diff --git a/PKG-INFO b/PKG-INFO index 8ae08004..0e847f7e 100644 --- a/PKG-INFO +++ b/PKG-INFO @@ -1,250 +1,250 @@ Metadata-Version: 2.1 Name: swh.storage -Version: 0.36.0 +Version: 0.37.0 Summary: Software Heritage storage manager Home-page: https://forge.softwareheritage.org/diffusion/DSTO/ Author: Software Heritage developers Author-email: swh-devel@inria.fr License: UNKNOWN Project-URL: Bug Reports, https://forge.softwareheritage.org/maniphest Project-URL: Funding, https://www.softwareheritage.org/donate Project-URL: Source, https://forge.softwareheritage.org/source/swh-storage Project-URL: Documentation, https://docs.softwareheritage.org/devel/swh-storage/ Platform: UNKNOWN Classifier: Programming Language :: Python :: 3 Classifier: Intended Audience :: Developers Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3) Classifier: Operating System :: OS Independent Classifier: Development Status :: 5 - Production/Stable Requires-Python: >=3.7 Description-Content-Type: text/markdown Provides-Extra: testing Provides-Extra: journal License-File: LICENSE License-File: AUTHORS swh-storage =========== Abstraction layer over the archive, allowing to access all stored source code artifacts as well as their metadata. See the [documentation](https://docs.softwareheritage.org/devel/swh-storage/index.html) for more details. ## Quick start ### Dependencies Python tests for this module include tests that cannot be run without a local Postgresql database, so you need the Postgresql server executable on your machine (no need to have a running Postgresql server). They also expect a cassandra server. #### Debian-like host ``` $ sudo apt install libpq-dev postgresql-11 cassandra ``` #### Non Debian-like host The tests expects the path to `cassandra` to either be unspecified, it is then looked up at `/usr/sbin/cassandra`, either specified through the environment variable `SWH_CASSANDRA_BIN`. Optionally, you can avoid running the cassandra tests. ``` (swh) :~/swh-storage$ tox -- -m 'not cassandra' ``` ### Installation It is strongly recommended to use a virtualenv. In the following, we consider you work in a virtualenv named `swh`. See the [developer setup guide](https://docs.softwareheritage.org/devel/developer-setup.html#developer-setup) for a more details on how to setup a working environment. You can install the package directly from [pypi](https://pypi.org/p/swh.storage): ``` (swh) :~$ pip install swh.storage [...] ``` Or from sources: ``` (swh) :~$ git clone https://forge.softwareheritage.org/source/swh-storage.git [...] (swh) :~$ cd swh-storage (swh) :~/swh-storage$ pip install . [...] ``` Then you can check it's properly installed: ``` (swh) :~$ swh storage --help Usage: swh storage [OPTIONS] COMMAND [ARGS]... Software Heritage Storage tools. Options: -h, --help Show this message and exit. Commands: rpc-serve Software Heritage Storage RPC server. ``` ## Tests The best way of running Python tests for this module is to use [tox](https://tox.readthedocs.io/). ``` (swh) :~$ pip install tox ``` ### tox From the sources directory, simply use tox: ``` (swh) :~/swh-storage$ tox [...] ========= 315 passed, 6 skipped, 15 warnings in 40.86 seconds ========== _______________________________ summary ________________________________ flake8: commands succeeded py3: commands succeeded congratulations :) ``` Note: it is possible to set the `JAVA_HOME` environment variable to specify the version of the JVM to be used by Cassandra. For example, at the time of writing this, Cassandra does not support java 14, so one may want to use for example java 11: ``` (swh) :~/swh-storage$ export JAVA_HOME=/usr/lib/jvm/java-14-openjdk-amd64/bin/java (swh) :~/swh-storage$ tox [...] ``` ## Development The storage server can be locally started. It requires a configuration file and a running Postgresql database. ### Sample configuration A typical configuration `storage.yml` file is: ``` storage: cls: postgresql db: "dbname=softwareheritage-dev user= password=" objstorage: cls: pathslicing root: /tmp/swh-storage/ slicing: 0:2/2:4/4:6 ``` which means, this uses: - a local storage instance whose db connection is to `softwareheritage-dev` local instance, - the objstorage uses a local objstorage instance whose: - `root` path is /tmp/swh-storage, - slicing scheme is `0:2/2:4/4:6`. This means that the identifier of the content (sha1) which will be stored on disk at first level with the first 2 hex characters, the second level with the next 2 hex characters and the third level with the next 2 hex characters. And finally the complete hash file holding the raw content. For example: 00062f8bd330715c4f819373653d97b3cd34394c will be stored at 00/06/2f/00062f8bd330715c4f819373653d97b3cd34394c Note that the `root` path should exist on disk before starting the server. ### Starting the storage server If the python package has been properly installed (e.g. in a virtual env), you should be able to use the command: ``` (swh) :~/swh-storage$ swh storage rpc-serve storage.yml ``` This runs a local swh-storage api at 5002 port. ``` (swh) :~/swh-storage$ curl http://127.0.0.1:5002 Software Heritage storage server

You have reached the Software Heritage storage server.
See its documentation and API for more information

``` ### And then what? In your upper layer ([loader-git](https://forge.softwareheritage.org/source/swh-loader-git/), [loader-svn](https://forge.softwareheritage.org/source/swh-loader-svn/), etc...), you can define a remote storage with this snippet of yaml configuration. ``` storage: cls: remote url: http://localhost:5002/ ``` You could directly define a postgresql storage with the following snippet: ``` storage: cls: postgresql db: service=swh-dev objstorage: cls: pathslicing root: /home/storage/swh-storage/ slicing: 0:2/2:4/4:6 ``` ## Cassandra As an alternative to PostgreSQL, swh-storage can use Cassandra as a database backend. It can be used like this: ``` storage: cls: cassandra hosts: - localhost objstorage: cls: pathslicing root: /home/storage/swh-storage/ slicing: 0:2/2:4/4:6 ``` The Cassandra swh-storage implementation supports both Cassandra >= 4.0-alpha2 and ScyllaDB >= 4.4 (and possibly earlier versions, but this is untested). While the main code supports both transparently, running tests or configuring the schema requires specific code when using ScyllaDB, enabled by setting the `SWH_USE_SCYLLADB=1` environment variable. diff --git a/requirements-swh.txt b/requirements-swh.txt index 489c6ec6..1c2878ea 100644 --- a/requirements-swh.txt +++ b/requirements-swh.txt @@ -1,3 +1,4 @@ swh.core[db,http] >= 0.14.0 +swh.counters >= v0.8.0 swh.model >= 2.1.0 swh.objstorage >= 0.2.2 diff --git a/swh.storage.egg-info/PKG-INFO b/swh.storage.egg-info/PKG-INFO index 8ae08004..0e847f7e 100644 --- a/swh.storage.egg-info/PKG-INFO +++ b/swh.storage.egg-info/PKG-INFO @@ -1,250 +1,250 @@ Metadata-Version: 2.1 Name: swh.storage -Version: 0.36.0 +Version: 0.37.0 Summary: Software Heritage storage manager Home-page: https://forge.softwareheritage.org/diffusion/DSTO/ Author: Software Heritage developers Author-email: swh-devel@inria.fr License: UNKNOWN Project-URL: Bug Reports, https://forge.softwareheritage.org/maniphest Project-URL: Funding, https://www.softwareheritage.org/donate Project-URL: Source, https://forge.softwareheritage.org/source/swh-storage Project-URL: Documentation, https://docs.softwareheritage.org/devel/swh-storage/ Platform: UNKNOWN Classifier: Programming Language :: Python :: 3 Classifier: Intended Audience :: Developers Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3) Classifier: Operating System :: OS Independent Classifier: Development Status :: 5 - Production/Stable Requires-Python: >=3.7 Description-Content-Type: text/markdown Provides-Extra: testing Provides-Extra: journal License-File: LICENSE License-File: AUTHORS swh-storage =========== Abstraction layer over the archive, allowing to access all stored source code artifacts as well as their metadata. See the [documentation](https://docs.softwareheritage.org/devel/swh-storage/index.html) for more details. ## Quick start ### Dependencies Python tests for this module include tests that cannot be run without a local Postgresql database, so you need the Postgresql server executable on your machine (no need to have a running Postgresql server). They also expect a cassandra server. #### Debian-like host ``` $ sudo apt install libpq-dev postgresql-11 cassandra ``` #### Non Debian-like host The tests expects the path to `cassandra` to either be unspecified, it is then looked up at `/usr/sbin/cassandra`, either specified through the environment variable `SWH_CASSANDRA_BIN`. Optionally, you can avoid running the cassandra tests. ``` (swh) :~/swh-storage$ tox -- -m 'not cassandra' ``` ### Installation It is strongly recommended to use a virtualenv. In the following, we consider you work in a virtualenv named `swh`. See the [developer setup guide](https://docs.softwareheritage.org/devel/developer-setup.html#developer-setup) for a more details on how to setup a working environment. You can install the package directly from [pypi](https://pypi.org/p/swh.storage): ``` (swh) :~$ pip install swh.storage [...] ``` Or from sources: ``` (swh) :~$ git clone https://forge.softwareheritage.org/source/swh-storage.git [...] (swh) :~$ cd swh-storage (swh) :~/swh-storage$ pip install . [...] ``` Then you can check it's properly installed: ``` (swh) :~$ swh storage --help Usage: swh storage [OPTIONS] COMMAND [ARGS]... Software Heritage Storage tools. Options: -h, --help Show this message and exit. Commands: rpc-serve Software Heritage Storage RPC server. ``` ## Tests The best way of running Python tests for this module is to use [tox](https://tox.readthedocs.io/). ``` (swh) :~$ pip install tox ``` ### tox From the sources directory, simply use tox: ``` (swh) :~/swh-storage$ tox [...] ========= 315 passed, 6 skipped, 15 warnings in 40.86 seconds ========== _______________________________ summary ________________________________ flake8: commands succeeded py3: commands succeeded congratulations :) ``` Note: it is possible to set the `JAVA_HOME` environment variable to specify the version of the JVM to be used by Cassandra. For example, at the time of writing this, Cassandra does not support java 14, so one may want to use for example java 11: ``` (swh) :~/swh-storage$ export JAVA_HOME=/usr/lib/jvm/java-14-openjdk-amd64/bin/java (swh) :~/swh-storage$ tox [...] ``` ## Development The storage server can be locally started. It requires a configuration file and a running Postgresql database. ### Sample configuration A typical configuration `storage.yml` file is: ``` storage: cls: postgresql db: "dbname=softwareheritage-dev user= password=" objstorage: cls: pathslicing root: /tmp/swh-storage/ slicing: 0:2/2:4/4:6 ``` which means, this uses: - a local storage instance whose db connection is to `softwareheritage-dev` local instance, - the objstorage uses a local objstorage instance whose: - `root` path is /tmp/swh-storage, - slicing scheme is `0:2/2:4/4:6`. This means that the identifier of the content (sha1) which will be stored on disk at first level with the first 2 hex characters, the second level with the next 2 hex characters and the third level with the next 2 hex characters. And finally the complete hash file holding the raw content. For example: 00062f8bd330715c4f819373653d97b3cd34394c will be stored at 00/06/2f/00062f8bd330715c4f819373653d97b3cd34394c Note that the `root` path should exist on disk before starting the server. ### Starting the storage server If the python package has been properly installed (e.g. in a virtual env), you should be able to use the command: ``` (swh) :~/swh-storage$ swh storage rpc-serve storage.yml ``` This runs a local swh-storage api at 5002 port. ``` (swh) :~/swh-storage$ curl http://127.0.0.1:5002 Software Heritage storage server

You have reached the Software Heritage storage server.
See its documentation and API for more information

``` ### And then what? In your upper layer ([loader-git](https://forge.softwareheritage.org/source/swh-loader-git/), [loader-svn](https://forge.softwareheritage.org/source/swh-loader-svn/), etc...), you can define a remote storage with this snippet of yaml configuration. ``` storage: cls: remote url: http://localhost:5002/ ``` You could directly define a postgresql storage with the following snippet: ``` storage: cls: postgresql db: service=swh-dev objstorage: cls: pathslicing root: /home/storage/swh-storage/ slicing: 0:2/2:4/4:6 ``` ## Cassandra As an alternative to PostgreSQL, swh-storage can use Cassandra as a database backend. It can be used like this: ``` storage: cls: cassandra hosts: - localhost objstorage: cls: pathslicing root: /home/storage/swh-storage/ slicing: 0:2/2:4/4:6 ``` The Cassandra swh-storage implementation supports both Cassandra >= 4.0-alpha2 and ScyllaDB >= 4.4 (and possibly earlier versions, but this is untested). While the main code supports both transparently, running tests or configuring the schema requires specific code when using ScyllaDB, enabled by setting the `SWH_USE_SCYLLADB=1` environment variable. diff --git a/swh.storage.egg-info/SOURCES.txt b/swh.storage.egg-info/SOURCES.txt index 1a49b49b..d8757763 100644 --- a/swh.storage.egg-info/SOURCES.txt +++ b/swh.storage.egg-info/SOURCES.txt @@ -1,330 +1,332 @@ .gitignore .pre-commit-config.yaml AUTHORS CODE_OF_CONDUCT.md CONTRIBUTORS LICENSE MANIFEST.in Makefile Makefile.local README.md conftest.py mypy.ini pyproject.toml pytest.ini requirements-swh-journal.txt requirements-swh.txt requirements-test.txt requirements.txt setup.cfg setup.py tox.ini ./requirements-swh-journal.txt ./requirements-swh.txt ./requirements-test.txt ./requirements.txt bin/swh-storage-add-dir docs/.gitignore docs/Makefile docs/Makefile.local docs/archive-copies.rst docs/cli.rst docs/conf.py docs/extrinsic-metadata-specification.rst docs/index.rst docs/sql-storage.rst docs/_static/.placeholder docs/_templates/.placeholder docs/images/.gitignore docs/images/Makefile docs/images/swh-archive-copies.dia sql/.gitignore sql/Makefile sql/TODO sql/clusters.dot sql/bin/db-upgrade sql/bin/dot_add_content sql/doc/json sql/doc/json/.gitignore sql/doc/json/Makefile sql/doc/json/entity.lister_metadata.schema.json sql/doc/json/entity.metadata.schema.json sql/doc/json/entity_history.lister_metadata.schema.json sql/doc/json/entity_history.metadata.schema.json sql/doc/json/fetch_history.result.schema.json sql/doc/json/list_history.result.schema.json sql/doc/json/listable_entity.list_params.schema.json sql/doc/json/origin_visit.metadata.json sql/doc/json/tool.tool_configuration.schema.json sql/json/.gitignore sql/json/Makefile sql/json/entity.lister_metadata.schema.json sql/json/entity.metadata.schema.json sql/json/entity_history.lister_metadata.schema.json sql/json/entity_history.metadata.schema.json sql/json/fetch_history.result.schema.json sql/json/list_history.result.schema.json sql/json/listable_entity.list_params.schema.json sql/json/origin_visit.metadata.json sql/json/tool.tool_configuration.schema.json sql/upgrades/015.sql sql/upgrades/016.sql sql/upgrades/017.sql sql/upgrades/018.sql sql/upgrades/019.sql sql/upgrades/020.sql sql/upgrades/021.sql sql/upgrades/022.sql sql/upgrades/023.sql sql/upgrades/024.sql sql/upgrades/025.sql sql/upgrades/026.sql sql/upgrades/027.sql sql/upgrades/028.sql sql/upgrades/029.sql sql/upgrades/030.sql sql/upgrades/032.sql sql/upgrades/033.sql sql/upgrades/034.sql sql/upgrades/035.sql sql/upgrades/036.sql sql/upgrades/037.sql sql/upgrades/038.sql sql/upgrades/039.sql sql/upgrades/040.sql sql/upgrades/041.sql sql/upgrades/042.sql sql/upgrades/043.sql sql/upgrades/044.sql sql/upgrades/045.sql sql/upgrades/046.sql sql/upgrades/047.sql sql/upgrades/048.sql sql/upgrades/049.sql sql/upgrades/050.sql sql/upgrades/051.sql sql/upgrades/052.sql sql/upgrades/053.sql sql/upgrades/054.sql sql/upgrades/055.sql sql/upgrades/056.sql sql/upgrades/057.sql sql/upgrades/058.sql sql/upgrades/059.sql sql/upgrades/060.sql sql/upgrades/061.sql sql/upgrades/062.sql sql/upgrades/063.sql sql/upgrades/064.sql sql/upgrades/065.sql sql/upgrades/066.sql sql/upgrades/067.sql sql/upgrades/068.sql sql/upgrades/069.sql sql/upgrades/070.sql sql/upgrades/071.sql sql/upgrades/072.sql sql/upgrades/073.sql sql/upgrades/074.sql sql/upgrades/075.sql sql/upgrades/076.sql sql/upgrades/077.sql sql/upgrades/078.sql sql/upgrades/079.sql sql/upgrades/080.sql sql/upgrades/081.sql sql/upgrades/082.sql sql/upgrades/083.sql sql/upgrades/084.sql sql/upgrades/085.sql sql/upgrades/086.sql sql/upgrades/087.sql sql/upgrades/088.sql sql/upgrades/089.sql sql/upgrades/090.sql sql/upgrades/091.sql sql/upgrades/092.sql sql/upgrades/093.sql sql/upgrades/094.sql sql/upgrades/095.sql sql/upgrades/096.sql sql/upgrades/097.sql sql/upgrades/098.sql sql/upgrades/099.sql sql/upgrades/100.sql sql/upgrades/101.sql sql/upgrades/102.sql sql/upgrades/103.sql sql/upgrades/104.sql sql/upgrades/105.sql sql/upgrades/106.sql sql/upgrades/107.sql sql/upgrades/108.sql sql/upgrades/109.sql sql/upgrades/110.sql sql/upgrades/111.sql sql/upgrades/112.sql sql/upgrades/113.sql sql/upgrades/114.sql sql/upgrades/115.sql sql/upgrades/116.sql sql/upgrades/117.sql sql/upgrades/118.sql sql/upgrades/119.sql sql/upgrades/120.sql sql/upgrades/121.sql sql/upgrades/122.sql sql/upgrades/123.sql sql/upgrades/124.sql sql/upgrades/125.sql sql/upgrades/126.sql sql/upgrades/127.sql sql/upgrades/128.sql sql/upgrades/129.sql sql/upgrades/130.sql sql/upgrades/131.sql sql/upgrades/132.sql sql/upgrades/133.sql sql/upgrades/134.sql sql/upgrades/135.sql sql/upgrades/136.sql sql/upgrades/137.sql sql/upgrades/138.sql sql/upgrades/139.sql sql/upgrades/140.sql sql/upgrades/141.sql sql/upgrades/142.sql sql/upgrades/143.sql sql/upgrades/144.sql sql/upgrades/145.sql sql/upgrades/146.sql sql/upgrades/147.sql sql/upgrades/148.sql sql/upgrades/149.sql sql/upgrades/150.sql sql/upgrades/151.sql sql/upgrades/152.sql sql/upgrades/153.sql sql/upgrades/154.sql sql/upgrades/155.sql sql/upgrades/156.sql sql/upgrades/157.sql sql/upgrades/158.sql sql/upgrades/159.sql sql/upgrades/160.sql sql/upgrades/161.sql sql/upgrades/162.sql sql/upgrades/163.sql sql/upgrades/164.sql sql/upgrades/165.sql sql/upgrades/166.sql sql/upgrades/167.sql sql/upgrades/168.sql sql/upgrades/169.sql sql/upgrades/170.sql sql/upgrades/171.sql sql/upgrades/172.sql sql/upgrades/173.sql sql/upgrades/174.sql sql/upgrades/175.sql sql/upgrades/176.sql sql/upgrades/177.sql swh/__init__.py swh.storage.egg-info/PKG-INFO swh.storage.egg-info/SOURCES.txt swh.storage.egg-info/dependency_links.txt swh.storage.egg-info/entry_points.txt swh.storage.egg-info/requires.txt swh.storage.egg-info/top_level.txt swh/storage/__init__.py swh/storage/backfill.py swh/storage/cli.py swh/storage/common.py swh/storage/exc.py swh/storage/fixer.py swh/storage/in_memory.py swh/storage/interface.py swh/storage/metrics.py swh/storage/migrate_extrinsic_metadata.py swh/storage/objstorage.py swh/storage/py.typed swh/storage/pytest_plugin.py swh/storage/replay.py swh/storage/utils.py swh/storage/writer.py swh/storage/algos/__init__.py swh/storage/algos/diff.py swh/storage/algos/dir_iterators.py swh/storage/algos/origin.py swh/storage/algos/revisions_walker.py swh/storage/algos/snapshot.py swh/storage/api/__init__.py swh/storage/api/client.py swh/storage/api/serializers.py swh/storage/api/server.py swh/storage/cassandra/__init__.py swh/storage/cassandra/common.py swh/storage/cassandra/converters.py swh/storage/cassandra/cql.py swh/storage/cassandra/model.py swh/storage/cassandra/schema.py swh/storage/cassandra/storage.py swh/storage/postgresql/__init__.py swh/storage/postgresql/converters.py swh/storage/postgresql/db.py swh/storage/postgresql/storage.py swh/storage/proxies/buffer.py +swh/storage/proxies/counter.py swh/storage/proxies/filter.py swh/storage/proxies/retry.py swh/storage/proxies/tenacious.py swh/storage/proxies/validate.py swh/storage/sql/10-superuser-init.sql swh/storage/sql/15-flavor.sql swh/storage/sql/20-enums.sql swh/storage/sql/30-schema.sql swh/storage/sql/40-funcs.sql swh/storage/sql/60-indexes.sql swh/storage/sql/logical_replication/replication_source.sql swh/storage/tests/__init__.py swh/storage/tests/conftest.py swh/storage/tests/storage_data.py swh/storage/tests/storage_tests.py swh/storage/tests/test_api_client.py swh/storage/tests/test_backfill.py swh/storage/tests/test_buffer.py swh/storage/tests/test_cassandra.py swh/storage/tests/test_cassandra_converters.py swh/storage/tests/test_cassandra_migration.py swh/storage/tests/test_cli.py +swh/storage/tests/test_counter.py swh/storage/tests/test_exception.py swh/storage/tests/test_filter.py swh/storage/tests/test_in_memory.py swh/storage/tests/test_init.py swh/storage/tests/test_kafka_writer.py swh/storage/tests/test_metrics.py swh/storage/tests/test_postgresql.py swh/storage/tests/test_postgresql_converters.py swh/storage/tests/test_pytest_plugin.py swh/storage/tests/test_replay.py swh/storage/tests/test_retry.py swh/storage/tests/test_revision_bw_compat.py swh/storage/tests/test_serializers.py swh/storage/tests/test_server.py swh/storage/tests/test_storage_data.py swh/storage/tests/test_tenacious.py swh/storage/tests/test_utils.py swh/storage/tests/test_validate.py swh/storage/tests/algos/__init__.py swh/storage/tests/algos/test_diff.py swh/storage/tests/algos/test_dir_iterator.py swh/storage/tests/algos/test_origin.py swh/storage/tests/algos/test_revisions_walker.py swh/storage/tests/algos/test_snapshot.py swh/storage/tests/data/storage.yml swh/storage/tests/migrate_extrinsic_metadata/test_cran.py swh/storage/tests/migrate_extrinsic_metadata/test_debian.py swh/storage/tests/migrate_extrinsic_metadata/test_deposit.py swh/storage/tests/migrate_extrinsic_metadata/test_gnu.py swh/storage/tests/migrate_extrinsic_metadata/test_nixguix.py swh/storage/tests/migrate_extrinsic_metadata/test_npm.py swh/storage/tests/migrate_extrinsic_metadata/test_pypi.py \ No newline at end of file diff --git a/swh.storage.egg-info/requires.txt b/swh.storage.egg-info/requires.txt index 398a78e1..22de944d 100644 --- a/swh.storage.egg-info/requires.txt +++ b/swh.storage.egg-info/requires.txt @@ -1,29 +1,30 @@ click flask psycopg2 aiohttp tenacity cassandra-driver!=3.21.0,>=3.19.0 deprecated typing-extensions mypy_extensions iso8601 swh.core[db,http]>=0.14.0 +swh.counters>=v0.8.0 swh.model>=2.1.0 swh.objstorage>=0.2.2 [journal] swh.journal>=0.6.2 [testing] hypothesis>=3.11.0 pytest pytest-mock swh.model[testing]>=0.0.50 pytz pytest-xdist types-python-dateutil types-pytz types-pyyaml types-requests swh.journal>=0.6.2 diff --git a/swh/storage/__init__.py b/swh/storage/__init__.py index 0a76a202..cd342de8 100644 --- a/swh/storage/__init__.py +++ b/swh/storage/__init__.py @@ -1,121 +1,122 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import importlib from typing import TYPE_CHECKING, Any, Dict, List import warnings if TYPE_CHECKING: from .interface import StorageInterface STORAGE_IMPLEMENTATIONS = { "remote": ".api.client.RemoteStorage", "memory": ".in_memory.InMemoryStorage", "cassandra": ".cassandra.CassandraStorage", "postgresql": ".postgresql.storage.Storage", # deprecated "local": ".postgresql.storage.Storage", # proxy storages - "filter": ".proxies.filter.FilteringProxyStorage", "buffer": ".proxies.buffer.BufferingProxyStorage", + "counter": ".proxies.counter.CountingProxyStorage", + "filter": ".proxies.filter.FilteringProxyStorage", "retry": ".proxies.retry.RetryingProxyStorage", - "validate": ".proxies.validate.ValidatingProxyStorage", "tenacious": ".proxies.tenacious.TenaciousProxyStorage", + "validate": ".proxies.validate.ValidatingProxyStorage", } def get_storage(cls: str, **kwargs) -> "StorageInterface": """Get a storage object of class `storage_class` with arguments `storage_args`. Args: cls (str): storage's class, can be: - ``local`` to use a postgresql database - ``cassandra`` to use a cassandra database - ``remote`` to connect to a swh-storage server - ``memory`` for an in-memory storage, useful for fast tests - ``filter``, ``buffer``, ... to use specific storage "proxies", see their respective documentations args (dict): dictionary with keys Returns: an instance of swh.storage.Storage or compatible class Raises: ValueError if passed an unknown storage class. """ if "args" in kwargs: warnings.warn( 'Explicit "args" key is deprecated, use keys directly instead.', DeprecationWarning, ) kwargs = kwargs["args"] if cls == "pipeline": return get_storage_pipeline(**kwargs) if cls == "local": warnings.warn( 'The "local" storage class is deprecated, use "postgresql" instead.', DeprecationWarning, ) class_path = STORAGE_IMPLEMENTATIONS.get(cls) if class_path is None: raise ValueError( "Unknown storage class `%s`. Supported: %s" % (cls, ", ".join(STORAGE_IMPLEMENTATIONS)) ) (module_path, class_name) = class_path.rsplit(".", 1) module = importlib.import_module(module_path, package=__package__) Storage = getattr(module, class_name) check_config = kwargs.pop("check_config", {}) storage = Storage(**kwargs) if check_config: if not storage.check_config(**check_config): raise EnvironmentError("storage check config failed") return storage def get_storage_pipeline( steps: List[Dict[str, Any]], check_config=None ) -> "StorageInterface": """Recursively get a storage object that may use other storage objects as backends. Args: steps (List[dict]): List of dicts that may be used as kwargs for `get_storage`. Returns: an instance of swh.storage.Storage or compatible class Raises: ValueError if passed an unknown storage class. """ storage_config = None for step in reversed(steps): if "args" in step: warnings.warn( 'Explicit "args" key is deprecated, use keys directly ' "instead.", DeprecationWarning, ) step = { "cls": step["cls"], **step["args"], } if storage_config: step["storage"] = storage_config step["check_config"] = check_config storage_config = step if storage_config is None: raise ValueError("'pipeline' has no steps.") return get_storage(**storage_config) diff --git a/swh/storage/cassandra/cql.py b/swh/storage/cassandra/cql.py index 7214dbdc..2bda8d51 100644 --- a/swh/storage/cassandra/cql.py +++ b/swh/storage/cassandra/cql.py @@ -1,1337 +1,1448 @@ # Copyright (C) 2019-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from collections import Counter import dataclasses import datetime import functools +import itertools import logging import random from typing import ( Any, Callable, Dict, Iterable, Iterator, List, Optional, + Sequence, Tuple, Type, TypeVar, Union, ) from cassandra import ConsistencyLevel, CoordinationFailure from cassandra.cluster import EXEC_PROFILE_DEFAULT, Cluster, ExecutionProfile, ResultSet +from cassandra.concurrent import execute_concurrent_with_args from cassandra.policies import DCAwareRoundRobinPolicy, TokenAwarePolicy from cassandra.query import BoundStatement, PreparedStatement, dict_factory from mypy_extensions import NamedArg from tenacity import ( retry, retry_if_exception_type, stop_after_attempt, wait_random_exponential, ) from swh.core.utils import grouper from swh.model.identifiers import CoreSWHID from swh.model.model import ( Content, Person, Sha1Git, SkippedContent, Timestamp, TimestampWithTimezone, ) from swh.storage.interface import ListOrder from ..utils import remove_keys from .common import TOKEN_BEGIN, TOKEN_END, hash_url from .model import ( MAGIC_NULL_PK, BaseRow, ContentRow, DirectoryEntryRow, DirectoryRow, ExtIDByTargetRow, ExtIDRow, MetadataAuthorityRow, MetadataFetcherRow, ObjectCountRow, OriginRow, OriginVisitRow, OriginVisitStatusRow, RawExtrinsicMetadataByIdRow, RawExtrinsicMetadataRow, ReleaseRow, RevisionParentRow, RevisionRow, SkippedContentRow, SnapshotBranchRow, SnapshotRow, content_index_table_name, ) from .schema import CREATE_TABLES_QUERIES, HASH_ALGORITHMS PARTITION_KEY_RESTRICTION_MAX_SIZE = 100 """Maximum number of restrictions in a single query. Usually this is a very low number (eg. SELECT ... FROM ... WHERE x=?), but some queries can request arbitrarily many (eg. SELECT ... FROM ... WHERE x IN ?). This can cause performance issues, as the node getting the query need to coordinate with other nodes to get the complete results. See for details and rationale. """ +BATCH_INSERT_MAX_SIZE = 1000 + logger = logging.getLogger(__name__) def get_execution_profiles( consistency_level: str = "ONE", ) -> Dict[object, ExecutionProfile]: if consistency_level not in ConsistencyLevel.name_to_value: raise ValueError( f"Configuration error: Unknown consistency level '{consistency_level}'" ) return { EXEC_PROFILE_DEFAULT: ExecutionProfile( load_balancing_policy=TokenAwarePolicy(DCAwareRoundRobinPolicy()), row_factory=dict_factory, consistency_level=ConsistencyLevel.name_to_value[consistency_level], ) } # Configuration for cassandra-driver's access to servers: # * hit the right server directly when sending a query (TokenAwarePolicy), # * if there's more than one, then pick one at random that's in the same # datacenter as the client (DCAwareRoundRobinPolicy) def create_keyspace( hosts: List[str], keyspace: str, port: int = 9042, *, durable_writes=True ): cluster = Cluster(hosts, port=port, execution_profiles=get_execution_profiles()) session = cluster.connect() extra_params = "" if not durable_writes: extra_params = "AND durable_writes = false" session.execute( """CREATE KEYSPACE IF NOT EXISTS "%s" WITH REPLICATION = { 'class' : 'SimpleStrategy', 'replication_factor' : 1 } %s; """ % (keyspace, extra_params) ) session.execute('USE "%s"' % keyspace) for query in CREATE_TABLES_QUERIES: session.execute(query) TRet = TypeVar("TRet") def _prepared_statement( query: str, ) -> Callable[[Callable[..., TRet]], Callable[..., TRet]]: """Returns a decorator usable on methods of CqlRunner, to inject them with a 'statement' argument, that is a prepared statement corresponding to the query. This only works on methods of CqlRunner, as preparing a statement requires a connection to a Cassandra server.""" def decorator(f): @functools.wraps(f) def newf(self, *args, **kwargs) -> TRet: if f.__name__ not in self._prepared_statements: statement: PreparedStatement = self._session.prepare(query) self._prepared_statements[f.__name__] = statement return f( self, *args, **kwargs, statement=self._prepared_statements[f.__name__] ) return newf return decorator TArg = TypeVar("TArg") TSelf = TypeVar("TSelf") +def _insert_query(row_class): + columns = row_class.cols() + return ( + f"INSERT INTO {row_class.TABLE} ({', '.join(columns)}) " + f"VALUES ({', '.join('?' for _ in columns)})" + ) + + def _prepared_insert_statement( row_class: Type[BaseRow], ) -> Callable[ [Callable[[TSelf, TArg, NamedArg(Any, "statement")], TRet]], # noqa Callable[[TSelf, TArg], TRet], ]: """Shorthand for using `_prepared_statement` for `INSERT INTO` statements.""" - columns = row_class.cols() - return _prepared_statement( - "INSERT INTO %s (%s) VALUES (%s)" - % (row_class.TABLE, ", ".join(columns), ", ".join("?" for _ in columns),) - ) + return _prepared_statement(_insert_query(row_class)) def _prepared_exists_statement( table_name: str, ) -> Callable[ [Callable[[TSelf, TArg, NamedArg(Any, "statement")], TRet]], # noqa Callable[[TSelf, TArg], TRet], ]: """Shorthand for using `_prepared_statement` for queries that only check which ids in a list exist in the table.""" return _prepared_statement(f"SELECT id FROM {table_name} WHERE id IN ?") def _prepared_select_statement( row_class: Type[BaseRow], clauses: str = "", cols: Optional[List[str]] = None, ) -> Callable[[Callable[..., TRet]], Callable[..., TRet]]: if cols is None: cols = row_class.cols() return _prepared_statement( f"SELECT {', '.join(cols)} FROM {row_class.TABLE} {clauses}" ) def _prepared_select_statements( row_class: Type[BaseRow], queries: Dict[Any, str], ) -> Callable[[Callable[..., TRet]], Callable[..., TRet]]: """Like _prepared_statement, but supports multiple statements, passed a dict, and passes a dict of prepared statements to the decorated method""" cols = row_class.cols() statement_start = f"SELECT {', '.join(cols)} FROM {row_class.TABLE} " def decorator(f): @functools.wraps(f) def newf(self, *args, **kwargs) -> TRet: if f.__name__ not in self._prepared_statements: self._prepared_statements[f.__name__] = { key: self._session.prepare(statement_start + query) for (key, query) in queries.items() } return f( self, *args, **kwargs, statements=self._prepared_statements[f.__name__] ) return newf return decorator def _next_bytes_value(value: bytes) -> bytes: """Returns the next bytes value by incrementing the integer representation of the provided value and converting it back to bytes. For instance when prefix is b"abcd", it returns b"abce". """ next_value_int = int.from_bytes(value, byteorder="big") + 1 return next_value_int.to_bytes( (next_value_int.bit_length() + 7) // 8, byteorder="big" ) class CqlRunner: """Class managing prepared statements and building queries to be sent to Cassandra.""" def __init__( self, hosts: List[str], keyspace: str, port: int, consistency_level: str ): self._cluster = Cluster( hosts, port=port, execution_profiles=get_execution_profiles(consistency_level), ) self._session = self._cluster.connect(keyspace) self._cluster.register_user_type( keyspace, "microtimestamp_with_timezone", TimestampWithTimezone ) self._cluster.register_user_type(keyspace, "microtimestamp", Timestamp) self._cluster.register_user_type(keyspace, "person", Person) # directly a PreparedStatement for methods decorated with # @_prepared_statements (and its wrappers, _prepared_insert_statement, # _prepared_exists_statement, and _prepared_select_statement); # and a dict of PreparedStatements with @_prepared_select_statements self._prepared_statements: Dict[ str, Union[PreparedStatement, Dict[Any, PreparedStatement]] ] = {} ########################## # Common utility functions ########################## MAX_RETRIES = 3 @retry( wait=wait_random_exponential(multiplier=1, max=10), stop=stop_after_attempt(MAX_RETRIES), retry=retry_if_exception_type(CoordinationFailure), ) - def _execute_with_retries(self, statement, args) -> ResultSet: + def _execute_with_retries(self, statement, args: Optional[Sequence]) -> ResultSet: return self._session.execute(statement, args, timeout=1000.0) - @_prepared_statement( - "UPDATE object_count SET count = count + ? " - "WHERE partition_key = 0 AND object_type = ?" + @retry( + wait=wait_random_exponential(multiplier=1, max=10), + stop=stop_after_attempt(MAX_RETRIES), + retry=retry_if_exception_type(CoordinationFailure), ) - def _increment_counter( - self, object_type: str, nb: int, *, statement: PreparedStatement - ) -> None: - self._execute_with_retries(statement, [nb, object_type]) + def _execute_many_with_retries( + self, statement, args_list: List[Tuple] + ) -> Iterable[BaseRow]: + for res in execute_concurrent_with_args(self._session, statement, args_list): + yield from res.result_or_exc def _add_one(self, statement, obj: BaseRow) -> None: - self._increment_counter(obj.TABLE, 1) self._execute_with_retries(statement, dataclasses.astuple(obj)) + def _add_many(self, statement, objs: Sequence[BaseRow]) -> None: + tables = {obj.TABLE for obj in objs} + assert len(tables) == 1, f"Cannot insert to multiple tables: {tables}" + rows = list(map(dataclasses.astuple, objs)) + for _ in self._execute_many_with_retries(statement, rows): + # Need to consume the generator to actually run the INSERTs + pass + _T = TypeVar("_T", bound=BaseRow) def _get_random_row(self, row_class: Type[_T], statement) -> Optional[_T]: # noqa """Takes a prepared statement of the form "SELECT * FROM WHERE token() > ? LIMIT 1" and uses it to return a random row""" token = random.randint(TOKEN_BEGIN, TOKEN_END) rows = self._execute_with_retries(statement, [token]) if not rows: # There are no row with a greater token; wrap around to get # the row with the smallest token rows = self._execute_with_retries(statement, [TOKEN_BEGIN]) if rows: return row_class.from_dict(rows.one()) # type: ignore else: return None def _missing(self, statement, ids): found_ids = set() for id_group in grouper(ids, PARTITION_KEY_RESTRICTION_MAX_SIZE): rows = self._execute_with_retries(statement, [list(id_group)]) found_ids.update(row["id"] for row in rows) return [id_ for id_ in ids if id_ not in found_ids] ########################## # 'content' table ########################## def _content_add_finalize(self, statement: BoundStatement) -> None: """Returned currified by content_add_prepare, to be called when the content row should be added to the primary table.""" self._execute_with_retries(statement, None) - self._increment_counter("content", 1) @_prepared_insert_statement(ContentRow) def content_add_prepare( self, content: ContentRow, *, statement ) -> Tuple[int, Callable[[], None]]: """Prepares insertion of a Content to the main 'content' table. Returns a token (to be used in secondary tables), and a function to be called to perform the insertion in the main table.""" statement = statement.bind(dataclasses.astuple(content)) # Type used for hashing keys (usually, it will be # cassandra.metadata.Murmur3Token) token_class = self._cluster.metadata.token_map.token_class # Token of the row when it will be inserted. This is equivalent to # "SELECT token({', '.join(ContentRow.PARTITION_KEY)}) FROM content WHERE ..." # after the row is inserted; but we need the token to insert in the # index tables *before* inserting to the main 'content' table token = token_class.from_key(statement.routing_key).value assert TOKEN_BEGIN <= token <= TOKEN_END # Function to be called after the indexes contain their respective # row finalizer = functools.partial(self._content_add_finalize, statement) return (token, finalizer) @_prepared_select_statement( ContentRow, f"WHERE {' AND '.join(map('%s = ?'.__mod__, HASH_ALGORITHMS))}" ) def content_get_from_pk( self, content_hashes: Dict[str, bytes], *, statement ) -> Optional[ContentRow]: rows = list( self._execute_with_retries( statement, [content_hashes[algo] for algo in HASH_ALGORITHMS] ) ) assert len(rows) <= 1 if rows: return ContentRow(**rows[0]) else: return None def content_missing_from_hashes( self, contents_hashes: List[Dict[str, bytes]] ) -> Iterator[Dict[str, bytes]]: for group in grouper(contents_hashes, PARTITION_KEY_RESTRICTION_MAX_SIZE): group = list(group) # Get all contents that share a sha256 with one of the contents in the group present = set( self._content_get_hashes_from_sha256( [content["sha256"] for content in group] ) ) for content in group: for algo in HASH_ALGORITHMS: assert content.get(algo) is not None, ( "content_missing_from_hashes must not be called with " "partial hashes." ) if tuple(content[algo] for algo in HASH_ALGORITHMS) not in present: yield content @_prepared_statement( f"SELECT {', '.join(HASH_ALGORITHMS)} FROM content WHERE sha256 IN ?" ) def _content_get_hashes_from_sha256( self, ids: List[bytes], *, statement ) -> Iterator[Tuple[bytes, bytes, bytes, bytes]]: for row in self._execute_with_retries(statement, [ids]): yield tuple(row[algo] for algo in HASH_ALGORITHMS) # type: ignore @_prepared_select_statement( ContentRow, f"WHERE token({', '.join(ContentRow.PARTITION_KEY)}) = ?" ) def content_get_from_token(self, token, *, statement) -> Iterable[ContentRow]: return map(ContentRow.from_dict, self._execute_with_retries(statement, [token])) @_prepared_select_statement( ContentRow, f"WHERE token({', '.join(ContentRow.PARTITION_KEY)}) > ? LIMIT 1" ) def content_get_random(self, *, statement) -> Optional[ContentRow]: return self._get_random_row(ContentRow, statement) @_prepared_statement( """ SELECT token({pk}) AS tok, {cols} FROM {table} WHERE token({pk}) >= ? AND token({pk}) <= ? LIMIT ? """.format( pk=", ".join(ContentRow.PARTITION_KEY), cols=", ".join(ContentRow.cols()), table=ContentRow.TABLE, ) ) def content_get_token_range( self, start: int, end: int, limit: int, *, statement ) -> Iterable[Tuple[int, ContentRow]]: """Returns an iterable of (token, row)""" return ( (row["tok"], ContentRow.from_dict(remove_keys(row, ("tok",)))) for row in self._execute_with_retries(statement, [start, end, limit]) ) ########################## # 'content_by_*' tables ########################## @_prepared_statement( f""" SELECT sha1_git AS id FROM {content_index_table_name("sha1_git", skipped_content=False)} WHERE sha1_git IN ? """ ) def content_missing_by_sha1_git( self, ids: List[bytes], *, statement ) -> List[bytes]: return self._missing(statement, ids) def content_index_add_one(self, algo: str, content: Content, token: int) -> None: """Adds a row mapping content[algo] to the token of the Content in the main 'content' table.""" query = f""" INSERT INTO {content_index_table_name(algo, skipped_content=False)} ({algo}, target_token) VALUES (%s, %s) """ self._execute_with_retries(query, [content.get_hash(algo), token]) - def content_get_tokens_from_single_hash( - self, algo: str, hash_: bytes + def content_get_tokens_from_single_algo( + self, algo: str, hashes: List[bytes] ) -> Iterable[int]: assert algo in HASH_ALGORITHMS query = f""" SELECT target_token FROM {content_index_table_name(algo, skipped_content=False)} WHERE {algo} = %s """ return ( - row["target_token"] for row in self._execute_with_retries(query, [hash_]) + row["target_token"] # type: ignore + for row in self._execute_many_with_retries( + query, [(hash_,) for hash_ in hashes] + ) ) ########################## # 'skipped_content' table ########################## def _skipped_content_add_finalize(self, statement: BoundStatement) -> None: """Returned currified by skipped_content_add_prepare, to be called when the content row should be added to the primary table.""" self._execute_with_retries(statement, None) - self._increment_counter("skipped_content", 1) @_prepared_insert_statement(SkippedContentRow) def skipped_content_add_prepare( self, content, *, statement ) -> Tuple[int, Callable[[], None]]: """Prepares insertion of a Content to the main 'skipped_content' table. Returns a token (to be used in secondary tables), and a function to be called to perform the insertion in the main table.""" # Replace NULLs (which are not allowed in the partition key) with # an empty byte string for key in SkippedContentRow.PARTITION_KEY: if getattr(content, key) is None: setattr(content, key, MAGIC_NULL_PK) statement = statement.bind(dataclasses.astuple(content)) # Type used for hashing keys (usually, it will be # cassandra.metadata.Murmur3Token) token_class = self._cluster.metadata.token_map.token_class # Token of the row when it will be inserted. This is equivalent to # "SELECT token({', '.join(SkippedContentRow.PARTITION_KEY)}) # FROM skipped_content WHERE ..." # after the row is inserted; but we need the token to insert in the # index tables *before* inserting to the main 'skipped_content' table token = token_class.from_key(statement.routing_key).value assert TOKEN_BEGIN <= token <= TOKEN_END # Function to be called after the indexes contain their respective # row finalizer = functools.partial(self._skipped_content_add_finalize, statement) return (token, finalizer) @_prepared_select_statement( SkippedContentRow, f"WHERE {' AND '.join(map('%s = ?'.__mod__, HASH_ALGORITHMS))}", ) def skipped_content_get_from_pk( self, content_hashes: Dict[str, bytes], *, statement ) -> Optional[SkippedContentRow]: rows = list( self._execute_with_retries( statement, [content_hashes[algo] or MAGIC_NULL_PK for algo in HASH_ALGORITHMS], ) ) assert len(rows) <= 1 if rows: return SkippedContentRow.from_dict(rows[0]) else: return None @_prepared_select_statement( SkippedContentRow, f"WHERE token({', '.join(SkippedContentRow.PARTITION_KEY)}) = ?", ) def skipped_content_get_from_token( self, token, *, statement ) -> Iterable[SkippedContentRow]: return map( SkippedContentRow.from_dict, self._execute_with_retries(statement, [token]) ) ########################## # 'skipped_content_by_*' tables ########################## def skipped_content_index_add_one( self, algo: str, content: SkippedContent, token: int ) -> None: """Adds a row mapping content[algo] to the token of the SkippedContent in the main 'skipped_content' table.""" query = ( f"INSERT INTO skipped_content_by_{algo} ({algo}, target_token) " f"VALUES (%s, %s)" ) self._execute_with_retries( query, [content.get_hash(algo) or MAGIC_NULL_PK, token] ) def skipped_content_get_tokens_from_single_hash( self, algo: str, hash_: bytes ) -> Iterable[int]: assert algo in HASH_ALGORITHMS query = f""" SELECT target_token FROM {content_index_table_name(algo, skipped_content=True)} WHERE {algo} = %s """ return ( row["target_token"] for row in self._execute_with_retries(query, [hash_]) ) ########################## # 'revision' table ########################## @_prepared_exists_statement("revision") def revision_missing(self, ids: List[bytes], *, statement) -> List[bytes]: return self._missing(statement, ids) @_prepared_insert_statement(RevisionRow) def revision_add_one(self, revision: RevisionRow, *, statement) -> None: self._add_one(statement, revision) @_prepared_statement(f"SELECT id FROM {RevisionRow.TABLE} WHERE id IN ?") def revision_get_ids(self, revision_ids, *, statement) -> Iterable[int]: return ( row["id"] for row in self._execute_with_retries(statement, [revision_ids]) ) @_prepared_select_statement(RevisionRow, "WHERE id IN ?") def revision_get( self, revision_ids: List[Sha1Git], *, statement ) -> Iterable[RevisionRow]: return map( RevisionRow.from_dict, self._execute_with_retries(statement, [revision_ids]) ) @_prepared_select_statement(RevisionRow, "WHERE token(id) > ? LIMIT 1") def revision_get_random(self, *, statement) -> Optional[RevisionRow]: return self._get_random_row(RevisionRow, statement) ########################## # 'revision_parent' table ########################## @_prepared_insert_statement(RevisionParentRow) def revision_parent_add_one( self, revision_parent: RevisionParentRow, *, statement ) -> None: self._add_one(statement, revision_parent) @_prepared_statement( f"SELECT parent_id FROM {RevisionParentRow.TABLE} WHERE id = ?" ) def revision_parent_get( self, revision_id: Sha1Git, *, statement ) -> Iterable[bytes]: return ( row["parent_id"] for row in self._execute_with_retries(statement, [revision_id]) ) ########################## # 'release' table ########################## @_prepared_exists_statement("release") def release_missing(self, ids: List[bytes], *, statement) -> List[bytes]: return self._missing(statement, ids) @_prepared_insert_statement(ReleaseRow) def release_add_one(self, release: ReleaseRow, *, statement) -> None: self._add_one(statement, release) @_prepared_select_statement(ReleaseRow, "WHERE id in ?") def release_get(self, release_ids: List[str], *, statement) -> Iterable[ReleaseRow]: return map( ReleaseRow.from_dict, self._execute_with_retries(statement, [release_ids]) ) @_prepared_select_statement(ReleaseRow, "WHERE token(id) > ? LIMIT 1") def release_get_random(self, *, statement) -> Optional[ReleaseRow]: return self._get_random_row(ReleaseRow, statement) ########################## # 'directory' table ########################## @_prepared_exists_statement("directory") def directory_missing(self, ids: List[bytes], *, statement) -> List[bytes]: return self._missing(statement, ids) @_prepared_insert_statement(DirectoryRow) def directory_add_one(self, directory: DirectoryRow, *, statement) -> None: """Called after all calls to directory_entry_add_one, to commit/finalize the directory.""" self._add_one(statement, directory) @_prepared_select_statement(DirectoryRow, "WHERE token(id) > ? LIMIT 1") def directory_get_random(self, *, statement) -> Optional[DirectoryRow]: return self._get_random_row(DirectoryRow, statement) ########################## # 'directory_entry' table ########################## @_prepared_insert_statement(DirectoryEntryRow) def directory_entry_add_one(self, entry: DirectoryEntryRow, *, statement) -> None: self._add_one(statement, entry) + @_prepared_insert_statement(DirectoryEntryRow) + def directory_entry_add_concurrent( + self, entries: List[DirectoryEntryRow], *, statement + ) -> None: + if len(entries) == 0: + # nothing to do + return + assert ( + len({entry.directory_id for entry in entries}) == 1 + ), "directory_entry_add_many must be called with entries for a single dir" + self._add_many(statement, entries) + + @_prepared_statement( + "BEGIN UNLOGGED BATCH\n" + + (_insert_query(DirectoryEntryRow) + ";\n") * BATCH_INSERT_MAX_SIZE + + "APPLY BATCH" + ) + def directory_entry_add_batch( + self, entries: List[DirectoryEntryRow], *, statement + ) -> None: + if len(entries) == 0: + # nothing to do + return + assert ( + len({entry.directory_id for entry in entries}) == 1 + ), "directory_entry_add_many must be called with entries for a single dir" + + for entry_group in grouper(entries, BATCH_INSERT_MAX_SIZE): + entry_group = list(entry_group) + if len(entry_group) == BATCH_INSERT_MAX_SIZE: + entry_group = list(map(dataclasses.astuple, entry_group)) + self._execute_with_retries( + statement, list(itertools.chain.from_iterable(entry_group)) + ) + else: + # Last group, with a smaller size than the BATCH we prepared. + # Creating a prepared BATCH just for this then discarding it would + # create too much churn on the server side; and using unprepared + # statements is annoying (we can't use _insert_query() as they have + # a different format) + # Fall back to inserting concurrently. + self.directory_entry_add_concurrent(entry_group) + @_prepared_select_statement(DirectoryEntryRow, "WHERE directory_id IN ?") def directory_entry_get( self, directory_ids, *, statement ) -> Iterable[DirectoryEntryRow]: return map( DirectoryEntryRow.from_dict, self._execute_with_retries(statement, [directory_ids]), ) @_prepared_select_statement( DirectoryEntryRow, "WHERE directory_id = ? AND name >= ? LIMIT ?" ) def directory_entry_get_from_name( self, directory_id: Sha1Git, from_: bytes, limit: int, *, statement ) -> Iterable[DirectoryEntryRow]: return map( DirectoryEntryRow.from_dict, self._execute_with_retries(statement, [directory_id, from_, limit]), ) ########################## # 'snapshot' table ########################## @_prepared_exists_statement("snapshot") def snapshot_missing(self, ids: List[bytes], *, statement) -> List[bytes]: return self._missing(statement, ids) @_prepared_insert_statement(SnapshotRow) def snapshot_add_one(self, snapshot: SnapshotRow, *, statement) -> None: self._add_one(statement, snapshot) @_prepared_select_statement(SnapshotRow, "WHERE token(id) > ? LIMIT 1") def snapshot_get_random(self, *, statement) -> Optional[SnapshotRow]: return self._get_random_row(SnapshotRow, statement) ########################## # 'snapshot_branch' table ########################## @_prepared_insert_statement(SnapshotBranchRow) def snapshot_branch_add_one(self, branch: SnapshotBranchRow, *, statement) -> None: self._add_one(statement, branch) @_prepared_statement( f""" SELECT ascii_bins_count(target_type) AS counts FROM {SnapshotBranchRow.TABLE} WHERE snapshot_id = ? AND name >= ? """ ) def snapshot_count_branches_from_name( self, snapshot_id: Sha1Git, from_: bytes, *, statement ) -> Dict[Optional[str], int]: row = self._execute_with_retries(statement, [snapshot_id, from_]).one() (nb_none, counts) = row["counts"] return {None: nb_none, **counts} @_prepared_statement( f""" SELECT ascii_bins_count(target_type) AS counts FROM {SnapshotBranchRow.TABLE} WHERE snapshot_id = ? AND name < ? """ ) def snapshot_count_branches_before_name( self, snapshot_id: Sha1Git, before: bytes, *, statement, ) -> Dict[Optional[str], int]: row = self._execute_with_retries(statement, [snapshot_id, before]).one() (nb_none, counts) = row["counts"] return {None: nb_none, **counts} def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Dict[Optional[str], int]: """Returns a dictionary from type names to the number of branches of that type.""" prefix = branch_name_exclude_prefix if prefix is None: return self.snapshot_count_branches_from_name(snapshot_id, b"") else: # counts branches before exclude prefix counts = Counter( self.snapshot_count_branches_before_name(snapshot_id, prefix) ) # no need to execute that part if each bit of the prefix equals 1 if prefix.replace(b"\xff", b"") != b"": # counts branches after exclude prefix and update counters counts.update( self.snapshot_count_branches_from_name( snapshot_id, _next_bytes_value(prefix) ) ) return counts @_prepared_select_statement( SnapshotBranchRow, "WHERE snapshot_id = ? AND name >= ? LIMIT ?" ) def snapshot_branch_get_from_name( self, snapshot_id: Sha1Git, from_: bytes, limit: int, *, statement ) -> Iterable[SnapshotBranchRow]: return map( SnapshotBranchRow.from_dict, self._execute_with_retries(statement, [snapshot_id, from_, limit]), ) @_prepared_select_statement( SnapshotBranchRow, "WHERE snapshot_id = ? AND name >= ? AND name < ? LIMIT ?" ) def snapshot_branch_get_range( self, snapshot_id: Sha1Git, from_: bytes, before: bytes, limit: int, *, statement, ) -> Iterable[SnapshotBranchRow]: return map( SnapshotBranchRow.from_dict, self._execute_with_retries(statement, [snapshot_id, from_, before, limit]), ) def snapshot_branch_get( self, snapshot_id: Sha1Git, from_: bytes, limit: int, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Iterable[SnapshotBranchRow]: prefix = branch_name_exclude_prefix if prefix is None: return self.snapshot_branch_get_from_name(snapshot_id, from_, limit) else: # get branches before the exclude prefix branches = list( self.snapshot_branch_get_range(snapshot_id, from_, prefix, limit) ) nb_branches = len(branches) # no need to execute that part if limit is reached # or if each bit of the prefix equals 1 if nb_branches < limit and prefix.replace(b"\xff", b"") != b"": # get branches after the exclude prefix and update list to return branches.extend( self.snapshot_branch_get_from_name( snapshot_id, _next_bytes_value(prefix), limit - nb_branches ) ) return branches ########################## # 'origin' table ########################## @_prepared_insert_statement(OriginRow) def origin_add_one(self, origin: OriginRow, *, statement) -> None: self._add_one(statement, origin) @_prepared_select_statement(OriginRow, "WHERE sha1 = ?") def origin_get_by_sha1(self, sha1: bytes, *, statement) -> Iterable[OriginRow]: return map(OriginRow.from_dict, self._execute_with_retries(statement, [sha1])) def origin_get_by_url(self, url: str) -> Iterable[OriginRow]: return self.origin_get_by_sha1(hash_url(url)) @_prepared_statement( f""" SELECT token(sha1) AS tok, {", ".join(OriginRow.cols())} FROM {OriginRow.TABLE} WHERE token(sha1) >= ? LIMIT ? """ ) def origin_list( self, start_token: int, limit: int, *, statement ) -> Iterable[Tuple[int, OriginRow]]: """Returns an iterable of (token, origin)""" return ( (row["tok"], OriginRow.from_dict(remove_keys(row, ("tok",)))) for row in self._execute_with_retries(statement, [start_token, limit]) ) @_prepared_select_statement(OriginRow) def origin_iter_all(self, *, statement) -> Iterable[OriginRow]: return map(OriginRow.from_dict, self._execute_with_retries(statement, [])) @_prepared_statement( f""" UPDATE {OriginRow.TABLE} SET next_visit_id=? WHERE sha1 = ? IF next_visit_id None: origin_sha1 = hash_url(origin_url) next_id = visit_id + 1 self._execute_with_retries(statement, [next_id, origin_sha1, next_id]) @_prepared_statement(f"SELECT next_visit_id FROM {OriginRow.TABLE} WHERE sha1 = ?") def _origin_get_next_visit_id(self, origin_sha1: bytes, *, statement) -> int: rows = list(self._execute_with_retries(statement, [origin_sha1])) assert len(rows) == 1 # TODO: error handling return rows[0]["next_visit_id"] @_prepared_statement( f""" UPDATE {OriginRow.TABLE} SET next_visit_id=? WHERE sha1 = ? IF next_visit_id=? """ ) def origin_generate_unique_visit_id(self, origin_url: str, *, statement) -> int: origin_sha1 = hash_url(origin_url) next_id = self._origin_get_next_visit_id(origin_sha1) while True: res = list( self._execute_with_retries( statement, [next_id + 1, origin_sha1, next_id] ) ) assert len(res) == 1 if res[0]["[applied]"]: # No data race return next_id else: # Someone else updated it before we did, let's try again next_id = res[0]["next_visit_id"] # TODO: abort after too many attempts return next_id ########################## # 'origin_visit' table ########################## @_prepared_select_statements( OriginVisitRow, { (True, ListOrder.ASC): ( "WHERE origin = ? AND visit > ? ORDER BY visit ASC LIMIT ?" ), (True, ListOrder.DESC): ( "WHERE origin = ? AND visit < ? ORDER BY visit DESC LIMIT ?" ), (False, ListOrder.ASC): "WHERE origin = ? ORDER BY visit ASC LIMIT ?", (False, ListOrder.DESC): "WHERE origin = ? ORDER BY visit DESC LIMIT ?", }, ) def origin_visit_get( self, origin_url: str, last_visit: Optional[int], limit: int, order: ListOrder, *, statements, ) -> Iterable[OriginVisitRow]: args: List[Any] = [origin_url] if last_visit is not None: args.append(last_visit) args.append(limit) statement = statements[(last_visit is not None, order)] return map( OriginVisitRow.from_dict, self._execute_with_retries(statement, args) ) @_prepared_insert_statement(OriginVisitRow) def origin_visit_add_one(self, visit: OriginVisitRow, *, statement) -> None: self._add_one(statement, visit) @_prepared_select_statement(OriginVisitRow, "WHERE origin = ? AND visit = ?") def origin_visit_get_one( self, origin_url: str, visit_id: int, *, statement ) -> Optional[OriginVisitRow]: # TODO: error handling rows = list(self._execute_with_retries(statement, [origin_url, visit_id])) if rows: return OriginVisitRow.from_dict(rows[0]) else: return None @_prepared_select_statement(OriginVisitRow, "WHERE origin = ?") def origin_visit_get_all( self, origin_url: str, *, statement ) -> Iterable[OriginVisitRow]: return map( OriginVisitRow.from_dict, self._execute_with_retries(statement, [origin_url]), ) @_prepared_select_statement(OriginVisitRow, "WHERE token(origin) >= ?") def _origin_visit_iter_from( self, min_token: int, *, statement ) -> Iterable[OriginVisitRow]: return map( OriginVisitRow.from_dict, self._execute_with_retries(statement, [min_token]) ) @_prepared_select_statement(OriginVisitRow, "WHERE token(origin) < ?") def _origin_visit_iter_to( self, max_token: int, *, statement ) -> Iterable[OriginVisitRow]: return map( OriginVisitRow.from_dict, self._execute_with_retries(statement, [max_token]) ) def origin_visit_iter(self, start_token: int) -> Iterator[OriginVisitRow]: """Returns all origin visits in order from this token, and wraps around the token space.""" yield from self._origin_visit_iter_from(start_token) yield from self._origin_visit_iter_to(start_token) ########################## # 'origin_visit_status' table ########################## @_prepared_select_statements( OriginVisitStatusRow, { (True, ListOrder.ASC): ( "WHERE origin = ? AND visit = ? AND date >= ? " "ORDER BY visit ASC LIMIT ?" ), (True, ListOrder.DESC): ( "WHERE origin = ? AND visit = ? AND date <= ? " "ORDER BY visit DESC LIMIT ?" ), (False, ListOrder.ASC): ( "WHERE origin = ? AND visit = ? ORDER BY visit ASC LIMIT ?" ), (False, ListOrder.DESC): ( "WHERE origin = ? AND visit = ? ORDER BY visit DESC LIMIT ?" ), }, ) def origin_visit_status_get_range( self, origin: str, visit: int, date_from: Optional[datetime.datetime], limit: int, order: ListOrder, *, statements, ) -> Iterable[OriginVisitStatusRow]: args: List[Any] = [origin, visit] if date_from is not None: args.append(date_from) args.append(limit) statement = statements[(date_from is not None, order)] return map( OriginVisitStatusRow.from_dict, self._execute_with_retries(statement, args) ) @_prepared_insert_statement(OriginVisitStatusRow) def origin_visit_status_add_one( self, visit_update: OriginVisitStatusRow, *, statement ) -> None: self._add_one(statement, visit_update) def origin_visit_status_get_latest( self, origin: str, visit: int, ) -> Optional[OriginVisitStatusRow]: """Given an origin visit id, return its latest origin_visit_status """ return next(self.origin_visit_status_get(origin, visit), None) @_prepared_select_statement( OriginVisitStatusRow, # 'visit DESC,' is optional with Cassandra 4, but ScyllaDB needs it "WHERE origin = ? AND visit = ? ORDER BY visit DESC, date DESC", ) def origin_visit_status_get( self, origin: str, visit: int, *, statement, ) -> Iterator[OriginVisitStatusRow]: """Return all origin visit statuses for a given visit """ return map( OriginVisitStatusRow.from_dict, self._execute_with_retries(statement, [origin, visit]), ) ########################## # 'metadata_authority' table ########################## @_prepared_insert_statement(MetadataAuthorityRow) def metadata_authority_add(self, authority: MetadataAuthorityRow, *, statement): self._add_one(statement, authority) @_prepared_select_statement(MetadataAuthorityRow, "WHERE type = ? AND url = ?") def metadata_authority_get( self, type, url, *, statement ) -> Optional[MetadataAuthorityRow]: rows = list(self._execute_with_retries(statement, [type, url])) if rows: return MetadataAuthorityRow.from_dict(rows[0]) else: return None ########################## # 'metadata_fetcher' table ########################## @_prepared_insert_statement(MetadataFetcherRow) def metadata_fetcher_add(self, fetcher, *, statement): self._add_one(statement, fetcher) @_prepared_select_statement(MetadataFetcherRow, "WHERE name = ? AND version = ?") def metadata_fetcher_get( self, name, version, *, statement ) -> Optional[MetadataFetcherRow]: rows = list(self._execute_with_retries(statement, [name, version])) if rows: return MetadataFetcherRow.from_dict(rows[0]) else: return None ######################### # 'raw_extrinsic_metadata_by_id' table ######################### @_prepared_insert_statement(RawExtrinsicMetadataByIdRow) def raw_extrinsic_metadata_by_id_add(self, row, *, statement): self._add_one(statement, row) @_prepared_select_statement(RawExtrinsicMetadataByIdRow, "WHERE id IN ?") def raw_extrinsic_metadata_get_by_ids( self, ids: List[Sha1Git], *, statement ) -> Iterable[RawExtrinsicMetadataByIdRow]: return map( RawExtrinsicMetadataByIdRow.from_dict, self._execute_with_retries(statement, [ids]), ) ######################### # 'raw_extrinsic_metadata' table ######################### @_prepared_insert_statement(RawExtrinsicMetadataRow) def raw_extrinsic_metadata_add(self, raw_extrinsic_metadata, *, statement): self._add_one(statement, raw_extrinsic_metadata) @_prepared_select_statement( RawExtrinsicMetadataRow, "WHERE target=? AND authority_url=? AND discovery_date>? AND authority_type=?", ) def raw_extrinsic_metadata_get_after_date( self, target: str, authority_type: str, authority_url: str, after: datetime.datetime, *, statement, ) -> Iterable[RawExtrinsicMetadataRow]: return map( RawExtrinsicMetadataRow.from_dict, self._execute_with_retries( statement, [target, authority_url, after, authority_type] ), ) @_prepared_select_statement( RawExtrinsicMetadataRow, # This is equivalent to: # WHERE target=? AND authority_type = ? AND authority_url = ? " # AND (discovery_date, id) > (?, ?)" # but it needs to be written this way to work with ScyllaDB. "WHERE target=? AND (authority_type, authority_url) <= (?, ?) " "AND (authority_type, authority_url, discovery_date, id) > (?, ?, ?, ?)", ) def raw_extrinsic_metadata_get_after_date_and_id( self, target: str, authority_type: str, authority_url: str, after_date: datetime.datetime, after_id: bytes, *, statement, ) -> Iterable[RawExtrinsicMetadataRow]: return map( RawExtrinsicMetadataRow.from_dict, self._execute_with_retries( statement, [ target, authority_type, authority_url, authority_type, authority_url, after_date, after_id, ], ), ) @_prepared_select_statement( RawExtrinsicMetadataRow, "WHERE target=? AND authority_url=? AND authority_type=?", ) def raw_extrinsic_metadata_get( self, target: str, authority_type: str, authority_url: str, *, statement ) -> Iterable[RawExtrinsicMetadataRow]: return map( RawExtrinsicMetadataRow.from_dict, self._execute_with_retries( statement, [target, authority_url, authority_type] ), ) @_prepared_statement( "SELECT authority_type, authority_url FROM raw_extrinsic_metadata " "WHERE target = ?" ) def raw_extrinsic_metadata_get_authorities( self, target: str, *, statement ) -> Iterable[Tuple[str, str]]: return ( (entry["authority_type"], entry["authority_url"]) for entry in self._execute_with_retries(statement, [target]) ) ########################## # 'extid' table ########################## def _extid_add_finalize(self, statement: BoundStatement) -> None: """Returned currified by extid_add_prepare, to be called when the extid row should be added to the primary table.""" self._execute_with_retries(statement, None) - self._increment_counter("extid", 1) @_prepared_insert_statement(ExtIDRow) def extid_add_prepare( self, extid: ExtIDRow, *, statement ) -> Tuple[int, Callable[[], None]]: statement = statement.bind(dataclasses.astuple(extid)) token_class = self._cluster.metadata.token_map.token_class token = token_class.from_key(statement.routing_key).value assert TOKEN_BEGIN <= token <= TOKEN_END # Function to be called after the indexes contain their respective # row finalizer = functools.partial(self._extid_add_finalize, statement) return (token, finalizer) @_prepared_select_statement( ExtIDRow, "WHERE extid_type=? AND extid=? AND extid_version=? " "AND target_type=? AND target=?", ) def extid_get_from_pk( self, extid_type: str, extid: bytes, extid_version: int, target: CoreSWHID, *, statement, ) -> Optional[ExtIDRow]: rows = list( self._execute_with_retries( statement, [ extid_type, extid, extid_version, target.object_type.value, target.object_id, ], ), ) assert len(rows) <= 1 if rows: return ExtIDRow(**rows[0]) else: return None @_prepared_select_statement( ExtIDRow, "WHERE token(extid_type, extid) = ?", ) def extid_get_from_token(self, token: int, *, statement) -> Iterable[ExtIDRow]: return map(ExtIDRow.from_dict, self._execute_with_retries(statement, [token]),) + # Rows are partitioned by token(extid_type, extid), then ordered (aka. "clustered") + # by (extid_type, extid, extid_version, ...). This means that, without knowing the + # exact extid_type and extid, we need to scan the whole partition; which should be + # reasonably small. We can change the schema later if this becomes an issue + @_prepared_select_statement( + ExtIDRow, + "WHERE token(extid_type, extid) = ? AND extid_version = ? ALLOW FILTERING", + ) + def extid_get_from_token_and_extid_version( + self, token: int, extid_version: int, *, statement + ) -> Iterable[ExtIDRow]: + return map( + ExtIDRow.from_dict, + self._execute_with_retries(statement, [token, extid_version]), + ) + @_prepared_select_statement( ExtIDRow, "WHERE extid_type=? AND extid=?", ) def extid_get_from_extid( self, extid_type: str, extid: bytes, *, statement ) -> Iterable[ExtIDRow]: return map( ExtIDRow.from_dict, self._execute_with_retries(statement, [extid_type, extid]), ) + @_prepared_select_statement( + ExtIDRow, "WHERE extid_type=? AND extid=? AND extid_version = ?", + ) + def extid_get_from_extid_and_version( + self, extid_type: str, extid: bytes, extid_version: int, *, statement + ) -> Iterable[ExtIDRow]: + return map( + ExtIDRow.from_dict, + self._execute_with_retries(statement, [extid_type, extid, extid_version]), + ) + def extid_get_from_target( - self, target_type: str, target: bytes + self, + target_type: str, + target: bytes, + extid_type: Optional[str] = None, + extid_version: Optional[int] = None, ) -> Iterable[ExtIDRow]: for token in self._extid_get_tokens_from_target(target_type, target): if token is not None: - for extid in self.extid_get_from_token(token): + if extid_type is not None and extid_version is not None: + extids = self.extid_get_from_token_and_extid_version( + token, extid_version + ) + else: + extids = self.extid_get_from_token(token) + + for extid in extids: # re-check the extid against target (in case of murmur3 collision) if ( extid is not None and extid.target_type == target_type and extid.target == target + and ( + (extid_version is None and extid_type is None) + or ( + ( + extid_version is not None + and extid.extid_version == extid_version + and extid_type is not None + and extid.extid_type == extid_type + ) + ) + ) ): yield extid ########################## # 'extid_by_target' table ########################## @_prepared_insert_statement(ExtIDByTargetRow) def extid_index_add_one(self, row: ExtIDByTargetRow, *, statement) -> None: """Adds a row mapping extid[target_type, target] to the token of the ExtID in the main 'extid' table.""" self._add_one(statement, row) @_prepared_statement( f""" SELECT target_token FROM {ExtIDByTargetRow.TABLE} WHERE target_type = ? AND target = ? """ ) def _extid_get_tokens_from_target( self, target_type: str, target: bytes, *, statement ) -> Iterable[int]: return ( row["target_token"] for row in self._execute_with_retries(statement, [target_type, target]) ) ########################## # Miscellaneous ########################## + def stat_counters(self) -> Iterable[ObjectCountRow]: + raise NotImplementedError( + "stat_counters is not implemented by the Cassandra backend" + ) + @_prepared_statement("SELECT uuid() FROM revision LIMIT 1;") def check_read(self, *, statement): self._execute_with_retries(statement, []) - - @_prepared_select_statement(ObjectCountRow, "WHERE partition_key=0") - def stat_counters(self, *, statement) -> Iterable[ObjectCountRow]: - return map(ObjectCountRow.from_dict, self._execute_with_retries(statement, [])) diff --git a/swh/storage/cassandra/schema.py b/swh/storage/cassandra/schema.py index 383cc04c..d4878fb7 100644 --- a/swh/storage/cassandra/schema.py +++ b/swh/storage/cassandra/schema.py @@ -1,343 +1,335 @@ # Copyright (C) 2019-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import os _use_scylla = bool(os.environ.get("SWH_USE_SCYLLADB", "")) UDF_LANGUAGE = "lua" if _use_scylla else "java" if UDF_LANGUAGE == "java": # For Cassandra CREATE_TABLES_QUERIES = [ """ CREATE OR REPLACE FUNCTION ascii_bins_count_sfunc ( state tuple>, -- (nb_none, map) bin_name ascii ) CALLED ON NULL INPUT RETURNS tuple> LANGUAGE java AS $$ if (bin_name == null) { state.setInt(0, state.getInt(0) + 1); } else { Map counters = state.getMap( 1, String.class, Integer.class); Integer nb = counters.get(bin_name); if (nb == null) { nb = 0; } counters.put(bin_name, nb + 1); state.setMap(1, counters, String.class, Integer.class); } return state; $$;""", """ CREATE OR REPLACE AGGREGATE ascii_bins_count ( ascii ) SFUNC ascii_bins_count_sfunc STYPE tuple> INITCOND (0, {}) ;""", ] elif UDF_LANGUAGE == "lua": # For ScyllaDB # TODO: this is not implementable yet, because ScyllaDB does not support # user-defined aggregates. https://github.com/scylladb/scylla/issues/7201 CREATE_TABLES_QUERIES = [] else: assert False, f"{UDF_LANGUAGE} must be 'lua' or 'java'" CREATE_TABLES_QUERIES = [ *CREATE_TABLES_QUERIES, """ CREATE TYPE IF NOT EXISTS microtimestamp ( seconds bigint, microseconds int );""", """ CREATE TYPE IF NOT EXISTS microtimestamp_with_timezone ( timestamp frozen, offset smallint, negative_utc boolean );""", """ CREATE TYPE IF NOT EXISTS person ( fullname blob, name blob, email blob );""", """ CREATE TABLE IF NOT EXISTS content ( sha1 blob, sha1_git blob, sha256 blob, blake2s256 blob, length bigint, ctime timestamp, -- creation time, i.e. time of (first) injection into the storage status ascii, PRIMARY KEY ((sha256), sha1, sha1_git, blake2s256) );""", """ CREATE TABLE IF NOT EXISTS skipped_content ( sha1 blob, sha1_git blob, sha256 blob, blake2s256 blob, length bigint, ctime timestamp, -- creation time, i.e. time of (first) injection into the storage status ascii, reason text, origin text, PRIMARY KEY ((sha1, sha1_git, sha256, blake2s256)) );""", """ CREATE TABLE IF NOT EXISTS revision ( id blob PRIMARY KEY, date microtimestamp_with_timezone, committer_date microtimestamp_with_timezone, type ascii, directory blob, -- source code "root" directory message blob, author person, committer person, synthetic boolean, -- true iff revision has been created by Software Heritage metadata text, -- extra metadata as JSON(tarball checksums, etc...) extra_headers frozen> > -- extra commit information as (tuple(key, value), ...) );""", """ CREATE TABLE IF NOT EXISTS revision_parent ( id blob, parent_rank int, -- parent position in merge commits, 0-based parent_id blob, PRIMARY KEY ((id), parent_rank) );""", """ CREATE TABLE IF NOT EXISTS release ( id blob PRIMARY KEY, target_type ascii, target blob, date microtimestamp_with_timezone, name blob, message blob, author person, synthetic boolean, -- true iff release has been created by Software Heritage );""", """ CREATE TABLE IF NOT EXISTS directory ( id blob PRIMARY KEY, );""", """ CREATE TABLE IF NOT EXISTS directory_entry ( directory_id blob, name blob, -- path name, relative to containing dir target blob, perms int, -- unix-like permissions type ascii, -- target type PRIMARY KEY ((directory_id), name) );""", """ CREATE TABLE IF NOT EXISTS snapshot ( id blob PRIMARY KEY, );""", """ -- For a given snapshot_id, branches are sorted by their name, -- allowing easy pagination. CREATE TABLE IF NOT EXISTS snapshot_branch ( snapshot_id blob, name blob, target_type ascii, target blob, PRIMARY KEY ((snapshot_id), name) );""", """ CREATE TABLE IF NOT EXISTS origin_visit ( origin text, visit bigint, date timestamp, type text, PRIMARY KEY ((origin), visit) );""", """ CREATE TABLE IF NOT EXISTS origin_visit_status ( origin text, visit bigint, date timestamp, type text, status ascii, metadata text, snapshot blob, PRIMARY KEY ((origin), visit, date) ) WITH CLUSTERING ORDER BY (visit DESC, date DESC) ;""", # 'WITH CLUSTERING ORDER BY' is optional with Cassandra 4, but ScyllaDB needs it """ CREATE TABLE IF NOT EXISTS origin ( sha1 blob PRIMARY KEY, url text, next_visit_id int, -- We need integer visit ids for compatibility with the pgsql -- storage, so we're using lightweight transactions with this trick: -- https://stackoverflow.com/a/29391877/539465 );""", """ CREATE TABLE IF NOT EXISTS metadata_authority ( url text, type ascii, PRIMARY KEY ((url), type) );""", """ CREATE TABLE IF NOT EXISTS metadata_fetcher ( name ascii, version ascii, PRIMARY KEY ((name), version) );""", """ CREATE TABLE IF NOT EXISTS raw_extrinsic_metadata ( id blob, type text, target text, -- metadata source authority_type text, authority_url text, discovery_date timestamp, fetcher_name ascii, fetcher_version ascii, -- metadata itself format ascii, metadata blob, -- context origin text, visit bigint, snapshot text, release text, revision text, path blob, directory text, PRIMARY KEY ((target), authority_type, authority_url, discovery_date, id) -- An explanation is in order for this primary key: -- -- Intuitively, the primary key should only be 'id', because two metadata -- entries are the same iff the id is the same; and 'id' is used for -- deduplication. -- -- However, we also want to query by -- (target, authority_type, authority_url, discovery_date) -- The naive solution to this would be an extra table, to use as index; -- but it means 1. extra code to keep them in sync 2. overhead when writing -- 3. overhead + random reads (instead of linear) when reading. -- -- Therefore, we use a single table for both, by adding the column -- we want to query with before the id. -- It solves both a) the query/order issues and b) the uniqueness issue because: -- -- a) adding the id at the end of the primary key does not change the rows' order: -- for two different rows, id1 != id2, so -- (target1, ..., date1) < (target2, ..., date2) -- <=> (target1, ..., date1, id1) < (target2, ..., date2, id2) -- -- b) the id is a hash of all the columns, so: -- rows are the same -- <=> id1 == id2 -- <=> (target1, ..., date1, id1) == (target2, ..., date2, id2) );""", """ CREATE TABLE IF NOT EXISTS raw_extrinsic_metadata_by_id ( id blob, target text, authority_type text, authority_url text, PRIMARY KEY ((id)) );""", """ -CREATE TABLE IF NOT EXISTS object_count ( - partition_key smallint, -- Constant, must always be 0 - object_type ascii, - count counter, - PRIMARY KEY ((partition_key), object_type) -);""", - """ CREATE TABLE IF NOT EXISTS extid ( extid_type ascii, extid blob, extid_version smallint, target_type ascii, target blob, PRIMARY KEY ((extid_type, extid), extid_version, target_type, target) );""", """ CREATE TABLE IF NOT EXISTS extid_by_target ( target_type ascii, target blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY ((target_type, target), target_token) );""", ] CONTENT_INDEX_TEMPLATE = """ -- Secondary table, used for looking up "content" from a single hash CREATE TABLE IF NOT EXISTS content_by_{main_algo} ( {main_algo} blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY (({main_algo}), target_token) ); CREATE TABLE IF NOT EXISTS skipped_content_by_{main_algo} ( {main_algo} blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY (({main_algo}), target_token) ); """ TABLES = [ "skipped_content", "content", "revision", "revision_parent", "release", "directory", "directory_entry", "snapshot", "snapshot_branch", "origin_visit", "origin", "raw_extrinsic_metadata", - "object_count", "origin_visit_status", "metadata_authority", "metadata_fetcher", "extid", "extid_by_target", ] HASH_ALGORITHMS = ["sha1", "sha1_git", "sha256", "blake2s256"] for main_algo in HASH_ALGORITHMS: CREATE_TABLES_QUERIES.extend( CONTENT_INDEX_TEMPLATE.format( main_algo=main_algo, other_algos=", ".join( [algo for algo in HASH_ALGORITHMS if algo != main_algo] ), ).split("\n\n") ) TABLES.append("content_by_%s" % main_algo) TABLES.append("skipped_content_by_%s" % main_algo) diff --git a/swh/storage/cassandra/storage.py b/swh/storage/cassandra/storage.py index 30e1ece8..6dbcd6fa 100644 --- a/swh/storage/cassandra/storage.py +++ b/swh/storage/cassandra/storage.py @@ -1,1556 +1,1707 @@ -# Copyright (C) 2019-2020 The Software Heritage developers +# Copyright (C) 2019-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 import datetime import itertools import operator import random import re from typing import ( Any, Callable, Counter, Dict, Iterable, List, Optional, Sequence, Set, Tuple, Union, ) import attr from swh.core.api.classes import stream_results from swh.core.api.serializers import msgpack_dumps, msgpack_loads from swh.model.hashutil import DEFAULT_ALGORITHMS, hash_to_hex from swh.model.identifiers import CoreSWHID, ExtendedObjectType, ExtendedSWHID from swh.model.identifiers import ObjectType as SwhidObjectType from swh.model.model import ( Content, Directory, DirectoryEntry, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Release, Revision, Sha1Git, SkippedContent, Snapshot, SnapshotBranch, TargetType, ) from swh.storage.interface import ( VISIT_STATUSES, ListOrder, PagedResult, PartialBranches, Sha1, ) +from swh.storage.metrics import process_metrics, send_metric, timed from swh.storage.objstorage import ObjStorage from swh.storage.utils import map_optional, now from swh.storage.writer import JournalWriter from . import converters from ..exc import HashCollision, StorageArgumentException from ..utils import remove_keys from .common import TOKEN_BEGIN, TOKEN_END, hash_url from .cql import CqlRunner from .model import ( ContentRow, DirectoryEntryRow, DirectoryRow, ExtIDByTargetRow, ExtIDRow, MetadataAuthorityRow, MetadataFetcherRow, OriginRow, OriginVisitRow, OriginVisitStatusRow, RawExtrinsicMetadataByIdRow, RawExtrinsicMetadataRow, RevisionParentRow, SkippedContentRow, SnapshotBranchRow, SnapshotRow, ) from .schema import HASH_ALGORITHMS # Max block size of contents to return BULK_BLOCK_CONTENT_LEN_MAX = 10000 +DIRECTORY_ENTRIES_INSERT_ALGOS = ["one-by-one", "concurrent", "batch"] + class CassandraStorage: def __init__( self, hosts, keyspace, objstorage, port=9042, journal_writer=None, allow_overwrite=False, consistency_level="ONE", + directory_entries_insert_algo="one-by-one", ): """ A backend of swh-storage backed by Cassandra Args: hosts: Seed Cassandra nodes, to start connecting to the cluster keyspace: Name of the Cassandra database to use objstorage: Passed as argument to :class:`ObjStorage` port: Cassandra port journal_writer: Passed as argument to :class:`JournalWriter` allow_overwrite: Whether ``*_add`` functions will check if an object already exists in the database before sending it in an INSERT. ``False`` is the default as it is more efficient when there is a moderately high probability the object is already known, but ``True`` can be useful to overwrite existing objects (eg. when applying a schema update), or when the database is known to be mostly empty. Note that a ``False`` value does not guarantee there won't be any overwrite. consistency_level: The default read/write consistency to use + directory_entries_insert_algo: Must be one of: + * one-by-one: naive, one INSERT per directory entry, serialized + * concurrent: one INSERT per directory entry, concurrent + * batch: using UNLOGGED BATCH to insert many entries in a few statements """ self._hosts = hosts self._keyspace = keyspace self._port = port self._consistency_level = consistency_level self._set_cql_runner() self.journal_writer: JournalWriter = JournalWriter(journal_writer) self.objstorage: ObjStorage = ObjStorage(objstorage) self._allow_overwrite = allow_overwrite + if directory_entries_insert_algo not in DIRECTORY_ENTRIES_INSERT_ALGOS: + raise ValueError( + f"directory_entries_insert_algo must be one of: " + f"{', '.join(DIRECTORY_ENTRIES_INSERT_ALGOS)}" + ) + self._directory_entries_insert_algo = directory_entries_insert_algo + def _set_cql_runner(self): """Used by tests when they need to reset the CqlRunner""" self._cql_runner: CqlRunner = CqlRunner( self._hosts, self._keyspace, self._port, self._consistency_level ) + @timed def check_config(self, *, check_write: bool) -> bool: self._cql_runner.check_read() return True - def _content_get_from_hash(self, algo, hash_) -> Iterable: + def _content_get_from_hashes(self, algo, hashes: List[bytes]) -> Iterable: """From the name of a hash algorithm and a value of that hash, looks up the "hash -> token" secondary table (content_by_{algo}) to get tokens. Then, looks up the main table (content) to get all contents with that token, and filters out contents whose hash doesn't match.""" - found_tokens = self._cql_runner.content_get_tokens_from_single_hash(algo, hash_) + found_tokens = self._cql_runner.content_get_tokens_from_single_algo( + algo, hashes + ) for token in found_tokens: assert isinstance(token, int), found_tokens # Query the main table ('content'). res = self._cql_runner.content_get_from_token(token) for row in res: # re-check the the hash (in case of murmur3 collision) - if getattr(row, algo) == hash_: + if getattr(row, algo) in hashes: yield row def _content_add(self, contents: List[Content], with_data: bool) -> Dict[str, int]: # Filter-out content already in the database. if not self._allow_overwrite: contents = [ c for c in contents if not self._cql_runner.content_get_from_pk(c.to_dict()) ] if with_data: # First insert to the objstorage, if the endpoint is # `content_add` (as opposed to `content_add_metadata`). # Must add to the objstorage before the DB and journal. Otherwise: # 1. in case of a crash the DB may "believe" we have the content, but # we didn't have time to write to the objstorage before the crash # 2. the objstorage mirroring, which reads from the journal, may attempt to # read from the objstorage before we finished writing it summary = self.objstorage.content_add( c for c in contents if c.status != "absent" ) content_add_bytes = summary["content:add:bytes"] self.journal_writer.content_add(contents) content_add = 0 for content in contents: content_add += 1 # Check for sha1 or sha1_git collisions. This test is not atomic # with the insertion, so it won't detect a collision if both # contents are inserted at the same time, but it's good enough. # # The proper way to do it would probably be a BATCH, but this # would be inefficient because of the number of partitions we # need to affect (len(HASH_ALGORITHMS)+1, which is currently 5) if not self._allow_overwrite: for algo in {"sha1", "sha1_git"}: collisions = [] # Get tokens of 'content' rows with the same value for # sha1/sha1_git - rows = self._content_get_from_hash(algo, content.get_hash(algo)) + # TODO: batch these requests, instead of sending them one by one + rows = self._content_get_from_hashes(algo, [content.get_hash(algo)]) for row in rows: if getattr(row, algo) != content.get_hash(algo): # collision of token(partition key), ignore this # row continue for other_algo in HASH_ALGORITHMS: if getattr(row, other_algo) != content.get_hash(other_algo): # This hash didn't match; discard the row. collisions.append( {k: getattr(row, k) for k in HASH_ALGORITHMS} ) if collisions: collisions.append(content.hashes()) raise HashCollision(algo, content.get_hash(algo), collisions) (token, insertion_finalizer) = self._cql_runner.content_add_prepare( ContentRow(**remove_keys(content.to_dict(), ("data",))) ) # Then add to index tables for algo in HASH_ALGORITHMS: self._cql_runner.content_index_add_one(algo, content, token) # Then to the main table insertion_finalizer() summary = { "content:add": content_add, } if with_data: summary["content:add:bytes"] = content_add_bytes return summary + @timed + @process_metrics def content_add(self, content: List[Content]) -> Dict[str, int]: to_add = { (c.sha1, c.sha1_git, c.sha256, c.blake2s256): c for c in content }.values() contents = [attr.evolve(c, ctime=now()) for c in to_add] return self._content_add(list(contents), with_data=True) + @timed def content_update( self, contents: List[Dict[str, Any]], keys: List[str] = [] ) -> None: raise NotImplementedError( "content_update is not supported by the Cassandra backend" ) + @timed + @process_metrics def content_add_metadata(self, content: List[Content]) -> Dict[str, int]: return self._content_add(content, with_data=False) + @timed def content_get_data(self, content: Sha1) -> Optional[bytes]: # FIXME: Make this method support slicing the `data` return self.objstorage.content_get(content) + @timed def content_get_partition( self, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Content]: if limit is None: raise StorageArgumentException("limit should not be None") # Compute start and end of the range of tokens covered by the # requested partition partition_size = (TOKEN_END - TOKEN_BEGIN) // nb_partitions range_start = TOKEN_BEGIN + partition_id * partition_size range_end = TOKEN_BEGIN + (partition_id + 1) * partition_size # offset the range start according to the `page_token`. if page_token is not None: if not (range_start <= int(page_token) <= range_end): raise StorageArgumentException("Invalid page_token.") range_start = int(page_token) next_page_token: Optional[str] = None rows = self._cql_runner.content_get_token_range( range_start, range_end, limit + 1 ) contents = [] for counter, (tok, row) in enumerate(rows): if row.status == "absent": continue row_d = row.to_dict() if counter >= limit: next_page_token = str(tok) break row_d.pop("ctime") contents.append(Content(**row_d)) assert len(contents) <= limit return PagedResult(results=contents, next_page_token=next_page_token) + @timed def content_get( self, contents: List[bytes], algo: str = "sha1" ) -> List[Optional[Content]]: if algo not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "algo should be one of {','.join(DEFAULT_ALGORITHMS)}" ) key = operator.attrgetter(algo) contents_by_hash: Dict[Sha1, Optional[Content]] = {} - for hash_ in contents: + for row in self._content_get_from_hashes(algo, contents): # Get all (sha1, sha1_git, sha256, blake2s256) whose sha1/sha1_git # matches the argument, from the index table ('content_by_*') - for row in self._content_get_from_hash(algo, hash_): - row_d = row.to_dict() - row_d.pop("ctime") - content = Content(**row_d) - contents_by_hash[key(content)] = content + row_d = row.to_dict() + row_d.pop("ctime") + content = Content(**row_d) + contents_by_hash[key(content)] = content return [contents_by_hash.get(hash_) for hash_ in contents] + @timed def content_find(self, content: Dict[str, Any]) -> List[Content]: + return self._content_find_many([content]) + + def _content_find_many(self, contents: List[Dict[str, Any]]) -> List[Content]: # Find an algorithm that is common to all the requested contents. # It will be used to do an initial filtering efficiently. - filter_algos = list(set(content).intersection(HASH_ALGORITHMS)) + filter_algos = set(HASH_ALGORITHMS) + for content in contents: + filter_algos &= set(content) if not filter_algos: raise StorageArgumentException( "content keys must contain at least one " f"of: {', '.join(sorted(HASH_ALGORITHMS))}" ) - common_algo = filter_algos[0] + common_algo = list(filter_algos)[0] results = [] - rows = self._content_get_from_hash(common_algo, content[common_algo]) + rows = self._content_get_from_hashes( + common_algo, [content[common_algo] for content in contents] + ) for row in rows: # Re-check all the hashes, in case of collisions (either of the # hash of the partition key, or the hashes in it) - for algo in HASH_ALGORITHMS: - if content.get(algo) and getattr(row, algo) != content[algo]: - # This hash didn't match; discard the row. + for content in contents: + for algo in HASH_ALGORITHMS: + if content.get(algo) and getattr(row, algo) != content[algo]: + # This hash didn't match; discard the row. + break + else: + # All hashes match, keep this row. + row_d = row.to_dict() + row_d["ctime"] = row.ctime.replace(tzinfo=datetime.timezone.utc) + results.append(Content(**row_d)) break else: - # All hashes match, keep this row. - row_d = row.to_dict() - row_d["ctime"] = row.ctime.replace(tzinfo=datetime.timezone.utc) - results.append(Content(**row_d)) + # No content matched; skip it + pass return results + @timed def content_missing( self, contents: List[Dict[str, Any]], key_hash: str = "sha1" ) -> Iterable[bytes]: if key_hash not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "key_hash should be one of {','.join(DEFAULT_ALGORITHMS)}" ) contents_with_all_hashes = [] contents_with_missing_hashes = [] for content in contents: if DEFAULT_ALGORITHMS <= set(content): contents_with_all_hashes.append(content) else: contents_with_missing_hashes.append(content) # These contents can be queried efficiently directly in the main table for content in self._cql_runner.content_missing_from_hashes( contents_with_all_hashes ): yield content[key_hash] # For these, we need the expensive index lookups + main table. for content in contents_with_missing_hashes: res = self.content_find(content) if not res: yield content[key_hash] + @timed def content_missing_per_sha1(self, contents: List[bytes]) -> Iterable[bytes]: return self.content_missing([{"sha1": c} for c in contents]) + @timed def content_missing_per_sha1_git( self, contents: List[Sha1Git] ) -> Iterable[Sha1Git]: return self.content_missing( [{"sha1_git": c} for c in contents], key_hash="sha1_git" ) + @timed def content_get_random(self) -> Sha1Git: content = self._cql_runner.content_get_random() assert content, "Could not find any content" return content.sha1_git def _skipped_content_add(self, contents: List[SkippedContent]) -> Dict[str, int]: # Filter-out content already in the database. if not self._allow_overwrite: contents = [ c for c in contents if not self._cql_runner.skipped_content_get_from_pk(c.to_dict()) ] self.journal_writer.skipped_content_add(contents) for content in contents: # Compute token of the row in the main table (token, insertion_finalizer) = self._cql_runner.skipped_content_add_prepare( SkippedContentRow.from_dict({"origin": None, **content.to_dict()}) ) # Then add to index tables for algo in HASH_ALGORITHMS: self._cql_runner.skipped_content_index_add_one(algo, content, token) # Then to the main table insertion_finalizer() return {"skipped_content:add": len(contents)} + @timed + @process_metrics def skipped_content_add(self, content: List[SkippedContent]) -> Dict[str, int]: contents = [attr.evolve(c, ctime=now()) for c in content] return self._skipped_content_add(contents) + @timed def skipped_content_missing( self, contents: List[Dict[str, Any]] ) -> Iterable[Dict[str, Any]]: for content in contents: if not self._cql_runner.skipped_content_get_from_pk(content): yield {algo: content[algo] for algo in DEFAULT_ALGORITHMS} + @timed + @process_metrics def directory_add(self, directories: List[Directory]) -> Dict[str, int]: to_add = {d.id: d for d in directories}.values() if not self._allow_overwrite: # Filter out directories that are already inserted. missing = self.directory_missing([dir_.id for dir_ in to_add]) directories = [dir_ for dir_ in directories if dir_.id in missing] self.journal_writer.directory_add(directories) for directory in directories: # Add directory entries to the 'directory_entry' table - for entry in directory.entries: - self._cql_runner.directory_entry_add_one( - DirectoryEntryRow(directory_id=directory.id, **entry.to_dict()) + rows = [ + DirectoryEntryRow(directory_id=directory.id, **entry.to_dict()) + for entry in directory.entries + ] + if self._directory_entries_insert_algo == "one-by-one": + for row in rows: + self._cql_runner.directory_entry_add_one(row) + elif self._directory_entries_insert_algo == "concurrent": + self._cql_runner.directory_entry_add_concurrent(rows) + elif self._directory_entries_insert_algo == "batch": + self._cql_runner.directory_entry_add_batch(rows) + else: + raise ValueError( + f"Unexpected value for directory_entries_insert_algo: " + f"{self._directory_entries_insert_algo}" ) # Add the directory *after* adding all the entries, so someone # calling snapshot_get_branch in the meantime won't end up # with half the entries. self._cql_runner.directory_add_one(DirectoryRow(id=directory.id)) return {"directory:add": len(directories)} + @timed def directory_missing(self, directories: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.directory_missing(directories) - def _join_dentry_to_content(self, dentry: DirectoryEntry) -> Dict[str, Any]: - contents: Union[List[Content], List[SkippedContentRow]] + def _join_dentry_to_content( + self, dentry: DirectoryEntry, contents: List[Content] + ) -> Dict[str, Any]: + content: Union[None, Content, SkippedContentRow] keys = ( "status", "sha1", "sha1_git", "sha256", "length", ) ret = dict.fromkeys(keys) ret.update(dentry.to_dict()) if ret["type"] == "file": - contents = self.content_find({"sha1_git": ret["target"]}) - if not contents: + for content in contents: + if dentry.target == content.sha1_git: + break + else: tokens = list( self._cql_runner.skipped_content_get_tokens_from_single_hash( "sha1_git", ret["target"] ) ) if tokens: - contents = list( + content = list( self._cql_runner.skipped_content_get_from_token(tokens[0]) - ) - if contents: - content = contents[0] + )[0] + else: + content = None + if content: for key in keys: ret[key] = getattr(content, key) return ret def _directory_ls( self, directory_id: Sha1Git, recursive: bool, prefix: bytes = b"" ) -> Iterable[Dict[str, Any]]: if self.directory_missing([directory_id]): return rows = list(self._cql_runner.directory_entry_get([directory_id])) + # TODO: dedup to be fast in case the directory contains the same subdir/file + # multiple times + contents = self._content_find_many([{"sha1_git": row.target} for row in rows]) + for row in rows: entry_d = row.to_dict() # Build and yield the directory entry dict del entry_d["directory_id"] entry = DirectoryEntry.from_dict(entry_d) - ret = self._join_dentry_to_content(entry) + ret = self._join_dentry_to_content(entry, contents) ret["name"] = prefix + ret["name"] ret["dir_id"] = directory_id yield ret if recursive and ret["type"] == "dir": yield from self._directory_ls( ret["target"], True, prefix + ret["name"] + b"/" ) + @timed def directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes] ) -> Optional[Dict[str, Any]]: return self._directory_entry_get_by_path(directory, paths, b"") def _directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes], prefix: bytes ) -> Optional[Dict[str, Any]]: if not paths: return None contents = list(self.directory_ls(directory)) if not contents: return None def _get_entry(entries, name): """Finds the entry with the requested name, prepends the prefix (to get its full path), and returns it. If no entry has that name, returns None.""" for entry in entries: if entry["name"] == name: entry = entry.copy() entry["name"] = prefix + entry["name"] return entry first_item = _get_entry(contents, paths[0]) if len(paths) == 1: return first_item if not first_item or first_item["type"] != "dir": return None return self._directory_entry_get_by_path( first_item["target"], paths[1:], prefix + paths[0] + b"/" ) + @timed def directory_ls( self, directory: Sha1Git, recursive: bool = False ) -> Iterable[Dict[str, Any]]: yield from self._directory_ls(directory, recursive) + @timed def directory_get_entries( self, directory_id: Sha1Git, page_token: Optional[bytes] = None, limit: int = 1000, ) -> Optional[PagedResult[DirectoryEntry]]: if self.directory_missing([directory_id]): return None entries_from: bytes = page_token or b"" rows = self._cql_runner.directory_entry_get_from_name( directory_id, entries_from, limit + 1 ) entries = [ DirectoryEntry.from_dict(remove_keys(row.to_dict(), ("directory_id",))) for row in rows ] if len(entries) > limit: last_entry = entries.pop() next_page_token = last_entry.name else: next_page_token = None return PagedResult(results=entries, next_page_token=next_page_token) + @timed def directory_get_random(self) -> Sha1Git: directory = self._cql_runner.directory_get_random() assert directory, "Could not find any directory" return directory.id + @timed + @process_metrics def revision_add(self, revisions: List[Revision]) -> Dict[str, int]: # Filter-out revisions already in the database if not self._allow_overwrite: to_add = {r.id: r for r in revisions}.values() missing = self.revision_missing([rev.id for rev in to_add]) revisions = [rev for rev in revisions if rev.id in missing] self.journal_writer.revision_add(revisions) for revision in revisions: revobject = converters.revision_to_db(revision) if revobject: # Add parents first for (rank, parent) in enumerate(revision.parents): self._cql_runner.revision_parent_add_one( RevisionParentRow( id=revobject.id, parent_rank=rank, parent_id=parent ) ) # Then write the main revision row. # Writing this after all parents were written ensures that # read endpoints don't return a partial view while writing # the parents self._cql_runner.revision_add_one(revobject) return {"revision:add": len(revisions)} + @timed def revision_missing(self, revisions: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.revision_missing(revisions) + @timed def revision_get(self, revision_ids: List[Sha1Git]) -> List[Optional[Revision]]: rows = self._cql_runner.revision_get(revision_ids) revisions: Dict[Sha1Git, Revision] = {} for row in rows: # TODO: use a single query to get all parents? # (it might have lower latency, but requires more code and more # bandwidth, because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(row.id)) # parent_rank is the clustering key, so results are already # sorted by rank. rev = converters.revision_from_db(row, parents=parents) revisions[rev.id] = rev return [revisions.get(rev_id) for rev_id in revision_ids] def _get_parent_revs( self, rev_ids: Iterable[Sha1Git], seen: Set[Sha1Git], limit: Optional[int], short: bool, ) -> Union[ Iterable[Dict[str, Any]], Iterable[Tuple[Sha1Git, Tuple[Sha1Git, ...]]], ]: if limit and len(seen) >= limit: return rev_ids = [id_ for id_ in rev_ids if id_ not in seen] if not rev_ids: return seen |= set(rev_ids) # We need this query, even if short=True, to return consistent # results (ie. not return only a subset of a revision's parents # if it is being written) if short: ids = self._cql_runner.revision_get_ids(rev_ids) for id_ in ids: # TODO: use a single query to get all parents? # (it might have less latency, but requires less code and more # bandwidth (because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(id_)) # parent_rank is the clustering key, so results are already # sorted by rank. yield (id_, parents) yield from self._get_parent_revs(parents, seen, limit, short) else: rows = self._cql_runner.revision_get(rev_ids) for row in rows: # TODO: use a single query to get all parents? # (it might have less latency, but requires less code and more # bandwidth (because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(row.id)) # parent_rank is the clustering key, so results are already # sorted by rank. rev = converters.revision_from_db(row, parents=parents) yield rev.to_dict() yield from self._get_parent_revs(parents, seen, limit, short) + @timed def revision_log( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Dict[str, Any]]]: seen: Set[Sha1Git] = set() yield from self._get_parent_revs(revisions, seen, limit, False) + @timed def revision_shortlog( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Tuple[Sha1Git, Tuple[Sha1Git, ...]]]]: seen: Set[Sha1Git] = set() yield from self._get_parent_revs(revisions, seen, limit, True) + @timed def revision_get_random(self) -> Sha1Git: revision = self._cql_runner.revision_get_random() assert revision, "Could not find any revision" return revision.id + @timed + @process_metrics def release_add(self, releases: List[Release]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {r.id: r for r in releases}.values() missing = set(self.release_missing([rel.id for rel in to_add])) releases = [rel for rel in to_add if rel.id in missing] self.journal_writer.release_add(releases) for release in releases: if release: self._cql_runner.release_add_one(converters.release_to_db(release)) return {"release:add": len(releases)} + @timed def release_missing(self, releases: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.release_missing(releases) + @timed def release_get(self, releases: List[Sha1Git]) -> List[Optional[Release]]: rows = self._cql_runner.release_get(releases) rels: Dict[Sha1Git, Release] = {} for row in rows: release = converters.release_from_db(row) rels[row.id] = release return [rels.get(rel_id) for rel_id in releases] + @timed def release_get_random(self) -> Sha1Git: release = self._cql_runner.release_get_random() assert release, "Could not find any release" return release.id + @timed + @process_metrics def snapshot_add(self, snapshots: List[Snapshot]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {s.id: s for s in snapshots}.values() missing = self._cql_runner.snapshot_missing([snp.id for snp in to_add]) snapshots = [snp for snp in snapshots if snp.id in missing] for snapshot in snapshots: self.journal_writer.snapshot_add([snapshot]) # Add branches for (branch_name, branch) in snapshot.branches.items(): if branch is None: target_type: Optional[str] = None target: Optional[bytes] = None else: target_type = branch.target_type.value target = branch.target self._cql_runner.snapshot_branch_add_one( SnapshotBranchRow( snapshot_id=snapshot.id, name=branch_name, target_type=target_type, target=target, ) ) # Add the snapshot *after* adding all the branches, so someone # calling snapshot_get_branch in the meantime won't end up # with half the branches. self._cql_runner.snapshot_add_one(SnapshotRow(id=snapshot.id)) return {"snapshot:add": len(snapshots)} + @timed def snapshot_missing(self, snapshots: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.snapshot_missing(snapshots) + @timed def snapshot_get(self, snapshot_id: Sha1Git) -> Optional[Dict[str, Any]]: d = self.snapshot_get_branches(snapshot_id) if d is None: return None return { "id": d["id"], "branches": { name: branch.to_dict() if branch else None for (name, branch) in d["branches"].items() }, "next_branch": d["next_branch"], } + @timed def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[Dict[Optional[str], int]]: if self._cql_runner.snapshot_missing([snapshot_id]): # Makes sure we don't fetch branches for a snapshot that is # being added. return None return self._cql_runner.snapshot_count_branches( snapshot_id, branch_name_exclude_prefix ) + @timed def snapshot_get_branches( self, snapshot_id: Sha1Git, branches_from: bytes = b"", branches_count: int = 1000, target_types: Optional[List[str]] = None, branch_name_include_substring: Optional[bytes] = None, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[PartialBranches]: if self._cql_runner.snapshot_missing([snapshot_id]): # Makes sure we don't fetch branches for a snapshot that is # being added. return None branches: List = [] while len(branches) < branches_count + 1: new_branches = list( self._cql_runner.snapshot_branch_get( snapshot_id, branches_from, branches_count + 1, branch_name_exclude_prefix, ) ) if not new_branches: break branches_from = new_branches[-1].name new_branches_filtered = new_branches # Filter by target_type if target_types: new_branches_filtered = [ branch for branch in new_branches_filtered if branch.target is not None and branch.target_type in target_types ] # Filter by branches_name_pattern if branch_name_include_substring: new_branches_filtered = [ branch for branch in new_branches_filtered if branch.name is not None and ( branch_name_include_substring is None or branch_name_include_substring in branch.name ) ] branches.extend(new_branches_filtered) if len(new_branches) < branches_count + 1: break if len(branches) > branches_count: last_branch = branches.pop(-1).name else: last_branch = None return PartialBranches( id=snapshot_id, branches={ branch.name: None if branch.target is None else SnapshotBranch( target=branch.target, target_type=TargetType(branch.target_type) ) for branch in branches }, next_branch=last_branch, ) + @timed def snapshot_get_random(self) -> Sha1Git: snapshot = self._cql_runner.snapshot_get_random() assert snapshot, "Could not find any snapshot" return snapshot.id + @timed def object_find_by_sha1_git(self, ids: List[Sha1Git]) -> Dict[Sha1Git, List[Dict]]: results: Dict[Sha1Git, List[Dict]] = {id_: [] for id_ in ids} missing_ids = set(ids) # Mind the order, revision is the most likely one for a given ID, # so we check revisions first. queries: List[Tuple[str, Callable[[List[Sha1Git]], List[Sha1Git]]]] = [ ("revision", self._cql_runner.revision_missing), ("release", self._cql_runner.release_missing), ("content", self._cql_runner.content_missing_by_sha1_git), ("directory", self._cql_runner.directory_missing), ] for (object_type, query_fn) in queries: found_ids = missing_ids - set(query_fn(list(missing_ids))) for sha1_git in found_ids: results[sha1_git].append( {"sha1_git": sha1_git, "type": object_type,} ) missing_ids.remove(sha1_git) if not missing_ids: # We found everything, skipping the next queries. break return results + @timed def origin_get(self, origins: List[str]) -> Iterable[Optional[Origin]]: return [self.origin_get_one(origin) for origin in origins] + @timed def origin_get_one(self, origin_url: str) -> Optional[Origin]: """Given an origin url, return the origin if it exists, None otherwise """ rows = list(self._cql_runner.origin_get_by_url(origin_url)) if rows: assert len(rows) == 1 return Origin(url=rows[0].url) else: return None + @timed def origin_get_by_sha1(self, sha1s: List[bytes]) -> List[Optional[Dict[str, Any]]]: results = [] for sha1 in sha1s: rows = list(self._cql_runner.origin_get_by_sha1(sha1)) origin = {"url": rows[0].url} if rows else None results.append(origin) return results + @timed def origin_list( self, page_token: Optional[str] = None, limit: int = 100 ) -> PagedResult[Origin]: # Compute what token to begin the listing from start_token = TOKEN_BEGIN if page_token: start_token = int(page_token) if not (TOKEN_BEGIN <= start_token <= TOKEN_END): raise StorageArgumentException("Invalid page_token.") next_page_token = None origins = [] # Take one more origin so we can reuse it as the next page token if any for (tok, row) in self._cql_runner.origin_list(start_token, limit + 1): origins.append(Origin(url=row.url)) # keep reference of the last id for pagination purposes last_id = tok if len(origins) > limit: # last origin id is the next page token next_page_token = str(last_id) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) + @timed def origin_search( self, url_pattern: str, page_token: Optional[str] = None, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, ) -> PagedResult[Origin]: # TODO: remove this endpoint, swh-search should be used instead. next_page_token = None offset = int(page_token) if page_token else 0 origin_rows = [row for row in self._cql_runner.origin_iter_all()] if regexp: pat = re.compile(url_pattern) origin_rows = [row for row in origin_rows if pat.search(row.url)] else: origin_rows = [row for row in origin_rows if url_pattern in row.url] if with_visit: origin_rows = [row for row in origin_rows if row.next_visit_id > 1] if visit_types: def _has_visit_types(origin, visit_types): for origin_visit in stream_results(self.origin_visit_get, origin): if origin_visit.type in visit_types: return True return False origin_rows = [ row for row in origin_rows if _has_visit_types(row.url, visit_types) ] origins = [Origin(url=row.url) for row in origin_rows] origins = origins[offset : offset + limit + 1] if len(origins) > limit: # next offset next_page_token = str(offset + limit) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) + @timed def origin_count( self, url_pattern: str, regexp: bool = False, with_visit: bool = False ) -> int: raise NotImplementedError( "The Cassandra backend does not implement origin_count" ) + @timed + @process_metrics def origin_add(self, origins: List[Origin]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {o.url: o for o in origins}.values() origins = [ori for ori in to_add if self.origin_get_one(ori.url) is None] self.journal_writer.origin_add(origins) for origin in origins: self._cql_runner.origin_add_one( OriginRow(sha1=hash_url(origin.url), url=origin.url, next_visit_id=1) ) return {"origin:add": len(origins)} + @timed def origin_visit_add(self, visits: List[OriginVisit]) -> Iterable[OriginVisit]: for visit in visits: origin = self.origin_get_one(visit.origin) if not origin: # Cannot add a visit without an origin raise StorageArgumentException("Unknown origin %s", visit.origin) all_visits = [] nb_visits = 0 for visit in visits: nb_visits += 1 if visit.visit: # Set origin.next_visit_id = max(origin.next_visit_id, visit.visit+1) # so the next loader run does not reuse the id. self._cql_runner.origin_bump_next_visit_id(visit.origin, visit.visit) else: visit_id = self._cql_runner.origin_generate_unique_visit_id( visit.origin ) visit = attr.evolve(visit, visit=visit_id) self.journal_writer.origin_visit_add([visit]) self._cql_runner.origin_visit_add_one(OriginVisitRow(**visit.to_dict())) assert visit.visit is not None all_visits.append(visit) self._origin_visit_status_add( OriginVisitStatus( origin=visit.origin, visit=visit.visit, date=visit.date, type=visit.type, status="created", snapshot=None, ) ) - + send_metric("origin_visit:add", count=nb_visits, method_name="origin_visit") return all_visits def _origin_visit_status_add(self, visit_status: OriginVisitStatus) -> None: """Add an origin visit status""" if visit_status.type is None: visit_row = self._cql_runner.origin_visit_get_one( visit_status.origin, visit_status.visit ) if visit_row is None: raise StorageArgumentException( f"Unknown origin visit {visit_status.visit} " f"of origin {visit_status.origin}" ) visit_status = attr.evolve(visit_status, type=visit_row.type) self.journal_writer.origin_visit_status_add([visit_status]) self._cql_runner.origin_visit_status_add_one( converters.visit_status_to_row(visit_status) ) + @timed + @process_metrics def origin_visit_status_add( self, visit_statuses: List[OriginVisitStatus] ) -> Dict[str, int]: # First round to check existence (fail early if any is ko) for visit_status in visit_statuses: origin_url = self.origin_get_one(visit_status.origin) if not origin_url: raise StorageArgumentException(f"Unknown origin {visit_status.origin}") for visit_status in visit_statuses: self._origin_visit_status_add(visit_status) return {"origin_visit_status:add": len(visit_statuses)} def _origin_visit_apply_status( self, visit: Dict[str, Any], visit_status: OriginVisitStatusRow ) -> Dict[str, Any]: """Retrieve the latest visit status information for the origin visit. Then merge it with the visit and return it. """ return { # default to the values in visit **visit, # override with the last update **visit_status.to_dict(), # visit['origin'] is the URL (via a join), while # visit_status['origin'] is only an id. "origin": visit["origin"], # but keep the date of the creation of the origin visit "date": visit["date"], # We use the visit type from origin visit # if it's not present on the origin visit status "type": visit_status.type or visit["type"], } def _origin_visit_get_latest_status(self, visit: OriginVisit) -> OriginVisitStatus: """Retrieve the latest visit status information for the origin visit object. """ assert visit.visit row = self._cql_runner.origin_visit_status_get_latest(visit.origin, visit.visit) assert row is not None visit_status = converters.row_to_visit_status(row) return attr.evolve(visit_status, origin=visit.origin) @staticmethod def _format_origin_visit_row(visit): return { **visit.to_dict(), "origin": visit.origin, "date": visit.date.replace(tzinfo=datetime.timezone.utc), } + @timed def origin_visit_get( self, origin: str, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisit]: if not isinstance(order, ListOrder): raise StorageArgumentException("order must be a ListOrder value") if page_token and not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") next_page_token = None visit_from = None if page_token is None else int(page_token) visits: List[OriginVisit] = [] extra_limit = limit + 1 rows = self._cql_runner.origin_visit_get(origin, visit_from, extra_limit, order) for row in rows: visits.append(converters.row_to_visit(row)) assert len(visits) <= extra_limit if len(visits) == extra_limit: visits = visits[:limit] next_page_token = str(visits[-1].visit) return PagedResult(results=visits, next_page_token=next_page_token) + @timed def origin_visit_status_get( self, origin: str, visit: int, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisitStatus]: next_page_token = None date_from = None if page_token is not None: date_from = datetime.datetime.fromisoformat(page_token) # Take one more visit status so we can reuse it as the next page token if any rows = self._cql_runner.origin_visit_status_get_range( origin, visit, date_from, limit + 1, order ) visit_statuses = [converters.row_to_visit_status(row) for row in rows] if len(visit_statuses) > limit: # last visit status date is the next page token next_page_token = str(visit_statuses[-1].date) # excluding that visit status from the result to respect the limit size visit_statuses = visit_statuses[:limit] return PagedResult(results=visit_statuses, next_page_token=next_page_token) + @timed def origin_visit_find_by_date( self, origin: str, visit_date: datetime.datetime ) -> Optional[OriginVisit]: # Iterator over all the visits of the origin # This should be ok for now, as there aren't too many visits # per origin. rows = list(self._cql_runner.origin_visit_get_all(origin)) def key(visit): dt = visit.date.replace(tzinfo=datetime.timezone.utc) - visit_date return (abs(dt), -visit.visit) if rows: return converters.row_to_visit(min(rows, key=key)) return None + @timed def origin_visit_get_by(self, origin: str, visit: int) -> Optional[OriginVisit]: row = self._cql_runner.origin_visit_get_one(origin, visit) if row: return converters.row_to_visit(row) return None + @timed def origin_visit_get_latest( self, origin: str, type: Optional[str] = None, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisit]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) # TODO: Do not fetch all visits rows = self._cql_runner.origin_visit_get_all(origin) latest_visit = None for row in rows: visit = self._format_origin_visit_row(row) for status_row in self._cql_runner.origin_visit_status_get( origin, visit["visit"] ): updated_visit = self._origin_visit_apply_status(visit, status_row) if type is not None and updated_visit["type"] != type: continue if allowed_statuses and updated_visit["status"] not in allowed_statuses: continue if require_snapshot and updated_visit["snapshot"] is None: continue # updated_visit is a candidate if latest_visit is not None: if updated_visit["date"] < latest_visit["date"]: continue assert ( updated_visit["visit"] >= latest_visit["visit"] ), "Cassandra returned visits not ordered by increasing visit id." latest_visit = updated_visit if latest_visit is None: return None return OriginVisit( origin=latest_visit["origin"], visit=latest_visit["visit"], date=latest_visit["date"], type=latest_visit["type"], ) + @timed def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisitStatus]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) rows = list(self._cql_runner.origin_visit_status_get(origin_url, visit)) # filtering is done python side as we cannot do it server side if allowed_statuses: rows = [row for row in rows if row.status in allowed_statuses] if require_snapshot: rows = [row for row in rows if row.snapshot is not None] if not rows: return None return converters.row_to_visit_status(rows[0]) + @timed def origin_visit_status_get_random(self, type: str) -> Optional[OriginVisitStatus]: back_in_the_day = now() - datetime.timedelta(weeks=12) # 3 months back # Random position to start iteration at start_token = random.randint(TOKEN_BEGIN, TOKEN_END) # Iterator over all visits, ordered by token(origins) then visit_id rows = self._cql_runner.origin_visit_iter(start_token) for row in rows: visit = converters.row_to_visit(row) visit_status = self._origin_visit_get_latest_status(visit) if visit.date > back_in_the_day and visit_status.status == "full": return visit_status return None + @timed def stat_counters(self): rows = self._cql_runner.stat_counters() keys = ( "content", "directory", "origin", "origin_visit", "release", "revision", "skipped_content", "snapshot", ) stats = {key: 0 for key in keys} stats.update({row.object_type: row.count for row in rows}) return stats + @timed def refresh_stat_counters(self): pass + @timed + @process_metrics def raw_extrinsic_metadata_add( self, metadata: List[RawExtrinsicMetadata] ) -> Dict[str, int]: self.journal_writer.raw_extrinsic_metadata_add(metadata) counter = Counter[ExtendedObjectType]() for metadata_entry in metadata: if not self._cql_runner.metadata_authority_get( metadata_entry.authority.type.value, metadata_entry.authority.url ): raise StorageArgumentException( f"Unknown authority {metadata_entry.authority}" ) if not self._cql_runner.metadata_fetcher_get( metadata_entry.fetcher.name, metadata_entry.fetcher.version ): raise StorageArgumentException( f"Unknown fetcher {metadata_entry.fetcher}" ) try: row = RawExtrinsicMetadataRow( id=metadata_entry.id, type=metadata_entry.target.object_type.name.lower(), target=str(metadata_entry.target), authority_type=metadata_entry.authority.type.value, authority_url=metadata_entry.authority.url, discovery_date=metadata_entry.discovery_date, fetcher_name=metadata_entry.fetcher.name, fetcher_version=metadata_entry.fetcher.version, format=metadata_entry.format, metadata=metadata_entry.metadata, origin=metadata_entry.origin, visit=metadata_entry.visit, snapshot=map_optional(str, metadata_entry.snapshot), release=map_optional(str, metadata_entry.release), revision=map_optional(str, metadata_entry.revision), path=metadata_entry.path, directory=map_optional(str, metadata_entry.directory), ) except TypeError as e: raise StorageArgumentException(*e.args) # Add to the index first self._cql_runner.raw_extrinsic_metadata_by_id_add( RawExtrinsicMetadataByIdRow( id=row.id, target=row.target, authority_type=row.authority_type, authority_url=row.authority_url, ) ) # Then to the main table self._cql_runner.raw_extrinsic_metadata_add(row) counter[metadata_entry.target.object_type] += 1 return { f"{type.value}_metadata:add": count for (type, count) in counter.items() } + @timed def raw_extrinsic_metadata_get( self, target: ExtendedSWHID, authority: MetadataAuthority, after: Optional[datetime.datetime] = None, page_token: Optional[bytes] = None, limit: int = 1000, ) -> PagedResult[RawExtrinsicMetadata]: if page_token is not None: (after_date, id_) = msgpack_loads(base64.b64decode(page_token)) if after and after_date < after: raise StorageArgumentException( "page_token is inconsistent with the value of 'after'." ) entries = self._cql_runner.raw_extrinsic_metadata_get_after_date_and_id( str(target), authority.type.value, authority.url, after_date, id_, ) elif after is not None: entries = self._cql_runner.raw_extrinsic_metadata_get_after_date( str(target), authority.type.value, authority.url, after ) else: entries = self._cql_runner.raw_extrinsic_metadata_get( str(target), authority.type.value, authority.url ) if limit: entries = itertools.islice(entries, 0, limit + 1) results = [] for entry in entries: assert str(target) == entry.target results.append(converters.row_to_raw_extrinsic_metadata(entry)) if len(results) > limit: results.pop() assert len(results) == limit last_result = results[-1] next_page_token: Optional[str] = base64.b64encode( msgpack_dumps((last_result.discovery_date, last_result.id,)) ).decode() else: next_page_token = None return PagedResult(next_page_token=next_page_token, results=results,) + @timed def raw_extrinsic_metadata_get_by_ids( self, ids: List[Sha1Git] ) -> List[RawExtrinsicMetadata]: keys = self._cql_runner.raw_extrinsic_metadata_get_by_ids(ids) results: Set[RawExtrinsicMetadata] = set() for key in keys: candidates = self._cql_runner.raw_extrinsic_metadata_get( key.target, key.authority_type, key.authority_url ) candidates = [ candidate for candidate in candidates if candidate.id == key.id ] if len(candidates) > 1: raise Exception( "Found multiple RawExtrinsicMetadata objects with the same id: " + hash_to_hex(key.id) ) results.update(map(converters.row_to_raw_extrinsic_metadata, candidates)) return list(results) + @timed def raw_extrinsic_metadata_get_authorities( self, target: ExtendedSWHID ) -> List[MetadataAuthority]: return [ MetadataAuthority( type=MetadataAuthorityType(authority_type), url=authority_url ) for (authority_type, authority_url) in set( self._cql_runner.raw_extrinsic_metadata_get_authorities(str(target)) ) ] + @timed + @process_metrics def metadata_fetcher_add(self, fetchers: List[MetadataFetcher]) -> Dict[str, int]: self.journal_writer.metadata_fetcher_add(fetchers) for fetcher in fetchers: self._cql_runner.metadata_fetcher_add( MetadataFetcherRow(name=fetcher.name, version=fetcher.version,) ) return {"metadata_fetcher:add": len(fetchers)} + @timed def metadata_fetcher_get( self, name: str, version: str ) -> Optional[MetadataFetcher]: fetcher = self._cql_runner.metadata_fetcher_get(name, version) if fetcher: return MetadataFetcher(name=fetcher.name, version=fetcher.version,) else: return None + @timed + @process_metrics def metadata_authority_add( self, authorities: List[MetadataAuthority] ) -> Dict[str, int]: self.journal_writer.metadata_authority_add(authorities) for authority in authorities: self._cql_runner.metadata_authority_add( MetadataAuthorityRow(url=authority.url, type=authority.type.value,) ) return {"metadata_authority:add": len(authorities)} + @timed def metadata_authority_get( self, type: MetadataAuthorityType, url: str ) -> Optional[MetadataAuthority]: authority = self._cql_runner.metadata_authority_get(type.value, url) if authority: return MetadataAuthority( type=MetadataAuthorityType(authority.type), url=authority.url, ) else: return None # ExtID tables + @timed + @process_metrics def extid_add(self, ids: List[ExtID]) -> Dict[str, int]: if not self._allow_overwrite: extids = [ extid for extid in ids if not self._cql_runner.extid_get_from_pk( extid_type=extid.extid_type, extid_version=extid.extid_version, extid=extid.extid, target=extid.target, ) ] else: extids = list(ids) self.journal_writer.extid_add(extids) inserted = 0 for extid in extids: target_type = extid.target.object_type.value target = extid.target.object_id + extid_version = extid.extid_version + extid_type = extid.extid_type extidrow = ExtIDRow( - extid_type=extid.extid_type, - extid_version=extid.extid_version, + extid_type=extid_type, + extid_version=extid_version, extid=extid.extid, target_type=target_type, target=target, ) (token, insertion_finalizer) = self._cql_runner.extid_add_prepare(extidrow) indexrow = ExtIDByTargetRow( target_type=target_type, target=target, target_token=token, ) self._cql_runner.extid_index_add_one(indexrow) insertion_finalizer() inserted += 1 return {"extid:add": inserted} - def extid_get_from_extid(self, id_type: str, ids: List[bytes]) -> List[ExtID]: + @timed + def extid_get_from_extid( + self, id_type: str, ids: List[bytes], version: Optional[int] = None + ) -> List[ExtID]: result: List[ExtID] = [] for extid in ids: - extidrows = list(self._cql_runner.extid_get_from_extid(id_type, extid)) + if version is not None: + extidrows = self._cql_runner.extid_get_from_extid_and_version( + id_type, extid, version + ) + else: + extidrows = self._cql_runner.extid_get_from_extid(id_type, extid) result.extend( ExtID( extid_type=extidrow.extid_type, extid_version=extidrow.extid_version, extid=extidrow.extid, target=CoreSWHID( object_type=extidrow.target_type, object_id=extidrow.target, ), ) for extidrow in extidrows ) return result + @timed def extid_get_from_target( - self, target_type: SwhidObjectType, ids: List[Sha1Git] + self, + target_type: SwhidObjectType, + ids: List[Sha1Git], + extid_type: Optional[str] = None, + extid_version: Optional[int] = None, ) -> List[ExtID]: + if (extid_version is not None and extid_type is None) or ( + extid_version is None and extid_type is not None + ): + raise ValueError("You must provide both extid_type and extid_version") + result: List[ExtID] = [] for target in ids: - extidrows = list( - self._cql_runner.extid_get_from_target(target_type.value, target) + extidrows = self._cql_runner.extid_get_from_target( + target_type.value, + target, + extid_type=extid_type, + extid_version=extid_version, ) + result.extend( ExtID( extid_type=extidrow.extid_type, extid_version=extidrow.extid_version, extid=extidrow.extid, target=CoreSWHID( object_type=SwhidObjectType(extidrow.target_type), object_id=extidrow.target, ), ) for extidrow in extidrows ) return result # Misc def clear_buffers(self, object_types: Sequence[str] = ()) -> None: """Do nothing """ return None def flush(self, object_types: Sequence[str] = ()) -> Dict[str, int]: return {} diff --git a/swh/storage/in_memory.py b/swh/storage/in_memory.py index 951f2e47..dbc5af74 100644 --- a/swh/storage/in_memory.py +++ b/swh/storage/in_memory.py @@ -1,738 +1,785 @@ -# Copyright (C) 2015-2020 The Software Heritage developers +# Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from collections import defaultdict import datetime import functools import itertools import random from typing import ( Any, Dict, Generic, Iterable, Iterator, List, Optional, Tuple, Type, TypeVar, Union, ) from swh.model.identifiers import ExtendedSWHID from swh.model.model import Content, Sha1Git, SkippedContent from swh.storage.cassandra import CassandraStorage from swh.storage.cassandra.model import ( BaseRow, ContentRow, DirectoryEntryRow, DirectoryRow, ExtIDByTargetRow, ExtIDRow, MetadataAuthorityRow, MetadataFetcherRow, ObjectCountRow, OriginRow, OriginVisitRow, OriginVisitStatusRow, RawExtrinsicMetadataByIdRow, RawExtrinsicMetadataRow, ReleaseRow, RevisionParentRow, RevisionRow, SkippedContentRow, SnapshotBranchRow, SnapshotRow, ) from swh.storage.interface import ListOrder from swh.storage.objstorage import ObjStorage from .common import origin_url_to_sha1 from .writer import JournalWriter TRow = TypeVar("TRow", bound=BaseRow) class Table(Generic[TRow]): def __init__(self, row_class: Type[TRow]): self.row_class = row_class self.primary_key_cols = row_class.PARTITION_KEY + row_class.CLUSTERING_KEY # Map from tokens to clustering keys to rows # These are not actually partitions (or rather, there is one partition # for each token) and they aren't sorted. # But it is good enough if we don't care about performance; # and makes the code a lot simpler. self.data: Dict[int, Dict[Tuple, TRow]] = defaultdict(dict) def __repr__(self): return f"<__module__.Table[{self.row_class.__name__}] object>" def partition_key(self, row: Union[TRow, Dict[str, Any]]) -> Tuple: """Returns the partition key of a row (ie. the cells which get hashed into the token.""" if isinstance(row, dict): row_d = row else: row_d = row.to_dict() return tuple(row_d[col] for col in self.row_class.PARTITION_KEY) def clustering_key(self, row: Union[TRow, Dict[str, Any]]) -> Tuple: """Returns the clustering key of a row (ie. the cells which are used for sorting rows within a partition.""" if isinstance(row, dict): row_d = row else: row_d = row.to_dict() return tuple(row_d[col] for col in self.row_class.CLUSTERING_KEY) def primary_key(self, row): return self.partition_key(row) + self.clustering_key(row) def primary_key_from_dict(self, d: Dict[str, Any]) -> Tuple: """Returns the primary key (ie. concatenation of partition key and clustering key) of the given dictionary interpreted as a row.""" return tuple(d[col] for col in self.primary_key_cols) def token(self, key: Tuple): """Returns the token of a row (ie. the hash of its partition key).""" return hash(key) def get_partition(self, token: int) -> Dict[Tuple, TRow]: """Returns the partition that contains this token.""" return self.data[token] def insert(self, row: TRow): partition = self.data[self.token(self.partition_key(row))] partition[self.clustering_key(row)] = row def split_primary_key(self, key: Tuple) -> Tuple[Tuple, Tuple]: """Returns (partition_key, clustering_key) from a partition key""" assert len(key) == len(self.primary_key_cols) partition_key = key[0 : len(self.row_class.PARTITION_KEY)] clustering_key = key[len(self.row_class.PARTITION_KEY) :] return (partition_key, clustering_key) def get_from_partition_key(self, partition_key: Tuple) -> Iterable[TRow]: """Returns at most one row, from its partition key.""" token = self.token(partition_key) for row in self.get_from_token(token): if self.partition_key(row) == partition_key: yield row def get_from_primary_key(self, primary_key: Tuple) -> Optional[TRow]: """Returns at most one row, from its primary key.""" (partition_key, clustering_key) = self.split_primary_key(primary_key) token = self.token(partition_key) partition = self.get_partition(token) return partition.get(clustering_key) def get_from_token(self, token: int) -> Iterable[TRow]: """Returns all rows whose token (ie. non-cryptographic hash of the partition key) is the one passed as argument.""" return (v for (k, v) in sorted(self.get_partition(token).items())) def iter_all(self) -> Iterator[Tuple[Tuple, TRow]]: return ( (self.primary_key(row), row) for (token, partition) in self.data.items() for (clustering_key, row) in partition.items() ) def get_random(self) -> Optional[TRow]: return random.choice([row for (pk, row) in self.iter_all()]) class InMemoryCqlRunner: def __init__(self): self._contents = Table(ContentRow) self._content_indexes = defaultdict(lambda: defaultdict(set)) self._skipped_contents = Table(ContentRow) self._skipped_content_indexes = defaultdict(lambda: defaultdict(set)) self._directories = Table(DirectoryRow) self._directory_entries = Table(DirectoryEntryRow) self._revisions = Table(RevisionRow) self._revision_parents = Table(RevisionParentRow) self._releases = Table(ReleaseRow) self._snapshots = Table(SnapshotRow) self._snapshot_branches = Table(SnapshotBranchRow) self._origins = Table(OriginRow) self._origin_visits = Table(OriginVisitRow) self._origin_visit_statuses = Table(OriginVisitStatusRow) self._metadata_authorities = Table(MetadataAuthorityRow) self._metadata_fetchers = Table(MetadataFetcherRow) self._raw_extrinsic_metadata = Table(RawExtrinsicMetadataRow) self._raw_extrinsic_metadata_by_id = Table(RawExtrinsicMetadataByIdRow) self._extid = Table(ExtIDRow) self._stat_counters = defaultdict(int) def increment_counter(self, object_type: str, nb: int): self._stat_counters[object_type] += nb def stat_counters(self) -> Iterable[ObjectCountRow]: for (object_type, count) in self._stat_counters.items(): yield ObjectCountRow(partition_key=0, object_type=object_type, count=count) ########################## # 'content' table ########################## def _content_add_finalize(self, content: ContentRow) -> None: self._contents.insert(content) self.increment_counter("content", 1) def content_add_prepare(self, content: ContentRow): finalizer = functools.partial(self._content_add_finalize, content) return (self._contents.token(self._contents.partition_key(content)), finalizer) def content_get_from_pk( self, content_hashes: Dict[str, bytes] ) -> Optional[ContentRow]: primary_key = self._contents.primary_key_from_dict(content_hashes) return self._contents.get_from_primary_key(primary_key) def content_get_from_token(self, token: int) -> Iterable[ContentRow]: return self._contents.get_from_token(token) def content_get_random(self) -> Optional[ContentRow]: return self._contents.get_random() def content_get_token_range( self, start: int, end: int, limit: int, ) -> Iterable[Tuple[int, ContentRow]]: matches = [ (token, row) for (token, partition) in self._contents.data.items() for (clustering_key, row) in partition.items() if start <= token <= end ] matches.sort() return matches[0:limit] def content_missing_from_hashes( self, contents_hashes: List[Dict[str, bytes]] ) -> Iterator[Dict[str, bytes]]: for content_hashes in contents_hashes: if not self.content_get_from_pk(content_hashes): yield content_hashes ########################## # 'content_by_*' tables ########################## def content_missing_by_sha1_git(self, ids: List[bytes]) -> List[bytes]: missing = [] for id_ in ids: if id_ not in self._content_indexes["sha1_git"]: missing.append(id_) return missing def content_index_add_one(self, algo: str, content: Content, token: int) -> None: self._content_indexes[algo][content.get_hash(algo)].add(token) - def content_get_tokens_from_single_hash( - self, algo: str, hash_: bytes + def content_get_tokens_from_single_algo( + self, algo: str, hashes: List[bytes] ) -> Iterable[int]: - return self._content_indexes[algo][hash_] + for hash_ in hashes: + yield from self._content_indexes[algo][hash_] ########################## # 'skipped_content' table ########################## def _skipped_content_add_finalize(self, content: SkippedContentRow) -> None: self._skipped_contents.insert(content) self.increment_counter("skipped_content", 1) def skipped_content_add_prepare(self, content: SkippedContentRow): finalizer = functools.partial(self._skipped_content_add_finalize, content) return ( self._skipped_contents.token(self._contents.partition_key(content)), finalizer, ) def skipped_content_get_from_pk( self, content_hashes: Dict[str, bytes] ) -> Optional[SkippedContentRow]: primary_key = self._skipped_contents.primary_key_from_dict(content_hashes) return self._skipped_contents.get_from_primary_key(primary_key) def skipped_content_get_from_token(self, token: int) -> Iterable[SkippedContentRow]: return self._skipped_contents.get_from_token(token) ########################## # 'skipped_content_by_*' tables ########################## def skipped_content_index_add_one( self, algo: str, content: SkippedContent, token: int ) -> None: self._skipped_content_indexes[algo][content.get_hash(algo)].add(token) def skipped_content_get_tokens_from_single_hash( self, algo: str, hash_: bytes ) -> Iterable[int]: return self._skipped_content_indexes[algo][hash_] ########################## # 'directory' table ########################## def directory_missing(self, ids: List[bytes]) -> List[bytes]: missing = [] for id_ in ids: if self._directories.get_from_primary_key((id_,)) is None: missing.append(id_) return missing def directory_add_one(self, directory: DirectoryRow) -> None: self._directories.insert(directory) self.increment_counter("directory", 1) def directory_get_random(self) -> Optional[DirectoryRow]: return self._directories.get_random() ########################## # 'directory_entry' table ########################## def directory_entry_add_one(self, entry: DirectoryEntryRow) -> None: self._directory_entries.insert(entry) def directory_entry_get( self, directory_ids: List[Sha1Git] ) -> Iterable[DirectoryEntryRow]: for id_ in directory_ids: yield from self._directory_entries.get_from_partition_key((id_,)) def directory_entry_get_from_name( self, directory_id: Sha1Git, from_: bytes, limit: int ) -> Iterable[DirectoryEntryRow]: # Get all entries entries = self._directory_entries.get_from_partition_key((directory_id,)) # Filter out the ones before from_ entries = itertools.dropwhile(lambda entry: entry.name < from_, entries) # Apply limit return itertools.islice(entries, limit) ########################## # 'revision' table ########################## def revision_missing(self, ids: List[bytes]) -> Iterable[bytes]: missing = [] for id_ in ids: if self._revisions.get_from_primary_key((id_,)) is None: missing.append(id_) return missing def revision_add_one(self, revision: RevisionRow) -> None: self._revisions.insert(revision) self.increment_counter("revision", 1) def revision_get_ids(self, revision_ids) -> Iterable[int]: for id_ in revision_ids: if self._revisions.get_from_primary_key((id_,)) is not None: yield id_ def revision_get(self, revision_ids: List[Sha1Git]) -> Iterable[RevisionRow]: for id_ in revision_ids: row = self._revisions.get_from_primary_key((id_,)) if row: yield row def revision_get_random(self) -> Optional[RevisionRow]: return self._revisions.get_random() ########################## # 'revision_parent' table ########################## def revision_parent_add_one(self, revision_parent: RevisionParentRow) -> None: self._revision_parents.insert(revision_parent) def revision_parent_get(self, revision_id: Sha1Git) -> Iterable[bytes]: for parent in self._revision_parents.get_from_partition_key((revision_id,)): yield parent.parent_id ########################## # 'release' table ########################## def release_missing(self, ids: List[bytes]) -> List[bytes]: missing = [] for id_ in ids: if self._releases.get_from_primary_key((id_,)) is None: missing.append(id_) return missing def release_add_one(self, release: ReleaseRow) -> None: self._releases.insert(release) self.increment_counter("release", 1) def release_get(self, release_ids: List[str]) -> Iterable[ReleaseRow]: for id_ in release_ids: row = self._releases.get_from_primary_key((id_,)) if row: yield row def release_get_random(self) -> Optional[ReleaseRow]: return self._releases.get_random() ########################## # 'snapshot' table ########################## def snapshot_missing(self, ids: List[bytes]) -> List[bytes]: missing = [] for id_ in ids: if self._snapshots.get_from_primary_key((id_,)) is None: missing.append(id_) return missing def snapshot_add_one(self, snapshot: SnapshotRow) -> None: self._snapshots.insert(snapshot) self.increment_counter("snapshot", 1) def snapshot_get_random(self) -> Optional[SnapshotRow]: return self._snapshots.get_random() ########################## # 'snapshot_branch' table ########################## def snapshot_branch_add_one(self, branch: SnapshotBranchRow) -> None: self._snapshot_branches.insert(branch) def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Dict[Optional[str], int]: """Returns a dictionary from type names to the number of branches of that type.""" counts: Dict[Optional[str], int] = defaultdict(int) for branch in self._snapshot_branches.get_from_partition_key((snapshot_id,)): if branch_name_exclude_prefix and branch.name.startswith( branch_name_exclude_prefix ): continue if branch.target_type is None: target_type = None else: target_type = branch.target_type counts[target_type] += 1 return counts def snapshot_branch_get( self, snapshot_id: Sha1Git, from_: bytes, limit: int, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Iterable[SnapshotBranchRow]: count = 0 for branch in self._snapshot_branches.get_from_partition_key((snapshot_id,)): prefix = branch_name_exclude_prefix if branch.name >= from_ and ( prefix is None or not branch.name.startswith(prefix) ): count += 1 yield branch if count >= limit: break ########################## # 'origin' table ########################## def origin_add_one(self, origin: OriginRow) -> None: self._origins.insert(origin) self.increment_counter("origin", 1) def origin_get_by_sha1(self, sha1: bytes) -> Iterable[OriginRow]: return self._origins.get_from_partition_key((sha1,)) def origin_get_by_url(self, url: str) -> Iterable[OriginRow]: return self.origin_get_by_sha1(origin_url_to_sha1(url)) def origin_list( self, start_token: int, limit: int ) -> Iterable[Tuple[int, OriginRow]]: """Returns an iterable of (token, origin)""" matches = [ (token, row) for (token, partition) in self._origins.data.items() for (clustering_key, row) in partition.items() if token >= start_token ] matches.sort() return matches[0:limit] def origin_iter_all(self) -> Iterable[OriginRow]: return ( row for (token, partition) in self._origins.data.items() for (clustering_key, row) in partition.items() ) def origin_bump_next_visit_id(self, origin_url: str, visit_id: int) -> None: origin = list(self.origin_get_by_url(origin_url))[0] origin.next_visit_id = max(origin.next_visit_id, visit_id + 1) def origin_generate_unique_visit_id(self, origin_url: str) -> int: origin = list(self.origin_get_by_url(origin_url))[0] visit_id = origin.next_visit_id origin.next_visit_id += 1 return visit_id ########################## # 'origin_visit' table ########################## def origin_visit_get( self, origin_url: str, last_visit: Optional[int], limit: int, order: ListOrder, ) -> Iterable[OriginVisitRow]: visits = list(self._origin_visits.get_from_partition_key((origin_url,))) if last_visit is not None: if order == ListOrder.ASC: visits = [v for v in visits if v.visit > last_visit] else: visits = [v for v in visits if v.visit < last_visit] visits.sort(key=lambda v: v.visit, reverse=order == ListOrder.DESC) visits = visits[0:limit] return visits def origin_visit_add_one(self, visit: OriginVisitRow) -> None: self._origin_visits.insert(visit) self.increment_counter("origin_visit", 1) def origin_visit_get_one( self, origin_url: str, visit_id: int ) -> Optional[OriginVisitRow]: return self._origin_visits.get_from_primary_key((origin_url, visit_id)) def origin_visit_get_all(self, origin_url: str) -> Iterable[OriginVisitRow]: return self._origin_visits.get_from_partition_key((origin_url,)) def origin_visit_iter(self, start_token: int) -> Iterator[OriginVisitRow]: """Returns all origin visits in order from this token, and wraps around the token space.""" return ( row for (token, partition) in self._origin_visits.data.items() for (clustering_key, row) in partition.items() ) ########################## # 'origin_visit_status' table ########################## def origin_visit_status_get_range( self, origin: str, visit: int, date_from: Optional[datetime.datetime], limit: int, order: ListOrder, ) -> Iterable[OriginVisitStatusRow]: statuses = list(self.origin_visit_status_get(origin, visit)) if date_from is not None: if order == ListOrder.ASC: statuses = [s for s in statuses if s.date >= date_from] else: statuses = [s for s in statuses if s.date <= date_from] statuses.sort(key=lambda s: s.date, reverse=order == ListOrder.DESC) return statuses[0:limit] def origin_visit_status_add_one(self, visit_update: OriginVisitStatusRow) -> None: self._origin_visit_statuses.insert(visit_update) self.increment_counter("origin_visit_status", 1) def origin_visit_status_get_latest( self, origin: str, visit: int, ) -> Optional[OriginVisitStatusRow]: """Given an origin visit id, return its latest origin_visit_status """ return next(self.origin_visit_status_get(origin, visit), None) def origin_visit_status_get( self, origin: str, visit: int, ) -> Iterator[OriginVisitStatusRow]: """Return all origin visit statuses for a given visit """ statuses = [ s for s in self._origin_visit_statuses.get_from_partition_key((origin,)) if s.visit == visit ] statuses.sort(key=lambda s: s.date, reverse=True) return iter(statuses) ########################## # 'metadata_authority' table ########################## def metadata_authority_add(self, authority: MetadataAuthorityRow): self._metadata_authorities.insert(authority) self.increment_counter("metadata_authority", 1) def metadata_authority_get(self, type, url) -> Optional[MetadataAuthorityRow]: return self._metadata_authorities.get_from_primary_key((url, type)) ########################## # 'metadata_fetcher' table ########################## def metadata_fetcher_add(self, fetcher: MetadataFetcherRow): self._metadata_fetchers.insert(fetcher) self.increment_counter("metadata_fetcher", 1) def metadata_fetcher_get(self, name, version) -> Optional[MetadataAuthorityRow]: return self._metadata_fetchers.get_from_primary_key((name, version)) ######################### # 'raw_extrinsic_metadata_by_id' table ######################### def raw_extrinsic_metadata_by_id_add( self, row: RawExtrinsicMetadataByIdRow ) -> None: self._raw_extrinsic_metadata_by_id.insert(row) def raw_extrinsic_metadata_get_by_ids( self, ids ) -> List[RawExtrinsicMetadataByIdRow]: results = [] for id_ in ids: result = self._raw_extrinsic_metadata_by_id.get_from_primary_key((id_,)) if result: results.append(result) return results ######################### # 'raw_extrinsic_metadata' table ######################### def raw_extrinsic_metadata_add(self, raw_extrinsic_metadata): self._raw_extrinsic_metadata.insert(raw_extrinsic_metadata) self.increment_counter("raw_extrinsic_metadata", 1) def raw_extrinsic_metadata_get_after_date( self, target: str, authority_type: str, authority_url: str, after: datetime.datetime, ) -> Iterable[RawExtrinsicMetadataRow]: metadata = self.raw_extrinsic_metadata_get( target, authority_type, authority_url ) return (m for m in metadata if m.discovery_date > after) def raw_extrinsic_metadata_get_after_date_and_id( self, target: str, authority_type: str, authority_url: str, after_date: datetime.datetime, after_id: bytes, ) -> Iterable[RawExtrinsicMetadataRow]: metadata = self._raw_extrinsic_metadata.get_from_partition_key((target,)) after_tuple = (after_date, after_id) return ( m for m in metadata if m.authority_type == authority_type and m.authority_url == authority_url and (m.discovery_date, m.id) > after_tuple ) def raw_extrinsic_metadata_get( self, target: str, authority_type: str, authority_url: str ) -> Iterable[RawExtrinsicMetadataRow]: metadata = self._raw_extrinsic_metadata.get_from_partition_key((target,)) return ( m for m in metadata if m.authority_type == authority_type and m.authority_url == authority_url ) def raw_extrinsic_metadata_get_authorities( self, target: str ) -> Iterable[Tuple[str, str]]: metadata = self._raw_extrinsic_metadata.get_from_partition_key((target,)) return ((m.authority_type, m.authority_url) for m in metadata) ######################### # 'extid' table ######################### def _extid_add_finalize(self, extid: ExtIDRow) -> None: self._extid.insert(extid) self.increment_counter("extid", 1) def extid_add_prepare(self, extid: ExtIDRow): finalizer = functools.partial(self._extid_add_finalize, extid) return (self._extid.token(self._extid.partition_key(extid)), finalizer) def extid_index_add_one(self, row: ExtIDByTargetRow) -> None: pass def extid_get_from_pk( self, extid_type: str, extid: bytes, extid_version: int, target: ExtendedSWHID, ) -> Optional[ExtIDRow]: primary_key = self._extid.primary_key_from_dict( dict( extid_type=extid_type, extid=extid, extid_version=extid_version, target_type=target.object_type.value, target=target.object_id, ) ) return self._extid.get_from_primary_key(primary_key) - def extid_get_from_extid(self, extid_type: str, extid: bytes) -> Iterable[ExtIDRow]: + def extid_get_from_extid( + self, extid_type: str, extid: bytes, + ) -> Iterable[ExtIDRow]: return ( row for pk, row in self._extid.iter_all() if row.extid_type == extid_type and row.extid == extid ) - def extid_get_from_target( - self, target_type: str, target: bytes + def extid_get_from_extid_and_version( + self, extid_type: str, extid: bytes, extid_version: int, + ) -> Iterable[ExtIDRow]: + return ( + row + for pk, row in self._extid.iter_all() + if row.extid_type == extid_type + and row.extid == extid + and (extid_version is None or row.extid_version == extid_version) + ) + + def _extid_get_from_target_with_type_and_version( + self, target_type: str, target: bytes, extid_type: str, extid_version: int, + ) -> Iterable[ExtIDRow]: + return ( + row + for pk, row in self._extid.iter_all() + if row.target_type == target_type + and row.target == target + and row.extid_version == extid_version + and row.extid_type == extid_type + ) + + def _extid_get_from_target( + self, target_type: str, target: bytes, ) -> Iterable[ExtIDRow]: return ( row for pk, row in self._extid.iter_all() if row.target_type == target_type and row.target == target ) + def extid_get_from_target( + self, + target_type: str, + target: bytes, + extid_type: Optional[str] = None, + extid_version: Optional[int] = None, + ) -> Iterable[ExtIDRow]: + if (extid_version is not None and extid_type is None) or ( + extid_version is None and extid_type is not None + ): + raise ValueError("You must provide both extid_type and extid_version") + + if extid_type is not None and extid_version is not None: + extids = self._extid_get_from_target_with_type_and_version( + target_type, target, extid_type, extid_version + ) + else: + extids = self._extid_get_from_target(target_type, target) + return extids + class InMemoryStorage(CassandraStorage): _cql_runner: InMemoryCqlRunner # type: ignore def __init__(self, journal_writer=None): self.reset() self.journal_writer = JournalWriter(journal_writer) self._allow_overwrite = False + self._directory_entries_insert_algo = "one-by-one" def reset(self): self._cql_runner = InMemoryCqlRunner() self.objstorage = ObjStorage({"cls": "memory"}) def check_config(self, *, check_write: bool) -> bool: return True diff --git a/swh/storage/interface.py b/swh/storage/interface.py index 1a8feeb9..fbf65d59 100644 --- a/swh/storage/interface.py +++ b/swh/storage/interface.py @@ -1,1308 +1,1322 @@ -# Copyright (C) 2015-2020 The Software Heritage developers +# Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime from enum import Enum from typing import Any, Dict, Iterable, List, Optional, Sequence, Tuple, TypeVar from typing_extensions import Protocol, TypedDict, runtime_checkable from swh.core.api import remote_api_endpoint from swh.core.api.classes import PagedResult as CorePagedResult from swh.model.identifiers import ExtendedSWHID, ObjectType from swh.model.model import ( Content, Directory, DirectoryEntry, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Release, Revision, Sha1, Sha1Git, SkippedContent, Snapshot, SnapshotBranch, ) class ListOrder(Enum): """Specifies the order for paginated endpoints returning sorted results.""" ASC = "asc" DESC = "desc" class PartialBranches(TypedDict): """Type of the dictionary returned by snapshot_get_branches""" id: Sha1Git """Identifier of the snapshot""" branches: Dict[bytes, Optional[SnapshotBranch]] """A dict of branches contained in the snapshot whose keys are the branches' names""" next_branch: Optional[bytes] """The name of the first branch not returned or :const:`None` if the snapshot has less than the request number of branches.""" TResult = TypeVar("TResult") PagedResult = CorePagedResult[TResult, str] # TODO: Make it an enum (too much impact) VISIT_STATUSES = ["created", "ongoing", "full", "partial"] def deprecated(f): f.deprecated_endpoint = True return f @runtime_checkable class StorageInterface(Protocol): @remote_api_endpoint("check_config") def check_config(self, *, check_write: bool) -> bool: """Check that the storage is configured and ready to go.""" ... @remote_api_endpoint("content/add") def content_add(self, content: List[Content]) -> Dict[str, int]: """Add content blobs to the storage Args: contents (iterable): iterable of dictionaries representing individual pieces of content to add. Each dictionary has the following keys: - data (bytes): the actual content - length (int): content length - one key for each checksum algorithm in :data:`swh.model.hashutil.ALGORITHMS`, mapped to the corresponding checksum - status (str): one of visible, hidden Raises: The following exceptions can occur: - HashCollision in case of collision - Any other exceptions raise by the db In case of errors, some of the content may have been stored in the DB and in the objstorage. Since additions to both idempotent, that should not be a problem. Returns: Summary dict with the following keys and associated values: content:add: New contents added content:add:bytes: Sum of the contents' length data """ ... @remote_api_endpoint("content/update") def content_update( self, contents: List[Dict[str, Any]], keys: List[str] = [] ) -> None: """Update content blobs to the storage. Does nothing for unknown contents or skipped ones. Args: content: iterable of dictionaries representing individual pieces of content to update. Each dictionary has the following keys: - data (bytes): the actual content - length (int): content length (default: -1) - one key for each checksum algorithm in :data:`swh.model.hashutil.ALGORITHMS`, mapped to the corresponding checksum - status (str): one of visible, hidden, absent keys (list): List of keys (str) whose values needs an update, e.g., new hash column """ ... @remote_api_endpoint("content/add_metadata") def content_add_metadata(self, content: List[Content]) -> Dict[str, int]: """Add content metadata to the storage (like `content_add`, but without inserting to the objstorage). Args: content (iterable): iterable of dictionaries representing individual pieces of content to add. Each dictionary has the following keys: - length (int): content length (default: -1) - one key for each checksum algorithm in :data:`swh.model.hashutil.ALGORITHMS`, mapped to the corresponding checksum - status (str): one of visible, hidden, absent - reason (str): if status = absent, the reason why - origin (int): if status = absent, the origin we saw the content in - ctime (datetime): time of insertion in the archive Returns: Summary dict with the following key and associated values: content:add: New contents added skipped_content:add: New skipped contents (no data) added """ ... @remote_api_endpoint("content/data") def content_get_data(self, content: Sha1) -> Optional[bytes]: """Given a content identifier, returns its associated data if any. Args: content: sha1 identifier Returns: raw content data (bytes) """ ... @remote_api_endpoint("content/partition") def content_get_partition( self, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Content]: """Splits contents into nb_partitions, and returns one of these based on partition_id (which must be in [0, nb_partitions-1]) There is no guarantee on how the partitioning is done, or the result order. Args: partition_id: index of the partition to fetch nb_partitions: total number of partitions to split into page_token: opaque token used for pagination. limit: Limit result (default to 1000) Returns: PagedResult of Content model objects within the partition. If next_page_token is None, there is no longer data to retrieve. """ ... @remote_api_endpoint("content/metadata") def content_get( self, contents: List[bytes], algo: str = "sha1" ) -> List[Optional[Content]]: """Retrieve content metadata in bulk Args: content: List of content identifiers algo: one of the checksum algorithm in :data:`swh.model.hashutil.DEFAULT_ALGORITHMS` Returns: List of contents model objects when they exist, None otherwise. """ ... @remote_api_endpoint("content/missing") def content_missing( self, contents: List[Dict[str, Any]], key_hash: str = "sha1" ) -> Iterable[bytes]: """List content missing from storage Args: content: iterable of dictionaries whose keys are either 'length' or an item of :data:`swh.model.hashutil.ALGORITHMS`; mapped to the corresponding checksum (or length). key_hash: name of the column to use as hash id result (default: 'sha1') Raises: StorageArgumentException when key_hash is unknown. TODO: an exception when we get a hash collision. Returns: iterable of missing content ids (as per the `key_hash` column) """ ... @remote_api_endpoint("content/missing/sha1") def content_missing_per_sha1(self, contents: List[bytes]) -> Iterable[bytes]: """List content missing from storage based only on sha1. Args: contents: List of sha1 to check for absence. Raises: TODO: an exception when we get a hash collision. Returns: Iterable of missing content ids (sha1) """ ... @remote_api_endpoint("content/missing/sha1_git") def content_missing_per_sha1_git( self, contents: List[Sha1Git] ) -> Iterable[Sha1Git]: """List content missing from storage based only on sha1_git. Args: contents (List): An iterable of content id (sha1_git) Yields: missing contents sha1_git """ ... @remote_api_endpoint("content/present") def content_find(self, content: Dict[str, Any]) -> List[Content]: """Find a content hash in db. Args: content: a dictionary representing one content hash, mapping checksum algorithm names (see swh.model.hashutil.ALGORITHMS) to checksum values Raises: ValueError: in case the key of the dictionary is not sha1, sha1_git nor sha256. Returns: an iterable of Content objects matching the search criteria if the content exist. Empty iterable otherwise. """ ... @remote_api_endpoint("content/get_random") def content_get_random(self) -> Sha1Git: """Finds a random content id. Returns: a sha1_git """ ... @remote_api_endpoint("content/skipped/add") def skipped_content_add(self, content: List[SkippedContent]) -> Dict[str, int]: """Add contents to the skipped_content list, which contains (partial) information about content missing from the archive. Args: contents (iterable): iterable of dictionaries representing individual pieces of content to add. Each dictionary has the following keys: - length (Optional[int]): content length (default: -1) - one key for each checksum algorithm in :data:`swh.model.hashutil.ALGORITHMS`, mapped to the corresponding checksum; each is optional - status (str): must be "absent" - reason (str): the reason why the content is absent - origin (int): if status = absent, the origin we saw the content in Raises: The following exceptions can occur: - HashCollision in case of collision - Any other exceptions raise by the backend In case of errors, some content may have been stored in the DB and in the objstorage. Since additions to both idempotent, that should not be a problem. Returns: Summary dict with the following key and associated values: skipped_content:add: New skipped contents (no data) added """ ... @remote_api_endpoint("content/skipped/missing") def skipped_content_missing( self, contents: List[Dict[str, Any]] ) -> Iterable[Dict[str, Any]]: """List skipped contents missing from storage. Args: contents: iterable of dictionaries containing the data for each checksum algorithm. Returns: Iterable of missing skipped contents as dict """ ... @remote_api_endpoint("directory/add") def directory_add(self, directories: List[Directory]) -> Dict[str, int]: """Add directories to the storage Args: directories (iterable): iterable of dictionaries representing the individual directories to add. Each dict has the following keys: - id (sha1_git): the id of the directory to add - entries (list): list of dicts for each entry in the directory. Each dict has the following keys: - name (bytes) - type (one of 'file', 'dir', 'rev'): type of the directory entry (file, directory, revision) - target (sha1_git): id of the object pointed at by the directory entry - perms (int): entry permissions Returns: Summary dict of keys with associated count as values: directory:add: Number of directories actually added """ ... @remote_api_endpoint("directory/missing") def directory_missing(self, directories: List[Sha1Git]) -> Iterable[Sha1Git]: """List directories missing from storage. Args: directories: list of directory ids Yields: missing directory ids """ ... @remote_api_endpoint("directory/ls") def directory_ls( self, directory: Sha1Git, recursive: bool = False ) -> Iterable[Dict[str, Any]]: """List entries for one directory. If `recursive=True`, names in the path of a dir/file not at the root are concatenated with a slash (`/`). Args: directory: the directory to list entries from. recursive: if flag on, this list recursively from this directory. Yields: directory entries for such directory. """ ... @remote_api_endpoint("directory/path") def directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes] ) -> Optional[Dict[str, Any]]: """Get the directory entry (either file or dir) from directory with path. Args: directory: directory id paths: path to lookup from the top level directory. From left (top) to right (bottom). Returns: The corresponding directory entry as dict if found, None otherwise. """ ... @remote_api_endpoint("directory/get_entries") def directory_get_entries( self, directory_id: Sha1Git, page_token: Optional[bytes] = None, limit: int = 1000, ) -> Optional[PagedResult[DirectoryEntry]]: """Get the content, possibly partial, of a directory with the given id The entries of the directory are not guaranteed to be returned in any particular order. The number of results is not guaranteed to be lower than the ``limit``. Args: directory_id: dentifier of the directory page_token: opaque string used to get the next results of a search limit: Number of entries to return Returns: None if the directory does not exist; a page of DirectoryEntry objects otherwise. """ ... @remote_api_endpoint("directory/get_random") def directory_get_random(self) -> Sha1Git: """Finds a random directory id. Returns: a sha1_git """ ... @remote_api_endpoint("revision/add") def revision_add(self, revisions: List[Revision]) -> Dict[str, int]: """Add revisions to the storage Args: revisions (List[dict]): iterable of dictionaries representing the individual revisions to add. Each dict has the following keys: - **id** (:class:`sha1_git`): id of the revision to add - **date** (:class:`dict`): date the revision was written - **committer_date** (:class:`dict`): date the revision got added to the origin - **type** (one of 'git', 'tar'): type of the revision added - **directory** (:class:`sha1_git`): the directory the revision points at - **message** (:class:`bytes`): the message associated with the revision - **author** (:class:`Dict[str, bytes]`): dictionary with keys: name, fullname, email - **committer** (:class:`Dict[str, bytes]`): dictionary with keys: name, fullname, email - **metadata** (:class:`jsonb`): extra information as dictionary - **synthetic** (:class:`bool`): revision's nature (tarball, directory creates synthetic revision`) - **parents** (:class:`list[sha1_git]`): the parents of this revision date dictionaries have the form defined in :mod:`swh.model`. Returns: Summary dict of keys with associated count as values revision:add: New objects actually stored in db """ ... @remote_api_endpoint("revision/missing") def revision_missing(self, revisions: List[Sha1Git]) -> Iterable[Sha1Git]: """List revisions missing from storage Args: revisions: revision ids Yields: missing revision ids """ ... @remote_api_endpoint("revision") def revision_get(self, revision_ids: List[Sha1Git]) -> List[Optional[Revision]]: """Get revisions from storage Args: revisions: revision ids Returns: list of revision object (if the revision exists or None otherwise) """ ... @remote_api_endpoint("extid/from_extid") - def extid_get_from_extid(self, id_type: str, ids: List[bytes]) -> List[ExtID]: + def extid_get_from_extid( + self, id_type: str, ids: List[bytes], version: Optional[int] = None + ) -> List[ExtID]: """Get ExtID objects from external IDs Args: id_type: type of the given external identifiers (e.g. 'mercurial') ids: list of external IDs + version: (Optional) version to use as filter Returns: list of ExtID objects """ ... @remote_api_endpoint("extid/from_target") def extid_get_from_target( - self, target_type: ObjectType, ids: List[Sha1Git] + self, + target_type: ObjectType, + ids: List[Sha1Git], + extid_type: Optional[str] = None, + extid_version: Optional[int] = None, ) -> List[ExtID]: """Get ExtID objects from target IDs and target_type Args: target_type: type the SWH object ids: list of target IDs + extid_type: (Optional) extid_type to use as filter. This cannot be empty if + extid_version is provided. + extid_version: (Optional) version to use as filter. This cannot be empty if + extid_type is provided. + + Raises: + ValueError if extid_version is provided without extid_type and vice versa. Returns: list of ExtID objects """ ... @remote_api_endpoint("extid/add") def extid_add(self, ids: List[ExtID]) -> Dict[str, int]: """Add a series of ExtID objects Args: ids: list of ExtID objects Returns: Summary dict of keys with associated count as values extid:add: New ExtID objects actually stored in db """ ... @remote_api_endpoint("revision/log") def revision_log( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Dict[str, Any]]]: """Fetch revision entry from the given root revisions. Args: revisions: array of root revisions to lookup limit: limitation on the output result. Default to None. Yields: revision entries log from the given root root revisions """ ... @remote_api_endpoint("revision/shortlog") def revision_shortlog( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Tuple[Sha1Git, Tuple[Sha1Git, ...]]]]: """Fetch the shortlog for the given revisions Args: revisions: list of root revisions to lookup limit: depth limitation for the output Yields: a list of (id, parents) tuples """ ... @remote_api_endpoint("revision/get_random") def revision_get_random(self) -> Sha1Git: """Finds a random revision id. Returns: a sha1_git """ ... @remote_api_endpoint("release/add") def release_add(self, releases: List[Release]) -> Dict[str, int]: """Add releases to the storage Args: releases (List[dict]): iterable of dictionaries representing the individual releases to add. Each dict has the following keys: - **id** (:class:`sha1_git`): id of the release to add - **revision** (:class:`sha1_git`): id of the revision the release points to - **date** (:class:`dict`): the date the release was made - **name** (:class:`bytes`): the name of the release - **comment** (:class:`bytes`): the comment associated with the release - **author** (:class:`Dict[str, bytes]`): dictionary with keys: name, fullname, email the date dictionary has the form defined in :mod:`swh.model`. Returns: Summary dict of keys with associated count as values release:add: New objects contents actually stored in db """ ... @remote_api_endpoint("release/missing") def release_missing(self, releases: List[Sha1Git]) -> Iterable[Sha1Git]: """List missing release ids from storage Args: releases: release ids Yields: a list of missing release ids """ ... @remote_api_endpoint("release") def release_get(self, releases: List[Sha1Git]) -> List[Optional[Release]]: """Given a list of sha1, return the releases's information Args: releases: list of sha1s Returns: List of releases matching the identifiers or None if the release does not exist. """ ... @remote_api_endpoint("release/get_random") def release_get_random(self) -> Sha1Git: """Finds a random release id. Returns: a sha1_git """ ... @remote_api_endpoint("snapshot/add") def snapshot_add(self, snapshots: List[Snapshot]) -> Dict[str, int]: """Add snapshots to the storage. Args: snapshot ([dict]): the snapshots to add, containing the following keys: - **id** (:class:`bytes`): id of the snapshot - **branches** (:class:`dict`): branches the snapshot contains, mapping the branch name (:class:`bytes`) to the branch target, itself a :class:`dict` (or ``None`` if the branch points to an unknown object) - **target_type** (:class:`str`): one of ``content``, ``directory``, ``revision``, ``release``, ``snapshot``, ``alias`` - **target** (:class:`bytes`): identifier of the target (currently a ``sha1_git`` for all object kinds, or the name of the target branch for aliases) Raises: ValueError: if the origin or visit id does not exist. Returns: Summary dict of keys with associated count as values snapshot:add: Count of object actually stored in db """ ... @remote_api_endpoint("snapshot/missing") def snapshot_missing(self, snapshots: List[Sha1Git]) -> Iterable[Sha1Git]: """List snapshots missing from storage Args: snapshots: snapshot ids Yields: missing snapshot ids """ ... @remote_api_endpoint("snapshot") def snapshot_get(self, snapshot_id: Sha1Git) -> Optional[Dict[str, Any]]: """Get the content, possibly partial, of a snapshot with the given id The branches of the snapshot are iterated in the lexicographical order of their names. .. warning:: At most 1000 branches contained in the snapshot will be returned for performance reasons. In order to browse the whole set of branches, the method :meth:`snapshot_get_branches` should be used instead. Args: snapshot_id: snapshot identifier Returns: dict: a dict with three keys: * **id**: identifier of the snapshot * **branches**: a dict of branches contained in the snapshot whose keys are the branches' names. * **next_branch**: the name of the first branch not returned or :const:`None` if the snapshot has less than 1000 branches. """ ... @remote_api_endpoint("snapshot/count_branches") def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[Dict[Optional[str], int]]: """Count the number of branches in the snapshot with the given id Args: snapshot_id: snapshot identifier branch_name_exclude_prefix: if provided, do not count branches whose name starts with given prefix Returns: A dict whose keys are the target types of branches and values their corresponding amount """ ... @remote_api_endpoint("snapshot/get_branches") def snapshot_get_branches( self, snapshot_id: Sha1Git, branches_from: bytes = b"", branches_count: int = 1000, target_types: Optional[List[str]] = None, branch_name_include_substring: Optional[bytes] = None, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[PartialBranches]: """Get the content, possibly partial, of a snapshot with the given id The branches of the snapshot are iterated in the lexicographical order of their names. Args: snapshot_id: identifier of the snapshot branches_from: optional parameter used to skip branches whose name is lesser than it before returning them branches_count: optional parameter used to restrain the amount of returned branches target_types: optional parameter used to filter the target types of branch to return (possible values that can be contained in that list are `'content', 'directory', 'revision', 'release', 'snapshot', 'alias'`) branch_name_include_substring: if provided, only return branches whose name contains given substring branch_name_exclude_prefix: if provided, do not return branches whose name contains given prefix Returns: dict: None if the snapshot does not exist; a dict with three keys otherwise: * **id**: identifier of the snapshot * **branches**: a dict of branches contained in the snapshot whose keys are the branches' names. * **next_branch**: the name of the first branch not returned or :const:`None` if the snapshot has less than `branches_count` branches after `branches_from` included. """ ... @remote_api_endpoint("snapshot/get_random") def snapshot_get_random(self) -> Sha1Git: """Finds a random snapshot id. Returns: a sha1_git """ ... @remote_api_endpoint("origin/visit/add") def origin_visit_add(self, visits: List[OriginVisit]) -> Iterable[OriginVisit]: """Add visits to storage. If the visits have no id, they will be created and assigned one. The resulted visits are visits with their visit id set. Args: visits: List of OriginVisit objects to add Raises: StorageArgumentException if some origin visit reference unknown origins Returns: List[OriginVisit] stored """ ... @remote_api_endpoint("origin/visit_status/add") def origin_visit_status_add( self, visit_statuses: List[OriginVisitStatus], ) -> Dict[str, int]: """Add origin visit statuses. If there is already a status for the same origin and visit id at the same date, the new one will be either dropped or will replace the existing one (it is unspecified which one of these two behaviors happens). Args: visit_statuses: origin visit statuses to add Raises: StorageArgumentException if the origin of the visit status is unknown """ ... @remote_api_endpoint("origin/visit/get") def origin_visit_get( self, origin: str, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisit]: """Retrieve page of OriginVisit information. Args: origin: The visited origin page_token: opaque string used to get the next results of a search order: Order on visit id fields to list origin visits (default to asc) limit: Number of visits to return Raises: StorageArgumentException if the order is wrong or the page_token type is mistyped. Returns: Page of OriginVisit data model objects. if next_page_token is None, there is no longer data to retrieve. """ ... @remote_api_endpoint("origin/visit/find_by_date") def origin_visit_find_by_date( self, origin: str, visit_date: datetime.datetime ) -> Optional[OriginVisit]: """Retrieves the origin visit whose date is closest to the provided timestamp. In case of a tie, the visit with largest id is selected. Args: origin: origin (URL) visit_date: expected visit date Returns: A visit if found, None otherwise """ ... @remote_api_endpoint("origin/visit/getby") def origin_visit_get_by(self, origin: str, visit: int) -> Optional[OriginVisit]: """Retrieve origin visit's information. Args: origin: origin (URL) visit: visit id Returns: The information on that particular OriginVisit or None if it does not exist """ ... @remote_api_endpoint("origin/visit/get_latest") def origin_visit_get_latest( self, origin: str, type: Optional[str] = None, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisit]: """Get the latest origin visit for the given origin, optionally looking only for those with one of the given allowed_statuses or for those with a snapshot. Args: origin: origin URL type: Optional visit type to filter on (e.g git, tar, dsc, svn, hg, npm, pypi, ...) allowed_statuses: list of visit statuses considered to find the latest visit. For instance, ``allowed_statuses=['full']`` will only consider visits that have successfully run to completion. require_snapshot: If True, only a visit with a snapshot will be returned. Raises: StorageArgumentException if values for the allowed_statuses parameters are unknown Returns: OriginVisit matching the criteria if found, None otherwise. Note that as OriginVisit no longer held reference on the visit status or snapshot, you may want to use origin_visit_status_get_latest for those information. """ ... @remote_api_endpoint("origin/visit_status/get") def origin_visit_status_get( self, origin: str, visit: int, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisitStatus]: """Retrieve page of OriginVisitStatus information. Args: origin: The visited origin visit: The visit identifier page_token: opaque string used to get the next results of a search order: Order on visit status objects to list (default to asc) limit: Number of visit statuses to return Returns: Page of OriginVisitStatus data model objects. if next_page_token is None, there is no longer data to retrieve. """ ... @remote_api_endpoint("origin/visit_status/get_latest") def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisitStatus]: """Get the latest origin visit status for the given origin visit, optionally looking only for those with one of the given allowed_statuses or with a snapshot. Args: origin: origin URL allowed_statuses: list of visit statuses considered to find the latest visit. Possible values are {created, ongoing, partial, full}. For instance, ``allowed_statuses=['full']`` will only consider visits that have successfully run to completion. require_snapshot: If True, only a visit with a snapshot will be returned. Raises: StorageArgumentException if values for the allowed_statuses parameters are unknown Returns: The OriginVisitStatus matching the criteria """ ... @remote_api_endpoint("origin/visit_status/get_random") def origin_visit_status_get_random(self, type: str) -> Optional[OriginVisitStatus]: """Randomly select one successful origin visit with made in the last 3 months. Returns: One random OriginVisitStatus matching the selection criteria """ ... @remote_api_endpoint("object/find_by_sha1_git") def object_find_by_sha1_git(self, ids: List[Sha1Git]) -> Dict[Sha1Git, List[Dict]]: """Return the objects found with the given ids. Args: ids: a generator of sha1_gits Returns: A dict from id to the list of objects found for that id. Each object found is itself a dict with keys: - sha1_git: the input id - type: the type of object found """ ... @remote_api_endpoint("origin/get") def origin_get(self, origins: List[str]) -> Iterable[Optional[Origin]]: """Return origins. Args: origin: a list of urls to find Returns: the list of associated existing origin model objects. The unknown origins will be returned as None at the same index as the input. """ ... @remote_api_endpoint("origin/get_sha1") def origin_get_by_sha1(self, sha1s: List[bytes]) -> List[Optional[Dict[str, Any]]]: """Return origins, identified by the sha1 of their URLs. Args: sha1s: a list of sha1s Returns: List of origins dict whose sha1 of their url match, None otherwise. """ ... @remote_api_endpoint("origin/list") def origin_list( self, page_token: Optional[str] = None, limit: int = 100 ) -> PagedResult[Origin]: """Returns the list of origins Args: page_token: opaque token used for pagination. limit: the maximum number of results to return Returns: Page of Origin data model objects. if next_page_token is None, there is no longer data to retrieve. """ ... @remote_api_endpoint("origin/search") def origin_search( self, url_pattern: str, page_token: Optional[str] = None, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, ) -> PagedResult[Origin]: """Search for origins whose urls contain a provided string pattern or match a provided regular expression. The search is performed in a case insensitive way. Args: url_pattern: the string pattern to search for in origin urls page_token: opaque token used for pagination limit: the maximum number of found origins to return regexp: if True, consider the provided pattern as a regular expression and return origins whose urls match it with_visit: if True, filter out origins with no visit visit_types: Only origins having any of the provided visit types (e.g. git, svn, pypi) will be returned Yields: PagedResult of Origin """ ... @deprecated @remote_api_endpoint("origin/count") def origin_count( self, url_pattern: str, regexp: bool = False, with_visit: bool = False ) -> int: """Count origins whose urls contain a provided string pattern or match a provided regular expression. The pattern search in origin urls is performed in a case insensitive way. Args: url_pattern (str): the string pattern to search for in origin urls regexp (bool): if True, consider the provided pattern as a regular expression and return origins whose urls match it with_visit (bool): if True, filter out origins with no visit Returns: int: The number of origins matching the search criterion. """ ... @remote_api_endpoint("origin/add_multi") def origin_add(self, origins: List[Origin]) -> Dict[str, int]: """Add origins to the storage Args: origins: list of dictionaries representing the individual origins, with the following keys: - type: the origin type ('git', 'svn', 'deb', ...) - url (bytes): the url the origin points to Returns: Summary dict of keys with associated count as values origin:add: Count of object actually stored in db """ ... def stat_counters(self): """compute statistics about the number of tuples in various tables Returns: dict: a dictionary mapping textual labels (e.g., content) to integer values (e.g., the number of tuples in table content) """ ... def refresh_stat_counters(self): """Recomputes the statistics for `stat_counters`.""" ... @remote_api_endpoint("raw_extrinsic_metadata/add") def raw_extrinsic_metadata_add( self, metadata: List[RawExtrinsicMetadata], ) -> Dict[str, int]: """Add extrinsic metadata on objects (contents, directories, ...). The authority and fetcher must be known to the storage before using this endpoint. If there is already metadata for the same object, authority, fetcher, and at the same date; the new one will be either dropped or will replace the existing one (it is unspecified which one of these two behaviors happens). Args: metadata: iterable of RawExtrinsicMetadata objects to be inserted. """ ... @remote_api_endpoint("raw_extrinsic_metadata/get") def raw_extrinsic_metadata_get( self, target: ExtendedSWHID, authority: MetadataAuthority, after: Optional[datetime.datetime] = None, page_token: Optional[bytes] = None, limit: int = 1000, ) -> PagedResult[RawExtrinsicMetadata]: """Retrieve list of all raw_extrinsic_metadata entries targeting the id Args: target: the SWHID of the objects to find metadata on authority: a dict containing keys `type` and `url`. after: minimum discovery_date for a result to be returned page_token: opaque token, used to get the next page of results limit: maximum number of results to be returned Returns: PagedResult of RawExtrinsicMetadata """ ... @remote_api_endpoint("raw_extrinsic_metadata/get_by_ids") def raw_extrinsic_metadata_get_by_ids( self, ids: List[Sha1Git] ) -> List[RawExtrinsicMetadata]: """Retrieve list of raw_extrinsic_metadata entries of the given id (unlike raw_extrinsic_metadata_get, which returns metadata entries **targeting** the id) Args: ids: list of hashes of RawExtrinsicMetadata objects """ ... @remote_api_endpoint("raw_extrinsic_metadata/get_authorities") def raw_extrinsic_metadata_get_authorities( self, target: ExtendedSWHID ) -> List[MetadataAuthority]: """Returns all authorities that provided metadata on the given object.""" ... @remote_api_endpoint("metadata_fetcher/add") def metadata_fetcher_add(self, fetchers: List[MetadataFetcher],) -> Dict[str, int]: """Add new metadata fetchers to the storage. Their `name` and `version` together are unique identifiers of this fetcher; and `metadata` is an arbitrary dict of JSONable data with information about this fetcher, which must not be `None` (but may be empty). Args: fetchers: iterable of MetadataFetcher to be inserted """ ... @remote_api_endpoint("metadata_fetcher/get") def metadata_fetcher_get( self, name: str, version: str ) -> Optional[MetadataFetcher]: """Retrieve information about a fetcher Args: name: the name of the fetcher version: version of the fetcher Returns: a MetadataFetcher object (with a non-None metadata field) if it is known, else None. """ ... @remote_api_endpoint("metadata_authority/add") def metadata_authority_add( self, authorities: List[MetadataAuthority] ) -> Dict[str, int]: """Add new metadata authorities to the storage. Their `type` and `url` together are unique identifiers of this authority; and `metadata` is an arbitrary dict of JSONable data with information about this authority, which must not be `None` (but may be empty). Args: authorities: iterable of MetadataAuthority to be inserted """ ... @remote_api_endpoint("metadata_authority/get") def metadata_authority_get( self, type: MetadataAuthorityType, url: str ) -> Optional[MetadataAuthority]: """Retrieve information about an authority Args: type: one of "deposit_client", "forge", or "registry" url: unique URI identifying the authority Returns: a MetadataAuthority object (with a non-None metadata field) if it is known, else None. """ ... @remote_api_endpoint("clear/buffer") def clear_buffers(self, object_types: Sequence[str] = ()) -> None: """For backend storages (pg, storage, in-memory), this is a noop operation. For proxy storages (especially filter, buffer), this is an operation which cleans internal state. """ @remote_api_endpoint("flush") def flush(self, object_types: Sequence[str] = ()) -> Dict[str, int]: """For backend storages (pg, storage, in-memory), this is expected to be a noop operation. For proxy storages (especially buffer), this is expected to trigger actual writes to the backend. """ ... diff --git a/swh/storage/migrate_extrinsic_metadata.py b/swh/storage/migrate_extrinsic_metadata.py index d390c147..4d4dba5b 100644 --- a/swh/storage/migrate_extrinsic_metadata.py +++ b/swh/storage/migrate_extrinsic_metadata.py @@ -1,1208 +1,1260 @@ #!/usr/bin/env python3 # Copyright (C) 2020-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """This is an executable script to migrate extrinsic revision metadata from the revision table to the new extrinsic metadata storage. This is designed to be as conservative as possible, following this principle: for each revision the script reads (in "handle_row"), it will read some of the fields, write them directly to the metadata storage, and remove them. Then it checks all the remaining fields are in a hardcoded list of fields that are known not to require migration. This means that every field that isn't migrated was explicitly reviewed while writing this script. Additionally, this script contains many assertions to prevent false positives in its heuristics. """ import datetime import hashlib +import itertools import json import os import re import sys import time from typing import Any, Dict, Optional from urllib.error import HTTPError from urllib.parse import unquote, urlparse from urllib.request import urlopen import iso8601 import psycopg2 from swh.core.db import BaseDb from swh.model.hashutil import hash_to_hex from swh.model.identifiers import ( CoreSWHID, ExtendedObjectType, ExtendedSWHID, ObjectType, QualifiedSWHID, ) from swh.model.model import ( MetadataAuthority, MetadataAuthorityType, MetadataFetcher, RawExtrinsicMetadata, Sha1Git, ) from swh.storage import get_storage from swh.storage.algos.origin import iter_origin_visit_statuses, iter_origin_visits from swh.storage.algos.snapshot import snapshot_get_all_branches # XML namespaces and fields for metadata coming from the deposit: CODEMETA_NS = "https://doi.org/10.5063/SCHEMA/CODEMETA-2.0" ATOM_NS = "http://www.w3.org/2005/Atom" ATOM_KEYS = ["id", "author", "external_identifier", "title"] # columns of the revision table (of the storage DB) REVISION_COLS = [ "id", "directory", "date", "committer_date", "type", "message", "metadata", ] # columns of the tables of the deposit DB DEPOSIT_COLS = [ "deposit.id", "deposit.external_id", "deposit.swhid_context", "deposit.status", "deposit_request.metadata", "deposit_request.date", "deposit_client.provider_url", "deposit_collection.name", "auth_user.username", ] # Formats we write to the extrinsic metadata storage OLD_DEPOSIT_FORMAT = ( "sword-v2-atom-codemeta-v2-in-json-with-expanded-namespaces" # before february 2018 ) NEW_DEPOSIT_FORMAT = "sword-v2-atom-codemeta-v2-in-json" # after february 2018 GNU_FORMAT = "gnu-tree-json" NIXGUIX_FORMAT = "nixguix-sources-json" NPM_FORMAT = "replicate-npm-package-json" ORIGINAL_ARTIFACT_FORMAT = "original-artifacts-json" PYPI_FORMAT = "pypi-project-json" # Information about this script, for traceability FETCHER = MetadataFetcher( name="migrate-extrinsic-metadata-from-revisions", version="0.0.1", ) # Authorities that we got the metadata from AUTHORITIES = { "npmjs": MetadataAuthority( type=MetadataAuthorityType.FORGE, url="https://npmjs.com/", metadata={} ), "pypi": MetadataAuthority( type=MetadataAuthorityType.FORGE, url="https://pypi.org/", metadata={} ), "gnu": MetadataAuthority( type=MetadataAuthorityType.FORGE, url="https://ftp.gnu.org/", metadata={} ), "swh": MetadataAuthority( type=MetadataAuthorityType.REGISTRY, url="https://softwareheritage.org/", metadata={}, ), # for original_artifact (which are checksums computed by SWH) } # Regular expression for the format of revision messages written by the # deposit loader deposit_revision_message_re = re.compile( b"(?P[a-z-]*): " b"Deposit (?P[0-9]+) in collection (?P[a-z-]+).*" ) # not reliable, because PyPI allows arbitrary names def pypi_project_from_filename(filename): original_filename = filename if filename.endswith(".egg"): return None elif filename == "mongomotor-0.13.0.n.tar.gz": return "mongomotor" elif re.match(r"datahaven-rev[0-9]+\.tar\.gz", filename): return "datahaven" elif re.match(r"Dtls-[0-9]\.[0-9]\.[0-9]\.sdist_with_openssl\..*", filename): return "Dtls" elif re.match(r"(gae)?pytz-20[0-9][0-9][a-z]\.(tar\.gz|zip)", filename): return filename.split("-", 1)[0] elif filename.startswith(("powny-", "obedient.powny-",)): return filename.split("-")[0] elif filename.startswith("devpi-theme-16-"): return "devpi-theme-16" elif re.match("[^-]+-[0-9]+.tar.gz", filename): return filename.split("-")[0] elif filename == "ohai-1!0.tar.gz": return "ohai" elif filename == "collective.topicitemsevent-0.1dvl.tar.gz": return "collective.topicitemsevent" elif filename.startswith( ("SpiNNStorageHandlers-1!", "sPyNNakerExternalDevicesPlugin-1!") ): return filename.split("-")[0] elif filename.startswith("limnoria-201"): return "limnoria" elif filename.startswith("pytz-20"): return "pytz" elif filename.startswith("youtube_dl_server-alpha."): return "youtube_dl_server" elif filename == "json-extensions-b76bc7d.tar.gz": return "json-extensions" elif filename == "LitReview-0.6989ev.tar.gz": # typo of "dev" return "LitReview" elif filename.startswith("django_options-r"): return "django_options" elif filename == "Greater than, equal, or less Library-0.1.tar.gz": return "Greater-than-equal-or-less-Library" elif filename.startswith("upstart--main-"): return "upstart" elif filename == "duckduckpy0.1.tar.gz": return "duckduckpy" elif filename == "QUI for MPlayer snapshot_9-14-2011.zip": return "QUI-for-MPlayer" elif filename == "Eddy's Memory Game-1.0.zip": return "Eddy-s-Memory-Game" elif filename == "jekyll2nikola-0-0-1.tar.gz": return "jekyll2nikola" elif filename.startswith("ore.workflowed"): return "ore.workflowed" elif re.match("instancemanager-[0-9]*", filename): return "instancemanager" elif filename == "OrzMC_W&L-1.0.0.tar.gz": return "OrzMC-W-L" elif filename == "use0mk.tar.gz": return "use0mk" elif filename == "play-0-develop-1-gd67cd85.tar.gz": return "play" + elif filename.startswith("mosaic-nist-"): + return "mosaic-nist" + elif filename.startswith("pypops-"): + return "pypops" + elif filename.startswith("pdfcomparator-"): + return "pdfcomparator" + elif filename.startswith("LabJackPython-"): + return "LabJackPython" + elif filename == "MD2K: Cerebral Cortex-3.0.0.tar.gz": + return "cerebralcortex-kernel" + elif filename.startswith("LyMaker-0 (copy)"): + return "LyMaker" + elif filename.startswith("python-tplink-smarthome-"): + return "python-tplink-smarthome" + elif filename.startswith("jtt=tm-utils-"): + return "jtt-tm-utils" + elif filename == "atproject0.1.tar.gz": + return "atproject" + elif filename == "labm8.tar.gz": + return "labm8" + elif filename == "Bugs Everywhere (BEurtle fork)-1.5.0.1.-2012-07-16-.zip": + return "Bugs-Everywhere-BEurtle-fork" + filename = filename.replace(" ", "-") match = re.match( r"^(?P[a-z_.-]+)" # project name r"\.(tar\.gz|tar\.bz2|tgz|zip)$", # extension filename, re.I, ) if match: return match.group("project_name") # First try with a rather strict format, but that allows accidentally # matching the version as part of the package name match = re.match( r"^(?P[a-z0-9_.]+?([-_][a-z][a-z0-9.]+?)*?)" # project name r"-v?" r"([0-9]+!)?" # epoch r"[0-9_.]+([a-z]+[0-9]+)?" # "main" version r"([.-]?(alpha|beta|dev|post|pre|rc)(\.?[0-9]+)?)*" # development status r"([.-]?20[012][0-9]{5,9})?" # date r"([.-]g?[0-9a-f]+)?" # git commit r"([-+]py(thon)?(3k|[23](\.?[0-9]{1,2})?))?" # python version r"\.(tar\.gz|tar\.bz2|tgz|zip)$", # extension filename, re.I, ) if match: return match.group("project_name") # If that doesn't work, give up on trying to parse version suffixes, # and just find the first version-like occurrence in the file name match = re.match( r"^(?P[a-z0-9_.-]+?)" # project name r"[-_.]v?" r"([0-9]+!)?" # epoch r"(" # "main" version r"[0-9_]+\.[0-9_.]+([a-z]+[0-9]+)?" # classic version number r"|20[012][0-9]{5,9}" # date as integer r"|20[012][0-9]-[01][0-9]-[0-3][0-9]" # date as ISO 8601 r")" # end of "main" version r"[a-z]?(dev|pre)?" # direct version suffix r"([._-].*)?" # extra suffixes r"\.(tar\.gz|tar\.bz2|tgz|zip)$", # extension filename, re.I, ) if match: return match.group("project_name") # If that still doesn't work, give one last chance if there's only one # dash or underscore in the name match = re.match( r"^(?P[^_-]+)" # project name r"[_-][^_-]+" # version r"\.(tar\.gz|tar\.bz2|tgz|zip)$", # extension filename, ) assert match, original_filename return match.group("project_name") def pypi_origin_from_project_name(project_name: str) -> str: return f"https://pypi.org/project/{project_name}/" def pypi_origin_from_filename(storage, rev_id: bytes, filename: str) -> Optional[str]: project_name = pypi_project_from_filename(filename) origin = pypi_origin_from_project_name(project_name) # But unfortunately, the filename is user-provided, and doesn't # necessarily match the package name on pypi. Therefore, we need # to check it. if _check_revision_in_origin(storage, origin, rev_id): return origin # if the origin we guessed does not exist, query the PyPI API with the # project name we guessed. If only the capitalisation and dash/underscores # are wrong (by far the most common case), PyPI kindly corrects them. try: resp = urlopen(f"https://pypi.org/pypi/{project_name}/json/") except HTTPError as e: assert e.code == 404 # nope; PyPI couldn't correct the wrong project name return None assert resp.code == 200, resp.code project_name = json.load(resp)["info"]["name"] origin = pypi_origin_from_project_name(project_name) if _check_revision_in_origin(storage, origin, rev_id): return origin else: # The origin exists, but the revision does not belong in it. # This happens sometimes, as the filename we guessed the origin # from is user-provided. return None def cran_package_from_url(filename): match = re.match( r"^https://cran\.r-project\.org/src/contrib/" r"(?P[a-zA-Z0-9.]+)_[0-9.-]+(\.tar\.gz)?$", filename, ) assert match, filename return match.group("package_name") def npm_package_from_source_url(package_source_url): match = re.match( "^https://registry.npmjs.org/(?P.*)/-/[^/]+.tgz$", package_source_url, ) assert match, package_source_url return unquote(match.group("package_name")) def remove_atom_codemeta_metadata_with_xmlns(metadata): """Removes all known Atom and Codemeta metadata fields from the dict, assuming this is a dict generated by xmltodict without expanding namespaces. """ keys_to_remove = ATOM_KEYS + ["@xmlns", "@xmlns:codemeta"] for key in list(metadata): if key.startswith("codemeta:") or key in keys_to_remove: del metadata[key] def remove_atom_codemeta_metadata_without_xmlns(metadata): """Removes all known Atom and Codemeta metadata fields from the dict, assuming this is a dict generated by xmltodict with expanded namespaces. """ for key in list(metadata): if key.startswith(("{%s}" % ATOM_NS, "{%s}" % CODEMETA_NS)): del metadata[key] def _check_revision_in_origin(storage, origin, revision_id): seen_snapshots = set() # no need to visit them again seen_revisions = set() for visit in iter_origin_visits(storage, origin): for status in iter_origin_visit_statuses(storage, origin, visit.visit): if status.snapshot is None: continue if status.snapshot in seen_snapshots: continue seen_snapshots.add(status.snapshot) snapshot = snapshot_get_all_branches(storage, status.snapshot) for (branch_name, branch) in snapshot.branches.items(): if branch is None: continue # If it's the revision passed as argument, then it is indeed in the # origin if branch.target == revision_id: return True # Else, let's make sure the branch doesn't have any other revision # Get the revision at the top of the branch. if branch.target in seen_revisions: continue seen_revisions.add(branch.target) revision = storage.revision_get([branch.target])[0] if revision is None: # https://forge.softwareheritage.org/T997 continue # Check it doesn't have parents (else we would have to # recurse) assert revision.parents == (), "revision with parents" return False def debian_origins_from_row(row, storage): """Guesses a Debian origin from a row. May return an empty list if it cannot reliably guess it, but all results are guaranteed to be correct.""" filenames = [entry["filename"] for entry in row["metadata"]["original_artifact"]] package_names = {filename.split("_")[0] for filename in filenames} assert len(package_names) == 1, package_names (package_name,) = package_names candidate_origins = [ f"deb://Debian/packages/{package_name}", f"deb://Debian-Security/packages/{package_name}", f"http://snapshot.debian.org/package/{package_name}/", ] return [ origin for origin in candidate_origins if _check_revision_in_origin(storage, origin, row["id"]) ] # Cache of origins that are known to exist _origins = set() def assert_origin_exists(storage, origin): assert check_origin_exists(storage, origin), origin def check_origin_exists(storage, origin): return ( ( hashlib.sha1(origin.encode()).digest() in _origins # very fast or storage.origin_get([origin])[0] is not None # slow, but up to date ), origin, ) def load_metadata( storage, revision_id, directory_id, discovery_date: datetime.datetime, metadata: Dict[str, Any], format: str, authority: MetadataAuthority, origin: Optional[str], dry_run: bool, ): """Does the actual loading to swh-storage.""" directory_swhid = ExtendedSWHID( object_type=ExtendedObjectType.DIRECTORY, object_id=directory_id ) revision_swhid = CoreSWHID(object_type=ObjectType.REVISION, object_id=revision_id) obj = RawExtrinsicMetadata( target=directory_swhid, discovery_date=discovery_date, authority=authority, fetcher=FETCHER, format=format, metadata=json.dumps(metadata).encode(), origin=origin, revision=revision_swhid, ) if not dry_run: storage.raw_extrinsic_metadata_add([obj]) def handle_deposit_row( row, discovery_date: Optional[datetime.datetime], origin, storage, deposit_cur, dry_run: bool, ): """Loads metadata from the deposit database (which is more reliable as the metadata on the revision object, as some versions of the deposit loader were a bit lossy; and they used very different format for the field in the revision table). """ parsed_message = deposit_revision_message_re.match(row["message"]) assert parsed_message is not None, row["message"] deposit_id = int(parsed_message.group("deposit_id")) collection = parsed_message.group("collection").decode() client_name = parsed_message.group("client").decode() deposit_cur.execute( f"SELECT {', '.join(DEPOSIT_COLS)} FROM deposit " f"INNER JOIN deposit_collection " f" ON (deposit.collection_id=deposit_collection.id) " f"INNER JOIN deposit_client ON (deposit.client_id=deposit_client.user_ptr_id) " f"INNER JOIN auth_user ON (deposit.client_id=auth_user.id) " f"INNER JOIN deposit_request ON (deposit.id=deposit_request.deposit_id) " f"WHERE deposit.id = %s", (deposit_id,), ) provider_urls = set() swhids = set() metadata_entries = [] dates = set() external_identifiers = set() for deposit_request_row in deposit_cur: deposit_request = dict(zip(DEPOSIT_COLS, deposit_request_row)) # Sanity checks to make sure we selected the right deposit assert deposit_request["deposit.id"] == deposit_id assert deposit_request["deposit_collection.name"] == collection, deposit_request if client_name != "": # Sometimes it's missing from the commit message assert deposit_request["auth_user.username"] == client_name # Date of the deposit request (either the initial request, of subsequent ones) date = deposit_request["deposit_request.date"] dates.add(date) - assert deposit_request["deposit.swhid_context"], deposit_request + if deposit_request["deposit.external_id"] == "hal-02355563": + # Failed deposit + swhids.add( + "swh:1:rev:9293f230baca9814490d4fff7ac53d487a20edb6" + ";origin=https://hal.archives-ouvertes.fr/hal-02355563" + ) + else: + assert deposit_request["deposit.swhid_context"], deposit_request + swhids.add(deposit_request["deposit.swhid_context"]) external_identifiers.add(deposit_request["deposit.external_id"]) - swhids.add(deposit_request["deposit.swhid_context"]) # Client of the deposit provider_urls.add(deposit_request["deposit_client.provider_url"]) metadata = deposit_request["deposit_request.metadata"] if metadata is not None: json.dumps(metadata).encode() # check it's valid if "@xmlns" in metadata: assert metadata["@xmlns"] == ATOM_NS assert metadata["@xmlns:codemeta"] in (CODEMETA_NS, [CODEMETA_NS]) format = NEW_DEPOSIT_FORMAT elif "{http://www.w3.org/2005/Atom}id" in metadata: assert ( "{https://doi.org/10.5063/SCHEMA/CODEMETA-2.0}author" in metadata or "{http://www.w3.org/2005/Atom}author" in metadata ) format = OLD_DEPOSIT_FORMAT else: # new format introduced in # https://forge.softwareheritage.org/D4065 # it's the same as the first case, but with the @xmlns # declarations stripped # Most of them should have the "id", but some revisions, # like 4d3890004fade1f4ec3bf7004a4af0c490605128, are missing # this field assert "id" in metadata or "title" in metadata assert "codemeta:author" in metadata format = NEW_DEPOSIT_FORMAT metadata_entries.append((date, format, metadata)) if discovery_date is None: discovery_date = max(dates) # Sanity checks to make sure deposit requests are consistent with each other assert len(metadata_entries) >= 1, deposit_id assert len(provider_urls) == 1, f"expected 1 provider url, got {provider_urls}" (provider_url,) = provider_urls assert len(swhids) == 1 (swhid,) = swhids assert ( len(external_identifiers) == 1 ), f"expected 1 external identifier, got {external_identifiers}" (external_identifier,) = external_identifiers # computed the origin from the external_identifier if we don't have one if origin is None: origin = f"{provider_url.strip('/')}/{external_identifier}" # explicit list of mistakes that happened in the past, but shouldn't # happen again: if origin == "https://hal.archives-ouvertes.fr/hal-01588781": # deposit id 75 origin = "https://inria.halpreprod.archives-ouvertes.fr/hal-01588781" elif origin == "https://hal.archives-ouvertes.fr/hal-01588782": # deposit id 76 origin = "https://inria.halpreprod.archives-ouvertes.fr/hal-01588782" elif origin == "https://hal.archives-ouvertes.fr/hal-01592430": # deposit id 143 origin = "https://hal-preprod.archives-ouvertes.fr/hal-01592430" elif origin == "https://hal.archives-ouvertes.fr/hal-01588927": origin = "https://inria.halpreprod.archives-ouvertes.fr/hal-01588927" elif origin == "https://hal.archives-ouvertes.fr/hal-01593875": # deposit id 175 origin = "https://hal-preprod.archives-ouvertes.fr/hal-01593875" elif deposit_id == 160: assert origin == "https://www.softwareheritage.org/je-suis-gpl", origin origin = "https://forge.softwareheritage.org/source/jesuisgpl/" elif origin == "https://hal.archives-ouvertes.fr/hal-01588942": # deposit id 90 origin = "https://inria.halpreprod.archives-ouvertes.fr/hal-01588942" elif origin == "https://hal.archives-ouvertes.fr/hal-01592499": # deposit id 162 origin = "https://hal-preprod.archives-ouvertes.fr/hal-01592499" elif origin == "https://hal.archives-ouvertes.fr/hal-01588935": # deposit id 89 origin = "https://hal-preprod.archives-ouvertes.fr/hal-01588935" assert_origin_exists(storage, origin) # check the origin we computed matches the one in the deposit db swhid_origin = QualifiedSWHID.from_string(swhid).origin if origin is not None: # explicit list of mistakes that happened in the past, but shouldn't # happen again: exceptions = [ ( # deposit id 229 "https://hal.archives-ouvertes.fr/hal-01243573", "https://hal-test.archives-ouvertes.fr/hal-01243573", ), ( # deposit id 199 "https://hal.archives-ouvertes.fr/hal-01243065", "https://hal-test.archives-ouvertes.fr/hal-01243065", ), ( # deposit id 164 "https://hal.archives-ouvertes.fr/hal-01593855", "https://hal-preprod.archives-ouvertes.fr/hal-01593855", ), ] if (origin, swhid_origin) not in exceptions: assert origin == swhid_origin, ( f"the origin we guessed from the deposit db or revision ({origin}) " f"doesn't match the one in the deposit db's SWHID ({swhid})" ) authority = MetadataAuthority( type=MetadataAuthorityType.DEPOSIT_CLIENT, url=provider_url, metadata={}, ) for (date, format, metadata) in metadata_entries: load_metadata( storage, row["id"], row["directory"], date, metadata, format, authority=authority, origin=origin, dry_run=dry_run, ) return (origin, discovery_date) def handle_row(row: Dict[str, Any], storage, deposit_cur, dry_run: bool): type_ = row["type"] # default date in case we can't find a better one discovery_date = row["date"] or row["committer_date"] metadata = row["metadata"] if metadata is None: return if type_ == "dsc": origin = None # it will be defined later, using debian_origins_from_row # TODO: the debian loader writes the changelog date as the revision's # author date and committer date. Instead, we should use the visit's date if "extrinsic" in metadata: extrinsic_files = metadata["extrinsic"]["raw"]["files"] for artifact_entry in metadata["original_artifact"]: extrinsic_file = extrinsic_files[artifact_entry["filename"]] for key in ("sha256",): assert artifact_entry["checksums"][key] == extrinsic_file[key] artifact_entry["url"] = extrinsic_file["uri"] del metadata["extrinsic"] elif type_ == "tar": provider = metadata.get("extrinsic", {}).get("provider") if provider is not None: # This is the format all the package loaders currently write, and # it is the easiest, thanks to the 'provider' and 'when' fields, # which have all the information we need to tell them easily # and generate accurate metadata discovery_date = iso8601.parse_date(metadata["extrinsic"]["when"]) # New versions of the loaders write the provider; use it. if provider.startswith("https://replicate.npmjs.com/"): # npm loader format 1 parsed_url = urlparse(provider) assert re.match("^/[^/]+/?$", parsed_url.path), parsed_url package_name = unquote(parsed_url.path.strip("/")) origin = "https://www.npmjs.com/package/" + package_name assert_origin_exists(storage, origin) load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["extrinsic"]["raw"], NPM_FORMAT, authority=AUTHORITIES["npmjs"], origin=origin, dry_run=dry_run, ) del metadata["extrinsic"] elif provider.startswith("https://pypi.org/"): # pypi loader format 1 match = re.match( "https://pypi.org/pypi/(?P.*)/json", provider ) assert match, f"unexpected provider URL format: {provider}" project_name = match.group("project_name") origin = f"https://pypi.org/project/{project_name}/" assert_origin_exists(storage, origin) load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["extrinsic"]["raw"], PYPI_FORMAT, authority=AUTHORITIES["pypi"], origin=origin, dry_run=dry_run, ) del metadata["extrinsic"] elif provider.startswith("https://cran.r-project.org/"): # cran loader provider = metadata["extrinsic"]["provider"] if provider.startswith("https://cran.r-project.org/package="): origin = metadata["extrinsic"]["provider"] else: package_name = cran_package_from_url(provider) origin = f"https://cran.r-project.org/package={package_name}" assert origin is not None # Ideally we should assert the origin exists, but we can't: # https://forge.softwareheritage.org/T2536 if ( hashlib.sha1(origin.encode()).digest() not in _origins and storage.origin_get([origin])[0] is None ): return raw_extrinsic_metadata = metadata["extrinsic"]["raw"] # this is actually intrinsic, ignore it if "version" in raw_extrinsic_metadata: del raw_extrinsic_metadata["version"] # Copy the URL to the original_artifacts metadata assert len(metadata["original_artifact"]) == 1 if "url" in metadata["original_artifact"][0]: assert ( metadata["original_artifact"][0]["url"] == raw_extrinsic_metadata["url"] ), row else: metadata["original_artifact"][0]["url"] = raw_extrinsic_metadata[ "url" ] del raw_extrinsic_metadata["url"] assert ( raw_extrinsic_metadata == {} ), f"Unexpected metadata keys: {list(raw_extrinsic_metadata)}" del metadata["extrinsic"] elif ( provider.startswith("https://nix-community.github.io/nixpkgs-swh/") or provider == "https://guix.gnu.org/sources.json" ): # nixguix loader origin = provider assert_origin_exists(storage, origin) authority = MetadataAuthority( type=MetadataAuthorityType.FORGE, url=provider, metadata={}, ) assert row["date"] is None # the nixguix loader does not write dates load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["extrinsic"]["raw"], NIXGUIX_FORMAT, authority=authority, origin=origin, dry_run=dry_run, ) del metadata["extrinsic"] elif provider.startswith("https://ftp.gnu.org/"): # archive loader format 1 origin = provider assert_origin_exists(storage, origin) assert len(metadata["original_artifact"]) == 1 metadata["original_artifact"][0]["url"] = metadata["extrinsic"]["raw"][ "url" ] # Remove duplicate keys of original_artifacts for key in ("url", "time", "length", "version", "filename"): del metadata["extrinsic"]["raw"][key] assert metadata["extrinsic"]["raw"] == {} del metadata["extrinsic"] elif provider.startswith("https://deposit.softwareheritage.org/"): origin = metadata["extrinsic"]["raw"]["origin"]["url"] assert_origin_exists(storage, origin) if "@xmlns" in metadata: assert metadata["@xmlns"] == ATOM_NS assert metadata["@xmlns:codemeta"] in (CODEMETA_NS, [CODEMETA_NS]) assert "intrinsic" not in metadata assert "extra_headers" not in metadata # deposit loader format 1 # in this case, the metadata seems to be both directly in metadata # and in metadata["extrinsic"]["raw"]["metadata"] (origin, discovery_date) = handle_deposit_row( row, discovery_date, origin, storage, deposit_cur, dry_run ) remove_atom_codemeta_metadata_with_xmlns(metadata) if "client" in metadata: del metadata["client"] del metadata["extrinsic"] else: # deposit loader format 2 actual_metadata = metadata["extrinsic"]["raw"]["origin_metadata"][ "metadata" ] if isinstance(actual_metadata, str): # new format introduced in # https://forge.softwareheritage.org/D4105 actual_metadata = json.loads(actual_metadata) if "@xmlns" in actual_metadata: assert actual_metadata["@xmlns"] == ATOM_NS assert actual_metadata["@xmlns:codemeta"] in ( CODEMETA_NS, [CODEMETA_NS], ) elif "{http://www.w3.org/2005/Atom}id" in actual_metadata: assert ( "{https://doi.org/10.5063/SCHEMA/CODEMETA-2.0}author" in actual_metadata ) else: # new format introduced in # https://forge.softwareheritage.org/D4065 # it's the same as the first case, but with the @xmlns # declarations stripped # Most of them should have the "id", but some revisions, # like 4d3890004fade1f4ec3bf7004a4af0c490605128, are missing # this field assert ( "id" in actual_metadata or "title" in actual_metadata or "atom:title" in actual_metadata ) assert "codemeta:author" in actual_metadata (origin, discovery_date) = handle_deposit_row( row, discovery_date, origin, storage, deposit_cur, dry_run ) del metadata["extrinsic"] else: assert False, f"unknown provider {provider}" # Older versions don't write the provider; use heuristics instead. elif ( metadata.get("package_source", {}) .get("url", "") .startswith("https://registry.npmjs.org/") ): # npm loader format 2 package_source_url = metadata["package_source"]["url"] package_name = npm_package_from_source_url(package_source_url) origin = "https://www.npmjs.com/package/" + package_name assert_origin_exists(storage, origin) load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["package"], NPM_FORMAT, authority=AUTHORITIES["npmjs"], origin=origin, dry_run=dry_run, ) del metadata["package"] assert "original_artifact" not in metadata # rebuild an "original_artifact"-like metadata dict from what we # can salvage of "package_source" package_source_metadata = metadata["package_source"] keep_keys = {"blake2s256", "filename", "sha1", "sha256", "url"} discard_keys = { "date", # is equal to the revision date "name", # was loaded above "version", # same } assert ( set(package_source_metadata) == keep_keys | discard_keys ), package_source_metadata # will be loaded below metadata["original_artifact"] = [ { "filename": package_source_metadata["filename"], "checksums": { "sha1": package_source_metadata["sha1"], "sha256": package_source_metadata["sha256"], "blake2s256": package_source_metadata["blake2s256"], }, "url": package_source_metadata["url"], } ] del metadata["package_source"] elif "@xmlns" in metadata: assert metadata["@xmlns:codemeta"] in (CODEMETA_NS, [CODEMETA_NS]) assert "intrinsic" not in metadata assert "extra_headers" not in metadata # deposit loader format 3 if row["message"] == b"swh: Deposit 159 in collection swh": # There is no deposit 159 in the deposit DB, for some reason assert ( hash_to_hex(row["id"]) == "8e9cee14a6ad39bca4347077b87fb5bbd8953bb1" ) return elif row["message"] == b"hal: Deposit 342 in collection hal": # They have status 'failed' and no swhid return origin = None # TODO discovery_date = None # TODO (origin, discovery_date) = handle_deposit_row( row, discovery_date, origin, storage, deposit_cur, dry_run ) remove_atom_codemeta_metadata_with_xmlns(metadata) if "client" in metadata: del metadata["client"] # found in the deposit db if "committer" in metadata: del metadata["committer"] # found on the revision object elif "{http://www.w3.org/2005/Atom}id" in metadata: assert ( "{https://doi.org/10.5063/SCHEMA/CODEMETA-2.0}author" in metadata or "{http://www.w3.org/2005/Atom}author" in metadata ) assert "intrinsic" not in metadata assert "extra_headers" not in metadata # deposit loader format 4 origin = None discovery_date = None # TODO (origin, discovery_date) = handle_deposit_row( row, discovery_date, origin, storage, deposit_cur, dry_run ) remove_atom_codemeta_metadata_without_xmlns(metadata) elif hash_to_hex(row["id"]) == "a86747d201ab8f8657d145df4376676d5e47cf9f": # deposit 91, is missing "{http://www.w3.org/2005/Atom}id" for some # reason, and has an invalid oririn return elif ( isinstance(metadata.get("original_artifact"), dict) and metadata["original_artifact"]["url"].startswith( "https://files.pythonhosted.org/" ) ) or ( isinstance(metadata.get("original_artifact"), list) and len(metadata.get("original_artifact")) == 1 and metadata["original_artifact"][0] .get("url", "") .startswith("https://files.pythonhosted.org/") ): if isinstance(metadata.get("original_artifact"), dict): metadata["original_artifact"] = [metadata["original_artifact"]] assert len(metadata["original_artifact"]) == 1 - origin = pypi_origin_from_filename( - storage, row["id"], metadata["original_artifact"][0]["filename"] - ) + version = metadata.get("project", {}).get("version") + filename = metadata["original_artifact"][0]["filename"] + if version: + origin = pypi_origin_from_project_name(filename.split("-" + version)[0]) + if not _check_revision_in_origin(storage, origin, row["id"]): + origin = None + else: + origin = None + if origin is None: + origin = pypi_origin_from_filename(storage, row["id"], filename) if "project" in metadata: # pypi loader format 2 load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["project"], PYPI_FORMAT, authority=AUTHORITIES["pypi"], origin=origin, dry_run=dry_run, ) del metadata["project"] else: assert set(metadata) == {"original_artifact"}, set(metadata) # pypi loader format 3 pass # nothing to do, there's no metadata elif row["message"] == b"synthetic revision message": assert isinstance(metadata["original_artifact"], list), metadata assert not any("url" in d for d in metadata["original_artifact"]) # archive loader format 2 origin = None elif deposit_revision_message_re.match(row["message"]): # deposit without metadata in the revision assert set(metadata) == {"original_artifact"}, metadata origin = None # TODO discovery_date = None (origin, discovery_date) = handle_deposit_row( row, discovery_date, origin, storage, deposit_cur, dry_run ) else: assert False, f"Unable to detect type of metadata for row: {row}" # Ignore common intrinsic metadata keys for key in ("intrinsic", "extra_headers"): if key in metadata: del metadata[key] # Ignore loader-specific intrinsic metadata keys if type_ == "hg": del metadata["node"] elif type_ == "dsc": if "package_info" in metadata: del metadata["package_info"] if "original_artifact" in metadata: for original_artifact in metadata["original_artifact"]: # Rename keys to the expected format of original-artifacts-json. rename_keys = [ ("name", "filename"), # eg. from old Debian loader ("size", "length"), # eg. from old PyPI loader ] for (old_name, new_name) in rename_keys: if old_name in original_artifact: assert new_name not in original_artifact original_artifact[new_name] = original_artifact.pop(old_name) # Move the checksums to their own subdict, which is the expected format # of original-artifacts-json. if "sha1" in original_artifact: assert "checksums" not in original_artifact original_artifact["checksums"] = {} for key in ("sha1", "sha256", "sha1_git", "blake2s256"): if key in original_artifact: original_artifact["checksums"][key] = original_artifact.pop(key) if "date" in original_artifact: # The information comes from the package repository rather than SWH, # so it shouldn't be in the 'original-artifacts' metadata # (which has SWH as authority). # Moreover, it's not a very useful information, so let's just drop it. del original_artifact["date"] allowed_keys = { "checksums", "filename", "length", "url", "archive_type", } assert set(original_artifact) <= allowed_keys, set(original_artifact) if type_ == "dsc": assert origin is None origins = debian_origins_from_row(row, storage) if not origins: - print("Missing Debian origin for revision: {hash_to_hex(row['id'])}") + print(f"Missing Debian origin for revision: {hash_to_hex(row['id'])}") else: origins = [origin] for origin in origins: load_metadata( storage, row["id"], row["directory"], discovery_date, metadata["original_artifact"], ORIGINAL_ARTIFACT_FORMAT, authority=AUTHORITIES["swh"], origin=origin, dry_run=dry_run, ) del metadata["original_artifact"] assert metadata == {}, ( f"remaining metadata keys for {row['id'].hex()} (type: {row['type']}): " f"{metadata}" ) def create_fetchers(db): with db.cursor() as cur: cur.execute( """ INSERT INTO metadata_fetcher (name, version, metadata) VALUES (%s, %s, %s) ON CONFLICT DO NOTHING """, (FETCHER.name, FETCHER.version, FETCHER.metadata), ) def iter_revision_rows(storage_dbconn: str, first_id: Sha1Git): after_id = first_id failures = 0 while True: try: storage_db = BaseDb.connect(storage_dbconn) with storage_db.cursor() as cur: while True: cur.execute( f"SELECT {', '.join(REVISION_COLS)} FROM revision " - f"WHERE id > %s AND metadata IS NOT NULL AND type != 'git'" + f"WHERE id >= %s AND metadata IS NOT NULL AND type != 'git'" f"ORDER BY id LIMIT 1000", (after_id,), ) new_rows = 0 for row in cur: new_rows += 1 row_d = dict(zip(REVISION_COLS, row)) yield row_d after_id = row_d["id"] if new_rows == 0: return except psycopg2.OperationalError as e: print(e) # most likely a temporary error, try again if failures >= 60: raise else: time.sleep(60) failures += 1 -def main(storage_dbconn, storage_url, deposit_dbconn, first_id, dry_run): +def main(storage_dbconn, storage_url, deposit_dbconn, first_id, limit, dry_run): storage_db = BaseDb.connect(storage_dbconn) deposit_db = BaseDb.connect(deposit_dbconn) storage = get_storage( "pipeline", steps=[ {"cls": "retry"}, { "cls": "postgresql", "db": storage_dbconn, "objstorage": {"cls": "memory", "args": {}}, }, ], ) if not dry_run: create_fetchers(storage_db) # Not creating authorities, as the loaders are presumably already running # and created them already. # This also helps make sure this script doesn't accidentally create # authorities that differ from what the loaders use. total_rows = 0 with deposit_db.cursor() as deposit_cur: - for row in iter_revision_rows(storage_dbconn, first_id): + rows = iter_revision_rows(storage_dbconn, first_id) + if limit is not None: + rows = itertools.islice(rows, limit) + for row in rows: handle_row(row, storage, deposit_cur, dry_run) total_rows += 1 if total_rows % 1000 == 0: percents = ( int.from_bytes(row["id"][0:4], byteorder="big") * 100 / (1 << 32) ) print( f"Processed {total_rows/1000000.:.2f}M rows " f"(~{percents:.1f}%, last revision: {row['id'].hex()})" ) if __name__ == "__main__": if len(sys.argv) == 4: (_, storage_dbconn, storage_url, deposit_dbconn) = sys.argv first_id = "00" * 20 elif len(sys.argv) == 5: (_, storage_dbconn, storage_url, deposit_dbconn, first_id) = sys.argv + limit = None + elif len(sys.argv) == 6: + (_, storage_dbconn, storage_url, deposit_dbconn, first_id, limit_str) = sys.argv + limit = int(limit_str) else: print( f"Syntax: {sys.argv[0]} " - f" []" + f" [ [limit]]" ) exit(1) if os.path.isfile("./origins.txt"): # You can generate this file with: # psql service=swh-replica \ # -c "\copy (select digest(url, 'sha1') from origin) to stdout" \ # | pv -l > origins.txt print("Loading origins...") with open("./origins.txt") as fd: for line in fd: digest = line.strip()[3:] _origins.add(bytes.fromhex(digest)) print("Done loading origins.") - main(storage_dbconn, storage_url, deposit_dbconn, bytes.fromhex(first_id), True) + main( + storage_dbconn, + storage_url, + deposit_dbconn, + bytes.fromhex(first_id), + limit, + True, + ) diff --git a/swh/storage/postgresql/db.py b/swh/storage/postgresql/db.py index f0821da0..86fbdaa4 100644 --- a/swh/storage/postgresql/db.py +++ b/swh/storage/postgresql/db.py @@ -1,1473 +1,1493 @@ -# Copyright (C) 2015-2020 The Software Heritage developers +# Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import logging import random import select from typing import Any, Dict, Iterable, List, Optional, Tuple from swh.core.db import BaseDb from swh.core.db.db_utils import execute_values_generator from swh.core.db.db_utils import jsonize as _jsonize from swh.core.db.db_utils import stored_procedure from swh.model.hashutil import DEFAULT_ALGORITHMS from swh.model.identifiers import ObjectType from swh.model.model import SHA1_SIZE, OriginVisit, OriginVisitStatus, Sha1Git from swh.storage.interface import ListOrder logger = logging.getLogger(__name__) def jsonize(d): return _jsonize(dict(d) if d is not None else None) class Db(BaseDb): """Proxy to the SWH DB, with wrappers around stored procedures """ current_version = 177 def mktemp_dir_entry(self, entry_type, cur=None): self._cursor(cur).execute( "SELECT swh_mktemp_dir_entry(%s)", (("directory_entry_%s" % entry_type),) ) @stored_procedure("swh_mktemp_revision") def mktemp_revision(self, cur=None): pass @stored_procedure("swh_mktemp_release") def mktemp_release(self, cur=None): pass @stored_procedure("swh_mktemp_snapshot_branch") def mktemp_snapshot_branch(self, cur=None): pass def register_listener(self, notify_queue, cur=None): """Register a listener for NOTIFY queue `notify_queue`""" self._cursor(cur).execute("LISTEN %s" % notify_queue) def listen_notifies(self, timeout): """Listen to notifications for `timeout` seconds""" if select.select([self.conn], [], [], timeout) == ([], [], []): return else: self.conn.poll() while self.conn.notifies: yield self.conn.notifies.pop(0) @stored_procedure("swh_content_add") def content_add_from_temp(self, cur=None): pass @stored_procedure("swh_directory_add") def directory_add_from_temp(self, cur=None): pass @stored_procedure("swh_skipped_content_add") def skipped_content_add_from_temp(self, cur=None): pass @stored_procedure("swh_revision_add") def revision_add_from_temp(self, cur=None): pass @stored_procedure("swh_extid_add") def extid_add_from_temp(self, cur=None): pass @stored_procedure("swh_release_add") def release_add_from_temp(self, cur=None): pass def content_update_from_temp(self, keys_to_update, cur=None): cur = self._cursor(cur) cur.execute( """select swh_content_update(ARRAY[%s] :: text[])""" % keys_to_update ) content_get_metadata_keys = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "status", ] content_add_keys = content_get_metadata_keys + ["ctime"] skipped_content_keys = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "reason", "status", "origin", ] def content_get_metadata_from_hashes( self, hashes: List[bytes], algo: str, cur=None ): cur = self._cursor(cur) assert algo in DEFAULT_ALGORITHMS query = f""" select {", ".join(self.content_get_metadata_keys)} from (values %s) as t (hash) inner join content on (content.{algo}=hash) """ yield from execute_values_generator( cur, query, ((hash_,) for hash_ in hashes), ) def content_get_range(self, start, end, limit=None, cur=None): """Retrieve contents within range [start, end]. """ cur = self._cursor(cur) query = """select %s from content where %%s <= sha1 and sha1 <= %%s order by sha1 limit %%s""" % ", ".join( self.content_get_metadata_keys ) cur.execute(query, (start, end, limit)) yield from cur content_hash_keys = ["sha1", "sha1_git", "sha256", "blake2s256"] def content_missing_from_list(self, contents, cur=None): cur = self._cursor(cur) keys = ", ".join(self.content_hash_keys) equality = " AND ".join( ("t.%s = c.%s" % (key, key)) for key in self.content_hash_keys ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(%s) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE %s ) """ % (keys, keys, equality), (tuple(c[key] for key in self.content_hash_keys) for c in contents), ) def content_missing_per_sha1(self, sha1s, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT t.sha1 FROM (VALUES %s) AS t(sha1) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE c.sha1 = t.sha1 )""", ((sha1,) for sha1 in sha1s), ) def content_missing_per_sha1_git(self, contents, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT t.sha1_git FROM (VALUES %s) AS t(sha1_git) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE c.sha1_git = t.sha1_git )""", ((sha1,) for sha1 in contents), ) def skipped_content_missing(self, contents, cur=None): if not contents: return [] cur = self._cursor(cur) query = """SELECT * FROM (VALUES %s) AS t (%s) WHERE not exists (SELECT 1 FROM skipped_content s WHERE s.sha1 is not distinct from t.sha1::sha1 and s.sha1_git is not distinct from t.sha1_git::sha1 and s.sha256 is not distinct from t.sha256::bytea);""" % ( (", ".join("%s" for _ in contents)), ", ".join(self.content_hash_keys), ) cur.execute( query, [tuple(cont[key] for key in self.content_hash_keys) for cont in contents], ) yield from cur def snapshot_exists(self, snapshot_id, cur=None): """Check whether a snapshot with the given id exists""" cur = self._cursor(cur) cur.execute("""SELECT 1 FROM snapshot where id=%s""", (snapshot_id,)) return bool(cur.fetchone()) def snapshot_missing_from_list(self, snapshots, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM snapshot d WHERE d.id = t.id ) """, ((id,) for id in snapshots), ) def snapshot_add(self, snapshot_id, cur=None): """Add a snapshot from the temporary table""" cur = self._cursor(cur) cur.execute("""SELECT swh_snapshot_add(%s)""", (snapshot_id,)) snapshot_count_cols = ["target_type", "count"] def snapshot_count_branches( self, snapshot_id, branch_name_exclude_prefix=None, cur=None, ): cur = self._cursor(cur) query = """\ SELECT %s FROM swh_snapshot_count_branches(%%s, %%s) """ % ", ".join( self.snapshot_count_cols ) cur.execute(query, (snapshot_id, branch_name_exclude_prefix)) yield from cur snapshot_get_cols = ["snapshot_id", "name", "target", "target_type"] def snapshot_get_by_id( self, snapshot_id, branches_from=b"", branches_count=None, target_types=None, branch_name_include_substring=None, branch_name_exclude_prefix=None, cur=None, ): cur = self._cursor(cur) query = """\ SELECT %s FROM swh_snapshot_get_by_id(%%s, %%s, %%s, %%s :: snapshot_target[], %%s, %%s) """ % ", ".join( self.snapshot_get_cols ) cur.execute( query, ( snapshot_id, branches_from, branches_count, target_types, branch_name_include_substring, branch_name_exclude_prefix, ), ) yield from cur def snapshot_get_random(self, cur=None): return self._get_random_row_from_table("snapshot", ["id"], "id", cur) content_find_cols = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "ctime", "status", ] def content_find( self, sha1: Optional[bytes] = None, sha1_git: Optional[bytes] = None, sha256: Optional[bytes] = None, blake2s256: Optional[bytes] = None, cur=None, ): """Find the content optionally on a combination of the following checksums sha1, sha1_git, sha256 or blake2s256. Args: sha1: sha1 content git_sha1: the sha1 computed `a la git` sha1 of the content sha256: sha256 content blake2s256: blake2s256 content Returns: The tuple (sha1, sha1_git, sha256, blake2s256) if found or None. """ cur = self._cursor(cur) checksum_dict = { "sha1": sha1, "sha1_git": sha1_git, "sha256": sha256, "blake2s256": blake2s256, } query_parts = [f"SELECT {','.join(self.content_find_cols)} FROM content WHERE "] query_params = [] where_parts = [] # Adds only those keys which have values exist for algorithm in checksum_dict: if checksum_dict[algorithm] is not None: where_parts.append(f"{algorithm} = %s") query_params.append(checksum_dict[algorithm]) query_parts.append(" AND ".join(where_parts)) query = "\n".join(query_parts) cur.execute(query, query_params) content = cur.fetchall() return content def content_get_random(self, cur=None): return self._get_random_row_from_table("content", ["sha1_git"], "sha1_git", cur) def directory_missing_from_list(self, directories, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM directory d WHERE d.id = t.id ) """, ((id,) for id in directories), ) directory_ls_cols = [ "dir_id", "type", "target", "name", "perms", "status", "sha1", "sha1_git", "sha256", "length", ] def directory_walk_one(self, directory, cur=None): cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_directory_walk_one(%%s)" % cols cur.execute(query, (directory,)) yield from cur def directory_walk(self, directory, cur=None): cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_directory_walk(%%s)" % cols cur.execute(query, (directory,)) yield from cur def directory_entry_get_by_path(self, directory, paths, cur=None): """Retrieve a directory entry by path. """ cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_find_directory_entry_by_path(%%s, %%s)" % cols cur.execute(query, (directory, paths)) data = cur.fetchone() if set(data) == {None}: return None return data directory_get_entries_cols = ["type", "target", "name", "perms"] def directory_get_entries(self, directory: Sha1Git, cur=None) -> List[Tuple]: cur = self._cursor(cur) cur.execute( "SELECT * FROM swh_directory_get_entries(%s::sha1_git)", (directory,) ) return list(cur) def directory_get_random(self, cur=None): return self._get_random_row_from_table("directory", ["id"], "id", cur) def revision_missing_from_list(self, revisions, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM revision r WHERE r.id = t.id ) """, ((id,) for id in revisions), ) revision_add_cols = [ "id", "date", "date_offset", "date_neg_utc_offset", "committer_date", "committer_date_offset", "committer_date_neg_utc_offset", "type", "directory", "message", "author_fullname", "author_name", "author_email", "committer_fullname", "committer_name", "committer_email", "metadata", "synthetic", "extra_headers", ] revision_get_cols = revision_add_cols + ["parents"] def origin_visit_add(self, origin, ts, type, cur=None): """Add a new origin_visit for origin origin at timestamp ts. Args: origin: origin concerned by the visit ts: the date of the visit type: type of loader for the visit Returns: The new visit index step for that origin """ cur = self._cursor(cur) self._cursor(cur).execute( "SELECT swh_origin_visit_add(%s, %s, %s)", (origin, ts, type) ) return cur.fetchone()[0] origin_visit_status_cols = [ "origin", "visit", "date", "type", "status", "snapshot", "metadata", ] def origin_visit_status_add( self, visit_status: OriginVisitStatus, cur=None ) -> None: """Add new origin visit status """ assert self.origin_visit_status_cols[0] == "origin" assert self.origin_visit_status_cols[-1] == "metadata" cols = self.origin_visit_status_cols[1:-1] cur = self._cursor(cur) cur.execute( f"WITH origin_id as (select id from origin where url=%s) " f"INSERT INTO origin_visit_status " f"(origin, {', '.join(cols)}, metadata) " f"VALUES ((select id from origin_id), " f"{', '.join(['%s']*len(cols))}, %s) " f"ON CONFLICT (origin, visit, date) do nothing", [visit_status.origin] + [getattr(visit_status, key) for key in cols] + [jsonize(visit_status.metadata)], ) origin_visit_cols = ["origin", "visit", "date", "type"] def origin_visit_add_with_id(self, origin_visit: OriginVisit, cur=None) -> None: """Insert origin visit when id are already set """ ov = origin_visit assert ov.visit is not None cur = self._cursor(cur) query = """INSERT INTO origin_visit ({cols}) VALUES ((select id from origin where url=%s), {values}) ON CONFLICT (origin, visit) DO NOTHING""".format( cols=", ".join(self.origin_visit_cols), values=", ".join("%s" for col in self.origin_visit_cols[1:]), ) cur.execute(query, (ov.origin, ov.visit, ov.date, ov.type)) origin_visit_get_cols = [ "origin", "visit", "date", "type", "status", "metadata", "snapshot", ] origin_visit_select_cols = [ "o.url AS origin", "ov.visit", "ov.date", "ov.type AS type", "ovs.status", - "ovs.metadata", "ovs.snapshot", + "ovs.metadata", ] origin_visit_status_select_cols = [ "o.url AS origin", "ovs.visit", "ovs.date", "ovs.type AS type", "ovs.status", "ovs.snapshot", "ovs.metadata", ] def _make_origin_visit_status( self, row: Optional[Tuple[Any]] ) -> Optional[Dict[str, Any]]: """Make an origin_visit_status dict out of a row """ if not row: return None return dict(zip(self.origin_visit_status_cols, row)) def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, cur=None, ) -> Optional[Dict[str, Any]]: """Given an origin visit id, return its latest origin_visit_status """ cur = self._cursor(cur) query_parts = [ "SELECT %s" % ", ".join(self.origin_visit_status_select_cols), "FROM origin_visit_status ovs ", "INNER JOIN origin o ON o.id = ovs.origin", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin_url] query_parts.append("AND ovs.visit = %s") query_params.append(visit) if require_snapshot: query_parts.append("AND ovs.snapshot is not null") if allowed_statuses: query_parts.append("AND ovs.status IN %s") query_params.append(tuple(allowed_statuses)) query_parts.append("ORDER BY ovs.date DESC LIMIT 1") query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) row = cur.fetchone() return self._make_origin_visit_status(row) def origin_visit_status_get_range( self, origin: str, visit: int, date_from: Optional[datetime.datetime], order: ListOrder, limit: int, cur=None, ): """Retrieve visit_status rows for visit (origin, visit) in a paginated way. """ cur = self._cursor(cur) query_parts = [ f"SELECT {', '.join(self.origin_visit_status_select_cols)} " "FROM origin_visit_status ovs ", "INNER JOIN origin o ON o.id = ovs.origin ", ] query_parts.append("WHERE o.url = %s AND ovs.visit = %s ") query_params: List[Any] = [origin, visit] if date_from is not None: op_comparison = ">=" if order == ListOrder.ASC else "<=" query_parts.append(f"and ovs.date {op_comparison} %s ") query_params.append(date_from) if order == ListOrder.ASC: query_parts.append("ORDER BY ovs.date ASC ") elif order == ListOrder.DESC: query_parts.append("ORDER BY ovs.date DESC ") else: assert False query_parts.append("LIMIT %s") query_params.append(limit) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) yield from cur def origin_visit_get_range( self, origin: str, visit_from: int, order: ListOrder, limit: int, cur=None, ): cur = self._cursor(cur) origin_visit_cols = ["o.url as origin", "ov.visit", "ov.date", "ov.type"] query_parts = [ f"SELECT {', '.join(origin_visit_cols)} FROM origin_visit ov ", "INNER JOIN origin o ON o.id = ov.origin ", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin] if visit_from > 0: op_comparison = ">" if order == ListOrder.ASC else "<" query_parts.append(f"and ov.visit {op_comparison} %s") query_params.append(visit_from) if order == ListOrder.ASC: query_parts.append("ORDER BY ov.visit ASC") elif order == ListOrder.DESC: query_parts.append("ORDER BY ov.visit DESC") query_parts.append("LIMIT %s") query_params.append(limit) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) yield from cur def origin_visit_get(self, origin_id, visit_id, cur=None): """Retrieve information on visit visit_id of origin origin_id. Args: origin_id: the origin concerned visit_id: The visit step for that origin Returns: The origin_visit information """ cur = self._cursor(cur) query = """\ SELECT %s FROM origin_visit ov INNER JOIN origin o ON o.id = ov.origin INNER JOIN origin_visit_status ovs USING (origin, visit) WHERE o.url = %%s AND ov.visit = %%s ORDER BY ovs.date DESC LIMIT 1 """ % ( ", ".join(self.origin_visit_select_cols) ) cur.execute(query, (origin_id, visit_id)) r = cur.fetchall() if not r: return None return r[0] def origin_visit_find_by_date(self, origin, visit_date, cur=None): cur = self._cursor(cur) cur.execute( "SELECT * FROM swh_visit_find_by_date(%s, %s)", (origin, visit_date) ) rows = cur.fetchall() if rows: visit = dict(zip(self.origin_visit_get_cols, rows[0])) visit["origin"] = origin return visit def origin_visit_exists(self, origin_id, visit_id, cur=None): """Check whether an origin visit with the given ids exists""" cur = self._cursor(cur) query = "SELECT 1 FROM origin_visit where origin = %s AND visit = %s" cur.execute(query, (origin_id, visit_id)) return bool(cur.fetchone()) def origin_visit_get_latest( self, origin_id: str, type: Optional[str], allowed_statuses: Optional[Iterable[str]], require_snapshot: bool, cur=None, ): """Retrieve the most recent origin_visit of the given origin, with optional filters. Args: origin_id: the origin concerned type: Optional visit type to filter on allowed_statuses: the visit statuses allowed for the returned visit require_snapshot (bool): If True, only a visit with a known snapshot will be returned. Returns: The origin_visit information, or None if no visit matches. """ cur = self._cursor(cur) query_parts = [ "SELECT %s" % ", ".join(self.origin_visit_select_cols), "FROM origin_visit ov ", "INNER JOIN origin o ON o.id = ov.origin", "INNER JOIN origin_visit_status ovs USING (origin, visit)", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin_id] if type is not None: query_parts.append("AND ov.type = %s") query_params.append(type) if require_snapshot: query_parts.append("AND ovs.snapshot is not null") if allowed_statuses: query_parts.append("AND ovs.status IN %s") query_params.append(tuple(allowed_statuses)) query_parts.append( "ORDER BY ov.date DESC, ov.visit DESC, ovs.date DESC LIMIT 1" ) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) r = cur.fetchone() if not r: return None return r def origin_visit_get_random(self, type, cur=None): """Randomly select one origin visit that was full and in the last 3 months """ cur = self._cursor(cur) columns = ",".join(self.origin_visit_select_cols) query = f"""select {columns} from origin_visit ov inner join origin o on ov.origin=o.id inner join origin_visit_status ovs using (origin, visit) where ovs.status='full' and ov.type=%s and ov.date > now() - '3 months'::interval and random() < 0.1 limit 1 """ cur.execute(query, (type,)) return cur.fetchone() @staticmethod def mangle_query_key(key, main_table): if key == "id": return "t.id" if key == "parents": return """ ARRAY( SELECT rh.parent_id::bytea FROM revision_history rh WHERE rh.id = t.id ORDER BY rh.parent_rank )""" if "_" not in key: return "%s.%s" % (main_table, key) head, tail = key.split("_", 1) if head in ("author", "committer") and tail in ( "name", "email", "id", "fullname", ): return "%s.%s" % (head, tail) return "%s.%s" % (main_table, key) def revision_get_from_list(self, revisions, cur=None): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "revision") for k in self.revision_get_cols ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(sortkey, id) LEFT JOIN revision ON t.id = revision.id LEFT JOIN person author ON revision.author = author.id LEFT JOIN person committer ON revision.committer = committer.id ORDER BY sortkey """ % query_keys, ((sortkey, id) for sortkey, id in enumerate(revisions)), ) extid_cols = ["extid", "extid_version", "extid_type", "target", "target_type"] - def extid_get_from_extid_list(self, extid_type, ids, cur=None): + def extid_get_from_extid_list( + self, extid_type: str, ids: List[bytes], version: Optional[int] = None, cur=None + ): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "extid") for k in self.extid_cols ) - sql = """ - SELECT %s - FROM (VALUES %%s) as t(sortkey, extid, extid_type) + filter_query = "" + if version is not None: + filter_query = cur.mogrify( + f"WHERE extid_version={version}", (version,) + ).decode() + + sql = f""" + SELECT {query_keys} + FROM (VALUES %s) as t(sortkey, extid, extid_type) LEFT JOIN extid USING (extid, extid_type) + {filter_query} ORDER BY sortkey - """ % ( - query_keys, - ) + """ yield from execute_values_generator( cur, sql, (((sortkey, extid, extid_type) for sortkey, extid in enumerate(ids))), ) - def extid_get_from_swhid_list(self, target_type, ids, cur=None): + def extid_get_from_swhid_list( + self, + target_type: str, + ids: List[bytes], + extid_version: Optional[int] = None, + extid_type: Optional[str] = None, + cur=None, + ): cur = self._cursor(cur) target_type = ObjectType( target_type ).name.lower() # aka "rev" -> "revision", ... query_keys = ", ".join( self.mangle_query_key(k, "extid") for k in self.extid_cols ) - sql = """ - SELECT %s - FROM (VALUES %%s) as t(sortkey, target, target_type) + filter_query = "" + if extid_version is not None and extid_type is not None: + filter_query = cur.mogrify( + "WHERE extid_version=%s AND extid_type=%s", (extid_version, extid_type,) + ).decode() + + sql = f""" + SELECT {query_keys} + FROM (VALUES %s) as t(sortkey, target, target_type) LEFT JOIN extid USING (target, target_type) + {filter_query} ORDER BY sortkey - """ % ( - query_keys, - ) + """ + yield from execute_values_generator( cur, sql, (((sortkey, target, target_type) for sortkey, target in enumerate(ids))), template=b"(%s,%s,%s::object_type)", ) def revision_log(self, root_revisions, limit=None, cur=None): cur = self._cursor(cur) query = """SELECT %s FROM swh_revision_log(%%s, %%s) """ % ", ".join( self.revision_get_cols ) cur.execute(query, (root_revisions, limit)) yield from cur revision_shortlog_cols = ["id", "parents"] def revision_shortlog(self, root_revisions, limit=None, cur=None): cur = self._cursor(cur) query = """SELECT %s FROM swh_revision_list(%%s, %%s) """ % ", ".join( self.revision_shortlog_cols ) cur.execute(query, (root_revisions, limit)) yield from cur def revision_get_random(self, cur=None): return self._get_random_row_from_table("revision", ["id"], "id", cur) def release_missing_from_list(self, releases, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM release r WHERE r.id = t.id ) """, ((id,) for id in releases), ) object_find_by_sha1_git_cols = ["sha1_git", "type"] def object_find_by_sha1_git(self, ids, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ WITH t (sha1_git) AS (VALUES %s), known_objects as (( select id as sha1_git, 'release'::object_type as type, object_id from release r where exists (select 1 from t where t.sha1_git = r.id) ) union all ( select id as sha1_git, 'revision'::object_type as type, object_id from revision r where exists (select 1 from t where t.sha1_git = r.id) ) union all ( select id as sha1_git, 'directory'::object_type as type, object_id from directory d where exists (select 1 from t where t.sha1_git = d.id) ) union all ( select sha1_git as sha1_git, 'content'::object_type as type, object_id from content c where exists (select 1 from t where t.sha1_git = c.sha1_git) )) select t.sha1_git as sha1_git, k.type from t left join known_objects k on t.sha1_git = k.sha1_git """, ((id,) for id in ids), ) def stat_counters(self, cur=None): cur = self._cursor(cur) cur.execute("SELECT * FROM swh_stat_counters()") yield from cur def origin_add(self, url, cur=None): """Insert a new origin and return the new identifier.""" insert = """INSERT INTO origin (url) values (%s) ON CONFLICT DO NOTHING """ cur.execute(insert, (url,)) return cur.rowcount origin_cols = ["url"] def origin_get_by_url(self, origins, cur=None): """Retrieve origin `(type, url)` from urls if found.""" cur = self._cursor(cur) query = """SELECT %s FROM (VALUES %%s) as t(url) LEFT JOIN origin ON t.url = origin.url """ % ",".join( "origin." + col for col in self.origin_cols ) yield from execute_values_generator(cur, query, ((url,) for url in origins)) def origin_get_by_sha1(self, sha1s, cur=None): """Retrieve origin urls from sha1s if found.""" cur = self._cursor(cur) query = """SELECT %s FROM (VALUES %%s) as t(sha1) LEFT JOIN origin ON t.sha1 = digest(origin.url, 'sha1') """ % ",".join( "origin." + col for col in self.origin_cols ) yield from execute_values_generator(cur, query, ((sha1,) for sha1 in sha1s)) def origin_id_get_by_url(self, origins, cur=None): """Retrieve origin `(type, url)` from urls if found.""" cur = self._cursor(cur) query = """SELECT id FROM (VALUES %s) as t(url) LEFT JOIN origin ON t.url = origin.url """ for row in execute_values_generator(cur, query, ((url,) for url in origins)): yield row[0] origin_get_range_cols = ["id", "url"] def origin_get_range(self, origin_from: int = 1, origin_count: int = 100, cur=None): """Retrieve ``origin_count`` origins whose ids are greater or equal than ``origin_from``. Origins are sorted by id before retrieving them. Args: origin_from: the minimum id of origins to retrieve origin_count: the maximum number of origins to retrieve """ cur = self._cursor(cur) query = """SELECT %s FROM origin WHERE id >= %%s ORDER BY id LIMIT %%s """ % ",".join( self.origin_get_range_cols ) cur.execute(query, (origin_from, origin_count)) yield from cur def _origin_query( self, url_pattern, count=False, offset=0, limit=50, regexp=False, with_visit=False, visit_types=None, cur=None, ): """ Method factorizing query creation for searching and counting origins. """ cur = self._cursor(cur) if count: origin_cols = "COUNT(*)" order_clause = "" else: origin_cols = ",".join(self.origin_cols) order_clause = "ORDER BY id" if not regexp: operator = "ILIKE" query_params = [f"%{url_pattern}%"] else: operator = "~*" query_params = [url_pattern] query = f""" WITH filtered_origins AS ( SELECT * FROM origin WHERE url {operator} %s {order_clause} ) SELECT {origin_cols} FROM filtered_origins AS o """ if with_visit or visit_types: visit_predicat = ( """ INNER JOIN origin_visit_status ovs USING (origin, visit) INNER JOIN snapshot ON ovs.snapshot=snapshot.id """ if with_visit else "" ) type_predicat = ( f"AND ov.type=any(ARRAY{visit_types})" if visit_types else "" ) query += f""" WHERE EXISTS ( SELECT 1 FROM origin_visit ov {visit_predicat} WHERE ov.origin=o.id {type_predicat} ) """ if not count: query += "OFFSET %s LIMIT %s" query_params.extend([offset, limit]) cur.execute(query, query_params) def origin_search( self, url_pattern: str, offset: int = 0, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, cur=None, ): """Search for origins whose urls contain a provided string pattern or match a provided regular expression. The search is performed in a case insensitive way. Args: url_pattern: the string pattern to search for in origin urls offset: number of found origins to skip before returning results limit: the maximum number of found origins to return regexp: if True, consider the provided pattern as a regular expression and returns origins whose urls match it with_visit: if True, filter out origins with no visit """ self._origin_query( url_pattern, offset=offset, limit=limit, regexp=regexp, with_visit=with_visit, visit_types=visit_types, cur=cur, ) yield from cur def origin_count(self, url_pattern, regexp=False, with_visit=False, cur=None): """Count origins whose urls contain a provided string pattern or match a provided regular expression. The pattern search in origin urls is performed in a case insensitive way. Args: url_pattern (str): the string pattern to search for in origin urls regexp (bool): if True, consider the provided pattern as a regular expression and returns origins whose urls match it with_visit (bool): if True, filter out origins with no visit """ self._origin_query( url_pattern, count=True, regexp=regexp, with_visit=with_visit, cur=cur ) return cur.fetchone()[0] release_add_cols = [ "id", "target", "target_type", "date", "date_offset", "date_neg_utc_offset", "name", "comment", "synthetic", "author_fullname", "author_name", "author_email", ] release_get_cols = release_add_cols def release_get_from_list(self, releases, cur=None): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "release") for k in self.release_get_cols ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(sortkey, id) LEFT JOIN release ON t.id = release.id LEFT JOIN person author ON release.author = author.id ORDER BY sortkey """ % query_keys, ((sortkey, id) for sortkey, id in enumerate(releases)), ) def release_get_random(self, cur=None): return self._get_random_row_from_table("release", ["id"], "id", cur) _raw_extrinsic_metadata_context_cols = [ "origin", "visit", "snapshot", "release", "revision", "path", "directory", ] """The list of context columns for all artifact types.""" _raw_extrinsic_metadata_insert_cols = [ "id", "type", "target", "authority_id", "fetcher_id", "discovery_date", "format", "metadata", *_raw_extrinsic_metadata_context_cols, ] """List of columns of the raw_extrinsic_metadata table, used when writing metadata.""" _raw_extrinsic_metadata_insert_query = f""" INSERT INTO raw_extrinsic_metadata ({', '.join(_raw_extrinsic_metadata_insert_cols)}) VALUES ({', '.join('%s' for _ in _raw_extrinsic_metadata_insert_cols)}) ON CONFLICT (id) DO NOTHING """ raw_extrinsic_metadata_get_cols = [ "raw_extrinsic_metadata.target", "raw_extrinsic_metadata.type", "discovery_date", "metadata_authority.type", "metadata_authority.url", "metadata_fetcher.id", "metadata_fetcher.name", "metadata_fetcher.version", *_raw_extrinsic_metadata_context_cols, "format", "raw_extrinsic_metadata.metadata", ] """List of columns of the raw_extrinsic_metadata, metadata_authority, and metadata_fetcher tables, used when reading object metadata.""" _raw_extrinsic_metadata_select_query = f""" SELECT {', '.join(raw_extrinsic_metadata_get_cols)} FROM raw_extrinsic_metadata INNER JOIN metadata_authority ON (metadata_authority.id=authority_id) INNER JOIN metadata_fetcher ON (metadata_fetcher.id=fetcher_id) """ def raw_extrinsic_metadata_add( self, id: bytes, type: str, target: str, discovery_date: datetime.datetime, authority_id: int, fetcher_id: int, format: str, metadata: bytes, origin: Optional[str], visit: Optional[int], snapshot: Optional[str], release: Optional[str], revision: Optional[str], path: Optional[bytes], directory: Optional[str], cur, ): query = self._raw_extrinsic_metadata_insert_query args: Dict[str, Any] = dict( id=id, type=type, target=target, authority_id=authority_id, fetcher_id=fetcher_id, discovery_date=discovery_date, format=format, metadata=metadata, origin=origin, visit=visit, snapshot=snapshot, release=release, revision=revision, path=path, directory=directory, ) params = [args[col] for col in self._raw_extrinsic_metadata_insert_cols] cur.execute(query, params) def raw_extrinsic_metadata_get( self, target: str, authority_id: int, after_time: Optional[datetime.datetime], after_fetcher: Optional[int], limit: int, cur, ): query_parts = [self._raw_extrinsic_metadata_select_query] query_parts.append("WHERE raw_extrinsic_metadata.target=%s AND authority_id=%s") args = [target, authority_id] if after_fetcher is not None: assert after_time query_parts.append("AND (discovery_date, fetcher_id) > (%s, %s)") args.extend([after_time, after_fetcher]) elif after_time is not None: query_parts.append("AND discovery_date > %s") args.append(after_time) query_parts.append("ORDER BY discovery_date, fetcher_id") if limit: query_parts.append("LIMIT %s") args.append(limit) cur.execute(" ".join(query_parts), args) yield from cur def raw_extrinsic_metadata_get_by_ids(self, ids: List[Sha1Git], cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, self._raw_extrinsic_metadata_select_query + "INNER JOIN (VALUES %s) AS t(id) ON t.id = raw_extrinsic_metadata.id", [(id_,) for id_ in ids], ) def raw_extrinsic_metadata_get_authorities(self, id: str, cur=None): cur = self._cursor(cur) cur.execute( """ SELECT DISTINCT metadata_authority.type, metadata_authority.url FROM raw_extrinsic_metadata INNER JOIN metadata_authority ON (metadata_authority.id=authority_id) WHERE raw_extrinsic_metadata.target = %s """, (id,), ) yield from cur metadata_fetcher_cols = ["name", "version"] def metadata_fetcher_add(self, name: str, version: str, cur=None) -> None: cur = self._cursor(cur) cur.execute( "INSERT INTO metadata_fetcher (name, version) " "VALUES (%s, %s) ON CONFLICT DO NOTHING", (name, version), ) def metadata_fetcher_get(self, name: str, version: str, cur=None): cur = self._cursor(cur) cur.execute( f"SELECT {', '.join(self.metadata_fetcher_cols)} " f"FROM metadata_fetcher " f"WHERE name=%s AND version=%s", (name, version), ) return cur.fetchone() def metadata_fetcher_get_id( self, name: str, version: str, cur=None ) -> Optional[int]: cur = self._cursor(cur) cur.execute( "SELECT id FROM metadata_fetcher WHERE name=%s AND version=%s", (name, version), ) row = cur.fetchone() if row: return row[0] else: return None metadata_authority_cols = ["type", "url"] def metadata_authority_add(self, type: str, url: str, cur=None) -> None: cur = self._cursor(cur) cur.execute( "INSERT INTO metadata_authority (type, url) " "VALUES (%s, %s) ON CONFLICT DO NOTHING", (type, url), ) def metadata_authority_get(self, type: str, url: str, cur=None): cur = self._cursor(cur) cur.execute( f"SELECT {', '.join(self.metadata_authority_cols)} " f"FROM metadata_authority " f"WHERE type=%s AND url=%s", (type, url), ) return cur.fetchone() def metadata_authority_get_id(self, type: str, url: str, cur=None) -> Optional[int]: cur = self._cursor(cur) cur.execute( "SELECT id FROM metadata_authority WHERE type=%s AND url=%s", (type, url) ) row = cur.fetchone() if row: return row[0] else: return None def _get_random_row_from_table(self, table_name, cols, id_col, cur=None): random_sha1 = bytes(random.randint(0, 255) for _ in range(SHA1_SIZE)) cur = self._cursor(cur) query = """ (SELECT {cols} FROM {table} WHERE {id_col} >= %s ORDER BY {id_col} LIMIT 1) UNION (SELECT {cols} FROM {table} WHERE {id_col} < %s ORDER BY {id_col} DESC LIMIT 1) LIMIT 1 """.format( cols=", ".join(cols), table=table_name, id_col=id_col ) cur.execute(query, (random_sha1, random_sha1)) row = cur.fetchone() if row: return row[0] dbversion_cols = ["version", "release", "description"] def dbversion(self): with self.transaction() as cur: cur.execute( f""" SELECT {', '.join(self.dbversion_cols)} FROM dbversion ORDER BY version DESC LIMIT 1 """ ) return dict(zip(self.dbversion_cols, cur.fetchone())) def check_dbversion(self): dbversion = self.dbversion()["version"] if dbversion != self.current_version: logger.warning( "database dbversion (%s) != %s current_version (%s)", dbversion, __name__, self.current_version, ) return dbversion == self.current_version diff --git a/swh/storage/postgresql/storage.py b/swh/storage/postgresql/storage.py index bb3ae398..41ea41e7 100644 --- a/swh/storage/postgresql/storage.py +++ b/swh/storage/postgresql/storage.py @@ -1,1548 +1,1572 @@ # Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 from collections import defaultdict import contextlib from contextlib import contextmanager import datetime import itertools import operator from typing import Any, Counter, Dict, Iterable, List, Optional, Sequence, Tuple import attr import psycopg2 import psycopg2.errors import psycopg2.pool from swh.core.api.serializers import msgpack_dumps, msgpack_loads from swh.core.db.common import db_transaction, db_transaction_generator from swh.model.hashutil import DEFAULT_ALGORITHMS, hash_to_bytes, hash_to_hex from swh.model.identifiers import ExtendedObjectType, ExtendedSWHID, ObjectType from swh.model.model import ( SHA1_SIZE, Content, Directory, DirectoryEntry, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Release, Revision, Sha1, Sha1Git, SkippedContent, Snapshot, SnapshotBranch, TargetType, ) from swh.storage.exc import HashCollision, StorageArgumentException, StorageDBError from swh.storage.interface import ( VISIT_STATUSES, ListOrder, PagedResult, PartialBranches, ) from swh.storage.metrics import process_metrics, send_metric, timed from swh.storage.objstorage import ObjStorage from swh.storage.utils import ( extract_collision_hash, get_partition_bounds_bytes, map_optional, now, ) from swh.storage.writer import JournalWriter from . import converters from .db import Db # Max block size of contents to return BULK_BLOCK_CONTENT_LEN_MAX = 10000 EMPTY_SNAPSHOT_ID = hash_to_bytes("1a8893e6a86f444e8be8e7bda6cb34fb1735a00e") """Identifier for the empty snapshot""" VALIDATION_EXCEPTIONS = ( KeyError, TypeError, ValueError, psycopg2.errors.CheckViolation, psycopg2.errors.IntegrityError, psycopg2.errors.InvalidTextRepresentation, psycopg2.errors.NotNullViolation, psycopg2.errors.NumericValueOutOfRange, psycopg2.errors.UndefinedFunction, # (raised on wrong argument typs) ) """Exceptions raised by postgresql when validation of the arguments failed.""" @contextlib.contextmanager def convert_validation_exceptions(): """Catches postgresql errors related to invalid arguments, and re-raises a StorageArgumentException.""" try: yield except tuple(VALIDATION_EXCEPTIONS) as e: raise StorageArgumentException(str(e)) class Storage: """SWH storage proxy, encompassing DB and object storage """ def __init__( self, db, objstorage, min_pool_conns=1, max_pool_conns=10, journal_writer=None ): """ Args: db_conn: either a libpq connection string, or a psycopg2 connection obj_root: path to the root of the object storage """ try: if isinstance(db, psycopg2.extensions.connection): self._pool = None self._db = Db(db) else: self._pool = psycopg2.pool.ThreadedConnectionPool( min_pool_conns, max_pool_conns, db ) self._db = None except psycopg2.OperationalError as e: raise StorageDBError(e) self.journal_writer = JournalWriter(journal_writer) self.objstorage = ObjStorage(objstorage) def get_db(self): if self._db: return self._db else: return Db.from_pool(self._pool) def put_db(self, db): if db is not self._db: db.put_conn() @contextmanager def db(self): db = None try: db = self.get_db() yield db finally: if db: self.put_db(db) @timed @db_transaction() def check_config(self, *, check_write: bool, db: Db, cur=None) -> bool: if not self.objstorage.check_config(check_write=check_write): return False if not db.check_dbversion(): return False # Check permissions on one of the tables if check_write: check = "INSERT" else: check = "SELECT" cur.execute("select has_table_privilege(current_user, 'content', %s)", (check,)) return cur.fetchone()[0] def _content_unique_key(self, hash, db): """Given a hash (tuple or dict), return a unique key from the aggregation of keys. """ keys = db.content_hash_keys if isinstance(hash, tuple): return hash return tuple([hash[k] for k in keys]) def _content_add_metadata(self, db, cur, content): """Add content to the postgresql database but not the object storage. """ # create temporary table for metadata injection db.mktemp("content", cur) db.copy_to( (c.to_dict() for c in content), "tmp_content", db.content_add_keys, cur ) # move metadata in place try: db.content_add_from_temp(cur) except psycopg2.IntegrityError as e: if e.diag.sqlstate == "23505" and e.diag.table_name == "content": message_detail = e.diag.message_detail if message_detail: hash_name, hash_id = extract_collision_hash(message_detail) collision_contents_hashes = [ c.hashes() for c in content if c.get_hash(hash_name) == hash_id ] else: constraint_to_hash_name = { "content_pkey": "sha1", "content_sha1_git_idx": "sha1_git", "content_sha256_idx": "sha256", } hash_name = constraint_to_hash_name.get(e.diag.constraint_name) hash_id = None collision_contents_hashes = None raise HashCollision( hash_name, hash_id, collision_contents_hashes ) from None else: raise @timed @process_metrics def content_add(self, content: List[Content]) -> Dict[str, int]: ctime = now() contents = [attr.evolve(c, ctime=ctime) for c in content] # Must add to the objstorage before the DB and journal. Otherwise: # 1. in case of a crash the DB may "believe" we have the content, but # we didn't have time to write to the objstorage before the crash # 2. the objstorage mirroring, which reads from the journal, may attempt to # read from the objstorage before we finished writing it objstorage_summary = self.objstorage.content_add(contents) with self.db() as db: with db.transaction() as cur: missing = list( self.content_missing( map(Content.to_dict, contents), key_hash="sha1_git", db=db, cur=cur, ) ) contents = [c for c in contents if c.sha1_git in missing] self.journal_writer.content_add(contents) self._content_add_metadata(db, cur, contents) return { "content:add": len(contents), "content:add:bytes": objstorage_summary["content:add:bytes"], } @timed @db_transaction() def content_update( self, contents: List[Dict[str, Any]], keys: List[str] = [], *, db: Db, cur=None ) -> None: # TODO: Add a check on input keys. How to properly implement # this? We don't know yet the new columns. self.journal_writer.content_update(contents) db.mktemp("content", cur) select_keys = list(set(db.content_get_metadata_keys).union(set(keys))) with convert_validation_exceptions(): db.copy_to(contents, "tmp_content", select_keys, cur) db.content_update_from_temp(keys_to_update=keys, cur=cur) @timed @process_metrics @db_transaction() def content_add_metadata( self, content: List[Content], *, db: Db, cur=None ) -> Dict[str, int]: missing = self.content_missing( (c.to_dict() for c in content), key_hash="sha1_git", db=db, cur=cur, ) contents = [c for c in content if c.sha1_git in missing] self.journal_writer.content_add_metadata(contents) self._content_add_metadata(db, cur, contents) return { "content:add": len(contents), } @timed def content_get_data(self, content: Sha1) -> Optional[bytes]: # FIXME: Make this method support slicing the `data` return self.objstorage.content_get(content) @timed @db_transaction() def content_get_partition( self, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, *, db: Db, cur=None, ) -> PagedResult[Content]: if limit is None: raise StorageArgumentException("limit should not be None") (start, end) = get_partition_bounds_bytes( partition_id, nb_partitions, SHA1_SIZE ) if page_token: start = hash_to_bytes(page_token) if end is None: end = b"\xff" * SHA1_SIZE next_page_token: Optional[str] = None contents = [] for counter, row in enumerate(db.content_get_range(start, end, limit + 1, cur)): row_d = dict(zip(db.content_get_metadata_keys, row)) content = Content(**row_d) if counter >= limit: # take the last content for the next page starting from this next_page_token = hash_to_hex(content.sha1) break contents.append(content) assert len(contents) <= limit return PagedResult(results=contents, next_page_token=next_page_token) @timed @db_transaction(statement_timeout=500) def content_get( self, contents: List[bytes], algo: str = "sha1", *, db: Db, cur=None ) -> List[Optional[Content]]: contents_by_hash: Dict[bytes, Optional[Content]] = {} if algo not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "algo should be one of {','.join(DEFAULT_ALGORITHMS)}" ) rows = db.content_get_metadata_from_hashes(contents, algo, cur) key = operator.attrgetter(algo) for row in rows: row_d = dict(zip(db.content_get_metadata_keys, row)) content = Content(**row_d) contents_by_hash[key(content)] = content return [contents_by_hash.get(sha1) for sha1 in contents] @timed @db_transaction_generator() def content_missing( self, contents: List[Dict[str, Any]], key_hash: str = "sha1", *, db: Db, cur=None, ) -> Iterable[bytes]: if key_hash not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "key_hash should be one of {','.join(DEFAULT_ALGORITHMS)}" ) keys = db.content_hash_keys key_hash_idx = keys.index(key_hash) for obj in db.content_missing_from_list(contents, cur): yield obj[key_hash_idx] @timed @db_transaction_generator() def content_missing_per_sha1( self, contents: List[bytes], *, db: Db, cur=None ) -> Iterable[bytes]: for obj in db.content_missing_per_sha1(contents, cur): yield obj[0] @timed @db_transaction_generator() def content_missing_per_sha1_git( self, contents: List[bytes], *, db: Db, cur=None ) -> Iterable[Sha1Git]: for obj in db.content_missing_per_sha1_git(contents, cur): yield obj[0] @timed @db_transaction() def content_find( self, content: Dict[str, Any], *, db: Db, cur=None ) -> List[Content]: if not set(content).intersection(DEFAULT_ALGORITHMS): raise StorageArgumentException( "content keys must contain at least one " f"of: {', '.join(sorted(DEFAULT_ALGORITHMS))}" ) rows = db.content_find( sha1=content.get("sha1"), sha1_git=content.get("sha1_git"), sha256=content.get("sha256"), blake2s256=content.get("blake2s256"), cur=cur, ) contents = [] for row in rows: row_d = dict(zip(db.content_find_cols, row)) contents.append(Content(**row_d)) return contents @timed @db_transaction() def content_get_random(self, *, db: Db, cur=None) -> Sha1Git: return db.content_get_random(cur) @staticmethod def _skipped_content_normalize(d): d = d.copy() if d.get("status") is None: d["status"] = "absent" if d.get("length") is None: d["length"] = -1 return d def _skipped_content_add_metadata(self, db, cur, content: List[SkippedContent]): origin_ids = db.origin_id_get_by_url([cont.origin for cont in content], cur=cur) content = [ attr.evolve(c, origin=origin_id) for (c, origin_id) in zip(content, origin_ids) ] db.mktemp("skipped_content", cur) db.copy_to( [c.to_dict() for c in content], "tmp_skipped_content", db.skipped_content_keys, cur, ) # move metadata in place db.skipped_content_add_from_temp(cur) @timed @process_metrics @db_transaction() def skipped_content_add( self, content: List[SkippedContent], *, db: Db, cur=None ) -> Dict[str, int]: ctime = now() content = [attr.evolve(c, ctime=ctime) for c in content] missing_contents = self.skipped_content_missing( (c.to_dict() for c in content), db=db, cur=cur, ) content = [ c for c in content if any( all( c.get_hash(algo) == missing_content.get(algo) for algo in DEFAULT_ALGORITHMS ) for missing_content in missing_contents ) ] self.journal_writer.skipped_content_add(content) self._skipped_content_add_metadata(db, cur, content) return { "skipped_content:add": len(content), } @timed @db_transaction_generator() def skipped_content_missing( self, contents: List[Dict[str, Any]], *, db: Db, cur=None ) -> Iterable[Dict[str, Any]]: contents = list(contents) for content in db.skipped_content_missing(contents, cur): yield dict(zip(db.content_hash_keys, content)) @timed @process_metrics @db_transaction() def directory_add( self, directories: List[Directory], *, db: Db, cur=None ) -> Dict[str, int]: summary = {"directory:add": 0} dirs = set() dir_entries: Dict[str, defaultdict] = { "file": defaultdict(list), "dir": defaultdict(list), "rev": defaultdict(list), } for cur_dir in directories: dir_id = cur_dir.id dirs.add(dir_id) for src_entry in cur_dir.entries: entry = src_entry.to_dict() entry["dir_id"] = dir_id dir_entries[entry["type"]][dir_id].append(entry) dirs_missing = set(self.directory_missing(dirs, db=db, cur=cur)) if not dirs_missing: return summary self.journal_writer.directory_add( dir_ for dir_ in directories if dir_.id in dirs_missing ) # Copy directory ids dirs_missing_dict = ({"id": dir} for dir in dirs_missing) db.mktemp("directory", cur) db.copy_to(dirs_missing_dict, "tmp_directory", ["id"], cur) # Copy entries for entry_type, entry_list in dir_entries.items(): entries = itertools.chain.from_iterable( entries_for_dir for dir_id, entries_for_dir in entry_list.items() if dir_id in dirs_missing ) db.mktemp_dir_entry(entry_type) db.copy_to( entries, "tmp_directory_entry_%s" % entry_type, ["target", "name", "perms", "dir_id"], cur, ) # Do the final copy db.directory_add_from_temp(cur) summary["directory:add"] = len(dirs_missing) return summary @timed @db_transaction_generator() def directory_missing( self, directories: List[Sha1Git], *, db: Db, cur=None ) -> Iterable[Sha1Git]: for obj in db.directory_missing_from_list(directories, cur): yield obj[0] @timed @db_transaction_generator(statement_timeout=20000) def directory_ls( self, directory: Sha1Git, recursive: bool = False, *, db: Db, cur=None ) -> Iterable[Dict[str, Any]]: if recursive: res_gen = db.directory_walk(directory, cur=cur) else: res_gen = db.directory_walk_one(directory, cur=cur) for line in res_gen: yield dict(zip(db.directory_ls_cols, line)) @timed @db_transaction(statement_timeout=2000) def directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes], *, db: Db, cur=None ) -> Optional[Dict[str, Any]]: res = db.directory_entry_get_by_path(directory, paths, cur) return dict(zip(db.directory_ls_cols, res)) if res else None @timed @db_transaction() def directory_get_random(self, *, db: Db, cur=None) -> Sha1Git: return db.directory_get_random(cur) @db_transaction() def directory_get_entries( self, directory_id: Sha1Git, page_token: Optional[bytes] = None, limit: int = 1000, *, db: Db, cur=None, ) -> Optional[PagedResult[DirectoryEntry]]: if list(self.directory_missing([directory_id], db=db, cur=cur)): return None if page_token is not None: raise StorageArgumentException("Unsupported page token") # TODO: actually paginate rows = db.directory_get_entries(directory_id, cur=cur) return PagedResult( results=[ DirectoryEntry(**dict(zip(db.directory_get_entries_cols, row))) for row in rows ], next_page_token=None, ) @timed @process_metrics @db_transaction() def revision_add( self, revisions: List[Revision], *, db: Db, cur=None ) -> Dict[str, int]: summary = {"revision:add": 0} revisions_missing = set( self.revision_missing( set(revision.id for revision in revisions), db=db, cur=cur ) ) if not revisions_missing: return summary db.mktemp_revision(cur) revisions_filtered = [ revision for revision in revisions if revision.id in revisions_missing ] self.journal_writer.revision_add(revisions_filtered) db_revisions_filtered = list(map(converters.revision_to_db, revisions_filtered)) parents_filtered: List[Dict[str, Any]] = [] with convert_validation_exceptions(): db.copy_to( db_revisions_filtered, "tmp_revision", db.revision_add_cols, cur, lambda rev: parents_filtered.extend(rev["parents"]), ) db.revision_add_from_temp(cur) db.copy_to( parents_filtered, "revision_history", ["id", "parent_id", "parent_rank"], cur, ) return {"revision:add": len(revisions_missing)} @timed @db_transaction_generator() def revision_missing( self, revisions: List[Sha1Git], *, db: Db, cur=None ) -> Iterable[Sha1Git]: if not revisions: return None for obj in db.revision_missing_from_list(revisions, cur): yield obj[0] @timed @db_transaction(statement_timeout=1000) def revision_get( self, revision_ids: List[Sha1Git], *, db: Db, cur=None ) -> List[Optional[Revision]]: revisions = [] for line in db.revision_get_from_list(revision_ids, cur): revision = converters.db_to_revision(dict(zip(db.revision_get_cols, line))) revisions.append(revision) return revisions @timed @db_transaction_generator(statement_timeout=2000) def revision_log( self, revisions: List[Sha1Git], limit: Optional[int] = None, *, db: Db, cur=None ) -> Iterable[Optional[Dict[str, Any]]]: for line in db.revision_log(revisions, limit, cur): data = converters.db_to_revision(dict(zip(db.revision_get_cols, line))) if not data: yield None continue yield data.to_dict() @timed @db_transaction_generator(statement_timeout=2000) def revision_shortlog( self, revisions: List[Sha1Git], limit: Optional[int] = None, *, db: Db, cur=None ) -> Iterable[Optional[Tuple[Sha1Git, Tuple[Sha1Git, ...]]]]: yield from db.revision_shortlog(revisions, limit, cur) @timed @db_transaction() def revision_get_random(self, *, db: Db, cur=None) -> Sha1Git: return db.revision_get_random(cur) @timed @db_transaction() def extid_get_from_extid( - self, id_type: str, ids: List[bytes], *, db: Db, cur=None + self, + id_type: str, + ids: List[bytes], + version: Optional[int] = None, + *, + db: Db, + cur=None, ) -> List[ExtID]: extids = [] - for row in db.extid_get_from_extid_list(id_type, ids, cur): + for row in db.extid_get_from_extid_list(id_type, ids, version=version, cur=cur): if row[0] is not None: extids.append(converters.db_to_extid(dict(zip(db.extid_cols, row)))) return extids @timed @db_transaction() def extid_get_from_target( - self, target_type: ObjectType, ids: List[Sha1Git], *, db: Db, cur=None + self, + target_type: ObjectType, + ids: List[Sha1Git], + extid_type: Optional[str] = None, + extid_version: Optional[int] = None, + *, + db: Db, + cur=None, ) -> List[ExtID]: extids = [] - for row in db.extid_get_from_swhid_list(target_type.value, ids, cur): + if (extid_version is not None and extid_type is None) or ( + extid_version is None and extid_type is not None + ): + raise ValueError("You must provide both extid_type and extid_version") + + for row in db.extid_get_from_swhid_list( + target_type.value, + ids, + extid_version=extid_version, + extid_type=extid_type, + cur=cur, + ): if row[0] is not None: extids.append(converters.db_to_extid(dict(zip(db.extid_cols, row)))) return extids @timed @db_transaction() def extid_add(self, ids: List[ExtID], *, db: Db, cur=None) -> Dict[str, int]: extid = [ { "extid": extid.extid, "extid_type": extid.extid_type, "extid_version": getattr(extid, "extid_version", 0), "target": extid.target.object_id, "target_type": extid.target.object_type.name.lower(), # arghh } for extid in ids ] db.mktemp("extid", cur) self.journal_writer.extid_add(ids) db.copy_to(extid, "tmp_extid", db.extid_cols, cur) # move metadata in place db.extid_add_from_temp(cur) return {"extid:add": len(extid)} @timed @process_metrics @db_transaction() def release_add( self, releases: List[Release], *, db: Db, cur=None ) -> Dict[str, int]: summary = {"release:add": 0} release_ids = set(release.id for release in releases) releases_missing = set(self.release_missing(release_ids, db=db, cur=cur)) if not releases_missing: return summary db.mktemp_release(cur) releases_filtered = [ release for release in releases if release.id in releases_missing ] self.journal_writer.release_add(releases_filtered) db_releases_filtered = list(map(converters.release_to_db, releases_filtered)) with convert_validation_exceptions(): db.copy_to(db_releases_filtered, "tmp_release", db.release_add_cols, cur) db.release_add_from_temp(cur) return {"release:add": len(releases_missing)} @timed @db_transaction_generator() def release_missing( self, releases: List[Sha1Git], *, db: Db, cur=None ) -> Iterable[Sha1Git]: if not releases: return for obj in db.release_missing_from_list(releases, cur): yield obj[0] @timed @db_transaction(statement_timeout=500) def release_get( self, releases: List[Sha1Git], *, db: Db, cur=None ) -> List[Optional[Release]]: rels = [] for release in db.release_get_from_list(releases, cur): data = converters.db_to_release(dict(zip(db.release_get_cols, release))) rels.append(data if data else None) return rels @timed @db_transaction() def release_get_random(self, *, db: Db, cur=None) -> Sha1Git: return db.release_get_random(cur) @timed @process_metrics @db_transaction() def snapshot_add( self, snapshots: List[Snapshot], *, db: Db, cur=None ) -> Dict[str, int]: created_temp_table = False count = 0 for snapshot in snapshots: if not db.snapshot_exists(snapshot.id, cur): if not created_temp_table: db.mktemp_snapshot_branch(cur) created_temp_table = True with convert_validation_exceptions(): db.copy_to( ( { "name": name, "target": info.target if info else None, "target_type": ( info.target_type.value if info else None ), } for name, info in snapshot.branches.items() ), "tmp_snapshot_branch", ["name", "target", "target_type"], cur, ) self.journal_writer.snapshot_add([snapshot]) db.snapshot_add(snapshot.id, cur) count += 1 return {"snapshot:add": count} @timed @db_transaction_generator() def snapshot_missing( self, snapshots: List[Sha1Git], *, db: Db, cur=None ) -> Iterable[Sha1Git]: for obj in db.snapshot_missing_from_list(snapshots, cur): yield obj[0] @timed @db_transaction(statement_timeout=2000) def snapshot_get( self, snapshot_id: Sha1Git, *, db: Db, cur=None ) -> Optional[Dict[str, Any]]: d = self.snapshot_get_branches(snapshot_id) if d is None: return d return { "id": d["id"], "branches": { name: branch.to_dict() if branch else None for (name, branch) in d["branches"].items() }, "next_branch": d["next_branch"], } @timed @db_transaction(statement_timeout=2000) def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, *, db: Db, cur=None, ) -> Optional[Dict[Optional[str], int]]: return dict( [ bc for bc in db.snapshot_count_branches( snapshot_id, branch_name_exclude_prefix, cur, ) ] ) @timed @db_transaction(statement_timeout=2000) def snapshot_get_branches( self, snapshot_id: Sha1Git, branches_from: bytes = b"", branches_count: int = 1000, target_types: Optional[List[str]] = None, branch_name_include_substring: Optional[bytes] = None, branch_name_exclude_prefix: Optional[bytes] = None, *, db: Db, cur=None, ) -> Optional[PartialBranches]: if snapshot_id == EMPTY_SNAPSHOT_ID: return PartialBranches(id=snapshot_id, branches={}, next_branch=None,) branches = {} next_branch = None fetched_branches = list( db.snapshot_get_by_id( snapshot_id, branches_from=branches_from, # the underlying SQL query can be quite expensive to execute for small # branches_count value, so we ensure a minimum branches limit of 10 for # optimal performances branches_count=max(branches_count + 1, 10), target_types=target_types, branch_name_include_substring=branch_name_include_substring, branch_name_exclude_prefix=branch_name_exclude_prefix, cur=cur, ) ) for row in fetched_branches[:branches_count]: branch_d = dict(zip(db.snapshot_get_cols, row)) del branch_d["snapshot_id"] name = branch_d.pop("name") if branch_d["target"] is None and branch_d["target_type"] is None: branch = None else: assert branch_d["target_type"] is not None branch = SnapshotBranch( target=branch_d["target"], target_type=TargetType(branch_d["target_type"]), ) branches[name] = branch if len(fetched_branches) > branches_count: next_branch = dict( zip(db.snapshot_get_cols, fetched_branches[branches_count]) )["name"] if branches: return PartialBranches( id=snapshot_id, branches=branches, next_branch=next_branch, ) return None @timed @db_transaction() def snapshot_get_random(self, *, db: Db, cur=None) -> Sha1Git: return db.snapshot_get_random(cur) @timed @db_transaction() def origin_visit_add( self, visits: List[OriginVisit], *, db: Db, cur=None ) -> Iterable[OriginVisit]: for visit in visits: origin = self.origin_get([visit.origin], db=db, cur=cur)[0] if not origin: # Cannot add a visit without an origin raise StorageArgumentException("Unknown origin %s", visit.origin) all_visits = [] nb_visits = 0 for visit in visits: nb_visits += 1 if not visit.visit: with convert_validation_exceptions(): visit_id = db.origin_visit_add( visit.origin, visit.date, visit.type, cur=cur ) visit = attr.evolve(visit, visit=visit_id) else: db.origin_visit_add_with_id(visit, cur=cur) assert visit.visit is not None all_visits.append(visit) # Forced to write after for the case when the visit has no id self.journal_writer.origin_visit_add([visit]) visit_status = OriginVisitStatus( origin=visit.origin, visit=visit.visit, date=visit.date, type=visit.type, status="created", snapshot=None, ) self._origin_visit_status_add(visit_status, db=db, cur=cur) send_metric("origin_visit:add", count=nb_visits, method_name="origin_visit") return all_visits def _origin_visit_status_add( self, visit_status: OriginVisitStatus, db, cur ) -> None: """Add an origin visit status""" self.journal_writer.origin_visit_status_add([visit_status]) db.origin_visit_status_add(visit_status, cur=cur) @timed @process_metrics @db_transaction() def origin_visit_status_add( self, visit_statuses: List[OriginVisitStatus], *, db: Db, cur=None, ) -> Dict[str, int]: visit_statuses_ = [] # First round to check existence (fail early if any is ko) for visit_status in visit_statuses: origin_url = self.origin_get([visit_status.origin], db=db, cur=cur)[0] if not origin_url: raise StorageArgumentException(f"Unknown origin {visit_status.origin}") if visit_status.type is None: origin_visit = self.origin_visit_get_by( visit_status.origin, visit_status.visit, db=db, cur=cur ) if origin_visit is None: raise StorageArgumentException( f"Unknown origin visit {visit_status.visit} " f"of origin {visit_status.origin}" ) origin_visit_status = attr.evolve(visit_status, type=origin_visit.type) else: origin_visit_status = visit_status visit_statuses_.append(origin_visit_status) for visit_status in visit_statuses_: self._origin_visit_status_add(visit_status, db, cur) return {"origin_visit_status:add": len(visit_statuses_)} @timed @db_transaction() def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, *, db: Db, cur=None, ) -> Optional[OriginVisitStatus]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) row_d = db.origin_visit_status_get_latest( origin_url, visit, allowed_statuses, require_snapshot, cur=cur ) if not row_d: return None return OriginVisitStatus(**row_d) @timed @db_transaction(statement_timeout=500) def origin_visit_get( self, origin: str, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, *, db: Db, cur=None, ) -> PagedResult[OriginVisit]: page_token = page_token or "0" if not isinstance(order, ListOrder): raise StorageArgumentException("order must be a ListOrder value") if not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") next_page_token = None visit_from = int(page_token) visits: List[OriginVisit] = [] extra_limit = limit + 1 for row in db.origin_visit_get_range( origin, visit_from=visit_from, order=order, limit=extra_limit, cur=cur ): row_d = dict(zip(db.origin_visit_cols, row)) visits.append( OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) ) assert len(visits) <= extra_limit if len(visits) == extra_limit: visits = visits[:limit] next_page_token = str(visits[-1].visit) return PagedResult(results=visits, next_page_token=next_page_token) @timed @db_transaction(statement_timeout=500) def origin_visit_find_by_date( self, origin: str, visit_date: datetime.datetime, *, db: Db, cur=None ) -> Optional[OriginVisit]: row_d = db.origin_visit_find_by_date(origin, visit_date, cur=cur) if not row_d: return None return OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) @timed @db_transaction(statement_timeout=500) def origin_visit_get_by( self, origin: str, visit: int, *, db: Db, cur=None ) -> Optional[OriginVisit]: row = db.origin_visit_get(origin, visit, cur) if row: row_d = dict(zip(db.origin_visit_get_cols, row)) return OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) return None @timed @db_transaction(statement_timeout=4000) def origin_visit_get_latest( self, origin: str, type: Optional[str] = None, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, *, db: Db, cur=None, ) -> Optional[OriginVisit]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) row = db.origin_visit_get_latest( origin, type=type, allowed_statuses=allowed_statuses, require_snapshot=require_snapshot, cur=cur, ) if row: row_d = dict(zip(db.origin_visit_get_cols, row)) visit = OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) return visit return None @timed @db_transaction(statement_timeout=500) def origin_visit_status_get( self, origin: str, visit: int, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, *, db: Db, cur=None, ) -> PagedResult[OriginVisitStatus]: next_page_token = None date_from = None if page_token is not None: date_from = datetime.datetime.fromisoformat(page_token) visit_statuses: List[OriginVisitStatus] = [] # Take one more visit status so we can reuse it as the next page token if any for row in db.origin_visit_status_get_range( origin, visit, date_from=date_from, order=order, limit=limit + 1, cur=cur, ): row_d = dict(zip(db.origin_visit_status_cols, row)) visit_statuses.append(OriginVisitStatus(**row_d)) if len(visit_statuses) > limit: # last visit status date is the next page token next_page_token = str(visit_statuses[-1].date) # excluding that visit status from the result to respect the limit size visit_statuses = visit_statuses[:limit] return PagedResult(results=visit_statuses, next_page_token=next_page_token) @timed @db_transaction() def origin_visit_status_get_random( self, type: str, *, db: Db, cur=None ) -> Optional[OriginVisitStatus]: row = db.origin_visit_get_random(type, cur) if row is not None: row_d = dict(zip(db.origin_visit_status_cols, row)) return OriginVisitStatus(**row_d) return None @timed @db_transaction(statement_timeout=2000) def object_find_by_sha1_git( self, ids: List[Sha1Git], *, db: Db, cur=None ) -> Dict[Sha1Git, List[Dict]]: ret: Dict[Sha1Git, List[Dict]] = {id: [] for id in ids} for retval in db.object_find_by_sha1_git(ids, cur=cur): if retval[1]: ret[retval[0]].append( dict(zip(db.object_find_by_sha1_git_cols, retval)) ) return ret @timed @db_transaction(statement_timeout=500) def origin_get( self, origins: List[str], *, db: Db, cur=None ) -> Iterable[Optional[Origin]]: rows = db.origin_get_by_url(origins, cur) result: List[Optional[Origin]] = [] for row in rows: origin_d = dict(zip(db.origin_cols, row)) url = origin_d["url"] result.append(None if url is None else Origin(url=url)) return result @timed @db_transaction(statement_timeout=500) def origin_get_by_sha1( self, sha1s: List[bytes], *, db: Db, cur=None ) -> List[Optional[Dict[str, Any]]]: return [ dict(zip(db.origin_cols, row)) if row[0] else None for row in db.origin_get_by_sha1(sha1s, cur) ] @timed @db_transaction_generator() def origin_get_range(self, origin_from=1, origin_count=100, *, db: Db, cur=None): for origin in db.origin_get_range(origin_from, origin_count, cur): yield dict(zip(db.origin_get_range_cols, origin)) @timed @db_transaction() def origin_list( self, page_token: Optional[str] = None, limit: int = 100, *, db: Db, cur=None ) -> PagedResult[Origin]: page_token = page_token or "0" if not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") origin_from = int(page_token) next_page_token = None origins: List[Origin] = [] # Take one more origin so we can reuse it as the next page token if any for row_d in self.origin_get_range(origin_from, limit + 1, db=db, cur=cur): origins.append(Origin(url=row_d["url"])) # keep the last_id for the pagination if needed last_id = row_d["id"] if len(origins) > limit: # data left for subsequent call # last origin id is the next page token next_page_token = str(last_id) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) @timed @db_transaction() def origin_search( self, url_pattern: str, page_token: Optional[str] = None, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, *, db: Db, cur=None, ) -> PagedResult[Origin]: next_page_token = None offset = int(page_token) if page_token else 0 origins = [] # Take one more origin so we can reuse it as the next page token if any for origin in db.origin_search( url_pattern, offset, limit + 1, regexp, with_visit, visit_types, cur ): row_d = dict(zip(db.origin_cols, origin)) origins.append(Origin(url=row_d["url"])) if len(origins) > limit: # next offset next_page_token = str(offset + limit) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) @timed @db_transaction() def origin_count( self, url_pattern: str, regexp: bool = False, with_visit: bool = False, *, db: Db, cur=None, ) -> int: return db.origin_count(url_pattern, regexp, with_visit, cur) @timed @process_metrics @db_transaction() def origin_add(self, origins: List[Origin], *, db: Db, cur=None) -> Dict[str, int]: urls = [o.url for o in origins] known_origins = set(url for (url,) in db.origin_get_by_url(urls, cur)) # keep only one occurrence of each given origin while keeping the list # sorted as originally given to_add = sorted(set(urls) - known_origins, key=urls.index) self.journal_writer.origin_add([Origin(url=url) for url in to_add]) added = 0 for url in to_add: if db.origin_add(url, cur): added += 1 return {"origin:add": added} @db_transaction(statement_timeout=500) def stat_counters(self, *, db: Db, cur=None): return {k: v for (k, v) in db.stat_counters()} @db_transaction() def refresh_stat_counters(self, *, db: Db, cur=None): keys = [ "content", "directory", "directory_entry_dir", "directory_entry_file", "directory_entry_rev", "origin", "origin_visit", "person", "release", "revision", "revision_history", "skipped_content", "snapshot", ] for key in keys: cur.execute("select * from swh_update_counter(%s)", (key,)) @timed @process_metrics @db_transaction() def raw_extrinsic_metadata_add( self, metadata: List[RawExtrinsicMetadata], db, cur, ) -> Dict[str, int]: metadata = list(metadata) self.journal_writer.raw_extrinsic_metadata_add(metadata) counter = Counter[ExtendedObjectType]() for metadata_entry in metadata: authority_id = self._get_authority_id(metadata_entry.authority, db, cur) fetcher_id = self._get_fetcher_id(metadata_entry.fetcher, db, cur) db.raw_extrinsic_metadata_add( id=metadata_entry.id, type=metadata_entry.target.object_type.name.lower(), target=str(metadata_entry.target), discovery_date=metadata_entry.discovery_date, authority_id=authority_id, fetcher_id=fetcher_id, format=metadata_entry.format, metadata=metadata_entry.metadata, origin=metadata_entry.origin, visit=metadata_entry.visit, snapshot=map_optional(str, metadata_entry.snapshot), release=map_optional(str, metadata_entry.release), revision=map_optional(str, metadata_entry.revision), path=metadata_entry.path, directory=map_optional(str, metadata_entry.directory), cur=cur, ) counter[metadata_entry.target.object_type] += 1 return { f"{type.value}_metadata:add": count for (type, count) in counter.items() } @db_transaction() def raw_extrinsic_metadata_get( self, target: ExtendedSWHID, authority: MetadataAuthority, after: Optional[datetime.datetime] = None, page_token: Optional[bytes] = None, limit: int = 1000, *, db: Db, cur=None, ) -> PagedResult[RawExtrinsicMetadata]: if page_token: (after_time, after_fetcher) = msgpack_loads(base64.b64decode(page_token)) if after and after_time < after: raise StorageArgumentException( "page_token is inconsistent with the value of 'after'." ) else: after_time = after after_fetcher = None authority_id = self._get_authority_id(authority, db, cur) if not authority_id: return PagedResult(next_page_token=None, results=[],) rows = db.raw_extrinsic_metadata_get( str(target), authority_id, after_time, after_fetcher, limit + 1, cur, ) rows = [dict(zip(db.raw_extrinsic_metadata_get_cols, row)) for row in rows] results = [] for row in rows: assert str(target) == row["raw_extrinsic_metadata.target"] results.append(converters.db_to_raw_extrinsic_metadata(row)) if len(results) > limit: results.pop() assert len(results) == limit last_returned_row = rows[-2] # rows[-1] corresponds to the popped result next_page_token: Optional[str] = base64.b64encode( msgpack_dumps( ( last_returned_row["discovery_date"], last_returned_row["metadata_fetcher.id"], ) ) ).decode() else: next_page_token = None return PagedResult(next_page_token=next_page_token, results=results,) @db_transaction() def raw_extrinsic_metadata_get_by_ids( self, ids: List[Sha1Git], *, db: Db, cur=None, ) -> List[RawExtrinsicMetadata]: return [ converters.db_to_raw_extrinsic_metadata( dict(zip(db.raw_extrinsic_metadata_get_cols, row)) ) for row in db.raw_extrinsic_metadata_get_by_ids(ids) ] @db_transaction() def raw_extrinsic_metadata_get_authorities( self, target: ExtendedSWHID, *, db: Db, cur=None, ) -> List[MetadataAuthority]: return [ MetadataAuthority( type=MetadataAuthorityType(authority_type), url=authority_url ) for ( authority_type, authority_url, ) in db.raw_extrinsic_metadata_get_authorities(str(target), cur) ] @timed @process_metrics @db_transaction() def metadata_fetcher_add( self, fetchers: List[MetadataFetcher], *, db: Db, cur=None ) -> Dict[str, int]: fetchers = list(fetchers) self.journal_writer.metadata_fetcher_add(fetchers) count = 0 for fetcher in fetchers: db.metadata_fetcher_add(fetcher.name, fetcher.version, cur=cur) count += 1 return {"metadata_fetcher:add": count} @timed @db_transaction(statement_timeout=500) def metadata_fetcher_get( self, name: str, version: str, *, db: Db, cur=None ) -> Optional[MetadataFetcher]: row = db.metadata_fetcher_get(name, version, cur=cur) if not row: return None return MetadataFetcher.from_dict(dict(zip(db.metadata_fetcher_cols, row))) @timed @process_metrics @db_transaction() def metadata_authority_add( self, authorities: List[MetadataAuthority], *, db: Db, cur=None ) -> Dict[str, int]: authorities = list(authorities) self.journal_writer.metadata_authority_add(authorities) count = 0 for authority in authorities: db.metadata_authority_add(authority.type.value, authority.url, cur=cur) count += 1 return {"metadata_authority:add": count} @timed @db_transaction() def metadata_authority_get( self, type: MetadataAuthorityType, url: str, *, db: Db, cur=None ) -> Optional[MetadataAuthority]: row = db.metadata_authority_get(type.value, url, cur=cur) if not row: return None return MetadataAuthority.from_dict(dict(zip(db.metadata_authority_cols, row))) def clear_buffers(self, object_types: Sequence[str] = ()) -> None: """Do nothing """ return None def flush(self, object_types: Sequence[str] = ()) -> Dict[str, int]: return {} def _get_authority_id(self, authority: MetadataAuthority, db, cur): authority_id = db.metadata_authority_get_id( authority.type.value, authority.url, cur ) if not authority_id: raise StorageArgumentException(f"Unknown authority {authority}") return authority_id def _get_fetcher_id(self, fetcher: MetadataFetcher, db, cur): fetcher_id = db.metadata_fetcher_get_id(fetcher.name, fetcher.version, cur) if not fetcher_id: raise StorageArgumentException(f"Unknown fetcher {fetcher}") return fetcher_id diff --git a/swh/storage/proxies/counter.py b/swh/storage/proxies/counter.py new file mode 100644 index 00000000..40633b27 --- /dev/null +++ b/swh/storage/proxies/counter.py @@ -0,0 +1,66 @@ +# Copyright (C) 2021 The Software Heritage developers +# See the AUTHORS file at the top-level directory of this distribution +# License: GNU General Public License version 3, or any later version +# See top-level LICENSE file for more information + + +from typing import Callable + +from swh.counters import get_counters +from swh.counters.interface import CountersInterface +from swh.storage import get_storage +from swh.storage.interface import StorageInterface + +OBJECT_TYPES = [ + "content", + "directory", + "snapshot", + "origin_visit_status", + "origin_visit", + "origin", +] + + +class CountingProxyStorage: + """Counting Storage Proxy. + + This is in charge of adding objects directly to swh-counters, without + going through Kafka/swh-journal. + This is meant as a simple way to setup counters for experiments; production + should use swh-journal to reduce load/latency of the storage server. + + Additionally, unlike the journal-based counting, it does not count persons + or the number of origins per netloc. + + Sample configuration use case for filtering storage: + + .. code-block: yaml + + storage: + cls: counter + counters: + cls: remote + url: http://counters.internal.staging.swh.network:5011/ + storage: + cls: remote + url: http://storage.internal.staging.swh.network:5002/ + + """ + + def __init__(self, counters, storage): + self.counters: CountersInterface = get_counters(**counters) + self.storage: StorageInterface = get_storage(**storage) + + def __getattr__(self, key): + if key == "storage": + raise AttributeError(key) + if key.endswith("_add"): + return self._adder(key[0:-4], getattr(self.storage, key)) + return getattr(self.storage, key) + + def _adder(self, collection: str, backend_function: Callable): + def f(objs): + self.counters.add(collection, [obj.unique_key() for obj in objs]) + return backend_function(objs) + + return f diff --git a/swh/storage/tests/migrate_extrinsic_metadata/test_pypi.py b/swh/storage/tests/migrate_extrinsic_metadata/test_pypi.py index 4220fe28..acaab9c6 100644 --- a/swh/storage/tests/migrate_extrinsic_metadata/test_pypi.py +++ b/swh/storage/tests/migrate_extrinsic_metadata/test_pypi.py @@ -1,651 +1,653 @@ # Copyright (C) 2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # flake8: noqa # because of long lines import copy import datetime import json import urllib.error import attr from swh.model.identifiers import CoreSWHID, ExtendedObjectType, ExtendedSWHID from swh.model.model import ( MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Snapshot, SnapshotBranch, TargetType, ) from swh.storage import get_storage from swh.storage.interface import PagedResult from swh.storage.migrate_extrinsic_metadata import ( handle_row, pypi_origin_from_filename, pypi_project_from_filename, ) FETCHER = MetadataFetcher( name="migrate-extrinsic-metadata-from-revisions", version="0.0.1", ) PYPI_AUTHORITY = MetadataAuthority( type=MetadataAuthorityType.FORGE, url="https://pypi.org/", ) SWH_AUTHORITY = MetadataAuthority( type=MetadataAuthorityType.REGISTRY, url="https://softwareheritage.org/", ) DIRECTORY_ID = b"a" * 20 DIRECTORY_SWHID = ExtendedSWHID( object_type=ExtendedObjectType.DIRECTORY, object_id=DIRECTORY_ID ) def now(): return datetime.datetime.now(tz=datetime.timezone.utc) def test_pypi_project_from_filename(): files = [ ("django-agent-trust-0.1.8.tar.gz", "django-agent-trust"), ("python_test-1.0.1.zip", "python_test"), ("py-evm-0.2.0a9.tar.gz", "py-evm"), ("collective.texttospeech-1.0rc1.tar.gz", "collective.texttospeech"), ("flatland-fork-0.4.post1.dev40550160.zip", "flatland-fork"), ("fake-factory-0.5.6-proper.tar.gz", "fake-factory"), ("ariane_procos-0.1.2-b05.tar.gz", "ariane_procos"), ("Yelpy-0.2.2dev.tar.gz", "Yelpy"), ("geventhttpclient_c-1.0a-t1.tar.gz", "geventhttpclient_c"), ("codeforlife-portal-1.0.0.post.dev618.tar.gz", "codeforlife-portal"), ("ChecklistDSL-0.0.1.alpha.1.tar.gz", "ChecklistDSL"), ("transifex-1.1.0beta.tar.gz", "transifex"), ("thespian-2.5.10.tar.bz2", "thespian"), ("janis pipelines-0.5.3.tar.gz", "janis-pipelines"), ("pants-1.0.0-beta.2.tar.gz", "pants"), ("uforge_python_sdk-3.8.4-RC15.tar.gz", "uforge_python_sdk"), ("virtuoso-0.11.0.48.b5865c2b46fb.tar.gz", "virtuoso"), ("cloud_ftp-v1.0.0.tar.gz", "cloud_ftp"), ("frozenordereddict-1.0.0.tgz", "frozenordereddict"), ("pywebsite-0.1.2pre.tar.gz", "pywebsite"), ("Flask Unchained-0.2.0.tar.gz", "Flask-Unchained"), ("mongomotor-0.13.0.n.tar.gz", "mongomotor"), ("datahaven-rev8784.tar.gz", "datahaven"), ("geopandas-0.1.0.dev-120d5ee.tar.gz", "geopandas"), ("aimmo-v0.1.1-alpha.post.dev61.tar.gz", "aimmo"), ("django-migrations-plus-0.1.0.dev5.gdd1abd3.tar.gz", "django-migrations-plus"), ("function_shield.tar.gz", "function_shield"), ("Dtls-0.1.0.sdist_with_openssl.mingw-win32.tar.gz", "Dtls"), ("pytz-2005m.tar.gz", "pytz"), ("python-librsync-0.1-3.tar.gz", "python-librsync"), ("powny-1.4.0-alpha-20141205-1452-f5a2b03.tar.gz", "powny"), ("stp-3pc-batch-0.1.11.tar.gz", "stp-3pc-batch"), ("obedient.powny-3.0.0-alpha-20141027-2102-9e53ebd.tar.gz", "obedient.powny"), ("mojimoji-0.0.9_2.tar.gz", "mojimoji"), ("devpi-theme-16-2.0.0.tar.gz", "devpi-theme-16"), ("Orange3-WONDER-1-1.0.7.tar.gz", "Orange3-WONDER-1"), ("obj-34.tar.gz", "obj"), ("pytorch-ignite-nightly-20190825.tar.gz", "pytorch-ignite-nightly"), ("tlds-2019081900.tar.gz", "tlds"), ("dominator-12.1.2-alpha-20141027-1446-ad46e0f.tar.gz", "dominator"), ("waferslim-1.0.0-py3.1.zip", "waferslim"), ("Beaver-21.tar.gz", "Beaver"), ("aimmo-0.post.dev460.tar.gz", "aimmo"), ("ohai-1!0.tar.gz", "ohai"), ("nevolution-risk-139.tar.gz", "nevolution-risk"), ("collective.topicitemsevent-0.1dvl.tar.gz", "collective.topicitemsevent"), ("lesscpy-0.9g.tar.gz", "lesscpy"), ("SpiNNStorageHandlers-1!4.0.0a1.tar.gz", "SpiNNStorageHandlers"), ("limnoria-2013-03-27T16:32:26+0100.tar.gz", "limnoria"), ( "sPyNNakerExternalDevicesPlugin-1!4.0.0a2.tar.gz", "sPyNNakerExternalDevicesPlugin", ), ("django-bootstrap-italia_0.1.tar.gz", "django-bootstrap-italia"), ("sPyNNaker8-1!4.0.0a1.tar.gz", "sPyNNaker8"), ("betahaus.openmember-0.1adev-r1651.tar.gz", "betahaus.openmember"), ("mailer.0.8.0.zip", "mailer"), ("pytz-2005k.tar.bz2", "pytz"), ("aha.plugin.microne-0.62bdev.tar.gz", "aha.plugin.microne"), ("youtube_dl_server-alpha.3.tar.gz", "youtube_dl_server"), ("json-extensions-b76bc7d.tar.gz", "json-extensions"), ("LitReview-0.6989ev.tar.gz", "LitReview"), ("django_options-r5.tar.gz", "django_options"), ("ddlib-2013-11-07.tar.gz", "ddlib"), ("python-morfeusz-0.3000+py3k.tar.gz", "python-morfeusz"), ("gaepytz-2011h.zip", "gaepytz"), ("ftldat-r3.tar.gz", "ftldat"), ("tigretoolbox-0.0.0-py2.7-linux-x86_64.egg", None), ( "Greater than, equal, or less Library-0.1.tar.gz", "Greater-than-equal-or-less-Library", ), ("upstart--main-.-VLazy.object.at.0x104ba8b50-.tar.gz", "upstart"), ("duckduckpy0.1.tar.gz", "duckduckpy"), ("QUI for MPlayer snapshot_9-14-2011.zip", "QUI-for-MPlayer"), ("Eddy's Memory Game-1.0.zip", "Eddy-s-Memory-Game"), ("jekyll2nikola-0-0-1.tar.gz", "jekyll2nikola"), ("ore.workflowed-0-6-2.tar.gz", "ore.workflowed"), ("instancemanager-1.0rc-r34317.tar.gz", "instancemanager"), ("OrzMC_W&L-1.0.0.tar.gz", "OrzMC-W-L"), ("use0mk.tar.gz", "use0mk"), ("play-0-develop-1-gd67cd85.tar.gz", "play"), + ("mosaic-nist-2.0b1+f98ae80.tar.gz", "mosaic-nist"), + ("pypops-201408-r3.tar.gz", "pypops"), ] for (filename, project) in files: assert pypi_project_from_filename(filename) == project def test_pypi_origin_from_project_name(mocker): origin_url = "https://pypi.org/project/ProjectName/" storage = get_storage("memory") revision_id = b"41" * 10 snapshot_id = b"42" * 10 storage.origin_add([Origin(url=origin_url)]) storage.origin_visit_add( [OriginVisit(origin=origin_url, visit=1, date=now(), type="pypi")] ) storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin_url, visit=1, date=now(), status="partial", snapshot=snapshot_id, ) ] ) storage.snapshot_add( [ Snapshot( id=snapshot_id, branches={ b"foo": SnapshotBranch( target_type=TargetType.REVISION, target=revision_id, ) }, ) ] ) class response: code = 200 def read(self): return b'{"info": {"name": "ProjectName"}}' mock_urlopen = mocker.patch( "swh.storage.migrate_extrinsic_metadata.urlopen", return_value=response(), ) assert ( pypi_origin_from_filename(storage, revision_id, "ProjectName-1.0.0.tar.gz") == origin_url ) mock_urlopen.assert_not_called() assert ( pypi_origin_from_filename(storage, revision_id, "projectname-1.0.0.tar.gz") == origin_url ) mock_urlopen.assert_called_once_with("https://pypi.org/pypi/projectname/json/") def test_pypi_1(): """Tests loading a revision generated by a new PyPI loader that has a provider.""" extrinsic_metadata = { "url": "https://files.pythonhosted.org/packages/70/89/a498245baf1bf3dde73d3da00b4b067a8aa7c7378ad83472078803ea3e43/m3-ui-2.2.73.tar.gz", "size": 3933168, "digests": { "md5": "a374ac3f655e97df5db5335e2142d344", "sha256": "1bc2756f7d0d2e15cf5880ca697682ff35e8b58116bf73eb9c78b3db358c5b7d", }, "has_sig": False, "filename": "m3-ui-2.2.73.tar.gz", "downloads": -1, "md5_digest": "a374ac3f655e97df5db5335e2142d344", "packagetype": "sdist", "upload_time": "2019-11-11T06:21:20", "comment_text": "", "python_version": "source", "requires_python": None, "upload_time_iso_8601": "2019-11-11T06:21:20.073082Z", } original_artifacts = [ { "length": 3933168, "filename": "m3-ui-2.2.73.tar.gz", "checksums": { "sha1": "9f4ec7ce64b7fea4b122e85d47ea31146c367b03", "sha256": "1bc2756f7d0d2e15cf5880ca697682ff35e8b58116bf73eb9c78b3db358c5b7d", }, } ] row = { "id": b"\x00\x00\x07a{S\xe7\xb1E\x8fi]\xd0}\xe4\xceU\xaf\x15\x17", "directory": DIRECTORY_ID, "date": datetime.datetime( 2019, 11, 11, 6, 21, 20, tzinfo=datetime.timezone.utc, ), "committer_date": datetime.datetime( 2019, 11, 11, 6, 21, 20, tzinfo=datetime.timezone.utc, ), "type": "tar", "message": b"2.2.73", "metadata": { "extrinsic": { "raw": extrinsic_metadata, "when": "2020-01-23T18:43:09.109407+00:00", "provider": "https://pypi.org/pypi/m3-ui/json", }, "intrinsic": { "raw": { "name": "m3-ui", "summary": "======", "version": "2.2.73", # ... "metadata_version": "1.1", }, "tool": "PKG-INFO", }, "original_artifact": original_artifacts, }, } origin_url = "https://pypi.org/project/m3-ui/" storage = get_storage("memory") storage.origin_add([Origin(url=origin_url)]) storage.metadata_authority_add( [ attr.evolve(PYPI_AUTHORITY, metadata={}), attr.evolve(SWH_AUTHORITY, metadata={}), ] ) storage.metadata_fetcher_add([FETCHER]) deposit_cur = None handle_row(copy.deepcopy(row), storage, deposit_cur, dry_run=False) revision_swhid = CoreSWHID.from_string( "swh:1:rev:000007617b53e7b1458f695dd07de4ce55af1517" ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=PYPI_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2020, 1, 23, 18, 43, 9, 109407, tzinfo=datetime.timezone.utc, ), authority=PYPI_AUTHORITY, fetcher=FETCHER, format="pypi-project-json", metadata=json.dumps(extrinsic_metadata).encode(), origin=origin_url, revision=revision_swhid, ), ], next_page_token=None, ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=SWH_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2020, 1, 23, 18, 43, 9, 109407, tzinfo=datetime.timezone.utc, ), authority=SWH_AUTHORITY, fetcher=FETCHER, format="original-artifacts-json", metadata=json.dumps(original_artifacts).encode(), origin=origin_url, revision=revision_swhid, ), ], next_page_token=None, ) def test_pypi_2(mocker): """Tests loading a revision generated by an old PyPI loader that does not have a provider, but has 'project' metadata.""" mocker.patch( "swh.storage.migrate_extrinsic_metadata.urlopen", side_effect=urllib.error.HTTPError(None, 404, "Not Found", None, None), ) extrinsic_metadata = { "name": "jupyterhub-simx", "author": "Jupyter Development Team", "license": "BSD", "summary": "JupyterHub: A multi-user server for Jupyter notebooks", "version": "1.0.5", # ... } source_original_artifacts = [ { "url": "https://files.pythonhosted.org/packages/72/28/a8098763d78e2c4607cb67602c0d726a97ac38d4c1f531aac28f49de2e1a/jupyterhub-simx-1.0.5.tar.gz", "date": "2019-01-23T22:10:55", "sha1": "ede3eadd5a06e70912e3ba7cfccef789c4ad3168", "size": 2346538, "sha256": "0399d7f5f0d90c525d369f0507ad0e8ef8729c1c7fa63aadfc46a27514d14a46", "filename": "jupyterhub-simx-1.0.5.tar.gz", "sha1_git": "734301124712182eb30fc90e97cc18cef5432f02", "blake2s256": "bb4aa82ffb5891a05dcf6d4dce3ad56fd2c18e9abdba9d20972910649d869322", "archive_type": "tar", } ] dest_original_artifacts = [ { "url": "https://files.pythonhosted.org/packages/72/28/a8098763d78e2c4607cb67602c0d726a97ac38d4c1f531aac28f49de2e1a/jupyterhub-simx-1.0.5.tar.gz", "filename": "jupyterhub-simx-1.0.5.tar.gz", "archive_type": "tar", "length": 2346538, "checksums": { "sha1": "ede3eadd5a06e70912e3ba7cfccef789c4ad3168", "sha256": "0399d7f5f0d90c525d369f0507ad0e8ef8729c1c7fa63aadfc46a27514d14a46", "sha1_git": "734301124712182eb30fc90e97cc18cef5432f02", "blake2s256": "bb4aa82ffb5891a05dcf6d4dce3ad56fd2c18e9abdba9d20972910649d869322", }, } ] row = { "id": b"\x00\x00\x04\xd68,J\xd4\xc0Q\x92fbl6U\x1f\x0eQ\xca", "directory": DIRECTORY_ID, "date": datetime.datetime( 2019, 1, 23, 22, 10, 55, tzinfo=datetime.timezone.utc ), "committer_date": datetime.datetime( 2019, 1, 23, 22, 10, 55, tzinfo=datetime.timezone.utc ), "type": "tar", "message": b"1.0.5", "metadata": { "project": extrinsic_metadata, "original_artifact": source_original_artifacts, }, } origin_url = "https://pypi.org/project/jupyterhub-simx/" storage = get_storage("memory") storage.origin_add([Origin(url=origin_url)]) storage.metadata_authority_add( [ attr.evolve(PYPI_AUTHORITY, metadata={}), attr.evolve(SWH_AUTHORITY, metadata={}), ] ) storage.metadata_fetcher_add([FETCHER]) deposit_cur = None handle_row(copy.deepcopy(row), storage, deposit_cur, dry_run=False) revision_swhid = CoreSWHID.from_string( "swh:1:rev:000004d6382c4ad4c0519266626c36551f0e51ca" ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=PYPI_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2019, 1, 23, 22, 10, 55, tzinfo=datetime.timezone.utc, ), authority=PYPI_AUTHORITY, fetcher=FETCHER, format="pypi-project-json", metadata=json.dumps(extrinsic_metadata).encode(), origin=None, revision=revision_swhid, ), ], next_page_token=None, ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=SWH_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2019, 1, 23, 22, 10, 55, tzinfo=datetime.timezone.utc, ), authority=SWH_AUTHORITY, fetcher=FETCHER, format="original-artifacts-json", metadata=json.dumps(dest_original_artifacts).encode(), origin=None, revision=revision_swhid, ), ], next_page_token=None, ) def test_pypi_3(mocker): """Tests loading a revision generated by a very old PyPI loader that does not have a provider or has 'project' metadata.""" mocker.patch( "swh.storage.migrate_extrinsic_metadata.urlopen", side_effect=urllib.error.HTTPError(None, 404, "Not Found", None, None), ) source_original_artifact = { "url": "https://files.pythonhosted.org/packages/34/4f/30087f22eaae8ad7077a28ce157342745a2977e264b8a8e4e7f804a8aa5e/PyPDFLite-0.1.32.tar.gz", "date": "2014-05-07T22:03:00", "sha1": "3289269f75b4111dd00eaea53e00330db9a1db12", "size": 46644, "sha256": "911497d655cf7ef6530c5b57773dad7da97e21cf4d608ad9ad1e38bd7bec7824", "filename": "PyPDFLite-0.1.32.tar.gz", "sha1_git": "1e5c38014731242cfa8594839bcba8a0c4e158c5", "blake2s256": "45792e57873f56d385c694e36c98a580cbba60d5ea91eb6fd0a2d1c71c1fb385", "archive_type": "tar", } dest_original_artifacts = [ { "url": "https://files.pythonhosted.org/packages/34/4f/30087f22eaae8ad7077a28ce157342745a2977e264b8a8e4e7f804a8aa5e/PyPDFLite-0.1.32.tar.gz", "filename": "PyPDFLite-0.1.32.tar.gz", "archive_type": "tar", "length": 46644, "checksums": { "sha1": "3289269f75b4111dd00eaea53e00330db9a1db12", "sha256": "911497d655cf7ef6530c5b57773dad7da97e21cf4d608ad9ad1e38bd7bec7824", "sha1_git": "1e5c38014731242cfa8594839bcba8a0c4e158c5", "blake2s256": "45792e57873f56d385c694e36c98a580cbba60d5ea91eb6fd0a2d1c71c1fb385", }, } ] row = { "id": b"N\xa9\x91|\xdfS\xcd\x13SJ\x04.N\xb3x{\x86\xc84\xd2", "directory": DIRECTORY_ID, "date": datetime.datetime(2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc), "committer_date": datetime.datetime( 2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc ), "type": "tar", "message": b"0.1.32", "metadata": {"original_artifact": source_original_artifact}, } origin_url = "https://pypi.org/project/PyPDFLite/" storage = get_storage("memory") storage.origin_add([Origin(url=origin_url)]) storage.metadata_authority_add( [ attr.evolve(PYPI_AUTHORITY, metadata={}), attr.evolve(SWH_AUTHORITY, metadata={}), ] ) storage.metadata_fetcher_add([FETCHER]) deposit_cur = None handle_row(copy.deepcopy(row), storage, deposit_cur, dry_run=False) revision_swhid = CoreSWHID.from_string( "swh:1:rev:4ea9917cdf53cd13534a042e4eb3787b86c834d2" ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=PYPI_AUTHORITY, ) == PagedResult(results=[], next_page_token=None,) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=SWH_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc, ), authority=SWH_AUTHORITY, fetcher=FETCHER, format="original-artifacts-json", metadata=json.dumps(dest_original_artifacts).encode(), origin=None, revision=revision_swhid, ), ], next_page_token=None, ) def test_pypi_good_origin(): """Tests loading a revision whose origin we can find""" source_original_artifact = { "url": "https://files.pythonhosted.org/packages/34/4f/30087f22eaae8ad7077a28ce157342745a2977e264b8a8e4e7f804a8aa5e/PyPDFLite-0.1.32.tar.gz", "date": "2014-05-07T22:03:00", "sha1": "3289269f75b4111dd00eaea53e00330db9a1db12", "size": 46644, "sha256": "911497d655cf7ef6530c5b57773dad7da97e21cf4d608ad9ad1e38bd7bec7824", "filename": "PyPDFLite-0.1.32.tar.gz", "sha1_git": "1e5c38014731242cfa8594839bcba8a0c4e158c5", "blake2s256": "45792e57873f56d385c694e36c98a580cbba60d5ea91eb6fd0a2d1c71c1fb385", "archive_type": "tar", } dest_original_artifacts = [ { "url": "https://files.pythonhosted.org/packages/34/4f/30087f22eaae8ad7077a28ce157342745a2977e264b8a8e4e7f804a8aa5e/PyPDFLite-0.1.32.tar.gz", "filename": "PyPDFLite-0.1.32.tar.gz", "archive_type": "tar", "length": 46644, "checksums": { "sha1": "3289269f75b4111dd00eaea53e00330db9a1db12", "sha256": "911497d655cf7ef6530c5b57773dad7da97e21cf4d608ad9ad1e38bd7bec7824", "sha1_git": "1e5c38014731242cfa8594839bcba8a0c4e158c5", "blake2s256": "45792e57873f56d385c694e36c98a580cbba60d5ea91eb6fd0a2d1c71c1fb385", }, } ] revision_id = b"N\xa9\x91|\xdfS\xcd\x13SJ\x04.N\xb3x{\x86\xc84\xd2" row = { "id": revision_id, "directory": DIRECTORY_ID, "date": datetime.datetime(2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc), "committer_date": datetime.datetime( 2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc ), "type": "tar", "message": b"0.1.32", "metadata": {"original_artifact": source_original_artifact}, } origin_url = "https://pypi.org/project/PyPDFLite/" storage = get_storage("memory") snapshot_id = b"42" * 10 storage.origin_add([Origin(url=origin_url)]) storage.origin_visit_add( [OriginVisit(origin=origin_url, visit=1, date=now(), type="pypi")] ) storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin_url, visit=1, date=now(), status="partial", snapshot=snapshot_id, ) ] ) storage.snapshot_add( [ Snapshot( id=snapshot_id, branches={ b"foo": SnapshotBranch( target_type=TargetType.REVISION, target=revision_id, ) }, ) ] ) storage.metadata_authority_add( [ attr.evolve(PYPI_AUTHORITY, metadata={}), attr.evolve(SWH_AUTHORITY, metadata={}), ] ) storage.metadata_fetcher_add([FETCHER]) deposit_cur = None handle_row(copy.deepcopy(row), storage, deposit_cur, dry_run=False) revision_swhid = CoreSWHID.from_string( "swh:1:rev:4ea9917cdf53cd13534a042e4eb3787b86c834d2" ) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=PYPI_AUTHORITY, ) == PagedResult(results=[], next_page_token=None,) assert storage.raw_extrinsic_metadata_get( DIRECTORY_SWHID, authority=SWH_AUTHORITY, ) == PagedResult( results=[ RawExtrinsicMetadata( target=DIRECTORY_SWHID, discovery_date=datetime.datetime( 2014, 5, 7, 22, 3, tzinfo=datetime.timezone.utc, ), authority=SWH_AUTHORITY, fetcher=FETCHER, format="original-artifacts-json", metadata=json.dumps(dest_original_artifacts).encode(), origin=origin_url, revision=revision_swhid, ), ], next_page_token=None, ) diff --git a/swh/storage/tests/storage_tests.py b/swh/storage/tests/storage_tests.py index cd270aed..c56305b0 100644 --- a/swh/storage/tests/storage_tests.py +++ b/swh/storage/tests/storage_tests.py @@ -1,4617 +1,4695 @@ # Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from collections import defaultdict import datetime from datetime import timedelta import inspect import itertools import math import random from typing import Any, ClassVar, Dict, Iterator, Optional from unittest.mock import MagicMock import attr from hypothesis import HealthCheck, given, settings, strategies import pytest +from swh.core.api import RemoteException from swh.core.api.classes import stream_results from swh.model import from_disk from swh.model.hashutil import DEFAULT_ALGORITHMS, hash_to_bytes from swh.model.hypothesis_strategies import objects from swh.model.identifiers import CoreSWHID, ObjectType from swh.model.model import ( Content, Directory, ExtID, Origin, OriginVisit, OriginVisitStatus, Person, RawExtrinsicMetadata, Revision, SkippedContent, Snapshot, SnapshotBranch, TargetType, ) from swh.storage import get_storage +from swh.storage.cassandra.storage import CassandraStorage from swh.storage.common import origin_url_to_sha1 as sha1 from swh.storage.exc import HashCollision, StorageArgumentException +from swh.storage.in_memory import InMemoryStorage from swh.storage.interface import ListOrder, PagedResult, StorageInterface from swh.storage.tests.conftest import function_scoped_fixture_check from swh.storage.utils import ( content_hex_hashes, now, remove_keys, round_to_milliseconds, ) def transform_entries( storage: StorageInterface, dir_: Directory, *, prefix: bytes = b"" ) -> Iterator[Dict[str, Any]]: """Iterate through a directory's entries, and yields the items 'directory_ls' is expected to return; including content metadata for file entries.""" for ent in dir_.entries: if ent.type == "dir": yield { "dir_id": dir_.id, "type": ent.type, "target": ent.target, "name": prefix + ent.name, "perms": ent.perms, "status": None, "sha1": None, "sha1_git": None, "sha256": None, "length": None, } elif ent.type == "file": contents = storage.content_find({"sha1_git": ent.target}) assert contents ent_dict = contents[0].to_dict() for key in ["ctime", "blake2s256"]: ent_dict.pop(key, None) ent_dict.update( { "dir_id": dir_.id, "type": ent.type, "target": ent.target, "name": prefix + ent.name, "perms": ent.perms, } ) yield ent_dict def assert_contents_ok( expected_contents, actual_contents, keys_to_check={"sha1", "data"} ): """Assert that a given list of contents matches on a given set of keys. """ for k in keys_to_check: expected_list = set([c.get(k) for c in expected_contents]) actual_list = set([c.get(k) for c in actual_contents]) assert actual_list == expected_list, k class LazyContent(Content): def with_data(self): return Content.from_dict({**self.to_dict(), "data": b"42\n"}) class TestStorage: """Main class for Storage testing. This class is used as-is to test local storage (see TestLocalStorage below) and remote storage (see TestRemoteStorage in test_remote_storage.py. We need to have the two classes inherit from this base class separately to avoid nosetests running the tests from the base class twice. """ maxDiff = None # type: ClassVar[Optional[int]] def test_types(self, swh_storage_backend_config): """Checks all methods of StorageInterface are implemented by this backend, and that they have the same signature.""" # Create an instance of the protocol (which cannot be instantiated # directly, so this creates a subclass, then instantiates it) interface = type("_", (StorageInterface,), {})() storage = get_storage(**swh_storage_backend_config) assert "content_add" in dir(interface) missing_methods = [] for meth_name in dir(interface): if meth_name.startswith("_"): continue interface_meth = getattr(interface, meth_name) try: concrete_meth = getattr(storage, meth_name) except AttributeError: if not getattr(interface_meth, "deprecated_endpoint", False): # The backend is missing a (non-deprecated) endpoint missing_methods.append(meth_name) continue expected_signature = inspect.signature(interface_meth) actual_signature = inspect.signature(concrete_meth) assert expected_signature == actual_signature, meth_name assert missing_methods == [] # If all the assertions above succeed, then this one should too. # But there's no harm in double-checking. # And we could replace the assertions above by this one, but unlike # the assertions above, it doesn't explain what is missing. assert isinstance(storage, StorageInterface) def test_check_config(self, swh_storage): assert swh_storage.check_config(check_write=True) assert swh_storage.check_config(check_write=False) def test_content_add(self, swh_storage, sample_data): cont = sample_data.content insertion_start_time = now() actual_result = swh_storage.content_add([cont]) insertion_end_time = now() assert actual_result == { "content:add": 1, "content:add:bytes": cont.length, } assert swh_storage.content_get_data(cont.sha1) == cont.data expected_cont = attr.evolve(cont, data=None) contents = [ obj for (obj_type, obj) in swh_storage.journal_writer.journal.objects if obj_type == "content" ] assert len(contents) == 1 for obj in contents: assert insertion_start_time <= obj.ctime assert obj.ctime <= insertion_end_time assert obj == expected_cont - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["content"] == 1 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["content"] == 1 def test_content_add_from_lazy_content(self, swh_storage, sample_data): cont = sample_data.content lazy_content = LazyContent.from_dict(cont.to_dict()) insertion_start_time = now() actual_result = swh_storage.content_add([lazy_content]) insertion_end_time = now() assert actual_result == { "content:add": 1, "content:add:bytes": cont.length, } # the fact that we retrieve the content object from the storage with # the correct 'data' field ensures it has been 'called' assert swh_storage.content_get_data(cont.sha1) == cont.data expected_cont = attr.evolve(lazy_content, data=None, ctime=None) contents = [ obj for (obj_type, obj) in swh_storage.journal_writer.journal.objects if obj_type == "content" ] assert len(contents) == 1 for obj in contents: assert insertion_start_time <= obj.ctime assert obj.ctime <= insertion_end_time assert attr.evolve(obj, ctime=None).to_dict() == expected_cont.to_dict() - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["content"] == 1 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["content"] == 1 def test_content_get_data_missing(self, swh_storage, sample_data): cont, cont2 = sample_data.contents[:2] swh_storage.content_add([cont]) # Query a single missing content actual_content_data = swh_storage.content_get_data(cont2.sha1) assert actual_content_data is None # Check content_get does not abort after finding a missing content actual_content_data = swh_storage.content_get_data(cont.sha1) assert actual_content_data == cont.data actual_content_data = swh_storage.content_get_data(cont2.sha1) assert actual_content_data is None def test_content_add_different_input(self, swh_storage, sample_data): cont, cont2 = sample_data.contents[:2] actual_result = swh_storage.content_add([cont, cont2]) assert actual_result == { "content:add": 2, "content:add:bytes": cont.length + cont2.length, } def test_content_add_twice(self, swh_storage, sample_data): cont, cont2 = sample_data.contents[:2] actual_result = swh_storage.content_add([cont]) assert actual_result == { "content:add": 1, "content:add:bytes": cont.length, } assert len(swh_storage.journal_writer.journal.objects) == 1 actual_result = swh_storage.content_add([cont, cont2]) assert actual_result == { "content:add": 1, "content:add:bytes": cont2.length, } assert 2 <= len(swh_storage.journal_writer.journal.objects) <= 3 assert len(swh_storage.content_find(cont.to_dict())) == 1 assert len(swh_storage.content_find(cont2.to_dict())) == 1 def test_content_add_collision(self, swh_storage, sample_data): cont1 = sample_data.content # create (corrupted) content with same sha1{,_git} but != sha256 sha256_array = bytearray(cont1.sha256) sha256_array[0] += 1 cont1b = attr.evolve(cont1, sha256=bytes(sha256_array)) with pytest.raises(HashCollision) as cm: swh_storage.content_add([cont1, cont1b]) exc = cm.value actual_algo = exc.algo assert actual_algo in ["sha1", "sha1_git"] actual_id = exc.hash_id assert actual_id == getattr(cont1, actual_algo).hex() collisions = exc.args[2] assert len(collisions) == 2 assert collisions == [ content_hex_hashes(cont1.hashes()), content_hex_hashes(cont1b.hashes()), ] assert exc.colliding_content_hashes() == [ cont1.hashes(), cont1b.hashes(), ] def test_content_add_duplicate(self, swh_storage, sample_data): cont = sample_data.content swh_storage.content_add([cont, cont]) assert swh_storage.content_get_data(cont.sha1) == cont.data def test_content_update(self, swh_storage, sample_data): cont1 = sample_data.content if hasattr(swh_storage, "journal_writer"): swh_storage.journal_writer.journal = None # TODO, not supported swh_storage.content_add([cont1]) # alter the sha1_git for example cont1b = attr.evolve( cont1, sha1_git=hash_to_bytes("3a60a5275d0333bf13468e8b3dcab90f4046e654") ) swh_storage.content_update([cont1b.to_dict()], keys=["sha1_git"]) actual_contents = swh_storage.content_get([cont1.sha1]) expected_content = attr.evolve(cont1b, data=None) assert actual_contents == [expected_content] def test_content_add_metadata(self, swh_storage, sample_data): cont = attr.evolve(sample_data.content, data=None, ctime=now()) actual_result = swh_storage.content_add_metadata([cont]) assert actual_result == { "content:add": 1, } expected_cont = cont assert swh_storage.content_get([cont.sha1]) == [expected_cont] contents = [ obj for (obj_type, obj) in swh_storage.journal_writer.journal.objects if obj_type == "content" ] assert len(contents) == 1 for obj in contents: obj = attr.evolve(obj, ctime=None) assert obj == cont def test_content_add_metadata_different_input(self, swh_storage, sample_data): contents = sample_data.contents[:2] cont = attr.evolve(contents[0], data=None, ctime=now()) cont2 = attr.evolve(contents[1], data=None, ctime=now()) actual_result = swh_storage.content_add_metadata([cont, cont2]) assert actual_result == { "content:add": 2, } def test_content_add_metadata_collision(self, swh_storage, sample_data): cont1 = attr.evolve(sample_data.content, data=None, ctime=now()) # create (corrupted) content with same sha1{,_git} but != sha256 sha1_git_array = bytearray(cont1.sha256) sha1_git_array[0] += 1 cont1b = attr.evolve(cont1, sha256=bytes(sha1_git_array)) with pytest.raises(HashCollision) as cm: swh_storage.content_add_metadata([cont1, cont1b]) exc = cm.value actual_algo = exc.algo assert actual_algo in ["sha1", "sha1_git", "blake2s256"] actual_id = exc.hash_id assert actual_id == getattr(cont1, actual_algo).hex() collisions = exc.args[2] assert len(collisions) == 2 assert collisions == [ content_hex_hashes(cont1.hashes()), content_hex_hashes(cont1b.hashes()), ] assert exc.colliding_content_hashes() == [ cont1.hashes(), cont1b.hashes(), ] def test_content_add_objstorage_first(self, swh_storage, sample_data): """Tests the objstorage is written to before the DB and journal""" cont = sample_data.content swh_storage.objstorage.content_add = MagicMock(side_effect=Exception("Oops")) # Try to add, but the objstorage crashes try: swh_storage.content_add([cont]) except Exception: pass # The DB must be written to after the objstorage, so the DB should be # unchanged if the objstorage crashed assert swh_storage.content_get_data(cont.sha1) is None # The journal too assert list(swh_storage.journal_writer.journal.objects) == [] def test_skipped_content_add(self, swh_storage, sample_data): contents = sample_data.skipped_contents[:2] cont = contents[0] cont2 = attr.evolve(contents[1], blake2s256=None) contents_dict = [c.to_dict() for c in [cont, cont2]] missing = list(swh_storage.skipped_content_missing(contents_dict)) assert missing == [cont.hashes(), cont2.hashes()] actual_result = swh_storage.skipped_content_add([cont, cont, cont2]) assert 2 <= actual_result.pop("skipped_content:add") <= 3 assert actual_result == {} missing = list(swh_storage.skipped_content_missing(contents_dict)) assert missing == [] def test_skipped_content_add_missing_hashes(self, swh_storage, sample_data): cont, cont2 = [ attr.evolve(c, sha1_git=None) for c in sample_data.skipped_contents[:2] ] contents_dict = [c.to_dict() for c in [cont, cont2]] missing = list(swh_storage.skipped_content_missing(contents_dict)) assert len(missing) == 2 actual_result = swh_storage.skipped_content_add([cont, cont, cont2]) assert 2 <= actual_result.pop("skipped_content:add") <= 3 assert actual_result == {} missing = list(swh_storage.skipped_content_missing(contents_dict)) assert missing == [] def test_skipped_content_missing_partial_hash(self, swh_storage, sample_data): cont = sample_data.skipped_content cont2 = attr.evolve(cont, sha1_git=None) contents_dict = [c.to_dict() for c in [cont, cont2]] missing = list(swh_storage.skipped_content_missing(contents_dict)) assert len(missing) == 2 actual_result = swh_storage.skipped_content_add([cont]) assert actual_result.pop("skipped_content:add") == 1 assert actual_result == {} missing = list(swh_storage.skipped_content_missing(contents_dict)) assert missing == [cont2.hashes()] @pytest.mark.property_based @settings( deadline=None, # this test is very slow suppress_health_check=function_scoped_fixture_check, ) @given( strategies.sets( elements=strategies.sampled_from(["sha256", "sha1_git", "blake2s256"]), min_size=0, ) ) def test_content_missing(self, swh_storage, sample_data, algos): algos |= {"sha1"} content, missing_content = [sample_data.content2, sample_data.skipped_content] swh_storage.content_add([content]) test_contents = [content.to_dict()] missing_per_hash = defaultdict(list) for i in range(256): test_content = missing_content.to_dict() for hash in algos: test_content[hash] = bytes([i]) + test_content[hash][1:] missing_per_hash[hash].append(test_content[hash]) test_contents.append(test_content) assert set(swh_storage.content_missing(test_contents)) == set( missing_per_hash["sha1"] ) for hash in algos: assert set( swh_storage.content_missing(test_contents, key_hash=hash) ) == set(missing_per_hash[hash]) @pytest.mark.property_based @settings(suppress_health_check=function_scoped_fixture_check,) @given( strategies.sets( elements=strategies.sampled_from(["sha256", "sha1_git", "blake2s256"]), min_size=0, ) ) def test_content_missing_unknown_algo(self, swh_storage, sample_data, algos): algos |= {"sha1"} content, missing_content = [sample_data.content2, sample_data.skipped_content] swh_storage.content_add([content]) test_contents = [content.to_dict()] missing_per_hash = defaultdict(list) for i in range(16): test_content = missing_content.to_dict() for hash in algos: test_content[hash] = bytes([i]) + test_content[hash][1:] missing_per_hash[hash].append(test_content[hash]) test_content["nonexisting_algo"] = b"\x00" test_contents.append(test_content) assert set(swh_storage.content_missing(test_contents)) == set( missing_per_hash["sha1"] ) for hash in algos: assert set( swh_storage.content_missing(test_contents, key_hash=hash) ) == set(missing_per_hash[hash]) def test_content_missing_per_sha1(self, swh_storage, sample_data): # given cont = sample_data.content cont2 = sample_data.content2 missing_cont = sample_data.skipped_content missing_cont2 = sample_data.skipped_content2 swh_storage.content_add([cont, cont2]) # when gen = swh_storage.content_missing_per_sha1( [cont.sha1, missing_cont.sha1, cont2.sha1, missing_cont2.sha1] ) # then assert list(gen) == [missing_cont.sha1, missing_cont2.sha1] def test_content_missing_per_sha1_git(self, swh_storage, sample_data): cont, cont2 = sample_data.contents[:2] missing_cont = sample_data.skipped_content missing_cont2 = sample_data.skipped_content2 swh_storage.content_add([cont, cont2]) contents = [ cont.sha1_git, cont2.sha1_git, missing_cont.sha1_git, missing_cont2.sha1_git, ] missing_contents = swh_storage.content_missing_per_sha1_git(contents) assert list(missing_contents) == [missing_cont.sha1_git, missing_cont2.sha1_git] missing_contents = swh_storage.content_missing_per_sha1_git([]) assert list(missing_contents) == [] def test_content_get_partition(self, swh_storage, swh_contents): """content_get_partition paginates results if limit exceeded""" expected_contents = [ attr.evolve(c, data=None) for c in swh_contents if c.status != "absent" ] actual_contents = [] for i in range(16): actual_result = swh_storage.content_get_partition(i, 16) assert actual_result.next_page_token is None actual_contents.extend(actual_result.results) assert len(actual_contents) == len(expected_contents) for content in actual_contents: assert content in expected_contents assert content.ctime is None def test_content_get_partition_full(self, swh_storage, swh_contents): """content_get_partition for a single partition returns all available contents """ expected_contents = [ attr.evolve(c, data=None) for c in swh_contents if c.status != "absent" ] actual_result = swh_storage.content_get_partition(0, 1) assert actual_result.next_page_token is None actual_contents = actual_result.results assert len(actual_contents) == len(expected_contents) for content in actual_contents: assert content in expected_contents def test_content_get_partition_empty(self, swh_storage, swh_contents): """content_get_partition when at least one of the partitions is empty""" expected_contents = { cont.sha1 for cont in swh_contents if cont.status != "absent" } # nb_partitions = smallest power of 2 such that at least one of # the partitions is empty nb_partitions = 1 << math.floor(math.log2(len(swh_contents)) + 1) seen_sha1s = [] for i in range(nb_partitions): actual_result = swh_storage.content_get_partition( i, nb_partitions, limit=len(swh_contents) + 1 ) for content in actual_result.results: seen_sha1s.append(content.sha1) # Limit is higher than the max number of results assert actual_result.next_page_token is None assert set(seen_sha1s) == expected_contents def test_content_get_partition_limit_none(self, swh_storage): """content_get_partition call with wrong limit input should fail""" with pytest.raises(StorageArgumentException, match="limit should not be None"): swh_storage.content_get_partition(1, 16, limit=None) def test_content_get_partition_pagination_generate(self, swh_storage, swh_contents): """content_get_partition returns contents within range provided""" expected_contents = [ attr.evolve(c, data=None) for c in swh_contents if c.status != "absent" ] # retrieve contents actual_contents = [] for i in range(4): page_token = None while True: actual_result = swh_storage.content_get_partition( i, 4, limit=3, page_token=page_token ) actual_contents.extend(actual_result.results) page_token = actual_result.next_page_token if page_token is None: break assert len(actual_contents) == len(expected_contents) for content in actual_contents: assert content in expected_contents @pytest.mark.parametrize("algo", sorted(DEFAULT_ALGORITHMS)) def test_content_get(self, swh_storage, sample_data, algo): cont1, cont2 = sample_data.contents[:2] swh_storage.content_add([cont1, cont2]) actual_contents = swh_storage.content_get( [getattr(cont1, algo), getattr(cont2, algo)], algo ) # we only retrieve the metadata so no data nor ctime within expected_contents = [attr.evolve(c, data=None) for c in [cont1, cont2]] assert actual_contents == expected_contents for content in actual_contents: assert content.ctime is None @pytest.mark.parametrize("algo", sorted(DEFAULT_ALGORITHMS)) def test_content_get_missing(self, swh_storage, sample_data, algo): cont1, cont2 = sample_data.contents[:2] assert cont1.sha1 != cont2.sha1 missing_cont = sample_data.skipped_content swh_storage.content_add([cont1, cont2]) actual_contents = swh_storage.content_get( [getattr(cont1, algo), getattr(cont2, algo), getattr(missing_cont, algo)], algo, ) expected_contents = [ attr.evolve(c, data=None) if c else None for c in [cont1, cont2, None] ] assert actual_contents == expected_contents def test_content_get_random(self, swh_storage, sample_data): cont, cont2, cont3 = sample_data.contents[:3] swh_storage.content_add([cont, cont2, cont3]) assert swh_storage.content_get_random() in { cont.sha1_git, cont2.sha1_git, cont3.sha1_git, } def test_directory_add(self, swh_storage, sample_data): content = sample_data.content - directory = sample_data.directories[1] + directory = sample_data.directory assert directory.entries[0].target == content.sha1_git swh_storage.content_add([content]) init_missing = list(swh_storage.directory_missing([directory.id])) assert [directory.id] == init_missing actual_result = swh_storage.directory_add([directory]) assert actual_result == {"directory:add": 1} assert ("directory", directory) in list( swh_storage.journal_writer.journal.objects ) actual_data = list(swh_storage.directory_ls(directory.id)) expected_data = list(transform_entries(swh_storage, directory)) for data in actual_data: assert data in expected_data after_missing = list(swh_storage.directory_missing([directory.id])) assert after_missing == [] - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["directory"] == 1 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["directory"] == 1 def test_directory_add_twice(self, swh_storage, sample_data): directory = sample_data.directories[1] actual_result = swh_storage.directory_add([directory]) assert actual_result == {"directory:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("directory", directory) ] actual_result = swh_storage.directory_add([directory]) assert actual_result == {"directory:add": 0} assert list(swh_storage.journal_writer.journal.objects) == [ ("directory", directory) ] def test_directory_ls_recursive(self, swh_storage, sample_data): # create consistent dataset regarding the directories we want to list content, content2 = sample_data.contents[:2] swh_storage.content_add([content, content2]) dir1, dir2, dir3 = sample_data.directories[:3] dir_ids = [d.id for d in [dir1, dir2, dir3]] init_missing = list(swh_storage.directory_missing(dir_ids)) assert init_missing == dir_ids actual_result = swh_storage.directory_add([dir1, dir2, dir3]) assert actual_result == {"directory:add": 3} # List directory containing one file actual_data = list(swh_storage.directory_ls(dir1.id, recursive=True)) expected_data = list(transform_entries(swh_storage, dir1)) for data in actual_data: assert data in expected_data # List directory containing a file and an unknown subdirectory actual_data = list(swh_storage.directory_ls(dir2.id, recursive=True)) expected_data = list(transform_entries(swh_storage, dir2)) for data in actual_data: assert data in expected_data # List directory containing both a known and unknown subdirectory, entries # should be both those of the directory and of the known subdir (up to contents) actual_data = list(swh_storage.directory_ls(dir3.id, recursive=True)) expected_data = list( itertools.chain( transform_entries(swh_storage, dir3), transform_entries(swh_storage, dir2, prefix=b"subdir/"), ) ) for data in actual_data: assert data in expected_data def test_directory_ls_non_recursive(self, swh_storage, sample_data): # create consistent dataset regarding the directories we want to list content, content2 = sample_data.contents[:2] swh_storage.content_add([content, content2]) dir1, dir2, dir3, _, dir5 = sample_data.directories[:5] dir_ids = [d.id for d in [dir1, dir2, dir3, dir5]] init_missing = list(swh_storage.directory_missing(dir_ids)) assert init_missing == dir_ids actual_result = swh_storage.directory_add([dir1, dir2, dir3, dir5]) assert actual_result == {"directory:add": 4} # List directory containing a file and an unknown subdirectory actual_data = list(swh_storage.directory_ls(dir1.id)) expected_data = list(transform_entries(swh_storage, dir1)) for data in actual_data: assert data in expected_data # List directory containing a single file actual_data = list(swh_storage.directory_ls(dir2.id)) expected_data = list(transform_entries(swh_storage, dir2)) for data in actual_data: assert data in expected_data # List directory containing a known subdirectory, entries should # only be those of the parent directory, not of the subdir actual_data = list(swh_storage.directory_ls(dir3.id)) expected_data = list(transform_entries(swh_storage, dir3)) for data in actual_data: assert data in expected_data def test_directory_ls_missing_content(self, swh_storage, sample_data): swh_storage.directory_add([sample_data.directory2]) assert list(swh_storage.directory_ls(sample_data.directory2.id)) == [ { "dir_id": sample_data.directory2.id, "length": None, "name": b"oof", "perms": 33188, "sha1": None, "sha1_git": None, "sha256": None, "status": None, "target": sample_data.directory2.entries[0].target, "type": "file", }, ] def test_directory_ls_skipped_content(self, swh_storage, sample_data): swh_storage.directory_add([sample_data.directory2]) cont = SkippedContent( sha1_git=sample_data.directory2.entries[0].target, sha1=b"c" * 20, sha256=None, blake2s256=None, length=42, status="absent", reason="You need a premium subscription to access this content", ) swh_storage.skipped_content_add([cont]) assert list(swh_storage.directory_ls(sample_data.directory2.id)) == [ { "dir_id": sample_data.directory2.id, "length": 42, "name": b"oof", "perms": 33188, "sha1": b"c" * 20, "sha1_git": sample_data.directory2.entries[0].target, "sha256": None, "status": "absent", "target": sample_data.directory2.entries[0].target, "type": "file", }, ] def test_directory_entry_get_by_path(self, swh_storage, sample_data): cont, content2 = sample_data.contents[:2] dir1, dir2, dir3, dir4, dir5 = sample_data.directories[:5] # given dir_ids = [d.id for d in [dir1, dir2, dir3, dir4, dir5]] init_missing = list(swh_storage.directory_missing(dir_ids)) assert init_missing == dir_ids actual_result = swh_storage.directory_add([dir3, dir4]) assert actual_result == {"directory:add": 2} expected_entries = [ { "dir_id": dir3.id, "name": b"foo", "type": "file", "target": cont.sha1_git, "sha1": None, "sha1_git": None, "sha256": None, "status": None, "perms": from_disk.DentryPerms.content, "length": None, }, { "dir_id": dir3.id, "name": b"subdir", "type": "dir", "target": dir2.id, "sha1": None, "sha1_git": None, "sha256": None, "status": None, "perms": from_disk.DentryPerms.directory, "length": None, }, { "dir_id": dir3.id, "name": b"hello", "type": "file", "target": content2.sha1_git, "sha1": None, "sha1_git": None, "sha256": None, "status": None, "perms": from_disk.DentryPerms.content, "length": None, }, ] # when (all must be found here) for entry, expected_entry in zip(dir3.entries, expected_entries): actual_entry = swh_storage.directory_entry_get_by_path( dir3.id, [entry.name] ) assert actual_entry == expected_entry # same, but deeper for entry, expected_entry in zip(dir3.entries, expected_entries): actual_entry = swh_storage.directory_entry_get_by_path( dir4.id, [b"subdir1", entry.name] ) expected_entry = expected_entry.copy() expected_entry["name"] = b"subdir1/" + expected_entry["name"] assert actual_entry == expected_entry # when (nothing should be found here since `dir` is not persisted.) for entry in dir2.entries: actual_entry = swh_storage.directory_entry_get_by_path( dir2.id, [entry.name] ) assert actual_entry is None def test_directory_get_entries_pagination(self, swh_storage, sample_data): - # Note: this test assumes entries are returned in lexicographic order, - # which is not actually guaranteed by the interface. dir_ = sample_data.directory3 entries = sorted(dir_.entries, key=lambda entry: entry.name) swh_storage.directory_add(sample_data.directories) # No pagination needed actual_data = swh_storage.directory_get_entries(dir_.id) - assert actual_data == PagedResult(results=entries, next_page_token=None) + assert sorted(actual_data.results) == sorted(entries) + assert actual_data.next_page_token is None, actual_data # A little pagination actual_data = swh_storage.directory_get_entries(dir_.id, limit=2) - assert actual_data.results == entries[0:2] - assert actual_data.next_page_token is not None + assert len(actual_data.results) == 2, actual_data + assert actual_data.next_page_token is not None, actual_data + + all_results = list(actual_data.results) actual_data = swh_storage.directory_get_entries( dir_.id, page_token=actual_data.next_page_token ) - assert actual_data == PagedResult(results=entries[2:], next_page_token=None) + assert len(actual_data.results) == len(entries) - 2, actual_data + assert actual_data.next_page_token is None, actual_data + + all_results.extend(actual_data.results) + assert sorted(all_results) == sorted(entries) @pytest.mark.parametrize("limit", [1, 2, 3, 4, 5]) def test_directory_get_entries(self, swh_storage, sample_data, limit): dir_ = sample_data.directory3 swh_storage.directory_add(sample_data.directories) actual_data = list( stream_results(swh_storage.directory_get_entries, dir_.id, limit=limit,) ) assert sorted(actual_data) == sorted(dir_.entries) def test_directory_get_random(self, swh_storage, sample_data): dir1, dir2, dir3 = sample_data.directories[:3] swh_storage.directory_add([dir1, dir2, dir3]) assert swh_storage.directory_get_random() in { dir1.id, dir2.id, dir3.id, } def test_revision_add(self, swh_storage, sample_data): revision = sample_data.revision init_missing = swh_storage.revision_missing([revision.id]) assert list(init_missing) == [revision.id] actual_result = swh_storage.revision_add([revision]) assert actual_result == {"revision:add": 1} end_missing = swh_storage.revision_missing([revision.id]) assert list(end_missing) == [] assert list(swh_storage.journal_writer.journal.objects) == [ ("revision", revision) ] # already there so nothing added actual_result = swh_storage.revision_add([revision]) assert actual_result == {"revision:add": 0} - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["revision"] == 1 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["revision"] == 1 def test_revision_add_twice(self, swh_storage, sample_data): revision, revision2 = sample_data.revisions[:2] actual_result = swh_storage.revision_add([revision]) assert actual_result == {"revision:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("revision", revision) ] actual_result = swh_storage.revision_add([revision, revision2]) assert actual_result == {"revision:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("revision", revision), ("revision", revision2), ] def test_revision_add_name_clash(self, swh_storage, sample_data): revision, revision2 = sample_data.revisions[:2] revision1 = attr.evolve( revision, author=Person( fullname=b"John Doe ", name=b"John Doe", email=b"john.doe@example.com", ), ) revision2 = attr.evolve( revision2, author=Person( fullname=b"John Doe ", name=b"John Doe ", email=b"john.doe@example.com ", ), ) actual_result = swh_storage.revision_add([revision1, revision2]) assert actual_result == {"revision:add": 2} def test_revision_get_order(self, swh_storage, sample_data): revision, revision2 = sample_data.revisions[:2] add_result = swh_storage.revision_add([revision, revision2]) assert add_result == {"revision:add": 2} # order 1 actual_revisions = swh_storage.revision_get([revision.id, revision2.id]) assert actual_revisions == [revision, revision2] # order 2 actual_revisions2 = swh_storage.revision_get([revision2.id, revision.id]) assert actual_revisions2 == [revision2, revision] def test_revision_log(self, swh_storage, sample_data): revision1, revision2, revision3, revision4 = sample_data.revisions[:4] # rev4 -is-child-of-> rev3 -> rev1, (rev2 -> rev1) swh_storage.revision_add([revision1, revision2, revision3, revision4]) # when results = list(swh_storage.revision_log([revision4.id])) # for comparison purposes actual_results = [Revision.from_dict(r) for r in results] assert len(actual_results) == 4 # rev4 -child-> rev3 -> rev1, (rev2 -> rev1) assert actual_results == [revision4, revision3, revision1, revision2] def test_revision_log_with_limit(self, swh_storage, sample_data): revision1, revision2, revision3, revision4 = sample_data.revisions[:4] # revision4 -is-child-of-> revision3 swh_storage.revision_add([revision3, revision4]) results = list(swh_storage.revision_log([revision4.id], 1)) actual_results = [Revision.from_dict(r) for r in results] assert len(actual_results) == 1 assert actual_results[0] == revision4 def test_revision_log_unknown_revision(self, swh_storage, sample_data): revision = sample_data.revision rev_log = list(swh_storage.revision_log([revision.id])) assert rev_log == [] def test_revision_shortlog(self, swh_storage, sample_data): revision1, revision2, revision3, revision4 = sample_data.revisions[:4] # rev4 -is-child-of-> rev3 -> (rev1, rev2); rev2 -> rev1 swh_storage.revision_add([revision1, revision2, revision3, revision4]) results = list(swh_storage.revision_shortlog([revision4.id])) actual_results = [[id, tuple(parents)] for (id, parents) in results] assert len(actual_results) == 4 assert actual_results == [ [revision4.id, revision4.parents], [revision3.id, revision3.parents], [revision1.id, revision1.parents], [revision2.id, revision2.parents], ] def test_revision_shortlog_with_limit(self, swh_storage, sample_data): revision1, revision2, revision3, revision4 = sample_data.revisions[:4] # revision4 -is-child-of-> revision3 swh_storage.revision_add([revision1, revision2, revision3, revision4]) results = list(swh_storage.revision_shortlog([revision4.id], 1)) actual_results = [[id, tuple(parents)] for (id, parents) in results] assert len(actual_results) == 1 assert list(actual_results[0]) == [revision4.id, revision4.parents] def test_revision_get(self, swh_storage, sample_data): revision, revision2 = sample_data.revisions[:2] swh_storage.revision_add([revision]) actual_revisions = swh_storage.revision_get([revision.id, revision2.id]) assert len(actual_revisions) == 2 assert actual_revisions == [revision, None] def test_revision_get_no_parents(self, swh_storage, sample_data): revision = sample_data.revision swh_storage.revision_add([revision]) actual_revision = swh_storage.revision_get([revision.id])[0] assert revision.parents == () assert actual_revision.parents == () # no parents on this one def test_revision_get_random(self, swh_storage, sample_data): revision1, revision2, revision3 = sample_data.revisions[:3] swh_storage.revision_add([revision1, revision2, revision3]) assert swh_storage.revision_get_random() in { revision1.id, revision2.id, revision3.id, } def test_revision_missing_many(self, swh_storage, sample_data): """Large number of revision ids to check can cause ScyllaDB to reject queries.""" revision = sample_data.revision ids = [bytes([b1, b2]) * 10 for b1 in range(256) for b2 in range(10)] ids.append(revision.id) ids.sort() init_missing = swh_storage.revision_missing(ids) assert set(init_missing) == set(ids) actual_result = swh_storage.revision_add([revision]) assert actual_result == {"revision:add": 1} end_missing = swh_storage.revision_missing(ids) assert set(end_missing) == set(ids) - {revision.id} def test_extid_add_git(self, swh_storage, sample_data): gitids = [ revision.id for revision in sample_data.revisions if revision.type.value == "git" ] extids = [ ExtID( extid=gitid, extid_type="git", target=CoreSWHID(object_id=gitid, object_type=ObjectType.REVISION,), ) for gitid in gitids ] assert swh_storage.extid_get_from_extid("git", gitids) == [] assert swh_storage.extid_get_from_target(ObjectType.REVISION, gitids) == [] summary = swh_storage.extid_add(extids) assert summary == {"extid:add": len(gitids)} assert swh_storage.extid_get_from_extid("git", gitids) == extids assert swh_storage.extid_get_from_target(ObjectType.REVISION, gitids) == extids assert swh_storage.extid_get_from_extid("hg", gitids) == [] assert swh_storage.extid_get_from_target(ObjectType.RELEASE, gitids) == [] # check ExtIDs have been added to the journal extids_in_journal = [ obj for (obj_type, obj) in swh_storage.journal_writer.journal.objects if obj_type == "extid" ] assert extids == extids_in_journal def test_extid_add_hg(self, swh_storage, sample_data): def get_node(revision): node = None if revision.extra_headers: node = dict(revision.extra_headers).get(b"node") if node is None and revision.metadata: node = hash_to_bytes(revision.metadata.get("node")) return node swhids = [ revision.id for revision in sample_data.revisions if revision.type.value == "hg" ] extids = [ get_node(revision) for revision in sample_data.revisions if revision.type.value == "hg" ] assert swh_storage.extid_get_from_extid("hg", extids) == [] assert swh_storage.extid_get_from_target(ObjectType.REVISION, swhids) == [] extid_objs = [ ExtID( extid=hgid, extid_type="hg", extid_version=1, target=CoreSWHID(object_id=swhid, object_type=ObjectType.REVISION,), ) for hgid, swhid in zip(extids, swhids) ] summary = swh_storage.extid_add(extid_objs) assert summary == {"extid:add": len(swhids)} assert swh_storage.extid_get_from_extid("hg", extids) == extid_objs assert ( swh_storage.extid_get_from_target(ObjectType.REVISION, swhids) == extid_objs ) assert swh_storage.extid_get_from_extid("git", extids) == [] assert swh_storage.extid_get_from_target(ObjectType.RELEASE, swhids) == [] # check ExtIDs have been added to the journal extids_in_journal = [ obj for (obj_type, obj) in swh_storage.journal_writer.journal.objects if obj_type == "extid" ] assert extid_objs == extids_in_journal def test_extid_add_twice(self, swh_storage, sample_data): gitids = [ revision.id for revision in sample_data.revisions if revision.type.value == "git" ] extids = [ ExtID( extid=gitid, extid_type="git", target=CoreSWHID(object_id=gitid, object_type=ObjectType.REVISION,), ) for gitid in gitids ] summary = swh_storage.extid_add(extids) assert summary == {"extid:add": len(gitids)} # add them again, should be noop summary = swh_storage.extid_add(extids) # assert summary == {"extid:add": 0} assert swh_storage.extid_get_from_extid("git", gitids) == extids assert swh_storage.extid_get_from_target(ObjectType.REVISION, gitids) == extids def test_extid_add_extid_multicity(self, swh_storage, sample_data): ids = [ revision.id for revision in sample_data.revisions if revision.type.value == "git" ] extids = [ ExtID( extid=extid, extid_type="git", extid_version=2, target=CoreSWHID(object_id=extid, object_type=ObjectType.REVISION,), ) for extid in ids ] swh_storage.extid_add(extids) # try to add "modified-extid" versions, should be added extids2 = [ ExtID( extid=extid, extid_type="hg", extid_version=2, target=CoreSWHID(object_id=extid, object_type=ObjectType.REVISION,), ) for extid in ids ] swh_storage.extid_add(extids2) assert swh_storage.extid_get_from_extid("git", ids) == extids assert swh_storage.extid_get_from_extid("hg", ids) == extids2 assert set(swh_storage.extid_get_from_target(ObjectType.REVISION, ids)) == { *extids, *extids2, } def test_extid_add_target_multicity(self, swh_storage, sample_data): ids = [ revision.id for revision in sample_data.revisions if revision.type.value == "git" ] extids = [ ExtID( extid=extid, extid_type="git", target=CoreSWHID(object_id=extid, object_type=ObjectType.REVISION,), ) for extid in ids ] swh_storage.extid_add(extids) # try to add "modified" versions, should be added extids2 = [ ExtID( extid=extid, extid_type="git", target=CoreSWHID(object_id=extid, object_type=ObjectType.RELEASE,), ) for extid in ids ] swh_storage.extid_add(extids2) assert set(swh_storage.extid_get_from_extid("git", ids)) == {*extids, *extids2} assert swh_storage.extid_get_from_target(ObjectType.REVISION, ids) == extids assert swh_storage.extid_get_from_target(ObjectType.RELEASE, ids) == extids2 def test_extid_version_behavior(self, swh_storage, sample_data): ids = [ revision.id for revision in sample_data.revisions if revision.type.value == "git" ] # Insert extids with several different versions extids = [ ExtID( extid=extid, extid_type="git", + extid_version=0, target=CoreSWHID(object_id=extid, object_type=ObjectType.REVISION,), ) for extid in ids ] + [ ExtID( extid=extid, extid_type="git", extid_version=1, target=CoreSWHID(object_id=extid, object_type=ObjectType.REVISION,), ) for extid in ids ] swh_storage.extid_add(extids) # Check that both versions get returned for git_id in ids: objs = swh_storage.extid_get_from_extid("git", [git_id]) assert len(objs) == 2 assert set(obj.extid_version for obj in objs) == {0, 1} for swhid in ids: objs = swh_storage.extid_get_from_target(ObjectType.REVISION, [swhid]) assert len(objs) == 2 assert set(obj.extid_version for obj in objs) == {0, 1} + for version in [0, 1]: + for git_id in ids: + objs = swh_storage.extid_get_from_extid( + "git", [git_id], version=version + ) + assert len(objs) == 1 + assert objs[0].extid_version == version + for swhid in ids: + objs = swh_storage.extid_get_from_target( + ObjectType.REVISION, + [swhid], + extid_version=version, + extid_type="git", + ) + assert len(objs) == 1 + assert objs[0].extid_version == version + assert objs[0].extid_type == "git" + + def test_extid_version_behavior_failure(self, swh_storage, sample_data): + """Calls with wrong input should raise""" + ids = [ + revision.id + for revision in sample_data.revisions + if revision.type.value == "git" + ] + + # Other edge cases + with pytest.raises( + (ValueError, RemoteException), match="both extid_type and extid_version" + ): + swh_storage.extid_get_from_target( + ObjectType.REVISION, [ids[0]], extid_version=0 + ) + + with pytest.raises( + (ValueError, RemoteException), match="both extid_type and extid_version" + ): + swh_storage.extid_get_from_target( + ObjectType.REVISION, [ids[0]], extid_type="git" + ) def test_release_add(self, swh_storage, sample_data): release, release2 = sample_data.releases[:2] init_missing = swh_storage.release_missing([release.id, release2.id]) assert list(init_missing) == [release.id, release2.id] actual_result = swh_storage.release_add([release, release2]) assert actual_result == {"release:add": 2} end_missing = swh_storage.release_missing([release.id, release2.id]) assert list(end_missing) == [] assert list(swh_storage.journal_writer.journal.objects) == [ ("release", release), ("release", release2), ] # already present so nothing added actual_result = swh_storage.release_add([release, release2]) assert actual_result == {"release:add": 0} - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["release"] == 2 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["release"] == 2 def test_release_add_no_author_date(self, swh_storage, sample_data): full_release = sample_data.release release = attr.evolve(full_release, author=None, date=None) actual_result = swh_storage.release_add([release]) assert actual_result == {"release:add": 1} end_missing = swh_storage.release_missing([release.id]) assert list(end_missing) == [] assert list(swh_storage.journal_writer.journal.objects) == [ ("release", release) ] def test_release_add_twice(self, swh_storage, sample_data): release, release2 = sample_data.releases[:2] actual_result = swh_storage.release_add([release]) assert actual_result == {"release:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("release", release) ] actual_result = swh_storage.release_add([release, release2, release, release2]) assert actual_result == {"release:add": 1} assert set(swh_storage.journal_writer.journal.objects) == set( [("release", release), ("release", release2),] ) def test_release_add_name_clash(self, swh_storage, sample_data): release, release2 = [ attr.evolve( c, author=Person( fullname=b"John Doe ", name=b"John Doe", email=b"john.doe@example.com", ), ) for c in sample_data.releases[:2] ] actual_result = swh_storage.release_add([release, release2]) assert actual_result == {"release:add": 2} def test_release_get(self, swh_storage, sample_data): release, release2, release3 = sample_data.releases[:3] # given swh_storage.release_add([release, release2]) # when actual_releases = swh_storage.release_get([release.id, release2.id]) # then assert actual_releases == [release, release2] unknown_releases = swh_storage.release_get([release3.id]) assert unknown_releases[0] is None def test_release_get_order(self, swh_storage, sample_data): release, release2 = sample_data.releases[:2] add_result = swh_storage.release_add([release, release2]) assert add_result == {"release:add": 2} # order 1 actual_releases = swh_storage.release_get([release.id, release2.id]) assert actual_releases == [release, release2] # order 2 actual_releases2 = swh_storage.release_get([release2.id, release.id]) assert actual_releases2 == [release2, release] def test_release_get_random(self, swh_storage, sample_data): release, release2, release3 = sample_data.releases[:3] swh_storage.release_add([release, release2, release3]) assert swh_storage.release_get_random() in { release.id, release2.id, release3.id, } def test_origin_add(self, swh_storage, sample_data): origins = list(sample_data.origins) origin_urls = [o.url for o in origins] assert swh_storage.origin_get(origin_urls) == [None] * len(origins) stats = swh_storage.origin_add(origins) assert stats == {"origin:add": len(origin_urls)} actual_origins = swh_storage.origin_get(origin_urls) assert actual_origins == origins assert set(swh_storage.journal_writer.journal.objects) == set( [("origin", origin) for origin in origins] ) - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["origin"] == len(origins) + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["origin"] == len(origins) def test_origin_add_twice(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] add1 = swh_storage.origin_add([origin, origin2]) assert set(swh_storage.journal_writer.journal.objects) == set( [("origin", origin), ("origin", origin2),] ) assert add1 == {"origin:add": 2} add2 = swh_storage.origin_add([origin, origin2]) assert set(swh_storage.journal_writer.journal.objects) == set( [("origin", origin), ("origin", origin2),] ) assert add2 == {"origin:add": 0} def test_origin_add_twice_at_once(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] add1 = swh_storage.origin_add([origin, origin2, origin, origin2]) assert set(swh_storage.journal_writer.journal.objects) == set( [("origin", origin), ("origin", origin2),] ) assert add1 == {"origin:add": 2} add2 = swh_storage.origin_add([origin, origin2, origin, origin2]) assert set(swh_storage.journal_writer.journal.objects) == set( [("origin", origin), ("origin", origin2),] ) assert add2 == {"origin:add": 0} def test_origin_get(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] assert swh_storage.origin_get([origin.url]) == [None] swh_storage.origin_add([origin]) actual_origins = swh_storage.origin_get([origin.url]) assert actual_origins == [origin] actual_origins = swh_storage.origin_get([origin.url, "not://exists"]) assert actual_origins == [origin, None] def _generate_random_visits(self, nb_visits=100, start=0, end=7): """Generate random visits within the last 2 months (to avoid computations) """ visits = [] today = now() for weeks in range(nb_visits, 0, -1): hours = random.randint(0, 24) minutes = random.randint(0, 60) seconds = random.randint(0, 60) days = random.randint(0, 28) weeks = random.randint(start, end) date_visit = today - timedelta( weeks=weeks, hours=hours, minutes=minutes, seconds=seconds, days=days ) visits.append(date_visit) return visits def test_origin_visit_get__unknown_origin(self, swh_storage): actual_page = swh_storage.origin_visit_get("foo") assert actual_page.next_page_token is None assert actual_page.results == [] assert actual_page == PagedResult() def test_origin_visit_get__validation_failure(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) with pytest.raises( StorageArgumentException, match="page_token must be a string" ): swh_storage.origin_visit_get(origin.url, page_token=10) # not bytes with pytest.raises( StorageArgumentException, match="order must be a ListOrder value" ): swh_storage.origin_visit_get(origin.url, order="foobar") # wrong order def test_origin_visit_get_all(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) ov1, ov2, ov3 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ), OriginVisit( origin=origin.url, date=sample_data.date_visit2, type=sample_data.type_visit2, ), OriginVisit( origin=origin.url, date=sample_data.date_visit2, type=sample_data.type_visit2, ), ] ) # order asc, no token, no limit actual_page = swh_storage.origin_visit_get(origin.url) assert actual_page.next_page_token is None assert actual_page.results == [ov1, ov2, ov3] # order asc, no token, limit actual_page = swh_storage.origin_visit_get(origin.url, limit=2) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov1, ov2] # order asc, token, no limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token ) assert actual_page.next_page_token is None assert actual_page.results == [ov3] # order asc, no token, limit actual_page = swh_storage.origin_visit_get(origin.url, limit=1) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov1] # order asc, token, no limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token ) assert actual_page.next_page_token is None assert actual_page.results == [ov2, ov3] # order asc, token, limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, limit=2 ) assert actual_page.next_page_token is None assert actual_page.results == [ov2, ov3] actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, limit=1 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov2] actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, limit=1 ) assert actual_page.next_page_token is None assert actual_page.results == [ov3] # order desc, no token, no limit actual_page = swh_storage.origin_visit_get(origin.url, order=ListOrder.DESC) assert actual_page.next_page_token is None assert actual_page.results == [ov3, ov2, ov1] # order desc, no token, limit actual_page = swh_storage.origin_visit_get( origin.url, limit=2, order=ListOrder.DESC ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov3, ov2] # order desc, token, no limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ov1] # order desc, no token, limit actual_page = swh_storage.origin_visit_get( origin.url, limit=1, order=ListOrder.DESC ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov3] # order desc, token, no limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ov2, ov1] # order desc, token, limit actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, order=ListOrder.DESC, limit=1 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ov2] actual_page = swh_storage.origin_visit_get( origin.url, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ov1] def test_origin_visit_status_get__unknown_cases(self, swh_storage, sample_data): origin = sample_data.origin actual_page = swh_storage.origin_visit_status_get("foobar", 1) assert actual_page.next_page_token is None assert actual_page.results == [] actual_page = swh_storage.origin_visit_status_get(origin.url, 1) assert actual_page.next_page_token is None assert actual_page.results == [] origin = sample_data.origin swh_storage.origin_add([origin]) ov1 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ), ] )[0] actual_page = swh_storage.origin_visit_status_get(origin.url, ov1.visit + 10) assert actual_page.next_page_token is None assert actual_page.results == [] def test_origin_visit_status_add_unknown_type(self, swh_storage, sample_data): ov = OriginVisit( origin=sample_data.origin.url, date=now(), type=sample_data.type_visit1, visit=42, ) ovs = OriginVisitStatus( origin=ov.origin, visit=ov.visit, date=now(), status="created", snapshot=None, ) with pytest.raises(StorageArgumentException): swh_storage.origin_visit_status_add([ovs]) swh_storage.origin_add([sample_data.origin]) with pytest.raises(StorageArgumentException): swh_storage.origin_visit_status_add([ovs]) swh_storage.origin_visit_add([ov]) swh_storage.origin_visit_status_add([ovs]) def test_origin_visit_status_get_all(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) date_visit3 = round_to_milliseconds(now()) date_visit1 = date_visit3 - datetime.timedelta(hours=2) date_visit2 = date_visit3 - datetime.timedelta(hours=1) assert date_visit1 < date_visit2 < date_visit3 ov1 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin.url, date=date_visit1, type=sample_data.type_visit1, ), ] )[0] ovs1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit1, type=ov1.type, status="created", snapshot=None, ) ovs2 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit2, type=ov1.type, status="partial", snapshot=None, ) ovs3 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit3, type=ov1.type, status="full", snapshot=sample_data.snapshot.id, metadata={}, ) swh_storage.origin_visit_status_add([ovs2, ovs3]) # order asc, no token, no limit actual_page = swh_storage.origin_visit_status_get(origin.url, ov1.visit) assert actual_page.next_page_token is None assert actual_page.results == [ovs1, ovs2, ovs3] # order asc, no token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, limit=2 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs1, ovs2] # order asc, token, no limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token ) assert actual_page.next_page_token is None assert actual_page.results == [ovs3] # order asc, no token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, limit=1 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs1] actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token ) assert actual_page.next_page_token is None assert actual_page.results == [ovs2, ovs3] # order asc, token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, limit=2 ) assert actual_page.next_page_token is None assert actual_page.results == [ovs2, ovs3] # order asc, no token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, limit=2 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs1, ovs2] actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, limit=1 ) assert actual_page.next_page_token is None assert actual_page.results == [ovs3] # order desc, no token, no limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ovs3, ovs2, ovs1] # order desc, no token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, limit=2, order=ListOrder.DESC ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs3, ovs2] actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ovs1] # order desc, no token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, order=ListOrder.DESC, limit=1 ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs3] # order desc, token, no limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ovs2, ovs1] # order desc, token, limit actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, order=ListOrder.DESC, limit=1, ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [ovs2] actual_page = swh_storage.origin_visit_status_get( origin.url, ov1.visit, page_token=next_page_token, order=ListOrder.DESC ) assert actual_page.next_page_token is None assert actual_page.results == [ovs1] def test_origin_visit_status_get_random(self, swh_storage, sample_data): origins = sample_data.origins[:2] swh_storage.origin_add(origins) # Add some random visits within the selection range visits = self._generate_random_visits() visit_type = "git" # Add visits to those origins for origin in origins: for date_visit in visits: visit = swh_storage.origin_visit_add( [OriginVisit(origin=origin.url, date=date_visit, type=visit_type,)] )[0] swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=visit.visit, date=now(), status="full", - snapshot=None, + snapshot=hash_to_bytes( + "9b922e6d8d5b803c1582aabe5525b7b91150788e" + ), ) ] ) - swh_storage.refresh_stat_counters() - - stats = swh_storage.stat_counters() - assert stats["origin"] == len(origins) - assert stats["origin_visit"] == len(origins) * len(visits) + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + stats = swh_storage.stat_counters() + assert stats["origin"] == len(origins) + assert stats["origin_visit"] == len(origins) * len(visits) random_ovs = swh_storage.origin_visit_status_get_random(visit_type) assert random_ovs assert random_ovs.origin is not None assert random_ovs.origin in [o.url for o in origins] assert random_ovs.type is not None def test_origin_visit_status_get_random_nothing_found( self, swh_storage, sample_data ): origins = sample_data.origins swh_storage.origin_add(origins) visit_type = "hg" # Add some visits outside of the random generation selection so nothing # will be found by the random selection visits = self._generate_random_visits(nb_visits=3, start=13, end=24) for origin in origins: for date_visit in visits: visit = swh_storage.origin_visit_add( [OriginVisit(origin=origin.url, date=date_visit, type=visit_type,)] )[0] swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=visit.visit, date=now(), status="full", snapshot=None, ) ] ) random_origin_visit = swh_storage.origin_visit_status_get_random(visit_type) assert random_origin_visit is None def test_origin_get_by_sha1(self, swh_storage, sample_data): origin = sample_data.origin assert swh_storage.origin_get([origin.url])[0] is None swh_storage.origin_add([origin]) origins = list(swh_storage.origin_get_by_sha1([sha1(origin.url)])) assert len(origins) == 1 assert origins[0]["url"] == origin.url def test_origin_get_by_sha1_not_found(self, swh_storage, sample_data): unknown_origin = sample_data.origin assert swh_storage.origin_get([unknown_origin.url])[0] is None origins = list(swh_storage.origin_get_by_sha1([sha1(unknown_origin.url)])) assert len(origins) == 1 assert origins[0] is None def test_origin_search_single_result(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] actual_page = swh_storage.origin_search(origin.url) assert actual_page.next_page_token is None assert actual_page.results == [] actual_page = swh_storage.origin_search(origin.url, regexp=True) assert actual_page.next_page_token is None assert actual_page.results == [] swh_storage.origin_add([origin]) actual_page = swh_storage.origin_search(origin.url) assert actual_page.next_page_token is None assert actual_page.results == [origin] actual_page = swh_storage.origin_search(f".{origin.url[1:-1]}.", regexp=True) assert actual_page.next_page_token is None assert actual_page.results == [origin] swh_storage.origin_add([origin2]) actual_page = swh_storage.origin_search(origin2.url) assert actual_page.next_page_token is None assert actual_page.results == [origin2] actual_page = swh_storage.origin_search(f".{origin2.url[1:-1]}.", regexp=True) assert actual_page.next_page_token is None assert actual_page.results == [origin2] def test_origin_search_no_regexp(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] swh_storage.origin_add([origin, origin2]) # no pagination actual_page = swh_storage.origin_search("/") assert actual_page.next_page_token is None assert actual_page.results == [origin, origin2] # offset=0 actual_page = swh_storage.origin_search("/", page_token=None, limit=1) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [origin] # offset=1 actual_page = swh_storage.origin_search( "/", page_token=next_page_token, limit=1 ) assert actual_page.next_page_token is None assert actual_page.results == [origin2] def test_origin_search_regexp_substring(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] swh_storage.origin_add([origin, origin2]) # no pagination actual_page = swh_storage.origin_search("/", regexp=True) assert actual_page.next_page_token is None assert actual_page.results == [origin, origin2] # offset=0 actual_page = swh_storage.origin_search( "/", page_token=None, limit=1, regexp=True ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [origin] # offset=1 actual_page = swh_storage.origin_search( "/", page_token=next_page_token, limit=1, regexp=True ) assert actual_page.next_page_token is None assert actual_page.results == [origin2] def test_origin_search_regexp_fullstring(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] swh_storage.origin_add([origin, origin2]) # no pagination actual_page = swh_storage.origin_search(".*/.*", regexp=True) assert actual_page.next_page_token is None assert actual_page.results == [origin, origin2] # offset=0 actual_page = swh_storage.origin_search( ".*/.*", page_token=None, limit=1, regexp=True ) next_page_token = actual_page.next_page_token assert next_page_token is not None assert actual_page.results == [origin] # offset=1 actual_page = swh_storage.origin_search( ".*/.*", page_token=next_page_token, limit=1, regexp=True ) assert actual_page.next_page_token is None assert actual_page.results == [origin2] def test_origin_search_no_visit_types(self, swh_storage, sample_data): origin = sample_data.origins[0] swh_storage.origin_add([origin]) actual_page = swh_storage.origin_search(origin.url, visit_types=["git"]) assert actual_page.next_page_token is None assert actual_page.results == [] def test_origin_search_with_visit_types(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] swh_storage.origin_add([origin, origin2]) swh_storage.origin_visit_add( [ OriginVisit(origin=origin.url, date=now(), type="git"), OriginVisit(origin=origin2.url, date=now(), type="svn"), ] ) actual_page = swh_storage.origin_search(origin.url, visit_types=["git"]) assert actual_page.next_page_token is None assert actual_page.results == [origin] actual_page = swh_storage.origin_search(origin2.url, visit_types=["svn"]) assert actual_page.next_page_token is None assert actual_page.results == [origin2] def test_origin_search_multiple_visit_types(self, swh_storage, sample_data): origin = sample_data.origins[0] swh_storage.origin_add([origin]) def _add_visit_type(visit_type): swh_storage.origin_visit_add( [OriginVisit(origin=origin.url, date=now(), type=visit_type)] ) def _check_visit_types(visit_types): actual_page = swh_storage.origin_search(origin.url, visit_types=visit_types) assert actual_page.next_page_token is None assert actual_page.results == [origin] _add_visit_type("git") _check_visit_types(["git"]) _check_visit_types(["git", "hg"]) _add_visit_type("hg") _check_visit_types(["hg"]) _check_visit_types(["git", "hg"]) def test_origin_visit_add(self, swh_storage, sample_data): origin1 = sample_data.origins[1] swh_storage.origin_add([origin1]) date_visit = now() date_visit2 = date_visit + datetime.timedelta(minutes=1) date_visit = round_to_milliseconds(date_visit) date_visit2 = round_to_milliseconds(date_visit2) visit1 = OriginVisit( origin=origin1.url, date=date_visit, type=sample_data.type_visit1, ) visit2 = OriginVisit( origin=origin1.url, date=date_visit2, type=sample_data.type_visit2, ) # add once ov1, ov2 = swh_storage.origin_visit_add([visit1, visit2]) # then again (will be ignored as they already exist) origin_visit1, origin_visit2 = swh_storage.origin_visit_add([ov1, ov2]) assert ov1 == origin_visit1 assert ov2 == origin_visit2 assert ov1.visit == 1 assert ov2.visit == 2 ovs1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit, type=ov1.type, status="created", snapshot=None, ) ovs2 = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=date_visit2, type=ov2.type, status="created", snapshot=None, ) actual_visits = swh_storage.origin_visit_get(origin1.url).results expected_visits = [ov1, ov2] assert len(expected_visits) == len(actual_visits) for visit in expected_visits: assert visit in actual_visits actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = list( [("origin", origin1)] + [("origin_visit", visit) for visit in expected_visits] * 2 + [("origin_visit_status", ovs) for ovs in [ovs1, ovs2]] ) for obj in expected_objects: assert obj in actual_objects def test_origin_visit_add_replayed(self, swh_storage, sample_data): """Tests adding a visit with an id makes sure the next id is higher""" origin1 = sample_data.origins[1] swh_storage.origin_add([origin1]) date_visit = now() date_visit2 = date_visit + datetime.timedelta(minutes=1) date_visit = round_to_milliseconds(date_visit) date_visit2 = round_to_milliseconds(date_visit2) visit1 = OriginVisit( origin=origin1.url, date=date_visit, type=sample_data.type_visit1, visit=42 ) visit2 = OriginVisit( origin=origin1.url, date=date_visit2, type=sample_data.type_visit2, ) # add once ov1, ov2 = swh_storage.origin_visit_add([visit1, visit2]) # then again (will be ignored as they already exist) origin_visit1, origin_visit2 = swh_storage.origin_visit_add([ov1, ov2]) assert ov1 == origin_visit1 assert ov2 == origin_visit2 assert ov1.visit == 42 assert ov2.visit == 43 visit3 = OriginVisit( origin=origin1.url, date=date_visit, type=sample_data.type_visit1, visit=12 ) visit4 = OriginVisit( origin=origin1.url, date=date_visit2, type=sample_data.type_visit2, ) # add once ov3, ov4 = swh_storage.origin_visit_add([visit3, visit4]) # then again (will be ignored as they already exist) origin_visit3, origin_visit4 = swh_storage.origin_visit_add([ov3, ov4]) assert ov3 == origin_visit3 assert ov4 == origin_visit4 assert ov3.visit == 12 assert ov4.visit == 44 def test_origin_visit_add_validation(self, swh_storage, sample_data): """Unknown origin when adding visits should raise""" visit = attr.evolve(sample_data.origin_visit, origin="something-unknonw") with pytest.raises(StorageArgumentException, match="Unknown origin"): swh_storage.origin_visit_add([visit]) objects = list(swh_storage.journal_writer.journal.objects) assert not objects def test_origin_visit_status_add_validation(self, swh_storage): """Wrong origin_visit_status input should raise storage argument error""" date_visit = now() visit_status1 = OriginVisitStatus( origin="unknown-origin-url", visit=10, date=date_visit, status="full", snapshot=None, ) with pytest.raises(StorageArgumentException, match="Unknown origin"): swh_storage.origin_visit_status_add([visit_status1]) objects = list(swh_storage.journal_writer.journal.objects) assert not objects def test_origin_visit_status_add(self, swh_storage, sample_data): """Correct origin visit statuses should add a new visit status """ snapshot = sample_data.snapshot origin1 = sample_data.origins[1] origin2 = Origin(url="new-origin") swh_storage.origin_add([origin1, origin2]) ov1, ov2 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin1.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ), OriginVisit( origin=origin2.url, date=sample_data.date_visit2, type=sample_data.type_visit2, ), ] ) ovs1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=sample_data.date_visit1, type=ov1.type, status="created", snapshot=None, ) ovs2 = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=sample_data.date_visit2, type=ov2.type, status="created", snapshot=None, ) date_visit_now = round_to_milliseconds(now()) visit_status1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit_now, type=ov1.type, status="full", snapshot=snapshot.id, ) date_visit_now = round_to_milliseconds(now()) visit_status2 = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=date_visit_now, type=ov2.type, status="ongoing", snapshot=None, metadata={"intrinsic": "something"}, ) stats = swh_storage.origin_visit_status_add([visit_status1, visit_status2]) assert stats == {"origin_visit_status:add": 2} visit = swh_storage.origin_visit_get_latest(origin1.url, require_snapshot=True) visit_status = swh_storage.origin_visit_status_get_latest( origin1.url, visit.visit, require_snapshot=True ) assert visit_status == visit_status1 visit = swh_storage.origin_visit_get_latest(origin2.url, require_snapshot=False) visit_status = swh_storage.origin_visit_status_get_latest( origin2.url, visit.visit, require_snapshot=False ) assert origin2.url != origin1.url assert visit_status == visit_status2 actual_objects = list(swh_storage.journal_writer.journal.objects) expected_origins = [origin1, origin2] expected_visits = [ov1, ov2] expected_visit_statuses = [ovs1, ovs2, visit_status1, visit_status2] expected_objects = ( [("origin", o) for o in expected_origins] + [("origin_visit", v) for v in expected_visits] + [("origin_visit_status", ovs) for ovs in expected_visit_statuses] ) for obj in expected_objects: assert obj in actual_objects def test_origin_visit_status_add_twice(self, swh_storage, sample_data): """Correct origin visit statuses should add a new visit status """ snapshot = sample_data.snapshot origin1 = sample_data.origins[1] swh_storage.origin_add([origin1]) ov1 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin1.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ), ] )[0] ovs1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=sample_data.date_visit1, type=ov1.type, status="created", snapshot=None, ) date_visit_now = round_to_milliseconds(now()) visit_status1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_visit_now, type=ov1.type, status="full", snapshot=snapshot.id, ) stats = swh_storage.origin_visit_status_add([visit_status1]) assert stats == {"origin_visit_status:add": 1} # second call will ignore existing entries (will send to storage though) stats = swh_storage.origin_visit_status_add([visit_status1]) # ...so the storage still returns it as an addition assert stats == {"origin_visit_status:add": 1} visit_status = swh_storage.origin_visit_status_get_latest(ov1.origin, ov1.visit) assert visit_status == visit_status1 actual_objects = list(swh_storage.journal_writer.journal.objects) expected_origins = [origin1] expected_visits = [ov1] expected_visit_statuses = [ovs1, visit_status1, visit_status1] # write twice in the journal expected_objects = ( [("origin", o) for o in expected_origins] + [("origin_visit", v) for v in expected_visits] + [("origin_visit_status", ovs) for ovs in expected_visit_statuses] ) for obj in expected_objects: assert obj in actual_objects def test_origin_visit_find_by_date(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) visit1 = OriginVisit( origin=origin.url, date=sample_data.date_visit2, type=sample_data.type_visit1, ) visit2 = OriginVisit( origin=origin.url, date=sample_data.date_visit3, type=sample_data.type_visit2, ) visit3 = OriginVisit( origin=origin.url, date=sample_data.date_visit2, type=sample_data.type_visit3, ) ov1, ov2, ov3 = swh_storage.origin_visit_add([visit1, visit2, visit3]) ovs1 = OriginVisitStatus( origin=origin.url, visit=ov1.visit, date=sample_data.date_visit2, status="ongoing", snapshot=None, ) ovs2 = OriginVisitStatus( origin=origin.url, visit=ov2.visit, date=sample_data.date_visit3, status="ongoing", snapshot=None, ) ovs3 = OriginVisitStatus( origin=origin.url, visit=ov3.visit, date=sample_data.date_visit2, status="ongoing", snapshot=None, ) swh_storage.origin_visit_status_add([ovs1, ovs2, ovs3]) # Simple case actual_visit = swh_storage.origin_visit_find_by_date( origin.url, sample_data.date_visit3 ) assert actual_visit == ov2 # There are two visits at the same date, the latest must be returned actual_visit = swh_storage.origin_visit_find_by_date( origin.url, sample_data.date_visit2 ) assert actual_visit == ov3 def test_origin_visit_find_by_date__unknown_origin(self, swh_storage, sample_data): actual_visit = swh_storage.origin_visit_find_by_date( "foo", sample_data.date_visit2 ) assert actual_visit is None def test_origin_visit_get_by(self, swh_storage, sample_data): snapshot = sample_data.snapshot origins = sample_data.origins[:2] swh_storage.origin_add(origins) origin_url, origin_url2 = [o.url for o in origins] visit = OriginVisit( origin=origin_url, date=sample_data.date_visit2, type=sample_data.type_visit2, ) origin_visit1 = swh_storage.origin_visit_add([visit])[0] swh_storage.snapshot_add([snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin_url, visit=origin_visit1.visit, date=now(), status="ongoing", snapshot=snapshot.id, ) ] ) # Add some other {origin, visit} entries visit2 = OriginVisit( origin=origin_url, date=sample_data.date_visit3, type=sample_data.type_visit3, ) visit3 = OriginVisit( origin=origin_url2, date=sample_data.date_visit3, type=sample_data.type_visit3, ) swh_storage.origin_visit_add([visit2, visit3]) # when visit1_metadata = { "contents": 42, "directories": 22, } swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin_url, visit=origin_visit1.visit, date=now(), status="full", snapshot=snapshot.id, metadata=visit1_metadata, ) ] ) actual_visit = swh_storage.origin_visit_get_by(origin_url, origin_visit1.visit) assert actual_visit == origin_visit1 def test_origin_visit_get_by__no_result(self, swh_storage, sample_data): actual_visit = swh_storage.origin_visit_get_by("unknown", 10) # unknown origin assert actual_visit is None origin = sample_data.origin swh_storage.origin_add([origin]) actual_visit = swh_storage.origin_visit_get_by(origin.url, 999) # unknown visit assert actual_visit is None def test_origin_visit_get_latest_edge_cases(self, swh_storage, sample_data): # unknown origin so no result assert swh_storage.origin_visit_get_latest("unknown-origin") is None # unknown type so no result origin = sample_data.origin swh_storage.origin_add([origin]) assert swh_storage.origin_visit_get_latest(origin.url, type="unknown") is None # unknown allowed statuses should raise with pytest.raises(StorageArgumentException, match="Unknown allowed statuses"): swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["unknown"] ) def test_origin_visit_get_latest_filter_type(self, swh_storage, sample_data): """Filtering origin visit get latest with filter type should be ok """ origin = sample_data.origin swh_storage.origin_add([origin]) visit1 = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type="git", ) visit2 = OriginVisit( origin=origin.url, date=sample_data.date_visit2, type="hg", ) date_now = round_to_milliseconds(now()) visit3 = OriginVisit(origin=origin.url, date=date_now, type="hg",) assert sample_data.date_visit1 < sample_data.date_visit2 assert sample_data.date_visit2 < date_now ov1, ov2, ov3 = swh_storage.origin_visit_add([visit1, visit2, visit3]) # Check type filter is ok actual_visit = swh_storage.origin_visit_get_latest(origin.url, type="git") assert actual_visit == ov1 actual_visit = swh_storage.origin_visit_get_latest(origin.url, type="hg") assert actual_visit == ov3 actual_visit_unknown_type = swh_storage.origin_visit_get_latest( origin.url, type="npm", # no visit matching that type ) assert actual_visit_unknown_type is None def test_origin_visit_get_latest(self, swh_storage, sample_data): empty_snapshot, complete_snapshot = sample_data.snapshots[1:3] origin = sample_data.origin swh_storage.origin_add([origin]) visit1 = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type="git", ) visit2 = OriginVisit( origin=origin.url, date=sample_data.date_visit2, type="hg", ) date_now = round_to_milliseconds(now()) visit3 = OriginVisit(origin=origin.url, date=date_now, type="hg",) assert visit1.date < visit2.date assert visit2.date < visit3.date ov1, ov2, ov3 = swh_storage.origin_visit_add([visit1, visit2, visit3]) # no filters, latest visit is the last one (whose date is most recent) actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov3 # 3 visits, none has snapshot so nothing is returned actual_visit = swh_storage.origin_visit_get_latest( origin.url, require_snapshot=True ) assert actual_visit is None # visit are created with "created" status, so nothing will get returned actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["partial"] ) assert actual_visit is None # visit are created with "created" status, so most recent again actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["created"] ) assert actual_visit == ov3 # Add snapshot to visit1; require_snapshot=True makes it return first visit swh_storage.snapshot_add([complete_snapshot]) visit_status_with_snapshot = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=round_to_milliseconds(now()), type=ov1.type, status="ongoing", snapshot=complete_snapshot.id, ) swh_storage.origin_visit_status_add([visit_status_with_snapshot]) # only the first visit has a snapshot now actual_visit = swh_storage.origin_visit_get_latest( origin.url, require_snapshot=True ) assert actual_visit == ov1 # only the first visit has a status ongoing now actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["ongoing"] ) assert actual_visit == ov1 actual_visit_status = swh_storage.origin_visit_status_get_latest( origin.url, ov1.visit, require_snapshot=True ) assert actual_visit_status == visit_status_with_snapshot # ... and require_snapshot=False (defaults) still returns latest visit (3rd) actual_visit = swh_storage.origin_visit_get_latest( origin.url, require_snapshot=False ) assert actual_visit == ov3 # no specific filter, this returns as before the latest visit actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov3 # Status filter: all three visits are status=ongoing, so no visit # returned actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["full"] ) assert actual_visit is None visit_status1_full = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=round_to_milliseconds(now()), type=ov1.type, status="full", snapshot=complete_snapshot.id, ) # Mark the first visit as completed and check status filter again swh_storage.origin_visit_status_add([visit_status1_full]) # only the first visit has the full status actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["full"] ) assert actual_visit == ov1 actual_visit_status = swh_storage.origin_visit_status_get_latest( origin.url, ov1.visit, allowed_statuses=["full"] ) assert actual_visit_status == visit_status1_full # no specific filter, this returns as before the latest visit actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov3 # Add snapshot to visit2 and check that the new snapshot is returned swh_storage.snapshot_add([empty_snapshot]) visit_status2_full = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=round_to_milliseconds(now()), type=ov2.type, status="ongoing", snapshot=empty_snapshot.id, ) swh_storage.origin_visit_status_add([visit_status2_full]) actual_visit = swh_storage.origin_visit_get_latest( origin.url, require_snapshot=True ) # 2nd visit is most recent with a snapshot assert actual_visit == ov2 actual_visit_status = swh_storage.origin_visit_status_get_latest( origin.url, ov2.visit, require_snapshot=True ) assert actual_visit_status == visit_status2_full # no specific filter, this returns as before the latest visit, 3rd one actual_origin = swh_storage.origin_visit_get_latest(origin.url) assert actual_origin == ov3 # full status is still the first visit actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["full"] ) assert actual_visit == ov1 # Add snapshot to visit3 (same date as visit2) visit_status3_with_snapshot = OriginVisitStatus( origin=ov3.origin, visit=ov3.visit, date=round_to_milliseconds(now()), type=ov3.type, status="ongoing", snapshot=complete_snapshot.id, ) swh_storage.origin_visit_status_add([visit_status3_with_snapshot]) # full status is still the first visit actual_visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["full"], require_snapshot=True, ) assert actual_visit == ov1 actual_visit_status = swh_storage.origin_visit_status_get_latest( origin.url, visit=actual_visit.visit, allowed_statuses=["full"], require_snapshot=True, ) assert actual_visit_status == visit_status1_full # most recent is still the 3rd visit actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov3 # 3rd visit has a snapshot now, so it's elected actual_visit = swh_storage.origin_visit_get_latest( origin.url, require_snapshot=True ) assert actual_visit == ov3 actual_visit_status = swh_storage.origin_visit_status_get_latest( origin.url, ov3.visit, require_snapshot=True ) assert actual_visit_status == visit_status3_with_snapshot def test_origin_visit_get_latest__same_date(self, swh_storage, sample_data): empty_snapshot, complete_snapshot = sample_data.snapshots[1:3] origin = sample_data.origin swh_storage.origin_add([origin]) visit1 = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type="git", ) visit2 = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type="hg", ) ov1, ov2 = swh_storage.origin_visit_add([visit1, visit2]) # ties should be broken by using the visit id actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov2 def test_origin_visit_get_latest_order(self, swh_storage, sample_data): empty_snapshot, complete_snapshot = sample_data.snapshots[1:3] origin = sample_data.origin id1 = 2 id2 = 1 id3 = 3 date1 = datetime.datetime(2021, 8, 2, tzinfo=datetime.timezone.utc) date2 = datetime.datetime(2021, 8, 3, tzinfo=datetime.timezone.utc) date3 = datetime.datetime(2021, 8, 1, tzinfo=datetime.timezone.utc) swh_storage.origin_add([origin]) visit1 = OriginVisit(origin=origin.url, visit=id1, date=date1, type="git",) visit2 = OriginVisit(origin=origin.url, visit=id2, date=date2, type="hg",) visit3 = OriginVisit(origin=origin.url, visit=id3, date=date3, type="tar",) ov1, ov2, ov3 = swh_storage.origin_visit_add([visit1, visit2, visit3]) # no filters, latest visit is the last one (whose date is most recent) actual_visit = swh_storage.origin_visit_get_latest(origin.url) assert actual_visit == ov2 def test_origin_visit_get_latest__not_last(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) visit1, visit2 = sample_data.origin_visits[:2] assert visit1.origin == origin.url swh_storage.origin_visit_add([visit1]) ov1 = swh_storage.origin_visit_get_latest(origin.url) # Add snapshot to visit1, latest snapshot = visit 1 snapshot complete_snapshot = sample_data.snapshots[2] swh_storage.snapshot_add([complete_snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=ov1.visit, date=visit2.date, status="partial", snapshot=None, ) ] ) assert visit1.date < visit2.date # no snapshot associated to the visit, so None visit = swh_storage.origin_visit_get_latest( origin.url, allowed_statuses=["partial"], require_snapshot=True, ) assert visit is None date_now = now() assert visit2.date < date_now swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=ov1.visit, date=date_now, status="full", snapshot=complete_snapshot.id, ) ] ) swh_storage.origin_visit_add( [OriginVisit(origin=origin.url, date=now(), type=visit1.type,)] ) visit = swh_storage.origin_visit_get_latest(origin.url, require_snapshot=True) assert visit is not None def test_origin_visit_status_get_latest__validation(self, swh_storage, sample_data): origin = sample_data.origin swh_storage.origin_add([origin]) visit1 = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type="git", ) # unknown allowed statuses should raise with pytest.raises(StorageArgumentException, match="Unknown allowed statuses"): swh_storage.origin_visit_status_get_latest( origin.url, visit1.visit, allowed_statuses=["unknown"] ) def test_origin_visit_status_get_latest(self, swh_storage, sample_data): snapshot = sample_data.snapshots[2] origin1 = sample_data.origin swh_storage.origin_add([origin1]) # to have some reference visits ov1, ov2 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin1.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ), OriginVisit( origin=origin1.url, date=sample_data.date_visit2, type=sample_data.type_visit2, ), ] ) swh_storage.snapshot_add([snapshot]) date_now = round_to_milliseconds(now()) assert sample_data.date_visit1 < sample_data.date_visit2 assert sample_data.date_visit2 < date_now ovs1 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=sample_data.date_visit1, type=ov1.type, status="partial", snapshot=None, ) ovs2 = OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=sample_data.date_visit2, type=ov1.type, status="ongoing", snapshot=None, ) ovs3 = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=sample_data.date_visit2 + datetime.timedelta(minutes=1), # to not be ignored type=ov2.type, status="ongoing", snapshot=None, ) ovs4 = OriginVisitStatus( origin=ov2.origin, visit=ov2.visit, date=date_now, type=ov2.type, status="full", snapshot=snapshot.id, metadata={"something": "wicked"}, ) swh_storage.origin_visit_status_add([ovs1, ovs2, ovs3, ovs4]) # unknown origin so no result actual_origin_visit = swh_storage.origin_visit_status_get_latest( "unknown-origin", ov1.visit ) assert actual_origin_visit is None # unknown visit so no result actual_origin_visit = swh_storage.origin_visit_status_get_latest( ov1.origin, ov1.visit + 10 ) assert actual_origin_visit is None # Two visits, both with no snapshot, take the most recent actual_origin_visit2 = swh_storage.origin_visit_status_get_latest( origin1.url, ov1.visit ) assert isinstance(actual_origin_visit2, OriginVisitStatus) assert actual_origin_visit2 == ovs2 assert ovs2.origin == origin1.url assert ovs2.visit == ov1.visit actual_origin_visit = swh_storage.origin_visit_status_get_latest( origin1.url, ov1.visit, require_snapshot=True ) # there is no visit with snapshot yet for that visit assert actual_origin_visit is None actual_origin_visit2 = swh_storage.origin_visit_status_get_latest( origin1.url, ov1.visit, allowed_statuses=["partial", "ongoing"] ) # visit status with partial status visit elected assert actual_origin_visit2 == ovs2 assert actual_origin_visit2.status == "ongoing" actual_origin_visit4 = swh_storage.origin_visit_status_get_latest( origin1.url, ov2.visit, require_snapshot=True ) assert actual_origin_visit4 == ovs4 assert actual_origin_visit4.snapshot == snapshot.id actual_origin_visit = swh_storage.origin_visit_status_get_latest( origin1.url, ov2.visit, require_snapshot=True, allowed_statuses=["ongoing"] ) # nothing matches so nothing assert actual_origin_visit is None # there is no visit with status full actual_origin_visit3 = swh_storage.origin_visit_status_get_latest( origin1.url, ov2.visit, allowed_statuses=["ongoing"] ) assert actual_origin_visit3 == ovs3 def test_person_fullname_unicity(self, swh_storage, sample_data): revision, rev2 = sample_data.revisions[0:2] # create a revision with same committer fullname but wo name and email revision2 = attr.evolve( rev2, committer=Person( fullname=revision.committer.fullname, name=None, email=None ), ) swh_storage.revision_add([revision, revision2]) # when getting added revisions revisions = swh_storage.revision_get([revision.id, revision2.id]) # then check committers are the same assert revisions[0].committer == revisions[1].committer def test_snapshot_add_get_empty(self, swh_storage, sample_data): empty_snapshot = sample_data.snapshots[1] empty_snapshot_dict = empty_snapshot.to_dict() origin = sample_data.origin swh_storage.origin_add([origin]) ov1 = swh_storage.origin_visit_add( [ OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ) ] )[0] actual_result = swh_storage.snapshot_add([empty_snapshot]) assert actual_result == {"snapshot:add": 1} date_now = now() swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=ov1.origin, visit=ov1.visit, date=date_now, type=ov1.type, status="full", snapshot=empty_snapshot.id, ) ] ) by_id = swh_storage.snapshot_get(empty_snapshot.id) assert by_id == {**empty_snapshot_dict, "next_branch": None} ovs1 = OriginVisitStatus.from_dict( { "origin": ov1.origin, "date": sample_data.date_visit1, "type": ov1.type, "visit": ov1.visit, "status": "created", "snapshot": None, "metadata": None, } ) ovs2 = OriginVisitStatus.from_dict( { "origin": ov1.origin, "date": date_now, "type": ov1.type, "visit": ov1.visit, "status": "full", "metadata": None, "snapshot": empty_snapshot.id, } ) actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = [ ("origin", origin), ("origin_visit", ov1), ("origin_visit_status", ovs1,), ("snapshot", empty_snapshot), ("origin_visit_status", ovs2,), ] for obj in expected_objects: assert obj in actual_objects def test_snapshot_add_get_complete(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] complete_snapshot_dict = complete_snapshot.to_dict() origin = sample_data.origin swh_storage.origin_add([origin]) visit = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ) origin_visit1 = swh_storage.origin_visit_add([visit])[0] actual_result = swh_storage.snapshot_add([complete_snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=origin_visit1.visit, date=now(), status="ongoing", snapshot=complete_snapshot.id, ) ] ) assert actual_result == {"snapshot:add": 1} by_id = swh_storage.snapshot_get(complete_snapshot.id) assert by_id == {**complete_snapshot_dict, "next_branch": None} def test_snapshot_add_many(self, swh_storage, sample_data): snapshot, _, complete_snapshot = sample_data.snapshots[:3] actual_result = swh_storage.snapshot_add([snapshot, complete_snapshot]) assert actual_result == {"snapshot:add": 2} assert swh_storage.snapshot_get(complete_snapshot.id) == { **complete_snapshot.to_dict(), "next_branch": None, } assert swh_storage.snapshot_get(snapshot.id) == { **snapshot.to_dict(), "next_branch": None, } - swh_storage.refresh_stat_counters() - assert swh_storage.stat_counters()["snapshot"] == 2 + if isinstance(swh_storage, InMemoryStorage) or not isinstance( + swh_storage, CassandraStorage + ): + swh_storage.refresh_stat_counters() + assert swh_storage.stat_counters()["snapshot"] == 2 def test_snapshot_add_many_incremental(self, swh_storage, sample_data): snapshot, _, complete_snapshot = sample_data.snapshots[:3] actual_result = swh_storage.snapshot_add([complete_snapshot]) assert actual_result == {"snapshot:add": 1} actual_result2 = swh_storage.snapshot_add([snapshot, complete_snapshot]) assert actual_result2 == {"snapshot:add": 1} assert swh_storage.snapshot_get(complete_snapshot.id) == { **complete_snapshot.to_dict(), "next_branch": None, } assert swh_storage.snapshot_get(snapshot.id) == { **snapshot.to_dict(), "next_branch": None, } def test_snapshot_add_twice(self, swh_storage, sample_data): snapshot, empty_snapshot = sample_data.snapshots[:2] actual_result = swh_storage.snapshot_add([empty_snapshot]) assert actual_result == {"snapshot:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("snapshot", empty_snapshot) ] actual_result = swh_storage.snapshot_add([snapshot]) assert actual_result == {"snapshot:add": 1} assert list(swh_storage.journal_writer.journal.objects) == [ ("snapshot", empty_snapshot), ("snapshot", snapshot), ] def test_snapshot_add_count_branches(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] actual_result = swh_storage.snapshot_add([complete_snapshot]) assert actual_result == {"snapshot:add": 1} snp_size = swh_storage.snapshot_count_branches(complete_snapshot.id) expected_snp_size = { "alias": 1, "content": 1, "directory": 2, "release": 1, "revision": 1, "snapshot": 1, None: 1, } assert snp_size == expected_snp_size def test_snapshot_add_count_branches_with_filtering(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] actual_result = swh_storage.snapshot_add([complete_snapshot]) assert actual_result == {"snapshot:add": 1} snp_size = swh_storage.snapshot_count_branches( complete_snapshot.id, branch_name_exclude_prefix=b"release" ) expected_snp_size = { "alias": 1, "content": 1, "directory": 2, "revision": 1, "snapshot": 1, None: 1, } assert snp_size == expected_snp_size def test_snapshot_add_count_branches_with_filtering_edge_cases( self, swh_storage, sample_data ): snapshot = Snapshot( branches={ b"\xaa\xff": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"\xaa\xff\x00": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"\xff\xff": SnapshotBranch( target=sample_data.release.id, target_type=TargetType.RELEASE, ), b"\xff\xff\x00": SnapshotBranch( target=sample_data.release.id, target_type=TargetType.RELEASE, ), b"dangling": None, }, ) swh_storage.snapshot_add([snapshot]) assert swh_storage.snapshot_count_branches( snapshot.id, branch_name_exclude_prefix=b"\xaa\xff" ) == {None: 1, "release": 2} assert swh_storage.snapshot_count_branches( snapshot.id, branch_name_exclude_prefix=b"\xff\xff" ) == {None: 1, "revision": 2} def test_snapshot_add_get_paginated(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] swh_storage.snapshot_add([complete_snapshot]) snp_id = complete_snapshot.id branches = complete_snapshot.branches branch_names = list(sorted(branches)) # Test branch_from snapshot = swh_storage.snapshot_get_branches(snp_id, branches_from=b"release") rel_idx = branch_names.index(b"release") expected_snapshot = { "id": snp_id, "branches": {name: branches[name] for name in branch_names[rel_idx:]}, "next_branch": None, } assert snapshot == expected_snapshot # Test branches_count snapshot = swh_storage.snapshot_get_branches(snp_id, branches_count=1) expected_snapshot = { "id": snp_id, "branches": {branch_names[0]: branches[branch_names[0]],}, "next_branch": b"content", } assert snapshot == expected_snapshot # test branch_from + branches_count snapshot = swh_storage.snapshot_get_branches( snp_id, branches_from=b"directory", branches_count=3 ) dir_idx = branch_names.index(b"directory") expected_snapshot = { "id": snp_id, "branches": { name: branches[name] for name in branch_names[dir_idx : dir_idx + 3] }, "next_branch": branch_names[dir_idx + 3], } assert snapshot == expected_snapshot def test_snapshot_add_get_filtered(self, swh_storage, sample_data): origin = sample_data.origin complete_snapshot = sample_data.snapshots[2] swh_storage.origin_add([origin]) visit = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ) origin_visit1 = swh_storage.origin_visit_add([visit])[0] swh_storage.snapshot_add([complete_snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=origin_visit1.visit, date=now(), status="ongoing", snapshot=complete_snapshot.id, ) ] ) snp_id = complete_snapshot.id branches = complete_snapshot.branches snapshot = swh_storage.snapshot_get_branches( snp_id, target_types=["release", "revision"] ) expected_snapshot = { "id": snp_id, "branches": { name: tgt for name, tgt in branches.items() if tgt and tgt.target_type in [TargetType.RELEASE, TargetType.REVISION] }, "next_branch": None, } assert snapshot == expected_snapshot snapshot = swh_storage.snapshot_get_branches(snp_id, target_types=["alias"]) expected_snapshot = { "id": snp_id, "branches": { name: tgt for name, tgt in branches.items() if tgt and tgt.target_type == TargetType.ALIAS }, "next_branch": None, } assert snapshot == expected_snapshot def test_snapshot_add_get_filtered_and_paginated(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] swh_storage.snapshot_add([complete_snapshot]) snp_id = complete_snapshot.id branches = complete_snapshot.branches branch_names = list(sorted(branches)) # Test branch_from snapshot = swh_storage.snapshot_get_branches( snp_id, target_types=["directory", "release"], branches_from=b"directory2" ) expected_snapshot = { "id": snp_id, "branches": {name: branches[name] for name in (b"directory2", b"release")}, "next_branch": None, } assert snapshot == expected_snapshot # Test branches_count snapshot = swh_storage.snapshot_get_branches( snp_id, target_types=["directory", "release"], branches_count=1 ) expected_snapshot = { "id": snp_id, "branches": {b"directory": branches[b"directory"]}, "next_branch": b"directory2", } assert snapshot == expected_snapshot # Test branches_count snapshot = swh_storage.snapshot_get_branches( snp_id, target_types=["directory", "release"], branches_count=2 ) expected_snapshot = { "id": snp_id, "branches": { name: branches[name] for name in (b"directory", b"directory2") }, "next_branch": b"release", } assert snapshot == expected_snapshot # test branch_from + branches_count snapshot = swh_storage.snapshot_get_branches( snp_id, target_types=["directory", "release"], branches_from=b"directory2", branches_count=1, ) dir_idx = branch_names.index(b"directory2") expected_snapshot = { "id": snp_id, "branches": {branch_names[dir_idx]: branches[branch_names[dir_idx]],}, "next_branch": b"release", } assert snapshot == expected_snapshot def test_snapshot_add_get_branch_by_type(self, swh_storage, sample_data): complete_snapshot = sample_data.snapshots[2] snapshot = complete_snapshot.to_dict() alias1 = b"alias1" alias2 = b"alias2" target1 = random.choice(list(snapshot["branches"].keys())) target2 = random.choice(list(snapshot["branches"].keys())) snapshot["branches"][alias2] = { "target": target2, "target_type": "alias", } snapshot["branches"][alias1] = { "target": target1, "target_type": "alias", } new_snapshot = Snapshot.from_dict(snapshot) swh_storage.snapshot_add([new_snapshot]) branches = swh_storage.snapshot_get_branches( new_snapshot.id, target_types=["alias"], branches_from=alias1, branches_count=1, )["branches"] assert len(branches) == 1 assert alias1 in branches def test_snapshot_add_get_by_branches_name_pattern(self, swh_storage, sample_data): snapshot = Snapshot( branches={ b"refs/heads/master": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"refs/heads/incoming": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"refs/pull/1": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"refs/pull/2": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"dangling": None, b"\xaa\xff": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"\xaa\xff\x00": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"\xff\xff": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), b"\xff\xff\x00": SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ), }, ) swh_storage.snapshot_add([snapshot]) for include_pattern, exclude_prefix, nb_results in ( (b"pull", None, 2), (b"incoming", None, 1), (b"dangling", None, 1), (None, b"refs/heads/", 7), (b"refs", b"refs/heads/master", 3), (b"refs", b"refs/heads/master", 3), (None, b"\xaa\xff", 7), (None, b"\xff\xff", 7), ): branches = swh_storage.snapshot_get_branches( snapshot.id, branch_name_include_substring=include_pattern, branch_name_exclude_prefix=exclude_prefix, )["branches"] expected_branches = [ branch_name for branch_name in snapshot.branches if (include_pattern is None or include_pattern in branch_name) and ( exclude_prefix is None or not branch_name.startswith(exclude_prefix) ) ] assert sorted(branches) == sorted(expected_branches) assert len(branches) == nb_results def test_snapshot_add_get_by_branches_name_pattern_filtered_paginated( self, swh_storage, sample_data ): pattern = b"foo" nb_branches_by_target_type = 10 branches = {} for i in range(nb_branches_by_target_type): branches[f"branch/directory/bar{i}".encode()] = SnapshotBranch( target=sample_data.directory.id, target_type=TargetType.DIRECTORY, ) branches[f"branch/revision/bar{i}".encode()] = SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ) branches[f"branch/directory/{pattern}{i}".encode()] = SnapshotBranch( target=sample_data.directory.id, target_type=TargetType.DIRECTORY, ) branches[f"branch/revision/{pattern}{i}".encode()] = SnapshotBranch( target=sample_data.revision.id, target_type=TargetType.REVISION, ) snapshot = Snapshot(branches=branches) swh_storage.snapshot_add([snapshot]) branches_count = nb_branches_by_target_type // 2 for target_type in ( TargetType.DIRECTORY, TargetType.REVISION, ): target_type_str = target_type.value partial_branches = swh_storage.snapshot_get_branches( snapshot.id, branch_name_include_substring=pattern, target_types=[target_type_str], branches_count=branches_count, ) branches = partial_branches["branches"] expected_branches = [ branch_name for branch_name, branch_data in snapshot.branches.items() if pattern in branch_name and branch_data.target_type == target_type ][:branches_count] assert sorted(branches) == sorted(expected_branches) assert ( partial_branches["next_branch"] == f"branch/{target_type_str}/{pattern}{branches_count}".encode() ) partial_branches = swh_storage.snapshot_get_branches( snapshot.id, branch_name_include_substring=pattern, target_types=[target_type_str], branches_from=partial_branches["next_branch"], ) branches = partial_branches["branches"] expected_branches = [ branch_name for branch_name, branch_data in snapshot.branches.items() if pattern in branch_name and branch_data.target_type == target_type ][branches_count:] assert sorted(branches) == sorted(expected_branches) assert partial_branches["next_branch"] is None def test_snapshot_add_get(self, swh_storage, sample_data): snapshot = sample_data.snapshot origin = sample_data.origin swh_storage.origin_add([origin]) visit = OriginVisit( origin=origin.url, date=sample_data.date_visit1, type=sample_data.type_visit1, ) ov1 = swh_storage.origin_visit_add([visit])[0] swh_storage.snapshot_add([snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=ov1.visit, date=now(), status="ongoing", snapshot=snapshot.id, ) ] ) expected_snapshot = {**snapshot.to_dict(), "next_branch": None} by_id = swh_storage.snapshot_get(snapshot.id) assert by_id == expected_snapshot actual_visit = swh_storage.origin_visit_get_by(origin.url, ov1.visit) assert actual_visit == ov1 visit_status = swh_storage.origin_visit_status_get_latest( origin.url, ov1.visit, require_snapshot=True ) assert visit_status.snapshot == snapshot.id def test_snapshot_get_random(self, swh_storage, sample_data): snapshot, empty_snapshot, complete_snapshot = sample_data.snapshots[:3] swh_storage.snapshot_add([snapshot, empty_snapshot, complete_snapshot]) assert swh_storage.snapshot_get_random() in { snapshot.id, empty_snapshot.id, complete_snapshot.id, } def test_snapshot_missing(self, swh_storage, sample_data): snapshot, missing_snapshot = sample_data.snapshots[:2] snapshots = [snapshot.id, missing_snapshot.id] swh_storage.snapshot_add([snapshot]) missing_snapshots = swh_storage.snapshot_missing(snapshots) assert list(missing_snapshots) == [missing_snapshot.id] def test_stat_counters(self, swh_storage, sample_data): + if isinstance(swh_storage, CassandraStorage) and not isinstance( + swh_storage, InMemoryStorage + ): + pytest.skip("Cassandra backend does not support stat counters") origin = sample_data.origin snapshot = sample_data.snapshot revision = sample_data.revision release = sample_data.release directory = sample_data.directory content = sample_data.content expected_keys = ["content", "directory", "origin", "revision"] # Initially, all counters are 0 swh_storage.refresh_stat_counters() counters = swh_storage.stat_counters() assert set(expected_keys) <= set(counters) for key in expected_keys: assert counters[key] == 0 # Add a content. Only the content counter should increase. swh_storage.content_add([content]) swh_storage.refresh_stat_counters() counters = swh_storage.stat_counters() assert set(expected_keys) <= set(counters) for key in expected_keys: if key != "content": assert counters[key] == 0 assert counters["content"] == 1 # Add other objects. Check their counter increased as well. swh_storage.origin_add([origin]) visit = OriginVisit( origin=origin.url, date=sample_data.date_visit2, type=sample_data.type_visit2, ) origin_visit1 = swh_storage.origin_visit_add([visit])[0] swh_storage.snapshot_add([snapshot]) swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin.url, visit=origin_visit1.visit, date=now(), status="ongoing", snapshot=snapshot.id, ) ] ) swh_storage.directory_add([directory]) swh_storage.revision_add([revision]) swh_storage.release_add([release]) swh_storage.refresh_stat_counters() counters = swh_storage.stat_counters() assert counters["content"] == 1 assert counters["directory"] == 1 assert counters["snapshot"] == 1 assert counters["origin"] == 1 assert counters["origin_visit"] == 1 assert counters["revision"] == 1 assert counters["release"] == 1 assert counters["snapshot"] == 1 if "person" in counters: assert counters["person"] == 3 def test_content_find_ctime(self, swh_storage, sample_data): origin_content = sample_data.content ctime = round_to_milliseconds(now()) content = attr.evolve(origin_content, data=None, ctime=ctime) swh_storage.content_add_metadata([content]) actually_present = swh_storage.content_find({"sha1": content.sha1}) assert actually_present[0] == content assert actually_present[0].ctime is not None assert actually_present[0].ctime.tzinfo is not None def test_content_find_with_present_content(self, swh_storage, sample_data): content = sample_data.content expected_content = attr.evolve(content, data=None) # 1. with something to find swh_storage.content_add([content]) actually_present = swh_storage.content_find({"sha1": content.sha1}) assert 1 == len(actually_present) assert actually_present[0] == expected_content # 2. with something to find actually_present = swh_storage.content_find({"sha1_git": content.sha1_git}) assert 1 == len(actually_present) assert actually_present[0] == expected_content # 3. with something to find actually_present = swh_storage.content_find({"sha256": content.sha256}) assert 1 == len(actually_present) assert actually_present[0] == expected_content # 4. with something to find actually_present = swh_storage.content_find(content.hashes()) assert 1 == len(actually_present) assert actually_present[0] == expected_content def test_content_find_with_non_present_content(self, swh_storage, sample_data): missing_content = sample_data.skipped_content # 1. with something that does not exist actually_present = swh_storage.content_find({"sha1": missing_content.sha1}) assert actually_present == [] # 2. with something that does not exist actually_present = swh_storage.content_find( {"sha1_git": missing_content.sha1_git} ) assert actually_present == [] # 3. with something that does not exist actually_present = swh_storage.content_find({"sha256": missing_content.sha256}) assert actually_present == [] def test_content_find_with_duplicate_input(self, swh_storage, sample_data): content = sample_data.content # Create fake data with colliding sha256 and blake2s256 sha1_array = bytearray(content.sha1) sha1_array[0] += 1 sha1git_array = bytearray(content.sha1_git) sha1git_array[0] += 1 duplicated_content = attr.evolve( content, sha1=bytes(sha1_array), sha1_git=bytes(sha1git_array) ) # Inject the data swh_storage.content_add([content, duplicated_content]) actual_result = swh_storage.content_find( { "blake2s256": duplicated_content.blake2s256, "sha256": duplicated_content.sha256, } ) expected_content = attr.evolve(content, data=None) expected_duplicated_content = attr.evolve(duplicated_content, data=None) for result in actual_result: assert result in [expected_content, expected_duplicated_content] def test_content_find_with_duplicate_sha256(self, swh_storage, sample_data): content = sample_data.content hashes = {} # Create fake data with colliding sha256 for hashalgo in ("sha1", "sha1_git", "blake2s256"): value = bytearray(getattr(content, hashalgo)) value[0] += 1 hashes[hashalgo] = bytes(value) duplicated_content = attr.evolve( content, sha1=hashes["sha1"], sha1_git=hashes["sha1_git"], blake2s256=hashes["blake2s256"], ) swh_storage.content_add([content, duplicated_content]) actual_result = swh_storage.content_find({"sha256": duplicated_content.sha256}) assert len(actual_result) == 2 expected_content = attr.evolve(content, data=None) expected_duplicated_content = attr.evolve(duplicated_content, data=None) for result in actual_result: assert result in [expected_content, expected_duplicated_content] # Find with both sha256 and blake2s256 actual_result = swh_storage.content_find( { "sha256": duplicated_content.sha256, "blake2s256": duplicated_content.blake2s256, } ) assert len(actual_result) == 1 assert actual_result == [expected_duplicated_content] def test_content_find_with_duplicate_blake2s256(self, swh_storage, sample_data): content = sample_data.content # Create fake data with colliding sha256 and blake2s256 sha1_array = bytearray(content.sha1) sha1_array[0] += 1 sha1git_array = bytearray(content.sha1_git) sha1git_array[0] += 1 sha256_array = bytearray(content.sha256) sha256_array[0] += 1 duplicated_content = attr.evolve( content, sha1=bytes(sha1_array), sha1_git=bytes(sha1git_array), sha256=bytes(sha256_array), ) swh_storage.content_add([content, duplicated_content]) actual_result = swh_storage.content_find( {"blake2s256": duplicated_content.blake2s256} ) expected_content = attr.evolve(content, data=None) expected_duplicated_content = attr.evolve(duplicated_content, data=None) for result in actual_result: assert result in [expected_content, expected_duplicated_content] # Find with both sha256 and blake2s256 actual_result = swh_storage.content_find( { "sha256": duplicated_content.sha256, "blake2s256": duplicated_content.blake2s256, } ) assert actual_result == [expected_duplicated_content] def test_content_find_bad_input(self, swh_storage): # 1. with no hash to lookup with pytest.raises(StorageArgumentException): swh_storage.content_find({}) # need at least one hash # 2. with bad hash with pytest.raises(StorageArgumentException): swh_storage.content_find({"unknown-sha1": "something"}) # not the right key def test_object_find_by_sha1_git(self, swh_storage, sample_data): content = sample_data.content directory = sample_data.directory revision = sample_data.revision release = sample_data.release sha1_gits = [b"00000000000000000000"] expected = { b"00000000000000000000": [], } swh_storage.content_add([content]) sha1_gits.append(content.sha1_git) expected[content.sha1_git] = [ {"sha1_git": content.sha1_git, "type": "content",} ] swh_storage.directory_add([directory]) sha1_gits.append(directory.id) expected[directory.id] = [{"sha1_git": directory.id, "type": "directory",}] swh_storage.revision_add([revision]) sha1_gits.append(revision.id) expected[revision.id] = [{"sha1_git": revision.id, "type": "revision",}] swh_storage.release_add([release]) sha1_gits.append(release.id) expected[release.id] = [{"sha1_git": release.id, "type": "release",}] ret = swh_storage.object_find_by_sha1_git(sha1_gits) assert expected == ret def test_metadata_fetcher_add_get(self, swh_storage, sample_data): fetcher = sample_data.metadata_fetcher actual_fetcher = swh_storage.metadata_fetcher_get(fetcher.name, fetcher.version) assert actual_fetcher is None # does not exist swh_storage.metadata_fetcher_add([fetcher]) res = swh_storage.metadata_fetcher_get(fetcher.name, fetcher.version) assert res == fetcher actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = [ ("metadata_fetcher", fetcher), ] for obj in expected_objects: assert obj in actual_objects def test_metadata_fetcher_add_zero(self, swh_storage, sample_data): fetcher = sample_data.metadata_fetcher actual_fetcher = swh_storage.metadata_fetcher_get(fetcher.name, fetcher.version) assert actual_fetcher is None # does not exist swh_storage.metadata_fetcher_add([]) def test_metadata_authority_add_get(self, swh_storage, sample_data): authority = sample_data.metadata_authority actual_authority = swh_storage.metadata_authority_get( authority.type, authority.url ) assert actual_authority is None # does not exist swh_storage.metadata_authority_add([authority]) res = swh_storage.metadata_authority_get(authority.type, authority.url) assert res == authority actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = [ ("metadata_authority", authority), ] for obj in expected_objects: assert obj in actual_objects def test_metadata_authority_add_zero(self, swh_storage, sample_data): authority = sample_data.metadata_authority actual_authority = swh_storage.metadata_authority_get( authority.type, authority.url ) assert actual_authority is None # does not exist swh_storage.metadata_authority_add([]) def test_content_metadata_add(self, swh_storage, sample_data): content = sample_data.content fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority content_metadata = sample_data.content_metadata[:2] swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add(content_metadata) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority ) assert result.next_page_token is None assert list(sorted(result.results, key=lambda x: x.discovery_date,)) == list( content_metadata ) actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = [ ("metadata_authority", authority), ("metadata_fetcher", fetcher), ] + [("raw_extrinsic_metadata", item) for item in content_metadata] for obj in expected_objects: assert obj in actual_objects def test_content_metadata_add_duplicate(self, swh_storage, sample_data): """Duplicates should be silently ignored.""" content = sample_data.content fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority content_metadata, content_metadata2 = sample_data.content_metadata[:2] swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([content_metadata, content_metadata2]) swh_storage.raw_extrinsic_metadata_add([content_metadata2, content_metadata]) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority ) assert result.next_page_token is None expected_results = (content_metadata, content_metadata2) assert ( tuple(sorted(result.results, key=lambda x: x.discovery_date,)) == expected_results ) def test_content_metadata_get(self, swh_storage, sample_data): content, content2 = sample_data.contents[:2] fetcher, fetcher2 = sample_data.fetchers[:2] authority, authority2 = sample_data.authorities[:2] ( content1_metadata1, content1_metadata2, content1_metadata3, ) = sample_data.content_metadata[:3] content2_metadata = RawExtrinsicMetadata.from_dict( { **remove_keys(content1_metadata2.to_dict(), ("id",)), # recompute id "target": str(content2.swhid()), } ) swh_storage.metadata_authority_add([authority, authority2]) swh_storage.metadata_fetcher_add([fetcher, fetcher2]) swh_storage.raw_extrinsic_metadata_add( [ content1_metadata1, content1_metadata2, content1_metadata3, content2_metadata, ] ) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority ) assert result.next_page_token is None assert [content1_metadata1, content1_metadata2] == list( sorted(result.results, key=lambda x: x.discovery_date,) ) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority2 ) assert result.next_page_token is None assert [content1_metadata3] == list( sorted(result.results, key=lambda x: x.discovery_date,) ) result = swh_storage.raw_extrinsic_metadata_get( content2.swhid().to_extended(), authority ) assert result.next_page_token is None assert [content2_metadata] == list(result.results,) def test_content_metadata_get_after(self, swh_storage, sample_data): content = sample_data.content fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority content_metadata, content_metadata2 = sample_data.content_metadata[:2] swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([content_metadata, content_metadata2]) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, after=content_metadata.discovery_date - timedelta(seconds=1), ) assert result.next_page_token is None assert [content_metadata, content_metadata2] == list( sorted(result.results, key=lambda x: x.discovery_date,) ) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, after=content_metadata.discovery_date, ) assert result.next_page_token is None assert result.results == [content_metadata2] result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, after=content_metadata2.discovery_date, ) assert result.next_page_token is None assert result.results == [] def test_content_metadata_get_paginate(self, swh_storage, sample_data): content = sample_data.content fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority content_metadata, content_metadata2 = sample_data.content_metadata[:2] swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([content_metadata, content_metadata2]) swh_storage.raw_extrinsic_metadata_get(content.swhid().to_extended(), authority) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, limit=1 ) assert result.next_page_token is not None assert result.results == [content_metadata] result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, limit=1, page_token=result.next_page_token, ) assert result.next_page_token is None assert result.results == [content_metadata2] def test_content_metadata_get_paginate_same_date(self, swh_storage, sample_data): content = sample_data.content fetcher1, fetcher2 = sample_data.fetchers[:2] authority = sample_data.metadata_authority content_metadata, content_metadata2 = sample_data.content_metadata[:2] swh_storage.metadata_fetcher_add([fetcher1, fetcher2]) swh_storage.metadata_authority_add([authority]) new_content_metadata2 = RawExtrinsicMetadata.from_dict( { **remove_keys(content_metadata2.to_dict(), ("id",)), # recompute id "discovery_date": content_metadata2.discovery_date, "fetcher": attr.evolve(fetcher2, metadata=None).to_dict(), } ) swh_storage.raw_extrinsic_metadata_add( [content_metadata, new_content_metadata2] ) result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, limit=1 ) assert result.next_page_token is not None assert result.results == [content_metadata] result = swh_storage.raw_extrinsic_metadata_get( content.swhid().to_extended(), authority, limit=1, page_token=result.next_page_token, ) assert result.next_page_token is None assert result.results[0].to_dict() == new_content_metadata2.to_dict() assert result.results == [new_content_metadata2] def test_content_metadata_get_by_ids(self, swh_storage, sample_data): content, content2 = sample_data.contents[:2] fetcher, fetcher2 = sample_data.fetchers[:2] authority, authority2 = sample_data.authorities[:2] ( content1_metadata1, content1_metadata2, content1_metadata3, ) = sample_data.content_metadata[:3] content2_metadata = RawExtrinsicMetadata.from_dict( { **remove_keys(content1_metadata2.to_dict(), ("id",)), # recompute id "target": str(content2.swhid()), } ) swh_storage.metadata_authority_add([authority, authority2]) swh_storage.metadata_fetcher_add([fetcher, fetcher2]) swh_storage.raw_extrinsic_metadata_add( [ content1_metadata1, content1_metadata2, content1_metadata3, content2_metadata, ] ) assert set( swh_storage.raw_extrinsic_metadata_get_by_ids( [content1_metadata1.id, b"\x00" * 20, content2_metadata.id] ) ) == {content1_metadata1, content2_metadata} def test_content_metadata_get_authorities(self, swh_storage, sample_data): content1, content2, content3 = sample_data.contents[:3] fetcher, fetcher2 = sample_data.fetchers[:2] authority, authority2 = sample_data.authorities[:2] ( content1_metadata1, content1_metadata2, content1_metadata3, ) = sample_data.content_metadata[:3] content2_metadata = RawExtrinsicMetadata.from_dict( { **remove_keys(content1_metadata2.to_dict(), ("id",)), # recompute id "target": str(content2.swhid()), } ) content1_metadata2 = RawExtrinsicMetadata.from_dict( { **remove_keys(content1_metadata2.to_dict(), ("id",)), # recompute id "authority": authority2.to_dict(), } ) swh_storage.metadata_authority_add([authority, authority2]) swh_storage.metadata_fetcher_add([fetcher, fetcher2]) swh_storage.raw_extrinsic_metadata_add( [ content1_metadata1, content1_metadata2, content1_metadata3, content2_metadata, ] ) assert swh_storage.raw_extrinsic_metadata_get_authorities(content1.swhid()) in ( [authority, authority2], [authority2, authority], ) assert swh_storage.raw_extrinsic_metadata_get_authorities(content2.swhid()) == [ authority ] assert ( swh_storage.raw_extrinsic_metadata_get_authorities(content3.swhid()) == [] ) def test_origin_metadata_add(self, swh_storage, sample_data): origin = sample_data.origin fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority ) assert result.next_page_token is None assert list(sorted(result.results, key=lambda x: x.discovery_date)) == [ origin_metadata, origin_metadata2, ] actual_objects = list(swh_storage.journal_writer.journal.objects) expected_objects = [ ("metadata_authority", authority), ("metadata_fetcher", fetcher), ("raw_extrinsic_metadata", origin_metadata), ("raw_extrinsic_metadata", origin_metadata2), ] for obj in expected_objects: assert obj in actual_objects def test_origin_metadata_add_duplicate(self, swh_storage, sample_data): """Duplicates should be silently updated.""" origin = sample_data.origin fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) swh_storage.raw_extrinsic_metadata_add([origin_metadata2, origin_metadata]) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority ) assert result.next_page_token is None # which of the two behavior happens is backend-specific. expected_results = (origin_metadata, origin_metadata2) assert ( tuple(sorted(result.results, key=lambda x: x.discovery_date,)) == expected_results ) def test_origin_metadata_get(self, swh_storage, sample_data): origin, origin2 = sample_data.origins[:2] fetcher, fetcher2 = sample_data.fetchers[:2] authority, authority2 = sample_data.authorities[:2] ( origin1_metadata1, origin1_metadata2, origin1_metadata3, ) = sample_data.origin_metadata[:3] assert swh_storage.origin_add([origin, origin2]) == {"origin:add": 2} origin2_metadata = RawExtrinsicMetadata.from_dict( { **remove_keys(origin1_metadata2.to_dict(), ("id",)), # recompute id "target": str(Origin(origin2.url).swhid()), } ) swh_storage.metadata_authority_add([authority, authority2]) swh_storage.metadata_fetcher_add([fetcher, fetcher2]) swh_storage.raw_extrinsic_metadata_add( [origin1_metadata1, origin1_metadata2, origin1_metadata3, origin2_metadata] ) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority ) assert result.next_page_token is None assert [origin1_metadata1, origin1_metadata2] == list( sorted(result.results, key=lambda x: x.discovery_date,) ) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority2 ) assert result.next_page_token is None assert [origin1_metadata3] == list( sorted(result.results, key=lambda x: x.discovery_date,) ) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin2.url).swhid(), authority ) assert result.next_page_token is None assert [origin2_metadata] == list(result.results,) def test_origin_metadata_get_after(self, swh_storage, sample_data): origin = sample_data.origin fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, after=origin_metadata.discovery_date - timedelta(seconds=1), ) assert result.next_page_token is None assert list(sorted(result.results, key=lambda x: x.discovery_date,)) == [ origin_metadata, origin_metadata2, ] result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, after=origin_metadata.discovery_date, ) assert result.next_page_token is None assert result.results == [origin_metadata2] result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, after=origin_metadata2.discovery_date, ) assert result.next_page_token is None assert result.results == [] def test_origin_metadata_get_paginate(self, swh_storage, sample_data): origin = sample_data.origin fetcher = sample_data.metadata_fetcher authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher]) swh_storage.metadata_authority_add([authority]) swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) swh_storage.raw_extrinsic_metadata_get(Origin(origin.url).swhid(), authority) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, limit=1 ) assert result.next_page_token is not None assert result.results == [origin_metadata] result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, limit=1, page_token=result.next_page_token, ) assert result.next_page_token is None assert result.results == [origin_metadata2] def test_origin_metadata_get_paginate_same_date(self, swh_storage, sample_data): origin = sample_data.origin fetcher1, fetcher2 = sample_data.fetchers[:2] authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher1, fetcher2]) swh_storage.metadata_authority_add([authority]) new_origin_metadata2 = RawExtrinsicMetadata.from_dict( { **remove_keys(origin_metadata2.to_dict(), ("id",)), # recompute id "discovery_date": origin_metadata2.discovery_date, "fetcher": attr.evolve(fetcher2, metadata=None).to_dict(), } ) swh_storage.raw_extrinsic_metadata_add([origin_metadata, new_origin_metadata2]) result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, limit=1 ) assert result.next_page_token is not None assert result.results == [origin_metadata] result = swh_storage.raw_extrinsic_metadata_get( Origin(origin.url).swhid(), authority, limit=1, page_token=result.next_page_token, ) assert result.next_page_token is None assert result.results == [new_origin_metadata2] def test_origin_metadata_add_missing_authority(self, swh_storage, sample_data): origin = sample_data.origin fetcher = sample_data.metadata_fetcher origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_fetcher_add([fetcher]) with pytest.raises(StorageArgumentException, match="authority"): swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) def test_origin_metadata_add_missing_fetcher(self, swh_storage, sample_data): origin = sample_data.origin authority = sample_data.metadata_authority origin_metadata, origin_metadata2 = sample_data.origin_metadata[:2] assert swh_storage.origin_add([origin]) == {"origin:add": 1} swh_storage.metadata_authority_add([authority]) with pytest.raises(StorageArgumentException, match="fetcher"): swh_storage.raw_extrinsic_metadata_add([origin_metadata, origin_metadata2]) class TestStorageGeneratedData: def test_generate_content_get_data(self, swh_storage, swh_contents): contents_with_data = [c for c in swh_contents if c.status != "absent"] # retrieve contents for content in contents_with_data: actual_content_data = swh_storage.content_get_data(content.sha1) assert actual_content_data is not None assert actual_content_data == content.data def test_generate_content_get(self, swh_storage, swh_contents): expected_contents = [ attr.evolve(c, data=None) for c in swh_contents if c.status != "absent" ] actual_contents = swh_storage.content_get([c.sha1 for c in expected_contents]) assert len(actual_contents) == len(expected_contents) assert actual_contents == expected_contents @pytest.mark.parametrize("limit", [1, 7, 10, 100, 1000]) def test_origin_list(self, swh_storage, swh_origins, limit): returned_origins = [] page_token = None i = 0 while True: actual_page = swh_storage.origin_list(page_token=page_token, limit=limit) assert len(actual_page.results) <= limit returned_origins.extend(actual_page.results) i += 1 page_token = actual_page.next_page_token if page_token is None: assert i * limit >= len(swh_origins) break else: assert len(actual_page.results) == limit assert sorted(returned_origins) == sorted(swh_origins) def test_origin_count(self, swh_storage, sample_data): swh_storage.origin_add(sample_data.origins) assert swh_storage.origin_count("github") == 3 assert swh_storage.origin_count("gitlab") == 2 assert swh_storage.origin_count(".*user.*", regexp=True) == 5 assert swh_storage.origin_count(".*user.*", regexp=False) == 0 assert swh_storage.origin_count(".*user1.*", regexp=True) == 2 assert swh_storage.origin_count(".*user1.*", regexp=False) == 0 def test_origin_count_with_visit_no_visits(self, swh_storage, sample_data): swh_storage.origin_add(sample_data.origins) # none of them have visits, so with_visit=True => 0 assert swh_storage.origin_count("github", with_visit=True) == 0 assert swh_storage.origin_count("gitlab", with_visit=True) == 0 assert swh_storage.origin_count(".*user.*", regexp=True, with_visit=True) == 0 assert swh_storage.origin_count(".*user.*", regexp=False, with_visit=True) == 0 assert swh_storage.origin_count(".*user1.*", regexp=True, with_visit=True) == 0 assert swh_storage.origin_count(".*user1.*", regexp=False, with_visit=True) == 0 def test_origin_count_with_visit_with_visits_no_snapshot( self, swh_storage, sample_data ): swh_storage.origin_add(sample_data.origins) origin_url = "https://github.com/user1/repo1" visit = OriginVisit(origin=origin_url, date=now(), type="git",) swh_storage.origin_visit_add([visit]) assert swh_storage.origin_count("github", with_visit=False) == 3 # it has a visit, but no snapshot, so with_visit=True => 0 assert swh_storage.origin_count("github", with_visit=True) == 0 assert swh_storage.origin_count("gitlab", with_visit=False) == 2 # these gitlab origins have no visit assert swh_storage.origin_count("gitlab", with_visit=True) == 0 assert ( swh_storage.origin_count("github.*user1", regexp=True, with_visit=False) == 1 ) assert ( swh_storage.origin_count("github.*user1", regexp=True, with_visit=True) == 0 ) assert swh_storage.origin_count("github", regexp=True, with_visit=True) == 0 def test_origin_count_with_visit_with_visits_and_snapshot( self, swh_storage, sample_data ): snapshot = sample_data.snapshot swh_storage.origin_add(sample_data.origins) swh_storage.snapshot_add([snapshot]) origin_url = "https://github.com/user1/repo1" visit = OriginVisit(origin=origin_url, date=now(), type="git",) visit = swh_storage.origin_visit_add([visit])[0] swh_storage.origin_visit_status_add( [ OriginVisitStatus( origin=origin_url, visit=visit.visit, date=now(), status="ongoing", snapshot=snapshot.id, ) ] ) assert swh_storage.origin_count("github", with_visit=False) == 3 # github/user1 has a visit and a snapshot, so with_visit=True => 1 assert swh_storage.origin_count("github", with_visit=True) == 1 assert ( swh_storage.origin_count("github.*user1", regexp=True, with_visit=False) == 1 ) assert ( swh_storage.origin_count("github.*user1", regexp=True, with_visit=True) == 1 ) assert swh_storage.origin_count("github", regexp=True, with_visit=True) == 1 @settings( suppress_health_check=[HealthCheck.too_slow] + function_scoped_fixture_check, ) @given(strategies.lists(objects(split_content=True), max_size=2)) def test_add_arbitrary(self, swh_storage, objects): for (obj_type, obj) in objects: if obj.object_type == "origin_visit": swh_storage.origin_add([Origin(url=obj.origin)]) visit = OriginVisit(origin=obj.origin, date=obj.date, type=obj.type,) swh_storage.origin_visit_add([visit]) elif obj.object_type == "raw_extrinsic_metadata": swh_storage.metadata_authority_add([obj.authority]) swh_storage.metadata_fetcher_add([obj.fetcher]) swh_storage.raw_extrinsic_metadata_add([obj]) else: method = getattr(swh_storage, obj_type + "_add") try: method([obj]) except HashCollision: pass diff --git a/swh/storage/tests/test_cassandra.py b/swh/storage/tests/test_cassandra.py index 970938d3..60fa042a 100644 --- a/swh/storage/tests/test_cassandra.py +++ b/swh/storage/tests/test_cassandra.py @@ -1,704 +1,745 @@ # Copyright (C) 2018-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import itertools import os import resource import signal import socket import subprocess import time from typing import Any, Dict import attr from cassandra.cluster import NoHostAvailable import pytest from swh.core.api.classes import stream_results +from swh.model import from_disk from swh.model.model import Directory, DirectoryEntry, Snapshot, SnapshotBranch from swh.storage import get_storage from swh.storage.cassandra import create_keyspace +from swh.storage.cassandra.cql import BATCH_INSERT_MAX_SIZE from swh.storage.cassandra.model import ContentRow, ExtIDRow from swh.storage.cassandra.schema import HASH_ALGORITHMS, TABLES +from swh.storage.cassandra.storage import DIRECTORY_ENTRIES_INSERT_ALGOS from swh.storage.tests.storage_data import StorageData from swh.storage.tests.storage_tests import ( TestStorageGeneratedData as _TestStorageGeneratedData, ) from swh.storage.tests.storage_tests import TestStorage as _TestStorage from swh.storage.utils import now, remove_keys CONFIG_TEMPLATE = """ data_file_directories: - {data_dir}/data commitlog_directory: {data_dir}/commitlog hints_directory: {data_dir}/hints saved_caches_directory: {data_dir}/saved_caches commitlog_sync: periodic commitlog_sync_period_in_ms: 1000000 partitioner: org.apache.cassandra.dht.Murmur3Partitioner endpoint_snitch: SimpleSnitch seed_provider: - class_name: org.apache.cassandra.locator.SimpleSeedProvider parameters: - seeds: "127.0.0.1" storage_port: {storage_port} native_transport_port: {native_transport_port} start_native_transport: true listen_address: 127.0.0.1 enable_user_defined_functions: true # speed-up by disabling period saving to disk key_cache_save_period: 0 row_cache_save_period: 0 trickle_fsync: false commitlog_sync_period_in_ms: 100000 """ SCYLLA_EXTRA_CONFIG_TEMPLATE = """ experimental_features: - udf view_hints_directory: {data_dir}/view_hints prometheus_port: 0 # disable prometheus server start_rpc: false # disable thrift server api_port: {api_port} """ def free_port(): sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) sock.bind(("127.0.0.1", 0)) port = sock.getsockname()[1] sock.close() return port def wait_for_peer(addr, port): wait_until = time.time() + 60 while time.time() < wait_until: try: sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) sock.connect((addr, port)) except ConnectionRefusedError: time.sleep(0.1) else: sock.close() return True return False @pytest.fixture(scope="session") def cassandra_cluster(tmpdir_factory): cassandra_conf = tmpdir_factory.mktemp("cassandra_conf") cassandra_data = tmpdir_factory.mktemp("cassandra_data") cassandra_log = tmpdir_factory.mktemp("cassandra_log") native_transport_port = free_port() storage_port = free_port() jmx_port = free_port() api_port = free_port() use_scylla = bool(os.environ.get("SWH_USE_SCYLLADB", "")) cassandra_bin = os.environ.get( "SWH_CASSANDRA_BIN", "/usr/bin/scylla" if use_scylla else "/usr/sbin/cassandra" ) if use_scylla: os.makedirs(cassandra_conf.join("conf")) config_path = cassandra_conf.join("conf/scylla.yaml") config_template = CONFIG_TEMPLATE + SCYLLA_EXTRA_CONFIG_TEMPLATE else: config_path = cassandra_conf.join("cassandra.yaml") config_template = CONFIG_TEMPLATE with open(str(config_path), "w") as fd: fd.write( config_template.format( data_dir=str(cassandra_data), storage_port=storage_port, native_transport_port=native_transport_port, api_port=api_port, ) ) if os.environ.get("SWH_CASSANDRA_LOG"): stdout = stderr = None else: stdout = stderr = subprocess.DEVNULL env = { "MAX_HEAP_SIZE": "300M", "HEAP_NEWSIZE": "50M", "JVM_OPTS": "-Xlog:gc=error:file=%s/gc.log" % cassandra_log, } if "JAVA_HOME" in os.environ: env["JAVA_HOME"] = os.environ["JAVA_HOME"] if use_scylla: env = { **env, "SCYLLA_HOME": cassandra_conf, } # prevent "NOFILE rlimit too low (recommended setting 200000, # minimum setting 10000; refusing to start." resource.setrlimit(resource.RLIMIT_NOFILE, (200000, 200000)) proc = subprocess.Popen( [cassandra_bin, "--developer-mode=1",], start_new_session=True, env=env, stdout=stdout, stderr=stderr, ) else: proc = subprocess.Popen( [ cassandra_bin, "-Dcassandra.config=file://%s/cassandra.yaml" % cassandra_conf, "-Dcassandra.logdir=%s" % cassandra_log, "-Dcassandra.jmx.local.port=%d" % jmx_port, "-Dcassandra-foreground=yes", ], start_new_session=True, env=env, stdout=stdout, stderr=stderr, ) listening = wait_for_peer("127.0.0.1", native_transport_port) if listening: yield (["127.0.0.1"], native_transport_port) if not listening or os.environ.get("SWH_CASSANDRA_LOG"): debug_log_path = str(cassandra_log.join("debug.log")) if os.path.exists(debug_log_path): with open(debug_log_path) as fd: print(fd.read()) if not listening: if proc.poll() is None: raise Exception("cassandra process unexpectedly not listening.") else: raise Exception("cassandra process unexpectedly stopped.") pgrp = os.getpgid(proc.pid) os.killpg(pgrp, signal.SIGKILL) class RequestHandler: def on_request(self, rf): if hasattr(rf.message, "query"): print() print(rf.message.query) @pytest.fixture(scope="session") def keyspace(cassandra_cluster): (hosts, port) = cassandra_cluster keyspace = os.urandom(10).hex() create_keyspace(hosts, keyspace, port) return keyspace # tests are executed using imported classes (TestStorage and # TestStorageGeneratedData) using overloaded swh_storage fixture # below @pytest.fixture def swh_storage_backend_config(cassandra_cluster, keyspace): (hosts, port) = cassandra_cluster storage_config = dict( cls="cassandra", hosts=hosts, port=port, keyspace=keyspace, journal_writer={"cls": "memory"}, objstorage={"cls": "memory"}, ) yield storage_config storage = get_storage(**storage_config) for table in TABLES: storage._cql_runner._session.execute('TRUNCATE TABLE "%s"' % table) storage._cql_runner._cluster.shutdown() @pytest.mark.cassandra class TestCassandraStorage(_TestStorage): def test_config_wrong_consistency_should_raise(self): storage_config = dict( cls="cassandra", hosts=["first"], port=9999, keyspace="any", consistency_level="fake", journal_writer={"cls": "memory"}, objstorage={"cls": "memory"}, ) with pytest.raises(ValueError, match="Unknown consistency"): get_storage(**storage_config) def test_config_consistency_used(self, swh_storage_backend_config): config_with_consistency = dict( swh_storage_backend_config, **{"consistency_level": "THREE"} ) storage = get_storage(**config_with_consistency) with pytest.raises(NoHostAvailable): storage.content_get_random() def test_content_add_murmur3_collision(self, swh_storage, mocker, sample_data): """The Murmur3 token is used as link from index tables to the main table; and non-matching contents with colliding murmur3-hash are filtered-out when reading the main table. This test checks the content methods do filter out these collision. """ called = 0 cont, cont2 = sample_data.contents[:2] # always return a token - def mock_cgtfsh(algo, hash_): + def mock_cgtfsa(algo, hashes): nonlocal called called += 1 assert algo in ("sha1", "sha1_git") return [123456] mocker.patch.object( - swh_storage._cql_runner, "content_get_tokens_from_single_hash", mock_cgtfsh, + swh_storage._cql_runner, "content_get_tokens_from_single_algo", mock_cgtfsa, ) # For all tokens, always return cont def mock_cgft(token): nonlocal called called += 1 return [ ContentRow( length=10, ctime=datetime.datetime.now(), status="present", **{algo: getattr(cont, algo) for algo in HASH_ALGORITHMS}, ) ] mocker.patch.object( swh_storage._cql_runner, "content_get_from_token", mock_cgft ) actual_result = swh_storage.content_add([cont2]) assert called == 4 assert actual_result == { "content:add": 1, "content:add:bytes": cont2.length, } def test_content_get_metadata_murmur3_collision( self, swh_storage, mocker, sample_data ): """The Murmur3 token is used as link from index tables to the main table; and non-matching contents with colliding murmur3-hash are filtered-out when reading the main table. This test checks the content methods do filter out these collisions. """ called = 0 cont, cont2 = [attr.evolve(c, ctime=now()) for c in sample_data.contents[:2]] # always return a token - def mock_cgtfsh(algo, hash_): + def mock_cgtfsa(algo, hashes): nonlocal called called += 1 assert algo in ("sha1", "sha1_git") return [123456] mocker.patch.object( - swh_storage._cql_runner, "content_get_tokens_from_single_hash", mock_cgtfsh, + swh_storage._cql_runner, "content_get_tokens_from_single_algo", mock_cgtfsa, ) # For all tokens, always return cont and cont2 cols = list(set(cont.to_dict()) - {"data"}) def mock_cgft(token): nonlocal called called += 1 return [ ContentRow(**{col: getattr(cont, col) for col in cols},) for cont in [cont, cont2] ] mocker.patch.object( swh_storage._cql_runner, "content_get_from_token", mock_cgft ) actual_result = swh_storage.content_get([cont.sha1]) assert called == 2 # dropping extra column not returned expected_cont = attr.evolve(cont, data=None) # but cont2 should be filtered out assert actual_result == [expected_cont] def test_content_find_murmur3_collision(self, swh_storage, mocker, sample_data): """The Murmur3 token is used as link from index tables to the main table; and non-matching contents with colliding murmur3-hash are filtered-out when reading the main table. This test checks the content methods do filter out these collisions. """ called = 0 cont, cont2 = [attr.evolve(c, ctime=now()) for c in sample_data.contents[:2]] # always return a token - def mock_cgtfsh(algo, hash_): + def mock_cgtfsa(algo, hashes): nonlocal called called += 1 assert algo in ("sha1", "sha1_git") return [123456] mocker.patch.object( - swh_storage._cql_runner, "content_get_tokens_from_single_hash", mock_cgtfsh, + swh_storage._cql_runner, "content_get_tokens_from_single_algo", mock_cgtfsa, ) # For all tokens, always return cont and cont2 cols = list(set(cont.to_dict()) - {"data"}) def mock_cgft(token): nonlocal called called += 1 return [ ContentRow(**{col: getattr(cont, col) for col in cols}) for cont in [cont, cont2] ] mocker.patch.object( swh_storage._cql_runner, "content_get_from_token", mock_cgft ) expected_content = attr.evolve(cont, data=None) actual_result = swh_storage.content_find({"sha1": cont.sha1}) assert called == 2 # but cont2 should be filtered out assert actual_result == [expected_content] def test_content_get_partition_murmur3_collision( self, swh_storage, mocker, sample_data ): """The Murmur3 token is used as link from index tables to the main table; and non-matching contents with colliding murmur3-hash are filtered-out when reading the main table. This test checks the content_get_partition endpoints return all contents, even the collisions. """ called = 0 rows: Dict[int, Dict] = {} for tok, content in enumerate(sample_data.contents): cont = attr.evolve(content, data=None, ctime=now()) row_d = {**cont.to_dict(), "tok": tok} rows[tok] = row_d # For all tokens, always return cont def mock_content_get_token_range(range_start, range_end, limit): nonlocal called called += 1 for tok in list(rows.keys()) * 3: # yield multiple times the same tok row_d = dict(rows[tok].items()) row_d.pop("tok") yield (tok, ContentRow(**row_d)) mocker.patch.object( swh_storage._cql_runner, "content_get_token_range", mock_content_get_token_range, ) actual_results = list( stream_results( swh_storage.content_get_partition, partition_id=0, nb_partitions=1 ) ) assert called > 0 # everything is listed, even collisions assert len(actual_results) == 3 * len(sample_data.contents) # as we duplicated the returned results, dropping duplicate should yield # the original length assert len(set(actual_results)) == len(sample_data.contents) @pytest.mark.skip("content_update is not yet implemented for Cassandra") def test_content_update(self): pass def test_extid_murmur3_collision(self, swh_storage, mocker, sample_data): """The Murmur3 token is used as link from index table to the main table; and non-matching extid with colliding murmur3-hash are filtered-out when reading the main table. This test checks the extid methods do filter out these collision. """ swh_storage.extid_add(sample_data.extids) # For any token, always return all extids, i.e. make as if all tokens # for all extid entries collide def mock_egft(token): return [ ExtIDRow( extid_type=extid.extid_type, extid=extid.extid, extid_version=extid.extid_version, target_type=extid.target.object_type.value, target=extid.target.object_id, ) for extid in sample_data.extids ] mocker.patch.object( swh_storage._cql_runner, "extid_get_from_token", mock_egft, ) for extid in sample_data.extids: extids = swh_storage.extid_get_from_target( target_type=extid.target.object_type, ids=[extid.target.object_id] ) assert extids == [extid] - def test_directory_add_atomic(self, swh_storage, sample_data, mocker): + def _directory_with_entries(self, sample_data, nb_entries): + """Returns a dir with ``nb_entries``, all pointing to + the same content""" + return Directory( + entries=tuple( + DirectoryEntry( + name=f"file{i:10}".encode(), + type="file", + target=sample_data.content.sha1_git, + perms=from_disk.DentryPerms.directory, + ) + for i in range(nb_entries) + ) + ) + + @pytest.mark.parametrize( + "insert_algo,nb_entries", + [ + ("one-by-one", 10), + ("concurrent", 10), + ("batch", 1), + ("batch", 2), + ("batch", BATCH_INSERT_MAX_SIZE - 1), + ("batch", BATCH_INSERT_MAX_SIZE), + ("batch", BATCH_INSERT_MAX_SIZE + 1), + ("batch", BATCH_INSERT_MAX_SIZE * 2), + ], + ) + def test_directory_add_algos( + self, swh_storage, sample_data, mocker, insert_algo, nb_entries, + ): + mocker.patch.object(swh_storage, "_directory_entries_insert_algo", insert_algo) + + class new_sample_data: + content = sample_data.content + directory = self._directory_with_entries(sample_data, nb_entries) + + self.test_directory_add(swh_storage, new_sample_data) + + @pytest.mark.parametrize("insert_algo", DIRECTORY_ENTRIES_INSERT_ALGOS) + def test_directory_add_atomic(self, swh_storage, sample_data, mocker, insert_algo): """Checks that a crash occurring after some directory entries were written does not cause the directory to be (partially) visible. ie. checks directories are added somewhat atomically.""" # Disable the journal writer, it would detect the CrashyEntry exception too # early for this test to be relevant swh_storage.journal_writer.journal = None - - class MyException(Exception): - pass + mocker.patch.object(swh_storage, "_directory_entries_insert_algo", insert_algo) class CrashyEntry(DirectoryEntry): def __init__(self): pass def to_dict(self): - raise MyException() + return {**directory.entries[0].to_dict(), "perms": "abcde"} - directory = sample_data.directory3 + directory = self._directory_with_entries(sample_data, BATCH_INSERT_MAX_SIZE) entries = directory.entries directory = attr.evolve(directory, entries=entries + (CrashyEntry(),)) - with pytest.raises(MyException): + with pytest.raises(TypeError): swh_storage.directory_add([directory]) # This should have written some of the entries to the database: entry_rows = swh_storage._cql_runner.directory_entry_get([directory.id]) assert {row.name for row in entry_rows} == {entry.name for entry in entries} # BUT, because not all the entries were written, the directory should # be considered not written. assert swh_storage.directory_missing([directory.id]) == [directory.id] assert list(swh_storage.directory_ls(directory.id)) == [] assert swh_storage.directory_get_entries(directory.id) is None def test_snapshot_add_atomic(self, swh_storage, sample_data, mocker): """Checks that a crash occurring after some snapshot branches were written does not cause the snapshot to be (partially) visible. ie. checks snapshots are added somewhat atomically.""" # Disable the journal writer, it would detect the CrashyBranch exception too # early for this test to be relevant swh_storage.journal_writer.journal = None class MyException(Exception): pass class CrashyBranch(SnapshotBranch): def __getattribute__(self, name): if name == "target" and should_raise: raise MyException() else: return super().__getattribute__(name) snapshot = sample_data.complete_snapshot branches = snapshot.branches should_raise = False # just so that we can construct the object crashy_branch = CrashyBranch.from_dict(branches[b"directory"].to_dict()) should_raise = True snapshot = attr.evolve( snapshot, branches={**branches, b"crashy": crashy_branch,}, ) with pytest.raises(MyException): swh_storage.snapshot_add([snapshot]) # This should have written some of the branches to the database: branch_rows = swh_storage._cql_runner.snapshot_branch_get(snapshot.id, b"", 10) assert {row.name for row in branch_rows} == set(branches) # BUT, because not all the branches were written, the snapshot should # be considered not written. assert swh_storage.snapshot_missing([snapshot.id]) == [snapshot.id] assert swh_storage.snapshot_get(snapshot.id) is None assert swh_storage.snapshot_count_branches(snapshot.id) is None assert swh_storage.snapshot_get_branches(snapshot.id) is None @pytest.mark.skip( 'The "person" table of the pgsql is a legacy thing, and not ' "supported by the cassandra backend." ) def test_person_fullname_unicity(self): pass @pytest.mark.skip( 'The "person" table of the pgsql is a legacy thing, and not ' "supported by the cassandra backend." ) def test_person_get(self): pass @pytest.mark.skip("Not supported by Cassandra") def test_origin_count(self): pass @pytest.mark.cassandra class TestCassandraStorageGeneratedData(_TestStorageGeneratedData): @pytest.mark.skip("Not supported by Cassandra") def test_origin_count(self): pass @pytest.mark.skip("Not supported by Cassandra") def test_origin_count_with_visit_no_visits(self): pass @pytest.mark.skip("Not supported by Cassandra") def test_origin_count_with_visit_with_visits_and_snapshot(self): pass @pytest.mark.skip("Not supported by Cassandra") def test_origin_count_with_visit_with_visits_no_snapshot(self): pass @pytest.mark.parametrize( "allow_overwrite,object_type", itertools.product( [False, True], # Note the absence of "content", it's tested above. ["directory", "revision", "release", "snapshot", "origin", "extid"], ), ) def test_allow_overwrite( allow_overwrite: bool, object_type: str, swh_storage_backend_config ): if object_type in ("origin", "extid"): pytest.skip( f"test_disallow_overwrite not implemented for {object_type} objects, " f"because all their columns are in the primary key." ) swh_storage = get_storage( allow_overwrite=allow_overwrite, **swh_storage_backend_config ) # directory_ls joins with content and directory table, and needs those to return # non-None entries: if object_type == "directory": swh_storage.directory_add([StorageData.directory5]) swh_storage.content_add([StorageData.content, StorageData.content2]) obj1: Any obj2: Any # Get two test objects if object_type == "directory": (obj1, obj2, *_) = StorageData.directories elif object_type == "snapshot": # StorageData.snapshots[1] is the empty snapshot, which is the corner case # that makes this test succeed for the wrong reasons obj1 = StorageData.snapshot obj2 = StorageData.complete_snapshot else: (obj1, obj2, *_) = getattr(StorageData, (object_type + "s")) # Let's make both objects have the same hash, but different content obj1 = attr.evolve(obj1, id=obj2.id) # Get the methods used to add and get these objects add = getattr(swh_storage, object_type + "_add") if object_type == "directory": def get(ids): return [ Directory( id=ids[0], entries=tuple( map( lambda entry: DirectoryEntry( name=entry["name"], type=entry["type"], target=entry["sha1_git"], perms=entry["perms"], ), swh_storage.directory_ls(ids[0]), ) ), ) ] elif object_type == "snapshot": def get(ids): return [ Snapshot.from_dict( remove_keys(swh_storage.snapshot_get(ids[0]), ("next_branch",)) ) ] else: get = getattr(swh_storage, object_type + "_get") # Add the first object add([obj1]) # It should be returned as-is assert get([obj1.id]) == [obj1] # Add the second object add([obj2]) if allow_overwrite: # obj1 was overwritten by obj2 expected = obj2 else: # obj2 was not written, because obj1 already exists and has the same hash expected = obj1 if allow_overwrite and object_type in ("directory", "snapshot"): # TODO pytest.xfail( "directory entries and snapshot branches are concatenated " "instead of being replaced" ) assert get([obj1.id]) == [expected] diff --git a/swh/storage/tests/test_counter.py b/swh/storage/tests/test_counter.py new file mode 100644 index 00000000..c44adee9 --- /dev/null +++ b/swh/storage/tests/test_counter.py @@ -0,0 +1,63 @@ +# Copyright (C) 2021 The Software Heritage developers +# See the AUTHORS file at the top-level directory of this distribution +# License: GNU General Public License version 3, or any later version +# See top-level LICENSE file for more information + +import attr +import pytest + +from swh.storage import get_storage + + +@pytest.fixture +def swh_storage(): + storage_config = { + "cls": "pipeline", + "steps": [ + {"cls": "counter", "counters": {"cls": "memory"}}, + {"cls": "memory"}, + ], + } + + return get_storage(**storage_config) + + +def test_counting_proxy_storage_content(swh_storage, sample_data): + assert swh_storage.counters.counters["content"] == set() + + swh_storage.content_add([sample_data.content]) + + assert swh_storage.counters.counters["content"] == {sample_data.content.sha1} + + swh_storage.content_add([sample_data.content2, sample_data.content3]) + + assert swh_storage.counters.counters["content"] == { + sample_data.content.sha1, + sample_data.content2.sha1, + sample_data.content3.sha1, + } + + assert [ + attr.evolve(cnt, ctime=None) + for cnt in swh_storage.content_find({"sha256": sample_data.content2.sha256}) + ] == [attr.evolve(sample_data.content2, data=None)] + + +def test_counting_proxy_storage_revision(swh_storage, sample_data): + assert swh_storage.counters.counters["revision"] == set() + + swh_storage.revision_add([sample_data.revision]) + + assert swh_storage.counters.counters["revision"] == {sample_data.revision.id} + + swh_storage.revision_add([sample_data.revision2, sample_data.revision3]) + + assert swh_storage.counters.counters["revision"] == { + sample_data.revision.id, + sample_data.revision2.id, + sample_data.revision3.id, + } + + assert swh_storage.revision_get([sample_data.revision2.id]) == [ + sample_data.revision2 + ]