diff --git a/swh/indexer/storage/__init__.py b/swh/indexer/storage/__init__.py index 2f4c285..787689b 100644 --- a/swh/indexer/storage/__init__.py +++ b/swh/indexer/storage/__init__.py @@ -1,658 +1,658 @@ # 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 from collections import Counter, defaultdict import itertools import json -from typing import Dict, Iterable, Iterator, List, Optional, Tuple +from typing import Dict, Iterable, List, Optional, Tuple import psycopg2 import psycopg2.pool from swh.core.db.common import db_transaction, db_transaction_generator from swh.model.hashutil import hash_to_bytes, hash_to_hex from swh.model.model import SHA1_SIZE from swh.storage.exc import StorageDBError from swh.storage.utils import get_partition_bounds_bytes from . import converters from .db import Db from .exc import DuplicateId, IndexerStorageArgumentException from .interface import PagedResult, Sha1 from .metrics import process_metrics, send_metric, timed from .model import ( ContentCtagsRow, ContentLanguageRow, ContentLicenseRow, ContentMetadataRow, ContentMimetypeRow, OriginIntrinsicMetadataRow, RevisionIntrinsicMetadataRow, ) INDEXER_CFG_KEY = "indexer_storage" MAPPING_NAMES = ["codemeta", "gemspec", "maven", "npm", "pkg-info"] def get_indexer_storage(cls, args): """Get an indexer storage object of class `storage_class` with arguments `storage_args`. Args: cls (str): storage's class, either 'local' or 'remote' args (dict): dictionary of arguments passed to the storage class constructor Returns: an instance of swh.indexer's storage (either local or remote) Raises: ValueError if passed an unknown storage class. """ if cls == "remote": from .api.client import RemoteStorage as IndexerStorage elif cls == "local": from . import IndexerStorage elif cls == "memory": from .in_memory import IndexerStorage else: raise ValueError("Unknown indexer storage class `%s`" % cls) return IndexerStorage(**args) def check_id_duplicates(data): """ If any two row models in `data` have the same unique key, raises a `ValueError`. Values associated to the key must be hashable. Args: data (List[dict]): List of dictionaries to be inserted >>> check_id_duplicates([ ... ContentLanguageRow(id=b'foo', indexer_configuration_id=42, lang="python"), ... ContentLanguageRow(id=b'foo', indexer_configuration_id=32, lang="python"), ... ]) >>> check_id_duplicates([ ... ContentLanguageRow(id=b'foo', indexer_configuration_id=42, lang="python"), ... ContentLanguageRow(id=b'foo', indexer_configuration_id=42, lang="python"), ... ]) Traceback (most recent call last): ... swh.indexer.storage.exc.DuplicateId: [{'id': b'foo', 'indexer_configuration_id': 42}] """ # noqa counter = Counter(tuple(sorted(item.unique_key().items())) for item in data) duplicates = [id_ for (id_, count) in counter.items() if count >= 2] if duplicates: raise DuplicateId(list(map(dict, duplicates))) class IndexerStorage: """SWH Indexer Storage """ def __init__(self, db, min_pool_conns=1, max_pool_conns=10): """ Args: db_conn: either a libpq connection string, or a psycopg2 connection """ 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) def get_db(self): if self._db: return self._db return Db.from_pool(self._pool) def put_db(self, db): if db is not self._db: db.put_conn() @timed @db_transaction() def check_config(self, *, check_write, db=None, cur=None): # Check permissions on one of the tables if check_write: check = "INSERT" else: check = "SELECT" cur.execute( "select has_table_privilege(current_user, 'content_mimetype', %s)", # noqa (check,), ) return cur.fetchone()[0] @timed @db_transaction_generator() def content_mimetype_missing( self, mimetypes: Iterable[Dict], db=None, cur=None - ) -> Iterator[Tuple[Sha1, int]]: + ) -> Iterable[Tuple[Sha1, int]]: for obj in db.content_mimetype_missing_from_list(mimetypes, cur): yield obj[0] @timed @db_transaction() def get_partition( self, indexer_type: str, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, with_textual_data=False, db=None, cur=None, ) -> PagedResult[Sha1]: """Retrieve ids of content with `indexer_type` within within partition partition_id bound by limit. Args: **indexer_type**: Type of data content to index (mimetype, language, etc...) **indexer_configuration_id**: The tool used to index data **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) **with_textual_data** (bool): Deal with only textual content (True) or all content (all contents by defaults, False) Raises: IndexerStorageArgumentException for; - limit to None - wrong indexer_type provided Returns: PagedResult of Sha1. If next_page_token is None, there is no more data to fetch """ if limit is None: raise IndexerStorageArgumentException("limit should not be None") if indexer_type not in db.content_indexer_names: err = f"Wrong type. Should be one of [{','.join(db.content_indexer_names)}]" raise IndexerStorageArgumentException(err) start, end = get_partition_bounds_bytes(partition_id, nb_partitions, SHA1_SIZE) if page_token is not None: start = hash_to_bytes(page_token) if end is None: end = b"\xff" * SHA1_SIZE next_page_token: Optional[str] = None ids = [ row[0] for row in db.content_get_range( indexer_type, start, end, indexer_configuration_id, limit=limit + 1, with_textual_data=with_textual_data, cur=cur, ) ] if len(ids) >= limit: next_page_token = hash_to_hex(ids[-1]) ids = ids[:limit] assert len(ids) <= limit return PagedResult(results=ids, next_page_token=next_page_token) @timed @db_transaction() def content_mimetype_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, db=None, cur=None, ) -> PagedResult[Sha1]: return self.get_partition( "mimetype", indexer_configuration_id, partition_id, nb_partitions, page_token=page_token, limit=limit, db=db, cur=cur, ) @timed @process_metrics @db_transaction() def content_mimetype_add( self, mimetypes: List[ContentMimetypeRow], conflict_update: bool = False, db=None, cur=None, ) -> Dict[str, int]: """Add mimetypes to the storage (if conflict_update is True, this will override existing data if any). Returns: A dict with the number of new elements added to the storage. """ check_id_duplicates(mimetypes) mimetypes.sort(key=lambda m: m.id) db.mktemp_content_mimetype(cur) db.copy_to( [m.to_dict() for m in mimetypes], "tmp_content_mimetype", ["id", "mimetype", "encoding", "indexer_configuration_id"], cur, ) count = db.content_mimetype_add_from_temp(conflict_update, cur) return {"content_mimetype:add": count} @timed @db_transaction_generator() def content_mimetype_get( self, ids: Iterable[Sha1], db=None, cur=None - ) -> Iterator[ContentMimetypeRow]: + ) -> Iterable[ContentMimetypeRow]: for c in db.content_mimetype_get_from_list(ids, cur): yield ContentMimetypeRow.from_dict( converters.db_to_mimetype(dict(zip(db.content_mimetype_cols, c))) ) @timed @db_transaction_generator() def content_language_missing(self, languages, db=None, cur=None): for obj in db.content_language_missing_from_list(languages, cur): yield obj[0] @timed @db_transaction_generator() def content_language_get(self, ids, db=None, cur=None): for c in db.content_language_get_from_list(ids, cur): yield converters.db_to_language(dict(zip(db.content_language_cols, c))) @timed @process_metrics @db_transaction() def content_language_add( self, languages: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: check_id_duplicates(map(ContentLanguageRow.from_dict, languages)) languages.sort(key=lambda m: m["id"]) db.mktemp_content_language(cur) # empty language is mapped to 'unknown' db.copy_to( ( { "id": lang["id"], "lang": "unknown" if not lang["lang"] else lang["lang"], "indexer_configuration_id": lang["indexer_configuration_id"], } for lang in languages ), "tmp_content_language", ["id", "lang", "indexer_configuration_id"], cur, ) count = db.content_language_add_from_temp(conflict_update, cur) return {"content_language:add": count} @timed @db_transaction_generator() def content_ctags_missing(self, ctags, db=None, cur=None): for obj in db.content_ctags_missing_from_list(ctags, cur): yield obj[0] @timed @db_transaction_generator() def content_ctags_get(self, ids, db=None, cur=None): for c in db.content_ctags_get_from_list(ids, cur): yield converters.db_to_ctags(dict(zip(db.content_ctags_cols, c))) @timed @process_metrics @db_transaction() def content_ctags_add( self, ctags: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: rows = list(itertools.chain.from_iterable(map(converters.ctags_to_db, ctags))) check_id_duplicates(map(ContentCtagsRow.from_dict, rows)) ctags.sort(key=lambda m: m["id"]) db.mktemp_content_ctags(cur) db.copy_to( rows, tblname="tmp_content_ctags", columns=["id", "name", "kind", "line", "lang", "indexer_configuration_id"], cur=cur, ) count = db.content_ctags_add_from_temp(conflict_update, cur) return {"content_ctags:add": count} @timed @db_transaction_generator() def content_ctags_search( self, expression, limit=10, last_sha1=None, db=None, cur=None ): for obj in db.content_ctags_search(expression, last_sha1, limit, cur=cur): yield converters.db_to_ctags(dict(zip(db.content_ctags_cols, obj))) @timed @db_transaction_generator() def content_fossology_license_get(self, ids, db=None, cur=None): d = defaultdict(list) for c in db.content_fossology_license_get_from_list(ids, cur): license = dict(zip(db.content_fossology_license_cols, c)) id_ = license["id"] d[id_].append(converters.db_to_fossology_license(license)) for id_, facts in d.items(): yield {id_: facts} @timed @process_metrics @db_transaction() def content_fossology_license_add( self, licenses: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: rows = list( itertools.chain.from_iterable( map(converters.fossology_license_to_db, licenses) ) ) check_id_duplicates(map(ContentLicenseRow.from_dict, rows)) licenses.sort(key=lambda m: m["id"]) db.mktemp_content_fossology_license(cur) db.copy_to( rows, tblname="tmp_content_fossology_license", columns=["id", "license", "indexer_configuration_id"], cur=cur, ) count = db.content_fossology_license_add_from_temp(conflict_update, cur) return {"content_fossology_license:add": count} @timed @db_transaction() def content_fossology_license_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, db=None, cur=None, ) -> PagedResult[Sha1]: return self.get_partition( "fossology_license", indexer_configuration_id, partition_id, nb_partitions, page_token=page_token, limit=limit, with_textual_data=True, db=db, cur=cur, ) @timed @db_transaction_generator() def content_metadata_missing(self, metadata, db=None, cur=None): for obj in db.content_metadata_missing_from_list(metadata, cur): yield obj[0] @timed @db_transaction_generator() def content_metadata_get(self, ids, db=None, cur=None): for c in db.content_metadata_get_from_list(ids, cur): yield converters.db_to_metadata(dict(zip(db.content_metadata_cols, c))) @timed @process_metrics @db_transaction() def content_metadata_add( self, metadata: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: check_id_duplicates(map(ContentMetadataRow.from_dict, metadata)) metadata.sort(key=lambda m: m["id"]) db.mktemp_content_metadata(cur) db.copy_to( metadata, "tmp_content_metadata", ["id", "metadata", "indexer_configuration_id"], cur, ) count = db.content_metadata_add_from_temp(conflict_update, cur) return { "content_metadata:add": count, } @timed @db_transaction_generator() def revision_intrinsic_metadata_missing(self, metadata, db=None, cur=None): for obj in db.revision_intrinsic_metadata_missing_from_list(metadata, cur): yield obj[0] @timed @db_transaction_generator() def revision_intrinsic_metadata_get(self, ids, db=None, cur=None): for c in db.revision_intrinsic_metadata_get_from_list(ids, cur): yield converters.db_to_metadata( dict(zip(db.revision_intrinsic_metadata_cols, c)) ) @timed @process_metrics @db_transaction() def revision_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: check_id_duplicates(map(RevisionIntrinsicMetadataRow.from_dict, metadata)) metadata.sort(key=lambda m: m["id"]) db.mktemp_revision_intrinsic_metadata(cur) db.copy_to( metadata, "tmp_revision_intrinsic_metadata", ["id", "metadata", "mappings", "indexer_configuration_id"], cur, ) count = db.revision_intrinsic_metadata_add_from_temp(conflict_update, cur) return { "revision_intrinsic_metadata:add": count, } @timed @process_metrics @db_transaction() def revision_intrinsic_metadata_delete( self, entries: List[Dict], db=None, cur=None ) -> Dict: count = db.revision_intrinsic_metadata_delete(entries, cur) return {"revision_intrinsic_metadata:del": count} @timed @db_transaction_generator() def origin_intrinsic_metadata_get(self, ids, db=None, cur=None): for c in db.origin_intrinsic_metadata_get_from_list(ids, cur): yield converters.db_to_metadata( dict(zip(db.origin_intrinsic_metadata_cols, c)) ) @timed @process_metrics @db_transaction() def origin_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False, db=None, cur=None ) -> Dict[str, int]: check_id_duplicates(map(OriginIntrinsicMetadataRow.from_dict, metadata)) metadata.sort(key=lambda m: m["id"]) db.mktemp_origin_intrinsic_metadata(cur) db.copy_to( metadata, "tmp_origin_intrinsic_metadata", ["id", "metadata", "indexer_configuration_id", "from_revision", "mappings"], cur, ) count = db.origin_intrinsic_metadata_add_from_temp(conflict_update, cur) return { "origin_intrinsic_metadata:add": count, } @timed @process_metrics @db_transaction() def origin_intrinsic_metadata_delete( self, entries: List[Dict], db=None, cur=None ) -> Dict: count = db.origin_intrinsic_metadata_delete(entries, cur) return { "origin_intrinsic_metadata:del": count, } @timed @db_transaction_generator() def origin_intrinsic_metadata_search_fulltext( self, conjunction, limit=100, db=None, cur=None ): for c in db.origin_intrinsic_metadata_search_fulltext( conjunction, limit=limit, cur=cur ): yield converters.db_to_metadata( dict(zip(db.origin_intrinsic_metadata_cols, c)) ) @timed @db_transaction() def origin_intrinsic_metadata_search_by_producer( self, page_token="", limit=100, ids_only=False, mappings=None, tool_ids=None, db=None, cur=None, ): assert isinstance(page_token, str) # we go to limit+1 to check whether we should add next_page_token in # the response res = db.origin_intrinsic_metadata_search_by_producer( page_token, limit + 1, ids_only, mappings, tool_ids, cur ) result = {} if ids_only: result["origins"] = [origin for (origin,) in res] if len(result["origins"]) > limit: result["origins"][limit:] = [] result["next_page_token"] = result["origins"][-1] else: result["origins"] = [ converters.db_to_metadata( dict(zip(db.origin_intrinsic_metadata_cols, c)) ) for c in res ] if len(result["origins"]) > limit: result["origins"][limit:] = [] result["next_page_token"] = result["origins"][-1]["id"] return result @timed @db_transaction() def origin_intrinsic_metadata_stats(self, db=None, cur=None): mapping_names = [m for m in MAPPING_NAMES] select_parts = [] # Count rows for each mapping for mapping_name in mapping_names: select_parts.append( ( "sum(case when (mappings @> ARRAY['%s']) " " then 1 else 0 end)" ) % mapping_name ) # Total select_parts.append("sum(1)") # Rows whose metadata has at least one key that is not '@context' select_parts.append( "sum(case when ('{}'::jsonb @> (metadata - '@context')) " " then 0 else 1 end)" ) cur.execute( "select " + ", ".join(select_parts) + " from origin_intrinsic_metadata" ) results = dict(zip(mapping_names + ["total", "non_empty"], cur.fetchone())) return { "total": results.pop("total"), "non_empty": results.pop("non_empty"), "per_mapping": results, } @timed @db_transaction_generator() def indexer_configuration_add(self, tools, db=None, cur=None): db.mktemp_indexer_configuration(cur) db.copy_to( tools, "tmp_indexer_configuration", ["tool_name", "tool_version", "tool_configuration"], cur, ) tools = db.indexer_configuration_add_from_temp(cur) count = 0 for line in tools: yield dict(zip(db.indexer_configuration_cols, line)) count += 1 send_metric( "indexer_configuration:add", count, method_name="indexer_configuration_add" ) @timed @db_transaction() def indexer_configuration_get(self, tool, db=None, cur=None): tool_conf = tool["tool_configuration"] if isinstance(tool_conf, dict): tool_conf = json.dumps(tool_conf) idx = db.indexer_configuration_get( tool["tool_name"], tool["tool_version"], tool_conf ) if not idx: return None return dict(zip(db.indexer_configuration_cols, idx)) diff --git a/swh/indexer/storage/in_memory.py b/swh/indexer/storage/in_memory.py index 245ab24..29dbdf9 100644 --- a/swh/indexer/storage/in_memory.py +++ b/swh/indexer/storage/in_memory.py @@ -1,539 +1,539 @@ # Copyright (C) 2018-2020 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, defaultdict import itertools import json import math import operator import re from typing import ( Any, Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, Type, TypeVar, ) from swh.core.collections import SortedList from swh.model.hashutil import hash_to_bytes, hash_to_hex from swh.model.model import SHA1_SIZE, Sha1Git from swh.storage.utils import get_partition_bounds_bytes from . import MAPPING_NAMES, check_id_duplicates, converters from .exc import IndexerStorageArgumentException from .interface import PagedResult, Sha1 from .model import ( BaseRow, ContentCtagsRow, ContentLanguageRow, ContentLicenseRow, ContentMetadataRow, ContentMimetypeRow, OriginIntrinsicMetadataRow, RevisionIntrinsicMetadataRow, ) SHA1_DIGEST_SIZE = 160 def _transform_tool(tool): return { "id": tool["id"], "name": tool["tool_name"], "version": tool["tool_version"], "configuration": tool["tool_configuration"], } def check_id_types(data: List[Dict[str, Any]]): """Checks all elements of the list have an 'id' whose type is 'bytes'.""" if not all(isinstance(item.get("id"), bytes) for item in data): raise IndexerStorageArgumentException("identifiers must be bytes.") def _key_from_dict(d): return tuple(sorted(d.items())) ToolId = int TValue = TypeVar("TValue", bound=BaseRow) class SubStorage(Generic[TValue]): """Implements common missing/get/add logic for each indexer type.""" _data: Dict[Sha1, Dict[Tuple, Dict[str, Any]]] _tools_per_id: Dict[Sha1, Set[ToolId]] def __init__(self, row_class: Type[TValue], tools): self.row_class = row_class self._tools = tools self._sorted_ids = SortedList[bytes, Sha1]() self._data = defaultdict(dict) self._tools_per_id = defaultdict(set) def _key_from_dict(self, d) -> Tuple: """Like the global _key_from_dict, but filters out dict keys that don't belong in the unique key.""" return _key_from_dict({k: d[k] for k in self.row_class.UNIQUE_KEY_FIELDS}) def missing(self, keys: Iterable[Dict]) -> Iterator[Sha1]: """List data missing from storage. Args: data (iterable): dictionaries with keys: - **id** (bytes): sha1 identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: missing sha1s """ for key in keys: tool_id = key["indexer_configuration_id"] id_ = key["id"] if tool_id not in self._tools_per_id.get(id_, set()): yield id_ def get(self, ids: Iterable[Sha1]) -> Iterator[TValue]: """Retrieve data per id. Args: ids (iterable): sha1 checksums Yields: dict: dictionaries with the following keys: - **id** (bytes) - **tool** (dict): tool used to compute metadata - arbitrary data (as provided to `add`) """ for id_ in ids: for entry in self._data[id_].values(): entry = entry.copy() tool_id = entry.pop("indexer_configuration_id") yield self.row_class( id=id_, tool=_transform_tool(self._tools[tool_id]), **entry, ) def get_all(self) -> Iterator[TValue]: yield from self.get(self._sorted_ids) def get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Sha1]: """Retrieve ids of content with `indexer_type` within partition partition_id bound by limit. Args: **indexer_type**: Type of data content to index (mimetype, language, etc...) **indexer_configuration_id**: The tool used to index data **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) **with_textual_data** (bool): Deal with only textual content (True) or all content (all contents by defaults, False) Raises: IndexerStorageArgumentException for; - limit to None - wrong indexer_type provided Returns: PagedResult of Sha1. If next_page_token is None, there is no more data to fetch """ if limit is None: raise IndexerStorageArgumentException("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 ids: List[Sha1] = [] sha1s = (sha1 for sha1 in self._sorted_ids.iter_from(start)) for counter, sha1 in enumerate(sha1s): if sha1 > end: break if counter >= limit: next_page_token = hash_to_hex(sha1) break ids.append(sha1) assert len(ids) <= limit return PagedResult(results=ids, next_page_token=next_page_token) def add(self, data: Iterable[TValue], conflict_update: bool) -> int: """Add data not present in storage. Args: data (iterable): dictionaries with keys: - **id**: sha1 - **indexer_configuration_id**: tool used to compute the results - arbitrary data conflict_update (bool): Flag to determine if we want to overwrite (true) or skip duplicates (false) """ data = list(data) check_id_duplicates(data) count = 0 for obj in data: item = obj.to_dict() id_ = item.pop("id") tool_id = item["indexer_configuration_id"] key = _key_from_dict(obj.unique_key()) if not conflict_update and key in self._data[id_]: # Duplicate, should not be updated continue self._data[id_][key] = item self._tools_per_id[id_].add(tool_id) count += 1 if id_ not in self._sorted_ids: self._sorted_ids.add(id_) return count def delete(self, entries: List[Dict]) -> int: """Delete entries and return the number of entries deleted. """ deleted = 0 for entry in entries: (id_, tool_id) = (entry["id"], entry["indexer_configuration_id"]) if tool_id in self._tools_per_id[id_]: self._tools_per_id[id_].remove(tool_id) if id_ in self._data: key = self._key_from_dict(entry) if key in self._data[id_]: deleted += 1 del self._data[id_][key] return deleted class IndexerStorage: """In-memory SWH indexer storage.""" def __init__(self): self._tools = {} self._mimetypes = SubStorage(ContentMimetypeRow, self._tools) self._languages = SubStorage(ContentLanguageRow, self._tools) self._content_ctags = SubStorage(ContentCtagsRow, self._tools) self._licenses = SubStorage(ContentLicenseRow, self._tools) self._content_metadata = SubStorage(ContentMetadataRow, self._tools) self._revision_intrinsic_metadata = SubStorage( RevisionIntrinsicMetadataRow, self._tools ) self._origin_intrinsic_metadata = SubStorage( OriginIntrinsicMetadataRow, self._tools ) def check_config(self, *, check_write): return True def content_mimetype_missing( self, mimetypes: Iterable[Dict] - ) -> Iterator[Tuple[Sha1, int]]: + ) -> Iterable[Tuple[Sha1, int]]: yield from self._mimetypes.missing(mimetypes) def content_mimetype_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Sha1]: return self._mimetypes.get_partition( indexer_configuration_id, partition_id, nb_partitions, page_token, limit ) def content_mimetype_add( self, mimetypes: List[ContentMimetypeRow], conflict_update: bool = False ) -> Dict[str, int]: added = self._mimetypes.add(mimetypes, conflict_update) return {"content_mimetype:add": added} - def content_mimetype_get(self, ids: Iterable[Sha1]) -> Iterator[ContentMimetypeRow]: + def content_mimetype_get(self, ids: Iterable[Sha1]) -> Iterable[ContentMimetypeRow]: yield from self._mimetypes.get(ids) def content_language_missing(self, languages): yield from self._languages.missing(languages) def content_language_get(self, ids): yield from (obj.to_dict() for obj in self._languages.get(ids)) def content_language_add( self, languages: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: check_id_types(languages) added = self._languages.add( map(ContentLanguageRow.from_dict, languages), conflict_update ) return {"content_language:add": added} def content_ctags_missing(self, ctags): yield from self._content_ctags.missing(ctags) def content_ctags_get(self, ids): for item in self._content_ctags.get(ids): yield {"id": item.id, "tool": item.tool, **item.to_dict()} def content_ctags_add( self, ctags: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: check_id_types(ctags) added = self._content_ctags.add( map( ContentCtagsRow.from_dict, itertools.chain.from_iterable(map(converters.ctags_to_db, ctags)), ), conflict_update, ) return {"content_ctags:add": added} def content_ctags_search(self, expression, limit=10, last_sha1=None): nb_matches = 0 items_per_id: Dict[Tuple[Sha1Git, ToolId], List[ContentCtagsRow]] = {} for item in sorted(self._content_ctags.get_all()): if item.id <= (last_sha1 or bytes(0 for _ in range(SHA1_DIGEST_SIZE))): continue items_per_id.setdefault( (item.id, item.indexer_configuration_id), [] ).append(item) for items in items_per_id.values(): ctags = [] for item in items: if item.name != expression: continue nb_matches += 1 if nb_matches > limit: break item_dict = item.to_dict() id_ = item_dict.pop("id") tool = item_dict.pop("tool") ctags.append(item_dict) if ctags: for ctag in ctags: yield {"id": id_, "tool": tool, **ctag} def content_fossology_license_get(self, ids): # Rewrites the output of SubStorage.get from the old format to # the new one. SubStorage.get should be updated once all other # *_get methods use the new format. # See: https://forge.softwareheritage.org/T1433 for id_ in ids: items = {} for obj in self._licenses.get([id_]): items.setdefault(obj.tool["id"], (obj.tool, []))[1].append(obj.license) if items: yield { id_: [ {"tool": tool, "licenses": licenses} for (tool, licenses) in items.values() ] } def content_fossology_license_add( self, licenses: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: check_id_types(licenses) added = self._licenses.add( map( ContentLicenseRow.from_dict, itertools.chain.from_iterable( map(converters.fossology_license_to_db, licenses) ), ), conflict_update, ) return {"content_fossology_license:add": added} def content_fossology_license_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Sha1]: return self._licenses.get_partition( indexer_configuration_id, partition_id, nb_partitions, page_token, limit ) def content_metadata_missing(self, metadata): yield from self._content_metadata.missing(metadata) def content_metadata_get(self, ids): yield from (obj.to_dict() for obj in self._content_metadata.get(ids)) def content_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: check_id_types(metadata) added = self._content_metadata.add( map(ContentMetadataRow.from_dict, metadata), conflict_update ) return {"content_metadata:add": added} def revision_intrinsic_metadata_missing(self, metadata): yield from self._revision_intrinsic_metadata.missing(metadata) def revision_intrinsic_metadata_get(self, ids): yield from (obj.to_dict() for obj in self._revision_intrinsic_metadata.get(ids)) def revision_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: check_id_types(metadata) added = self._revision_intrinsic_metadata.add( map(RevisionIntrinsicMetadataRow.from_dict, metadata), conflict_update ) return {"revision_intrinsic_metadata:add": added} def revision_intrinsic_metadata_delete(self, entries: List[Dict]) -> Dict: deleted = self._revision_intrinsic_metadata.delete(entries) return {"revision_intrinsic_metadata:del": deleted} def origin_intrinsic_metadata_get(self, ids): yield from (obj.to_dict() for obj in self._origin_intrinsic_metadata.get(ids)) def origin_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: added = self._origin_intrinsic_metadata.add( map(OriginIntrinsicMetadataRow.from_dict, metadata), conflict_update ) return {"origin_intrinsic_metadata:add": added} def origin_intrinsic_metadata_delete(self, entries: List[Dict]) -> Dict: deleted = self._origin_intrinsic_metadata.delete(entries) return {"origin_intrinsic_metadata:del": deleted} def origin_intrinsic_metadata_search_fulltext(self, conjunction, limit=100): # A very crude fulltext search implementation, but that's enough # to work on English metadata tokens_re = re.compile("[a-zA-Z0-9]+") search_tokens = list(itertools.chain(*map(tokens_re.findall, conjunction))) def rank(data): # Tokenize the metadata text = json.dumps(data.metadata) text_tokens = tokens_re.findall(text) text_token_occurences = Counter(text_tokens) # Count the number of occurrences of search tokens in the text score = 0 for search_token in search_tokens: if text_token_occurences[search_token] == 0: # Search token is not in the text. return 0 score += text_token_occurences[search_token] # Normalize according to the text's length return score / math.log(len(text_tokens)) results = [ (rank(data), data) for data in self._origin_intrinsic_metadata.get_all() ] results = [(rank_, data) for (rank_, data) in results if rank_ > 0] results.sort( key=operator.itemgetter(0), reverse=True # Don't try to order 'data' ) for (rank_, result) in results[:limit]: yield result.to_dict() def origin_intrinsic_metadata_search_by_producer( self, page_token="", limit=100, ids_only=False, mappings=None, tool_ids=None ): assert isinstance(page_token, str) nb_results = 0 if mappings is not None: mappings = frozenset(mappings) if tool_ids is not None: tool_ids = frozenset(tool_ids) origins = [] # we go to limit+1 to check whether we should add next_page_token in # the response for entry in self._origin_intrinsic_metadata.get_all(): if entry.id <= page_token: continue if nb_results >= (limit + 1): break if mappings is not None and mappings.isdisjoint(entry.mappings): continue if tool_ids is not None and entry.tool["id"] not in tool_ids: continue origins.append(entry.to_dict()) nb_results += 1 result = {} if len(origins) > limit: origins = origins[:limit] result["next_page_token"] = origins[-1]["id"] if ids_only: origins = [origin["id"] for origin in origins] result["origins"] = origins return result def origin_intrinsic_metadata_stats(self): mapping_count = {m: 0 for m in MAPPING_NAMES} total = non_empty = 0 for data in self._origin_intrinsic_metadata.get_all(): total += 1 if set(data.metadata) - {"@context"}: non_empty += 1 for mapping in data.mappings: mapping_count[mapping] += 1 return {"per_mapping": mapping_count, "total": total, "non_empty": non_empty} def indexer_configuration_add(self, tools): inserted = [] for tool in tools: tool = tool.copy() id_ = self._tool_key(tool) tool["id"] = id_ self._tools[id_] = tool inserted.append(tool) return inserted def indexer_configuration_get(self, tool): return self._tools.get(self._tool_key(tool)) def _tool_key(self, tool): return hash( ( tool["tool_name"], tool["tool_version"], json.dumps(tool["tool_configuration"], sort_keys=True), ) ) diff --git a/swh/indexer/storage/interface.py b/swh/indexer/storage/interface.py index 87f1399..a05d460 100644 --- a/swh/indexer/storage/interface.py +++ b/swh/indexer/storage/interface.py @@ -1,613 +1,613 @@ # 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 -from typing import Dict, Iterable, Iterator, List, Optional, Tuple, TypeVar +from typing import Dict, Iterable, List, Optional, Tuple, TypeVar from swh.core.api import remote_api_endpoint from swh.core.api.classes import PagedResult as CorePagedResult from swh.indexer.storage.model import ContentMimetypeRow TResult = TypeVar("TResult") PagedResult = CorePagedResult[TResult, str] Sha1 = bytes class IndexerStorageInterface: @remote_api_endpoint("check_config") def check_config(self, *, check_write): """Check that the storage is configured and ready to go.""" ... @remote_api_endpoint("content_mimetype/missing") def content_mimetype_missing( self, mimetypes: Iterable[Dict] - ) -> Iterator[Tuple[Sha1, int]]: + ) -> Iterable[Tuple[Sha1, int]]: """Generate mimetypes missing from storage. Args: mimetypes (iterable): iterable of dict with keys: - **id** (bytes): sha1 identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: tuple (id, indexer_configuration_id): missing id """ ... @remote_api_endpoint("content_mimetype/range") def content_mimetype_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Sha1]: """Retrieve mimetypes within partition partition_id bound by limit. Args: **indexer_configuration_id**: The tool used to index data **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) Raises: IndexerStorageArgumentException for; - limit to None - wrong indexer_type provided Returns: PagedResult of Sha1. If next_page_token is None, there is no more data to fetch """ ... @remote_api_endpoint("content_mimetype/add") def content_mimetype_add( self, mimetypes: List[ContentMimetypeRow], conflict_update: bool = False ) -> Dict[str, int]: """Add mimetypes not present in storage. Args: mimetypes: mimetype rows to be added conflict_update: Flag to determine if we want to overwrite (``True``) or skip duplicates (``False``, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("content_mimetype") - def content_mimetype_get(self, ids: Iterable[Sha1]) -> Iterator[ContentMimetypeRow]: + def content_mimetype_get(self, ids: Iterable[Sha1]) -> Iterable[ContentMimetypeRow]: """Retrieve full content mimetype per ids. Args: ids (iterable): sha1 identifier Yields: mimetypes (iterable): dictionaries with keys: - **id** (bytes): sha1 identifier - **mimetype** (bytes): raw content's mimetype - **encoding** (bytes): raw content's encoding - **tool** (dict): Tool used to compute the language """ ... @remote_api_endpoint("content_language/missing") def content_language_missing(self, languages): """List languages missing from storage. Args: languages (iterable): dictionaries with keys: - **id** (bytes): sha1 identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: an iterable of missing id for the tuple (id, indexer_configuration_id) """ ... @remote_api_endpoint("content_language") def content_language_get(self, ids): """Retrieve full content language per ids. Args: ids (iterable): sha1 identifier Yields: languages (iterable): dictionaries with keys: - **id** (bytes): sha1 identifier - **lang** (bytes): raw content's language - **tool** (dict): Tool used to compute the language """ ... @remote_api_endpoint("content_language/add") def content_language_add( self, languages: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add languages not present in storage. Args: languages (iterable): dictionaries with keys: - **id** (bytes): sha1 - **lang** (bytes): language detected conflict_update (bool): Flag to determine if we want to overwrite (true) or skip duplicates (false, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("content/ctags/missing") def content_ctags_missing(self, ctags): """List ctags missing from storage. Args: ctags (iterable): dicts with keys: - **id** (bytes): sha1 identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: an iterable of missing id for the tuple (id, indexer_configuration_id) """ ... @remote_api_endpoint("content/ctags") def content_ctags_get(self, ids): """Retrieve ctags per id. Args: ids (iterable): sha1 checksums Yields: Dictionaries with keys: - **id** (bytes): content's identifier - **name** (str): symbol's name - **kind** (str): symbol's kind - **lang** (str): language for that content - **tool** (dict): tool used to compute the ctags' info """ ... @remote_api_endpoint("content/ctags/add") def content_ctags_add( self, ctags: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add ctags not present in storage Args: ctags (iterable): dictionaries with keys: - **id** (bytes): sha1 - **ctags** ([list): List of dictionary with keys: name, kind, line, lang Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("content/ctags/search") def content_ctags_search(self, expression, limit=10, last_sha1=None): """Search through content's raw ctags symbols. Args: expression (str): Expression to search for limit (int): Number of rows to return (default to 10). last_sha1 (str): Offset from which retrieving data (default to ''). Yields: rows of ctags including id, name, lang, kind, line, etc... """ ... @remote_api_endpoint("content/fossology_license") def content_fossology_license_get(self, ids): """Retrieve licenses per id. Args: ids (iterable): sha1 checksums Yields: dict: ``{id: facts}`` where ``facts`` is a dict with the following keys: - **licenses** ([str]): associated licenses for that content - **tool** (dict): Tool used to compute the license """ ... @remote_api_endpoint("content/fossology_license/add") def content_fossology_license_add( self, licenses: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add licenses not present in storage. Args: licenses (iterable): dictionaries with keys: - **id**: sha1 - **licenses** ([bytes]): List of licenses associated to sha1 - **tool** (str): nomossa conflict_update: Flag to determine if we want to overwrite (true) or skip duplicates (false, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("content/fossology_license/range") def content_fossology_license_get_partition( self, indexer_configuration_id: int, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Sha1]: """Retrieve licenses within the partition partition_id bound by limit. Args: **indexer_configuration_id**: The tool used to index data **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) Raises: IndexerStorageArgumentException for; - limit to None - wrong indexer_type provided Returns: PagedResult of Sha1. If next_page_token is None, there is no more data to fetch """ ... @remote_api_endpoint("content_metadata/missing") def content_metadata_missing(self, metadata): """List metadata missing from storage. Args: metadata (iterable): dictionaries with keys: - **id** (bytes): sha1 identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: missing sha1s """ ... @remote_api_endpoint("content_metadata") def content_metadata_get(self, ids): """Retrieve metadata per id. Args: ids (iterable): sha1 checksums Yields: dictionaries with the following keys: id (bytes) metadata (str): associated metadata tool (dict): tool used to compute metadata """ ... @remote_api_endpoint("content_metadata/add") def content_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add metadata not present in storage. Args: metadata (iterable): dictionaries with keys: - **id**: sha1 - **metadata**: arbitrary dict conflict_update: Flag to determine if we want to overwrite (true) or skip duplicates (false, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("revision_intrinsic_metadata/missing") def revision_intrinsic_metadata_missing(self, metadata): """List metadata missing from storage. Args: metadata (iterable): dictionaries with keys: - **id** (bytes): sha1_git revision identifier - **indexer_configuration_id** (int): tool used to compute the results Yields: missing ids """ ... @remote_api_endpoint("revision_intrinsic_metadata") def revision_intrinsic_metadata_get(self, ids): """Retrieve revision metadata per id. Args: ids (iterable): sha1 checksums Yields: : dictionaries with the following keys: - **id** (bytes) - **metadata** (str): associated metadata - **tool** (dict): tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata """ ... @remote_api_endpoint("revision_intrinsic_metadata/add") def revision_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add metadata not present in storage. Args: metadata (iterable): dictionaries with keys: - **id**: sha1_git of revision - **metadata**: arbitrary dict - **indexer_configuration_id**: tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata conflict_update: Flag to determine if we want to overwrite (true) or skip duplicates (false, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("revision_intrinsic_metadata/delete") def revision_intrinsic_metadata_delete(self, entries: List[Dict]) -> Dict: """Remove revision metadata from the storage. Args: entries (dict): dictionaries with the following keys: - **id** (bytes): revision identifier - **indexer_configuration_id** (int): tool used to compute metadata Returns: Summary of number of rows deleted """ ... @remote_api_endpoint("origin_intrinsic_metadata") def origin_intrinsic_metadata_get(self, ids): """Retrieve origin metadata per id. Args: ids (iterable): origin identifiers Yields: list: dictionaries with the following keys: - **id** (str): origin url - **from_revision** (bytes): which revision this metadata was extracted from - **metadata** (str): associated metadata - **tool** (dict): tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata """ ... @remote_api_endpoint("origin_intrinsic_metadata/add") def origin_intrinsic_metadata_add( self, metadata: List[Dict], conflict_update: bool = False ) -> Dict[str, int]: """Add origin metadata not present in storage. Args: metadata (iterable): dictionaries with keys: - **id**: origin urls - **from_revision**: sha1 id of the revision used to generate these metadata. - **metadata**: arbitrary dict - **indexer_configuration_id**: tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata conflict_update: Flag to determine if we want to overwrite (true) or skip duplicates (false, the default) Returns: Dict summary of number of rows added """ ... @remote_api_endpoint("origin_intrinsic_metadata/delete") def origin_intrinsic_metadata_delete(self, entries: List[Dict]) -> Dict: """Remove origin metadata from the storage. Args: entries (dict): dictionaries with the following keys: - **id** (str): origin urls - **indexer_configuration_id** (int): tool used to compute metadata Returns: Summary of number of rows deleted """ ... @remote_api_endpoint("origin_intrinsic_metadata/search/fulltext") def origin_intrinsic_metadata_search_fulltext(self, conjunction, limit=100): """Returns the list of origins whose metadata contain all the terms. Args: conjunction (List[str]): List of terms to be searched for. limit (int): The maximum number of results to return Yields: list: dictionaries with the following keys: - **id** (str): origin urls - **from_revision**: sha1 id of the revision used to generate these metadata. - **metadata** (str): associated metadata - **tool** (dict): tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata """ ... @remote_api_endpoint("origin_intrinsic_metadata/search/by_producer") def origin_intrinsic_metadata_search_by_producer( self, page_token="", limit=100, ids_only=False, mappings=None, tool_ids=None ): """Returns the list of origins whose metadata contain all the terms. Args: page_token (str): Opaque token used for pagination. limit (int): The maximum number of results to return ids_only (bool): Determines whether only origin urls are returned or the content as well mappings (List[str]): Returns origins whose intrinsic metadata were generated using at least one of these mappings. Returns: dict: dict with the following keys: - **next_page_token** (str, optional): opaque token to be used as `page_token` for retrieving the next page. If absent, there is no more pages to gather. - **origins** (list): list of origin url (str) if `ids_only=True` else dictionaries with the following keys: - **id** (str): origin urls - **from_revision**: sha1 id of the revision used to generate these metadata. - **metadata** (str): associated metadata - **tool** (dict): tool used to compute metadata - **mappings** (List[str]): list of mappings used to translate these metadata """ ... @remote_api_endpoint("origin_intrinsic_metadata/stats") def origin_intrinsic_metadata_stats(self): """Returns counts of indexed metadata per origins, broken down into metadata types. Returns: dict: dictionary with keys: - total (int): total number of origins that were indexed (possibly yielding an empty metadata dictionary) - non_empty (int): total number of origins that we extracted a non-empty metadata dictionary from - per_mapping (dict): a dictionary with mapping names as keys and number of origins whose indexing used this mapping. Note that indexing a given origin may use 0, 1, or many mappings. """ ... @remote_api_endpoint("indexer_configuration/add") def indexer_configuration_add(self, tools): """Add new tools to the storage. Args: tools ([dict]): List of dictionary representing tool to insert in the db. Dictionary with the following keys: - **tool_name** (str): tool's name - **tool_version** (str): tool's version - **tool_configuration** (dict): tool's configuration (free form dict) Returns: List of dict inserted in the db (holding the id key as well). The order of the list is not guaranteed to match the order of the initial list. """ ... @remote_api_endpoint("indexer_configuration/data") def indexer_configuration_get(self, tool): """Retrieve tool information. Args: tool (dict): Dictionary representing a tool with the following keys: - **tool_name** (str): tool's name - **tool_version** (str): tool's version - **tool_configuration** (dict): tool's configuration (free form dict) Returns: The same dictionary with an `id` key, None otherwise. """ ...