diff --git a/mypy.ini b/mypy.ini --- a/mypy.ini +++ b/mypy.ini @@ -20,5 +20,8 @@ [mypy-pytest.*] ignore_missing_imports = True +[mypy-pytest_postgresql.*] +ignore_missing_imports = True + [mypy-psycopg2.*] ignore_missing_imports = True diff --git a/swh/provenance/tests/conftest.py b/swh/provenance/tests/conftest.py --- a/swh/provenance/tests/conftest.py +++ b/swh/provenance/tests/conftest.py @@ -4,12 +4,14 @@ # See top-level LICENSE file for more information from os import path +import tempfile from typing import Any, Dict, Iterable, Iterator from _pytest.fixtures import SubRequest import msgpack import psycopg2.extensions import pytest +from pytest_postgresql import factories from swh.journal.serializers import msgpack_ext_hook from swh.provenance import get_provenance, get_provenance_storage @@ -49,23 +51,42 @@ @pytest.fixture(params=["local", "remote"]) -def provenance( +def storage( request: SubRequest, populated_db: Dict[str, str], swh_rpc_client: RemoteProvenanceStorage, -) -> ProvenanceInterface: - """Return a working and initialized ProvenanceInterface object""" +) -> ProvenanceStorageInterface: + """Return a working and initialized ProvenanceStorageInterface object""" if request.param == "remote": - from swh.provenance.provenance import Provenance - assert isinstance(swh_rpc_client, ProvenanceStorageInterface) - return Provenance(swh_rpc_client) + return swh_rpc_client else: # in test sessions, we DO want to raise any exception occurring at commit time - prov = get_provenance(cls=request.param, db=populated_db, raise_on_commit=True) - return prov + storage = get_provenance_storage( + cls=request.param, db=populated_db, raise_on_commit=True + ) + return storage + + +# using the factory to create a postgresql instance +socket_dir = tempfile.TemporaryDirectory() +postgresql2_proc = factories.postgresql_proc(port=None, unixsocketdir=socket_dir.name) +postgresql2 = factories.postgresql("postgresql2_proc") + + +@pytest.fixture +def provenance(postgresql2: psycopg2.extensions.connection) -> ProvenanceInterface: + """Return a working and initialized ProvenanceInterface object""" + + from swh.core.cli.db import populate_database_for_package + + populate_database_for_package("swh.provenance", postgresql2.dsn, flavor="with-path") + # in test sessions, we DO want to raise any exception occurring at commit time + return get_provenance( + cls="local", db=postgresql2.get_dsn_parameters(), raise_on_commit=True + ) @pytest.fixture diff --git a/swh/provenance/tests/test_provenance_storage.py b/swh/provenance/tests/test_provenance_storage.py --- a/swh/provenance/tests/test_provenance_storage.py +++ b/swh/provenance/tests/test_provenance_storage.py @@ -3,9 +3,307 @@ # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information +from datetime import datetime import inspect +import os +from typing import Any, Dict, Iterable, Optional, Set -from ..interface import ProvenanceInterface, ProvenanceStorageInterface +import pytest + +from swh.model.hashutil import hash_to_bytes +from swh.model.identifiers import origin_identifier +from swh.model.model import Sha1Git +from swh.provenance.interface import ( + EntityType, + ProvenanceInterface, + ProvenanceResult, + ProvenanceStorageInterface, + RelationData, + RelationType, +) +from swh.provenance.tests.conftest import load_repo_data +from swh.provenance.tests.test_provenance_db import ts2dt + + +def relation_add_and_compare_result( + relation: RelationType, + data: Set[RelationData], + refstorage: ProvenanceStorageInterface, + storage: ProvenanceStorageInterface, + with_path: bool = True, +) -> None: + assert data + assert refstorage.relation_add(relation, data) == storage.relation_add( + relation, data + ) + + assert relation_compare_result( + refstorage.relation_get(relation, (reldata.src for reldata in data)), + storage.relation_get(relation, (reldata.src for reldata in data)), + with_path, + ) + assert relation_compare_result( + refstorage.relation_get( + relation, + (reldata.dst for reldata in data), + reverse=True, + ), + storage.relation_get( + relation, + (reldata.dst for reldata in data), + reverse=True, + ), + with_path, + ) + assert relation_compare_result( + refstorage.relation_get_all(relation), + storage.relation_get_all(relation), + with_path, + ) + + +def relation_compare_result( + expected: Set[RelationData], computed: Set[RelationData], with_path: bool +) -> bool: + return { + RelationData(reldata.src, reldata.dst, reldata.path if with_path else None) + for reldata in expected + } == computed + + +def dircontent( + data: Dict[str, Any], + ref: Sha1Git, + dir: Dict[str, Any], + prefix: bytes = b"", +) -> Iterable[RelationData]: + content = { + RelationData(entry["target"], ref, os.path.join(prefix, entry["name"])) + for entry in dir["entries"] + if entry["type"] == "file" + } + for entry in dir["entries"]: + if entry["type"] == "dir": + child = next( + subdir + for subdir in data["directory"] + if subdir["id"] == entry["target"] + ) + content.update( + dircontent(data, ref, child, os.path.join(prefix, entry["name"])) + ) + return content + + +@pytest.mark.parametrize( + "repo", + ("cmdbts2", "out-of-order", "with-merges"), +) +def test_provenance_storage( + provenance: ProvenanceInterface, storage: ProvenanceStorageInterface, repo: str +) -> None: + """Tests every ProvenanceStorageInterface implementation against the one provided + for provenance.storage.""" + # Read data/README.md for more details on how these datasets are generated. + data = load_repo_data(repo) + + # Assuming provenance.storage has the 'with-path' flavor. + assert provenance.storage.with_path() + + # Test content methods. + # Add all content present in the current repo to both storages, just assigning their + # creation dates. Then check that the inserted content is the same in both cases. + cnt_dates = {cnt["sha1_git"]: cnt["ctime"] for cnt in data["content"]} + assert cnt_dates + assert provenance.storage.content_set_date(cnt_dates) == storage.content_set_date( + cnt_dates + ) + + assert provenance.storage.content_get(cnt_dates) == storage.content_get(cnt_dates) + assert provenance.storage.entity_get_all( + EntityType.CONTENT + ) == storage.entity_get_all(EntityType.CONTENT) + + # Test directory methods. + # Of all directories present in the current repo, only assign a date to those + # containing blobs (picking the max date among the available ones). Then check that + # the inserted data is the same in both storages. + def getmaxdate( + dir: Dict[str, Any], cnt_dates: Dict[Sha1Git, datetime] + ) -> Optional[datetime]: + dates = [ + cnt_dates[entry["target"]] + for entry in dir["entries"] + if entry["type"] == "file" + ] + return max(dates) if dates else None + + dir_dates = {dir["id"]: getmaxdate(dir, cnt_dates) for dir in data["directory"]} + assert dir_dates + assert provenance.storage.directory_set_date( + {sha1: date for sha1, date in dir_dates.items() if date is not None} + ) == storage.directory_set_date( + {sha1: date for sha1, date in dir_dates.items() if date is not None} + ) + + assert provenance.storage.directory_get(dir_dates) == storage.directory_get( + dir_dates + ) + assert provenance.storage.entity_get_all( + EntityType.DIRECTORY + ) == storage.entity_get_all(EntityType.DIRECTORY) + + # Test origin methods. + # Add all origins present in the current repo to both storages. Then check that the + # inserted data is the same in both cases. + org_urls = { + hash_to_bytes(origin_identifier(org)): org["url"] for org in data["origin"] + } + assert org_urls + assert provenance.storage.origin_set_url(org_urls) == storage.origin_set_url( + org_urls + ) + + assert provenance.storage.origin_get(org_urls) == storage.origin_get(org_urls) + assert provenance.storage.entity_get_all( + EntityType.ORIGIN + ) == storage.entity_get_all(EntityType.ORIGIN) + + # Test revision methods. + # Add all revisions present in the current repo to both storages, assigning their + # dataes and an arbitrary origin to each one. Then check that the inserted data is + # the same in both cases. + rev_dates = {rev["id"]: ts2dt(rev["date"]) for rev in data["revision"]} + assert rev_dates + assert provenance.storage.revision_set_date(rev_dates) == storage.revision_set_date( + rev_dates + ) + + rev_origins = { + rev["id"]: next(iter(org_urls)) # any arbitrary origin will do + for rev in data["revision"] + } + assert rev_origins + assert provenance.storage.revision_set_origin( + rev_origins + ) == storage.revision_set_origin(rev_origins) + + assert provenance.storage.revision_get(rev_dates) == storage.revision_get(rev_dates) + assert provenance.storage.entity_get_all( + EntityType.REVISION + ) == storage.entity_get_all(EntityType.REVISION) + + # Test content-in-revision relation. + # Create flat models of every root directory for the revisions in the dataset. + cnt_in_rev: Set[RelationData] = set() + for rev in data["revision"]: + root = next( + subdir for subdir in data["directory"] if subdir["id"] == rev["directory"] + ) + cnt_in_rev.update(dircontent(data, rev["id"], root)) + + relation_add_and_compare_result( + RelationType.CNT_EARLY_IN_REV, + cnt_in_rev, + provenance.storage, + storage, + storage.with_path(), + ) + + # Test content-in-directory relation. + # Create flat models for every directory in the dataset. + cnt_in_dir: Set[RelationData] = set() + for dir in data["directory"]: + cnt_in_dir.update(dircontent(data, dir["id"], dir)) + + relation_add_and_compare_result( + RelationType.CNT_IN_DIR, + cnt_in_dir, + provenance.storage, + storage, + storage.with_path(), + ) + + # Test content-in-directory relation. + # Add root directories to their correspondent revision in the dataset. + dir_in_rev = { + RelationData(rev["directory"], rev["id"], b".") for rev in data["revision"] + } + + relation_add_and_compare_result( + RelationType.DIR_IN_REV, + dir_in_rev, + provenance.storage, + storage, + storage.with_path(), + ) + + # Test revision-in-origin relation. + # Add all revisions that are head of some snapshot branch to the corresponding + # origin. + rev_in_org = { + RelationData( + branch["target"], + hash_to_bytes(origin_identifier({"url": status["origin"]})), + None, + ) + for status in data["origin_visit_status"] + if status["snapshot"] is not None + for snapshot in data["snapshot"] + if snapshot["id"] == status["snapshot"] + for _, branch in snapshot["branches"].items() + if branch["target_type"] == "revision" + } + + relation_add_and_compare_result( + RelationType.REV_IN_ORG, + rev_in_org, + provenance.storage, + storage, + ) + + # Test revision-before-revision relation. + # For each revision in the data set add an entry for each parent to the relation. + rev_before_rev = { + RelationData(parent, rev["id"], None) + for rev in data["revision"] + for parent in rev["parents"] + } + + relation_add_and_compare_result( + RelationType.REV_BEFORE_REV, + rev_before_rev, + provenance.storage, + storage, + ) + + # Test location_get. + if storage.with_path(): + assert provenance.storage.location_get() == storage.location_get() + + # Test content_find_first and content_find_all. + def adapt_result( + result: Optional[ProvenanceResult], with_path: bool + ) -> Optional[ProvenanceResult]: + if result is not None: + return ProvenanceResult( + result.content, + result.revision, + result.date, + result.origin, + result.path if with_path else b"", + ) + return result + + for cnt in cnt_dates: + assert adapt_result( + provenance.storage.content_find_first(cnt), storage.with_path() + ) == storage.content_find_first(cnt) + + assert { + adapt_result(occur, storage.with_path()) + for occur in provenance.storage.content_find_all(cnt) + } == set(storage.content_find_all(cnt)) def test_types(provenance: ProvenanceInterface) -> None: