def test_multiple_filename_only_matches_to_one_accession(self):
session = Session()
accession = AccessionBuilder().build()
restore1 = create_perfect_match(accession)
restore1.md5 = 'filename_only_md5_1'
restore1.bytes = restore1.bytes + 100
restore2 = create_perfect_match(accession)
restore2.md5 = 'filename_only_md5_2'
restore2.bytes = restore2.bytes + 100
session.add(accession)
session.add(restore1)
session.add(restore2)
session.commit()
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
self.assertEqual(2, len(new_matches_found))
self.assertEqual(2, len(accession.filename_only_matches))
self.assertEqual(1, len(restore1.filename_only_matches))
self.assertEqual(accession, restore1.filename_only_matches[0])
self.assertEqual(1, len(restore2.filename_only_matches))
self.assertEqual(accession, restore2.filename_only_matches[0])
def test_accession_with_perfect_match_and_altered_md5_match_and_filename_only_match(
self):
session = Session()
accession = AccessionBuilder().build()
perfect_restore = create_perfect_match(accession)
altered_restore = create_perfect_match(accession)
altered_restore.md5 = 'altered_md5'
filename_only_restore = create_perfect_match(accession)
filename_only_restore.md5 = 'filename_only_md5'
filename_only_restore.bytes = filename_only_restore.bytes + 100
session.add(accession)
session.add(perfect_restore)
session.add(altered_restore)
session.add(filename_only_restore)
session.commit()
accessions = session.query(Accession)
perfect_matches_found = find_perfect_matches(session, accessions)
altered_md5_matches_found = find_altered_md5_matches(
session, accessions)
filename_only_matches_found = find_filename_only_matches(
session, accessions)
self.assertEqual(1, len(perfect_matches_found))
self.assertEqual(1, len(altered_md5_matches_found))
self.assertEqual(1, len(filename_only_matches_found))
self.assertEqual(perfect_restore, accession.perfect_matches[0])
self.assertEqual(altered_restore, accession.altered_md5_matches[0])
self.assertEqual(filename_only_restore,
accession.filename_only_matches[0])
def test_deleting_accession_using_raw_sql_should_delete_filename_only_match(
self):
session = Session()
accession = AccessionBuilder().set_batch("batch_to_delete").build()
restore = create_perfect_match(accession)
restore.md5 = 'filename_only_md5'
restore.bytes = restore.bytes + 100
session.add(accession)
session.add(restore)
session.commit()
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
session.commit()
self.assertEqual(1, len(new_matches_found))
self.assertEqual(1, len(accession.filename_only_matches))
self.assertEqual(1, len(restore.filename_only_matches))
# Delete using raw SQL, not SQLAlchemy, to test ON DELETE CASCADE
session.execute(
"DELETE FROM accessions where accessions.batch = 'batch_to_delete'"
)
session.commit()
accessions_count = session.query(Accession).count()
restores_count = session.query(Restore).count()
matches_count = session.query(filename_only_matches_table).count()
self.assertEqual(0, accessions_count)
self.assertEqual(1, restores_count) # Restores are not affected
self.assertEqual(0, matches_count)
self.assertEqual([], restore.filename_only_matches)
def test_deleting_accession_using_orm_should_delete_filename_only_match(
self):
session = Session()
accession = AccessionBuilder().set_batch("batch_to_delete").build()
restore = create_perfect_match(accession)
restore.md5 = 'filename_only_md5'
restore.bytes = restore.bytes + 100
session.add(accession)
session.add(restore)
session.commit()
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
session.commit()
self.assertEqual(1, len(new_matches_found))
self.assertEqual(1, len(accession.filename_only_matches))
self.assertEqual(1, len(restore.filename_only_matches))
# Delete using SQLAlchemy ORM
session.delete(accession)
session.commit()
accessions_count = session.query(Accession).count()
restores_count = session.query(Restore).count()
matches_count = session.query(filename_only_matches_table).count()
self.assertEqual(0, accessions_count)
self.assertEqual(1, restores_count) # Restores are not affected
self.assertEqual(0, matches_count)
self.assertEqual([], restore.filename_only_matches)
def test_filename_only_match_does_not_include_altered_md_match(self):
# In this test, we are running find_filename_only_matches without first
# running find_altered_md5_matches, to ensure an altered MD match
# is not added to the filename only matches
session = Session()
accession = AccessionBuilder().build()
altered_restore = create_perfect_match(accession)
altered_restore.md5 = 'altered_md5'
filename_only_restore = create_perfect_match(accession)
filename_only_restore.md5 = 'filename_only_md5'
filename_only_restore.bytes = filename_only_restore.bytes + 100
session.add(accession)
session.add(altered_restore)
session.add(filename_only_restore)
session.commit()
accessions = session.query(Accession)
filename_only_matches_found = find_filename_only_matches(
session, accessions)
self.assertEqual(0, len(accession.altered_md5_matches))
self.assertEqual(1, len(filename_only_matches_found))
self.assertEqual(filename_only_restore,
accession.filename_only_matches[0])
def test_finding_filename_only_matches_more_than_once(self):
session = Session()
accession = AccessionBuilder().build()
restore = create_perfect_match(accession)
restore.md5 = 'filename_only_md5'
restore.bytes = restore.bytes + 100
session.add(accession)
session.add(restore)
session.commit()
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
self.assertEqual(1, len(new_matches_found))
self.assertEqual(1, len(accession.filename_only_matches))
self.assertEqual(1, len(restore.filename_only_matches))
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
self.assertEqual(0, len(new_matches_found))
self.assertEqual(1, len(accession.filename_only_matches))
self.assertEqual(1, len(restore.filename_only_matches))
def test_deleting_accession_using_raw_sql_should_delete_filename_only_match_not_affect_other_matches(
self):
session = Session()
accession1 = AccessionBuilder().set_batch("batch_to_delete").build()
restore1 = create_perfect_match(accession1)
restore1.md5 = 'filename_only_md5'
restore1.bytes = restore1.bytes + 100
accession2 = AccessionBuilder().set_batch("batch_to_preserve").build()
restore2 = create_perfect_match(accession2)
restore2.md5 = 'filename_only_md5'
restore2.bytes = restore2.bytes + 100
session.add(accession1)
session.add(restore1)
session.add(accession2)
session.add(restore2)
session.commit()
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
session.commit()
self.assertEqual(2, len(new_matches_found))
self.assertEqual(1, len(accession1.filename_only_matches))
self.assertEqual(1, len(restore1.filename_only_matches))
self.assertEqual(1, len(accession2.filename_only_matches))
self.assertEqual(1, len(restore2.filename_only_matches))
# Delete using raw SQL, not SQLAlchemy, to test ON DELETE CASCADE
connection = session.connection()
connection.execute(
"DELETE FROM accessions where accessions.batch = 'batch_to_delete'"
)
session.commit()
accessions_count = session.query(Accession).count()
restores_count = session.query(Restore).count()
matches_count = session.query(filename_only_matches_table).count()
self.assertEqual(1, accessions_count)
self.assertEqual(2, restores_count) # Restores are no affected
self.assertEqual(1, matches_count)
self.assertEqual([], restore1.filename_only_matches)
self.assertIn(accession2, restore2.filename_only_matches)
self.assertIn(restore2, accession2.filename_only_matches)
def test_same_bytes_and_filename_but_different_md5(self):
session = Session()
accession = AccessionBuilder().build()
restore = create_perfect_match(accession)
restore.md5 = 'filename_only_md5'
session.add(accession)
session.add(restore)
session.commit()
# Verify that filename and bytes are the same, but MD5 differs
self.assertEqual(accession.bytes, restore.bytes)
self.assertEqual(accession.filename, restore.filename)
self.assertNotEqual(accession.md5, restore.md5)
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
self.assertEqual(0, len(new_matches_found))
self.assertEqual(0, len(accession.filename_only_matches))
self.assertEqual(0, len(restore.filename_only_matches))
def test_no_filename_only_match(self):
session = Session()
accession = AccessionBuilder().build()
restore = RestoreBuilder().build()
session.add(accession)
session.add(restore)
session.commit()
# Verify that MD5 checksums, filenames, and bytes are not equal
self.assertNotEqual(accession.md5, restore.md5)
self.assertNotEqual(accession.filename, restore.filename)
self.assertNotEqual(accession.bytes, restore.bytes)
accessions = session.query(Accession)
new_matches_found = find_filename_only_matches(session, accessions)
# No perfect match should be found
self.assertEqual(0, len(new_matches_found))
self.assertEqual(0, len(accession.filename_only_matches))
self.assertEqual(0, len(restore.filename_only_matches))
