def test_it_logs_when_its_called(self, caplog, db_session, duplicate_docs):
caplog.set_level(logging.INFO)
merge_documents(db_session, duplicate_docs)
assert caplog.record_tuples == [("h.models.document._document", 20,
"Merging 3 documents")]
def test_it_deletes_all_but_the_first(self, db_session, duplicate_docs):
merge_documents(db_session, duplicate_docs)
db_session.flush()
assert not (db_session.query(Document).filter(
Document.id.in_([duplicate_docs[1].id, duplicate_docs[2].id
])).count())
def test_it_raises_retryable_error_when_flush_fails(
self, db_session, duplicate_docs, monkeypatch):
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, "flush", err)
with pytest.raises(ConcurrentUpdateError):
merge_documents(db_session, duplicate_docs)
def test_it_moves_annotations_to_the_first(self, db_session,
duplicate_docs):
merge_documents(db_session, duplicate_docs)
db_session.flush()
for document, expected_count in zip(duplicate_docs, (3, 0, 0)):
count = (db_session.query(
models.Annotation).filter_by(document_id=document.id).count())
assert count == expected_count
def test_it_moves_sub_items_to_the_first(self, db_session, duplicate_docs,
datetime, updated, sub_item):
items = []
for doc in duplicate_docs[1:]:
items.extend(getattr(doc, sub_item))
master = merge_documents(db_session, duplicate_docs, updated=updated)
db_session.flush()
assert [len(getattr(doc, sub_item))
for doc in duplicate_docs] == [3, 0, 0]
expected_date = updated if updated else datetime.utcnow.return_value
for item in items:
assert item.document == master
assert item.updated == expected_date
def test_it_returns_the_first_doc(self, db_session, duplicate_docs):
merged = merge_documents(db_session, duplicate_docs)
assert merged == duplicate_docs[0]