def test_merge_documents_deletes_duplicate_documents(self, merge_data):
_, duplicate = merge_data
document.merge_documents(db.Session, merge_data)
db.Session.flush()
assert document.Document.query.get(duplicate.id) is None
def test_raises_retryable_error_when_flush_fails(self, db_session, merge_data, monkeypatch):
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, 'flush', err)
with pytest.raises(transaction.interfaces.TransientError):
document.merge_documents(db_session, merge_data)
def test_merge_documents_deletes_duplicate_documents(self, db_session, merge_data):
_, duplicate = merge_data
document.merge_documents(db_session, merge_data)
db_session.flush()
assert db_session.query(document.Document).get(duplicate.id) is None
def test_merge_documents_deletes_duplicate_documents(merge_data):
master, duplicate = merge_data
merge_documents(db.Session, merge_data)
db.Session.flush()
assert Document.query.get(duplicate.id) is None
def test_merge_documents_rewires_document_meta(self, db_session, merge_data):
master, duplicate = merge_data
document.merge_documents(db_session, merge_data)
db_session.flush()
assert len(master.meta) == 2
assert len(duplicate.meta) == 0
def test_merge_documents_rewires_document_meta(merge_data):
master, duplicate = merge_data
merge_documents(db.Session, merge_data)
db.Session.flush()
assert len(master.meta) == 2
assert len(duplicate.meta) == 0
def test_merge_documents_rewires_document_uris(merge_data):
master, duplicate = merge_data
merge_documents(db.Session, merge_data)
db.Session.flush()
assert len(master.document_uris) == 2
assert len(duplicate.document_uris) == 0
def test_merge_documents_rewires_document_uris(self, merge_data):
master, duplicate = merge_data
document.merge_documents(db.Session, merge_data)
db.Session.flush()
assert len(master.document_uris) == 2
assert len(duplicate.document_uris) == 0
def create_or_update_document_objects(es_ann):
es_doc = es_ann.document
if not es_doc:
return
uris = [u.uri for u in es_doc.document_uris]
documents = Document.find_or_create_by_uris(Session,
es_ann.target_uri,
uris,
created=es_doc.created,
updated=es_doc.updated)
if documents.count() > 1:
document = merge_documents(Session, documents, updated=es_doc.updated)
else:
document = documents.first()
document.updated = es_doc.updated
for uri_ in es_doc.document_uris:
create_or_update_document_uri(uri_, document)
for meta in es_doc.meta:
create_or_update_document_meta(meta, document)
def create_or_update_document_objects(es_ann):
es_doc = es_ann.document
if not es_doc:
return
uris = [u.uri for u in es_doc.document_uris]
documents = Document.find_or_create_by_uris(Session, es_ann.target_uri, uris,
created=es_doc.created,
updated=es_doc.updated)
if documents.count() > 1:
document = merge_documents(Session, documents, updated=es_doc.updated)
else:
document = documents.first()
document.updated = es_doc.updated
for uri_ in es_doc.document_uris:
create_or_update_document_uri(uri_, document)
for meta in es_doc.meta:
create_or_update_document_meta(meta, document)
def test_merge_documents_returns_master(self, db_session, merge_data):
master, _ = merge_data
merged_master = document.merge_documents(db_session, merge_data)
assert merged_master == master
def test_merge_documents_returns_master(merge_data):
master, duplicate = merge_data
merged_master = merge_documents(db.Session, merge_data)
assert merged_master == master
def test_merge_documents_returns_master(self, merge_data):
master, _ = merge_data
merged_master = document.merge_documents(db.Session, merge_data)
assert merged_master == master