def test_you_cannot_add_duplicate_document_uris(self, db_session):
# You can't add DocumentURI's with the same claimant, uri, type and
# content_type, even if they have different documents.
attrs = {
"claimant": "http://www.example.com",
"uri": "http://www.example.com",
"type": "foo",
"content_type": "bar",
}
db_session.add(DocumentURI(**attrs, document=Document()))
db_session.add(DocumentURI(**attrs, document=Document()))
with pytest.raises(sa.exc.IntegrityError):
db_session.commit()
def test_repr(self):
uri = DocumentURI(id=1234)
repr_string = repr(uri)
assert "DocumentURI" in repr_string
assert "1234" in repr_string
def test_it_creates_a_new_DocumentURI_if_there_is_no_existing_one(
self, db_session, doc_uri_attrs):
original_attrs = doc_uri_attrs
updated_attrs = dict(original_attrs,
created=datetime.now(),
updated=datetime.now())
# Add one non-matching DocumentURI to the database.
db_session.add(
DocumentURI(**dict(original_attrs, content_type="different")))
create_or_update_document_uri(session=db_session, **updated_attrs)
document_uri = (db_session.query(DocumentURI).order_by(
DocumentURI.created.desc()).first())
assert document_uri == Any.object.with_attrs(updated_attrs)
def test_it_updates_the_existing_DocumentURI_if_there_is_one(
self, db_session, doc_uri_attrs):
original_attrs = doc_uri_attrs
updated_attrs = dict(original_attrs,
created=datetime.now(),
updated=datetime.now())
document_uri = DocumentURI(**original_attrs)
db_session.add(document_uri)
create_or_update_document_uri(session=db_session, **updated_attrs)
assert document_uri.created == original_attrs["created"]
assert document_uri.updated == updated_attrs["updated"]
assert (len(db_session.query(DocumentURI).all()) == 1
), "It shouldn't have added any new objects to the db"
def find_or_create_by_uris( # pylint: disable=too-many-arguments
cls,
session,
claimant_uri,
uris,
created=None,
updated=None):
"""
Find or create documents from a claimant uri and a list of uris.
It tries to find a document based on the claimant and the set of uris.
If none can be found it will return a new document with the claimant
uri as its only document uri as a self-claim. It is the callers
responsibility to create any other document uris.
"""
finduris = [claimant_uri] + uris
documents = cls.find_by_uris(session, finduris)
if not documents.count():
doc = Document(created=created, updated=updated)
DocumentURI(
document=doc,
claimant=claimant_uri,
uri=claimant_uri,
type="self-claim",
created=created,
updated=updated,
)
session.add(doc)
try:
session.flush()
except sa.exc.IntegrityError as err:
raise ConcurrentUpdateError(
"concurrent document creation") from err
return documents
def test_it_normalizes_the_uri(self):
document_uri = DocumentURI(uri="http://example.com/")
assert document_uri.uri_normalized == "httpx://example.com"