def test_it_logs_a_warning_if_document_ids_differ(self, log):
"""
It should log a warning on Document objects mismatch.
If there's an existing DocumentURI and its .document property is
different to the given document it shoulg log a warning.
"""
session = mock_db_session()
# existing_document_uri.document won't be equal to the given document.
existing_document_uri = mock.Mock(document=mock_document())
session.query.return_value.filter.return_value.first.return_value = (
existing_document_uri
)
document.create_or_update_document_uri(
session=session,
claimant="http://example.com/example_claimant.html",
uri="http://example.com/example_uri.html",
type="self-claim",
content_type=None,
document=mock_document(),
created=now(),
updated=now(),
)
assert log.warning.call_count == 1
def test_it_updates_the_existing_DocumentURI_if_there_is_one(self, db_session):
claimant = 'http://example.com/example_claimant.html'
uri = 'http://example.com/example_uri.html'
type_ = 'self-claim'
content_type = ''
document_ = document.Document()
created = yesterday()
updated = yesterday()
document_uri = document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=created,
updated=updated,
)
db_session.add(document_uri)
now_ = now()
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now_,
updated=now_,
)
assert document_uri.created == created
assert document_uri.updated == now_
assert len(db_session.query(document.DocumentURI).all()) == 1, (
"It shouldn't have added any new objects to the db")
def test_it_updates_the_existing_DocumentURI_if_there_is_one(self, db_session):
claimant = 'http://example.com/example_claimant.html'
uri = 'http://example.com/example_uri.html'
type_ = 'self-claim'
content_type = ''
document_ = document.Document()
created = yesterday()
updated = yesterday()
document_uri = document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=created,
updated=updated,
)
db_session.add(document_uri)
now_ = now()
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now_,
updated=now_,
)
assert document_uri.created == created
assert document_uri.updated == now_
assert len(db_session.query(document.DocumentURI).all()) == 1, (
"It shouldn't have added any new objects to the db")
def test_it_logs_a_warning_if_document_ids_differ(self, log):
"""
It should log a warning on Document objects mismatch.
If there's an existing DocumentURI and its .document property is
different to the given document it shoulg log a warning.
"""
session = mock_db_session()
# existing_document_uri.document won't be equal to the given document.
existing_document_uri = mock.Mock(document=mock_document())
session.query.return_value.filter.return_value.first.return_value = (
existing_document_uri)
document.create_or_update_document_uri(
session=session,
claimant="http://example.com/example_claimant.html",
uri="http://example.com/example_uri.html",
type="self-claim",
content_type=None,
document=mock_document(),
created=now(),
updated=now(),
)
assert log.warning.call_count == 1
def test_it_skips_denormalizing_http_uris_to_document(
self, db_session, doc_uri_attrs):
doc_uri_attrs["document"] = document = Document(
web_uri="http://example.com/first_uri.html")
db_session.add(document)
create_or_update_document_uri(session=db_session, **doc_uri_attrs)
document_ = db_session.query(Document).get(document.id)
assert document_.web_uri == "http://example.com/first_uri.html"
def test_raises_retryable_error_when_flush_fails(
self, db_session, monkeypatch, doc_uri_attrs
):
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, "flush", err)
with pytest.raises(ConcurrentUpdateError):
with db_session.no_autoflush: # prevent premature IntegrityError
create_or_update_document_uri(session=db_session, **doc_uri_attrs)
def test_it_logs_a_warning_if_document_ids_differ(
self, log, mock_db_session, factories, doc_uri_attrs
):
# Ensure the document we use, and that returned by filter first are
# different
mock_db_session.query.return_value.filter.return_value.first.return_value = (
factories.DocumentURI()
)
different_document = factories.Document()
create_or_update_document_uri(
session=mock_db_session, **dict(doc_uri_attrs, document=different_document)
)
assert log.warning.call_count == 1
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 test_it_creates_a_new_DocumentURI_if_there_is_no_existing_one(self, db_session):
claimant = "http://example.com/example_claimant.html"
uri = "http://example.com/example_uri.html"
type_ = "self-claim"
content_type = ""
document_ = document.Document()
created = yesterday()
updated = yesterday()
# Add one non-matching DocumentURI to the database.
db_session.add(
document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
# Different content_type means this DocumentURI should not match
# the query.
content_type="different",
document=document_,
created=created,
updated=updated,
)
)
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now(),
updated=now(),
)
document_uri = (
db_session.query(document.DocumentURI)
.order_by(document.DocumentURI.created.desc())
.first()
)
assert document_uri.claimant == claimant
assert document_uri.uri == uri
assert document_uri.type == type_
assert document_uri.content_type == content_type
assert document_uri.document == document_
assert document_uri.created > created
assert document_uri.updated > updated
def test_it_skips_denormalizing_http_s_uri_to_document(self, db_session):
document_ = document.Document(web_uri='http://example.com/first_uri.html')
db_session.add(document_)
document.create_or_update_document_uri(
session=db_session,
claimant='http://example.com/example_claimant.html',
uri='http://example.com/second_uri.html',
type='self-claim',
content_type='',
document=document_,
created=now(),
updated=now(),
)
document_ = db_session.query(document.Document).get(document_.id)
assert document_.web_uri == 'http://example.com/first_uri.html'
def test_it_skips_denormalizing_http_s_uri_to_document(self, db_session):
document_ = document.Document(web_uri="http://example.com/first_uri.html")
db_session.add(document_)
document.create_or_update_document_uri(
session=db_session,
claimant="http://example.com/example_claimant.html",
uri="http://example.com/second_uri.html",
type="self-claim",
content_type="",
document=document_,
created=now(),
updated=now(),
)
document_ = db_session.query(document.Document).get(document_.id)
assert document_.web_uri == "http://example.com/first_uri.html"
def test_it_creates_a_new_DocumentURI_if_there_is_no_existing_one(
self, db_session):
claimant = "http://example.com/example_claimant.html"
uri = "http://example.com/example_uri.html"
type_ = "self-claim"
content_type = ""
document_ = document.Document()
created = yesterday()
updated = yesterday()
# Add one non-matching DocumentURI to the database.
db_session.add(
document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
# Different content_type means this DocumentURI should not match
# the query.
content_type="different",
document=document_,
created=created,
updated=updated,
))
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now(),
updated=now(),
)
document_uri = (db_session.query(document.DocumentURI).order_by(
document.DocumentURI.created.desc()).first())
assert document_uri.claimant == claimant
assert document_uri.uri == uri
assert document_uri.type == type_
assert document_uri.content_type == content_type
assert document_uri.document == document_
assert document_uri.created > created
assert document_uri.updated > updated
def test_raises_retryable_error_when_flush_fails(self, db_session, monkeypatch):
document_ = document.Document()
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, 'flush', err)
with pytest.raises(transaction.interfaces.TransientError):
with db_session.no_autoflush: # prevent premature IntegrityError
document.create_or_update_document_uri(
session=db_session,
claimant='http://example.com',
uri='http://example.org',
type='rel-canonical',
content_type='text/html',
document=document_,
created=now(),
updated=now(),
)
def test_it_creates_a_new_DocumentURI_if_there_is_no_existing_one(
self, db_session):
claimant = 'http://example.com/example_claimant.html'
uri = 'http://example.com/example_uri.html'
type_ = 'self-claim'
content_type = ''
document_ = document.Document()
created = yesterday()
updated = yesterday()
# Add one non-matching DocumentURI to the database.
db_session.add(
document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
# Different content_type means this DocumentURI should not match
# the query.
content_type='different',
document=document_,
created=created,
updated=updated,
))
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now(),
updated=now(),
)
document_uri = db_session.query(document.DocumentURI).all()[-1]
assert document_uri.claimant == claimant
assert document_uri.uri == uri
assert document_uri.type == type_
assert document_uri.content_type == content_type
assert document_uri.document == document_
assert document_uri.created > created
assert document_uri.updated > updated
def test_raises_retryable_error_when_flush_fails(self, db_session, monkeypatch):
document_ = document.Document()
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, 'flush', err)
with pytest.raises(transaction.interfaces.TransientError):
with db_session.no_autoflush: # prevent premature IntegrityError
document.create_or_update_document_uri(
session=db_session,
claimant='http://example.com',
uri='http://example.org',
type='rel-canonical',
content_type='text/html',
document=document_,
created=now(),
updated=now(),
)
def test_it_creates_a_new_DocumentURI_if_there_is_no_existing_one(self, db_session):
claimant = 'http://example.com/example_claimant.html'
uri = 'http://example.com/example_uri.html'
type_ = 'self-claim'
content_type = ''
document_ = document.Document()
created = yesterday()
updated = yesterday()
# Add one non-matching DocumentURI to the database.
db_session.add(document.DocumentURI(
claimant=claimant,
uri=uri,
type=type_,
# Different content_type means this DocumentURI should not match
# the query.
content_type='different',
document=document_,
created=created,
updated=updated,
))
document.create_or_update_document_uri(
session=db_session,
claimant=claimant,
uri=uri,
type=type_,
content_type=content_type,
document=document_,
created=now(),
updated=now(),
)
document_uri = db_session.query(document.DocumentURI).all()[-1]
assert document_uri.claimant == claimant
assert document_uri.uri == uri
assert document_uri.type == type_
assert document_uri.content_type == content_type
assert document_uri.document == document_
assert document_uri.created > created
assert document_uri.updated > updated
def test_raises_retryable_error_when_flush_fails(self, db_session,
monkeypatch):
document_ = document.Document()
def err():
raise sa.exc.IntegrityError(None, None, None)
monkeypatch.setattr(db_session, "flush", err)
with pytest.raises(ConcurrentUpdateError):
with db_session.no_autoflush: # prevent premature IntegrityError
document.create_or_update_document_uri(
session=db_session,
claimant="http://example.com",
uri="http://example.org",
type="rel-canonical",
content_type="text/html",
document=document_,
created=now(),
updated=now(),
)
