def test_it_calls_first(self, annotation, session, Document):
"""If it finds only one document it calls first()."""
Document.find_or_create_by_uris.return_value = mock.Mock(
count=mock.Mock(return_value=1))
document.update_document_metadata(session, annotation, [], [])
Document.find_or_create_by_uris.return_value.first.assert_called_once_with(
)
def test_it_calls_first(self, annotation, session, Document):
"""If it finds only one document it calls first()."""
Document.find_or_create_by_uris.return_value = mock.Mock(
count=mock.Mock(return_value=1)
)
document.update_document_metadata(session, annotation, [], [])
Document.find_or_create_by_uris.return_value.first.assert_called_once_with()
def make_anno(data, dbdocs):
#document_uri_dicts = data['document']['document_uri_dicts']
#document_meta_dicts = data['document']['document_meta_dicts']
#del data['document']
#data = {k:v for k, v in data.items() if k != 'document'} # prevent overwrite on batch load
annotation = models.Annotation(
**data) # FIXME for batch the overhead here is stupid beyond belief
annotation.document_id = dbdocs[uri_normalize(annotation.target_uri)].id
#for k, v in data.items():
#print(k, v)
#setattr(annotation, k, v)
#id, created, updated = extra
#annotation.id = id
#annotation.created = created
#annotation.updated = updated
return annotation
# this baby is super slow
document = update_document_metadata(session,
annotation.target_uri,
document_meta_dicts,
document_uri_dicts,
created=created,
updated=updated)
annotation.document = document
return annotation
def test_it_updates_document_web_uri(self, annotation, Document, factories,
session):
document_ = mock.Mock(web_uri=None)
Document.find_or_create_by_uris.return_value.count.return_value = 1
Document.find_or_create_by_uris.return_value.first.return_value = document_
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
document_.update_web_uri.assert_called_once_with()
def create_anno(self, row):
datum = validate(row)
document_dict = datum.pop('document')
document_uri_dicts = document_dict['document_uri_dicts']
document_meta_dicts = document_dict['document_meta_dicts']
id = row['id']
target_uri = datum['target_uri']
created = row['created']
updated = row['updated']
annotation = models.Annotation(**datum)
document = update_document_metadata( # TODO update normalization rules
self.session,
target_uri,
document_meta_dicts,
document_uri_dicts,
created=created, # FIXME doesn't quite seem right, would klobber
updated=updated)
print(id)
annotation.document = document
annotation.id = id
annotation.target_uri = target_uri
annotation.created = created
annotation.updated = updated
self.session.add(annotation)
self.session.flush()
self.session.commit() # FIXME hypothesis doesn't call this
def test_it_updates_document_updated(self, annotation, Document,
merge_documents, session):
yesterday_ = "yesterday"
document_ = merge_documents.return_value = mock.Mock(
updated=yesterday_)
Document.find_or_create_by_uris.return_value.first.return_value = document_
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
assert document_.updated == annotation.updated
def test_it_updates_document_updated(
self, annotation, Document, merge_documents, session
):
yesterday_ = "yesterday"
document_ = merge_documents.return_value = mock.Mock(updated=yesterday_)
Document.find_or_create_by_uris.return_value.first.return_value = document_
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
assert document_.updated == annotation.updated
def test_it_updates_document_web_uri(
self, annotation, Document, factories, session
):
document_ = mock.Mock(web_uri=None)
Document.find_or_create_by_uris.return_value.count.return_value = 1
Document.find_or_create_by_uris.return_value.first.return_value = document_
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
document_.update_web_uri.assert_called_once_with()
def test_it_saves_all_the_document_uris(
self, session, annotation, Document, create_or_update_document_uri
):
"""It creates or updates a DocumentURI for each document URI dict."""
Document.find_or_create_by_uris.return_value.count.return_value = 1
document_uri_dicts = [
{
"uri": "http://example.com/example_1",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_2",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_3",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
]
document.update_document_metadata(
session,
annotation.target_uri,
[],
document_uri_dicts,
annotation.created,
annotation.updated,
)
assert create_or_update_document_uri.call_count == 3
for doc_uri_dict in document_uri_dicts:
create_or_update_document_uri.assert_any_call(
session=session,
document=Document.find_or_create_by_uris.return_value.first.return_value,
created=annotation.created,
updated=annotation.updated,
**doc_uri_dict
)
def test_it_saves_all_the_document_uris(self, session, annotation,
Document,
create_or_update_document_uri):
"""It creates or updates a DocumentURI for each document URI dict."""
Document.find_or_create_by_uris.return_value.count.return_value = 1
document_uri_dicts = [
{
"uri": "http://example.com/example_1",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_2",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_3",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
]
document.update_document_metadata(
session,
annotation.target_uri,
[],
document_uri_dicts,
annotation.created,
annotation.updated,
)
assert create_or_update_document_uri.call_count == 3
for doc_uri_dict in document_uri_dicts:
create_or_update_document_uri.assert_any_call(
session=session,
document=Document.find_or_create_by_uris.return_value.first.
return_value,
created=annotation.created,
updated=annotation.updated,
**doc_uri_dict)
def test_it_uses_the_target_uri_to_get_the_document(
self, annotation, Document, session
):
document_uri_dicts = [
{
"uri": "http://example.com/example_1",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_2",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_3",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
]
document.update_document_metadata(
session,
annotation.target_uri,
[],
document_uri_dicts,
annotation.created,
annotation.updated,
)
Document.find_or_create_by_uris.assert_called_once_with(
session,
annotation.target_uri,
[
"http://example.com/example_1",
"http://example.com/example_2",
"http://example.com/example_3",
],
created=annotation.created,
updated=annotation.updated,
)
def test_it_uses_the_target_uri_to_get_the_document(
self, annotation, Document, session):
document_uri_dicts = [
{
"uri": "http://example.com/example_1",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_2",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
{
"uri": "http://example.com/example_3",
"claimant": "http://example.com/claimant",
"type": "type",
"content_type": None,
},
]
document.update_document_metadata(
session,
annotation.target_uri,
[],
document_uri_dicts,
annotation.created,
annotation.updated,
)
Document.find_or_create_by_uris.assert_called_once_with(
session,
annotation.target_uri,
[
"http://example.com/example_1",
"http://example.com/example_2",
"http://example.com/example_3",
],
created=annotation.created,
updated=annotation.updated,
)
def test_it_saves_all_the_document_metas(
self, annotation, create_or_update_document_meta, Document, session
):
"""It creates or updates a DocumentMeta for each document meta dict."""
Document.find_or_create_by_uris.return_value.count.return_value = 1
document_meta_dicts = [
{
"claimant": "http://example.com/claimant",
"type": "title",
"value": "foo",
},
{
"type": "article title",
"value": "bar",
"claimant": "http://example.com/claimant",
},
{
"type": "site title",
"value": "gar",
"claimant": "http://example.com/claimant",
},
]
document.update_document_metadata(
session,
annotation.target_uri,
document_meta_dicts,
[],
annotation.created,
annotation.updated,
)
assert create_or_update_document_meta.call_count == 3
for document_meta_dict in document_meta_dicts:
create_or_update_document_meta.assert_any_call(
session=session,
document=Document.find_or_create_by_uris.return_value.first.return_value,
created=annotation.created,
updated=annotation.updated,
**document_meta_dict
)
def test_it_saves_all_the_document_metas(self, annotation,
create_or_update_document_meta,
Document, session):
"""It creates or updates a DocumentMeta for each document meta dict."""
Document.find_or_create_by_uris.return_value.count.return_value = 1
document_meta_dicts = [
{
"claimant": "http://example.com/claimant",
"type": "title",
"value": "foo",
},
{
"type": "article title",
"value": "bar",
"claimant": "http://example.com/claimant",
},
{
"type": "site title",
"value": "gar",
"claimant": "http://example.com/claimant",
},
]
document.update_document_metadata(
session,
annotation.target_uri,
document_meta_dicts,
[],
annotation.created,
annotation.updated,
)
assert create_or_update_document_meta.call_count == 3
for document_meta_dict in document_meta_dicts:
create_or_update_document_meta.assert_any_call(
session=session,
document=Document.find_or_create_by_uris.return_value.first.
return_value,
created=annotation.created,
updated=annotation.updated,
**document_meta_dict)
def test_if_there_are_multiple_documents_it_merges_them_into_one(
self, annotation, Document, merge_documents, session):
"""If it finds more than one document it calls merge_documents()."""
Document.find_or_create_by_uris.return_value = mock.Mock(
count=mock.Mock(return_value=3))
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
merge_documents.assert_called_once_with(
session,
Document.find_or_create_by_uris.return_value,
updated=annotation.updated,
)
def test_if_there_are_multiple_documents_it_merges_them_into_one(
self, annotation, Document, merge_documents, session
):
"""If it finds more than one document it calls merge_documents()."""
Document.find_or_create_by_uris.return_value = mock.Mock(
count=mock.Mock(return_value=3)
)
document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
merge_documents.assert_called_once_with(
session,
Document.find_or_create_by_uris.return_value,
updated=annotation.updated,
)
def test_it_returns_a_document(
self, annotation, create_or_update_document_meta, Document, session
):
Document.find_or_create_by_uris.return_value.count.return_value = 1
result = document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
assert result == Document.find_or_create_by_uris.return_value.first.return_value
def test_it_returns_a_document(self, annotation,
create_or_update_document_meta, Document,
session):
Document.find_or_create_by_uris.return_value.count.return_value = 1
result = document.update_document_metadata(
session,
annotation.target_uri,
[],
[],
annotation.created,
annotation.updated,
)
assert result == Document.find_or_create_by_uris.return_value.first.return_value
def make_metadata(self, create, extracted, **kwargs):
"""Create associated document metadata for the annotation."""
# The metadata objects are going to be added to the db, so if we're not
# using the create strategy then simply don't make any.
if not create:
return
def document_uri_dict():
"""
Return a randomly generated DocumentURI dict for this annotation.
This doesn't add anything to the database session yet.
"""
document_uri = DocumentURI.build(
document=None, claimant=self.target_uri, uri=self.target_uri
)
return dict(
claimant=document_uri.claimant,
uri=document_uri.uri,
type=document_uri.type,
content_type=document_uri.content_type,
)
document_uri_dicts = [document_uri_dict() for _ in range(random.randint(1, 3))]
def document_meta_dict(type_=None):
"""
Return a randomly generated DocumentMeta dict for this annotation.
This doesn't add anything to the database session yet.
"""
kwargs = {"document": None, "claimant": self.target_uri}
if type_ is not None:
kwargs["type"] = type_
document_meta = DocumentMeta.build(**kwargs)
return dict(
claimant=document_meta.claimant,
type=document_meta.type,
value=document_meta.value,
)
document_meta_dicts = [
document_meta_dict() for _ in range(random.randint(1, 3))
]
# Make sure that there's always at least one DocumentMeta with
# type='title', so that we never get annotation.document.title is None:
if "title" not in [m["type"] for m in document_meta_dicts]:
document_meta_dicts.append(document_meta_dict(type_="title"))
self.document = update_document_metadata(
orm.object_session(self),
self.target_uri,
document_meta_dicts=document_meta_dicts,
document_uri_dicts=document_uri_dicts,
created=self.created,
updated=self.updated,
)