def test_document_title_returns_first():
doc = Document()
DocumentMeta(type='title', value='The US Title', document=doc, claimant='http://example.com')
DocumentMeta(type='title', value='The UK Title', document=doc, claimant='http://example.co.uk')
db.Session.add(doc)
db.Session.flush()
assert doc.title == 'The US Title'
def test_document_title_meta_not_found():
doc = Document()
DocumentMeta(type='other', value='something', document=doc, claimant='http://example.com')
db.Session.add(doc)
db.Session.flush()
assert doc.title is None
def create_or_update_document_meta(es_meta, pg_document):
meta = DocumentMeta.query.filter(
DocumentMeta.claimant_normalized == es_meta.claimant_normalized,
DocumentMeta.type == es_meta.type).one_or_none()
if meta is None:
meta = DocumentMeta(claimant=es_meta.claimant,
type=es_meta.type,
value=es_meta.value,
created=es_meta.created,
updated=es_meta.updated,
document=pg_document)
Session.add(meta)
else:
meta.value = es_meta.value
meta.updated = es_meta.updated
if not meta.document == pg_document:
log.warn('Found DocumentMeta with id {:d} does not match expected document with id {:d}', meta.id, pg_document.id)
def create_or_update_document_meta(es_meta, pg_document):
meta = DocumentMeta.query.filter(
DocumentMeta.claimant_normalized == es_meta.claimant_normalized,
DocumentMeta.type == es_meta.normalized_type).one_or_none()
if meta is None:
meta = DocumentMeta(claimant=es_meta.claimant,
type=es_meta.normalized_type,
value=es_meta.value,
created=es_meta.created,
updated=es_meta.updated,
document=pg_document)
Session.add(meta)
else:
meta.value = es_meta.value
meta.updated = es_meta.updated
if not meta.document == pg_document:
log.warn(
'Found DocumentMeta with id %d does not match expected document with id %d',
meta.id, pg_document.id)
def merge_data(request):
master = Document(document_uris=[DocumentURI(
claimant='https://en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='self-claim')],
meta=[DocumentMeta(
claimant='https://en.wikipedia.org/wiki/Main_Page',
type='title',
value='Wikipedia, the free encyclopedia')])
duplicate = Document(document_uris=[DocumentURI(
claimant='https://m.en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='rel-canonical')],
meta=[DocumentMeta(
claimant='https://m.en.wikipedia.org/wiki/Main_Page',
type='title',
value='Wikipedia, the free encyclopedia')])
db.Session.add_all([master, duplicate])
db.Session.flush()
return (master, duplicate)