def merge_data(self, db_session, request):
master = document.Document(
document_uris=[
document.DocumentURI(
claimant='https://en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='self-claim')
],
meta=[
document.DocumentMeta(
claimant='https://en.wikipedia.org/wiki/Main_Page',
type='title',
value='Wikipedia, the free encyclopedia')
])
duplicate = document.Document(
document_uris=[
document.DocumentURI(
claimant='https://m.en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='rel-canonical')
],
meta=[
document.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)
def merge_data(self, db_session, request):
master = document.Document(
document_uris=[
document.DocumentURI(
claimant='https://en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='self-claim')
],
meta=[
document.DocumentMeta(
claimant='https://en.wikipedia.org/wiki/Main_Page',
type='title',
value='Wikipedia, the free encyclopedia')
])
duplicate_1 = document.Document(
document_uris=[
document.DocumentURI(
claimant='https://m.en.wikipedia.org/wiki/Main_Page',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='rel-canonical')
],
meta=[
document.DocumentMeta(
claimant='https://m.en.wikipedia.org/wiki/Main_Page',
type='title',
value='Wikipedia, the free encyclopedia')
])
duplicate_2 = document.Document(
document_uris=[
document.DocumentURI(
claimant='https://en.wikipedia.org/wiki/Home',
uri='https://en.wikipedia.org/wiki/Main_Page',
type='rel-canonical')
],
meta=[
document.DocumentMeta(
claimant='https://en.wikipedia.org/wiki/Home',
type='title',
value='Wikipedia, the free encyclopedia')
])
db_session.add_all([master, duplicate_1, duplicate_2])
db_session.flush()
master_ann_1 = models.Annotation(userid='luke', document_id=master.id)
master_ann_2 = models.Annotation(userid='alice', document_id=master.id)
duplicate_1_ann_1 = models.Annotation(userid='lucy',
document_id=duplicate_1.id)
duplicate_1_ann_2 = models.Annotation(userid='bob',
document_id=duplicate_1.id)
duplicate_2_ann_1 = models.Annotation(userid='amy',
document_id=duplicate_2.id)
duplicate_2_ann_2 = models.Annotation(userid='dan',
document_id=duplicate_2.id)
db_session.add_all([
master_ann_1, master_ann_2, duplicate_1_ann_1, duplicate_1_ann_2,
duplicate_2_ann_1, duplicate_2_ann_2
])
return (master, duplicate_1, duplicate_2)
def test_it_returns_the_value_of_the_first_title_DocumentMeta(
self, db_session):
doc = document.Document()
document.DocumentMeta(type='title',
value=['The US Title'],
document=doc,
claimant='http://example.com')
document.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_it_updates_an_existing_DocumentMeta_if_there_is_one(self, db_session):
claimant = 'http://example.com/claimant'
type_ = 'title'
value = 'the title'
document_ = document.Document()
created = yesterday()
updated = now()
document_meta = document.DocumentMeta(
claimant=claimant,
type=type_,
value=value,
document=document_,
created=created,
updated=updated,
)
db_session.add(document_meta)
new_updated = now()
document.create_or_update_document_meta(
session=db_session,
claimant=claimant,
type=type_,
value='new value',
document=document.Document(), # This should be ignored.
created=now(), # This should be ignored.
updated=new_updated,
)
assert document_meta.value == 'new value'
assert document_meta.updated == new_updated
assert document_meta.created == created, "It shouldn't update created"
assert document_meta.document == document_, (
"It shouldn't update document")
assert len(db_session.query(document.DocumentMeta).all()) == 1, (
"It shouldn't have added any new objects to the db")
def test_it_returns_None_if_there_are_no_title_DocumentMetas(
self, db_session):
doc = document.Document()
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 test_it_returns_the_value_of_the_one_title_DocumentMeta(
self, db_session):
"""When there's only one DocumentMeta it should return its title."""
doc = document.Document()
document.DocumentMeta(type='title',
value=['The Title'],
document=doc,
claimant='http://example.com')
db_session.add(doc)
db_session.flush()
assert doc.title == 'The Title'
def test_it_creates_a_new_DocumentMeta_if_there_is_no_existing_one(
self, db_session):
claimant = 'http://example.com/claimant'
type_ = 'title'
value = 'the title'
document_ = document.Document()
created = yesterday()
updated = now()
# Add one non-matching DocumentMeta to the database.
# This should be ignored.
db_session.add(
document.DocumentMeta(
claimant=claimant,
# Different type means this should not match the query.
type='different',
value=value,
document=document_,
created=created,
updated=updated,
))
document.create_or_update_document_meta(
session=db_session,
claimant=claimant,
type=type_,
value=value,
document=document_,
created=created,
updated=updated,
)
document_meta = db_session.query(document.DocumentMeta).all()[-1]
assert document_meta.claimant == claimant
assert document_meta.type == type_
assert document_meta.value == value
assert document_meta.document == document_
assert document_meta.created == created
assert document_meta.updated == updated