def extract_acts_by_person(person, db_uri=None):
"""Extracts movies or tv shows the passed *person* played in."""
person_source = process_url(u'http://www.imdb.com/name/' + person.id)
for title_id in parser.roles(person_source):
title = get_cached_model(title_id)
if title is None:
# We don't have the movie or tv show cached; index it first.
source = process_url(
urlparse.urljoin(constants.url_base, 'title/' + title_id))
new_title = factory.model_builder(source)
if new_title:
if store_model(new_title):
title = get_cached_model(title_id)
if title is not None:
store_cast_member(title, person)
def models_from_source(source, db_uri=None):
for model_id in parser.model_ids(source):
# Search the database for a cached model matching the current model id.
cached_model = get_cached_model(model_id, db_uri)
if cached_model:
yield cached_model
continue
# When we reach this point in the loop the model is not cached in our database.
# Build the model, cache and yield it.
_model_url = model_url(model_id)
if _model_url is None:
continue
model_source = process_url(_model_url)
model = factory.model_builder(model_source, db_uri)
if model:
if store_model(model, db_uri):
yield get_cached_model(model_id, db_uri)
def models_from_json(json_data, db_uri=None):
"""Parses the passed search result *json_data* and extracts the models contained in it."""
if not json_data or not json_data.get('d'):
return
for result in json_data.get('d'):
cached_model = get_cached_model(result['id'], db_uri)
if cached_model:
store_search_result(json_data.get('q'), cached_model, db_uri)
yield cached_model
continue
_model_url = model_url(result['id'])
if _model_url is None:
continue
model_source = process_url(_model_url)
model = factory.model_builder(model_source, db_uri)
if model:
store_search_result(json_data.get('q'), model, db_uri)
if store_model(model, db_uri):
yield get_cached_model(result['id'], db_uri)
def test_person_builder(self):
with open(os.path.join(self.here, 'files',
'morgan_freeman.html')) as f:
person_source = f.read()
person = factory.model_builder(person_source)
assert isinstance(person, models.Person) == True
def test_tvshow_builder(self):
with open(os.path.join(self.here, 'files', 'black_mirror.html')) as f:
tvshow_source = f.read()
tvshow = factory.model_builder(tvshow_source, self.db_uri)
assert isinstance(tvshow, models.TVShow) == True
def test_movie_builder(self):
with open(os.path.join(self.here, 'files', 'the_matrix.html')) as f:
movie_source = f.read()
movie = factory.model_builder(movie_source, self.db_uri)
assert isinstance(movie, models.Movie) == True