def read_issuers(file_path, corpus):
if not os.path.isfile(file_path):
return
data = jsonfile.read_json_file(file_path)
for issuer_idx, issuer_data in data.items():
issuer_type = issuer_data.get('type', None)
issuer_info = issuer_data.get('info', {})
if issuer_type == 'speaker':
gender = issuers.Gender(
issuer_data.get('gender', 'unknown').lower())
age_group = issuers.AgeGroup(
issuer_data.get('age_group', 'unknown').lower())
native_language = issuer_data.get('native_language', None)
issuer = issuers.Speaker(issuer_idx,
gender=gender,
age_group=age_group,
native_language=native_language,
info=issuer_info)
elif issuer_type == 'artist':
name = issuer_data.get('name', None)
issuer = issuers.Artist(issuer_idx,
name=name,
info=issuer_info)
else:
issuer = issuers.Issuer(issuer_idx, info=issuer_info)
corpus.import_issuers(issuer)
def generate_issuers(n, rand=None):
if rand is None:
rand = random.Random()
items = []
for issuer_index in range(n):
issuer_idx = 'issuer-{}'.format(issuer_index)
issuer_type = rand.randint(1, 3)
if issuer_type == 1:
issuer = issuers.Speaker(
issuer_idx,
gender=issuers.Gender.UNKNOWN,
age_group=issuers.AgeGroup.CHILD,
native_language='de'
)
elif issuer_type == 2:
issuer = issuers.Artist(issuer_idx, 'badam')
else:
issuer = issuers.Issuer(issuer_idx)
items.append(issuer)
return items
def _create_or_get_music_issuer(corpus, file_idx, labels):
if file_idx not in labels:
return None
issuer_idx = labels[file_idx][2]
if issuer_idx not in corpus.issuers:
issuer = issuers.Artist(issuer_idx, name=issuer_idx)
corpus.import_issuers(issuer)
return issuer_idx