def delete_corpus_helper(corpus=None, force=False):
"""Deletes one or several corpora.
Args:
corpus: string if you want to delete a specific corpus, None for
every corpus.
"""
if corpus is None:
if force or query_yes_no(
'Do you want to delete every working directory?'):
corpora = read_corpus()
write_corpus({})
for _, value in corpora.iteritems():
rmtree(value)
else:
corpora = read_corpus()
if corpus in corpora:
try:
rmtree(corpora[corpus], ignore_errors=False)
except Exception as exc:
logger.warning("Could not delete {0}: {1}".format(
corpora[corpus], exc))
del corpora[corpus]
else:
raise ValueError('This working directory does not exist.')
write_corpus(corpora)
def delete_broadcast_helper(broadcast=None, corpus=None, force=False):
"""Deletes one or all broadcast(s) from a corpus.
Args:
broadcast: Name of the broadcast (default: None). None to delete all broadcast
corpus: Name of the corpus (default: None)
force (boolean): do not ask for confirmation on deleting all
"""
corpora = read_corpus()
corpus_path = corpora[corpus]
if broadcast is None:
if force or query_yes_no('Do you want to delete every '
'broadcast in this directory?'):
corpora = [
os.path.join(corpus_path, item)
for item in os.listdir(corpus_path)
if item != 'references' and item != 'raw'
]
corpora = [
item for item in corpora
if os.path.exists(os.path.join(item, 'references.json'))
]
for broadcast in corpora:
# Get only the end of the path
rmtree(broadcast, ignore_errors=False)
else:
if os.path.exists(
os.path.join(corpus_path, broadcast, 'references.json')):
rmtree(os.path.join(corpus_path, broadcast), ignore_errors=False)
def list_broadcast_helper(corpus):
"""List every broadcast in the corpus
Args:
corpus: Name of the corpus
Returns:
a dict of {broadcast_name: broadcast_path}
"""
corpora = read_corpus()
if corpus not in corpora:
raise ValueError('The working directory you stated does not exist')
corpus_path = corpora[corpus]
broadcasts = [
os.path.join(corpus_path, item) for item in os.listdir(corpus_path)
if item != 'references' and item != 'raw'
]
broadcasts = [
item for item in broadcasts
if os.path.exists(os.path.join(item, 'references.json'))
]
broadcast_dict = {}
for broadcast in broadcasts:
# Get only the end of the path
broadcast_name = os.path.relpath(broadcast, os.path.dirname(broadcast))
broadcast_dict[broadcast_name] = broadcast
return broadcast_dict
def init_corpus_helper(path, corpus):
"""Creates a new corpus.
Args:
path: the path where the corpus should be created.
corpus: name of the corpus
"""
corpus_path = os.path.join(path, corpus)
corpora = read_corpus()
if corpus in corpora:
raise ValueError('Corpus already exists')
if not os.path.exists(corpus_path):
os.mkdir(corpus_path)
if not os.path.exists(os.path.join(corpus_path, 'references')):
os.mkdir(os.path.join(corpus_path, 'references'))
corpora[corpus] = corpus_path
write_corpus(corpora)
def init_broadcast_helper(broadcast, corpus):
"""Initialize a broadcast inside a corpus.
Args:
broadcast: Name of the broadcast
corpus: Name of the corpus
"""
corpora = read_corpus()
if corpus not in corpora:
raise ValueError('The corpus you stated does not exist')
broadcast_path = os.path.join(corpora[corpus], broadcast)
try:
os.mkdir(broadcast_path)
except OSError as e:
if 'File exists' in e:
logger.warning(
'The broadcast dir already exists. Still initializing')
else:
raise e
open(os.path.join(broadcast_path, 'groundtruth.xml'), 'a').close()
open(os.path.join(broadcast_path, 'references.json'), 'a').close()
return corpora[corpus], broadcast
def list_corpus_helper():
"""List all the corpora on a computer.
"""
corpora = read_corpus()
return corpora