def __init__(self,
synonyms_collection_filepath,
frames_collection_filepath,
init_ner=True,
init_stemmer=True,
init_frames=True,
use_ner_cache_only=False,
ner_name=supported.ONTONOTES_BERT_MULT_NAME):
assert (isinstance(init_ner, bool))
assert (isinstance(init_frames, bool))
assert (isinstance(init_stemmer, bool))
assert (isinstance(ner_name, str))
self.__auth_objects = None
self.__use_ner_cache_only = use_ner_cache_only
self.__synonyms = None
self.__stemmer = None
self.__frame_variants = None
self.__frames = None
self.__pos_tagger = None
self.__syntax = None
self.__use_auth_list = False
self.__frames_cache = None
# NER
self.__ner_cache = None
self.__ner_class_type = Default.get_class_by_ner_name(ner_name)
self.__ner = None
if init_stemmer:
self.__stemmer = Default.create_default_stemmer()
if self.__stemmer is not None:
self.__pos_tagger = POSMystemWrapper(self.__stemmer.MystemInstance)
if init_frames:
self.__frames = Default.create_default_frames_collection(
frames_collection_filepath)
if self.__stemmer is not None and self.__frames is not None:
self.__frame_variants = Default.create_default_frame_variants_collection(
frames=self.__frames, stemmer=self.__stemmer)
if self.__frame_variants is not None:
self.__frames_helper = FramesHelper(self.__frame_variants)
if init_ner and not use_ner_cache_only:
self.__ner = self.__ner_class_type()
self.__synonyms = Default.create_default_synonyms_collection(
filepath=synonyms_collection_filepath,
stemmer=None
if self.DISABLE_LEMMA_FOR_SYNONYMS else self.__stemmer)
self.__auth_objects = AuthorizedObjectsCollection(OrderedDict())
NewsSourceDirArg.add_argument(parser)
SourceNewsReaderArg.add_argument(parser)
NewsStartFromIndexArg.add_argument(parser)
ParseFramesInSentencesArgs.add_argument(parser)
RuSentiFramesCacheArgs.add_argument(parser)
# Parsing arguments.
args = parser.parse_args()
# Reading arguments.
src_dir = NewsSourceDirArg.read_argument(args)
reader = SourceNewsReaderArg.read_argument(args)
frames_filepath = RuSentiFramesCacheArgs.read_argument(args)
parse_frames_in_sents = ParseFramesInSentencesArgs.read_argument(args)
start_from_index = NewsStartFromIndexArg.read_argument(args)
stemmer = Default.create_default_stemmer()
frames = Default.create_default_frames_collection(frames_filepath)
f_var = Default.create_default_frame_variants_collection(frames=frames,
stemmer=stemmer)
run_frames_cache(reader=reader,
src_dir=src_dir,
version=basename(frames_filepath),
frames=frames,
frames_helper=FramesHelper(f_var),
stemmer=stemmer,
parse_frames_in_sentences=parse_frames_in_sents,
start_from_index=start_from_index,
miniter_count=2000000)