def init_person_name_corrections(self, object_list):
if len(object_list) == 0:
name = self.field_params.get('person_name')
if name is not None and len(name) > 5:
fio = TRussianFio(name, from_search_request=False)
if fio.is_resolved:
name = TRussianFio.convert_to_rml_encoding(fio.get_normalized_person_name())
corrections = FIO_MISSPELL_CORRECTOR.correct_misspell(name)
if len(corrections) > 0:
if name != corrections[0]:
self.person_name_corrections = list(TRussianFio.convert_from_rml_encoding(c) for c in corrections[:10])
self.log('person names corrections count = {}'.format(len(self.person_name_corrections)))
def init_person_info(self, section_json):
person_info = section_json.get('person')
if person_info is None:
fio = section_json.get('fio')
if fio is None:
raise TSmartParserSectionJson.SerializerException(
"cannot find nor 'person' neither 'fio' key in json")
else:
fio = person_info.get('name', person_info.get('name_raw'))
if fio is None:
raise TSmartParserSectionJson.SerializerException(
"cannot find 'name' or 'name_raw' in json")
fio = normalize_fio_before_db_insert(fio)
resolved_fio = TRussianFio(fio)
if not resolved_fio.is_resolved:
raise TSmartParserSectionJson.SerializerException(
"cannot resolve person name {}".format(fio))
self.section.person_name = resolved_fio.get_normalized_person_name()
self.section.position = person_info.get("role")
self.section.department = person_info.get("department")
def write_list(self, outp):
used = set()
for person_name in self.build_person_names():
fio = TRussianFio(person_name.strip(),
from_search_request=False,
make_lower=False)
if fio.is_resolved:
norm = TRussianFio.convert_to_rml_encoding(
fio.get_normalized_person_name())
if not is_in_rml_alphabet(norm):
self.logger.debug("bad alphabet {}".format(
person_name.strip()))
else:
if norm not in used:
outp.write("{}\n".format(norm))
used.add(norm)
norm = TRussianFio.convert_to_rml_encoding(
fio.get_abridged_normalized_person_name())
if norm not in used:
outp.write("{}\n".format(norm))
used.add(norm)
self.logger.info("create {} person names in {}".format(
len(used), outp.name))
assert len(used) > 0