def find_landlord_tenant(text: str):
# landlord, tenant = find_landlord_tenant_re(text)
# if landlord or tenant:
# return landlord, tenant
companies = list(get_companies(text, detail_type=True, name_upper=True, strict=True))
text = text.lower()
min_index_landlord = min_index_of_word(text, _LANDLORD_SYNONYMS)
min_index_tenant = min_index_of_word(text, _TENANT_SYNONYMS)
if min_index_landlord < min_index_tenant:
landlord = companies[0] if len(companies) > 0 else None
tenant = companies[1] if len(companies) > 1 else None
else:
tenant = companies[0] if len(companies) > 0 else None
landlord = companies[1] if len(companies) > 1 else None
if landlord is not None and landlord[0] is not None:
landlord = '{0}{1}'.format(landlord[0].upper(),
(' ' + landlord[1].upper()) if landlord[1] is not None else '')
if tenant is not None and tenant[0] is not None:
tenant = '{0}{1}'.format(tenant[0].upper(),
(' ' + tenant[1].upper()) if tenant[1] is not None else '')
return landlord, tenant
def get_employer_name(text, return_source=False):
definitions = list(get_definitions(text))
companies = []
defined_employer_found = False
defined_employee_found = False
first_company_string = None
for d in definitions:
if d.lower() in TRIGGER_LIST_COMPANY:
defined_employer_found = True
if d.lower() in TRIGGER_LIST_EMPLOYEE:
defined_employee_found = True
if defined_employee_found is True and defined_employer_found is True:
break
if defined_employer_found and defined_employee_found:
companies = list(get_companies(text))
if len(companies) > 0:
# take first employer found
first_company_string = ', '.join(str(s) for s in companies[0])
if return_source:
return first_company_string, text
else:
return first_company_string
def parse(self,
log: ProcessLogger,
text,
text_unit_id,
_text_unit_lang,
document_initial_load: bool = False,
**kwargs) -> Optional[ParseResults]:
# Here we override saving logic to workaround race conditions on party creation vs party usage saving
if not document_initial_load:
PartyUsage.objects.filter(text_unit_id=text_unit_id).delete()
found = list(
get_companies(text,
count_unique=True,
detail_type=True,
name_upper=True))
if found:
for _party in found:
name, _type, type_abbr, type_label, type_desc, count = _party
defaults = dict(type=_type,
type_label=type_label,
type_description=type_desc)
party, created = Party.objects.get_or_create(
name=name, type_abbr=type_abbr or '', defaults=defaults)
return ParseResults({
PartyUsage: [
PartyUsage(text_unit_id=text_unit_id,
party=party,
count=count)
]
})
def _extract_variants_from_text(self, field, text: str, **kwargs):
companies = list(get_companies(text, detail_type=True, name_upper=True, strict=True))
if not companies:
return None
return ['{0}{1}'.format(company[0].upper(),
(' ' + company[1].upper()) if company[1] is not None else '')
for company in companies]
def get_employee_name(text, return_source=False):
definitions = list(get_definitions(text))
fake_person = False
found_employee = None
defined_employee_found = False
for d in definitions:
if d.lower() in TRIGGER_LIST_EMPLOYEE:
defined_employee_found = True
break
if defined_employee_found:
persons = list(get_persons(text))
companies = list(get_companies(text))
for p in persons:
person_is_a_company = False
for f in FALSE_PEOPLE:
if f in str(p).lower():
fake_person = True
if not fake_person:
for c in companies:
# persons and companies return slightly different values for same text
# so need to standardize to compare
if len(c) > 0:
if c[1] is not None and c[0] is not None:
company_full_string = str(
clean(c[0]) + clean(c[1]))
else:
company_full_string = str(clean(c[0]))
employee_full_string = str(clean(p))
# handle this- where get_companies picks up more surrounding text
# than get_persons: EMPLOYMENT AGREEMENT WHEREAS, Kensey Nash Corporation,
# a Delaware corporation (the “Company”) and Todd M. DeWitt
# (the “Executive”) entered into that certain Amended
# and Restated Employment Agreement,...
if (employee_full_string == company_full_string or
employee_full_string in company_full_string):
person_is_a_company = True
if not person_is_a_company and not fake_person:
found_employee = str(p)
# take first person found meeting our employee criteria
break
fake_person = False # reset for next person
if return_source:
return found_employee, text
else:
return found_employee
def extraction_function(self, field, possible_value, text):
if possible_value:
return possible_value
if possible_value is None and not text:
return None
companies = list(
get_companies(text, detail_type=True, name_upper=True,
strict=True))
company = ValueExtractionHint.get_value(companies, field.item_number)
if company:
return '{0}{1}'.format(
company[0].upper(),
(' ' + company[1].upper()) if company[1] is not None else '')
else:
return None
def get_companies(cls,
text: str,
strict: bool = False,
use_gnp: bool = False,
detail_type: bool = False,
count_unique: bool = False,
name_upper: bool = False,
parse_name_abbr: bool = False,
return_source: bool = False):
_filter = cls.get_banlist_filter()
return get_companies(text,
strict=strict,
use_gnp=use_gnp,
detail_type=detail_type,
count_unique=count_unique,
name_upper=name_upper,
parse_name_abbr=parse_name_abbr,
return_source=return_source,
banlist_usage=_filter)
def parse(self, text, text_unit_id, _text_unit_lang,
**kwargs) -> ParseResults:
found = list(
get_companies(text,
count_unique=True,
detail_type=True,
name_upper=True))
if found:
pu_list = []
for _party in found:
name, _type, type_abbr, type_label, type_desc, count = _party
defaults = dict(type=_type,
type_label=type_label,
type_description=type_desc)
party, _ = Party.objects.get_or_create(name=name,
type_abbr=type_abbr
or '',
defaults=defaults)
pu_list.append(
PartyUsage(text_unit_id=text_unit_id,
party=party,
count=count))
return ParseResults({PartyUsage: pu_list})
def extract_companies(self, text=None):
if not text:
text = self.text
return list(lex_entities.get_companies(text))
def getCompanies(self):
mem = []
companies = list(get_companies(self.bill_text))
for company in companies:
mem.append(str(company[0] + " " + str(company[1])))
self.bill.info['companies'] = mem