def test_date(): assert get_date("le 12 janvier 2013 !") == [(3, 18, 'DATE_1')] assert get_date("le 12/01/2016 !") == [(3, 13, 'DATE_1')] assert get_date("le 12 / 01/2016 !") == [(3, 15, 'DATE_1')] assert get_date("le 12 / 01/16 !") == [(3, 13, 'DATE_1')] assert get_date("ARRÊT DU HUIT FÉVRIER DEUX MILLE TREIZE") == [(9, 39, 'DATE_1')] assert get_date("le 1er janvier 2013 !") == [(3, 19, 'DATE_1')] assert get_date("le 552-4-1 !") == []
rg_matcher = MatchRg(case_id=previous_case_id) for current_paragraph, current_xml_offset in zip( current_case_paragraphs, current_case_offsets): match_from_headers = headers_matcher.get_matched_entities( current_paragraph) company_names_offset = get_company_names(current_paragraph) # full_name_pp = current_doc_extend_pp_name_pattern.get_extended_names(text=current_paragraph) # partie_pp = get_partie_pers(current_paragraph) # judge_names = get_judge_name(current_paragraph) # clerk_names = get_clerk_name(current_paragraph) # lawyer_names = get_lawyer_name(current_paragraph) # addresses = get_addresses(current_paragraph) # court_name = get_juridictions(current_paragraph) case_dates = get_date(current_paragraph) # rg_from_regex = get_rg_from_regex(text=current_paragraph) # bar = get_bar(current_paragraph) # postal_code_matches = postal_code_city_matcher.get_matches(text=current_paragraph) # court_names_matches = court_names_matcher.get_matches(text=current_paragraph) # # frequent_entities = frequent_entities_matcher.get_matches(text=current_paragraph) # licence_plate_number = get_licence_plate(text=current_paragraph) # phone_numbers = get_phone_number(text=current_paragraph) all_matches = ( match_from_headers + current_xml_offset + company_names_offset + # full_name_pp + # judge_names + # clerk_names + # lawyer_names +