日付の構造化

transduce_date.py

import re
import unicodedata
from transduce_jpnumber import kansuji_to_arabic_numeral
_date_re = re.compile('((?P<year>\d{4})[-/年](?P<ysuf>[^0-9\-/年月日\s]+)?)?((?P<month>\d{1,2})[-/月](?P<msuf>[^0-9\-/年月日\s]+)?)?((?P<date>\d{1,2})[日]?(?P<dsuf>[^0-9\-/年月日\s]+)?)?')

def transduce_jpdate(datestr):
    datestr_n = kansuji_to_arabic_numeral(datestr)
    datestr_n = unicodedata.normalize('NFKC', datestr_n)
    m = _date_re.match(datestr_n.strip())
    result = {}
    if m is not None:
        return {'year': m.group('year'), 
              'month': m.group('month'), 
              'date': m.group('date'),
              'year_suffix': m.group('ysuf'), 
              'month_suffix': m.group('msuf'), 
              'date_suffix': m.group('dsuf')}
    return result
transduce_jpdate('2018年3月上旬')