/ svn-cassiopeia --username yuuma.tomita@gmail.com/cas_helper/xldb/utils/parsers/kansuji2arabic.py

#!/usr/bin/env python

# -*- coding: utf-8 -*-



import re



def kansuji2arabic(text):

  """???????????????"""



  KANNUM_PATTERN = re.compile(u'(?P<kansuji>[????????????????????????????????,\d]+)')



  index = 0

  while index < len(text):

      matched = KANNUM_PATTERN.search(text[index:])

      if matched:

          kansuji = matched.group('kansuji')

          startindex = matched.start('kansuji') + index

          endindex = matched.end('kansuji') + index

          result = 0

          digit = 1

          numgroup = 1

          kanindex = len(kansuji)

          while kanindex > 0:

              c = kansuji[(kanindex - 1):kanindex]

              kanindex -= 1

              if c == u'??0':

                  digit *= 10

              elif c in u'??':

                  digit = 10

              elif c == u'?':

                  digit = 100

              elif c == u'?':

                  digit = 1000

              elif c in u'??':

                  numgroup = 10000

                  digit = 1

              elif c in u'?':

                  numgroup = 10000*10000

                  digit = 1

              elif c in u'?':

                  numgroup = 10000*10000*10000

                  digit = 1

              elif c in u'?,':

                  pass

              else:

                  if c in u'???1':

                      result += digit * numgroup

                  elif c in u'???2':

                      result += 2 * digit * numgroup

                  elif c in u'???3':

                      result += 3 * digit * numgroup

                  elif c in u'??4':

                      result += 4 * digit * numgroup

                  elif c in u'??5':

                      result += 5 * digit * numgroup

                  elif c in u'??6':

                      result += 6 * digit * numgroup

                  elif c in u'??7':

                      result += 7 * digit * numgroup

                  elif c in u'??8':

                      result += 8 * digit * numgroup

                  elif c in u'??9':

                      result += 9 * digit * numgroup

                  digit *= 10

          text = u'%s%d%s' % (text[:startindex], result, text[endindex:])

          index = startindex + len('%d' % (result,))

      else:

          break

  return text



k2a = lambda str: kansuji2arabic(unicode(str)) #extended shortcut