Localizable.stringsを CSV に変換する Python スクリプト

$ python ls2csv.py tols data.csv

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys
import os
import re

def ls2csv(ls_dir):
  langs = []
  section_data = {'No Section':{}}
  all_data = []
  
  # get lang list
  res_dir = os.getcwd()
  if not os.path.exists(ls_dir):
    print "指定されたディレクトリ(%s)は存在しません。" % (ls_dir)
    sys.exit(0)
  res_dir = os.path.join(res_dir, ls_dir)
  for (root, dirs, files) in os.walk(res_dir):
    for filename in dirs:
      (basename, ext) = os.path.splitext(filename)
      if ext.lower() == '.lproj':
        langs.append(basename)
  
  # get key list
  key_val_regex = re.compile('"(?P.+)"s+=s+"(?P.+)";')
  section_regex = re.compile('//s+(?P.+)')
  first_lang = True
  current_section_name = None
  for lang in langs:
    lang_res_file = os.path.join(res_dir, '%s.lproj/Localizable.strings' % (lang,))
    content = open(lang_res_file, 'r').read()
    for line in content.split('n'):
      m = section_regex.match(line)
      if m:
        current_section_name = m.group('section_name')
        if first_lang:
          if current_section not in section_data.keys():
            section_data[current_section_name] = {}
            all_data.append({'section_name':current_section_name, 'section_data':section_data[current_section_name]})
      if not current_section_name or current_section_name not in section_data.keys():
        current_section_name = 'No Section'
      current_section = section_data[current_section_name]
      m = key_val_regex.match(line)
      if m:
        key = m.group('key')
        val = m.group('val')
        found = False
        if key not in current_section:
          for section_name in section_data.keys():
            section = section_data[section_name]
            if key in section:
              section[key][lang] = val
              found = True
              break
        if not found:
          if key not in current_section:
            current_section[key] = {}
          current_section[key][lang] = val
  
    first_lang = False
  all_data.append({'section_name':'No Section', 'section_data':section_data['No Section']})
  
  # output csv
  csv = 'key,'
  for i, lang in enumerate(langs):
    csv += lang
    if i < len(langs) - 1:
      csv += ','
  csv += 'n'
  for section in all_data:
    csv += '// %sn' % (section['section_name'])
    for key in section['section_data'].keys():
      csv += '%s,' % (key,)
      for i, lang in enumerate(langs):
        try:
          val = section['section_data'][key][lang]
        except:
          val = ''
        csv += val
        if i < len(langs) - 1:
          csv += ','
      csv += 'n'
    csv += 'n'
  print csv

def csv2ls(csv_file):
  csv_data = open(csv_file, 'r').read().split('n')

  head_line = csv_data[0]
  langs = [lang.replace('"', '') for lang in  head_line.split(',')[1:]]
  all_data = {}
  
  for lang in langs:
    all_data[lang] = []

  # get all sections
  current_section_name = None
  for line in csv_data[1:]:
    params = line.split(',')
    if len(params) != len(langs) + 1:
      current_section_name = line
      for lang in langs:
        all_data[lang].append({'section_name':current_section_name, 'section_data':{}})
    else:
      key = params[0]
      for i, lang in enumerate(langs):
        val = params[i + 1]
        section_data = all_data[lang][-1]['section_data']
        section_data[key] = val

  # output
  for lang_name in all_data:
    lang_name = lang_name.replace('"', '')
    lang_data = all_data[lang_name]
    result = ''
    for section in lang_data:
      result += '%sn' % (section['section_name'].replace('"', ''))
      for key, value in section['section_data'].items():
        key = key.replace('"', '')
        value = value.replace('"', '')
        result += '"%(key)s" = "%(val)s";n' % {'key':key, 'val':value}

    f = open('%s.strings' % (lang_name,), 'w')
    f.write(result)
    f.close()
    

def printUsageAndExit():
    print '''Usage:
    ls2csv.py tocsv [target_dir]
    ls2csv.py tols [csv_file]
'''
    sys.exit(0)

if __name__ == '__main__':
  if len(sys.argv) < 3:
    printUsageAndExit()
  command = sys.argv[1]
  if command == 'tocsv':
    ls2csv(sys.argv[2])
  elif command == 'tols':
    csv2ls(sys.argv[2])
  else:
    printUsageAndExit()