- there is no ftp.task.gda.pl/pub/mozilla/js

[packages/rpm-build-tools.git] / spec_utf8

#!/usr/bin/python

import os, os.path, re, sys, locale

langs={
    'bg':'windows-1251',
    'br':'iso8859-1',
    'ca':'iso8859-1',
    'cs':'iso8859-2',
    'da':'iso8859-1',
    'de':'iso8859-1',
    'en':'iso8859-1',
    'eo':'iso8859-3',
    'es':'iso8859-1',
    'fi':'iso8859-1',
    'fo':'iso8859-1',
    'fr':'iso8859-1',
    'gl':'iso8859-1',
    'he':'iso8859-8',
    'id':'iso8859-1',
    'is':'iso8859-1',
    'it':'iso8859-1',
    'ja':'euc-jp',
    'ko':'euc-kr',
    'nb':'iso8859-1',
    'nl':'iso8859-1', 
    'pl':'iso8859-2',
    'pt':'iso8859-1',
    'pt_BR':'iso8859-1',
    'ro':'iso8859-2',
    'ru':'KOI8-R',
    'se':'UTF-8',
    'sk':'iso8859-2',
    'sl':'iso8859-2',
    'sv':'iso8859-1',
    'tr':'iso8859-9',
    'uk':'KOI8-U',
    'wa':'iso8859-1',
    'zh_CN':'GB2312',
    'zh_HK':'BIG5-HKSCS',
    'zh_TW':'BIG5',
    0:0}

def find_encoding(lang):
  r = re.match("^([^.]+)(\.[^@]+)?$", lang)
  pure_lang = r.group(1)
  if r.group(2) == None:
    try:
      enc = langs[lang]
    except KeyError:
      enc = None
  else:
    # strip dot
    enc = r.group(2)[1:]
  return (enc, pure_lang)

def parse_spec(infile, outfile):
  re_summary = re.compile("^Summary\(([^\)]+)\):\t+(.*)$")
  re_utf = re.compile("^utf-8$", re.I)
  re_desc = re.compile("^(%description.*\s)-l\s+([\S]+)($|\s.*$)")
  re_proc = re.compile("^%")
  in_desc = False

  for l in infile:
      outline = l
      r = re_summary.match(l)
      if r:
        (enc, pure_lang) = find_encoding(r.group(1))
        if enc == None:
          outfile.write("#unknow lang code Summary(%s)\n" % (lang))
        elif not re_utf.search(enc):
          try:
            desc = unicode(r.group(2), enc).encode("UTF-8")
            l = "Summary(%s.UTF-8):   %s\n" % (pure_lang, desc)
          except UnicodeDecodeError:
            outfile.write("#transcoding error Summary(%s)\n" % (lang))
      if in_desc:
        if re_proc.search(l):
          in_desc = False
        else:
          if not re_utf.search(enc):
            try:
              l = unicode(l, enc).encode("UTF-8")
            except UnicodeDecodeError:
              outfile.write("#transcoding error %%description -l %s\n" % (lang))

      r = re_desc.match(l)
      if r:
        (enc, pure_lang) = find_encoding(r.group(2))
        if enc == None:
          outfile.write("#unknown lang code in %%description -l %s\n" % (lang))
        elif not re_utf.search(enc):
          in_desc = True
          l = "%s-l %s.UTF-8%s\n" % (r.group(1), pure_lang, r.group(3))

      outfile.write(l)

def main():
  parse_spec(sys.stdin, sys.stdout)

if __name__ == "__main__":
  main()