0_tools/po-merge.py

   1 #!/usr/bin/env python
   2
   3 import os, sys
   4
   5 def parse_translation(f):
   6     """Read a single translation entry from the file F and return a
   7     tuple with the comments, msgid and msgstr.  The comments is returned
   8     as a list of lines which do not end in new-lines.  The msgid and
   9     msgstr are strings, possibly with embedded newlines"""
  10     line = f.readline()
  11
  12     # Parse comments
  13     comments = []
  14     while 1:
  15         if line.strip() == '':
  16             return comments, None, None
  17         elif line[0] == '#':
  18             comments.append(line[:-1])
  19         else:
  20             break
  21         line = f.readline()
  22
  23     # Parse msgid
  24     if line[:7] != 'msgid "' or line[-2] != '"':
  25         raise RuntimeError("parse error")
  26     msgid = line[6:-1]
  27     while 1:
  28         line = f.readline()
  29         if line[0] != '"':
  30             break
  31         msgid += '\n' + line[:-1]
  32
  33     # Parse msgstr
  34     if line[:8] != 'msgstr "' or line[-2] != '"':
  35         raise RuntimeError("parse error")
  36     msgstr = line[7:-1]
  37     while 1:
  38         line = f.readline()
  39         if len(line) == 0 or line[0] != '"':
  40             break
  41         msgstr += '\n' + line[:-1]
  42
  43     if line.strip() != '':
  44         raise RuntimeError("parse error")
  45
  46     return comments, msgid, msgstr
  47
  48 def split_comments(comments):
  49     """Split COMMENTS into flag comments and other comments.  Flag
  50     comments are those that begin with '#,', e.g. '#,fuzzy'."""
  51     flags = []
  52     other = []
  53     for c in comments:
  54         if len(c) > 1 and c[1] == ',':
  55             flags.append(c)
  56         else:
  57             other.append(c)
  58     return flags, other
  59
  60 def main(argv):
  61     if len(argv) != 2:
  62         argv0 = os.path.basename(argv[0])
  63         sys.exit('Usage: %s <lang.po>\n'
  64                  '\n'
  65                  'This script will replace the translations and flags in lang.po with\n'
  66                  'the translations and flags in the source po file read from standard\n'
  67                  'input.  Strings that are not found in the source file are left untouched.\n'
  68                  'A backup copy of lang.po is saved as lang.po.bak.\n'
  69                  '\n'
  70                  'Example:\n'
  71                  '    svn cat http://svn.collab.net/repos/svn/trunk/subversion/po/sv.po | \\\n'
  72                  '        %s sv.po' % (argv0, argv0))
  73
  74     # Read the source po file into a hash
  75     source = {}
  76     while 1:
  77         comments, msgid, msgstr = parse_translation(sys.stdin)
  78         if not comments and msgid is None:
  79             break
  80         if msgid is not None:
  81             source[msgid] = msgstr, split_comments(comments)[0]
  82
  83     # Make a backup of the output file, open the copy for reading
  84     # and the original for writing.
  85     os.rename(argv[1], argv[1] + '.bak')
  86     infile = open(argv[1] + '.bak')
  87     outfile = open(argv[1], 'w')
  88
  89     # Loop thought the original and replace stuff as we go
  90     first = 1
  91     string_count = 0
  92     update_count = 0
  93     untranslated = 0
  94     while 1:
  95         comments, msgid, msgstr = parse_translation(infile)
  96         if not comments and msgid is None:
  97             break
  98         if not first:
  99             outfile.write('\n')
 100         first = 0
 101         if msgid is None:
 102             outfile.write('\n'.join(comments) + '\n')
 103         else:
 104             string_count += 1
 105             # Do not update the header, and only update if the source
 106             # has a non-empty translation.
 107             if msgid != '""' and source.get(msgid, ['""', []])[0] != '""':
 108                 other = split_comments(comments)[1]
 109                 new_msgstr, new_flags = source[msgid]
 110                 new_comments = other + new_flags
 111                 if new_msgstr != msgstr or new_comments != comments:
 112                     update_count += 1
 113                     msgstr = new_msgstr
 114                     comments = new_comments
 115             outfile.write('\n'.join(comments) + '\n')
 116             outfile.write('msgid ' + msgid + '\n')
 117             outfile.write('msgstr ' + msgstr + '\n')
 118         if msgstr == '""':
 119             untranslated += 1
 120
 121     # We're done.  Tell the user what we did.
 122     print('%d strings updated. '
 123           '%d of %d strings are still untranslated (%.0f%%).' %
 124           (update_count, untranslated, string_count,
 125            100.0 * untranslated / string_count))
 126
 127 if __name__ == '__main__':
 128     main(sys.argv)