Tools/i18n/msgfmt.py

   1 #! /usr/bin/env python
   2 # -*- coding: iso-8859-1 -*-
   3 # Written by Martin v. Löwis <loewis@informatik.hu-berlin.de>
   4
   5 """Generate binary message catalog from textual translation description.
   6
   7 This program converts a textual Uniforum-style message catalog (.po file) into
   8 a binary GNU catalog (.mo file).  This is essentially the same function as the
   9 GNU msgfmt program, however, it is a simpler implementation.
  10
  11 Usage: msgfmt.py [OPTIONS] filename.po
  12
  13 Options:
  14     -o file
  15     --output-file=file
  16         Specify the output file to write to.  If omitted, output will go to a
  17         file named filename.mo (based off the input file name).
  18
  19     -h
  20     --help
  21         Print this message and exit.
  22
  23     -V
  24     --version
  25         Display version information and exit.
  26 """
  27
  28 import sys
  29 import os
  30 import getopt
  31 import struct
  32 import array
  33
  34 __version__ = "1.1"
  35
  36 MESSAGES = {}
  37
  38
  39 \f
  40 def usage(code, msg=''):
  41     print >> sys.stderr, __doc__
  42     if msg:
  43         print >> sys.stderr, msg
  44     sys.exit(code)
  45
  46
  47 \f
  48 def add(id, str, fuzzy):
  49     "Add a non-fuzzy translation to the dictionary."
  50     global MESSAGES
  51     if not fuzzy and str:
  52         MESSAGES[id] = str
  53
  54
  55 \f
  56 def generate():
  57     "Return the generated output."
  58     global MESSAGES
  59     keys = MESSAGES.keys()
  60     # the keys are sorted in the .mo file
  61     keys.sort()
  62     offsets = []
  63     ids = strs = ''
  64     for id in keys:
  65         # For each string, we need size and file offset.  Each string is NUL
  66         # terminated; the NUL does not count into the size.
  67         offsets.append((len(ids), len(id), len(strs), len(MESSAGES[id])))
  68         ids += id + '\0'
  69         strs += MESSAGES[id] + '\0'
  70     output = ''
  71     # The header is 7 32-bit unsigned integers.  We don't use hash tables, so
  72     # the keys start right after the index tables.
  73     # translated string.
  74     keystart = 7*4+16*len(keys)
  75     # and the values start after the keys
  76     valuestart = keystart + len(ids)
  77     koffsets = []
  78     voffsets = []
  79     # The string table first has the list of keys, then the list of values.
  80     # Each entry has first the size of the string, then the file offset.
  81     for o1, l1, o2, l2 in offsets:
  82         koffsets += [l1, o1+keystart]
  83         voffsets += [l2, o2+valuestart]
  84     offsets = koffsets + voffsets
  85     output = struct.pack("Iiiiiii",
  86                          0x950412deL,       # Magic
  87                          0,                 # Version
  88                          len(keys),         # # of entries
  89                          7*4,               # start of key index
  90                          7*4+len(keys)*8,   # start of value index
  91                          0, 0)              # size and offset of hash table
  92     output += array.array("i", offsets).tostring()
  93     output += ids
  94     output += strs
  95     return output
  96
  97
  98 \f
  99 def make(filename, outfile):
 100     ID = 1
 101     STR = 2
 102
 103     # Compute .mo name from .po name and arguments
 104     if filename.endswith('.po'):
 105         infile = filename
 106     else:
 107         infile = filename + '.po'
 108     if outfile is None:
 109         outfile = os.path.splitext(infile)[0] + '.mo'
 110
 111     try:
 112         lines = open(infile).readlines()
 113     except IOError, msg:
 114         print >> sys.stderr, msg
 115         sys.exit(1)
 116
 117     section = None
 118     fuzzy = 0
 119
 120     # Parse the catalog
 121     lno = 0
 122     for l in lines:
 123         lno += 1
 124         # If we get a comment line after a msgstr, this is a new entry
 125         if l[0] == '#' and section == STR:
 126             add(msgid, msgstr, fuzzy)
 127             section = None
 128             fuzzy = 0
 129         # Record a fuzzy mark
 130         if l[:2] == '#,' and 'fuzzy' in l:
 131             fuzzy = 1
 132         # Skip comments
 133         if l[0] == '#':
 134             continue
 135         # Now we are in a msgid section, output previous section
 136         if l.startswith('msgid'):
 137             if section == STR:
 138                 add(msgid, msgstr, fuzzy)
 139             section = ID
 140             l = l[5:]
 141             msgid = msgstr = ''
 142         # Now we are in a msgstr section
 143         elif l.startswith('msgstr'):
 144             section = STR
 145             l = l[6:]
 146         # Skip empty lines
 147         l = l.strip()
 148         if not l:
 149             continue
 150         # XXX: Does this always follow Python escape semantics?
 151         l = eval(l)
 152         if section == ID:
 153             msgid += l
 154         elif section == STR:
 155             msgstr += l
 156         else:
 157             print >> sys.stderr, 'Syntax error on %s:%d' % (infile, lno), \
 158                   'before:'
 159             print >> sys.stderr, l
 160             sys.exit(1)
 161     # Add last entry
 162     if section == STR:
 163         add(msgid, msgstr, fuzzy)
 164
 165     # Compute output
 166     output = generate()
 167
 168     try:
 169         open(outfile,"wb").write(output)
 170     except IOError,msg:
 171         print >> sys.stderr, msg
 172
 173
 174 \f
 175 def main():
 176     try:
 177         opts, args = getopt.getopt(sys.argv[1:], 'hVo:',
 178                                    ['help', 'version', 'output-file='])
 179     except getopt.error, msg:
 180         usage(1, msg)
 181
 182     outfile = None
 183     # parse options
 184     for opt, arg in opts:
 185         if opt in ('-h', '--help'):
 186             usage(0)
 187         elif opt in ('-V', '--version'):
 188             print >> sys.stderr, "msgfmt.py", __version__
 189             sys.exit(0)
 190         elif opt in ('-o', '--output-file'):
 191             outfile = arg
 192     # do it
 193     if not args:
 194         print >> sys.stderr, 'No input file given'
 195         print >> sys.stderr, "Try `msgfmt --help' for more information."
 196         return
 197
 198     for filename in args:
 199         make(filename, outfile)
 200
 201
 202 if __name__ == '__main__':
 203     main()