po/lyx_pot.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 # file lyx_pot.py
   5 # This file is part of LyX, the document processor.
   6 # Licence details can be found in the file COPYING.
   7 #
   8 # \author Bo Peng
   9 #
  10 # Full author contact details are available in file CREDITS
  11
  12 # Usage: use
  13 #     lyx_pot.py -h
  14 # to get usage message
  15
  16 # This script will extract translatable strings from input files and write
  17 # to output in gettext .pot format.
  18 #
  19 import sys, os, re, getopt
  20 if sys.version_info < (2, 4, 0):
  21     from sets import Set as set
  22
  23 def relativePath(path, base):
  24     '''return relative path from top source dir'''
  25     # full pathname of path
  26     path1 = os.path.normpath(os.path.realpath(path)).split(os.sep)
  27     path2 = os.path.normpath(os.path.realpath(base)).split(os.sep)
  28     if path1[:len(path2)] != path2:
  29         print "Path %s is not under top source directory" % path
  30     path3 = os.path.join(*path1[len(path2):]);
  31     # replace all \ by / such that we get the same comments on Windows and *nix
  32     path3 = path3.replace('\\', '/')
  33     return path3
  34
  35
  36 def writeString(outfile, infile, basefile, lineno, string):
  37     string = string.replace('\\', '\\\\').replace('"', '')
  38     if string == "":
  39         return
  40     print >> outfile, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
  41         (relativePath(infile, basefile), lineno, string)
  42
  43
  44 def ui_l10n(input_files, output, base):
  45     '''Generate pot file from lib/ui/*'''
  46     output = open(output, 'w')
  47     Submenu = re.compile(r'^[^#]*Submenu\s+"([^"]*)"')
  48     Popupmenu = re.compile(r'^[^#]*PopupMenu\s+"[^"]+"\s+"([^"]*)"')
  49     IconPalette = re.compile(r'^[^#]*IconPalette\s+"[^"]+"\s+"([^"]*)"')
  50     Toolbar = re.compile(r'^[^#]*Toolbar\s+"[^"]+"\s+"([^"]*)"')
  51     Item = re.compile(r'[^#]*Item\s+"([^"]*)"')
  52     TableInsert = re.compile(r'[^#]*TableInsert\s+"([^"]*)"')
  53     for src in input_files:
  54         input = open(src)
  55         for lineno, line in enumerate(input.readlines()):
  56             if Submenu.match(line):
  57                 (string,) = Submenu.match(line).groups()
  58                 string = string.replace('_', ' ')
  59             elif Popupmenu.match(line):
  60                 (string,) = Popupmenu.match(line).groups()
  61             elif IconPalette.match(line):
  62                 (string,) = IconPalette.match(line).groups()
  63             elif Toolbar.match(line):
  64                 (string,) = Toolbar.match(line).groups()
  65             elif Item.match(line):
  66                 (string,) = Item.match(line).groups()
  67             elif TableInsert.match(line):
  68                 (string,) = TableInsert.match(line).groups()
  69             else:
  70                 continue
  71             string = string.replace('"', '')
  72             if string != "":
  73                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
  74                     (relativePath(src, base), lineno+1, string)
  75         input.close()
  76     output.close()
  77
  78
  79 def layouts_l10n(input_files, output, base, layouttranslations):
  80     '''Generate pot file from lib/layouts/*.{layout,inc,module}'''
  81     Style = re.compile(r'^Style\s+(.*)', re.IGNORECASE)
  82     # include ???LabelString???, but exclude comment lines
  83     LabelString = re.compile(r'^[^#]*LabelString\S*\s+(.*)')
  84     GuiName = re.compile(r'\s*GuiName\s+(.*)')
  85     ListName = re.compile(r'\s*ListName\s+(.*)')
  86     CategoryName = re.compile(r'\s*Category\s+(.*)')
  87     NameRE = re.compile(r'DeclareLyXModule.*{(.*)}')
  88     InsetLayout = re.compile(r'^InsetLayout\s+\"?(.*)\"?')
  89     FlexCheck = re.compile(r'^Flex:(.*)')
  90     DescBegin = re.compile(r'#+\s*DescriptionBegin\s*$')
  91     DescEnd = re.compile(r'#+\s*DescriptionEnd\s*$')
  92     Category = re.compile(r'#Category: (.*)$')
  93     I18nPreamble = re.compile(r'\s*(Lang)|(Babel)Preamble\s*$')
  94     EndI18nPreamble = re.compile(r'\s*End(Lang)|(Babel)Preamble\s*$')
  95     I18nString = re.compile(r'_\(([^\)]+)\)')
  96     CounterFormat = re.compile(r'\s*PrettyFormat\s+"?(.*)"?')
  97     CiteFormat = re.compile(r'\s*CiteFormat')
  98     KeyVal = re.compile(r'^\s*_\w+\s+(.*)$')
  99     Float = re.compile(r'\s*Float')
 100     End = re.compile(r'\s*End')
 101     Comment = re.compile(r'\s*#')
 102     Translation = re.compile(r'\s*Translation\s+(.*)\s*$')
 103     KeyValPair = re.compile(r'\s*"(.*)"\s+"(.*)"')
 104
 105     oldlanguages = []
 106     languages = []
 107     keyset = set()
 108     oldtrans = dict()
 109     if layouttranslations:
 110         linguas_file = os.path.join(base, 'po/LINGUAS')
 111         for line in open(linguas_file).readlines():
 112             if Comment.search(line) == None:
 113                 languages.extend(line.split())
 114
 115         # read old translations if available
 116         try:
 117             input = open(output)
 118             lang = ''
 119             for line in input.readlines():
 120                 res = Comment.search(line)
 121                 if res:
 122                     continue
 123                 if line.strip() == '':
 124                     continue
 125                 res = Translation.search(line)
 126                 if res:
 127                     lang = res.group(1)
 128                     if lang not in languages:
 129                         oldlanguages.append(lang)
 130                         languages.append(lang)
 131                     oldtrans[lang] = dict()
 132                     continue
 133                 res = End.search(line)
 134                 if res:
 135                     lang = ''
 136                     continue
 137                 res = KeyValPair.search(line)
 138                 if res and lang != '':
 139                     key = res.group(1).decode('utf-8')
 140                     val = res.group(2).decode('utf-8')
 141                     key = key.replace('\\"', '"').replace('\\\\', '\\')
 142                     val = val.replace('\\"', '"').replace('\\\\', '\\')
 143                     oldtrans[lang][key] = val
 144                     keyset.add(key)
 145                     continue
 146                 print "Error: Unable to handle line:"
 147                 print line
 148         except IOError:
 149             print "Warning: Unable to open %s for reading." % output
 150             print "         Old translations will be lost."
 151
 152         # walon is not a known document language
 153         # FIXME: Do not hardcode, read from lib/languages!
 154         if 'wa' in languages:
 155             languages.remove('wa')
 156
 157     out = open(output, 'w')
 158     for src in input_files:
 159         readingDescription = False
 160         readingI18nPreamble = False
 161         readingFloat = False
 162         readingCiteFormats = False
 163         descStartLine = -1
 164         descLines = []
 165         lineno = 0
 166         for line in open(src).readlines():
 167             lineno += 1
 168             if readingDescription:
 169                 res = DescEnd.search(line)
 170                 if res != None:
 171                     readingDescription = False
 172                     desc = " ".join(descLines)
 173                     if not layouttranslations:
 174                         writeString(out, src, base, lineno + 1, desc)
 175                     continue
 176                 descLines.append(line[1:].strip())
 177                 continue
 178             res = DescBegin.search(line)
 179             if res != None:
 180                 readingDescription = True
 181                 descStartLine = lineno
 182                 continue
 183             if readingI18nPreamble:
 184                 res = EndI18nPreamble.search(line)
 185                 if res != None:
 186                     readingI18nPreamble = False
 187                     continue
 188                 res = I18nString.search(line)
 189                 if res != None:
 190                     string = res.group(1)
 191                     if layouttranslations:
 192                         keyset.add(string)
 193                     else:
 194                         writeString(out, src, base, lineno, string)
 195                 continue
 196             res = I18nPreamble.search(line)
 197             if res != None:
 198                 readingI18nPreamble = True
 199                 continue
 200             res = NameRE.search(line)
 201             if res != None:
 202                 string = res.group(1)
 203                 if not layouttranslations:
 204                     writeString(out, src, base, lineno + 1, string)
 205                 continue
 206             res = Style.search(line)
 207             if res != None:
 208                 string = res.group(1)
 209                 string = string.replace('_', ' ')
 210                 if not layouttranslations:
 211                     writeString(out, src, base, lineno, string)
 212                 continue
 213             res = LabelString.search(line)
 214             if res != None:
 215                 string = res.group(1)
 216                 if not layouttranslations:
 217                     writeString(out, src, base, lineno, string)
 218                 continue
 219             res = GuiName.search(line)
 220             if res != None:
 221                 string = res.group(1)
 222                 if layouttranslations:
 223                     # gui name must only be added for floats
 224                     if readingFloat:
 225                         keyset.add(string)
 226                 else:
 227                     writeString(out, src, base, lineno, string)
 228                 continue
 229             res = CategoryName.search(line)
 230             if res != None:
 231                 string = res.group(1)
 232                 if not layouttranslations:
 233                     writeString(out, src, base, lineno, string)
 234                 continue
 235             res = ListName.search(line)
 236             if res != None:
 237                 string = res.group(1)
 238                 if layouttranslations:
 239                     keyset.add(string.strip('"'))
 240                 else:
 241                     writeString(out, src, base, lineno, string)
 242                 continue
 243             res = InsetLayout.search(line)
 244             if res != None:
 245                 string = res.group(1)
 246                 string = string.replace('_', ' ')
 247                 #Flex:xxx is not used in translation
 248                 #if not layouttranslations:
 249                 #    writeString(out, src, base, lineno, string)
 250                 m = FlexCheck.search(string)
 251                 if m:
 252                     if not layouttranslations:
 253                         writeString(out, src, base, lineno, m.group(1))
 254                 continue
 255             res = Category.search(line)
 256             if res != None:
 257                 string = res.group(1)
 258                 if not layouttranslations:
 259                     writeString(out, src, base, lineno, string)
 260                 continue
 261             res = CounterFormat.search(line)
 262             if res != None:
 263                 string = res.group(1)
 264                 if not layouttranslations:
 265                     writeString(out, src, base, lineno, string)
 266                 continue
 267             res = Float.search(line)
 268             if res != None:
 269                 readingFloat = True
 270                 continue
 271             res = CiteFormat.search(line)
 272             if res != None:
 273                 readingCiteFormats = True
 274             res = End.search(line)
 275             if res != None:
 276                 readingCiteFormats = False
 277                 readingFloat = False
 278             if readingCiteFormats:
 279                 res = KeyVal.search(line)
 280                 if res != None:
 281                     val = res.group(1)
 282                     if not layouttranslations:
 283                         writeString(out, src, base, lineno, val)
 284
 285     if layouttranslations:
 286         # Extract translations of layout files
 287         import polib
 288
 289         # Sort languages and key to minimize the diff between different runs
 290         # with changed translations
 291         languages.sort()
 292         keys = []
 293         for key in keyset:
 294             keys.append(key)
 295         keys.sort()
 296
 297         print >> out, '''# This file has been automatically generated by po/lyx_pot.py.
 298 # PLEASE MODIFY ONLY THE LAGUAGES HAVING NO .po FILE! If you want to regenerate
 299 # this file from the translations, run `make ../lib/layouttranslations' in po.
 300 # Python polib library is needed for building the output file.
 301 #
 302 # This file should remain fixed during minor LyX releases.
 303 # For more comments see README.localization file.'''
 304         for lang in languages:
 305             print >> out, '\nTranslation %s' % lang
 306             if lang in oldtrans.keys():
 307                 trans = oldtrans[lang]
 308             else:
 309                 trans = dict()
 310             if not lang in oldlanguages:
 311                 poname = os.path.join(base, 'po/' + lang + '.po')
 312                 po = polib.pofile(poname)
 313                 # Iterate through po entries and not keys for speed reasons.
 314                 # FIXME: The code is still too slow
 315                 for entry in po:
 316                     if not entry.translated():
 317                         continue
 318                     if entry.msgid in keys:
 319                         key = entry.msgid
 320                         val = entry.msgstr
 321                         # some translators keep untranslated entries
 322                         if val != key:
 323                             trans[key] = val
 324             for key in keys:
 325                 if key in trans.keys():
 326                     val = trans[key].replace('\\', '\\\\').replace('"', '\\"')
 327                     key = key.replace('\\', '\\\\').replace('"', '\\"')
 328                     print >> out, '\t"%s" "%s"' % \
 329                              (key.encode('utf-8'), val.encode('utf-8'))
 330                 # also print untranslated entries to help translators
 331                 elif not lang in oldlanguages:
 332                     key = key.replace('\\', '\\\\').replace('"', '\\"')
 333                     print >> out, '\t"%s" "%s"' % \
 334                              (key.encode('utf-8'), key.encode('utf-8'))
 335             print >> out, 'End'
 336
 337     out.close()
 338
 339
 340 def qt4_l10n(input_files, output, base):
 341     '''Generate pot file from src/frontends/qt4/ui/*.ui'''
 342     output = open(output, 'w')
 343     pat = re.compile(r'\s*<string>(.*)</string>')
 344     prop = re.compile(r'\s*<property.*name.*=.*shortcut')
 345     for src in input_files:
 346         input = open(src)
 347         skipNextLine = False
 348         for lineno, line in enumerate(input.readlines()):
 349             # skip the line after <property name=shortcut>
 350             if skipNextLine:
 351                 skipNextLine = False
 352                 continue
 353             if prop.match(line):
 354                 skipNextLine = True
 355                 continue
 356             # get lines that match <string>...</string>
 357             if pat.match(line):
 358                 (string,) = pat.match(line).groups()
 359                 string = string.replace('&amp;', '&').replace('&quot;', '"')
 360                 string = string.replace('&lt;', '<').replace('&gt;', '>')
 361                 string = string.replace('\\', '\\\\').replace('"', r'\"')
 362                 string = string.replace('&#x0a;', r'\n')
 363                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 364                     (relativePath(src, base), lineno+1, string)
 365         input.close()
 366     output.close()
 367
 368
 369 def languages_l10n(input_files, output, base):
 370     '''Generate pot file from lib/languages'''
 371     out = open(output, 'w')
 372     GuiName = re.compile(r'^[^#]*GuiName\s+(.*)')
 373
 374     for src in input_files:
 375         descStartLine = -1
 376         descLines = []
 377         lineno = 0
 378         for line in open(src).readlines():
 379             lineno += 1
 380             res = GuiName.search(line)
 381             if res != None:
 382                 string = res.group(1)
 383                 writeString(out, src, base, lineno, string)
 384                 continue
 385
 386     out.close()
 387
 388
 389 def external_l10n(input_files, output, base):
 390     '''Generate pot file from lib/external_templates'''
 391     output = open(output, 'w')
 392     Template = re.compile(r'^Template\s+(.*)')
 393     GuiName = re.compile(r'\s*GuiName\s+(.*)')
 394     HelpTextStart = re.compile(r'\s*HelpText\s')
 395     HelpTextSection = re.compile(r'\s*(\S.*)\s*$')
 396     HelpTextEnd = re.compile(r'\s*HelpTextEnd\s')
 397     i = -1
 398     for src in input_files:
 399         input = open(src)
 400         inHelp = False
 401         hadHelp = False
 402         prev_help_string = ''
 403         for lineno, line in enumerate(input.readlines()):
 404             if Template.match(line):
 405                 (string,) = Template.match(line).groups()
 406             elif GuiName.match(line):
 407                 (string,) = GuiName.match(line).groups()
 408             elif inHelp:
 409                 if HelpTextEnd.match(line):
 410                     if hadHelp:
 411                         print >> output, '\nmsgstr ""\n'
 412                     inHelp = False
 413                     hadHelp = False
 414                     prev_help_string = ''
 415                 elif HelpTextSection.match(line):
 416                     (help_string,) = HelpTextSection.match(line).groups()
 417                     help_string = help_string.replace('"', '')
 418                     if help_string != "" and prev_help_string == '':
 419                         print >> output, '#: %s:%d\nmsgid ""\n"%s\\n"' % \
 420                             (relativePath(src, base), lineno+1, help_string)
 421                         hadHelp = True
 422                     elif help_string != "":
 423                         print >> output, '"%s\\n"' % help_string
 424                     prev_help_string = help_string
 425             elif HelpTextStart.match(line):
 426                 inHelp = True
 427                 prev_help_string = ''
 428             else:
 429                 continue
 430             string = string.replace('"', '')
 431             if string != "" and not inHelp:
 432                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 433                     (relativePath(src, base), lineno+1, string)
 434         input.close()
 435     output.close()
 436
 437
 438 def formats_l10n(input_files, output, base):
 439     '''Generate pot file from configure.py'''
 440     output = open(output, 'w')
 441     GuiName = re.compile(r'.*\Format\s+\S+\s+\S+\s+"([^"]*)"\s+(\S*)\s+.*')
 442     GuiName2 = re.compile(r'.*\Format\s+\S+\s+\S+\s+([^"]\S+)\s+(\S*)\s+.*')
 443     input = open(input_files[0])
 444     for lineno, line in enumerate(input.readlines()):
 445         label = ""
 446         labelsc = ""
 447         if GuiName.match(line):
 448             label = GuiName.match(line).group(1)
 449             shortcut = GuiName.match(line).group(2).replace('"', '')
 450         elif GuiName2.match(line):
 451             label = GuiName2.match(line).group(1)
 452             shortcut = GuiName2.match(line).group(2).replace('"', '')
 453         else:
 454             continue
 455         label = label.replace('\\', '\\\\').replace('"', '')
 456         if shortcut != "":
 457             labelsc = label + "|" + shortcut
 458         if label != "":
 459             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 460                 (relativePath(input_files[0], base), lineno+1, label)
 461         if labelsc != "":
 462             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 463                 (relativePath(input_files[0], base), lineno+1, labelsc)
 464     input.close()
 465     output.close()
 466
 467
 468 def encodings_l10n(input_files, output, base):
 469     '''Generate pot file from lib/encodings'''
 470     output = open(output, 'w')
 471     # assuming only one encodings file
 472     #                 Encoding utf8      utf8    "Unicode (utf8)" UTF-8    variable inputenc
 473     reg = re.compile('Encoding [\w-]+\s+[\w-]+\s+"([\w \-\(\)]+)"\s+[\w-]+\s+(fixed|variable)\s+\w+.*')
 474     input = open(input_files[0])
 475     for lineno, line in enumerate(input.readlines()):
 476         if not line.startswith('Encoding'):
 477             continue
 478         if reg.match(line):
 479             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 480                 (relativePath(input_files[0], base), lineno+1, reg.match(line).groups()[0])
 481         else:
 482             print "Error: Unable to handle line:"
 483             print line
 484             # No need to abort if the parsing fails
 485             # sys.exit(1)
 486     input.close()
 487     output.close()
 488
 489
 490
 491 Usage = '''
 492 lyx_pot.py [-b|--base top_src_dir] [-o|--output output_file] [-h|--help] [-s|src_file filename] -t|--type input_type input_files
 493
 494 where
 495     --base:
 496         path to the top source directory. default to '.'
 497     --output:
 498         output pot file, default to './lyx.pot'
 499     --src_file
 500         filename that contains a list of input files in each line
 501     --input_type can be
 502         ui: lib/ui/*
 503         layouts: lib/layouts/*
 504         layouttranslations: create lib/layouttranslations from po/*.po and lib/layouts/*
 505         qt4: qt4 ui files
 506         languages: file lib/languages
 507         encodings: file lib/encodings
 508         external: external templates file
 509         formats: formats predefined in lib/configure.py
 510 '''
 511
 512 if __name__ == '__main__':
 513     input_type = None
 514     output = 'lyx.pot'
 515     base = '.'
 516     input_files = []
 517     #
 518     optlist, args = getopt.getopt(sys.argv[1:], 'ht:o:b:s:',
 519         ['help', 'type=', 'output=', 'base=', 'src_file='])
 520     for (opt, value) in optlist:
 521         if opt in ['-h', '--help']:
 522             print Usage
 523             sys.exit(0)
 524         elif opt in ['-o', '--output']:
 525             output = value
 526         elif opt in ['-b', '--base']:
 527             base = value
 528         elif opt in ['-t', '--type']:
 529             input_type = value
 530         elif opt in ['-s', '--src_file']:
 531             input_files = [f.strip() for f in open(value)]
 532
 533     if input_type not in ['ui', 'layouts', 'layouttranslations', 'qt4', 'languages', 'encodings', 'external', 'formats'] or output is None:
 534         print 'Wrong input type or output filename.'
 535         sys.exit(1)
 536
 537     input_files += args
 538
 539     if input_type == 'ui':
 540         ui_l10n(input_files, output, base)
 541     elif input_type == 'layouts':
 542         layouts_l10n(input_files, output, base, False)
 543     elif input_type == 'layouttranslations':
 544         layouts_l10n(input_files, output, base, True)
 545     elif input_type == 'qt4':
 546         qt4_l10n(input_files, output, base)
 547     elif input_type == 'external':
 548         external_l10n(input_files, output, base)
 549     elif input_type == 'formats':
 550         formats_l10n(input_files, output, base)
 551     elif input_type == 'encodings':
 552         encodings_l10n(input_files, output, base)
 553     else:
 554         languages_l10n(input_files, output, base)
 555
 556