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             pass
 150
 151         # walon is not a known document language
 152         # FIXME: Do not hardcode, read from lib/languages!
 153         if 'wa' in languages:
 154             languages.remove('wa')
 155
 156     out = open(output, 'w')
 157     for src in input_files:
 158         readingDescription = False
 159         readingI18nPreamble = False
 160         readingFloat = False
 161         readingCiteFormats = False
 162         descStartLine = -1
 163         descLines = []
 164         lineno = 0
 165         for line in open(src).readlines():
 166             lineno += 1
 167             if readingDescription:
 168                 res = DescEnd.search(line)
 169                 if res != None:
 170                     readingDescription = False
 171                     desc = " ".join(descLines)
 172                     if not layouttranslations:
 173                         writeString(out, src, base, lineno + 1, desc)
 174                     continue
 175                 descLines.append(line[1:].strip())
 176                 continue
 177             res = DescBegin.search(line)
 178             if res != None:
 179                 readingDescription = True
 180                 descStartLine = lineno
 181                 continue
 182             if readingI18nPreamble:
 183                 res = EndI18nPreamble.search(line)
 184                 if res != None:
 185                     readingI18nPreamble = False
 186                     continue
 187                 res = I18nString.search(line)
 188                 if res != None:
 189                     string = res.group(1)
 190                     if layouttranslations:
 191                         keyset.add(string)
 192                     else:
 193                         writeString(out, src, base, lineno, string)
 194                 continue
 195             res = I18nPreamble.search(line)
 196             if res != None:
 197                 readingI18nPreamble = True
 198                 continue
 199             res = NameRE.search(line)
 200             if res != None:
 201                 string = res.group(1)
 202                 if not layouttranslations:
 203                     writeString(out, src, base, lineno + 1, string)
 204                 continue
 205             res = Style.search(line)
 206             if res != None:
 207                 string = res.group(1)
 208                 string = string.replace('_', ' ')
 209                 if not layouttranslations:
 210                     writeString(out, src, base, lineno, string)
 211                 continue
 212             res = LabelString.search(line)
 213             if res != None:
 214                 string = res.group(1)
 215                 if not layouttranslations:
 216                     writeString(out, src, base, lineno, string)
 217                 continue
 218             res = GuiName.search(line)
 219             if res != None:
 220                 string = res.group(1)
 221                 if layouttranslations:
 222                     # gui name must only be added for floats
 223                     if readingFloat:
 224                         keyset.add(string)
 225                 else:
 226                     writeString(out, src, base, lineno, string)
 227                 continue
 228             res = CategoryName.search(line)
 229             if res != None:
 230                 string = res.group(1)
 231                 if not layouttranslations:
 232                     writeString(out, src, base, lineno, string)
 233                 continue
 234             res = ListName.search(line)
 235             if res != None:
 236                 string = res.group(1)
 237                 if layouttranslations:
 238                     keyset.add(string.strip('"'))
 239                 else:
 240                     writeString(out, src, base, lineno, string)
 241                 continue
 242             res = InsetLayout.search(line)
 243             if res != None:
 244                 string = res.group(1)
 245                 string = string.replace('_', ' ')
 246                 #Flex:xxx is not used in translation
 247                 #if not layouttranslations:
 248                 #    writeString(out, src, base, lineno, string)
 249                 m = FlexCheck.search(string)
 250                 if m:
 251                     if not layouttranslations:
 252                         writeString(out, src, base, lineno, m.group(1))
 253                 continue
 254             res = Category.search(line)
 255             if res != None:
 256                 string = res.group(1)
 257                 if not layouttranslations:
 258                     writeString(out, src, base, lineno, string)
 259                 continue
 260             res = CounterFormat.search(line)
 261             if res != None:
 262                 string = res.group(1)
 263                 if not layouttranslations:
 264                     writeString(out, src, base, lineno, string)
 265                 continue
 266             res = Float.search(line)
 267             if res != None:
 268                 readingFloat = True
 269                 continue
 270             res = CiteFormat.search(line)
 271             if res != None:
 272                 readingCiteFormats = True
 273             res = End.search(line)
 274             if res != None:
 275                 readingCiteFormats = False
 276                 readingFloat = False
 277             if readingCiteFormats:
 278                 res = KeyVal.search(line)
 279                 if res != None:
 280                     val = res.group(1)
 281                     if not layouttranslations:
 282                         writeString(out, src, base, lineno, val)
 283
 284     if layouttranslations:
 285         # Extract translations of layout files
 286         import polib
 287
 288         # Sort languages and key to minimize the diff between different runs
 289         # with changed translations
 290         languages.sort()
 291         keys = []
 292         for key in keyset:
 293             keys.append(key)
 294         keys.sort()
 295
 296         print >> out, '''# This file has been automatically generated by po/lyx_pot.py.
 297 # PLEASE MODIFY ONLY THE LAGUAGES HAVING NO .po FILE! If you want to regenerate
 298 # this file from the translations, run `make ../lib/layouttranslations' in po.
 299 # Python polib library is needed for building the output file.
 300 #
 301 # This file should remain fixed during minor LyX releases.'''
 302         for lang in languages:
 303             print >> out, '\nTranslation %s' % lang
 304             if lang in oldtrans.keys():
 305                 trans = oldtrans[lang]
 306             else:
 307                 trans = dict()
 308             if not lang in oldlanguages:
 309                 poname = os.path.join(base, 'po/' + lang + '.po')
 310                 po = polib.pofile(poname)
 311                 # Iterate through po entries and not keys for speed reasons.
 312                 # FIXME: The code is still too slow
 313                 for entry in po:
 314                     if not entry.translated():
 315                         continue
 316                     if entry.msgid in keys:
 317                         key = entry.msgid
 318                         val = entry.msgstr
 319                         # some translators keep untranslated entries
 320                         if val != key:
 321                             trans[key] = val
 322             for key in keys:
 323                 if key in trans.keys():
 324                     val = trans[key].replace('\\', '\\\\').replace('"', '\\"')
 325                     key = key.replace('\\', '\\\\').replace('"', '\\"')
 326                     print >> out, '\t"%s" "%s"' % \
 327                              (key.encode('utf-8'), val.encode('utf-8'))
 328                 # to print untranslated entries, uncomment the following lines
 329                 #else:
 330                 #    key = key.replace('\\', '\\\\').replace('"', '\\"')
 331                 #    print >> out, '\t"%s" "%s"' % \
 332                 #             (key.encode('utf-8'), key.encode('utf-8'))
 333             print >> out, 'End'
 334
 335     out.close()
 336
 337
 338 def qt4_l10n(input_files, output, base):
 339     '''Generate pot file from src/frontends/qt4/ui/*.ui'''
 340     output = open(output, 'w')
 341     pat = re.compile(r'\s*<string>(.*)</string>')
 342     prop = re.compile(r'\s*<property.*name.*=.*shortcut')
 343     for src in input_files:
 344         input = open(src)
 345         skipNextLine = False
 346         for lineno, line in enumerate(input.readlines()):
 347             # skip the line after <property name=shortcut>
 348             if skipNextLine:
 349                 skipNextLine = False
 350                 continue
 351             if prop.match(line):
 352                 skipNextLine = True
 353                 continue
 354             # get lines that match <string>...</string>
 355             if pat.match(line):
 356                 (string,) = pat.match(line).groups()
 357                 string = string.replace('&amp;', '&').replace('&quot;', '"')
 358                 string = string.replace('&lt;', '<').replace('&gt;', '>')
 359                 string = string.replace('\\', '\\\\').replace('"', r'\"')
 360                 string = string.replace('&#x0a;', r'\n')
 361                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 362                     (relativePath(src, base), lineno+1, string)
 363         input.close()
 364     output.close()
 365
 366
 367 def languages_l10n(input_files, output, base):
 368     '''Generate pot file from lib/languages'''
 369     out = open(output, 'w')
 370     GuiName = re.compile(r'^[^#]*GuiName\s+(.*)')
 371
 372     for src in input_files:
 373         descStartLine = -1
 374         descLines = []
 375         lineno = 0
 376         for line in open(src).readlines():
 377             lineno += 1
 378             res = GuiName.search(line)
 379             if res != None:
 380                 string = res.group(1)
 381                 writeString(out, src, base, lineno, string)
 382                 continue
 383
 384     out.close()
 385
 386
 387 def external_l10n(input_files, output, base):
 388     '''Generate pot file from lib/external_templates'''
 389     output = open(output, 'w')
 390     Template = re.compile(r'^Template\s+(.*)')
 391     GuiName = re.compile(r'\s*GuiName\s+(.*)')
 392     HelpTextStart = re.compile(r'\s*HelpText\s')
 393     HelpTextSection = re.compile(r'\s*(\S.*)\s*$')
 394     HelpTextEnd = re.compile(r'\s*HelpTextEnd\s')
 395     i = -1
 396     for src in input_files:
 397         input = open(src)
 398         inHelp = False
 399         hadHelp = False
 400         prev_help_string = ''
 401         for lineno, line in enumerate(input.readlines()):
 402             if Template.match(line):
 403                 (string,) = Template.match(line).groups()
 404             elif GuiName.match(line):
 405                 (string,) = GuiName.match(line).groups()
 406             elif inHelp:
 407                 if HelpTextEnd.match(line):
 408                     if hadHelp:
 409                         print >> output, '\nmsgstr ""\n'
 410                     inHelp = False
 411                     hadHelp = False
 412                     prev_help_string = ''
 413                 elif HelpTextSection.match(line):
 414                     (help_string,) = HelpTextSection.match(line).groups()
 415                     help_string = help_string.replace('"', '')
 416                     if help_string != "" and prev_help_string == '':
 417                         print >> output, '#: %s:%d\nmsgid ""\n"%s\\n"' % \
 418                             (relativePath(src, base), lineno+1, help_string)
 419                         hadHelp = True
 420                     elif help_string != "":
 421                         print >> output, '"%s\\n"' % help_string
 422                     prev_help_string = help_string
 423             elif HelpTextStart.match(line):
 424                 inHelp = True
 425                 prev_help_string = ''
 426             else:
 427                 continue
 428             string = string.replace('"', '')
 429             if string != "" and not inHelp:
 430                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 431                     (relativePath(src, base), lineno+1, string)
 432         input.close()
 433     output.close()
 434
 435
 436 def formats_l10n(input_files, output, base):
 437     '''Generate pot file from configure.py'''
 438     output = open(output, 'w')
 439     GuiName = re.compile(r'.*\Format\s+\S+\s+\S+\s+"([^"]*)"\s+(\S*)\s+.*')
 440     GuiName2 = re.compile(r'.*\Format\s+\S+\s+\S+\s+([^"]\S+)\s+(\S*)\s+.*')
 441     input = open(input_files[0])
 442     for lineno, line in enumerate(input.readlines()):
 443         label = ""
 444         labelsc = ""
 445         if GuiName.match(line):
 446             label = GuiName.match(line).group(1)
 447             shortcut = GuiName.match(line).group(2).replace('"', '')
 448         elif GuiName2.match(line):
 449             label = GuiName2.match(line).group(1)
 450             shortcut = GuiName2.match(line).group(2).replace('"', '')
 451         else:
 452             continue
 453         label = label.replace('\\', '\\\\').replace('"', '')
 454         if shortcut != "":
 455             labelsc = label + "|" + shortcut
 456         if label != "":
 457             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 458                 (relativePath(input_files[0], base), lineno+1, label)
 459         if labelsc != "":
 460             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 461                 (relativePath(input_files[0], base), lineno+1, labelsc)
 462     input.close()
 463     output.close()
 464
 465
 466 def encodings_l10n(input_files, output, base):
 467     '''Generate pot file from lib/encodings'''
 468     output = open(output, 'w')
 469     # assuming only one encodings file
 470     #                 Encoding utf8      utf8    "Unicode (utf8)" UTF-8    variable inputenc
 471     reg = re.compile('Encoding [\w-]+\s+[\w-]+\s+"([\w \-\(\)]+)"\s+[\w-]+\s+(fixed|variable)\s+\w+.*')
 472     input = open(input_files[0])
 473     for lineno, line in enumerate(input.readlines()):
 474         if not line.startswith('Encoding'):
 475             continue
 476         if reg.match(line):
 477             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 478                 (relativePath(input_files[0], base), lineno+1, reg.match(line).groups()[0])
 479         else:
 480             print "Error: Unable to handle line:"
 481             print line
 482             # No need to abort if the parsing fails
 483             # sys.exit(1)
 484     input.close()
 485     output.close()
 486
 487
 488
 489 Usage = '''
 490 lyx_pot.py [-b|--base top_src_dir] [-o|--output output_file] [-h|--help] [-s|src_file filename] -t|--type input_type input_files
 491
 492 where
 493     --base:
 494         path to the top source directory. default to '.'
 495     --output:
 496         output pot file, default to './lyx.pot'
 497     --src_file
 498         filename that contains a list of input files in each line
 499     --input_type can be
 500         ui: lib/ui/*
 501         layouts: lib/layouts/*
 502         layouttranslations: create lib/layouttranslations from po/*.po and lib/layouts/*
 503         qt4: qt4 ui files
 504         languages: file lib/languages
 505         encodings: file lib/encodings
 506         external: external templates file
 507         formats: formats predefined in lib/configure.py
 508 '''
 509
 510 if __name__ == '__main__':
 511     input_type = None
 512     output = 'lyx.pot'
 513     base = '.'
 514     input_files = []
 515     #
 516     optlist, args = getopt.getopt(sys.argv[1:], 'ht:o:b:s:',
 517         ['help', 'type=', 'output=', 'base=', 'src_file='])
 518     for (opt, value) in optlist:
 519         if opt in ['-h', '--help']:
 520             print Usage
 521             sys.exit(0)
 522         elif opt in ['-o', '--output']:
 523             output = value
 524         elif opt in ['-b', '--base']:
 525             base = value
 526         elif opt in ['-t', '--type']:
 527             input_type = value
 528         elif opt in ['-s', '--src_file']:
 529             input_files = [f.strip() for f in open(value)]
 530
 531     if input_type not in ['ui', 'layouts', 'layouttranslations', 'qt4', 'languages', 'encodings', 'external', 'formats'] or output is None:
 532         print 'Wrong input type or output filename.'
 533         sys.exit(1)
 534
 535     input_files += args
 536
 537     if input_type == 'ui':
 538         ui_l10n(input_files, output, base)
 539     elif input_type == 'layouts':
 540         layouts_l10n(input_files, output, base, False)
 541     elif input_type == 'layouttranslations':
 542         layouts_l10n(input_files, output, base, True)
 543     elif input_type == 'qt4':
 544         qt4_l10n(input_files, output, base)
 545     elif input_type == 'external':
 546         external_l10n(input_files, output, base)
 547     elif input_type == 'formats':
 548         formats_l10n(input_files, output, base)
 549     elif input_type == 'encodings':
 550         encodings_l10n(input_files, output, base)
 551     else:
 552         languages_l10n(input_files, output, base)
 553
 554