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         for lang in languages:
 300             print >> out, '\nTranslation %s' % lang
 301             if lang in oldtrans.keys():
 302                 trans = oldtrans[lang]
 303             else:
 304                 trans = dict()
 305             if not lang in oldlanguages:
 306                 poname = os.path.join(base, 'po/' + lang + '.po')
 307                 po = polib.pofile(poname)
 308                 # Iterate through po entries and not keys for speed reasons.
 309                 # FIXME: The code is still too slow
 310                 for entry in po:
 311                     if not entry.translated():
 312                         continue
 313                     if entry.msgid in keys:
 314                         key = entry.msgid
 315                         val = entry.msgstr
 316                         # some translators keep untranslated entries
 317                         if val != key:
 318                             trans[key] = val
 319             for key in keys:
 320                 if key in trans.keys():
 321                     val = trans[key].replace('\\', '\\\\').replace('"', '\\"')
 322                     key = key.replace('\\', '\\\\').replace('"', '\\"')
 323                     print >> out, '\t"%s" "%s"' % \
 324                              (key.encode('utf-8'), val.encode('utf-8'))
 325                 # to print untranslated entries, uncomment the following lines
 326                 #else:
 327                 #    key = key.replace('\\', '\\\\').replace('"', '\\"')
 328                 #    print >> out, '\t"%s" "%s"' % \
 329                 #             (key.encode('utf-8'), key.encode('utf-8'))
 330             print >> out, 'End'
 331
 332     out.close()
 333
 334
 335 def qt4_l10n(input_files, output, base):
 336     '''Generate pot file from src/frontends/qt4/ui/*.ui'''
 337     output = open(output, 'w')
 338     pat = re.compile(r'\s*<string>(.*)</string>')
 339     prop = re.compile(r'\s*<property.*name.*=.*shortcut')
 340     for src in input_files:
 341         input = open(src)
 342         skipNextLine = False
 343         for lineno, line in enumerate(input.readlines()):
 344             # skip the line after <property name=shortcut>
 345             if skipNextLine:
 346                 skipNextLine = False
 347                 continue
 348             if prop.match(line):
 349                 skipNextLine = True
 350                 continue
 351             # get lines that match <string>...</string>
 352             if pat.match(line):
 353                 (string,) = pat.match(line).groups()
 354                 string = string.replace('&amp;', '&').replace('&quot;', '"')
 355                 string = string.replace('&lt;', '<').replace('&gt;', '>')
 356                 string = string.replace('\\', '\\\\').replace('"', r'\"')
 357                 string = string.replace('&#x0a;', r'\n')
 358                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 359                     (relativePath(src, base), lineno+1, string)
 360         input.close()
 361     output.close()
 362
 363
 364 def languages_l10n(input_files, output, base):
 365     '''Generate pot file from lib/languages'''
 366     out = open(output, 'w')
 367     GuiName = re.compile(r'^[^#]*GuiName\s+(.*)')
 368
 369     for src in input_files:
 370         descStartLine = -1
 371         descLines = []
 372         lineno = 0
 373         for line in open(src).readlines():
 374             lineno += 1
 375             res = GuiName.search(line)
 376             if res != None:
 377                 string = res.group(1)
 378                 writeString(out, src, base, lineno, string)
 379                 continue
 380
 381     out.close()
 382
 383
 384 def external_l10n(input_files, output, base):
 385     '''Generate pot file from lib/external_templates'''
 386     output = open(output, 'w')
 387     Template = re.compile(r'^Template\s+(.*)')
 388     GuiName = re.compile(r'\s*GuiName\s+(.*)')
 389     HelpTextStart = re.compile(r'\s*HelpText\s')
 390     HelpTextSection = re.compile(r'\s*(\S.*)\s*$')
 391     HelpTextEnd = re.compile(r'\s*HelpTextEnd\s')
 392     i = -1
 393     for src in input_files:
 394         input = open(src)
 395         inHelp = False
 396         hadHelp = False
 397         prev_help_string = ''
 398         for lineno, line in enumerate(input.readlines()):
 399             if Template.match(line):
 400                 (string,) = Template.match(line).groups()
 401             elif GuiName.match(line):
 402                 (string,) = GuiName.match(line).groups()
 403             elif inHelp:
 404                 if HelpTextEnd.match(line):
 405                     if hadHelp:
 406                         print >> output, '\nmsgstr ""\n'
 407                     inHelp = False
 408                     hadHelp = False
 409                     prev_help_string = ''
 410                 elif HelpTextSection.match(line):
 411                     (help_string,) = HelpTextSection.match(line).groups()
 412                     help_string = help_string.replace('"', '')
 413                     if help_string != "" and prev_help_string == '':
 414                         print >> output, '#: %s:%d\nmsgid ""\n"%s\\n"' % \
 415                             (relativePath(src, base), lineno+1, help_string)
 416                         hadHelp = True
 417                     elif help_string != "":
 418                         print >> output, '"%s\\n"' % help_string
 419                     prev_help_string = help_string
 420             elif HelpTextStart.match(line):
 421                 inHelp = True
 422                 prev_help_string = ''
 423             else:
 424                 continue
 425             string = string.replace('"', '')
 426             if string != "" and not inHelp:
 427                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 428                     (relativePath(src, base), lineno+1, string)
 429         input.close()
 430     output.close()
 431
 432
 433 def formats_l10n(input_files, output, base):
 434     '''Generate pot file from configure.py'''
 435     output = open(output, 'w')
 436     GuiName = re.compile(r'.*\Format\s+\S+\s+\S+\s+"([^"]*)"\s+(\S*)\s+.*')
 437     GuiName2 = re.compile(r'.*\Format\s+\S+\s+\S+\s+([^"]\S+)\s+(\S*)\s+.*')
 438     input = open(input_files[0])
 439     for lineno, line in enumerate(input.readlines()):
 440         label = ""
 441         labelsc = ""
 442         if GuiName.match(line):
 443             label = GuiName.match(line).group(1)
 444             shortcut = GuiName.match(line).group(2).replace('"', '')
 445         elif GuiName2.match(line):
 446             label = GuiName2.match(line).group(1)
 447             shortcut = GuiName2.match(line).group(2).replace('"', '')
 448         else:
 449             continue
 450         label = label.replace('\\', '\\\\').replace('"', '')
 451         if shortcut != "":
 452             labelsc = label + "|" + shortcut
 453         if label != "":
 454             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 455                 (relativePath(input_files[0], base), lineno+1, label)
 456         if labelsc != "":
 457             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 458                 (relativePath(input_files[0], base), lineno+1, labelsc)
 459     input.close()
 460     output.close()
 461
 462
 463 def encodings_l10n(input_files, output, base):
 464     '''Generate pot file from lib/encodings'''
 465     output = open(output, 'w')
 466     # assuming only one encodings file
 467     #                 Encoding utf8      utf8    "Unicode (utf8)" UTF-8    variable inputenc
 468     reg = re.compile('Encoding [\w-]+\s+[\w-]+\s+"([\w \-\(\)]+)"\s+[\w-]+\s+(fixed|variable)\s+\w+.*')
 469     input = open(input_files[0])
 470     for lineno, line in enumerate(input.readlines()):
 471         if not line.startswith('Encoding'):
 472             continue
 473         if reg.match(line):
 474             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 475                 (relativePath(input_files[0], base), lineno+1, reg.match(line).groups()[0])
 476         else:
 477             print "Error: Unable to handle line:"
 478             print line
 479             # No need to abort if the parsing fails
 480             # sys.exit(1)
 481     input.close()
 482     output.close()
 483
 484
 485
 486 Usage = '''
 487 lyx_pot.py [-b|--base top_src_dir] [-o|--output output_file] [-h|--help] [-s|src_file filename] -t|--type input_type input_files
 488
 489 where
 490     --base:
 491         path to the top source directory. default to '.'
 492     --output:
 493         output pot file, default to './lyx.pot'
 494     --src_file
 495         filename that contains a list of input files in each line
 496     --input_type can be
 497         ui: lib/ui/*
 498         layouts: lib/layouts/*
 499         layouttranslations: create lib/layouttranslations from po/*.po and lib/layouts/*
 500         qt4: qt4 ui files
 501         languages: file lib/languages
 502         encodings: file lib/encodings
 503         external: external templates file
 504         formats: formats predefined in lib/configure.py
 505 '''
 506
 507 if __name__ == '__main__':
 508     input_type = None
 509     output = 'lyx.pot'
 510     base = '.'
 511     input_files = []
 512     #
 513     optlist, args = getopt.getopt(sys.argv[1:], 'ht:o:b:s:',
 514         ['help', 'type=', 'output=', 'base=', 'src_file='])
 515     for (opt, value) in optlist:
 516         if opt in ['-h', '--help']:
 517             print Usage
 518             sys.exit(0)
 519         elif opt in ['-o', '--output']:
 520             output = value
 521         elif opt in ['-b', '--base']:
 522             base = value
 523         elif opt in ['-t', '--type']:
 524             input_type = value
 525         elif opt in ['-s', '--src_file']:
 526             input_files = [f.strip() for f in open(value)]
 527
 528     if input_type not in ['ui', 'layouts', 'layouttranslations', 'qt4', 'languages', 'encodings', 'external', 'formats'] or output is None:
 529         print 'Wrong input type or output filename.'
 530         sys.exit(1)
 531
 532     input_files += args
 533
 534     if input_type == 'ui':
 535         ui_l10n(input_files, output, base)
 536     elif input_type == 'layouts':
 537         layouts_l10n(input_files, output, base, False)
 538     elif input_type == 'layouttranslations':
 539         layouts_l10n(input_files, output, base, True)
 540     elif input_type == 'qt4':
 541         qt4_l10n(input_files, output, base)
 542     elif input_type == 'external':
 543         external_l10n(input_files, output, base)
 544     elif input_type == 'formats':
 545         formats_l10n(input_files, output, base)
 546     elif input_type == 'encodings':
 547         encodings_l10n(input_files, output, base)
 548     else:
 549         languages_l10n(input_files, output, base)
 550
 551