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
  21 def relativePath(path, base):
  22     '''return relative path from top source dir'''
  23     # full pathname of path
  24     path1 = os.path.normpath(os.path.realpath(path)).split(os.sep)
  25     path2 = os.path.normpath(os.path.realpath(base)).split(os.sep)
  26     if path1[:len(path2)] != path2:
  27         print "Path %s is not under top source directory" % path
  28     path3 = os.path.join(*path1[len(path2):]);
  29     # replace all \ by / such that we get the same comments on Windows and *nix
  30     path3 = path3.replace('\\', '/')
  31     return path3
  32
  33
  34 def writeString(outfile, infile, basefile, lineno, string):
  35     string = string.replace('\\', '\\\\').replace('"', '')
  36     if string == "":
  37         return
  38     print >> outfile, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
  39         (relativePath(infile, basefile), lineno, string)
  40
  41
  42 def ui_l10n(input_files, output, base):
  43     '''Generate pot file from lib/ui/*'''
  44     output = open(output, 'w')
  45     Submenu = re.compile(r'^[^#]*Submenu\s+"([^"]*)"')
  46     Popupmenu = re.compile(r'^[^#]*PopupMenu\s+"[^"]+"\s+"([^"]*)"')
  47     Toolbar = re.compile(r'^[^#]*Toolbar\s+"[^"]+"\s+"([^"]*)"')
  48     Item = re.compile(r'[^#]*Item\s+"([^"]*)"')
  49     TableInsert = re.compile(r'[^#]*TableInsert\s+"([^"]*)"')
  50     for src in input_files:
  51         input = open(src)
  52         for lineno, line in enumerate(input.readlines()):
  53             if Submenu.match(line):
  54                 (string,) = Submenu.match(line).groups()
  55                 string = string.replace('_', ' ')
  56             elif Popupmenu.match(line):
  57                 (string,) = Popupmenu.match(line).groups()
  58             elif Toolbar.match(line):
  59                 (string,) = Toolbar.match(line).groups()
  60             elif Item.match(line):
  61                 (string,) = Item.match(line).groups()
  62             elif TableInsert.match(line):
  63                 (string,) = TableInsert.match(line).groups()
  64             else:
  65                 continue
  66             string = string.replace('"', '')
  67             if string != "":
  68                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
  69                     (relativePath(src, base), lineno+1, string)
  70         input.close()
  71     output.close()
  72
  73
  74 def layouts_l10n(input_files, output, base):
  75     '''Generate pot file from lib/layouts/*.{layout,inc,module}'''
  76     out = open(output, 'w')
  77     Style = re.compile(r'^Style\s+(.*)')
  78     # include ???LabelString???, but exclude comment lines
  79     LabelString = re.compile(r'^[^#]*LabelString\S*\s+(.*)')
  80     GuiName = re.compile(r'\s*GuiName\s+(.*)')
  81     ListName = re.compile(r'\s*ListName\s+(.*)')
  82     CategoryName = re.compile(r'\s*Category\s+(.*)')
  83     NameRE = re.compile(r'DeclareLyXModule.*{(.*)}')
  84     InsetLayout = re.compile(r'^InsetLayout\s+(.*)')
  85     DescBegin = re.compile(r'#+\s*DescriptionBegin\s*$')
  86     DescEnd = re.compile(r'#+\s*DescriptionEnd\s*$')
  87     Category = re.compile(r'#Category: (.*)$')
  88     I18nPreamble = re.compile(r'\s*(Lang)|(Babel)Preamble\s*$')
  89     EndI18nPreamble = re.compile(r'\s*End(Lang)|(Babel)Preamble\s*$')
  90     I18nString = re.compile(r'_\(([^\)]+)\)')
  91
  92     for src in input_files:
  93         readingDescription = False
  94         readingI18nPreamble = False
  95         descStartLine = -1
  96         descLines = []
  97         lineno = 0
  98         for line in open(src).readlines():
  99             lineno += 1
 100             if readingDescription:
 101                 res = DescEnd.search(line)
 102                 if res != None:
 103                     readingDescription = False
 104                     desc = " ".join(descLines)
 105                     writeString(out, src, base, lineno + 1, desc)
 106                     continue
 107                 descLines.append(line[1:].strip())
 108                 continue
 109             res = DescBegin.search(line)
 110             if res != None:
 111                 readingDescription = True
 112                 descStartLine = lineno
 113                 continue
 114             if readingI18nPreamble:
 115                 res = EndI18nPreamble.search(line)
 116                 if res != None:
 117                     readingI18nPreamble = False
 118                     continue
 119                 res = I18nString.search(line)
 120                 if res != None:
 121                     string = res.group(1)
 122                     writeString(out, src, base, lineno, string)
 123                 continue
 124             res = I18nPreamble.search(line)
 125             if res != None:
 126                 readingI18nPreamble = True
 127                 continue
 128             res = NameRE.search(line)
 129             if res != None:
 130                 string = res.group(1)
 131                 string = string.replace('\\', '\\\\').replace('"', '')
 132                 if string != "":
 133                     print >> out, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 134                         (relativePath(src, base), lineno + 1, string)
 135                 continue
 136             res = Style.search(line)
 137             if res != None:
 138                 string = res.group(1)
 139                 string = string.replace('_', ' ')
 140                 writeString(out, src, base, lineno, string)
 141                 continue
 142             res = LabelString.search(line)
 143             if res != None:
 144                 string = res.group(1)
 145                 writeString(out, src, base, lineno, string)
 146                 continue
 147             res = GuiName.search(line)
 148             if res != None:
 149                 string = res.group(1)
 150                 writeString(out, src, base, lineno, string)
 151                 continue
 152             res = CategoryName.search(line)
 153             if res != None:
 154                 string = res.group(1)
 155                 writeString(out, src, base, lineno, string)
 156                 continue
 157             res = ListName.search(line)
 158             if res != None:
 159                 string = res.group(1)
 160                 writeString(out, src, base, lineno, string)
 161                 continue
 162             res = InsetLayout.search(line)
 163             if res != None:
 164                 string = res.group(1)
 165                 string = string.replace('_', ' ')
 166                 writeString(out, src, base, lineno, string)
 167                 continue
 168             res = Category.search(line)
 169             if res != None:
 170                 string = res.group(1)
 171                 writeString(out, src, base, lineno, string)
 172                 continue
 173     out.close()
 174
 175
 176 def qt4_l10n(input_files, output, base):
 177     '''Generate pot file from src/frontends/qt4/ui/*.ui'''
 178     output = open(output, 'w')
 179     pat = re.compile(r'\s*<string>(.*)</string>')
 180     prop = re.compile(r'\s*<property.*name.*=.*shortcut')
 181     for src in input_files:
 182         input = open(src)
 183         skipNextLine = False
 184         for lineno, line in enumerate(input.readlines()):
 185             # skip the line after <property name=shortcut>
 186             if skipNextLine:
 187                 skipNextLine = False
 188                 continue
 189             if prop.match(line):
 190                 skipNextLine = True
 191                 continue
 192             # get lines that match <string>...</string>
 193             if pat.match(line):
 194                 (string,) = pat.match(line).groups()
 195                 string = string.replace('&amp;', '&').replace('&lt;', '<').replace('&gt;', '>')
 196                 string = string.replace('\\', '\\\\').replace('"', r'\"')
 197                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 198                     (relativePath(src, base), lineno+1, string)
 199         input.close()
 200     output.close()
 201
 202
 203 def languages_l10n(input_files, output, base):
 204     '''Generate pot file from lib/language'''
 205     output = open(output, 'w')
 206     # assuming only one language file
 207     reg = re.compile('[\w-]+\s+[\w"]+\s+"([\w \-\(\),]+)"\s+(true|false)\s+[\w-]+\s+\w+\s+"[^"]*"')
 208     input = open(input_files[0])
 209     for lineno, line in enumerate(input.readlines()):
 210         if line[0] == '#':
 211             continue
 212         # From:
 213         #   afrikaans   afrikaans       "Afrikaans"     false  iso8859-15 af_ZA  ""
 214         # To:
 215         #   #: lib/languages:2
 216         #   msgid "Afrikaans"
 217         #   msgstr ""
 218         if reg.match(line):
 219             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 220                 (relativePath(input_files[0], base), lineno+1, reg.match(line).groups()[0])
 221         else:
 222             print "Error: Unable to handle line:"
 223             print line
 224             # No need to abort if the parsing fails (e.g. "ignore" language has no encoding)
 225             # sys.exit(1)
 226     input.close()
 227     output.close()
 228
 229
 230 def external_l10n(input_files, output, base):
 231     '''Generate pot file from lib/external_templates'''
 232     output = open(output, 'w')
 233     Template = re.compile(r'^Template\s+(.*)')
 234     GuiName = re.compile(r'\s*GuiName\s+(.*)')
 235     HelpTextStart = re.compile(r'\s*HelpText\s')
 236     HelpTextSection = re.compile(r'\s*(\S.*)\s*$')
 237     HelpTextEnd = re.compile(r'\s*HelpTextEnd\s')
 238     i = -1
 239     for src in input_files:
 240         input = open(src)
 241         inHelp = False
 242         hadHelp = False
 243         prev_help_string = ''
 244         for lineno, line in enumerate(input.readlines()):
 245             if Template.match(line):
 246                 (string,) = Template.match(line).groups()
 247             elif GuiName.match(line):
 248                 (string,) = GuiName.match(line).groups()
 249             elif inHelp:
 250                 if HelpTextEnd.match(line):
 251                     if hadHelp:
 252                         print >> output, '\nmsgstr ""\n'
 253                     inHelp = False
 254                     hadHelp = False
 255                     prev_help_string = ''
 256                 elif HelpTextSection.match(line):
 257                     (help_string,) = HelpTextSection.match(line).groups()
 258                     help_string = help_string.replace('"', '')
 259                     if help_string != "" and prev_help_string == '':
 260                         print >> output, '#: %s:%d\nmsgid ""\n"%s\\n"' % \
 261                             (relativePath(src, base), lineno+1, help_string)
 262                         hadHelp = True
 263                     elif help_string != "":
 264                         print >> output, '"%s\\n"' % help_string
 265                     prev_help_string = help_string
 266             elif HelpTextStart.match(line):
 267                 inHelp = True
 268                 prev_help_string = ''
 269             else:
 270                 continue
 271             string = string.replace('"', '')
 272             if string != "" and not inHelp:
 273                 print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 274                     (relativePath(src, base), lineno+1, string)
 275         input.close()
 276     output.close()
 277
 278
 279 def formats_l10n(input_files, output, base):
 280     '''Generate pot file from configure.py'''
 281     output = open(output, 'w')
 282     GuiName = re.compile(r'.*\Format\s+\S+\s+\S+\s+"([^"]*)"\s+(\S*)\s+.*')
 283     GuiName2 = re.compile(r'.*\Format\s+\S+\s+\S+\s+([^"]\S+)\s+(\S*)\s+.*')
 284     input = open(input_files[0])
 285     for lineno, line in enumerate(input.readlines()):
 286         label = ""
 287         labelsc = ""
 288         if GuiName.match(line):
 289             label = GuiName.match(line).group(1)
 290             shortcut = GuiName.match(line).group(2).replace('"', '')
 291         elif GuiName2.match(line):
 292             label = GuiName2.match(line).group(1)
 293             shortcut = GuiName2.match(line).group(2).replace('"', '')
 294         else:
 295             continue
 296         label = label.replace('\\', '\\\\').replace('"', '')
 297         if shortcut != "":
 298             labelsc = label + "|" + shortcut
 299         if label != "":
 300             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 301                 (relativePath(input_files[0], base), lineno+1, label)
 302         if labelsc != "":
 303             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 304                 (relativePath(input_files[0], base), lineno+1, labelsc)
 305     input.close()
 306     output.close()
 307
 308
 309 def encodings_l10n(input_files, output, base):
 310     '''Generate pot file from lib/encodings'''
 311     output = open(output, 'w')
 312     # assuming only one encodings file
 313     #                 Encoding utf8      utf8    "Unicode (utf8)" UTF-8    variable inputenc
 314     reg = re.compile('Encoding [\w-]+\s+[\w-]+\s+"([\w \-\(\)]+)"\s+[\w-]+\s+(fixed|variable)\s+\w+.*')
 315     input = open(input_files[0])
 316     for lineno, line in enumerate(input.readlines()):
 317         if not line.startswith('Encoding'):
 318             continue
 319         if reg.match(line):
 320             print >> output, '#: %s:%d\nmsgid "%s"\nmsgstr ""\n' % \
 321                 (relativePath(input_files[0], base), lineno+1, reg.match(line).groups()[0])
 322         else:
 323             print "Error: Unable to handle line:"
 324             print line
 325             # No need to abort if the parsing fails
 326             # sys.exit(1)
 327     input.close()
 328     output.close()
 329
 330
 331
 332 Usage = '''
 333 lyx_pot.py [-b|--base top_src_dir] [-o|--output output_file] [-h|--help] [-s|src_file filename] -t|--type input_type input_files
 334
 335 where
 336     --base:
 337         path to the top source directory. default to '.'
 338     --output:
 339         output pot file, default to './lyx.pot'
 340     --src_file
 341         filename that contains a list of input files in each line
 342     --input_type can be
 343         ui: lib/ui/*
 344         layouts: lib/layouts/*
 345         qt4: qt4 ui files
 346         languages: file lib/languages
 347         encodings: file lib/encodings
 348         external: external templates file
 349         formats: formats predefined in lib/configure.py
 350 '''
 351
 352 if __name__ == '__main__':
 353     input_type = None
 354     output = 'lyx.pot'
 355     base = '.'
 356     input_files = []
 357     #
 358     optlist, args = getopt.getopt(sys.argv[1:], 'ht:o:b:s:',
 359         ['help', 'type=', 'output=', 'base=', 'src_file='])
 360     for (opt, value) in optlist:
 361         if opt in ['-h', '--help']:
 362             print Usage
 363             sys.exit(0)
 364         elif opt in ['-o', '--output']:
 365             output = value
 366         elif opt in ['-b', '--base']:
 367             base = value
 368         elif opt in ['-t', '--type']:
 369             input_type = value
 370         elif opt in ['-s', '--src_file']:
 371             input_files = [f.strip() for f in open(value)]
 372
 373     if input_type not in ['ui', 'layouts', 'modules', 'qt4', 'languages', 'encodings', 'external', 'formats'] or output is None:
 374         print 'Wrong input type or output filename.'
 375         sys.exit(1)
 376
 377     input_files += args
 378
 379     if input_type == 'ui':
 380         ui_l10n(input_files, output, base)
 381     elif input_type == 'layouts':
 382         layouts_l10n(input_files, output, base)
 383     elif input_type == 'qt4':
 384         qt4_l10n(input_files, output, base)
 385     elif input_type == 'external':
 386         external_l10n(input_files, output, base)
 387     elif input_type == 'formats':
 388         formats_l10n(input_files, output, base)
 389     elif input_type == 'encodings':
 390         encodings_l10n(input_files, output, base)
 391     else:
 392         languages_l10n(input_files, output, base)
 393
 394