po/extract-android-po.pl

   1 #!/usr/bin/env perl
   2 #
   3 # Quick hack to build a .pot file from an Android strings.xml file
   4 #
   5 # Copyright: Morphoss Ltd <http://www.morphoss.com/>
   6 # Author: Andrew McMillan <andrew@mcmillan.net.nz>
   7 # License: CC-0, Public Domain, GPL v2, GPL v3, Apache v2 or Simplified BSD.
   8 #          Other licenses considered on request.
   9 #
  10
  11 use strict;
  12 use warnings;
  13
  14 use IO::Handle;
  15 use Getopt::Long qw(:config permute);  # allow mixed args.
  16
  17
  18 # Stuff that should not be hard-coded, but is :-)
  19 my $resources_dir = "../res";
  20 my @extract_filenames = ( "strings", "timezonenames" );
  21 my $build_filename = "strings";
  22 my $messages_filename = "messages.pot";
  23
  24 my $debug = 0;
  25 my $extractmode = 0;
  26 my $buildmode = 0;
  27 my $helpmeplease = 0;
  28
  29 GetOptions ('debug!'    => \$debug,
  30
  31             'extract!'  => \$extractmode,
  32             'build!'    => \$buildmode,
  33
  34             'help'      => \$helpmeplease  );
  35
  36 usage() if ( $helpmeplease || ($extractmode && $buildmode) );
  37
  38 my $code_dir = $0;
  39 $code_dir =~ s{/[^/]*$}{};
  40 chdir $code_dir;
  41
  42 if ( $extractmode ) {
  43   # Update/extract the strings for the messages.po file
  44   extract_po_file(\@extract_filenames, $messages_filename);
  45 }
  46 elsif ( $buildmode ) {
  47   # From the language po files build the various strings files.
  48   build_strings_files($messages_filename, $build_filename );
  49 }
  50 else {
  51   usage();
  52 }
  53
  54
  55 =item
  56 Provides basic help to the user.
  57 =cut
  58 sub usage {
  59   print "You must specify either '--extract' or '--build'\n";
  60   exit 0
  61 }
  62
  63
  64 =item
  65 Extracts the strings from some Android strings.xml into a messages.po file
  66 =cut
  67 sub extract_po_file {
  68   my $filenames = shift;
  69   my $outfile = shift;
  70
  71   my @xgettext = ( "xgettext", "-L", "PO", "-o", $outfile, "-" );
  72   open( XGETTEXT, "|-", @xgettext );
  73   autoflush XGETTEXT 1;
  74   printf( "Extracting to '%s'\n", $outfile ) if ( $debug );
  75
  76   for my $filename ( @$filenames ) {
  77     $filename = $resources_dir ."/values/". $filename . ".xml";
  78     printf( XGETTEXT "#SourceFileStart:%s\n", $filename );
  79     printf( "Extracting strings from '%s'\n", $filename ) if ( $debug );
  80     open( XMLFILE, "<", $filename );
  81     while( <XMLFILE> ) {
  82       m{<string .*?name="(.*?)".*?>(.*?)</string>} && do {
  83         my $msgid = $1;
  84         my $msgstr = $2;
  85         $msgstr =~ s{\\'}{'}g;
  86         $msgstr =~ s/"/&quot;/g;
  87         printf( XGETTEXT 'msgid "%s"%s', $msgid, "\n" );
  88         printf( XGETTEXT 'msgstr "%s"%s', $msgstr, "\n\n" );
  89       };
  90       m{<!--(.*?)-->} && do {
  91         printf( XGETTEXT "#%s\n", $1 );
  92       };
  93     }
  94     close(XMLFILE);
  95     printf( XGETTEXT "#SourceFileEnd:%s\n\n", $filename );
  96   }
  97   printf( "Extraction completed\n" ) if ( $debug );
  98   # close(XGETTEXT);
  99
 100 }
 101
 102
 103 =item
 104 Finds the translated .po files in a directory and uses them to construct a new strings.xml file
 105 for each one by merging the details from the original strings.xml with the translated strings
 106 from the la_NG.po file.
 107 =cut
 108 sub build_strings_files {
 109   my $template_file = shift;
 110   my $stringsfile = shift;
 111
 112   # Now merge into each translation
 113   opendir(my $dh, ".");
 114   while( my $fn = readdir($dh) ) {
 115     if( $fn =~ m{^([a-z]{2}(_[A-Z]{2})?)\.po$} ) {
 116       my $lang = $1;
 117       printf( "Building language: %s\n", $lang );
 118       my $strings = get_translated_strings($fn);
 119       merge_into_xml( $lang, $strings, $stringsfile );
 120     }
 121   }
 122   closedir($dh);
 123 }
 124
 125
 126 =item
 127 =cut
 128 sub get_translated_strings {
 129   my $filename = shift;
 130
 131   my $strings = {};
 132   open( TRANSLATED, "<", $filename );
 133
 134   my $msgid = undef;
 135   while( <TRANSLATED> ) {
 136     next if ( /^\s*#/ );
 137
 138     if ( /^\s*msgid \"(.*)"\r?\n?$/ ) {
 139       $msgid = $1;
 140       $strings->{$msgid} = "";
 141     }
 142     elsif ( defined($msgid) ) {
 143       /^\s*(msgstr )?"(.*)"\r?\n?$/ && do {
 144         $strings->{$msgid} .= $2;
 145       };
 146     }
 147   }
 148
 149   close(TRANSLATED);
 150
 151 #  foreach my $s ( keys %{$strings} ) {
 152 #    printf( STDOUT "<string name=\"%s\">%s</string>\n", $s, $strings->{$s} );
 153 #  }
 154
 155   return $strings;
 156 }
 157
 158
 159 =item
 160 =cut
 161 sub merge_into_xml {
 162   my $lang = shift;
 163   my $strings = shift;
 164   my $strings_filename = shift;
 165
 166   $lang =~ s{_([A-Z]{2})}{-r$1};
 167   my $in_filename = sprintf( '%s/values/%s.xml', $resources_dir, $strings_filename);
 168
 169   my $outdir = sprintf('%s/values-%s', $resources_dir, $lang);
 170   mkdir $outdir unless( -d $outdir );
 171   my $out_filename = sprintf( '%s/%s.xml', $outdir, $strings_filename);
 172   open( XMLIN, "<", $in_filename );
 173   open( XMLOUT, ">", $out_filename );
 174
 175   while( <XMLIN> ) {
 176     if ( ! m{<!--} && m{<string (.*?)name="(.*?)"(.*?)>(.*?)</string>} ) {
 177       my $preamble = (defined($1)?$1:"");
 178       my $msgid = $2;
 179       my $postamble = (defined($3)?$3:"");
 180       my $msgstr = $4;
 181       $msgstr =~ s{\\'}{'}g;
 182       next if ( ! defined($strings->{$msgid}) || $msgstr eq $strings->{$msgid} );
 183       next if ( $strings->{$msgid} eq "" );
 184       $strings->{$msgid} =~ s{"}{&quot;}g;
 185       $strings->{$msgid} =~ s{(['\\])}{\\$1}g;
 186       printf( XMLOUT '<string %sname="%s"%s>%s</string>%s',
 187                         $preamble, $msgid, $postamble, $strings->{$msgid}, "\n" );
 188     }
 189     else {
 190       print XMLOUT unless( $_ =~ m{^\s*$} );
 191     }
 192   }
 193
 194   close(XMLIN);
 195   close(XMLOUT);
 196 }
 197