po/pocheck.pl

   1 #! /usr/bin/perl -w
   2
   3 # file pocheck.pl
   4 #
   5 # This file is part of LyX, the document processor.
   6 # Licence details can be found in the file COPYING.
   7 #
   8 # author: Michael Gerz, michael.gerz@teststep.org
   9 #
  10 # This script performs some consistency checks on po files:
  11 #
  12 #   1. Uniform translation of messages that are identical except
  13 #      for capitalization, shortcuts, and shortcut notation.
  14 #   2. Usage of the following elements in both the original and
  15 #      the translated message (or no usage at all):
  16 #      shortcuts ("&" and "|..."), trailing space, trailing colon
  17 #
  18 # Invocation:
  19 #    pocheck.pl po_file po_file ...
  20
  21 foreach $pofilename ( @ARGV )
  22 {
  23   print "Processing po file '$pofilename'...\n";
  24
  25   open( INPUT, "<$pofilename" )
  26     || die "Cannot read po file '$pofilename'";
  27   @pofile = <INPUT>;
  28   close( INPUT );
  29
  30   undef( %trans );
  31   keys( %trans ) = 10000;
  32
  33   $noOfLines = $#pofile;
  34
  35   $warn = 0;
  36
  37   $i = 0;
  38   while ($i <= $noOfLines) {
  39     if ( ( $msgid ) = ( $pofile[$i] =~ m/^msgid "(.*)"/ ) ) {
  40       $i++;
  41       while ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) {
  42         $msgid = $msgid . $more;
  43         $i++;
  44       }
  45
  46       until ( ( $msgstr ) = ( $pofile[$i] =~ m/^msgstr "(.*)"/ ) ) { $i++; };
  47       $i++;
  48       while ( ( $i <= $noOfLines ) &&
  49               ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) ) {
  50         $msgstr = $msgstr . $more;
  51         $i++;
  52       }
  53
  54       if ( $msgid ne "" && $msgstr ne "" ) {
  55
  56         # Check colon at the end of a message
  57         if ( ( $msgid =~ m/: *(\|.*)?$/ ) != ( $msgstr =~ m/: *(\|.*)?$/ ) ) {
  58           print( "Missing or unexpected colon:\n" );
  59           print( "  '$msgid' => '$msgstr'\n" );
  60           $warn++;
  61         }
  62
  63         # Check period at the end of a message; uncomment code if you are paranoid
  64         #if ( ( $msgid =~ m/\. *(\|.*)?$/ ) != ( $msgstr =~ m/\. *(\|.*)?$/ ) ) {
  65         #  print( "Missing or unexpected period:\n" );
  66         #  print( "  '$msgid' => '$msgstr'\n" );
  67         #  $warn++;
  68         #}
  69
  70         # Check space at the end of a message
  71         if ( ( $msgid =~ m/  *?(\|.*)?$/ ) != ( $msgstr =~ m/  *?(\|.*)?$/ ) ) {
  72           print( "Missing or unexpected space:\n" );
  73           print( "  '$msgid' => '$msgstr'\n" );
  74           $warn++;
  75         }
  76
  77         # Check for "&" shortcuts
  78         if ( ( $msgid =~ m/&[^ ]/ ) != ( $msgstr =~ m/&[^ ]/ ) ) {
  79           print( "Missing or unexpected Qt shortcut:\n" );
  80           print( "  '$msgid' => '$msgstr'\n" );
  81           $warn++;
  82         }
  83
  84         # Check for "|..." shortcuts
  85         if ( ( $msgid =~ m/\|[^ ]/ ) != ( $msgstr =~ m/\|[^ ]/ ) ) {
  86           print( "Missing or unexpected menu shortcut:\n" );
  87           print( "  '$msgid' => '$msgstr'\n" );
  88           $warn++;
  89         }
  90
  91         $msgid_clean  = lc($msgid);
  92         $msgstr_clean = lc($msgstr);
  93
  94         $msgid_clean  =~ s/(.*)\|.*?$/$1/;  # strip menu shortcuts
  95         $msgstr_clean =~ s/(.*)\|.*?$/$1/;
  96         $msgid_clean  =~ s/&([^ ])/$1/;     # strip Qt shortcuts
  97         $msgstr_clean =~ s/&([^ ])/$1/;
  98
  99         $trans{$msgid_clean}{$msgstr_clean} = [ $msgid, $msgstr ];
 100       }
 101     } else {
 102       $i++;
 103     }
 104   }
 105
 106   foreach $msgid ( keys %trans ) {
 107     $ref = $trans{$msgid};
 108     @msgstrkeys = keys %$ref;
 109
 110     if ( $#msgstrkeys > 0 ) {
 111       print( "Different translations for '$msgid':\n" );
 112       foreach $msgstr ( @msgstrkeys ) {
 113         print( "  '" . $trans{$msgid}{$msgstr}[0] . "' => '" . $trans{$msgid}{$msgstr}[1] . "'\n" );
 114       }
 115       $warn++;
 116     }
 117   }
 118
 119   print( "\nTotal number of warnings: $warn\n\n" );
 120 }