5 # This file is part of LyX, the document processor.
6 # Licence details can be found in the file COPYING.
8 # author: Michael Gerz, michael.gerz@teststep.org
10 # This script performs some consistency checks on po files:
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
19 # pocheck.pl po_file po_file ...
26 foreach my $pofilename ( @ARGV )
28 print "Processing po file '$pofilename'...\n";
30 open( INPUT, "<$pofilename" )
31 || die "Cannot read po file '$pofilename'";
36 keys( %trans ) = 10000;
38 my $noOfLines = $#pofile;
43 my ($msgid, $msgstr, $more);
45 while ($i <= $noOfLines) {
46 ( $msgid ) = ( $pofile[$i] =~ m/^msgid "(.*)"/ );
50 # some msgid's are more than one line long, so add those.
51 while ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) {
52 $msgid = $msgid . $more;
56 # now look for the associated msgstr.
57 until ( ( $msgstr ) = ( $pofile[$i] =~ m/^msgstr "(.*)"/ ) ) { $i++; };
59 # again collect any extra lines.
60 while ( ( $i <= $noOfLines ) &&
61 ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) ) {
62 $msgstr = $msgstr . $more;
66 # nothing to do if one of them is empty.
67 # (surely that is always $msgstr?)
68 next if ($msgid eq "" or $msgstr eq "");
70 # Check for matching %1$s, etc.
71 my @argstrs = ( $msgid =~ m/%(\d)\$s/g );
74 foreach my $arg (@argstrs) { $n = $arg if $arg > $n; }
76 print "Problem finding arguments in:\n $msgid!\n";
79 foreach my $i (1..$n) {
81 if ( $msgstr !~ m/$arg/ ) {
82 print "Missing argument `$arg'\n '$msgid' ==> '$msgstr'\n";
89 # Check colon at the end of a message
90 if ( ( $msgid =~ m/: *(\|.*)?$/ ) != ( $msgstr =~ m/: *(\|.*)?$/ ) ) {
91 print( "Missing or unexpected colon:\n" );
92 print( " '$msgid' => '$msgstr'\n" );
96 # Check period at the end of a message; uncomment code if you are paranoid
97 #if ( ( $msgid =~ m/\. *(\|.*)?$/ ) != ( $msgstr =~ m/\. *(\|.*)?$/ ) ) {
98 # print( "Missing or unexpected period:\n" );
99 # print( " '$msgid' => '$msgstr'\n" );
103 # Check space at the end of a message
104 if ( ( $msgid =~ m/ *?(\|.*)?$/ ) != ( $msgstr =~ m/ *?(\|.*)?$/ ) ) {
105 print( "Missing or unexpected space:\n" );
106 print( " '$msgid' => '$msgstr'\n" );
110 # Check for "&" shortcuts
111 if ( ( $msgid =~ m/&[^ ]/ ) != ( $msgstr =~ m/&[^ ]/ ) ) {
112 print( "Missing or unexpected Qt shortcut:\n" );
113 print( " '$msgid' => '$msgstr'\n" );
117 # Check for "|..." shortcuts
118 if ( ( $msgid =~ m/\|[^ ]/ ) != ( $msgstr =~ m/\|[^ ]/ ) ) {
119 print( "Missing or unexpected menu shortcut:\n" );
120 print( " '$msgid' => '$msgstr'\n" );
124 # we now collect these translations in a hash.
125 # this will allow us to check below if we have translated
126 # anything more than one way.
127 my $msgid_clean = lc($msgid);
128 my $msgstr_clean = lc($msgstr);
130 $msgid_clean =~ s/(.*)\|.*?$/$1/; # strip menu shortcuts
131 $msgstr_clean =~ s/(.*)\|.*?$/$1/;
132 $msgid_clean =~ s/&([^ ])/$1/; # strip Qt shortcuts
133 $msgstr_clean =~ s/&([^ ])/$1/;
135 # this is a hash of hashes. the keys of the outer hash are
136 # cleaned versions of ORIGINAL strings. the keys of the inner hash
137 # are the cleaned versions of their TRANSLATIONS. The value for the
138 # inner hash is an array of the orignal string and translation.
139 $trans{$msgid_clean}{$msgstr_clean} = [ $msgid, $msgstr ];
142 foreach $msgid ( keys %trans ) {
143 # so $ref is a reference to the inner hash.
144 my $ref = $trans{$msgid};
145 # @msgstrkeys is an array of the keys of that inner hash.
146 my @msgstrkeys = keys %$ref;
148 # do we have more than one such key?
149 if ( $#msgstrkeys > 0 ) {
150 print( "Different translations for '$msgid':\n" );
151 foreach $msgstr ( @msgstrkeys ) {
152 print( " '" . $trans{$msgid}{$msgstr}[0] . "' => '" . $trans{$msgid}{$msgstr}[1] . "'\n" );
158 print( "\nTotal number of warnings: $warn\n\n" );