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
16 pocheck.pl [-acmpqst] po_file [po_file] ...
18 This script performs some consistency checks on po files.
20 We check for everything listed here, unless one or more of these
21 options is given, in which case we checks only for those requested.
22 -a: Check arguments, like %1\$s
23 -c: Check for colons at end
24 -m: Check for menu shortcuts
25 -p: Check for period at end
26 -q: Check Qt shortcuts
27 -s: Check for space at end
28 -t: Check for uniform translation
29 These options can be given with or without other options.
30 -f: Ignore fuzzy translations
31 -w: Only report summary total of errors
32 -i: Silent mode, report only errors
36 getopts(":hacfmpqstwi", \%options);
38 if (defined($options{h})) {
43 my $only_total = defined($options{w});
44 delete $options{w} if $only_total;
45 my $ignore_fuzzy = defined($options{f});
46 delete $options{f} if $ignore_fuzzy;
47 my $silent_mode = defined($options{i});
48 delete $options{i} if $silent_mode;
50 my $check_args = (!%options or defined($options{a}));
51 my $check_colons = (!%options or defined($options{c}));
52 my $check_spaces = (!%options or defined($options{s}));
53 my $check_periods = (!%options or defined($options{p}));
54 my $check_qt = (!%options or defined($options{q}));
55 my $check_menu = (!%options or defined($options{m}));
56 my $check_trans = (!%options or defined($options{t}));
60 foreach my $pofilename ( @ARGV ) {
63 print "Processing po file '$pofilename'...\n";
66 open( INPUT, "<$pofilename" )
67 || die "Cannot read po file '$pofilename'";
72 keys( %trans ) = 10000;
74 my $noOfLines = $#pofile;
79 my ($msgid, $msgstr, $more);
81 while ($i <= $noOfLines) {
83 ( $msgid ) = ( $pofile[$i] =~ m/^msgid "(.*)"/ );
87 my $previous = $pofile[$i - 2];
88 next if $previous =~ m/#,.*fuzzy/;
91 # some msgid's are more than one line long, so add those.
92 while ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) {
93 $msgid = $msgid . $more;
97 # now look for the associated msgstr.
98 until ( ( $msgstr ) = ( $pofile[$i] =~ m/^msgstr "(.*)"/ ) ) { $i++; };
100 # again collect any extra lines.
101 while ( ( $i <= $noOfLines ) &&
102 ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) ) {
103 $msgstr = $msgstr . $more;
107 # nothing to do if one of them is empty.
108 # (surely that is always $msgstr?)
109 next if ($msgid eq "" or $msgstr eq "");
111 # discard [[...]] from the end of msgid, this is used only as hint to translation
112 $msgid =~ s/\[\[.*\]\]$//;
114 # Check for matching %1$s, etc.
116 my @argstrs = ( $msgid =~ m/%(\d)\$s/g );
119 foreach my $arg (@argstrs) { $n = $arg if $arg > $n; }
121 print "$pofilename, line $linenum: Problem finding arguments in:\n $msgid!\n"
123 ++$bad{"Missing arguments"};
126 foreach my $i (1..$n) {
127 my $arg = "%$i\\\$s";
128 if ( $msgstr !~ m/$arg/ ) {
129 print "$pofilename, line $linenum: Missing argument `$arg'\n '$msgid' ==> '$msgstr'\n"
131 ++$bad{"Missing arguments"};
140 # Check colon at the end of a message
141 if ( ( $msgid =~ m/: *(\|.*)?$/ ) != ( $msgstr =~ m/: *(\|.*)?$/ ) ) {
142 print "Line $linenum: Missing or unexpected colon:\n '$msgid' => '$msgstr'\n"
144 ++$bad{"Bad colons"};
149 if ($check_periods) {
150 # Check period at the end of a message; uncomment code if you are paranoid
151 if ( ( $msgid =~ m/\. *(\|.*)?$/ ) != ( $msgstr =~ m/\. *(\|.*)?$/ ) ) {
152 print "Line $linenum: Missing or unexpected period:\n '$msgid' => '$msgstr'\n"
154 ++$bad{"Bad periods"};
160 # Check space at the end of a message
161 if ( ( $msgid =~ m/ *?(\|.*)?$/ ) != ( $msgstr =~ m/ *?(\|.*)?$/ ) ) {
162 print "Line $linenum: Missing or unexpected space:\n '$msgid' => '$msgstr'\n"
164 ++$bad{"Bad spaces"};
170 # Check for "&" shortcuts
171 if ( ( $msgid =~ m/&[^ ]/ ) != ( $msgstr =~ m/&[^ ]/ ) ) {
172 print "Line $linenum: Missing or unexpected Qt shortcut:\n '$msgid' => '$msgstr'\n"
174 ++$bad{"Bad Qt shortcuts"};
180 # Check for "|..." shortcuts
181 if ( ( $msgid =~ m/\|[^ ]/ ) != ( $msgstr =~ m/\|[^ ]/ ) ) {
182 print "Line $linenum: Missing or unexpected menu shortcut:\n '$msgid' => '$msgstr'\n"
184 ++$bad{"Bad menu shortcuts"};
189 next unless $check_trans;
191 # we now collect these translations in a hash.
192 # this will allow us to check below if we have translated
193 # anything more than one way.
194 my $msgid_clean = lc($msgid);
195 my $msgstr_clean = lc($msgstr);
197 $msgid_clean =~ s/(.*)\|.*?$/$1/; # strip menu shortcuts
198 $msgstr_clean =~ s/(.*)\|.*?$/$1/;
199 $msgid_clean =~ s/&([^ ])/$1/; # strip Qt shortcuts
200 $msgstr_clean =~ s/&([^ ])/$1/;
202 # this is a hash of hashes. the keys of the outer hash are
203 # cleaned versions of ORIGINAL strings. the keys of the inner hash
204 # are the cleaned versions of their TRANSLATIONS. The value for the
205 # inner hash is an array of the orignal string and translation.
206 $trans{$msgid_clean}{$msgstr_clean} = [ $msgid, $msgstr, $linenum ];
210 foreach $msgid ( keys %trans ) {
211 # so $ref is a reference to the inner hash.
212 my $ref = $trans{$msgid};
213 # @msgstrkeys is an array of the keys of that inner hash.
214 my @msgstrkeys = keys %$ref;
216 # do we have more than one such key?
217 if ( $#msgstrkeys > 0 ) {
219 print "Different translations for '$msgid':\n";
220 foreach $msgstr ( @msgstrkeys ) {
221 print "Line $ref->{$msgstr}[2]: '" .
222 $ref->{$msgstr}[0] . "' => '" .
223 $ref->{$msgstr}[1] . "'\n";
226 ++$bad{"Inconsistent translations"};
233 while (my ($k, $v) = each %bad) { print "$k: $v\n"; }
234 if (scalar(keys %bad) > 1) {
235 print "Total warnings: $warn\n";
238 print "No warnings!\n";