1 #! /usr/bin/perl -w
2 # Try to detect markup errors in translations.
4 # Author: Peter Moulder <pmoulder@mail.csse.monash.edu.au>
5 # Copyright (C) 2004 Monash University
6 # License: GNU GPL v2 or (at your option) any later version.
8 # Initial egrep version:
9 #mydir=`dirname "$0"`
10 #egrep '<b>[^<>]*(>|<([^/]|/([^b"]|b[^>])))' "$mydir"/*.po
11 # Somewhat simplified by use of negative lookahead in perl.
12 # (The egrep version as written can't detect problems that span a line,
13 # e.g. unterminated `<b>'. One way of doing the s/"\n"//g thing would be with
14 # tr and sed, but that requires a sed that allows arbitrary line lengths, which
15 # many non-GNU seds don't.)
17 use strict;
19 my $com = qr/(?:\#[^\n]*\n)/;
20 my $str = qr/(?:"(?:[^"\\]|\\.)*")/;
21 my $attrsRE = qr/(?: +[^<>]*)?/;
22 my $span_attr = qr/(?:\ +(?:font_(?:desc|family)|face|size|style|weight|variant|stretch|(?:fore|back)ground|underline|rise|strikethrough|fallback|lang)\=\\\"[^\\\"]*\\\")/;
24 my $rc = 0;
26 sub po_error ($) {
27 my ($msg) = @_;
28 my $name = $ARGV;
29 $name =~ s,.*/,,;
30 print "$name: $msg:\n$_";
31 $rc = 1;
32 }
34 # Returns true iff successful.
35 sub check_str ($) {
36 my ($str) = @_;
38 $str =~ s/\A"// or die "Bug: No leading `\"' in `$str'";
39 $str =~ s/"\Z// or die "Bug: No trailing `\"' in `$str'";
41 if ($str =~ /\AProject-Id-Version:.*POT-Creation-Date/
42 or $str =~ /\A<[^<>]*>\Z/) {
43 # Not a Pango string.
44 return 1;
45 }
47 my $is_xml = 0;
49 # Remove valid sequences.
50 while ($str =~ s{<([bisu]|big|su[bp]|small|tt|span)(${attrsRE})>[^<>]*</\1>}{}) {
51 $is_xml = 1;
52 my ($tag, $attrs) = ($1, $2);
53 if ($tag eq 'span') {
54 $attrs =~ s/${span_attr}*//g;
55 if ($attrs ne '') {
56 $attrs =~ s/\A *//;
57 $attrs =~ s/\\"/"/g;
58 po_error("Unexpected <span> attributes `$attrs'");
59 return 0;
60 }
61 } else {
62 if ($attrs !~ /\A *\Z/) {
63 po_error("<$tag> can't have attributes in Pango");
64 return 0;
65 }
66 }
67 }
69 # Check for attributes etc. in non-<span> element.
70 if ($str =~ m{<([bisu]|big|su[bp]|small|tt)\b(?! *)>}) {
71 po_error("Unexpected characters in <$1> tag");
72 return 0;
73 }
75 if ($str =~ m{<([bisu]|big|su[bp]|small|span|tt)${attrsRE}>}) {
76 po_error("unclosed <$1>");
77 return 0;
78 }
80 if ($str =~ m{</\ *([bisu]|big|su[bp]|small|span|tt)\ *>}) {
81 po_error("Unmatched closing </$1>");
82 return 0;
83 }
85 if (!$is_xml) {
86 $str =~ s/<(?:defs|image|rect|svg)>//g;
87 $str =~ s/<[ 01]//g;
88 $str =~ s/\A>+//;
89 $str =~ s/<+\Z//;
90 $str =~ s/\([<>][01]\)//g;
91 $str =~ s/ -> //g;
93 # Quoting.
94 $str =~ s/\[[<>]\]//g;
95 $str =~ s/\\"[<>]\\"//g;
96 $str =~ s/\xe2\x80\x9e[<>]\xe2\x80\x9c//g;
97 $str =~ s/\xc2\xab[<>]\xc2\xbb//;
98 }
100 $str =~ s/\A[^<>]*//;
101 $str =~ s/[^<>]*\Z//;
103 if ($str =~ /\A([<>])\Z/) {
104 if ($is_xml) {
105 po_error("Unescaped `$1'");
106 return 0;
107 } else {
108 return 1;
109 }
110 }
112 if ($str ne '') {
113 po_error("parsing error for `$str'");
114 return 0;
115 }
117 return 1;
118 }
120 sub check_strs (@) {
121 if ($#_ < 1) {
122 die "check_strs: expecting >= 2 strings";
123 }
124 if ((($_[0] eq '""') && ($_[1] =~ /Project-Id-Version:.*POT-Creation-Date:/s))
125 or ($_[0] eq '"> and < scale by:"')) {
126 # Not a Pango string.
127 return 1;
128 }
129 foreach my $str (@_) {
130 $str eq '""' or check_str($str) or return 0;
131 }
132 return 1;
133 }
135 $/ = '';
137 # Reference for the markup language:
138 # http://developer.gnome.org/doc/API/2.0/pango/PangoMarkupFormat.html
139 # (though not all translation strings will be pango markup strings).
140 ENTRY: while(<>) {
141 if (m{\A${com}*\Z}) {
142 next ENTRY;
143 }
145 s/"\n"//g;
147 if (!m{\A${com}*msgid[^\n]*\n${com}*msgstr[^\n]*\n${com}*\Z} &&
148 !m{\A${com}*msgid[^\n]*\n${com}*msgid_plural[^\n]*\n${com}*(msgstr\[[^\n]*\n${com}*)+\Z}) {
149 po_error('Not in msg format');
150 next ENTRY;
151 }
152 if (!m{\A${com}*msgid ${str}\s*\n${com}*msgstr ${str}\s*\n${com}*\Z} &&
153 !m{\A${com}*msgid ${str}\s*\n${com}*msgid_plural ${str}\s*\n${com}*(msgstr\[\d+\] ${str}\s*\n${com}*)+\Z}) {
154 po_error('Mismatched quotes');
155 next ENTRY;
156 }
158 if (m{\n\#,\ fuzzy}) {
159 # Fuzzy entries aren't used, so ignore them.
160 # (This prevents warnings about mismatching <>/ pattern.)
161 next ENTRY;
162 }
164 if (m{\A${com}*msgid\ (${str})\n
165 ${com}*msgstr\ (${str})\n
166 ${com}*\Z}x) {
167 check_strs($1, $2) or next ENTRY;
168 }
169 elsif (m{\A${com}*msgid\ (${str})\n
170 ${com}*msgid_plural\ (${str})\n
171 ((?:${com}*msgstr\[\d+\]\ ${str}\n${com}*)+)\Z}x) {
172 my ($s1, $s2, $rest) = ($1, $2, $3);
173 my @strs = ($s1, $s2);
174 while ($rest =~ s/\A${com}*msgstr\[\d+\]\ (${str})\n${com}*//) {
175 push @strs, ($1);
176 }
177 $rest eq '' or die "BUG: unparsed plural entries `$rest'";
178 check_strs(@strs) or next ENTRY;
179 }
180 elsif (m{$str[ \t]}) {
181 po_error('Trailing whitespace');
182 next ENTRY;
183 } else {
184 po_error("parse error; may be a bug in po/check-markup");
185 }
186 }
188 # Some makefiles (currently the top-level Makefile.am) expect this script to
189 # exit 1 if any problems found.
190 exit $rc;