Cross Reference: /inkscape/po/check-markup

	check-markup revision 75a68d6ceb269d78054ba138a5d8fd6b5a37c9b8
#! /usr/bin/perl -w
# Try to detect markup errors in translations.

# Author: Peter Moulder <pmoulder@mail.csse.monash.edu.au>
# Copyright (C) 2004 Monash University
# License: GNU GPL v2 or (at your option) any later version.

# Initial egrep version:
#mydir=`dirname "$0"`
#egrep '<b>[^<>]*(>|<([^/]|/([^b"]|b[^>])))' "$mydir"/*.po
# Somewhat simplified by use of negative lookahead in perl.
# (The egrep version as written can't detect problems that span a line,
# e.g. unterminated `<b>'.  One way of doing the s/"\n"//g thing would be with
# tr and sed, but that requires a sed that allows arbitrary line lengths, which
# many non-GNU seds don't.)

use strict;

my $com = qr/(?:\#[^\n]*\n)/;
my $str = qr/(?:"(?:[^"\\]|\\.)*")/;
my $attrsRE = qr/(?: +[^<>]*)?/;
my $span_attr = qr/(?:\ +(?:font_(?:desc|family)|face|size|style|weight|variant|stretch|(?:fore|back)ground|underline|rise|strikethrough|fallback|lang)\=\\\"[^\\\"]*\\\")/;

my $rc = 0;

sub po_error ($) {
    my ($msg) = @_;
    my $name = $ARGV;
    $name =~ s,.*/,,;
    print "$name: $msg:\n$_";
    $rc = 1;
}

# Returns true iff successful.
sub check_str ($) {
    my ($str) = @_;

    $str =~ s/\A"// or die "Bug: No leading `\"' in `$str'";
    $str =~ s/"\Z// or die "Bug: No trailing `\"' in `$str'";

    if ($str =~ /\AProject-Id-Version:.*POT-Creation-Date/
        or $str =~ /\A<[^<>]*>\Z/) {
    # Not a Pango string.
    return 1;
    }

    my $is_xml = 0;

    # Remove valid sequences.
    while ($str =~ s{<([bisu]|big|su[bp]|small|tt|span)(${attrsRE})>[^<>]*</\1>}{}) {
    $is_xml = 1;
    my ($tag, $attrs) = ($1, $2);
    if ($tag eq 'span') {
        $attrs =~ s/${span_attr}*//g;
        if ($attrs ne '') {
        $attrs =~ s/\A *//;
        $attrs =~ s/\\"/"/g;
        po_error("Unexpected <span> attributes `$attrs'");
        return 0;
        }
    } else {
        if ($attrs !~ /\A *\Z/) {
        po_error("<$tag> can't have attributes in Pango");
        return 0;
        }
    }
    }

    if ($str =~ m{&#[^0-9]+;}) {
    po_error("Entity declaration error (must look like '&#123;')");
    return 0;
    }

    # Check for attributes etc. in non-<span> element.
    if ($str =~ m{<([bisu]|big|su[bp]|small|tt)\b(?! *)>}) {
    po_error("Unexpected characters in <$1> tag");
    return 0;
    }

    if ($str =~ m{<([bisu]|big|su[bp]|small|span|tt)${attrsRE}>}) {
    po_error("unclosed <$1>");
    return 0;
    }

    if ($str =~ m{</\ *([bisu]|big|su[bp]|small|span|tt)\ *>}) {
    po_error("Unmatched closing </$1>");
    return 0;
    }

    if (!$is_xml) {
    $str =~ s/<(?:defs|image|rect|svg)>//g;
    $str =~ s/<[ 01]//g;
    $str =~ s/\A>+//;
    $str =~ s/<+\Z//;
    $str =~ s/\([<>][01]\)//g;
    $str =~ s/ -> //g;

    # Quoting.
    $str =~ s/\[[<>]\]//g;
    $str =~ s/\\"[<>]\\"//g;
    $str =~ s/\xe2\x80\x9e[<>]\xe2\x80\x9c//g;
    $str =~ s/\xc2\xab[<>]\xc2\xbb//;
    }

    $str =~ s/\A[^<>]*//;
    $str =~ s/[^<>]*\Z//;

    if ($str =~ /\A([<>])\Z/) {
    if ($is_xml) {
        po_error("Unescaped `$1'");
        return 0;
    } else {
        return 1;
    }
    }

    if ($str ne '') {
    po_error("parsing error for `$str'");
    return 0;
    }

    return 1;
}

sub check_strs (@) {
    if ($#_ < 1) {
    die "check_strs: expecting >= 2 strings";
    }
    if ((($_[0] eq '""') && ($_[1] =~ /Project-Id-Version:.*POT-Creation-Date:/s))
    or ($_[0] eq '"> and < scale by:"')) {
    # Not a Pango string.
    return 1;
    }
    foreach my $str (@_) {
    $str eq '""' or check_str($str) or return 0;
    }
    return 1;
}

$/ = '';

# Reference for the markup language:
# http://developer.gnome.org/doc/API/2.0/pango/PangoMarkupFormat.html
# (though not all translation strings will be pango markup strings).
ENTRY: while(<>) {
    if (m{\A${com}*\Z}) {
        next ENTRY;
    }

    s/"\n"//g;

    if (!m{\A${com}*msgid[^\n]*\n${com}*msgstr[^\n]*\n${com}*\Z} &&
        !m{\A${com}*msgid[^\n]*\n${com}*msgid_plural[^\n]*\n${com}*(msgstr\[[^\n]*\n${com}*)+\Z}) {
        po_error('Not in msg format');
        next ENTRY;
    }
    if (!m{\A${com}*msgid ${str}\s*\n${com}*msgstr ${str}\s*\n${com}*\Z} &&
           !m{\A${com}*msgid ${str}\s*\n${com}*msgid_plural ${str}\s*\n${com}*(msgstr\[\d+\] ${str}\s*\n${com}*)+\Z}) {
        po_error('Mismatched quotes');
        next ENTRY;
    }

    if (m{\n\#,\ fuzzy}) {
        # Fuzzy entries aren't used, so ignore them.
        # (This prevents warnings about mismatching <>/ pattern.)
        next ENTRY;
    }

    if (m{\A${com}*msgid\ (${str})\n
          ${com}*msgstr\ (${str})\n
          ${com}*\Z}x) {
        check_strs($1, $2) or next ENTRY;
    }
    elsif (m{\A${com}*msgid\ (${str})\n
         ${com}*msgid_plural\ (${str})\n
         ((?:${com}*msgstr\[\d+\]\ ${str}\n${com}*)+)\Z}x) {
        my ($s1, $s2, $rest) = ($1, $2, $3);
        my @strs = ($s1, $s2);
        while ($rest =~ s/\A${com}*msgstr\[\d+\]\ (${str})\n${com}*//) {
        push @strs, ($1);
        }
        $rest eq '' or die "BUG: unparsed plural entries `$rest'";
        check_strs(@strs) or next ENTRY;
    }
    elsif (m{$str[ \t]}) {
        po_error('Trailing whitespace');
        next ENTRY;
    } else {
        po_error("parse error; may be a bug in po/check-markup");
    }
}

# Some makefiles (currently the top-level Makefile.am) expect this script to
# exit 1 if any problems found.
exit $rc;