Parent Directory
| 
Revision Log
++ ChangeLog 20 May 2007 08:14:29 -0000 2007-05-19 Wakaba <wakaba@suika.fam.cx> * parser-interface.en.html: A checkbox to enable DOM5 HTML conformance checking is added. * parser.cgi: Check DOM5 HTML conformance if |dom5| parameter is specified.
| 1 | wakaba | 1.1 | #!/usr/bin/perl |
| 2 | use strict; | ||
| 3 | |||
| 4 | wakaba | 1.2 | use lib qw[/home/httpd/html/www/markup/html/whatpm |
| 5 | /home/wakaba/public_html/-temp/wiki/lib]; | ||
| 6 | wakaba | 1.1 | |
| 7 | use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module | ||
| 8 | |||
| 9 | my $http = SuikaWiki::Input::HTTP->new; | ||
| 10 | |||
| 11 | ## TODO: _charset_ | ||
| 12 | |||
| 13 | my $mode = $http->meta_variable ('PATH_INFO'); | ||
| 14 | ## TODO: decode unreserved characters | ||
| 15 | |||
| 16 | wakaba | 1.2 | if ($mode eq '/html' or $mode eq '/test') { |
| 17 | wakaba | 1.1 | require Encode; |
| 18 | wakaba | 1.3 | require Whatpm::HTML; |
| 19 | require Whatpm::NanoDOM; | ||
| 20 | wakaba | 1.1 | |
| 21 | my $s = $http->parameter ('s'); | ||
| 22 | if (length $s > 1000_000) { | ||
| 23 | print STDOUT "Status: 400 Document Too Long\nContent-Type: text/plain; charset=us-ascii\n\nToo long"; | ||
| 24 | exit; | ||
| 25 | } | ||
| 26 | |||
| 27 | $s = Encode::decode ('utf-8', $s); | ||
| 28 | |||
| 29 | print STDOUT "Content-Type: text/plain; charset=utf-8\n\n"; | ||
| 30 | |||
| 31 | wakaba | 1.2 | print STDOUT "#errors\n"; |
| 32 | |||
| 33 | my $onerror = sub { | ||
| 34 | wakaba | 1.4 | my (%opt) = @_; |
| 35 | print STDOUT "$opt{line},$opt{column},$opt{type}\n"; | ||
| 36 | wakaba | 1.2 | }; |
| 37 | wakaba | 1.1 | |
| 38 | wakaba | 1.3 | my $doc = Whatpm::HTML->parse_string |
| 39 | ($s => Whatpm::NanoDOM::Document->new, $onerror); | ||
| 40 | wakaba | 1.1 | |
| 41 | wakaba | 1.2 | print "#document\n"; |
| 42 | |||
| 43 | my $out; | ||
| 44 | if ($mode eq '/html') { | ||
| 45 | wakaba | 1.3 | $out = Whatpm::HTML->get_inner_html ($doc); |
| 46 | wakaba | 1.2 | } else { # test |
| 47 | $out = test_serialize ($doc); | ||
| 48 | wakaba | 1.1 | } |
| 49 | wakaba | 1.2 | print STDOUT Encode::encode ('utf-8', $$out); |
| 50 | wakaba | 1.4 | print STDOUT "\n"; |
| 51 | |||
| 52 | if ($http->parameter ('dom5')) { | ||
| 53 | require Whatpm::ContentChecker; | ||
| 54 | print STDOUT "#domerrors\n"; | ||
| 55 |