Parent Directory
|
Revision Log
++ ChangeLog 23 Nov 2007 06:36:14 -0000 2007-11-23 Wakaba <wakaba@suika.fam.cx> * error-description-source.xml: New error descriptions. * cc.cgi: |Accept-Encoding: *; q=0| is a request for server to send a 406 page! |identity| encoding is added.
| 1 | wakaba | 1.1 | #!/usr/bin/perl |
| 2 | use strict; | ||
| 3 | wakaba | 1.23 | use utf8; |
| 4 | wakaba | 1.1 | |
| 5 | use lib qw[/home/httpd/html/www/markup/html/whatpm | ||
| 6 | wakaba | 1.16 | /home/wakaba/work/manakai2/lib]; |
| 7 | wakaba | 1.1 | use CGI::Carp qw[fatalsToBrowser]; |
| 8 | wakaba | 1.2 | use Scalar::Util qw[refaddr]; |
| 9 | wakaba | 1.16 | use Time::HiRes qw/time/; |
| 10 | wakaba | 1.1 | |
| 11 | wakaba | 1.2 | sub htescape ($) { |
| 12 | my $s = $_[0]; | ||
| 13 | $s =~ s/&/&/g; | ||
| 14 | $s =~ s/</</g; | ||
| 15 | $s =~ s/>/>/g; | ||
| 16 | $s =~ s/"/"/g; | ||
| 17 | wakaba | 1.12 | $s =~ s{([\x00-\x09\x0B-\x1F\x7F-\xA0\x{FEFF}\x{FFFC}-\x{FFFF}])}{ |
| 18 | sprintf '<var>U+%04X</var>', ord $1; | ||
| 19 | }ge; | ||
| 20 | wakaba | 1.2 | return $s; |
| 21 | } # htescape | ||
| 22 | |||
| 23 | wakaba | 1.16 | use Message::CGI::HTTP; |
| 24 | my $http = Message::CGI::HTTP->new; | ||
| 25 | wakaba | 1.1 | |
| 26 | wakaba | 1.16 | if ($http->get_meta_variable ('PATH_INFO') ne '/') { |
| 27 | wakaba | 1.8 | print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400"; |
| 28 | exit; | ||
| 29 | } | ||
| 30 | |||
| 31 | wakaba | 1.12 | binmode STDOUT, ':utf8'; |
| 32 | wakaba | 1.14 | $| = 1; |
| 33 | wakaba | 1.12 | |
| 34 | wakaba | 1.9 | require Message::DOM::DOMImplementation; |
| 35 | my $dom = Message::DOM::DOMImplementation->new; | ||
| 36 | |||
| 37 | wakaba | 1.7 | load_text_catalog ('en'); ## TODO: conneg |
| 38 | |||
| 39 | wakaba | 1.3 | my @nav; |
| 40 | wakaba | 1.2 | print STDOUT qq[Content-Type: text/html; charset=utf-8 |
| 41 | |||
| 42 | <!DOCTYPE html> | ||
| 43 | <html lang="en"> | ||
| 44 | <head> | ||
| 45 | <title>Web Document Conformance Checker (BETA)</title> | ||
| 46 | wakaba | 1.3 | <link rel="stylesheet" href="../cc-style.css" type="text/css"> |
| 47 | wakaba | 1.2 | </head> |
| 48 | <body> | ||
| 49 | wakaba | 1.13 | <h1><a href="../cc-interface">Web Document Conformance Checker</a> |
| 50 | (<em>beta</em>)</h1> | ||
| 51 | wakaba | 1.14 | ]; |
| 52 | wakaba | 1.2 | |
| 53 | wakaba | 1.14 | $| = 0; |
| 54 | my $input = get_input_document ($http, $dom); | ||
| 55 | wakaba | 1.16 | my $char_length = 0; |
| 56 | my %time; | ||
| 57 | wakaba | 1.14 | |
| 58 | print qq[ | ||
| 59 | wakaba | 1.4 | <div id="document-info" class="section"> |
| 60 | wakaba | 1.2 | <dl> |
| 61 | wakaba | 1.9 | <dt>Request URI</dt> |
| 62 | <dd><code class="URI" lang=""><<a href="@{[htescape $input->{request_uri}]}">@{[htescape $input->{request_uri}]}</a>></code></dd> | ||
| 63 | wakaba | 1.2 | <dt>Document URI</dt> |
| 64 | wakaba | 1.25 | <dd><code class="URI" lang=""><<a href="@{[htescape $input->{uri}]}" id=anchor-document-uri>@{[htescape $input->{uri}]}</a>></code> |
| 65 | <script> | ||
| 66 | document.title = '<' | ||
| 67 | + document.getElementById ('anchor-document-uri').href + '> \\u2014 ' | ||
| 68 | + document.title; | ||
| 69 | </script></dd> | ||
| 70 | wakaba | 1.2 | ]; # no </dl> yet |
| 71 | wakaba | 1.3 | push @nav, ['#document-info' => 'Information']; |
| 72 | wakaba | 1.1 | |
| 73 | wakaba | 1.9 | if (defined $input->{s}) { |
| 74 | wakaba | 1.16 | $char_length = length $input->{s}; |
| 75 | wakaba | 1.9 | |
| 76 | print STDOUT qq[ | ||
| 77 | <dt>Base URI</dt> | ||
| 78 | <dd><code class="URI" lang=""><<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>></code></dd> | ||
| 79 | <dt>Internet Media Type</dt> | ||
| 80 | <dd><code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code> | ||
| 81 | wakaba | 1.25 | @{[$input->{media_type_overridden} ? '<em>(overridden)</em>' : defined $input->{official_type} ? $input->{media_type} eq $input->{official_type} ? '' : '<em>(sniffed; official type is: <code class=MIME lang=en>'.htescape ($input->{official_type}).'</code>)' : '<em>(sniffed)</em>']}</dd> |
| 82 | wakaba | 1.9 | <dt>Character Encoding</dt> |
| 83 | <dd>@{[defined $input->{charset} ? '<code class="charset" lang="en">'.htescape ($input->{charset}).'</code>' : '(none)']} | ||
| 84 | @{[$input->{charset_overridden} ? '<em>(overridden)</em>' : '']}</dd> | ||
| 85 | wakaba | 1.16 | <dt>Length</dt> |
| 86 | <dd>$char_length byte@{[$char_length == 1 ? '' : 's']}</dd> | ||
| 87 | wakaba | 1.9 | </dl> |
| 88 | </div> | ||
| 89 | ]; | ||
| 90 | |||
| 91 | wakaba | 1.20 | my $result = {conforming_min => 1, conforming_max => 1}; |
| 92 | wakaba | 1.19 | print_http_header_section ($input, $result); |
| 93 | wakaba | 1.9 | |
| 94 | wakaba | 1.1 | my $doc; |
| 95 | my $el; | ||
| 96 | wakaba | 1.22 | my $manifest; |
| 97 | wakaba | 1.1 | |
| 98 | wakaba | 1.9 | if ($input->{media_type} eq 'text/html') { |
| 99 | wakaba | 1.19 | ($doc, $el) = print_syntax_error_html_section ($input, $result); |
| 100 | wakaba | 1.25 | print_source_string_section |
| 101 | (\($input->{s}), $input->{charset} || $doc->input_encoding); | ||
| 102 | wakaba | 1.9 | } elsif ({ |
| 103 | 'text/xml' => 1, | ||
| 104 | wakaba | 1.16 | 'application/atom+xml' => 1, |
| 105 | 'application/rss+xml' => 1, | ||
| 106 | 'application/svg+xml' => 1, | ||
| 107 | wakaba | 1.9 | 'application/xhtml+xml' => 1, |
| 108 | 'application/xml' => 1, | ||
| 109 | }->{$input->{media_type}}) { | ||
| 110 | wakaba | 1.19 | ($doc, $el) = print_syntax_error_xml_section ($input, $result); |
| 111 | wakaba | 1.9 | print_source_string_section (\($input->{s}), $doc->input_encoding); |
| 112 | wakaba | 1.22 | } elsif ($input->{media_type} eq 'text/cache-manifest') { |
| 113 | ## TODO: MUST be text/cache-manifest | ||
| 114 | $manifest = print_syntax_error_manifest_section ($input, $result); | ||
| 115 | print_source_string_section (\($input->{s}), 'utf-8'); | ||
| 116 | wakaba | 1.2 | } else { |
| 117 | wakaba | 1.9 | ## TODO: Change HTTP status code?? |
| 118 | wakaba | 1.24 | print_result_unknown_type_section ($input, $result); |
| 119 | wakaba | 1.2 | } |
| 120 | |||
| 121 | if (defined $doc or defined $el) { | ||
| 122 | wakaba | 1.22 | print_structure_dump_dom_section ($doc, $el); |
| 123 | my $elements = print_structure_error_dom_section ($doc, $el, $result); | ||
| 124 | wakaba | 1.18 | print_table_section ($elements->{table}) if @{$elements->{table}}; |
| 125 | print_id_section ($elements->{id}) if keys %{$elements->{id}}; | ||
| 126 | print_term_section ($elements->{term}) if keys %{$elements->{term}}; | ||
| 127 | print_class_section ($elements->{class}) if keys %{$elements->{class}}; | ||
| 128 | wakaba | 1.22 | } elsif (defined $manifest) { |
| 129 | print_structure_dump_manifest_section ($manifest); | ||
| 130 | print_structure_error_manifest_section ($manifest, $result); | ||
| 131 | wakaba | 1.1 | } |
| 132 | |||
| 133 | wakaba | 1.19 | print_result_section ($result); |
| 134 | wakaba | 1.9 | } else { |
| 135 | wakaba | 1.18 | print STDOUT qq[</dl></div>]; |
| 136 | print_result_input_error_section ($input); | ||
| 137 | wakaba | 1.9 | } |
| 138 | wakaba | 1.3 | |
| 139 | wakaba | 1.2 | print STDOUT qq[ |
| 140 | wakaba | 1.3 | <ul class="navigation" id="nav-items"> |
| 141 | ]; | ||
| 142 | for (@nav) { | ||
| 143 | print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>]; | ||
| 144 | } | ||
| 145 | print STDOUT qq[ | ||
| 146 | </ul> | ||
| 147 | wakaba | 1.2 | </body> |
| 148 | </html> | ||
| 149 | ]; | ||
| 150 | wakaba | 1.1 | |
| 151 | wakaba | 1.24 | for (qw/decode parse parse_html parse_xml parse_manifest |
| 152 | check check_manifest/) { | ||
| 153 | wakaba | 1.16 | next unless defined $time{$_}; |
| 154 | open my $file, '>>', ".cc-$_.txt" or die ".cc-$_.txt: $!"; | ||
| 155 | print $file $char_length, "\t", $time{$_}, "\n"; | ||
| 156 | } | ||
| 157 | |||
| 158 | wakaba | 1.1 | exit; |
| 159 | |||
| 160 | wakaba | 1.19 | sub add_error ($$$) { |
| 161 |