Parent Directory
|
Revision Log
++ ChangeLog 30 Jun 2007 14:51:07 -0000 * cc.cgi: "Terms" section. "Tree" was sometimes non-conforming to HTML5. (get_node_link): New function. 2007-06-30 Wakaba <wakaba@suika.fam.cx>
| 1 | wakaba | 1.1 | #!/usr/bin/perl |
| 2 | use strict; | ||
| 3 | |||
| 4 | use lib qw[/home/httpd/html/www/markup/html/whatpm | ||
| 5 | /home/wakaba/work/manakai/lib | ||
| 6 | /home/wakaba/public_html/-temp/wiki/lib]; | ||
| 7 | use CGI::Carp qw[fatalsToBrowser]; | ||
| 8 | wakaba | 1.2 | use Scalar::Util qw[refaddr]; |
| 9 | wakaba | 1.1 | |
| 10 | use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module | ||
| 11 | |||
| 12 | wakaba | 1.2 | sub htescape ($) { |
| 13 | my $s = $_[0]; | ||
| 14 | $s =~ s/&/&/g; | ||
| 15 | $s =~ s/</</g; | ||
| 16 | $s =~ s/>/>/g; | ||
| 17 | $s =~ s/"/"/g; | ||
| 18 | $s =~ s!([\x00-\x09\x0B-\x1F\x7F-\x80])!sprintf '<var>U+%04X</var>', ord $1!ge; | ||
| 19 | return $s; | ||
| 20 | } # htescape | ||
| 21 | |||
| 22 | wakaba | 1.1 | my $http = SuikaWiki::Input::HTTP->new; |
| 23 | |||
| 24 | ## TODO: _charset_ | ||
| 25 | |||
| 26 | wakaba | 1.2 | my $input_format = $http->parameter ('i') || 'text/html'; |
| 27 | my $inner_html_element = $http->parameter ('e'); | ||
| 28 | my $input_uri = 'thismessage:/'; | ||
| 29 | wakaba | 1.1 | |
| 30 | my $s = $http->parameter ('s'); | ||
| 31 | if (length $s > 1000_000) { | ||
| 32 | print STDOUT "Status: 400 Document Too Long\nContent-Type: text/plain; charset=us-ascii\n\nToo long"; | ||
| 33 | exit; | ||
| 34 | } | ||
| 35 | wakaba | 1.2 | |
| 36 | wakaba | 1.3 | my @nav; |
| 37 | wakaba | 1.2 | print STDOUT qq[Content-Type: text/html; charset=utf-8 |
| 38 | |||
| 39 | <!DOCTYPE html> | ||
| 40 | <html lang="en"> | ||
| 41 | <head> | ||
| 42 | <title>Web Document Conformance Checker (BETA)</title> | ||
| 43 | wakaba | 1.3 | <link rel="stylesheet" href="../cc-style.css" type="text/css"> |
| 44 | wakaba | 1.2 | </head> |
| 45 | <body> | ||
| 46 | <h1>Web Document Conformance Checker (<em>beta</em>)</h1> | ||
| 47 | |||
| 48 | wakaba | 1.4 | <div id="document-info" class="section"> |
| 49 | wakaba | 1.2 | <dl> |
| 50 | <dt>Document URI</dt> | ||
| 51 | <dd><code class="URI" lang=""><<a href="@{[htescape $input_uri]}">@{[htescape $input_uri]}</a>></code></dd> | ||
| 52 | <dt>Internet Media Type</dt> | ||
| 53 | <dd><code class="MIME" lang="en">@{[htescape $input_format]}</code></dd> | ||
| 54 | ]; # no </dl> yet | ||
| 55 | wakaba | 1.3 | push @nav, ['#document-info' => 'Information']; |
| 56 | wakaba | 1.1 | |
| 57 | require Message::DOM::DOMImplementation; | ||
| 58 | my $dom = Message::DOM::DOMImplementation->____new; | ||
| 59 | my $doc; | ||
| 60 | my $el; | ||
| 61 | |||
| 62 | wakaba | 1.2 | if ($input_format eq 'text/html') { |
| 63 | require Encode; | ||
| 64 | require Whatpm::HTML; | ||
| 65 | |||
| 66 | $s = Encode::decode ('utf-8', $s); | ||
| 67 | |||
| 68 | print STDOUT qq[ | ||
| 69 | <dt>Character Encoding</dt> | ||
| 70 | <dd>(none)</dd> | ||
| 71 | </dl> | ||
| 72 | wakaba | 1.3 | </div> |
| 73 | wakaba | 1.2 | |
| 74 | <div id="source-string" class="section"> | ||
| 75 | wakaba | 1.3 | <h2>Document Source</h2> |
| 76 | wakaba | 1.2 | ]; |
| 77 | wakaba | 1.3 | push @nav, ['#source-string' => 'Source']; |
| 78 | wakaba | 1.2 | print_source_string (\$s); |
| 79 | print STDOUT qq[ | ||
| 80 | </div> | ||
| 81 | wakaba | 1.1 | |
| 82 | wakaba | 1.2 | <div id="parse-errors" class="section"> |
| 83 | <h2>Parse Errors</h2> | ||
| 84 | |||
| 85 | wakaba | 1.4 | <dl> |
| 86 | wakaba | 1.2 | ]; |
| 87 | wakaba | 1.3 | push @nav, ['#parse-errors' => 'Parse Error']; |
| 88 | wakaba | 1.1 | |
| 89 | my $onerror = sub { | ||
| 90 | my (%opt) = @_; | ||
| 91 | wakaba | 1.2 | if ($opt{column} > 0) { |
| 92 | wakaba | 1.4 | print STDOUT qq[<dt><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n]; |
| 93 | wakaba | 1.2 | } else { |
| 94 | $opt{line}--; | ||
| 95 | wakaba | 1.4 | print STDOUT qq[<dt><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n]; |
| 96 | wakaba | 1.2 | } |
| 97 | wakaba | 1.4 | print STDOUT qq[<dd>@{[htescape $opt{type}]}</dd>\n]; |
| 98 | wakaba | 1.1 | }; |
| 99 | |||
| 100 | $doc = $dom->create_document; | ||
| 101 | wakaba | 1.2 | if (defined $inner_html_element and length $inner_html_element) { |
| 102 | wakaba | 1.1 | $el = $doc->create_element_ns |
| 103 | wakaba | 1.2 | ('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]); |
| 104 | wakaba | 1.1 | Whatpm::HTML->set_inner_html ($el, $s, $onerror); |
| 105 | } else { | ||
| 106 | Whatpm::HTML->parse_string ($s => $doc, $onerror); | ||
| 107 | } | ||
| 108 | |||
| 109 | wakaba | 1.2 | print STDOUT qq[ |
| 110 | wakaba | 1.4 | </dl> |
| 111 | wakaba | 1.2 | </div> |
| 112 | ]; | ||
| 113 | } elsif ($input_format eq 'application/xhtml+xml') { | ||
| 114 | require Message::DOM::XMLParserTemp; | ||
| 115 | require Encode; | ||
| 116 | |||
| 117 | my $t = Encode::decode ('utf-8', $s); | ||
| 118 | |||
| 119 | print STDOUT qq[ | ||
| 120 | <dt>Character Encoding</dt> | ||
| 121 | <dd>(none)</dd> | ||
| 122 | </dl> | ||
| 123 | wakaba | 1.3 | </div> |
| 124 | wakaba | 1.2 | |
| 125 | <div id="source-string" class="section"> | ||
| 126 | wakaba | 1.3 | <h2>Document Source</h2> |
| 127 | wakaba | 1.2 | ]; |
| 128 | wakaba | 1.3 | push @nav, ['#source-string' => 'Source']; |
| 129 | wakaba | 1.2 | print_source_string (\$t); |
| 130 | print STDOUT qq[ | ||
| 131 | </div> | ||
| 132 | wakaba | 1.1 | |
| 133 | wakaba | 1.2 | <div id="parse-errors" class="section"> |
| 134 | <h2>Parse Errors</h2> | ||
| 135 | wakaba | 1.1 | |
| 136 | wakaba | 1.4 | <dl> |
| 137 | wakaba | 1.2 | ]; |
| 138 | wakaba | 1.3 | push @nav, ['#parse-errors' => 'Parse Error']; |
| 139 | wakaba | 1.1 | |
| 140 | my $onerror = sub { | ||
| 141 | my $err = shift; | ||
| 142 | wakaba | 1.2 | my $line = $err->location->line_number; |
| 143 | wakaba | 1.4 | print STDOUT qq[<dt><a href="#line-$line">Line $line</a> column ]; |
| 144 | print STDOUT $err->location->column_number, "</dt><dd>"; | ||
| 145 | print STDOUT htescape $err->text, "</dd>\n"; | ||
| 146 | wakaba | 1.1 | return 1; |
| 147 | }; | ||
| 148 | |||
| 149 | open my $fh, '<', \$s; | ||
| 150 | $doc = Message::DOM::XMLParserTemp->parse_byte_stream | ||
| 151 | ($fh => $dom, $onerror, charset => 'utf-8'); | ||
| 152 | |||
| 153 | wakaba | 1.2 | print STDOUT qq[ |
| 154 | wakaba | 1.4 | </dl> |
| 155 | wakaba | 1.2 | </div> |
| 156 | ]; | ||
| 157 | } else { | ||
| 158 | print STDOUT qq[ | ||
| 159 | </dl> | ||
| 160 | wakaba | 1.4 | </div> |
| 161 | wakaba | 1.2 | |
| 162 | wakaba | 1.3 | <div id="result-summary" class="section"> |
| 163 | wakaba | 1.2 | <p><em>Media type <code class="MIME" lang="en">@{[htescape $input_format]}</code> is not supported!</em></p> |
| 164 | wakaba | 1.3 | </div> |
| 165 | wakaba | 1.2 | ]; |
| 166 | wakaba | 1.3 | push @nav, ['#result-summary' => 'Result']; |
| 167 | wakaba | 1.2 | } |
| 168 | |||
| 169 | |||
| 170 | if (defined $doc or defined $el) { | ||
| 171 | print STDOUT qq[ | ||
| 172 | <div id="document-tree" class="section"> | ||
| 173 | <h2>Document Tree</h2> | ||
| 174 | ]; | ||
| 175 | wakaba | 1.3 | push @nav, ['#document-tree' => 'Tree']; |
| 176 | wakaba | 1.2 | |
| 177 | print_document_tree ($el || $doc); | ||
| 178 | |||
| 179 | print STDOUT qq[ | ||
| 180 | </div> | ||
| 181 | wakaba | 1.1 | |
| 182 | wakaba | 1.2 | <div id="document-errors" class="section"> |
| 183 | <h2>Document Errors</h2> | ||
| 184 | |||
| 185 | wakaba | 1.4 | <dl> |
| 186 | wakaba | 1.2 | ]; |
| 187 | wakaba | 1.3 | push @nav, ['#document-errors' => 'Document Error']; |
| 188 | wakaba | 1.1 | |
| 189 | require Whatpm::ContentChecker; | ||
| 190 | my $onerror = sub { | ||
| 191 | my %opt = @_; | ||
| 192 | wakaba | 1.6 | print STDOUT qq[<dt>] . get_node_link ($opt{node}) . |
| 193 | "</dt>\n<dd>", htescape $opt{type}, "</dd>\n"; | ||
| 194 | wakaba | 1.1 | }; |
| 195 | wakaba | 1.2 | |
| 196 | wakaba | 1.5 | my $elements; |
| 197 | wakaba | 1.1 | if ($el) { |
| 198 | wakaba | 1.5 | $elements = Whatpm::ContentChecker->check_element ($el, $onerror); |
| 199 | wakaba | 1.1 | } else { |
| 200 | wakaba | 1.5 | $elements = Whatpm::ContentChecker->check_document ($doc, $onerror); |
| 201 | wakaba | 1.1 | } |
| 202 | wakaba | 1.2 | |
| 203 | print STDOUT qq[ | ||
| 204 | wakaba | 1.4 | </dl> |
| 205 | wakaba | 1.2 | </div> |
| 206 | ]; | ||
| 207 | wakaba | 1.5 | |
| 208 | if (@{$elements->{table}}) { | ||
| 209 | require JSON; | ||
| 210 | |||
| 211 | print STDOUT qq[ | ||
| 212 | <div id="tables" class="section"> | ||
| 213 | <h2>Tables</h2> | ||
| 214 | |||
| 215 | <!--[if IE]><script type="text/javascript" src="../excanvas.js"></script><![endif]--> | ||
| 216 | <script src="../table-script.js" type="text/javascript"></script> | ||
| 217 | <noscript> | ||
| 218 | <p><em>Structure of tables are visualized here if scripting is enabled.</em></p> | ||
| 219 | </noscript> | ||
| 220 | ]; | ||
| 221 | |||
| 222 | my $i = 0; | ||
| 223 | for my $table_el (@{$elements->{table}}) { | ||
| 224 | $i++; | ||
| 225 | wakaba | 1.6 | print STDOUT qq[<div class="section" id="table-$i"><h3>] . |
| 226 | get_node_link ($table_el) . q[</h3>]; | ||
| 227 | wakaba | 1.5 | |
| 228 | my $table = Whatpm::HTMLTable->form_table ($table_el); | ||
| 229 | |||
| 230 | for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) { | ||
| 231 | next unless $_; | ||
| 232 | delete $_->{element}; | ||
| 233 | } | ||
| 234 | |||
| 235 | for (@{$table->{row_group}}) { | ||
| 236 | next unless $_; | ||
| 237 | next unless $_->{element}; | ||
| 238 | $_->{type} = $_->{element}->manakai_local_name; | ||
| 239 | delete $_->{element}; | ||
| 240 | } | ||
| 241 | |||
| 242 | for (@{$table->{cell}}) { | ||
| 243 | next unless $_; | ||
| 244 | for (@{$_}) { | ||
| 245 | next unless $_; | ||
| 246 | for (@$_) { | ||
| 247 | $_->{id} = refaddr $_->{element} if defined $_->{element}; | ||
| 248 | delete $_->{element}; | ||
| 249 | } | ||
| 250 | } | ||
| 251 | } | ||
| 252 | |||
| 253 | print STDOUT '</div><script type="text/javascript">tableToCanvas ('; | ||
| 254 | print STDOUT JSON::objToJson ($table); | ||
| 255 | print STDOUT qq[, document.getElementById ('table-$i'));</script>]; | ||
| 256 | } | ||
| 257 | |||
| 258 | print STDOUT qq[</div>]; | ||
| 259 | } | ||
| 260 | wakaba | 1.6 | |
| 261 | if (keys %{$elements->{term}}) { | ||
| 262 | print STDOUT qq[ | ||
| 263 | <div id="terms" class="section"> | ||
| 264 | <h2>Terms</h2> | ||
| 265 | |||
| 266 | & |