| 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 |
use Scalar::Util qw[refaddr]; |
| 9 |
|
| 10 |
use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module |
| 11 |
|
| 12 |
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 |
my $http = SuikaWiki::Input::HTTP->new; |
| 23 |
|
| 24 |
## TODO: _charset_ |
| 25 |
|
| 26 |
my $input_format = $http->parameter ('i') || 'text/html'; |
| 27 |
my $inner_html_element = $http->parameter ('e'); |
| 28 |
my $input_uri = 'thismessage:/'; |
| 29 |
|
| 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 |
|
| 36 |
my @nav; |
| 37 |
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 |
<link rel="stylesheet" href="../cc-style.css" type="text/css"> |
| 44 |
</head> |
| 45 |
<body> |
| 46 |
<h1>Web Document Conformance Checker (<em>beta</em>)</h1> |
| 47 |
|
| 48 |
<div id="document-info" class="section"> |
| 49 |
<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 |
push @nav, ['#document-info' => 'Information']; |
| 56 |
|
| 57 |
require Message::DOM::DOMImplementation; |
| 58 |
my $dom = Message::DOM::DOMImplementation->____new; |
| 59 |
my $doc; |
| 60 |
my $el; |
| 61 |
|
| 62 |
if ($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 |
</div> |
| 73 |
|
| 74 |
<div id="source-string" class="section"> |
| 75 |
<h2>Document Source</h2> |
| 76 |
]; |
| 77 |
push @nav, ['#source-string' => 'Source']; |
| 78 |
print_source_string (\$s); |
| 79 |
print STDOUT qq[ |
| 80 |
</div> |
| 81 |
|
| 82 |
<div id="parse-errors" class="section"> |
| 83 |
<h2>Parse Errors</h2> |
| 84 |
|
| 85 |
<dl> |
| 86 |
]; |
| 87 |
push @nav, ['#parse-errors' => 'Parse Error']; |
| 88 |
|
| 89 |
my $onerror = sub { |
| 90 |
my (%opt) = @_; |
| 91 |
if ($opt{column} > 0) { |
| 92 |
print STDOUT qq[<dt><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n]; |
| 93 |
} else { |
| 94 |
$opt{line}--; |
| 95 |
print STDOUT qq[<dt><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n]; |
| 96 |
} |
| 97 |
print STDOUT qq[<dd>@{[htescape $opt{type}]}</dd>\n]; |
| 98 |
}; |
| 99 |
|
| 100 |
$doc = $dom->create_document; |
| 101 |
if (defined $inner_html_element and length $inner_html_element) { |
| 102 |
$el = $doc->create_element_ns |
| 103 |
('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]); |
| 104 |
Whatpm::HTML->set_inner_html ($el, $s, $onerror); |
| 105 |
} else { |
| 106 |
Whatpm::HTML->parse_string ($s => $doc, $onerror); |
| 107 |
} |
| 108 |
|
| 109 |
print STDOUT qq[ |
| 110 |
</dl> |
| 111 |
</div> |
| 112 |
]; |
| 113 |
} elsif ($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 |
</div> |
| 124 |
|
| 125 |
<div id="source-string" class="section"> |
| 126 |
<h2>Document Source</h2> |
| 127 |
]; |
| 128 |
push @nav, ['#source-string' => 'Source']; |
| 129 |
print_source_string (\$t); |
| 130 |
print STDOUT qq[ |
| 131 |
</div> |
| 132 |
|
| 133 |
<div id="parse-errors" class="section"> |
| 134 |
<h2>Parse Errors</h2> |
| 135 |
|
| 136 |
<dl> |
| 137 |
]; |
| 138 |
push @nav, ['#parse-errors' => 'Parse Error']; |
| 139 |
|
| 140 |
my $onerror = sub { |
| 141 |
my $err = shift; |
| 142 |
my $line = $err->location->line_number; |
| 143 |
print STDOUT qq[<dt><a href="#line-$line">Line $line</a> column ]; |
| 144 |
print STDOUT $err->location->column_number, "</dt><dd>"; |
| 145 |
print STDOUT htescape $err->text, "</dd>\n"; |
| 146 |
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 |
print STDOUT qq[ |
| 154 |
</dl> |
| 155 |
</div> |
| 156 |
]; |
| 157 |
} else { |
| 158 |
print STDOUT qq[ |
| 159 |
</dl> |
| 160 |
</div> |
| 161 |
|
| 162 |
<div id="result-summary" class="section"> |
| 163 |
<p><em>Media type <code class="MIME" lang="en">@{[htescape $input_format]}</code> is not supported!</em></p> |
| 164 |
</div> |
| 165 |
]; |
| 166 |
push @nav, ['#result-summary' => 'Result']; |
| 167 |
} |
| 168 |
|
| 169 |
|
| 170 |
if (defined $doc or defined $el) { |
| 171 |
print STDOUT qq[ |
| 172 |
<div id="document-tree" class="section"> |
| 173 |
<h2>Document Tree</h2> |
| 174 |
]; |
| 175 |
push @nav, ['#document-tree' => 'Tree']; |
| 176 |
|
| 177 |
print_document_tree ($el || $doc); |
| 178 |
|
| 179 |
print STDOUT qq[ |
| 180 |
</div> |
| 181 |
|
| 182 |
<div id="document-errors" class="section"> |
| 183 |
<h2>Document Errors</h2> |
| 184 |
|
| 185 |
<dl> |
| 186 |
]; |
| 187 |
push @nav, ['#document-errors' => 'Document Error']; |
| 188 |
|
| 189 |
require Whatpm::ContentChecker; |
| 190 |
my $onerror = sub { |
| 191 |
my %opt = @_; |
| 192 |
print STDOUT qq[<dt><a href="#node-@{[refaddr $opt{node}]}">], |
| 193 |
htescape get_node_path ($opt{node}), |
| 194 |
"</a></dt>\n<dd>", htescape $opt{type}, "</dd>\n"; |
| 195 |
}; |
| 196 |
|
| 197 |
my $elements; |
| 198 |
if ($el) { |
| 199 |
$elements = Whatpm::ContentChecker->check_element ($el, $onerror); |
| 200 |
} else { |
| 201 |
$elements = Whatpm::ContentChecker->check_document ($doc, $onerror); |
| 202 |
} |
| 203 |
|
| 204 |
print STDOUT qq[ |
| 205 |
</dl> |
| 206 |
</div> |
| 207 |
]; |
| 208 |
|
| 209 |
if (@{$elements->{table}}) { |
| 210 |
require JSON; |
| 211 |
|
| 212 |
print STDOUT qq[ |
| 213 |
<div id="tables" class="section"> |
| 214 |
<h2>Tables</h2> |
| 215 |
|
| 216 |
<!--[if IE]><script type="text/javascript" src="../excanvas.js"></script><![endif]--> |
| 217 |
<script src="../table-script.js" type="text/javascript"></script> |
| 218 |
<noscript> |
| 219 |
<p><em>Structure of tables are visualized here if scripting is enabled.</em></p> |
| 220 |
</noscript> |
| 221 |
]; |
| 222 |
|
| 223 |
my $i = 0; |
| 224 |
for my $table_el (@{$elements->{table}}) { |
| 225 |
$i++; |
| 226 |
print STDOUT qq[<div class="section" id="table-$i"><h3>]; |
| 227 |
print STDOUT qq[<a href="#node-@{[refaddr $table_el]}">], |
| 228 |
htescape get_node_path ($table_el); |
| 229 |
print STDOUT qq[</a></h3>\n]; |
| 230 |
|
| 231 |
my $table = Whatpm::HTMLTable->form_table ($table_el); |
| 232 |
|
| 233 |
for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) { |
| 234 |
next unless $_; |
| 235 |
delete $_->{element}; |
| 236 |
} |
| 237 |
|
| 238 |
for (@{$table->{row_group}}) { |
| 239 |
next unless $_; |
| 240 |
next unless $_->{element}; |
| 241 |
$_->{type} = $_->{element}->manakai_local_name; |
| 242 |
delete $_->{element}; |
| 243 |
} |
| 244 |
|
| 245 |
for (@{$table->{cell}}) { |
| 246 |
next unless $_; |
| 247 |
for (@{$_}) { |
| 248 |
next unless $_; |
| 249 |
for (@$_) { |
| 250 |
$_->{id} = refaddr $_->{element} if defined $_->{element}; |
| 251 |
delete $_->{element}; |
| 252 |
} |
| 253 |
} |
| 254 |
} |
| 255 |
|
| 256 |
print STDOUT '</div><script type="text/javascript">tableToCanvas ('; |
| 257 |
print STDOUT JSON::objToJson ($table); |
| 258 |
print STDOUT qq[, document.getElementById ('table-$i'));</script>]; |
| 259 |
} |
| 260 |
|
| 261 |
print STDOUT qq[</div>]; |
| 262 |
} |
| 263 |
} |
| 264 |
|
| 265 |
## TODO: Show result |
| 266 |
|
| 267 |
print STDOUT qq[ |
| 268 |
<ul class="navigation" id="nav-items"> |
| 269 |
]; |
| 270 |
for (@nav) { |
| 271 |
print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>]; |
| 272 |
} |
| 273 |
print STDOUT qq[ |
| 274 |
</ul> |
| 275 |
</body> |
| 276 |
</html> |
| 277 |
]; |
| 278 |
|
| 279 |
exit; |
| 280 |
|
| 281 |
sub print_source_string ($) { |
| 282 |
my $s = $_[0]; |
| 283 |
my $i = 1; |
| 284 |
print STDOUT qq[<ol lang="">\n]; |
| 285 |
while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) { |
| 286 |
print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n"; |
| 287 |
$i++; |
| 288 |
} |
| 289 |
if ($$s =~ /\G([^\x0A]+)/gc) { |
| 290 |
print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n"; |
| 291 |
} |
| 292 |
print STDOUT "</ol>"; |
| 293 |
} # print_input_string |
| 294 |
|
| 295 |
sub print_document_tree ($) { |
| 296 |
my $node = shift; |
| 297 |
my $r = '<ol class="xoxo">'; |
| 298 |
|
| 299 |
my @node = ($node); |
| 300 |
while (@node) { |
| 301 |
my $child = shift @node; |
| 302 |
unless (ref $child) { |
| 303 |
$r .= $child; |
| 304 |
next; |
| 305 |
} |
| 306 |
|
| 307 |
my $node_id = 'node-'.refaddr $child; |
| 308 |
my $nt = $child->node_type; |
| 309 |
if ($nt == $child->ELEMENT_NODE) { |
| 310 |
my $child_nsuri = $child->namespace_uri; |
| 311 |
$r .= qq[<li id="$node_id" class="tree-element"><code title="@{[defined $child_nsuri ? $child_nsuri : '']}">] . htescape ($child->tag_name) . |
| 312 |
'</code>'; ## ISSUE: case |
| 313 |
|
| 314 |
if ($child->has_attributes) { |
| 315 |
$r .= '<ul class="attributes">'; |
| 316 |
for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] } |
| 317 |
@{$child->attributes}) { |
| 318 |
$r .= qq[<li id="$attr->[3]" class="tree-attribute"><code title="@{[defined $_->[2] ? $_->[2] : '']}">] . htescape ($attr->[0]) . '</code> = '; ## ISSUE: case? |
| 319 |
$r .= '<q>' . htescape ($attr->[1]) . '</q></li>'; ## TODO: children |
| 320 |
} |
| 321 |
$r .= '</ul>'; |
| 322 |
} |
| 323 |
|
| 324 |
if ($node->has_child_nodes) { |
| 325 |
$r .= '<ol class="children">'; |
| 326 |
unshift @node, @{$child->child_nodes}, '</ol>'; |
| 327 |
} |
| 328 |
} elsif ($nt == $child->TEXT_NODE) { |
| 329 |
$r .= qq'<li id="$node_id" class="tree-text"><q lang="">' . htescape ($child->data) . '</q></li>'; |
| 330 |
} elsif ($nt == $child->CDATA_SECTION_NODE) { |
| 331 |
$r .= qq'<li id="$node_id" class="tree-cdata"><code><[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]></code></li>'; |
| 332 |
} elsif ($nt == $child->COMMENT_NODE) { |
| 333 |
$r .= qq'<li id="$node_id" class="tree-comment"><code><!--</code><q lang="">' . htescape ($child->data) . '</q><code>--></code></li>'; |
| 334 |
} elsif ($nt == $child->DOCUMENT_NODE) { |
| 335 |
$r .= qq'<li id="$node_id" class="tree-document">Document</li>'; |
| 336 |
if ($child->has_child_nodes) { |
| 337 |
$r .= '<ol>'; |
| 338 |
unshift @node, @{$child->child_nodes}, '</ol>'; |
| 339 |
} |
| 340 |
} elsif ($nt == $child->DOCUMENT_TYPE_NODE) { |
| 341 |
$r .= qq'<li id="$node_id" class="tree-doctype"><code><!DOCTYPE></code><ul class="attributes">'; |
| 342 |
$r .= qq[<li class="tree-doctype-name">Name = <q>@{[htescape ($child->name)]}</q></li>]; |
| 343 |
$r .= qq[<li class="tree-doctype-publicid">Public identifier = <q>@{[htescape ($child->public_id)]}</q></li>]; |
| 344 |
$r .= qq[<li class="tree-doctype-systemid">System identifier = <q>@{[htescape ($child->system_id)]}</q></li>]; |
| 345 |
$r .= '</ul></li>'; |
| 346 |
} elsif ($nt == $child->PROCESSING_INSTRUCTION_NODE) { |
| 347 |
$r .= qq'<li id="$node_id" class="tree-id"><code><?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?></code></li>'; |
| 348 |
} else { |
| 349 |
$r .= qq'<li id="$node_id" class="tree-unknown">@{[$child->node_type]} @{[htescape ($child->node_name)]}</li>'; # error |
| 350 |
} |
| 351 |
} |
| 352 |
|
| 353 |
$r .= '</ol>'; |
| 354 |
print STDOUT $r; |
| 355 |
} # print_document_tree |
| 356 |
|
| 357 |
sub get_node_path ($) { |
| 358 |
my $node = shift; |
| 359 |
my @r; |
| 360 |
while (defined $node) { |
| 361 |
my $rs; |
| 362 |
if ($node->node_type == 1) { |
| 363 |
$rs = $node->manakai_local_name; |
| 364 |
$node = $node->parent_node; |
| 365 |
} elsif ($node->node_type == 2) { |
| 366 |
$rs = '@' . $node->manakai_local_name; |
| 367 |
$node = $node->owner_element; |
| 368 |
} elsif ($node->node_type == 3) { |
| 369 |
$rs = '"' . $node->data . '"'; |
| 370 |
$node = $node->parent_node; |
| 371 |
} elsif ($node->node_type == 9) { |
| 372 |
$rs = ''; |
| 373 |
$node = $node->parent_node; |
| 374 |
} else { |
| 375 |
$rs = '#' . $node->node_type; |
| 376 |
$node = $node->parent_node; |
| 377 |
} |
| 378 |
unshift @r, $rs; |
| 379 |
} |
| 380 |
return join '/', @r; |
| 381 |
} # get_node_path |
| 382 |
|
| 383 |
=head1 AUTHOR |
| 384 |
|
| 385 |
Wakaba <w@suika.fam.cx>. |
| 386 |
|
| 387 |
=head1 LICENSE |
| 388 |
|
| 389 |
Copyright 2007 Wakaba <w@suika.fam.cx> |
| 390 |
|
| 391 |
This library is free software; you can redistribute it |
| 392 |
and/or modify it under the same terms as Perl itself. |
| 393 |
|
| 394 |
=cut |
| 395 |
|
| 396 |
## $Date: 2007/06/27 14:36:45 $ |
Parent Directory
| 
Revision Log