| 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 $http = SuikaWiki::Input::HTTP->new; |
| 23 |
|
| 24 |
## TODO: _charset_ |
| 25 |
|
| 26 |
if ($http->meta_variable ('PATH_INFO') ne '/') { |
| 27 |
print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400"; |
| 28 |
exit; |
| 29 |
} |
| 30 |
|
| 31 |
require Message::DOM::DOMImplementation; |
| 32 |
my $dom = Message::DOM::DOMImplementation->new; |
| 33 |
|
| 34 |
my $input = get_input_document ($http, $dom); |
| 35 |
my $inner_html_element = $http->parameter ('e'); |
| 36 |
|
| 37 |
load_text_catalog ('en'); ## TODO: conneg |
| 38 |
|
| 39 |
my @nav; |
| 40 |
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 |
<link rel="stylesheet" href="../cc-style.css" type="text/css"> |
| 47 |
</head> |
| 48 |
<body> |
| 49 |
<h1>Web Document Conformance Checker (<em>beta</em>)</h1> |
| 50 |
|
| 51 |
<div id="document-info" class="section"> |
| 52 |
<dl> |
| 53 |
<dt>Request URI</dt> |
| 54 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{request_uri}]}">@{[htescape $input->{request_uri}]}</a>></code></dd> |
| 55 |
<dt>Document URI</dt> |
| 56 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{uri}]}">@{[htescape $input->{uri}]}</a>></code></dd> |
| 57 |
]; # no </dl> yet |
| 58 |
push @nav, ['#document-info' => 'Information']; |
| 59 |
|
| 60 |
if (defined $input->{s}) { |
| 61 |
|
| 62 |
print STDOUT qq[ |
| 63 |
<dt>Base URI</dt> |
| 64 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>></code></dd> |
| 65 |
<dt>Internet Media Type</dt> |
| 66 |
<dd><code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code> |
| 67 |
@{[$input->{media_type_overridden} ? '<em>(overridden)</em>' : '']}</dd> |
| 68 |
<dt>Character Encoding</dt> |
| 69 |
<dd>@{[defined $input->{charset} ? '<code class="charset" lang="en">'.htescape ($input->{charset}).'</code>' : '(none)']} |
| 70 |
@{[$input->{charset_overridden} ? '<em>(overridden)</em>' : '']}</dd> |
| 71 |
</dl> |
| 72 |
</div> |
| 73 |
]; |
| 74 |
|
| 75 |
print_http_header_section ($input); |
| 76 |
|
| 77 |
my $doc; |
| 78 |
my $el; |
| 79 |
|
| 80 |
if ($input->{media_type} eq 'text/html') { |
| 81 |
require Encode; |
| 82 |
require Whatpm::HTML; |
| 83 |
|
| 84 |
my $t = Encode::decode ($input->{charset}, $input->{s}); |
| 85 |
|
| 86 |
print STDOUT qq[ |
| 87 |
<div id="parse-errors" class="section"> |
| 88 |
<h2>Parse Errors</h2> |
| 89 |
|
| 90 |
<dl> |
| 91 |
]; |
| 92 |
push @nav, ['#parse-errors' => 'Parse Error']; |
| 93 |
|
| 94 |
my $onerror = sub { |
| 95 |
my (%opt) = @_; |
| 96 |
my ($cls, $msg) = get_text ($opt{type}, $opt{level}); |
| 97 |
if ($opt{column} > 0) { |
| 98 |
print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n]; |
| 99 |
} else { |
| 100 |
$opt{line} = $opt{line} - 1 || 1; |
| 101 |
print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n]; |
| 102 |
} |
| 103 |
$opt{type} =~ tr/ /-/; |
| 104 |
$opt{type} =~ s/\|/%7C/g; |
| 105 |
$msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]]; |
| 106 |
print STDOUT qq[<dd class="$cls">$msg</dd>\n]; |
| 107 |
}; |
| 108 |
|
| 109 |
$doc = $dom->create_document; |
| 110 |
if (defined $inner_html_element and length $inner_html_element) { |
| 111 |
$el = $doc->create_element_ns |
| 112 |
('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]); |
| 113 |
Whatpm::HTML->set_inner_html ($el, $t, $onerror); |
| 114 |
} else { |
| 115 |
Whatpm::HTML->parse_string ($t => $doc, $onerror); |
| 116 |
} |
| 117 |
|
| 118 |
print STDOUT qq[ |
| 119 |
</dl> |
| 120 |
</div> |
| 121 |
]; |
| 122 |
|
| 123 |
print_source_string_section (\($input->{s}), $input->{charset}); |
| 124 |
} elsif ({ |
| 125 |
'text/xml' => 1, |
| 126 |
'application/xhtml+xml' => 1, |
| 127 |
'application/xml' => 1, |
| 128 |
}->{$input->{media_type}}) { |
| 129 |
require Message::DOM::XMLParserTemp; |
| 130 |
|
| 131 |
print STDOUT qq[ |
| 132 |
<div id="parse-errors" class="section"> |
| 133 |
<h2>Parse Errors</h2> |
| 134 |
|
| 135 |
<dl>]; |
| 136 |
push @nav, ['#parse-errors' => 'Parse Error']; |
| 137 |
|
| 138 |
my $onerror = sub { |
| 139 |
my $err = shift; |
| 140 |
my $line = $err->location->line_number; |
| 141 |
print STDOUT qq[<dt><a href="#line-$line">Line $line</a> column ]; |
| 142 |
print STDOUT $err->location->column_number, "</dt><dd>"; |
| 143 |
print STDOUT htescape $err->text, "</dd>\n"; |
| 144 |
return 1; |
| 145 |
}; |
| 146 |
|
| 147 |
open my $fh, '<', \($input->{s}); |
| 148 |
$doc = Message::DOM::XMLParserTemp->parse_byte_stream |
| 149 |
($fh => $dom, $onerror, charset => $input->{charset}); |
| 150 |
|
| 151 |
print STDOUT qq[</dl> |
| 152 |
</div> |
| 153 |
|
| 154 |
]; |
| 155 |
print_source_string_section (\($input->{s}), $doc->input_encoding); |
| 156 |
} else { |
| 157 |
## TODO: Change HTTP status code?? |
| 158 |
print STDOUT qq[ |
| 159 |
<div id="result-summary" class="section"> |
| 160 |
<p><em>Media type <code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code> is not supported!</em></p> |
| 161 |
</div> |
| 162 |
]; |
| 163 |
push @nav, ['#result-summary' => 'Result']; |
| 164 |
} |
| 165 |
|
| 166 |
|
| 167 |
if (defined $doc or defined $el) { |
| 168 |
print STDOUT qq[ |
| 169 |
<div id="document-tree" class="section"> |
| 170 |
<h2>Document Tree</h2> |
| 171 |
]; |
| 172 |
push @nav, ['#document-tree' => 'Tree']; |
| 173 |
|
| 174 |
print_document_tree ($el || $doc); |
| 175 |
|
| 176 |
print STDOUT qq[ |
| 177 |
</div> |
| 178 |
|
| 179 |
<div id="document-errors" class="section"> |
| 180 |
<h2>Document Errors</h2> |
| 181 |
|
| 182 |
<dl>]; |
| 183 |
push @nav, ['#document-errors' => 'Document Error']; |
| 184 |
|
| 185 |
require Whatpm::ContentChecker; |
| 186 |
my $onerror = sub { |
| 187 |
my %opt = @_; |
| 188 |
my ($cls, $msg) = get_text ($opt{type}, $opt{level}); |
| 189 |
$opt{type} = $opt{level} . ':' . $opt{type} if defined $opt{level}; |
| 190 |
$opt{type} =~ tr/ /-/; |
| 191 |
$opt{type} =~ s/\|/%7C/g; |
| 192 |
$msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]]; |
| 193 |
print STDOUT qq[<dt class="$cls">] . get_node_link ($opt{node}) . |
| 194 |
qq[</dt>\n<dd class="$cls">], $msg, "</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, $onerror); |
| 202 |
} |
| 203 |
|
| 204 |
print STDOUT qq[</dl> |
| 205 |
</div> |
| 206 |
]; |
| 207 |
|
| 208 |
if (@{$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 |
$i++; |
| 225 |
print STDOUT qq[<div class="section" id="table-$i"><h3>] . |
| 226 |
get_node_link ($table_el) . q[</h3>]; |
| 227 |
|
| 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-$i'));</script>]; |
| 256 |
} |
| 257 |
|
| 258 |
print STDOUT qq[</div>]; |
| 259 |
} |
| 260 |
|
| 261 |
if (keys %{$elements->{term}}) { |
| 262 |
print STDOUT qq[ |
| 263 |
<div id="terms" class="section"> |
| 264 |
<h2>Terms</h2> |
| 265 |
|
| 266 |
<dl> |
| 267 |
]; |
| 268 |
for my $term (sort {$a cmp $b} keys %{$elements->{term}}) { |
| 269 |
print STDOUT qq[<dt>@{[htescape $term]}</dt>]; |
| 270 |
for (@{$elements->{term}->{$term}}) { |
| 271 |
print STDOUT qq[<dd>].get_node_link ($_).qq[</dd>]; |
| 272 |
} |
| 273 |
} |
| 274 |
print STDOUT qq[</dl></div>]; |
| 275 |
} |
| 276 |
} |
| 277 |
|
| 278 |
## TODO: Show result |
| 279 |
} else { |
| 280 |
print STDOUT qq[ |
| 281 |
</dl> |
| 282 |
</div> |
| 283 |
|
| 284 |
<div class="section" id="result-summary"> |
| 285 |
<p><em><strong>Input Error</strong>: @{[htescape ($input->{error_status_text})]}</em></p> |
| 286 |
</div> |
| 287 |
]; |
| 288 |
push @nav, ['#result-summary' => 'Result']; |
| 289 |
|
| 290 |
} |
| 291 |
|
| 292 |
print STDOUT qq[ |
| 293 |
<ul class="navigation" id="nav-items"> |
| 294 |
]; |
| 295 |
for (@nav) { |
| 296 |
print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>]; |
| 297 |
} |
| 298 |
print STDOUT qq[ |
| 299 |
</ul> |
| 300 |
</body> |
| 301 |
</html> |
| 302 |
]; |
| 303 |
|
| 304 |
exit; |
| 305 |
|
| 306 |
sub print_http_header_section ($) { |
| 307 |
my $input = shift; |
| 308 |
return unless defined $input->{header_status_code} or |
| 309 |
defined $input->{header_status_text} or |
| 310 |
@{$input->{header_field}}; |
| 311 |
|
| 312 |
push @nav, ['#source-header' => 'HTTP Header']; |
| 313 |
print STDOUT qq[<div id="source-header" class="section"> |
| 314 |
<h2>HTTP Header</h2> |
| 315 |
|
| 316 |
<p><strong>Note</strong>: Due to the limitation of the |
| 317 |
network library in use, the content of this section might |
| 318 |
not be the real header.</p> |
| 319 |
|
| 320 |
<table><tbody> |
| 321 |
]; |
| 322 |
|
| 323 |
if (defined $input->{header_status_code}) { |
| 324 |
print STDOUT qq[<tr><th scope="row">Status code</th>]; |
| 325 |
print STDOUT qq[<td><code>@{[htescape ($input->{header_status_code})]}</code></td></tr>]; |
| 326 |
} |
| 327 |
if (defined $input->{header_status_text}) { |
| 328 |
print STDOUT qq[<tr><th scope="row">Status text</th>]; |
| 329 |
print STDOUT qq[<td><code>@{[htescape ($input->{header_status_text})]}</code></td></tr>]; |
| 330 |
} |
| 331 |
|
| 332 |
for (@{$input->{header_field}}) { |
| 333 |
print STDOUT qq[<tr><th scope="row"><code>@{[htescape ($_->[0])]}</code></th>]; |
| 334 |
print STDOUT qq[<td><code>@{[htescape ($_->[1])]}</code></td></tr>]; |
| 335 |
} |
| 336 |
|
| 337 |
print STDOUT qq[</tbody></table></div>]; |
| 338 |
} # print_http_header_section |
| 339 |
|
| 340 |
sub print_source_string_section ($$) { |
| 341 |
require Encode; |
| 342 |
my $enc = Encode::find_encoding ($_[1]); ## TODO: charset name -> Perl name |
| 343 |
return unless $enc; |
| 344 |
|
| 345 |
my $s = \($enc->decode (${$_[0]})); |
| 346 |
my $i = 1; |
| 347 |
push @nav, ['#source-string' => 'Source']; |
| 348 |
print STDOUT qq[<div id="source-string" class="section"> |
| 349 |
<h2>Document Source</h2> |
| 350 |
<ol lang="">\n]; |
| 351 |
if (length $$s) { |
| 352 |
while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) { |
| 353 |
print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n"; |
| 354 |
$i++; |
| 355 |
} |
| 356 |
if ($$s =~ /\G([^\x0A]+)/gc) { |
| 357 |
print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n"; |
| 358 |
} |
| 359 |
} else { |
| 360 |
print STDOUT q[<li id="line-1"></li>]; |
| 361 |
} |
| 362 |
print STDOUT "</ol></div>"; |
| 363 |
} # print_input_string_section |
| 364 |
|
| 365 |
sub print_document_tree ($) { |
| 366 |
my $node = shift; |
| 367 |
my $r = '<ol class="xoxo">'; |
| 368 |
|
| 369 |
my @node = ($node); |
| 370 |
while (@node) { |
| 371 |
my $child = shift @node; |
| 372 |
unless (ref $child) { |
| 373 |
$r .= $child; |
| 374 |
next; |
| 375 |
} |
| 376 |
|
| 377 |
my $node_id = 'node-'.refaddr $child; |
| 378 |
my $nt = $child->node_type; |
| 379 |
if ($nt == $child->ELEMENT_NODE) { |
| 380 |
my $child_nsuri = $child->namespace_uri; |
| 381 |
$r .= qq[<li id="$node_id" class="tree-element"><code title="@{[defined $child_nsuri ? $child_nsuri : '']}">] . htescape ($child->tag_name) . |
| 382 |
'</code>'; ## ISSUE: case |
| 383 |
|
| 384 |
if ($child->has_attributes) { |
| 385 |
$r .= '<ul class="attributes">'; |
| 386 |
for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] } |
| 387 |
@{$child->attributes}) { |
| 388 |
$r .= qq[<li id="$attr->[3]" class="tree-attribute"><code title="@{[defined $_->[2] ? $_->[2] : '']}">] . htescape ($attr->[0]) . '</code> = '; ## ISSUE: case? |
| 389 |
$r .= '<q>' . htescape ($attr->[1]) . '</q></li>'; ## TODO: children |
| 390 |
} |
| 391 |
$r .= '</ul>'; |
| 392 |
} |
| 393 |
|
| 394 |
if ($child->has_child_nodes) { |
| 395 |
$r .= '<ol class="children">'; |
| 396 |
unshift @node, @{$child->child_nodes}, '</ol></li>'; |
| 397 |
} else { |
| 398 |
$r .= '</li>'; |
| 399 |
} |
| 400 |
} elsif ($nt == $child->TEXT_NODE) { |
| 401 |
$r .= qq'<li id="$node_id" class="tree-text"><q lang="">' . htescape ($child->data) . '</q></li>'; |
| 402 |
} elsif ($nt == $child->CDATA_SECTION_NODE) { |
| 403 |
$r .= qq'<li id="$node_id" class="tree-cdata"><code><[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]></code></li>'; |
| 404 |
} elsif ($nt == $child->COMMENT_NODE) { |
| 405 |
$r .= qq'<li id="$node_id" class="tree-comment"><code><!--</code><q lang="">' . htescape ($child->data) . '</q><code>--></code></li>'; |
| 406 |
} elsif ($nt == $child->DOCUMENT_NODE) { |
| 407 |
$r .= qq'<li id="$node_id" class="tree-document">Document'; |
| 408 |
$r .= qq[<ul class="attributes">]; |
| 409 |
$r .= qq[<li>@{[scalar get_text ('manakaiIsHTML:'.($child->manakai_is_html?1:0))]}</li>]; |
| 410 |
$r .= qq[<li>@{[scalar get_text ('manakaiCompatMode:'.$child->manakai_compat_mode)]}</li>]; |
| 411 |
unless ($child->manakai_is_html) { |
| 412 |
$r .= qq[<li>XML version = <code>@{[htescape ($child->xml_version)]}</code></li>]; |
| 413 |
if (defined $child->xml_encoding) { |
| 414 |
$r .= qq[<li>XML encoding = <code>@{[htescape ($child->xml_encoding)]}</code></li>]; |
| 415 |
} else { |
| 416 |
$r .= qq[<li>XML encoding = (null)</li>]; |
| 417 |
} |
| 418 |
$r .= qq[<li>XML standalone = @{[$child->xml_standalone ? 'true' : 'false']}</li>]; |
| 419 |
} |
| 420 |
$r .= qq[</ul>]; |
| 421 |
if ($child->has_child_nodes) { |
| 422 |
$r .= '<ol class="children">'; |
| 423 |
unshift @node, @{$child->child_nodes}, '</ol></li>'; |
| 424 |
} |
| 425 |
} elsif ($nt == $child->DOCUMENT_TYPE_NODE) { |
| 426 |
$r .= qq'<li id="$node_id" class="tree-doctype"><code><!DOCTYPE></code><ul class="attributes">'; |
| 427 |
$r .= qq[<li class="tree-doctype-name">Name = <q>@{[htescape ($child->name)]}</q></li>]; |
| 428 |
$r .= qq[<li class="tree-doctype-publicid">Public identifier = <q>@{[htescape ($child->public_id)]}</q></li>]; |
| 429 |
$r .= qq[<li class="tree-doctype-systemid">System identifier = <q>@{[htescape ($child->system_id)]}</q></li>]; |
| 430 |
$r .= '</ul></li>'; |
| 431 |
} elsif ($nt == $child->PROCESSING_INSTRUCTION_NODE) { |
| 432 |
$r .= qq'<li id="$node_id" class="tree-id"><code><?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?></code></li>'; |
| 433 |
} else { |
| 434 |
$r .= qq'<li id="$node_id" class="tree-unknown">@{[$child->node_type]} @{[htescape ($child->node_name)]}</li>'; # error |
| 435 |
} |
| 436 |
} |
| 437 |
|
| 438 |
$r .= '</ol>'; |
| 439 |
print STDOUT $r; |
| 440 |
} # print_document_tree |
| 441 |
|
| 442 |
sub get_node_path ($) { |
| 443 |
my $node = shift; |
| 444 |
my @r; |
| 445 |
while (defined $node) { |
| 446 |
my $rs; |
| 447 |
if ($node->node_type == 1) { |
| 448 |
$rs = $node->manakai_local_name; |
| 449 |
$node = $node->parent_node; |
| 450 |
} elsif ($node->node_type == 2) { |
| 451 |
$rs = '@' . $node->manakai_local_name; |
| 452 |
$node = $node->owner_element; |
| 453 |
} elsif ($node->node_type == 3) { |
| 454 |
$rs = '"' . $node->data . '"'; |
| 455 |
$node = $node->parent_node; |
| 456 |
} elsif ($node->node_type == 9) { |
| 457 |
@r = ('') unless @r; |
| 458 |
$rs = ''; |
| 459 |
$node = $node->parent_node; |
| 460 |
} else { |
| 461 |
$rs = '#' . $node->node_type; |
| 462 |
$node = $node->parent_node; |
| 463 |
} |
| 464 |
unshift @r, $rs; |
| 465 |
} |
| 466 |
return join '/', @r; |
| 467 |
} # get_node_path |
| 468 |
|
| 469 |
sub get_node_link ($) { |
| 470 |
return qq[<a href="#node-@{[refaddr $_[0]]}">] . |
| 471 |
htescape (get_node_path ($_[0])) . qq[</a>]; |
| 472 |
} # get_node_link |
| 473 |
|
| 474 |
{ |
| 475 |
my $Msg = {}; |
| 476 |
|
| 477 |
sub load_text_catalog ($) { |
| 478 |
my $lang = shift; # MUST be a canonical lang name |
| 479 |
open my $file, '<', "cc-msg.$lang.txt" or die "$0: cc-msg.$lang.txt: $!"; |
| 480 |
while (<$file>) { |
| 481 |
if (s/^([^;]+);([^;]*);//) { |
| 482 |
my ($type, $cls, $msg) = ($1, $2, $_); |
| 483 |
$msg =~ tr/\x0D\x0A//d; |
| 484 |
$Msg->{$type} = [$cls, $msg]; |
| 485 |
} |
| 486 |
} |
| 487 |
} # load_text_catalog |
| 488 |
|
| 489 |
sub get_text ($) { |
| 490 |
my ($type, $level) = @_; |
| 491 |
$type = $level . ':' . $type if defined $level; |
| 492 |
my @arg; |
| 493 |
{ |
| 494 |
if (defined $Msg->{$type}) { |
| 495 |
my $msg = $Msg->{$type}->[1]; |
| 496 |
$msg =~ s/\$([0-9]+)/defined $arg[$1] ? htescape ($arg[$1]) : '(undef)'/ge; |
| 497 |
return ($Msg->{$type}->[0], $msg); |
| 498 |
} elsif ($type =~ s/:([^:]*)$//) { |
| 499 |
unshift @arg, $1; |
| 500 |
redo; |
| 501 |
} |
| 502 |
} |
| 503 |
return ('', htescape ($_[0])); |
| 504 |
} # get_text |
| 505 |
|
| 506 |
} |
| 507 |
|
| 508 |
sub get_input_document ($$) { |
| 509 |
my ($http, $dom) = @_; |
| 510 |
|
| 511 |
my $request_uri = $http->parameter ('uri'); |
| 512 |
my $r = {}; |
| 513 |
if (defined $request_uri and length $request_uri) { |
| 514 |
my $uri = $dom->create_uri_reference ($request_uri); |
| 515 |
unless ({ |
| 516 |
http => 1, |
| 517 |
}->{lc $uri->uri_scheme}) { |
| 518 |
return {uri => $request_uri, request_uri => $request_uri, |
| 519 |
error_status_text => 'URI scheme not allowed'}; |
| 520 |
} |
| 521 |
|
| 522 |
require Message::Util::HostPermit; |
| 523 |
my $host_permit = new Message::Util::HostPermit; |
| 524 |
$host_permit->add_rule (<<EOH); |
| 525 |
Allow host=suika port=80 |
| 526 |
Deny host=suika |
| 527 |
Allow host=suika.fam.cx port=80 |
| 528 |
Deny host=suika.fam.cx |
| 529 |
Deny host=localhost |
| 530 |
Deny host=*.localdomain |
| 531 |
Deny ipv4=0.0.0.0/8 |
| 532 |
Deny ipv4=10.0.0.0/8 |
| 533 |
Deny ipv4=127.0.0.0/8 |
| 534 |
Deny ipv4=169.254.0.0/16 |
| 535 |
Deny ipv4=172.0.0.0/11 |
| 536 |
Deny ipv4=192.0.2.0/24 |
| 537 |
Deny ipv4=192.88.99.0/24 |
| 538 |
Deny ipv4=192.168.0.0/16 |
| 539 |
Deny ipv4=198.18.0.0/15 |
| 540 |
Deny ipv4=224.0.0.0/4 |
| 541 |
Deny ipv4=255.255.255.255/32 |
| 542 |
Deny ipv6=0::0/0 |
| 543 |
Allow host=* |
| 544 |
EOH |
| 545 |
unless ($host_permit->check ($uri->uri_host, $uri->uri_port || 80)) { |
| 546 |
return {uri => $request_uri, request_uri => $request_uri, |
| 547 |
error_status_text => 'Connection to the host is forbidden'}; |
| 548 |
} |
| 549 |
|
| 550 |
require LWP::UserAgent; |
| 551 |
my $ua = WDCC::LWPUA->new; |
| 552 |
$ua->{wdcc_dom} = $dom; |
| 553 |
$ua->{wdcc_host_permit} = $host_permit; |
| 554 |
$ua->agent ('Mozilla'); ## TODO: for now. |
| 555 |
$ua->parse_head (0); |
| 556 |
$ua->protocols_allowed ([qw/http/]); |
| 557 |
$ua->max_size (1000_000); |
| 558 |
my $req = HTTP::Request->new (GET => $request_uri); |
| 559 |
my $res = $ua->request ($req); |
| 560 |
if ($res->is_success or $http->parameter ('error-page')) { |
| 561 |
$r->{base_uri} = $res->base; ## NOTE: It does check |Content-Base|, |Content-Location|, and <base>. ## TODO: Use our own code! |
| 562 |
$r->{uri} = $res->request->uri; |
| 563 |
$r->{request_uri} = $request_uri; |
| 564 |
|
| 565 |
## TODO: More strict parsing... |
| 566 |
my $ct = $res->header ('Content-Type'); |
| 567 |
if (defined $ct and $ct =~ m#^([0-9A-Za-z._+-]+/[0-9A-Za-z._+-]+)#) { |
| 568 |
$r->{media_type} = lc $1; |
| 569 |
} |
| 570 |
if (defined $ct and $ct =~ /;\s*charset\s*=\s*"?(\S+)"?/i) { |
| 571 |
$r->{charset} = lc $1; |
| 572 |
$r->{charset} =~ tr/\\//d; |
| 573 |
} |
| 574 |
|
| 575 |
my $input_charset = $http->parameter ('charset'); |
| 576 |
if (defined $input_charset and length $input_charset) { |
| 577 |
$r->{charset_overridden} |
| 578 |
= (not defined $r->{charset} or $r->{charset} ne $input_charset); |
| 579 |
$r->{charset} = $input_charset; |
| 580 |
} |
| 581 |
|
| 582 |
$r->{s} = ''.$res->content; |
| 583 |
} else { |
| 584 |
$r->{uri} = $res->request->uri; |
| 585 |
$r->{request_uri} = $request_uri; |
| 586 |
$r->{error_status_text} = $res->status_line; |
| 587 |
} |
| 588 |
|
| 589 |
$r->{header_field} = []; |
| 590 |
$res->scan (sub { |
| 591 |
push @{$r->{header_field}}, [$_[0], $_[1]]; |
| 592 |
}); |
| 593 |
$r->{header_status_code} = $res->code; |
| 594 |
$r->{header_status_text} = $res->message; |
| 595 |
} else { |
| 596 |
$r->{s} = ''.$http->parameter ('s'); |
| 597 |
$r->{uri} = q<thismessage:/>; |
| 598 |
$r->{request_uri} = q<thismessage:/>; |
| 599 |
$r->{base_uri} = q<thismessage:/>; |
| 600 |
$r->{charset} = ''.$http->parameter ('_charset_'); |
| 601 |
$r->{charset} =~ s/\s+//g; |
| 602 |
$r->{charset} = 'utf-8' if $r->{charset} eq ''; |
| 603 |
$r->{header_field} = []; |
| 604 |
} |
| 605 |
|
| 606 |
my $input_format = $http->parameter ('i'); |
| 607 |
if (defined $input_format and length $input_format) { |
| 608 |
$r->{media_type_overridden} |
| 609 |
= (not defined $r->{media_type} or $input_format ne $r->{media_type}); |
| 610 |
$r->{media_type} = $input_format; |
| 611 |
} |
| 612 |
if (defined $r->{s} and not defined $r->{media_type}) { |
| 613 |
$r->{media_type} = 'text/html'; |
| 614 |
$r->{media_type_overridden} = 1; |
| 615 |
} |
| 616 |
|
| 617 |
if ($r->{media_type} eq 'text/xml') { |
| 618 |
unless (defined $r->{charset}) { |
| 619 |
$r->{charset} = 'us-ascii'; |
| 620 |
} elsif ($r->{charset_overridden} and $r->{charset} eq 'us-ascii') { |
| 621 |
$r->{charset_overridden} = 0; |
| 622 |
} |
| 623 |
} |
| 624 |
|
| 625 |
if (length $r->{s} > 1000_000) { |
| 626 |
$r->{error_status_text} = 'Entity-body too large'; |
| 627 |
delete $r->{s}; |
| 628 |
return $r; |
| 629 |
} |
| 630 |
|
| 631 |
return $r; |
| 632 |
} # get_input_document |
| 633 |
|
| 634 |
package WDCC::LWPUA; |
| 635 |
BEGIN { push our @ISA, 'LWP::UserAgent'; } |
| 636 |
|
| 637 |
sub redirect_ok { |
| 638 |
my $ua = shift; |
| 639 |
unless ($ua->SUPER::redirect_ok (@_)) { |
| 640 |
return 0; |
| 641 |
} |
| 642 |
|
| 643 |
my $uris = $_[1]->header ('Location'); |
| 644 |
return 0 unless $uris; |
| 645 |
my $uri = $ua->{wdcc_dom}->create_uri_reference ($uris); |
| 646 |
unless ({ |
| 647 |
http => 1, |
| 648 |
}->{lc $uri->uri_scheme}) { |
| 649 |
return 0; |
| 650 |
} |
| 651 |
unless ($ua->{wdcc_host_permit}->check ($uri->uri_host, $uri->uri_port || 80)) { |
| 652 |
return 0; |
| 653 |
} |
| 654 |
return 1; |
| 655 |
} # redirect_ok |
| 656 |
|
| 657 |
=head1 AUTHOR |
| 658 |
|
| 659 |
Wakaba <w@suika.fam.cx>. |
| 660 |
|
| 661 |
=head1 LICENSE |
| 662 |
|
| 663 |
Copyright 2007 Wakaba <w@suika.fam.cx> |
| 664 |
|
| 665 |
This library is free software; you can redistribute it |
| 666 |
and/or modify it under the same terms as Perl itself. |
| 667 |
|
| 668 |
=cut |
| 669 |
|
| 670 |
## $Date: 2007/07/01 10:02:24 $ |
Parent Directory
| 
Revision Log