/[suikacvs]/test/html-webhacc/cc.cgi
Suika

Contents of /test/html-webhacc/cc.cgi

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.21 - (hide annotations) (download)
Tue Sep 11 08:25:23 2007 UTC (19 years, 2 months ago) by wakaba
Branch: MAIN
Changes since 1.20: +14 -9 lines
Scoring bug fixes

1 wakaba 1.1 #!/usr/bin/perl
2     use strict;
3    
4     use lib qw[/home/httpd/html/www/markup/html/whatpm
5 wakaba 1.16 /home/wakaba/work/manakai2/lib];
6 wakaba 1.1 use CGI::Carp qw[fatalsToBrowser];
7 wakaba 1.2 use Scalar::Util qw[refaddr];
8 wakaba 1.16 use Time::HiRes qw/time/;
9 wakaba 1.1
10 wakaba 1.2 sub htescape ($) {
11     my $s = $_[0];
12     $s =~ s/&/&/g;
13     $s =~ s/</&lt;/g;
14     $s =~ s/>/&gt;/g;
15     $s =~ s/"/&quot;/g;
16 wakaba 1.12 $s =~ s{([\x00-\x09\x0B-\x1F\x7F-\xA0\x{FEFF}\x{FFFC}-\x{FFFF}])}{
17     sprintf '<var>U+%04X</var>', ord $1;
18     }ge;
19 wakaba 1.2 return $s;
20     } # htescape
21    
22 wakaba 1.16 use Message::CGI::HTTP;
23     my $http = Message::CGI::HTTP->new;
24 wakaba 1.1
25 wakaba 1.16 if ($http->get_meta_variable ('PATH_INFO') ne '/') {
26 wakaba 1.8 print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400";
27     exit;
28     }
29    
30 wakaba 1.12 binmode STDOUT, ':utf8';
31 wakaba 1.14 $| = 1;
32 wakaba 1.12
33 wakaba 1.9 require Message::DOM::DOMImplementation;
34     my $dom = Message::DOM::DOMImplementation->new;
35    
36 wakaba 1.7 load_text_catalog ('en'); ## TODO: conneg
37    
38 wakaba 1.3 my @nav;
39 wakaba 1.2 print STDOUT qq[Content-Type: text/html; charset=utf-8
40    
41     <!DOCTYPE html>
42     <html lang="en">
43     <head>
44     <title>Web Document Conformance Checker (BETA)</title>
45 wakaba 1.3 <link rel="stylesheet" href="../cc-style.css" type="text/css">
46 wakaba 1.2 </head>
47     <body>
48 wakaba 1.13 <h1><a href="../cc-interface">Web Document Conformance Checker</a>
49     (<em>beta</em>)</h1>
50 wakaba 1.14 ];
51 wakaba 1.2
52 wakaba 1.14 $| = 0;
53     my $input = get_input_document ($http, $dom);
54 wakaba 1.16 my $inner_html_element = $http->get_parameter ('e');
55     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="">&lt;<a href="@{[htescape $input->{request_uri}]}">@{[htescape $input->{request_uri}]}</a>&gt;</code></dd>
63 wakaba 1.2 <dt>Document URI</dt>
64 wakaba 1.9 <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{uri}]}">@{[htescape $input->{uri}]}</a>&gt;</code></dd>
65 wakaba 1.2 ]; # no </dl> yet
66 wakaba 1.3 push @nav, ['#document-info' => 'Information'];
67 wakaba 1.1
68 wakaba 1.9 if (defined $input->{s}) {
69 wakaba 1.16 $char_length = length $input->{s};
70 wakaba 1.9
71     print STDOUT qq[
72     <dt>Base URI</dt>
73     <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>&gt;</code></dd>
74     <dt>Internet Media Type</dt>
75     <dd><code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code>
76     @{[$input->{media_type_overridden} ? '<em>(overridden)</em>' : '']}</dd>
77     <dt>Character Encoding</dt>
78     <dd>@{[defined $input->{charset} ? '<code class="charset" lang="en">'.htescape ($input->{charset}).'</code>' : '(none)']}
79     @{[$input->{charset_overridden} ? '<em>(overridden)</em>' : '']}</dd>
80 wakaba 1.16 <dt>Length</dt>
81     <dd>$char_length byte@{[$char_length == 1 ? '' : 's']}</dd>
82 wakaba 1.9 </dl>
83     </div>
84     ];
85    
86 wakaba 1.20 my $result = {conforming_min => 1, conforming_max => 1};
87 wakaba 1.19 print_http_header_section ($input, $result);
88 wakaba 1.9
89 wakaba 1.1 my $doc;
90     my $el;
91    
92 wakaba 1.9 if ($input->{media_type} eq 'text/html') {
93 wakaba 1.19 ($doc, $el) = print_syntax_error_html_section ($input, $result);
94 wakaba 1.9 print_source_string_section (\($input->{s}), $input->{charset});
95     } elsif ({
96     'text/xml' => 1,
97 wakaba 1.16 'application/atom+xml' => 1,
98     'application/rss+xml' => 1,
99     'application/svg+xml' => 1,
100 wakaba 1.9 'application/xhtml+xml' => 1,
101     'application/xml' => 1,
102     }->{$input->{media_type}}) {
103 wakaba 1.19 ($doc, $el) = print_syntax_error_xml_section ($input, $result);
104 wakaba 1.9 print_source_string_section (\($input->{s}), $doc->input_encoding);
105 wakaba 1.2 } else {
106 wakaba 1.9 ## TODO: Change HTTP status code??
107 wakaba 1.18 print_result_unknown_type_section ($input);
108 wakaba 1.2 }
109    
110     if (defined $doc or defined $el) {
111 wakaba 1.18 print_structure_dump_section ($doc, $el);
112 wakaba 1.19 my $elements = print_structure_error_section ($doc, $el, $result);
113 wakaba 1.18 print_table_section ($elements->{table}) if @{$elements->{table}};
114     print_id_section ($elements->{id}) if keys %{$elements->{id}};
115     print_term_section ($elements->{term}) if keys %{$elements->{term}};
116     print_class_section ($elements->{class}) if keys %{$elements->{class}};
117 wakaba 1.1 }
118    
119 wakaba 1.19 print_result_section ($result);
120 wakaba 1.9 } else {
121 wakaba 1.18 print STDOUT qq[</dl></div>];
122     print_result_input_error_section ($input);
123 wakaba 1.9 }
124 wakaba 1.3
125 wakaba 1.2 print STDOUT qq[
126 wakaba 1.3 <ul class="navigation" id="nav-items">
127     ];
128     for (@nav) {
129     print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>];
130     }
131     print STDOUT qq[
132     </ul>
133 wakaba 1.2 </body>
134     </html>
135     ];
136 wakaba 1.1
137 wakaba 1.16 for (qw/decode parse parse_xml check/) {
138     next unless defined $time{$_};
139     open my $file, '>>', ".cc-$_.txt" or die ".cc-$_.txt: $!";
140     print $file $char_length, "\t", $time{$_}, "\n";
141     }
142    
143 wakaba 1.1 exit;
144    
145 wakaba 1.19 sub add_error ($$$) {
146     my ($layer, $err, $result) = @_;
147     if (defined $err->{level}) {
148     if ($err->{level} eq 's') {
149     $result->{$layer}->{should}++;
150     $result->{$layer}->{score_min} -= 2;
151     $result->{conforming_min} = 0;
152     } elsif ($err->{level} eq 'w' or $err->{level} eq 'g') {
153     $result->{$layer}->{warning}++;
154     } elsif ($err->{level} eq 'unsupported') {
155     $result->{$layer}->{unsupported}++;
156     $result->{unsupported} = 1;
157     } else {
158     $result->{$layer}->{must}++;
159     $result->{$layer}->{score_max} -= 2;
160     $result->{$layer}->{score_min} -= 2;
161     $result->{conforming_min} = 0;
162     $result->{conforming_max} = 0;
163     }
164     } else {
165     $result->{$layer}->{must}++;
166     $result->{$layer}->{score_max} -= 2;
167     $result->{$layer}->{score_min} -= 2;
168     $result->{conforming_min} = 0;
169     $result->{conforming_max} = 0;
170     }
171     } # add_error
172    
173     sub print_http_header_section ($$) {
174     my ($input, $result) = @_;
175 wakaba 1.9 return unless defined $input->{header_status_code} or
176     defined $input->{header_status_text} or
177     @{$input->{header_field}};
178    
179     push @nav, ['#source-header' => 'HTTP Header'];
180     print STDOUT qq[<div id="source-header" class="section">
181     <h2>HTTP Header</h2>
182    
183     <p><strong>Note</strong>: Due to the limitation of the
184     network library in use, the content of this section might
185     not be the real header.</p>
186    
187     <table><tbody>
188     ];
189    
190     if (defined $input->{header_status_code}) {
191     print STDOUT qq[<tr><th scope="row">Status code</th>];
192     print STDOUT qq[<td><code>@{[htescape ($input->{header_status_code})]}</code></td></tr>];
193     }
194     if (defined $input->{header_status_text}) {
195     print STDOUT qq[<tr><th scope="row">Status text</th>];
196     print STDOUT qq[<td><code>@{[htescape ($input->{header_status_text})]}</code></td></tr>];
197     }
198    
199     for (@{$input->{header_field}}) {
200     print STDOUT qq[<tr><th scope="row"><code>@{[htescape ($_->[0])]}</code></th>];
201     print STDOUT qq[<td><code>@{[htescape ($_->[1])]}</code></td></tr>];
202     }
203    
204     print STDOUT qq[</tbody></table></div>];
205     } # print_http_header_section
206    
207 wakaba 1.19 sub print_syntax_error_html_section ($$) {
208     my ($input, $result) = @_;
209 wakaba 1.18
210     require Encode;
211     require Whatpm::HTML;
212    
213     $input->{charset} ||= 'ISO-8859-1'; ## TODO: for now.
214    
215     my $time1 = time;
216     my $t = Encode::decode ($input->{charset}, $input->{s});
217     $time{decode} = time - $time1;
218    
219     print STDOUT qq[
220     <div id="parse-errors" class="section">
221     <h2>Parse Errors</h2>
222    
223     <dl>];
224     push @nav, ['#parse-errors' => 'Parse Error'];
225    
226     my $onerror = sub {
227     my (%opt) = @_;
228     my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level});
229     if ($opt{column} > 0) {
230     print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n];
231     } else {
232     $opt{line} = $opt{line} - 1 || 1;
233     print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n];
234     }
235     $type =~ tr/ /-/;
236     $type =~ s/\|/%7C/g;
237     $msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]];
238     print STDOUT qq[<dd class="$cls">$msg</dd>\n];
239 wakaba 1.19
240     add_error ('syntax', \%opt => $result);
241 wakaba 1.18 };
242    
243     my $doc = $dom->create_document;
244     my $el;
245     $time1 = time;
246     if (defined $inner_html_element and length $inner_html_element) {
247     $el = $doc->create_element_ns
248     ('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
249     Whatpm::HTML->set_inner_html ($el, $t, $onerror);
250     } else {
251     Whatpm::HTML->parse_string ($t => $doc, $onerror);
252     }
253     $time{parse} = time - $time1;
254    
255     print STDOUT qq[</dl></div>];
256    
257     return ($doc, $el);
258     } # print_syntax_error_html_section
259    
260 wakaba 1.19 sub print_syntax_error_xml_section ($$) {
261     my ($input, $result) = @_;
262 wakaba 1.18
263     require Message::DOM::XMLParserTemp;
264    
265     print STDOUT qq[
266     <div id="parse-errors" class="section">
267     <h2>Parse Errors</h2>
268