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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.44 - (hide annotations) (download)
Mon Mar 17 13:52:48 2008 UTC (18 years, 8 months ago) by wakaba
Branch: MAIN
Changes since 1.43: +5 -3 lines
++ ChangeLog	17 Mar 2008 13:52:22 -0000
	* cc.cgi (get_error_label): No-parent case was not considered.

2008-03-17  Wakaba  <wakaba@suika.fam.cx>

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/&/&amp;/g;
14     $s =~ s/</&lt;/g;
15     $s =~ s/>/&gt;/g;
16     $s =~ s/"/&quot;/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 1.35 my @nav;
24     my %time;
25     require Message::DOM::DOMImplementation;
26     my $dom = Message::DOM::DOMImplementation->new;
27     {
28 wakaba 1.16 use Message::CGI::HTTP;
29     my $http = Message::CGI::HTTP->new;
30 wakaba 1.1
31 wakaba 1.16 if ($http->get_meta_variable ('PATH_INFO') ne '/') {
32 wakaba 1.8 print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400";
33     exit;
34     }
35    
36 wakaba 1.12 binmode STDOUT, ':utf8';
37 wakaba 1.14 $| = 1;
38 wakaba 1.12
39 wakaba 1.7 load_text_catalog ('en'); ## TODO: conneg
40    
41 wakaba 1.2 print STDOUT qq[Content-Type: text/html; charset=utf-8
42    
43     <!DOCTYPE html>
44     <html lang="en">
45     <head>
46     <title>Web Document Conformance Checker (BETA)</title>
47 wakaba 1.3 <link rel="stylesheet" href="../cc-style.css" type="text/css">
48 wakaba 1.2 </head>
49     <body>
50 wakaba 1.13 <h1><a href="../cc-interface">Web Document Conformance Checker</a>
51     (<em>beta</em>)</h1>
52 wakaba 1.14 ];
53 wakaba 1.2
54 wakaba 1.14 $| = 0;
55     my $input = get_input_document ($http, $dom);
56 wakaba 1.16 my $char_length = 0;
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 1.25 <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{uri}]}" id=anchor-document-uri>@{[htescape $input->{uri}]}</a>&gt;</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="">&lt;<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>&gt;</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 wakaba 1.39
90     <script src="../cc-script.js"></script>
91 wakaba 1.9 ];
92    
93 wakaba 1.35 $input->{id_prefix} = '';
94     #$input->{nested} = 0;
95 wakaba 1.20 my $result = {conforming_min => 1, conforming_max => 1};
96 wakaba 1.31 check_and_print ($input => $result);
97 wakaba 1.19 print_result_section ($result);
98 wakaba 1.9 } else {
99 wakaba 1.18 print STDOUT qq[</dl></div>];
100     print_result_input_error_section ($input);
101 wakaba 1.9 }
102 wakaba 1.3
103 wakaba 1.2 print STDOUT qq[
104 wakaba 1.3 <ul class="navigation" id="nav-items">
105     ];
106     for (@nav) {
107     print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>];
108     }
109     print STDOUT qq[
110     </ul>
111 wakaba 1.2 </body>
112     </html>
113     ];
114 wakaba 1.1
115 wakaba 1.24 for (qw/decode parse parse_html parse_xml parse_manifest
116     check check_manifest/) {
117 wakaba 1.16 next unless defined $time{$_};
118     open my $file, '>>', ".cc-$_.txt" or die ".cc-$_.txt: $!";
119     print $file $char_length, "\t", $time{$_}, "\n";
120     }
121    
122 wakaba 1.1 exit;
123 wakaba 1.35 }
124 wakaba 1.1
125 wakaba 1.19 sub add_error ($$$) {
126     my ($layer, $err, $result) = @_;
127     if (defined $err->{level}) {
128     if ($err->{level} eq 's') {
129     $result->{$layer}->{should}++;
130     $result->{$layer}->{score_min} -= 2;
131     $result->{conforming_min} = 0;
132     } elsif ($err->{level} eq 'w' or $err->{level} eq 'g') {
133     $result->{$layer}->{warning}++;
134 wakaba 1.30 } elsif ($err->{level} eq 'u' or $err->{level} eq 'unsupported') {
135 wakaba 1.19 $result->{$layer}->{unsupported}++;
136     $result->{unsupported} = 1;
137 wakaba 1.37 } elsif ($err->{level} eq 'i') {
138     #
139 wakaba 1.19 } else {
140     $result->{$layer}->{must}++;
141     $result->{$layer}->{score_max} -= 2;
142     $result->{$layer}->{score_min} -= 2;
143     $result->{conforming_min} = 0;
144     $result->{conforming_max} = 0;
145     }
146     } else {
147     $result->{$layer}->{must}++;
148     $result->{$layer}->{score_max} -= 2;
149     $result->{$layer}->{score_min} -= 2;
150     $result->{conforming_min} = 0;
151     $result->{conforming_max} = 0;
152     }
153     } # add_error
154    
155 wakaba 1.31 sub check_and_print ($$) {
156     my ($input, $result) = @_;
157    
158     print_http_header_section ($input, $result);
159    
160     my $doc;
161     my $el;
162 wakaba 1.35 my $cssom;
163 wakaba 1.31 my $manifest;
164 wakaba 1.34 my @subdoc;
165 wakaba 1.31
166     if ($input->{media_type} eq 'text/html') {
167     ($doc, $el) = print_syntax_error_html_section ($input, $result);
168     print_source_string_section
169 wakaba 1.35 ($input,
170     \($input->{s}),
171     $input->{charset} || $doc->input_encoding);
172 wakaba 1.31 } elsif ({
173     'text/xml' => 1,
174     'application/atom+xml' => 1,
175     'application/rss+xml' => 1,
176     'application/svg+xml' => 1,
177     'application/xhtml+xml' => 1,
178     'application/xml' => 1,
179     }->{$input->{media_type}}) {
180     ($doc, $el) = print_syntax_error_xml_section ($input, $result);
181 wakaba 1.35 print_source_string_section ($input,
182     \($input->{s}),
183     $doc->input_encoding);
184     } elsif ($input->{media_type} eq 'text/css') {
185     $cssom = print_syntax_error_css_section ($input, $result);
186     print_source_string_section
187     ($input, \($input->{s}),
188     $cssom->manakai_input_encoding);
189 wakaba 1.31 } elsif ($input->{media_type} eq 'text/cache-manifest') {
190     ## TODO: MUST be text/cache-manifest
191     $manifest = print_syntax_error_manifest_section ($input, $result);
192 wakaba 1.35 print_source_string_section ($input, \($input->{s}),
193     'utf-8');
194 wakaba 1.31 } else {
195     ## TODO: Change HTTP status code??
196     print_result_unknown_type_section ($input, $result);
197     }
198    
199     if (defined $doc or defined $el) {
200 wakaba 1.34 $doc->document_uri ($input->{uri});
201     $doc->manakai_entity_base_uri ($input->{base_uri});
202 wakaba 1.32 print_structure_dump_dom_section ($input, $doc, $el);
203     my $elements = print_structure_error_dom_section
204 wakaba 1.34 ($input, $doc, $el, $result, sub {
205     push @subdoc, shift;
206     });
207 wakaba 1.32 print_table_section ($input, $elements->{table}) if @{$elements->{table}};
208 wakaba 1.33 print_listing_section ({
209     id => 'identifiers', label => 'IDs', heading => 'Identifiers',
210     }, $input, $elements->{id}) if keys %{$elements->{id}};
211     print_listing_section ({
212     id => 'terms', label => 'Terms', heading => 'Terms',
213     }, $input, $elements->{term}) if keys %{$elements->{term}};
214     print_listing_section ({
215     id => 'classes', label => 'Classes', heading => 'Classes',
216     }, $input, $elements->{class}) if keys %{$elements->{class}};
217 wakaba 1.35 } elsif (defined $cssom) {
218     print_structure_dump_cssom_section ($input, $cssom);
219     ## TODO: CSSOM validation
220 wakaba 1.36 add_error ('structure', {level => 'u'} => $result);
221 wakaba 1.31 } elsif (defined $manifest) {
222 wakaba 1.32 print_structure_dump_manifest_section ($input, $manifest);
223     print_structure_error_manifest_section ($input, $manifest, $result);
224 wakaba 1.31 }
225 wakaba 1.34
226     my $id_prefix = 0;
227     for my $subinput (@subdoc) {
228     $subinput->{id_prefix} = 'subdoc-' . ++$id_prefix;
229     $subinput->{nested} = 1;
230     $subinput->{base_uri} = $subinput->{container_node}->base_uri
231     unless defined $subinput->{base_uri};
232     my $ebaseuri = htescape ($subinput->{base_uri});
233     push @nav, ['#' . $subinput->{id_prefix} => 'Sub #' . $id_prefix];
234     print STDOUT qq[<div id="$subinput->{id_prefix}" class=section>
235     <h2>Subdocument #$id_prefix</h2>
236    
237     <dl>
238     <dt>Internet Media Type</dt>
239     <dd><code class="MIME" lang="en">@{[htescape $subinput->{media_type}]}</code>
240     <dt>Container Node</dt>
241     <dd>@{[get_node_link ($input, $subinput->{container_node})]}</dd>
242     <dt>Base <abbr title="Uniform Resource Identifiers">URI</abbr></dt>
243     <dd><code class=URI>&lt;<a href="$ebaseuri">$ebaseuri</a>></code></dd>
244     </dl>];
245    
246 wakaba 1.35 $subinput->{id_prefix} .= '-';
247 wakaba 1.34 check_and_print ($subinput => $result);
248    
249     print STDOUT qq[</div>];
250     }
251 wakaba 1.31 } # check_and_print
252    
253 wakaba 1.19 sub print_http_header_section ($$) {
254     my ($input, $result) = @_;
255 wakaba 1.9 return unless defined $input->{header_status_code} or
256     defined $input->{header_status_text} or
257 wakaba 1.34 @{$input->{header_field} or []};
258 wakaba 1.9
259 wakaba 1.32 push @nav, ['#source-header' => 'HTTP Header'] unless $input->{nested};
260     print STDOUT qq[<div id="$input->{id_prefix}source-header" class="section">
261 wakaba 1.9 <h2>HTTP Header</h2>
262    
263     <p><strong>Note</strong>: Due to the limitation of the
264     network library in use, the content of this section might
265     not be the real header.</p>
266