6 |
/home/wakaba/work/manakai2/lib]; |
/home/wakaba/work/manakai2/lib]; |
7 |
use CGI::Carp qw[fatalsToBrowser]; |
use CGI::Carp qw[fatalsToBrowser]; |
8 |
use Scalar::Util qw[refaddr]; |
use Scalar::Util qw[refaddr]; |
|
use Time::HiRes qw/time/; |
|
9 |
|
|
10 |
sub htescape ($) { |
require WebHACC::Input; |
11 |
my $s = $_[0]; |
require WebHACC::Result; |
12 |
$s =~ s/&/&/g; |
require WebHACC::Output; |
13 |
$s =~ s/</</g; |
|
14 |
$s =~ s/>/>/g; |
my $out; |
|
$s =~ s/"/"/g; |
|
|
$s =~ s{([\x00-\x09\x0B-\x1F\x7F-\xA0\x{FEFF}\x{FFFC}-\x{FFFF}])}{ |
|
|
sprintf '<var>U+%04X</var>', ord $1; |
|
|
}ge; |
|
|
return $s; |
|
|
} # htescape |
|
15 |
|
|
|
my @nav; |
|
|
my %time; |
|
16 |
require Message::DOM::DOMImplementation; |
require Message::DOM::DOMImplementation; |
17 |
my $dom = Message::DOM::DOMImplementation->new; |
my $dom = Message::DOM::DOMImplementation->new; |
18 |
{ |
{ |
23 |
print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400"; |
print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400"; |
24 |
exit; |
exit; |
25 |
} |
} |
26 |
|
|
|
binmode STDOUT, ':utf8'; |
|
|
$| = 1; |
|
|
|
|
27 |
load_text_catalog ('en'); ## TODO: conneg |
load_text_catalog ('en'); ## TODO: conneg |
28 |
|
|
29 |
print STDOUT qq[Content-Type: text/html; charset=utf-8 |
$out = WebHACC::Output->new; |
30 |
|
$out->handle (*STDOUT); |
31 |
|
$out->set_utf8; |
32 |
|
$out->set_flush; |
33 |
|
$out->html (qq[Content-Type: text/html; charset=utf-8 |
34 |
|
|
35 |
<!DOCTYPE html> |
<!DOCTYPE html> |
36 |
<html lang="en"> |
<html lang="en"> |
41 |
<body> |
<body> |
42 |
<h1><a href="../cc-interface">Web Document Conformance Checker</a> |
<h1><a href="../cc-interface">Web Document Conformance Checker</a> |
43 |
(<em>beta</em>)</h1> |
(<em>beta</em>)</h1> |
44 |
]; |
]); |
45 |
|
|
|
$| = 0; |
|
46 |
my $input = get_input_document ($http, $dom); |
my $input = get_input_document ($http, $dom); |
47 |
|
$out->input ($input); |
48 |
|
$out->unset_flush; |
49 |
|
|
50 |
my $char_length = 0; |
my $char_length = 0; |
51 |
|
|
52 |
print qq[ |
$out->start_section (id => 'document-info', title => 'Information'); |
53 |
<div id="document-info" class="section"> |
$out->html (qq[<dl> |
54 |
<dl> |
<dt>Request URL</dt> |
55 |
<dt>Request URI</dt> |
<dd>]); |
56 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{request_uri}]}">@{[htescape $input->{request_uri}]}</a>></code></dd> |
$out->url ($input->{request_uri}); |
57 |
<dt>Document URI</dt> |
$out->html (q[<dt>Document URL<!-- HTML5 document's address? --> |
58 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{uri}]}" id=anchor-document-uri>@{[htescape $input->{uri}]}</a>></code> |
<dd>]); |
59 |
|
$out->url ($input->{uri}, id => 'anchor-document-url'); |
60 |
|
$out->html (q[ |
61 |
<script> |
<script> |
62 |
document.title = '<' |
document.title = '<' |
63 |
+ document.getElementById ('anchor-document-uri').href + '> \\u2014 ' |
+ document.getElementById ('anchor-document-url').href + '> \\u2014 ' |
64 |
+ document.title; |
+ document.title; |
65 |
</script></dd> |
</script>]); |
66 |
]; # no </dl> yet |
## NOTE: no </dl> yet |
|
push @nav, ['#document-info' => 'Information']; |
|
67 |
|
|
68 |
if (defined $input->{s}) { |
if (defined $input->{s}) { |
69 |
$char_length = length $input->{s}; |
$char_length = length $input->{s}; |
70 |
|
|
71 |
print STDOUT qq[ |
$out->html (qq[<dt>Base URI<dd>]); |
72 |
<dt>Base URI</dt> |
$out->url ($input->{base_uri}); |
73 |
<dd><code class="URI" lang=""><<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>></code></dd> |
$out->html (qq[<dt>Internet Media Type</dt> |
74 |
<dt>Internet Media Type</dt> |
<dd><code class="MIME" lang="en">]); |
75 |
<dd><code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code> |
$out->text ($input->{media_type}); |
76 |
@{[$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> |
$out->html (qq[</code> ]); |
77 |
<dt>Character Encoding</dt> |
if ($input->{media_type_overridden}) { |
78 |
<dd>@{[defined $input->{charset} ? '<code class="charset" lang="en">'.htescape ($input->{charset}).'</code>' : '(none)']} |
$out->html ('<em>(overridden)</em>'); |
79 |
@{[$input->{charset_overridden} ? '<em>(overridden)</em>' : '']}</dd> |
} elsif (defined $input->{official_type}) { |
80 |
|
if ($input->{media_type} eq $input->{official_type}) { |
81 |
|
# |
82 |
|
} else { |
83 |
|
$out->html ('<em>(sniffed; official type is: <code class=MIME lang=en>'); |
84 |
|
$out->text ($input->{official_type}); |
85 |
|
$out->html ('</code>)'); |
86 |
|
} |
87 |
|
} else { |
88 |
|
$out->html ('<em>(sniffed)</em>'); |
89 |
|
} |
90 |
|
$out->html (q[<dt>Character Encoding<dd>]); |
91 |
|
if (defined $input->{charset}) { |
92 |
|
$out->html ('<code class="charset" lang="en">'); |
93 |
|
$out->text ($input->{charset}); |
94 |
|
$out->html ('</code>'); |
95 |
|
} else { |
96 |
|
$out->text ('(none)'); |
97 |
|
} |
98 |
|
$out->html (' <em>overridden</em>') if $input->{charset_overridden}; |
99 |
|
$out->html (qq[ |
100 |
<dt>Length</dt> |
<dt>Length</dt> |
101 |
<dd>$char_length byte@{[$char_length == 1 ? '' : 's']}</dd> |
<dd>$char_length byte@{[$char_length == 1 ? '' : 's']}</dd> |
102 |
</dl> |
</dl> |
|
</div> |
|
|
]; |
|
103 |
|
|
104 |
$input->{id_prefix} = ''; |
<script src="../cc-script.js"></script> |
105 |
#$input->{nested} = 0; |
]); |
106 |
my $result = {conforming_min => 1, conforming_max => 1}; |
$out->end_section; |
107 |
check_and_print ($input => $result); |
|
108 |
print_result_section ($result); |
my $result = WebHACC::Result->new; |
109 |
} else { |
$result->output ($out); |
110 |
print STDOUT qq[</dl></div>]; |
$result->{conforming_min} = 1; |
111 |
print_result_input_error_section ($input); |
$result->{conforming_max} = 1; |
112 |
} |
check_and_print ($input => $result => $out); |
113 |
|
$result->generate_result_section; |
114 |
|
} else { |
115 |
|
$out->html ('</dl>'); |
116 |
|
$out->end_section; |
117 |
|
|
118 |
print STDOUT qq[ |
my $result = WebHACC::Result->new; |
119 |
<ul class="navigation" id="nav-items"> |
$result->output ($out); |
120 |
]; |
$result->{conforming_min} = 0; |
121 |
for (@nav) { |
$result->{conforming_max} = 1; |
|
print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>]; |
|
|
} |
|
|
print STDOUT qq[ |
|
|
</ul> |
|
|
</body> |
|
|
</html> |
|
|
]; |
|
122 |
|
|
123 |
for (qw/decode parse parse_html parse_xml parse_manifest |
$input->generate_transfer_sections ($result); |
124 |
check check_manifest/) { |
$result->generate_result_section; |
|
next unless defined $time{$_}; |
|
|
open my $file, '>>', ".cc-$_.txt" or die ".cc-$_.txt: $!"; |
|
|
print $file $char_length, "\t", $time{$_}, "\n"; |
|
125 |
} |
} |
126 |
|
|
127 |
exit; |
$out->nav_list; |
|
} |
|
128 |
|
|
129 |
sub add_error ($$$) { |
exit; |
130 |
my ($layer, $err, $result) = @_; |
} |
|
if (defined $err->{level}) { |
|
|
if ($err->{level} eq 's') { |
|
|
$result->{$layer}->{should}++; |
|
|
$result->{$layer}->{score_min} -= 2; |
|
|
$result->{conforming_min} = 0; |
|
|
} elsif ($err->{level} eq 'w' or $err->{level} eq 'g') { |
|
|
$result->{$layer}->{warning}++; |
|
|
} elsif ($err->{level} eq 'u' or $err->{level} eq 'unsupported') { |
|
|
$result->{$layer}->{unsupported}++; |
|
|
$result->{unsupported} = 1; |
|
|
} elsif ($err->{level} eq 'i') { |
|
|
# |
|
|
} else { |
|
|
$result->{$layer}->{must}++; |
|
|
$result->{$layer}->{score_max} -= 2; |
|
|
$result->{$layer}->{score_min} -= 2; |
|
|
$result->{conforming_min} = 0; |
|
|
$result->{conforming_max} = 0; |
|
|
} |
|
|
} else { |
|
|
$result->{$layer}->{must}++; |
|
|
$result->{$layer}->{score_max} -= 2; |
|
|
$result->{$layer}->{score_min} -= 2; |
|
|
$result->{conforming_min} = 0; |
|
|
$result->{conforming_max} = 0; |
|
|
} |
|
|
} # add_error |
|
131 |
|
|
132 |
sub check_and_print ($$) { |
sub check_and_print ($$$) { |
133 |
my ($input, $result) = @_; |
my ($input, $result, $out) = @_; |
134 |
|
my $original_input = $out->input; |
135 |
|
$out->input ($input); |
136 |
|
|
137 |
print_http_header_section ($input, $result); |
$input->generate_transfer_sections ($result); |
138 |
|
|
|
my $doc; |
|
|
my $el; |
|
|
my $cssom; |
|
|
my $manifest; |
|
139 |
my @subdoc; |
my @subdoc; |
140 |
|
|
141 |
if ($input->{media_type} eq 'text/html') { |
my $checker_class = { |
142 |
($doc, $el) = print_syntax_error_html_section ($input, $result); |
'text/cache-manifest' => 'WebHACC::Language::CacheManifest', |
143 |
print_source_string_section |
'text/css' => 'WebHACC::Language::CSS', |
144 |
($input, |
'text/html' => 'WebHACC::Language::HTML', |
145 |
\($input->{s}), |
'text/x-webidl' => 'WebHACC::Language::WebIDL', |
146 |
$input->{charset} || $doc->input_encoding); |
|
147 |
} elsif ({ |
'text/xml' => 'WebHACC::Language::XML', |
148 |
'text/xml' => 1, |
'application/atom+xml' => 'WebHACC::Language::XML', |
149 |
'application/atom+xml' => 1, |
'application/rss+xml' => 'WebHACC::Language::XML', |
150 |
'application/rss+xml' => 1, |
'image/svg+xml' => 'WebHACC::Language::XML', |
151 |
'application/svg+xml' => 1, |
'application/xhtml+xml' => 'WebHACC::Language::XML', |
152 |
'application/xhtml+xml' => 1, |
'application/xml' => 'WebHACC::Language::XML', |
153 |
'application/xml' => 1, |
## TODO: Should we make all XML MIME Types fall |
154 |
}->{$input->{media_type}}) { |
## into this category? |
155 |
($doc, $el) = print_syntax_error_xml_section ($input, $result); |
|
156 |
print_source_string_section ($input, |
## NOTE: This type has different model from normal XML types. |
157 |
\($input->{s}), |
'application/rdf+xml' => 'WebHACC::Language::XML', |
158 |
$doc->input_encoding); |
}->{$input->{media_type}} || 'WebHACC::Language::Default'; |
159 |
} elsif ($input->{media_type} eq 'text/css') { |
|
160 |
$cssom = print_syntax_error_css_section ($input, $result); |
eval qq{ require $checker_class } or die "$0: Loading $checker_class: $@"; |
161 |
print_source_string_section |
my $checker = $checker_class->new; |
162 |
($input, \($input->{s}), |
$checker->input ($input); |
163 |
$cssom->manakai_input_encoding); |
$checker->output ($out); |
164 |
} elsif ($input->{media_type} eq 'text/cache-manifest') { |
$checker->result ($result); |
165 |
## TODO: MUST be text/cache-manifest |
|
166 |
$manifest = print_syntax_error_manifest_section ($input, $result); |
## TODO: A cache manifest MUST be text/cache-manifest |
167 |
print_source_string_section ($input, \($input->{s}), |
## TODO: WebIDL media type "text/x-webidl" |
168 |
'utf-8'); |
|
169 |
} else { |
$checker->generate_syntax_error_section; |
170 |
## TODO: Change HTTP status code?? |
$checker->generate_source_string_section; |
171 |
print_result_unknown_type_section ($input, $result); |
|
172 |
} |
$checker->onsubdoc (sub { |
173 |
|
push @subdoc, shift; |
174 |
|
}); |
175 |
|
|
176 |
|
$checker->generate_structure_dump_section; |
177 |
|
$checker->generate_structure_error_section; |
178 |
|
$checker->generate_additional_sections; |
179 |
|
|
180 |
|
=pod |
181 |
|
|
182 |
if (defined $doc or defined $el) { |
if (defined $doc or defined $el) { |
183 |
$doc->document_uri ($input->{uri}); |
|
|
$doc->manakai_entity_base_uri ($input->{base_uri}); |
|
|
print_structure_dump_dom_section ($input, $doc, $el); |
|
|
my $elements = print_structure_error_dom_section |
|
|
($input, $doc, $el, $result, sub { |
|
|
push @subdoc, shift; |
|
|
}); |
|
184 |
print_table_section ($input, $elements->{table}) if @{$elements->{table}}; |
print_table_section ($input, $elements->{table}) if @{$elements->{table}}; |
185 |
print_listing_section ({ |
print_listing_section ({ |
186 |
id => 'identifiers', label => 'IDs', heading => 'Identifiers', |
id => 'identifiers', label => 'IDs', heading => 'Identifiers', |
191 |
print_listing_section ({ |
print_listing_section ({ |
192 |
id => 'classes', label => 'Classes', heading => 'Classes', |
id => 'classes', label => 'Classes', heading => 'Classes', |
193 |
}, $input, $elements->{class}) if keys %{$elements->{class}}; |
}, $input, $elements->{class}) if keys %{$elements->{class}}; |
194 |
} elsif (defined $cssom) { |
|
195 |
print_structure_dump_cssom_section ($input, $cssom); |
print_rdf_section ($input, $elements->{rdf}) if @{$elements->{rdf}}; |
|
## TODO: CSSOM validation |
|
|
add_error ('structure', {level => 'u'} => $result); |
|
|
} elsif (defined $manifest) { |
|
|
print_structure_dump_manifest_section ($input, $manifest); |
|
|
print_structure_error_manifest_section ($input, $manifest, $result); |
|
196 |
} |
} |
197 |
|
|
198 |
|
=cut |
199 |
|
|
200 |
my $id_prefix = 0; |
my $id_prefix = 0; |
201 |
for my $subinput (@subdoc) { |
for my $_subinput (@subdoc) { |
202 |
$subinput->{id_prefix} = 'subdoc-' . ++$id_prefix; |
my $subinput = WebHACC::Input->new; |
203 |
$subinput->{nested} = 1; |
$subinput->{$_} = $_subinput->{$_} for keys %$_subinput; |
204 |
|
$subinput->id_prefix ('subdoc-' . ++$id_prefix); |
205 |
|
$subinput->nested (1); |
206 |
$subinput->{base_uri} = $subinput->{container_node}->base_uri |
$subinput->{base_uri} = $subinput->{container_node}->base_uri |
207 |
unless defined $subinput->{base_uri}; |
unless defined $subinput->{base_uri}; |
208 |
my $ebaseuri = htescape ($subinput->{base_uri}); |
my $ebaseuri = htescape ($subinput->{base_uri}); |
209 |
push @nav, ['#' . $subinput->{id_prefix} => 'Sub #' . $id_prefix]; |
$out->start_section (id => $subinput->id_prefix, |
210 |
print STDOUT qq[<div id="$subinput->{id_prefix}" class=section> |
title => qq[Subdocument #$id_prefix]); |
211 |
<h2>Subdocument #$id_prefix</h2> |
print STDOUT qq[ |
|
|
|
212 |
<dl> |
<dl> |
213 |
<dt>Internet Media Type</dt> |
<dt>Internet Media Type</dt> |
214 |
<dd><code class="MIME" lang="en">@{[htescape $subinput->{media_type}]}</code> |
<dd><code class="MIME" lang="en">@{[htescape $subinput->{media_type}]}</code> |
219 |
</dl>]; |
</dl>]; |
220 |
|
|
221 |
$subinput->{id_prefix} .= '-'; |
$subinput->{id_prefix} .= '-'; |
222 |
check_and_print ($subinput => $result); |
check_and_print ($subinput => $result => $out); |
|
|
|
|
print STDOUT qq[</div>]; |
|
|
} |
|
|
} # check_and_print |
|
|
|
|
|
sub print_http_header_section ($$) { |
|
|
my ($input, $result) = @_; |
|
|
return unless defined $input->{header_status_code} or |
|
|
defined $input->{header_status_text} or |
|
|
@{$input->{header_field} or []}; |
|
|
|
|
|
push @nav, ['#source-header' => 'HTTP Header'] unless $input->{nested}; |
|
|
print STDOUT qq[<div id="$input->{id_prefix}source-header" class="section"> |
|
|
<h2>HTTP Header</h2> |
|
|
|
|
|
<p><strong>Note</strong>: Due to the limitation of the |
|
|
network library in use, the content of this section might |
|
|
not be the real header.</p> |
|
|
|
|
|
<table><tbody> |
|
|
]; |
|
|
|
|
|
if (defined $input->{header_status_code}) { |
|
|
print STDOUT qq[<tr><th scope="row">Status code</th>]; |
|
|
print STDOUT qq[<td><code>@{[htescape ($input->{header_status_code})]}</code></td></tr>]; |
|
|
} |
|
|
if (defined $input->{header_status_text}) { |
|
|
print STDOUT qq[<tr><th scope="row">Status text</th>]; |
|
|
print STDOUT qq[<td><code>@{[htescape ($input->{header_status_text})]}</code></td></tr>]; |
|
|
} |
|
|
|
|
|
for (@{$input->{header_field}}) { |
|
|
print STDOUT qq[<tr><th scope="row"><code>@{[htescape ($_->[0])]}</code></th>]; |
|
|
print STDOUT qq[<td><code>@{[htescape ($_->[1])]}</code></td></tr>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[</tbody></table></div>]; |
|
|
} # print_http_header_section |
|
|
|
|
|
sub print_syntax_error_html_section ($$) { |
|
|
my ($input, $result) = @_; |
|
|
|
|
|
require Encode; |
|
|
require Whatpm::HTML; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}parse-errors" class="section"> |
|
|
<h2>Parse Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, ['#parse-errors' => 'Parse Error'] unless $input->{nested}; |
|
|
|
|
|
my $onerror = sub { |
|
|
my (%opt) = @_; |
|
|
my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level}); |
|
|
print STDOUT qq[<dt class="$cls">], get_error_label ($input, \%opt), |
|
|
qq[</dt>]; |
|
|
$type =~ tr/ /-/; |
|
|
$type =~ s/\|/%7C/g; |
|
|
$msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]]; |
|
|
print STDOUT qq[<dd class="$cls">], get_error_level_label (\%opt); |
|
|
print STDOUT qq[$msg</dd>\n]; |
|
|
|
|
|
add_error ('syntax', \%opt => $result); |
|
|
}; |
|
|
|
|
|
my $doc = $dom->create_document; |
|
|
my $el; |
|
|
my $inner_html_element = $input->{inner_html_element}; |
|
|
if (defined $inner_html_element and length $inner_html_element) { |
|
|
$input->{charset} ||= 'windows-1252'; ## TODO: for now. |
|
|
my $time1 = time; |
|
|
my $t = Encode::decode ($input->{charset}, $input->{s}); |
|
|
$time{decode} = time - $time1; |
|
|
|
|
|
$el = $doc->create_element_ns |
|
|
('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]); |
|
|
$time1 = time; |
|
|
Whatpm::HTML->set_inner_html ($el, $t, $onerror); |
|
|
$time{parse} = time - $time1; |
|
|
} else { |
|
|
my $time1 = time; |
|
|
Whatpm::HTML->parse_byte_string |
|
|
($input->{charset}, $input->{s} => $doc, $onerror); |
|
|
$time{parse_html} = time - $time1; |
|
|
} |
|
|
$doc->manakai_charset ($input->{official_charset}) |
|
|
if defined $input->{official_charset}; |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
|
|
|
return ($doc, $el); |
|
|
} # print_syntax_error_html_section |
|
|
|
|
|
sub print_syntax_error_xml_section ($$) { |
|
|
my ($input, $result) = @_; |
|
|
|
|
|
require Message::DOM::XMLParserTemp; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}parse-errors" class="section"> |
|
|
<h2>Parse Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, ['#parse-errors' => 'Parse Error'] unless $input->{prefix}; |
|
|
|
|
|
my $onerror = sub { |
|
|
my $err = shift; |
|
|
my $line = $err->location->line_number; |
|
|
print STDOUT qq[<dt><a href="#$input->{id_prefix}line-$line">Line $line</a> column ]; |
|
|
print STDOUT $err->location->column_number, "</dt><dd>"; |
|
|
print STDOUT htescape $err->text, "</dd>\n"; |
|
|
|
|
|
add_error ('syntax', {type => $err->text, |
|
|
level => [ |
|
|
$err->SEVERITY_FATAL_ERROR => 'm', |
|
|
$err->SEVERITY_ERROR => 'm', |
|
|
$err->SEVERITY_WARNING => 's', |
|
|
]->[$err->severity]} => $result); |
|
|
|
|
|
return 1; |
|
|
}; |
|
|
|
|
|
my $time1 = time; |
|
|
open my $fh, '<', \($input->{s}); |
|
|
my $doc = Message::DOM::XMLParserTemp->parse_byte_stream |
|
|
($fh => $dom, $onerror, charset => $input->{charset}); |
|
|
$time{parse_xml} = time - $time1; |
|
|
$doc->manakai_charset ($input->{official_charset}) |
|
|
if defined $input->{official_charset}; |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
|
|
|
return ($doc, undef); |
|
|
} # print_syntax_error_xml_section |
|
|
|
|
|
sub get_css_parser () { |
|
|
our $CSSParser; |
|
|
return $CSSParser if $CSSParser; |
|
|
|
|
|
require Whatpm::CSS::Parser; |
|
|
my $p = Whatpm::CSS::Parser->new; |
|
|
|
|
|
$p->{prop}->{$_} = 1 for qw/ |
|
|
alignment-baseline |
|
|
background background-attachment background-color background-image |
|
|
background-position background-position-x background-position-y |
|
|
background-repeat border border-bottom border-bottom-color |
|
|
border-bottom-style border-bottom-width border-collapse border-color |
|
|
border-left border-left-color |
|
|
border-left-style border-left-width border-right border-right-color |
|
|
border-right-style border-right-width |
|
|
border-spacing -manakai-border-spacing-x -manakai-border-spacing-y |
|
|
border-style border-top border-top-color border-top-style border-top-width |
|
|
border-width bottom |
|
|
caption-side clear clip color content counter-increment counter-reset |
|
|
cursor direction display dominant-baseline empty-cells float font |
|
|
font-family font-size font-size-adjust font-stretch |
|
|
font-style font-variant font-weight height left |
|
|
letter-spacing line-height |
|
|
list-style list-style-image list-style-position list-style-type |
|
|
margin margin-bottom margin-left margin-right margin-top marker-offset |
|
|
marks max-height max-width min-height min-width opacity -moz-opacity |
|
|
orphans outline outline-color outline-style outline-width overflow |
|
|
overflow-x overflow-y |
|
|
padding padding-bottom padding-left padding-right padding-top |
|
|
page page-break-after page-break-before page-break-inside |
|
|
position quotes right size table-layout |
|
|
text-align text-anchor text-decoration text-indent text-transform |
|
|
top unicode-bidi vertical-align visibility white-space width widows |
|
|
word-spacing writing-mode z-index |
|
|
/; |
|
|
$p->{prop_value}->{display}->{$_} = 1 for qw/ |
|
|
block clip inline inline-block inline-table list-item none |
|
|
table table-caption table-cell table-column table-column-group |
|
|
table-header-group table-footer-group table-row table-row-group |
|
|
compact marker |
|
|
/; |
|
|
$p->{prop_value}->{position}->{$_} = 1 for qw/ |
|
|
absolute fixed relative static |
|
|
/; |
|
|
$p->{prop_value}->{float}->{$_} = 1 for qw/ |
|
|
left right none |
|
|
/; |
|
|
$p->{prop_value}->{clear}->{$_} = 1 for qw/ |
|
|
left right none both |
|
|
/; |
|
|
$p->{prop_value}->{direction}->{ltr} = 1; |
|
|
$p->{prop_value}->{direction}->{rtl} = 1; |
|
|
$p->{prop_value}->{marks}->{crop} = 1; |
|
|
$p->{prop_value}->{marks}->{cross} = 1; |
|
|
$p->{prop_value}->{'unicode-bidi'}->{$_} = 1 for qw/ |
|
|
normal bidi-override embed |
|
|
/; |
|
|
for my $prop_name (qw/overflow overflow-x overflow-y/) { |
|
|
$p->{prop_value}->{$prop_name}->{$_} = 1 for qw/ |
|
|
visible hidden scroll auto -webkit-marquee -moz-hidden-unscrollable |
|
|
/; |
|
|
} |
|
|
$p->{prop_value}->{visibility}->{$_} = 1 for qw/ |
|
|
visible hidden collapse |
|
|
/; |
|
|
$p->{prop_value}->{'list-style-type'}->{$_} = 1 for qw/ |
|
|
disc circle square decimal decimal-leading-zero |
|
|
lower-roman upper-roman lower-greek lower-latin |
|
|
upper-latin armenian georgian lower-alpha upper-alpha none |
|
|
hebrew cjk-ideographic hiragana katakana hiragana-iroha |
|
|
katakana-iroha |
|
|
/; |
|
|
$p->{prop_value}->{'list-style-position'}->{outside} = 1; |
|
|
$p->{prop_value}->{'list-style-position'}->{inside} = 1; |
|
|
$p->{prop_value}->{'page-break-before'}->{$_} = 1 for qw/ |
|
|
auto always avoid left right |
|
|
/; |
|
|
$p->{prop_value}->{'page-break-after'}->{$_} = 1 for qw/ |
|
|
auto always avoid left right |
|
|
/; |
|
|
$p->{prop_value}->{'page-break-inside'}->{auto} = 1; |
|
|
$p->{prop_value}->{'page-break-inside'}->{avoid} = 1; |
|
|
$p->{prop_value}->{'background-repeat'}->{$_} = 1 for qw/ |
|
|
repeat repeat-x repeat-y no-repeat |
|
|
/; |
|
|
$p->{prop_value}->{'background-attachment'}->{scroll} = 1; |
|
|
$p->{prop_value}->{'background-attachment'}->{fixed} = 1; |
|
|
$p->{prop_value}->{'font-size'}->{$_} = 1 for qw/ |
|
|
xx-small x-small small medium large x-large xx-large |
|
|
-manakai-xxx-large -webkit-xxx-large |
|
|
larger smaller |
|
|
/; |
|
|
$p->{prop_value}->{'font-style'}->{normal} = 1; |
|
|
$p->{prop_value}->{'font-style'}->{italic} = 1; |
|
|
$p->{prop_value}->{'font-style'}->{oblique} = 1; |
|
|
$p->{prop_value}->{'font-variant'}->{normal} = 1; |
|
|
$p->{prop_value}->{'font-variant'}->{'small-caps'} = 1; |
|
|
$p->{prop_value}->{'font-stretch'}->{$_} = 1 for |
|
|
qw/normal wider narrower ultra-condensed extra-condensed |
|
|
condensed semi-condensed semi-expanded expanded |
|
|
extra-expanded ultra-expanded/; |
|
|
$p->{prop_value}->{'text-align'}->{$_} = 1 for qw/ |
|
|
left right center justify begin end |
|
|
/; |
|
|
$p->{prop_value}->{'text-transform'}->{$_} = 1 for qw/ |
|
|
capitalize uppercase lowercase none |
|
|
/; |
|
|
$p->{prop_value}->{'white-space'}->{$_} = 1 for qw/ |
|
|
normal pre nowrap pre-line pre-wrap -moz-pre-wrap |
|
|
/; |
|
|
$p->{prop_value}->{'writing-mode'}->{$_} = 1 for qw/ |
|
|
lr rl tb lr-tb rl-tb tb-rl |
|
|
/; |
|
|
$p->{prop_value}->{'text-anchor'}->{$_} = 1 for qw/ |
|
|
start middle end |
|
|
/; |
|
|
$p->{prop_value}->{'dominant-baseline'}->{$_} = 1 for qw/ |
|
|
auto use-script no-change reset-size ideographic alphabetic |
|
|
hanging mathematical central middle text-after-edge text-before-edge |
|
|
/; |
|
|
$p->{prop_value}->{'alignment-baseline'}->{$_} = 1 for qw/ |
|
|
auto baseline before-edge text-before-edge middle central |
|
|
after-edge text-after-edge ideographic alphabetic hanging |
|
|
mathematical |
|
|
/; |
|
|
$p->{prop_value}->{'text-decoration'}->{$_} = 1 for qw/ |
|
|
none blink underline overline line-through |
|
|
/; |
|
|
$p->{prop_value}->{'caption-side'}->{$_} = 1 for qw/ |
|
|
top bottom left right |
|
|
/; |
|
|
$p->{prop_value}->{'table-layout'}->{auto} = 1; |
|
|
$p->{prop_value}->{'table-layout'}->{fixed} = 1; |
|
|
$p->{prop_value}->{'border-collapse'}->{collapse} = 1; |
|
|
$p->{prop_value}->{'border-collapse'}->{separate} = 1; |
|
|
$p->{prop_value}->{'empty-cells'}->{show} = 1; |
|
|
$p->{prop_value}->{'empty-cells'}->{hide} = 1; |
|
|
$p->{prop_value}->{cursor}->{$_} = 1 for qw/ |
|
|
auto crosshair default pointer move e-resize ne-resize nw-resize n-resize |
|
|
se-resize sw-resize s-resize w-resize text wait help progress |
|
|
/; |
|
|
for my $prop (qw/border-top-style border-left-style |
|
|
border-bottom-style border-right-style outline-style/) { |
|
|
$p->{prop_value}->{$prop}->{$_} = 1 for qw/ |
|
|
none hidden dotted dashed solid double groove ridge inset outset |
|
|
/; |
|
|
} |
|
|
for my $prop (qw/color background-color |
|
|
border-bottom-color border-left-color border-right-color |
|
|
border-top-color border-color/) { |
|
|
$p->{prop_value}->{$prop}->{transparent} = 1; |
|
|
$p->{prop_value}->{$prop}->{flavor} = 1; |
|
|
$p->{prop_value}->{$prop}->{'-manakai-default'} = 1; |
|
|
} |
|
|
$p->{prop_value}->{'outline-color'}->{invert} = 1; |
|
|
$p->{prop_value}->{'outline-color'}->{'-manakai-invert-or-currentcolor'} = 1; |
|
|
$p->{pseudo_class}->{$_} = 1 for qw/ |
|
|
active checked disabled empty enabled first-child first-of-type |
|
|
focus hover indeterminate last-child last-of-type link only-child |
|
|
only-of-type root target visited |
|
|
lang nth-child nth-last-child nth-of-type nth-last-of-type not |
|
|
-manakai-contains -manakai-current |
|
|
/; |
|
|
$p->{pseudo_element}->{$_} = 1 for qw/ |
|
|
after before first-letter first-line |
|
|
/; |
|
|
|
|
|
return $CSSParser = $p; |
|
|
} # get_css_parser |
|
|
|
|
|
sub print_syntax_error_css_section ($$) { |
|
|
my ($input, $result) = @_; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}parse-errors" class="section"> |
|
|
<h2>Parse Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, ['#parse-errors' => 'Parse Error'] unless $input->{nested}; |
|
|
|
|
|
my $p = get_css_parser (); |
|
|
$p->init; |
|
|
$p->{onerror} = sub { |
|
|
my (%opt) = @_; |
|
|
my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level}); |
|
|
if ($opt{token}) { |
|
|
print STDOUT qq[<dt class="$cls"><a href="#$input->{id_prefix}line-$opt{token}->{line}">Line $opt{token}->{line}</a> column $opt{token}->{column}]; |
|
|
} else { |
|
|
print STDOUT qq[<dt class="$cls">Unknown location]; |
|
|
} |
|
|
if (defined $opt{value}) { |
|
|
print STDOUT qq[ (<code>@{[htescape ($opt{value})]}</code>)]; |
|
|
} elsif (defined $opt{token}) { |
|
|
print STDOUT qq[ (<code>@{[htescape (Whatpm::CSS::Tokenizer->serialize_token ($opt{token}))]}</code>)]; |
|
|
} |
|
|
$type =~ tr/ /-/; |
|
|
$type =~ s/\|/%7C/g; |
|
|
$msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]]; |
|
|
print STDOUT qq[<dd class="$cls">], get_error_level_label (\%opt); |
|
|
print STDOUT qq[$msg</dd>\n]; |
|
|
|
|
|
add_error ('syntax', \%opt => $result); |
|
|
}; |
|
|
$p->{href} = $input->{uri}; |
|
|
$p->{base_uri} = $input->{base_uri}; |
|
|
|
|
|
# if ($parse_mode eq 'q') { |
|
|
# $p->{unitless_px} = 1; |
|
|
# $p->{hashless_color} = 1; |
|
|
# } |
|
|
|
|
|
## TODO: Make $input->{s} a ref. |
|
|
|
|
|
my $s = \$input->{s}; |
|
|
my $charset; |
|
|
unless ($input->{is_char_string}) { |
|
|
require Encode; |
|
|
if (defined $input->{charset}) {## TODO: IANA->Perl |
|
|
$charset = $input->{charset}; |
|
|
$s = \(Encode::decode ($input->{charset}, $$s)); |
|
|
} else { |
|
|
## TODO: charset detection |
|
|
$s = \(Encode::decode ($charset = 'utf-8', $$s)); |
|
|
} |
|
|
} |
|
|
|
|
|
my $cssom = $p->parse_char_string ($$s); |
|
|
$cssom->manakai_input_encoding ($charset) if defined $charset; |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
|
|
|
return $cssom; |
|
|
} # print_syntax_error_css_section |
|
|
|
|
|
sub print_syntax_error_manifest_section ($$) { |
|
|
my ($input, $result) = @_; |
|
|
|
|
|
require Whatpm::CacheManifest; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}parse-errors" class="section"> |
|
|
<h2>Parse Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, ['#parse-errors' => 'Parse Error'] unless $input->{nested}; |
|
|
|
|
|
my $onerror = sub { |
|
|
my (%opt) = @_; |
|
|
my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level}); |
|
|
print STDOUT qq[<dt class="$cls">], get_error_label ($input, \%opt), |
|
|
qq[</dt>]; |
|
|
$type =~ tr/ /-/; |
|
|
$type =~ s/\|/%7C/g; |
|
|
$msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]]; |
|
|
print STDOUT qq[<dd class="$cls">], get_error_level_label (\%opt); |
|
|
print STDOUT qq[$msg</dd>\n]; |
|
|
|
|
|
add_error ('syntax', \%opt => $result); |
|
|
}; |
|
|
|
|
|
my $time1 = time; |
|
|
my $manifest = Whatpm::CacheManifest->parse_byte_string |
|
|
($input->{s}, $input->{uri}, $input->{base_uri}, $onerror); |
|
|
$time{parse_manifest} = time - $time1; |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
|
|
|
return $manifest; |
|
|
} # print_syntax_error_manifest_section |
|
|
|
|
|
sub print_source_string_section ($$$) { |
|
|
my $input = shift; |
|
|
my $s; |
|
|
unless ($input->{is_char_string}) { |
|
|
require Encode; |
|
|
my $enc = Encode::find_encoding ($_[1]); ## TODO: charset name -> Perl name |
|
|
return unless $enc; |
|
|
|
|
|
$s = \($enc->decode (${$_[0]})); |
|
|
} else { |
|
|
$s = $_[0]; |
|
|
} |
|
223 |
|
|
224 |
my $i = 1; |
$out->end_section; |
|
push @nav, ['#source-string' => 'Source'] unless $input->{nested}; |
|
|
print STDOUT qq[<div id="$input->{id_prefix}source-string" class="section"> |
|
|
<h2>Document Source</h2> |
|
|
<ol lang="">\n]; |
|
|
if (length $$s) { |
|
|
while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) { |
|
|
print STDOUT qq[<li id="$input->{id_prefix}line-$i">], htescape $1, |
|
|
"</li>\n"; |
|
|
$i++; |
|
|
} |
|
|
if ($$s =~ /\G([^\x0A]+)/gc) { |
|
|
print STDOUT qq[<li id="$input->{id_prefix}line-$i">], htescape $1, |
|
|
"</li>\n"; |
|
|
} |
|
|
} else { |
|
|
print STDOUT q[<li id="$input->{id_prefix}line-1"></li>]; |
|
|
} |
|
|
print STDOUT "</ol></div>"; |
|
|
} # print_input_string_section |
|
|
|
|
|
sub print_document_tree ($$) { |
|
|
my ($input, $node) = @_; |
|
|
|
|
|
my $r = '<ol class="xoxo">'; |
|
|
|
|
|
my @node = ($node); |
|
|
while (@node) { |
|
|
my $child = shift @node; |
|
|
unless (ref $child) { |
|
|
$r .= $child; |
|
|
next; |
|
|
} |
|
|
|
|
|
my $node_id = $input->{id_prefix} . 'node-'.refaddr $child; |
|
|
my $nt = $child->node_type; |
|
|
if ($nt == $child->ELEMENT_NODE) { |
|
|
my $child_nsuri = $child->namespace_uri; |
|
|
$r .= qq[<li id="$node_id" class="tree-element"><code title="@{[defined $child_nsuri ? $child_nsuri : '']}">] . htescape ($child->tag_name) . |
|
|
'</code>'; ## ISSUE: case |
|
|
|
|
|
if ($child->has_attributes) { |
|
|
$r .= '<ul class="attributes">'; |
|
|
for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] } |
|
|
@{$child->attributes}) { |
|
|
$r .= qq[<li id="$input->{id_prefix}$attr->[3]" class="tree-attribute"><code title="@{[defined $attr->[2] ? htescape ($attr->[2]) : '']}">] . htescape ($attr->[0]) . '</code> = '; ## ISSUE: case? |
|
|
$r .= '<q>' . htescape ($attr->[1]) . '</q></li>'; ## TODO: children |
|
|
} |
|
|
$r .= '</ul>'; |
|
|
} |
|
|
|
|
|
if ($child->has_child_nodes) { |
|
|
$r .= '<ol class="children">'; |
|
|
unshift @node, @{$child->child_nodes}, '</ol></li>'; |
|
|
} else { |
|
|
$r .= '</li>'; |
|
|
} |
|
|
} elsif ($nt == $child->TEXT_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-text"><q lang="">' . htescape ($child->data) . '</q></li>'; |
|
|
} elsif ($nt == $child->CDATA_SECTION_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-cdata"><code><[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]></code></li>'; |
|
|
} elsif ($nt == $child->COMMENT_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-comment"><code><!--</code><q lang="">' . htescape ($child->data) . '</q><code>--></code></li>'; |
|
|
} elsif ($nt == $child->DOCUMENT_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-document">Document'; |
|
|
$r .= qq[<ul class="attributes">]; |
|
|
my $cp = $child->manakai_charset; |
|
|
if (defined $cp) { |
|
|
$r .= qq[<li><code>charset</code> parameter = <code>]; |
|
|
$r .= htescape ($cp) . qq[</code></li>]; |
|
|
} |
|
|
$r .= qq[<li><code>inputEncoding</code> = ]; |
|
|
my $ie = $child->input_encoding; |
|
|
if (defined $ie) { |
|
|
$r .= qq[<code>@{[htescape ($ie)]}</code>]; |
|
|
if ($child->manakai_has_bom) { |
|
|
$r .= qq[ (with <code class=charname><abbr>BOM</abbr></code>)]; |
|
|
} |
|
|
} else { |
|
|
$r .= qq[(<code>null</code>)]; |
|
|
} |
|
|
$r .= qq[<li>@{[scalar get_text ('manakaiIsHTML:'.($child->manakai_is_html?1:0))]}</li>]; |
|
|
$r .= qq[<li>@{[scalar get_text ('manakaiCompatMode:'.$child->manakai_compat_mode)]}</li>]; |
|
|
unless ($child->manakai_is_html) { |
|
|
$r .= qq[<li>XML version = <code>@{[htescape ($child->xml_version)]}</code></li>]; |
|
|
if (defined $child->xml_encoding) { |
|
|
$r .= qq[<li>XML encoding = <code>@{[htescape ($child->xml_encoding)]}</code></li>]; |
|
|
} else { |
|
|
$r .= qq[<li>XML encoding = (null)</li>]; |
|
|
} |
|
|
$r .= qq[<li>XML standalone = @{[$child->xml_standalone ? 'true' : 'false']}</li>]; |
|
|
} |
|
|
$r .= qq[</ul>]; |
|
|
if ($child->has_child_nodes) { |
|
|
$r .= '<ol class="children">'; |
|
|
unshift @node, @{$child->child_nodes}, '</ol></li>'; |
|
|
} |
|
|
} elsif ($nt == $child->DOCUMENT_TYPE_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-doctype"><code><!DOCTYPE></code><ul class="attributes">'; |
|
|
$r .= qq[<li class="tree-doctype-name">Name = <q>@{[htescape ($child->name)]}</q></li>]; |
|
|
$r .= qq[<li class="tree-doctype-publicid">Public identifier = <q>@{[htescape ($child->public_id)]}</q></li>]; |
|
|
$r .= qq[<li class="tree-doctype-systemid">System identifier = <q>@{[htescape ($child->system_id)]}</q></li>]; |
|
|
$r .= '</ul></li>'; |
|
|
} elsif ($nt == $child->PROCESSING_INSTRUCTION_NODE) { |
|
|
$r .= qq'<li id="$node_id" class="tree-id"><code><?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?></code></li>'; |
|
|
} else { |
|
|
$r .= qq'<li id="$node_id" class="tree-unknown">@{[$child->node_type]} @{[htescape ($child->node_name)]}</li>'; # error |
|
|
} |
|
|
} |
|
|
|
|
|
$r .= '</ol>'; |
|
|
print STDOUT $r; |
|
|
} # print_document_tree |
|
|
|
|
|
sub print_structure_dump_dom_section ($$$) { |
|
|
my ($input, $doc, $el) = @_; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}document-tree" class="section"> |
|
|
<h2>Document Tree</h2> |
|
|
]; |
|
|
push @nav, [qq[#$input->{id_prefix}document-tree] => 'Tree'] |
|
|
unless $input->{nested}; |
|
|
|
|
|
print_document_tree ($input, $el || $doc); |
|
|
|
|
|
print STDOUT qq[</div>]; |
|
|
} # print_structure_dump_dom_section |
|
|
|
|
|
sub print_structure_dump_cssom_section ($$) { |
|
|
my ($input, $cssom) = @_; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}document-tree" class="section"> |
|
|
<h2>Document Tree</h2> |
|
|
]; |
|
|
push @nav, [qq[#$input->{id_prefix}document-tree] => 'Tree'] |
|
|
unless $input->{nested}; |
|
|
|
|
|
## TODO: |
|
|
print STDOUT "<pre>".htescape ($cssom->css_text)."</pre>"; |
|
|
|
|
|
print STDOUT qq[</div>]; |
|
|
} # print_structure_dump_cssom_section |
|
|
|
|
|
sub print_structure_dump_manifest_section ($$) { |
|
|
my ($input, $manifest) = @_; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="$input->{id_prefix}dump-manifest" class="section"> |
|
|
<h2>Cache Manifest</h2> |
|
|
]; |
|
|
push @nav, [qq[#$input->{id_prefix}dump-manifest] => 'Cache Manifest'] |
|
|
unless $input->{nested}; |
|
|
|
|
|
print STDOUT qq[<dl><dt>Explicit entries</dt>]; |
|
|
my $i = 0; |
|
|
for my $uri (@{$manifest->[0]}) { |
|
|
my $euri = htescape ($uri); |
|
|
print STDOUT qq[<dd id="$input->{id_prefix}index-@{[$i++]}"><code class=uri><<a href="$euri">$euri</a>></code></dd>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[<dt>Fallback entries</dt><dd> |
|
|
<table><thead><tr><th scope=row>Oppotunistic Caching Namespace</th> |
|
|
<th scope=row>Fallback Entry</tr><tbody>]; |
|
|
for my $uri (sort {$a cmp $b} keys %{$manifest->[1]}) { |
|
|
my $euri = htescape ($uri); |
|
|
my $euri2 = htescape ($manifest->[1]->{$uri}); |
|
|
print STDOUT qq[<tr><td id="$input->{id_prefix}index-@{[$i++]}"><code class=uri><<a href="$euri">$euri</a>></code></td> |
|
|
<td id="$input->{id_prefix}index-@{[$i++]}"><code class=uri><<a href="$euri2">$euri2</a>></code></td>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[</table><dt>Online whitelist</dt>]; |
|
|
for my $uri (@{$manifest->[2]}) { |
|
|
my $euri = htescape ($uri); |
|
|
print STDOUT qq[<dd id="$input->{id_prefix}index-@{[$i++]}"><code class=uri><<a href="$euri">$euri</a>></code></dd>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
} # print_structure_dump_manifest_section |
|
|
|
|
|
sub print_structure_error_dom_section ($$$$$) { |
|
|
my ($input, $doc, $el, $result, $onsubdoc) = @_; |
|
|
|
|
|
print STDOUT qq[<div id="$input->{id_prefix}document-errors" class="section"> |
|
|
<h2>Document Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, [qq[#$input->{id_prefix}document-errors] => 'Document Error'] |
|
|
unless $input->{nested}; |
|
|
|
|
|
require Whatpm::ContentChecker; |
|
|
my $onerror = sub { |
|
|
my %opt = @_; |
|
|
my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level}, $opt{node}); |
|
|
$type =~ tr/ /-/; |
|
|
$type =~ s/\|/%7C/g; |
|
|
$msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]]; |
|
|
print STDOUT qq[<dt class="$cls">] . get_error_label ($input, \%opt) . |
|
|
qq[</dt>\n<dd class="$cls">], get_error_level_label (\%opt); |
|
|
print STDOUT $msg, "</dd>\n"; |
|
|
add_error ('structure', \%opt => $result); |
|
|
}; |
|
|
|
|
|
my $elements; |
|
|
my $time1 = time; |
|
|
if ($el) { |
|
|
$elements = Whatpm::ContentChecker->check_element |
|
|
($el, $onerror, $onsubdoc); |
|
|
} else { |
|
|
$elements = Whatpm::ContentChecker->check_document |
|
|
($doc, $onerror, $onsubdoc); |
|
225 |
} |
} |
|
$time{check} = time - $time1; |
|
|
|
|
|
print STDOUT qq[</dl></div>]; |
|
|
|
|
|
return $elements; |
|
|
} # print_structure_error_dom_section |
|
|
|
|
|
sub print_structure_error_manifest_section ($$$) { |
|
|
my ($input, $manifest, $result) = @_; |
|
|
|
|
|
print STDOUT qq[<div id="$input->{id_prefix}document-errors" class="section"> |
|
|
<h2>Document Errors</h2> |
|
|
|
|
|
<dl>]; |
|
|
push @nav, [qq[#$input->{id_prefix}document-errors] => 'Document Error'] |
|
|
unless $input->{nested}; |
|
226 |
|
|
227 |
require Whatpm::CacheManifest; |
$out->input ($original_input); |
228 |
Whatpm::CacheManifest->check_manifest ($manifest, sub { |
} # check_and_print |
|
my %opt = @_; |
|
|
my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level}, $opt{node}); |
|
|
$type =~ tr/ /-/; |
|
|
$type =~ s/\|/%7C/g; |
|
|
$msg .= qq[ [<a href="../error-description#@{[htescape ($type)]}">Description</a>]]; |
|
|
print STDOUT qq[<dt class="$cls">] . get_error_label ($input, \%opt) . |
|
|
qq[</dt>\n<dd class="$cls">], $msg, "</dd>\n"; |
|
|
add_error ('structure', \%opt => $result); |
|
|
}); |
|
|
|
|
|
print STDOUT qq[</div>]; |
|
|
} # print_structure_error_manifest_section |
|
229 |
|
|
230 |
sub print_table_section ($$) { |
sub print_table_section ($$) { |
231 |
my ($input, $tables) = @_; |
my ($input, $tables) = @_; |
232 |
|
|
233 |
push @nav, [qq[#$input->{id_prefix}tables] => 'Tables'] |
# push @nav, [qq[#$input->{id_prefix}tables] => 'Tables'] |
234 |
unless $input->{nested}; |
# unless $input->{nested}; |
235 |
print STDOUT qq[ |
print STDOUT qq[ |
236 |
<div id="$input->{id_prefix}tables" class="section"> |
<div id="$input->{id_prefix}tables" class="section"> |
237 |
<h2>Tables</h2> |
<h2>Tables</h2> |
246 |
require JSON; |
require JSON; |
247 |
|
|
248 |
my $i = 0; |
my $i = 0; |
249 |
for my $table_el (@$tables) { |
for my $table (@$tables) { |
250 |
$i++; |
$i++; |
251 |
print STDOUT qq[<div class="section" id="$input->{id_prefix}table-$i"><h3>] . |
print STDOUT qq[<div class="section" id="$input->{id_prefix}table-$i"><h3>] . |
252 |
get_node_link ($input, $table_el) . q[</h3>]; |
get_node_link ($input, $table->{element}) . q[</h3>]; |
253 |
|
|
254 |
## TODO: Make |ContentChecker| return |form_table| result |
delete $table->{element}; |
255 |
## so that this script don't have to run the algorithm twice. |
|
256 |
my $table = Whatpm::HTMLTable->form_table ($table_el); |
for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}, |
257 |
|
@{$table->{row}}) { |
|
for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) { |
|
258 |
next unless $_; |
next unless $_; |
259 |
delete $_->{element}; |
delete $_->{element}; |
260 |
} |
} |
290 |
sub print_listing_section ($$$) { |
sub print_listing_section ($$$) { |
291 |
my ($opt, $input, $ids) = @_; |
my ($opt, $input, $ids) = @_; |
292 |
|
|
293 |
push @nav, ['#' . $input->{id_prefix} . $opt->{id} => $opt->{label}] |
# push @nav, ['#' . $input->{id_prefix} . $opt->{id} => $opt->{label}] |
294 |
unless $input->{nested}; |
# unless $input->{nested}; |
295 |
print STDOUT qq[ |
print STDOUT qq[ |
296 |
<div id="$input->{id_prefix}$opt->{id}" class="section"> |
<div id="$input->{id_prefix}$opt->{id}" class="section"> |
297 |
<h2>$opt->{heading}</h2> |
<h2>$opt->{heading}</h2> |
307 |
print STDOUT qq[</dl></div>]; |
print STDOUT qq[</dl></div>]; |
308 |
} # print_listing_section |
} # print_listing_section |
309 |
|
|
|
sub print_result_section ($) { |
|
|
my $result = shift; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<div id="result-summary" class="section"> |
|
|
<h2>Result</h2>]; |
|
|
|
|
|
if ($result->{unsupported} and $result->{conforming_max}) { |
|
|
print STDOUT qq[<p class=uncertain id=result-para>The conformance |
|
|
checker cannot decide whether the document is conforming or |
|
|
not, since the document contains one or more unsupported |
|
|
features. The document might or might not be conforming.</p>]; |
|
|
} elsif ($result->{conforming_min}) { |
|
|
print STDOUT qq[<p class=PASS id=result-para>No conformance-error is |
|
|
found in this document.</p>]; |
|
|
} elsif ($result->{conforming_max}) { |
|
|
print STDOUT qq[<p class=SEE-RESULT id=result-para>This document |
|
|
is <strong>likely <em>non</em>-conforming</strong>, but in rare case |
|
|
it might be conforming.</p>]; |
|
|
} else { |
|
|
print STDOUT qq[<p class=FAIL id=result-para>This document is |
|
|
<strong><em>non</em>-conforming</strong>.</p>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[<table> |
|
|
<colgroup><col><colgroup><col><col><col><colgroup><col> |
|
|
<thead> |
|
|
<tr><th scope=col></th> |
|
|
<th scope=col><a href="../error-description#level-m"><em class=rfc2119>MUST</em>‐level |
|
|
Errors</a></th> |
|
|
<th scope=col><a href="../error-description#level-s"><em class=rfc2119>SHOULD</em>‐level |
|
|
Errors</a></th> |
|
|
<th scope=col><a href="../error-description#level-w">Warnings</a></th> |
|
|
<th scope=col>Score</th></tr></thead><tbody>]; |
|
|
|
|
|
my $must_error = 0; |
|
|
my $should_error = 0; |
|
|
my $warning = 0; |
|
|
my $score_min = 0; |
|
|
my $score_max = 0; |
|
|
my $score_base = 20; |
|
|
my $score_unit = $score_base / 100; |
|
|
for ( |
|
|
[Transfer => 'transfer', ''], |
|
|
[Character => 'char', ''], |
|
|
[Syntax => 'syntax', '#parse-errors'], |
|
|
[Structure => 'structure', '#document-errors'], |
|
|
) { |
|
|
$must_error += ($result->{$_->[1]}->{must} += 0); |
|
|
$should_error += ($result->{$_->[1]}->{should} += 0); |
|
|
$warning += ($result->{$_->[1]}->{warning} += 0); |
|
|
$score_min += (($result->{$_->[1]}->{score_min} *= $score_unit) += $score_base); |
|
|
$score_max += (($result->{$_->[1]}->{score_max} *= $score_unit) += $score_base); |
|
|
|
|
|
my $uncertain = $result->{$_->[1]}->{unsupported} ? '?' : ''; |
|
|
my $label = $_->[0]; |
|
|
if ($result->{$_->[1]}->{must} or |
|
|
$result->{$_->[1]}->{should} or |
|
|
$result->{$_->[1]}->{warning} or |
|
|
$result->{$_->[1]}->{unsupported}) { |
|
|
$label = qq[<a href="$_->[2]">$label</a>]; |
|
|
} |
|
|
|
|
|
print STDOUT qq[<tr class="@{[$uncertain ? 'uncertain' : '']}"><th scope=row>$label</th><td class="@{[$result->{$_->[1]}->{must} ? 'FAIL' : '']}">$result->{$_->[1]}->{must}$uncertain</td><td class="@{[$result->{$_->[1]}->{should} ? 'SEE-RESULT' : '']}">$result->{$_->[1]}->{should}$uncertain</td><td>$result->{$_->[1]}->{warning}$uncertain</td>]; |
|
|
if ($uncertain) { |
|
|
print qq[<td class="@{[$result->{$_->[1]}->{must} ? 'FAIL' : $result->{$_->[1]}->{should} ? 'SEE-RESULT' : '']}">−∞..$result->{$_->[1]}->{score_max}</td>]; |
|
|
} elsif ($result->{$_->[1]}->{score_min} != $result->{$_->[1]}->{score_max}) { |
|
|
print qq[<td class="@{[$result->{$_->[1]}->{must} ? 'FAIL' : 'SEE-RESULT']}">$result->{$_->[1]}->{score_min}..$result->{$_->[1]}->{score_max}</td></tr>]; |
|
|
} else { |
|
|
print qq[<td class="@{[$result->{$_->[1]}->{must} ? 'FAIL' : '']}">$result->{$_->[1]}->{score_min}</td></tr>]; |
|
|
} |
|
|
} |
|
|
|
|
|
$score_max += $score_base; |
|
|
|
|
|
print STDOUT qq[ |
|
|
<tr class=uncertain><th scope=row>Semantics</th><td>0?</td><td>0?</td><td>0?</td><td>−∞..$score_base</td></tr> |
|
|
</tbody> |
|
|
<tfoot><tr class=uncertain><th scope=row>Total</th> |
|
|
<td class="@{[$must_error ? 'FAIL' : '']}">$must_error?</td> |
|
|
<td class="@{[$should_error ? 'SEE-RESULT' : '']}">$should_error?</td> |
|
|
<td>$warning?</td> |
|
|
<td class="@{[$must_error ? 'FAIL' : $should_error ? 'SEE-RESULT' : '']}"><strong>−∞..$score_max</strong></td></tr></tfoot> |
|
|
</table> |
|
|
|
|
|
<p><strong>Important</strong>: This conformance checking service |
|
|
is <em>under development</em>. The result above might be <em>wrong</em>.</p> |
|
|
</div>]; |
|
|
push @nav, ['#result-summary' => 'Result']; |
|
|
} # print_result_section |
|
|
|
|
|
sub print_result_unknown_type_section ($$) { |
|
|
my ($input, $result) = @_; |
|
310 |
|
|
311 |
my $euri = htescape ($input->{uri}); |
sub print_rdf_section ($$$) { |
312 |
|
my ($input, $rdfs) = @_; |
313 |
|
|
314 |
|
# push @nav, ['#' . $input->{id_prefix} . 'rdf' => 'RDF'] |
315 |
|
# unless $input->{nested}; |
316 |
print STDOUT qq[ |
print STDOUT qq[ |
317 |
<div id="$input->{id_prefix}parse-errors" class="section"> |
<div id="$input->{id_prefix}rdf" class="section"> |
318 |
<h2>Errors</h2> |
<h2>RDF Triples</h2> |
319 |
|
|
320 |
<dl> |
<dl>]; |
321 |
<dt class=unsupported><code><<a href="$euri">$euri</a>></code></dt> |
my $i = 0; |
322 |
<dd class=unsupported><strong><a href="../error-description#level-u">Not |
for my $rdf (@$rdfs) { |
323 |
supported</a></strong>: |
print STDOUT qq[<dt id="$input->{id_prefix}rdf-@{[$i++]}">]; |
324 |
Media type |
print STDOUT get_node_link ($input, $rdf->[0]); |
325 |
<code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code> |
print STDOUT qq[<dd><dl>]; |
326 |
is not supported.</dd> |
for my $triple (@{$rdf->[1]}) { |
327 |
</dl> |
print STDOUT '<dt>' . get_node_link ($input, $triple->[0]) . '<dd>'; |
328 |
</div> |
print STDOUT get_rdf_resource_html ($triple->[1]); |
329 |
]; |
print STDOUT ' '; |
330 |
push @nav, [qq[#$input->{id_prefix}parse-errors] => 'Errors'] |
print STDOUT get_rdf_resource_html ($triple->[2]); |
331 |
unless $input->{nested}; |
print STDOUT ' '; |
332 |
add_error (char => {level => 'u'} => $result); |
print STDOUT get_rdf_resource_html ($triple->[3]); |
|
add_error (syntax => {level => 'u'} => $result); |
|
|
add_error (structure => {level => 'u'} => $result); |
|
|
} # print_result_unknown_type_section |
|
|
|
|
|
sub print_result_input_error_section ($) { |
|
|
my $input = shift; |
|
|
print STDOUT qq[<div class="section" id="result-summary"> |
|
|
<p><em><strong>Input Error</strong>: @{[htescape ($input->{error_status_text})]}</em></p> |
|
|
</div>]; |
|
|
push @nav, ['#result-summary' => 'Result']; |
|
|
} # print_result_input_error_section |
|
|
|
|
|
sub get_error_label ($$) { |
|
|
my ($input, $err) = @_; |
|
|
|
|
|
my $r = ''; |
|
|
|
|
|
if (defined $err->{line}) { |
|
|
if ($err->{column} > 0) { |
|
|
$r = qq[<a href="#$input->{id_prefix}line-$err->{line}">Line $err->{line}</a> column $err->{column}]; |
|
|
} else { |
|
|
$err->{line} = $err->{line} - 1 || 1; |
|
|
$r = qq[<a href="#$input->{id_prefix}line-$err->{line}">Line $err->{line}</a>]; |
|
333 |
} |
} |
334 |
|
print STDOUT qq[</dl>]; |
335 |
} |
} |
336 |
|
print STDOUT qq[</dl></div>]; |
337 |
|
} # print_rdf_section |
338 |
|
|
339 |
if (defined $err->{node}) { |
sub get_rdf_resource_html ($) { |
340 |
$r .= ' ' if length $r; |
my $resource = shift; |
341 |
$r = get_node_link ($input, $err->{node}); |
if (defined $resource->{uri}) { |
342 |
} |
my $euri = htescape ($resource->{uri}); |
343 |
|
return '<code class=uri><<a href="' . $euri . '">' . $euri . |
344 |
if (defined $err->{index}) { |
'</a>></code>'; |
345 |
if (length $r) { |
} elsif (defined $resource->{bnodeid}) { |
346 |
$r .= ', Index ' . (0+$err->{index}); |
return htescape ('_:' . $resource->{bnodeid}); |
347 |
} else { |
} elsif ($resource->{nodes}) { |
348 |
$r .= "<a href='#$input->{id_prefix}index-@{[0+$err->{index}]}'>Index " |
return '(rdf:XMLLiteral)'; |
349 |
. (0+$err->{index}) . '</a>'; |
} elsif (defined $resource->{value}) { |
350 |
|
my $elang = htescape (defined $resource->{language} |
351 |
|
? $resource->{language} : ''); |
352 |
|
my $r = qq[<q lang="$elang">] . htescape ($resource->{value}) . '</q>'; |
353 |
|
if (defined $resource->{datatype}) { |
354 |
|
my $euri = htescape ($resource->{datatype}); |
355 |
|
$r .= '^^<code class=uri><<a href="' . $euri . '">' . $euri . |
356 |
|
'</a>></code>'; |
357 |
|
} elsif (length $resource->{language}) { |
358 |
|
$r .= '@' . htescape ($resource->{language}); |
359 |
} |
} |
360 |
} |
return $r; |
|
|
|
|
if (defined $err->{value}) { |
|
|
$r .= ' ' if length $r; |
|
|
$r .= '<q><code>' . htescape ($err->{value}) . '</code></q>'; |
|
|
} |
|
|
|
|
|
return $r; |
|
|
} # get_error_label |
|
|
|
|
|
sub get_error_level_label ($) { |
|
|
my $err = shift; |
|
|
|
|
|
my $r = ''; |
|
|
|
|
|
if (not defined $err->{level} or $err->{level} eq 'm') { |
|
|
$r = qq[<strong><a href="../error-description#level-m"><em class=rfc2119>MUST</em>‐level |
|
|
error</a></strong>: ]; |
|
|
} elsif ($err->{level} eq 's') { |
|
|
$r = qq[<strong><a href="../error-description#level-s"><em class=rfc2119>SHOULD</em>‐level |
|
|
error</a></strong>: ]; |
|
|
} elsif ($err->{level} eq 'w') { |
|
|
$r = qq[<strong><a href="../error-description#level-w">Warning</a></strong>: |
|
|
]; |
|
|
} elsif ($err->{level} eq 'u' or $err->{level} eq 'unsupported') { |
|
|
$r = qq[<strong><a href="../error-description#level-u">Not |
|
|
supported</a></strong>: ]; |
|
|
} elsif ($err->{level} eq 'i') { |
|
|
$r = qq[<strong><a href="../error-description#level-i">Information</a></strong>: ]; |
|
361 |
} else { |
} else { |
362 |
my $elevel = htescape ($err->{level}); |
return '??'; |
|
$r = qq[<strong><a href="../error-description#level-$elevel">$elevel</a></strong>: |
|
|
]; |
|
|
} |
|
|
|
|
|
return $r; |
|
|
} # get_error_level_label |
|
|
|
|
|
sub get_node_path ($) { |
|
|
my $node = shift; |
|
|
my @r; |
|
|
while (defined $node) { |
|
|
my $rs; |
|
|
if ($node->node_type == 1) { |
|
|
$rs = $node->manakai_local_name; |
|
|
$node = $node->parent_node; |
|
|
} elsif ($node->node_type == 2) { |
|
|
$rs = '@' . $node->manakai_local_name; |
|
|
$node = $node->owner_element; |
|
|
} elsif ($node->node_type == 3) { |
|
|
$rs = '"' . $node->data . '"'; |
|
|
$node = $node->parent_node; |
|
|
} elsif ($node->node_type == 9) { |
|
|
@r = ('') unless @r; |
|
|
$rs = ''; |
|
|
$node = $node->parent_node; |
|
|
} else { |
|
|
$rs = '#' . $node->node_type; |
|
|
$node = $node->parent_node; |
|
|
} |
|
|
unshift @r, $rs; |
|
363 |
} |
} |
364 |
return join '/', @r; |
} # get_rdf_resource_html |
|
} # get_node_path |
|
|
|
|
|
sub get_node_link ($$) { |
|
|
return qq[<a href="#$_[0]->{id_prefix}node-@{[refaddr $_[1]]}">] . |
|
|
htescape (get_node_path ($_[1])) . qq[</a>]; |
|
|
} # get_node_link |
|
365 |
|
|
366 |
{ |
{ |
367 |
my $Msg = {}; |
my $Msg = {}; |
368 |
|
|
369 |
sub load_text_catalog ($) { |
sub load_text_catalog ($) { |
370 |
|
# my $self = shift; |
371 |
my $lang = shift; # MUST be a canonical lang name |
my $lang = shift; # MUST be a canonical lang name |
372 |
open my $file, '<:utf8', "cc-msg.$lang.txt" |
open my $file, '<:utf8', "cc-msg.$lang.txt" |
373 |
or die "$0: cc-msg.$lang.txt: $!"; |
or die "$0: cc-msg.$lang.txt: $!"; |
380 |
} |
} |
381 |
} # load_text_catalog |
} # load_text_catalog |
382 |
|
|
383 |
sub get_text ($) { |
sub get_text ($;$$) { |
384 |
|
# my $self = shift; |
385 |
my ($type, $level, $node) = @_; |
my ($type, $level, $node) = @_; |
386 |
$type = $level . ':' . $type if defined $level; |
$type = $level . ':' . $type if defined $level; |
387 |
$level = 'm' unless defined $level; |
$level = 'm' unless defined $level; |
390 |
if (defined $Msg->{$type}) { |
if (defined $Msg->{$type}) { |
391 |
my $msg = $Msg->{$type}->[1]; |
my $msg = $Msg->{$type}->[1]; |
392 |
$msg =~ s{<var>\$([0-9]+)</var>}{ |
$msg =~ s{<var>\$([0-9]+)</var>}{ |
393 |
defined $arg[$1] ? htescape ($arg[$1]) : '(undef)'; |
defined $arg[$1] ? ($arg[$1]) : '(undef)'; |
394 |
}ge; |
}ge; ##BUG: ^ must be escaped |
395 |
$msg =~ s{<var>{\@([A-Za-z0-9:_.-]+)}</var>}{ |
$msg =~ s{<var>{\@([A-Za-z0-9:_.-]+)}</var>}{ |
396 |
UNIVERSAL::can ($node, 'get_attribute_ns') |
UNIVERSAL::can ($node, 'get_attribute_ns') |
397 |
? htescape ($node->get_attribute_ns (undef, $1)) : '' |
? ($node->get_attribute_ns (undef, $1)) : '' |
398 |
}ge; |
}ge; ## BUG: ^ must be escaped |
399 |
$msg =~ s{<var>{\@}</var>}{ |
$msg =~ s{<var>{\@}</var>}{ ## BUG: v must be escaped |
400 |
UNIVERSAL::can ($node, 'value') ? htescape ($node->value) : '' |
UNIVERSAL::can ($node, 'value') ? ($node->value) : '' |
401 |
}ge; |
}ge; |
402 |
$msg =~ s{<var>{local-name}</var>}{ |
$msg =~ s{<var>{local-name}</var>}{ |
403 |
UNIVERSAL::can ($node, 'manakai_local_name') |
UNIVERSAL::can ($node, 'manakai_local_name') |
404 |
? htescape ($node->manakai_local_name) : '' |
? ($node->manakai_local_name) : '' |
405 |
}ge; |
}ge; ## BUG: ^ must be escaped |
406 |
$msg =~ s{<var>{element-local-name}</var>}{ |
$msg =~ s{<var>{element-local-name}</var>}{ |
407 |
(UNIVERSAL::can ($node, 'owner_element') and |
(UNIVERSAL::can ($node, 'owner_element') and |
408 |
$node->owner_element) |
$node->owner_element) |
409 |
? htescape ($node->owner_element->manakai_local_name) |
? ($node->owner_element->manakai_local_name) |
410 |
: '' |
: '' ## BUG: ^ must be escaped |
411 |
}ge; |
}ge; |
412 |
return ($type, 'level-' . $level . ' ' . $Msg->{$type}->[0], $msg); |
return ($type, 'level-' . $level . ' ' . $Msg->{$type}->[0], $msg); |
413 |
} elsif ($type =~ s/:([^:]*)$//) { |
} elsif ($type =~ s/:([^:]*)$//) { |
415 |
redo; |
redo; |
416 |
} |
} |
417 |
} |
} |
418 |
return ($type, 'level-'.$level, htescape ($_[0])); |
return ($type, 'level-'.$level, ($_[0])); |
419 |
|
## BUG: ^ must be escaped |
420 |
} # get_text |
} # get_text |
421 |
|
|
422 |
} |
} |
424 |
sub get_input_document ($$) { |
sub get_input_document ($$) { |
425 |
my ($http, $dom) = @_; |
my ($http, $dom) = @_; |
426 |
|
|
427 |
my $request_uri = $http->get_parameter ('uri'); |
require Encode; |
428 |
my $r = {}; |
my $request_uri = Encode::decode ('utf-8', $http->get_parameter ('uri')); |
429 |
|
my $r = WebHACC::Input->new; |
430 |
if (defined $request_uri and length $request_uri) { |
if (defined $request_uri and length $request_uri) { |
431 |
my $uri = $dom->create_uri_reference ($request_uri); |
my $uri = $dom->create_uri_reference ($request_uri); |
432 |
unless ({ |
unless ({ |
433 |
http => 1, |
http => 1, |
434 |
}->{lc $uri->uri_scheme}) { |
}->{lc $uri->uri_scheme}) { |
435 |
return {uri => $request_uri, request_uri => $request_uri, |
$r = WebHACC::Input::Error->new; |
436 |
error_status_text => 'URI scheme not allowed'}; |
$r->{uri} = $request_uri; |
437 |
|
$r->{request_uri} = $request_uri; |
438 |
|
$r->{error_status_text} = 'URL scheme not allowed'; |
439 |
} |
} |
440 |
|
|
441 |
require Message::Util::HostPermit; |
require Message::Util::HostPermit; |
462 |
Allow host=* |
Allow host=* |
463 |
EOH |
EOH |
464 |
unless ($host_permit->check ($uri->uri_host, $uri->uri_port || 80)) { |
unless ($host_permit->check ($uri->uri_host, $uri->uri_port || 80)) { |
465 |
return {uri => $request_uri, request_uri => $request_uri, |
my $r = WebHACC::Input::Error->new; |
466 |
error_status_text => 'Connection to the host is forbidden'}; |
$r->{uri} = $request_uri; |
467 |
|
$r->{request_uri} = $request_uri; |
468 |
|
$r->{error_status_text} = 'Connection to the host is forbidden'; |
469 |
|
return $r; |
470 |
} |
} |
471 |
|
|
472 |
require LWP::UserAgent; |
require LWP::UserAgent; |