test/html-webhacc/cc.cgi

#!/usr/bin/perl
use strict;

use lib qw[/home/httpd/html/www/markup/html/whatpm
           /home/wakaba/work/manakai2/lib];
use CGI::Carp qw[fatalsToBrowser];

  require WebHACC::Input;

{
  require Message::CGI::HTTP;
  my $http = Message::CGI::HTTP->new;

  require WebHACC::Output;
  my $out = WebHACC::Output->new;
  $out->handle (*STDOUT);
  $out->set_utf8;

  if ($http->get_meta_variable ('PATH_INFO') ne '/') {
    $out->http_error (404);
    exit;
  }

  ## TODO: We need real conneg support...
  my $primary_language = 'en';
  if ($ENV{HTTP_ACCEPT_LANGUAGE} =~ /ja/) {
    $primary_language = 'ja';
  }
  $out->load_text_catalog ($primary_language);
  
  $out->set_flush;
  $out->http_header;
  $out->html_header;
  $out->unset_flush;

  my $input = get_input_document ($http);
  $out->input ($input);

  require WebHACC::Result;
  my $result = WebHACC::Result->new;
  $result->output ($out);
  $result->{conforming_min} = 1;
  $result->{conforming_max} = 1;

  $out->html ('<script src="../cc-script.js"></script>');

  check_and_print ($input => $result => $out);
  
  $result->generate_result_section;

  $out->nav_list;

  exit;
}

sub check_and_print ($$$) {
  my ($input, $result, $out) = @_;
  my $original_input = $out->input;
  $out->input ($input);

  $input->generate_info_section ($result);

  $input->generate_transfer_sections ($result);

  unless (defined $input->{s}) {
    $result->{conforming_min} = 0;
    return;
  }

  my $checker_class = {
    'text/cache-manifest' => 'WebHACC::Language::CacheManifest',
    'text/css' => 'WebHACC::Language::CSS',
    'text/html' => 'WebHACC::Language::HTML',
    'text/x-webidl' => 'WebHACC::Language::WebIDL',

    'text/xml' => 'WebHACC::Language::XML',
    'application/atom+xml' => 'WebHACC::Language::XML',
    'application/rss+xml' => 'WebHACC::Language::XML',
    'image/svg+xml' => 'WebHACC::Language::XML',
    'application/xhtml+xml' => 'WebHACC::Language::XML',
    'application/xml' => 'WebHACC::Language::XML',
    ## TODO: Should we make all XML MIME Types fall
    ## into this category?

    ## NOTE: This type has different model from normal XML types.
    'application/rdf+xml' => 'WebHACC::Language::XML',
  }->{$input->{media_type}} || 'WebHACC::Language::Default';

  eval qq{ require $checker_class } or die "$0: Loading $checker_class: $@";
  my $checker = $checker_class->new;
  $checker->input ($input);
  $checker->output ($out);
  $checker->result ($result);

  ## TODO: A cache manifest MUST be text/cache-manifest
  ## TODO: WebIDL media type "text/x-webidl"

  $checker->generate_syntax_error_section;
  $checker->generate_source_string_section;

  my @subdoc;
  $checker->onsubdoc (sub {
    push @subdoc, shift;
  });

  $checker->generate_structure_dump_section;
  $checker->generate_structure_error_section;
  $checker->generate_additional_sections;

  my $id_prefix = 0;
  for my $_subinput (@subdoc) {
    my $subinput = WebHACC::Input::Subdocument->new (++$id_prefix);
    $subinput->{$_} = $_subinput->{$_} for keys %$_subinput;
    $subinput->{base_uri} = $subinput->{container_node}->base_uri
        unless defined $subinput->{base_uri};
    $subinput->{parent_input} = $input;

    $subinput->start_section ($result);
    check_and_print ($subinput => $result => $out);
    $subinput->end_section ($result);
  }

  $out->input ($original_input);
} # check_and_print

sub get_input_document ($) {
  my $http = shift;

  require Message::DOM::DOMImplementation;
  my $dom = Message::DOM::DOMImplementation->new;

  require Encode;
  my $request_uri = Encode::decode ('utf-8', $http->get_parameter ('uri'));
  my $r = WebHACC::Input->new;
  if (defined $request_uri and length $request_uri) {
    my $uri = $dom->create_uri_reference ($request_uri);
    unless ({
             http => 1,
            }->{lc $uri->uri_scheme}) {
      $r = WebHACC::Input::Error->new;
      $r->{uri} = $request_uri;
      $r->{request_uri} = $request_uri;
      $r->{error_status_text} = 'URL scheme not allowed';
    }

    require Message::Util::HostPermit;
    my $host_permit = new Message::Util::HostPermit;
    $host_permit->add_rule (<<EOH);
Allow host=suika port=80
Deny host=suika
Allow host=suika.fam.cx port=80
Deny host=suika.fam.cx
Deny host=localhost
Deny host=*.localdomain
Deny ipv4=0.0.0.0/8
Deny ipv4=10.0.0.0/8
Deny ipv4=127.0.0.0/8
Deny ipv4=169.254.0.0/16
Deny ipv4=172.0.0.0/11
Deny ipv4=192.0.2.0/24
Deny ipv4=192.88.99.0/24
Deny ipv4=192.168.0.0/16
Deny ipv4=198.18.0.0/15
Deny ipv4=224.0.0.0/4
Deny ipv4=255.255.255.255/32
Deny ipv6=0::0/0
Allow host=*
EOH
    unless ($host_permit->check ($uri->uri_host, $uri->uri_port || 80)) {
      my $r = WebHACC::Input::Error->new;
      $r->{uri} = $request_uri;
      $r->{request_uri} = $request_uri;
      $r->{error_status_text} = 'Connection to the host is forbidden';
      return $r;
    }

    require LWP::UserAgent;
    my $ua = WDCC::LWPUA->new;
    $ua->{wdcc_dom} = $dom;
    $ua->{wdcc_host_permit} = $host_permit;
    $ua->agent ('Mozilla'); ## TODO: for now.
    $ua->parse_head (0);
    $ua->protocols_allowed ([qw/http/]);
    $ua->max_size (1000_000);
    my $req = HTTP::Request->new (GET => $request_uri);
    $req->header ('Accept-Encoding' => 'identity, *; q=0');
    my $res = $ua->request ($req);
    ## TODO: 401 sets |is_success| true.
    if ($res->is_success or $http->get_parameter ('error-page')) {
      $r->{base_uri} = $res->base; ## NOTE: It does check |Content-Base|, |Content-Location|, and <base>. ## TODO: Use our own code!
      $r->{uri} = $res->request->uri;
      $r->{request_uri} = $request_uri;

      ## TODO: More strict parsing...
      my $ct = $res->header ('Content-Type');
      if (defined $ct and $ct =~ /;\s*charset\s*=\s*"?([^\s;"]+)"?/i) {
        $r->{charset} = lc $1;
        $r->{charset} =~ tr/\\//d;
        $r->{official_charset} = $r->{charset};
      }

      my $input_charset = $http->get_parameter ('charset');
      if (defined $input_charset and length $input_charset) {
        $r->{charset_overridden}
            = (not defined $r->{charset} or $r->{charset} ne $input_charset);
        $r->{charset} = $input_charset;
      }

      ## TODO: Support for HTTP Content-Encoding

      $r->{s} = ''.$res->content;

      require Whatpm::ContentType;
      ($r->{official_type}, $r->{media_type})
          = Whatpm::ContentType->get_sniffed_type
              (get_file_head => sub {
                 return substr $r->{s}, 0, shift;
               },
               http_content_type_byte => $ct,
               has_http_content_encoding =>
                   defined $res->header ('Content-Encoding'),
               supported_image_types => {});
    } else {
      $r->{uri} = $res->request->uri;
      $r->{request_uri} = $request_uri;
      $r->{error_status_text} = $res->status_line;
    }

    $r->{header_field} = [];
    $res->scan (sub {
      push @{$r->{header_field}}, [$_[0], $_[1]];
    });
    $r->{header_status_code} = $res->code;
    $r->{header_status_text} = $res->message;
  } else {
    $r->{s} = ''.$http->get_parameter ('s');
    $r->{uri} = q<thismessage:/>;
    $r->{request_uri} = q<thismessage:/>;
    $r->{base_uri} = q<thismessage:/>;
    $r->{charset} = ''.$http->get_parameter ('_charset_');
    $r->{charset} =~ s/\s+//g;
    $r->{charset} = 'utf-8' if $r->{charset} eq '';
    $r->{official_charset} = $r->{charset};
    $r->{header_field} = [];

    require Whatpm::ContentType;
    ($r->{official_type}, $r->{media_type})
        = Whatpm::ContentType->get_sniffed_type
            (get_file_head => sub {
               return substr $r->{s}, 0, shift;
             },
             http_content_type_byte => undef,
             has_http_content_encoding => 0,
             supported_image_types => {});
  }

  my $input_format = $http->get_parameter ('i');
  if (defined $input_format and length $input_format) {
    $r->{media_type_overridden}
        = (not defined $r->{media_type} or $input_format ne $r->{media_type});
    $r->{media_type} = $input_format;
  }
  if (defined $r->{s} and not defined $r->{media_type}) {
    $r->{media_type} = 'text/html';
    $r->{media_type_overridden} = 1;
  }

  if ($r->{media_type} eq 'text/xml') {
    unless (defined $r->{charset}) {
      $r->{charset} = 'us-ascii';
      $r->{official_charset} = $r->{charset};
    } elsif ($r->{charset_overridden} and $r->{charset} eq 'us-ascii') {
      $r->{charset_overridden} = 0;
    }
  }

  if (length $r->{s} > 1000_000) {
    $r->{error_status_text} = 'Entity-body too large';
    delete $r->{s};
    return $r;
  }

  $r->{inner_html_element} = $http->get_parameter ('e');

  return $r;
} # get_input_document

package WDCC::LWPUA;
BEGIN { push our @ISA, 'LWP::UserAgent'; }

sub redirect_ok {
  my $ua = shift;
  unless ($ua->SUPER::redirect_ok (@_)) {
    return 0;
  }

  my $uris = $_[1]->header ('Location');
  return 0 unless $uris;
  my $uri = $ua->{wdcc_dom}->create_uri_reference ($uris);
  unless ({
           http => 1,
          }->{lc $uri->uri_scheme}) {
    return 0;
  }
  unless ($ua->{wdcc_host_permit}->check ($uri->uri_host, $uri->uri_port || 80)) {
    return 0;
  }
  return 1;
} # redirect_ok

=head1 AUTHOR

Wakaba <w@suika.fam.cx>.

=head1 LICENSE

Copyright 2007-2008 Wakaba <w@suika.fam.cx>

This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.

=cut

## $Date: 2008/07/21 09:40:59 $
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
8	wakaba	1.53	require WebHACC::Input;
9	wakaba	1.2
10	wakaba	1.35	{
11	wakaba	1.58	require Message::CGI::HTTP;
12	wakaba	1.16	my $http = Message::CGI::HTTP->new;
13	wakaba	1.1
14	wakaba	1.58	require WebHACC::Output;
15			my $out = WebHACC::Output->new;
16			$out->handle (*STDOUT);
17			$out->set_utf8;
18
19	wakaba	1.16	if ($http->get_meta_variable ('PATH_INFO') ne '/') {
20	wakaba	1.58	$out->http_error (404);
21	wakaba	1.8	exit;
22			}
23	wakaba	1.59
24			## TODO: We need real conneg support...
25			my $primary_language = 'en';
26			if ($ENV{HTTP_ACCEPT_LANGUAGE} =~ /ja/) {
27			$primary_language = 'ja';
28			}
29			$out->load_text_catalog ($primary_language);
30	wakaba	1.53
31			$out->set_flush;
32	wakaba	1.58	$out->http_header;
33			$out->html_header;
34			$out->unset_flush;
35	wakaba	1.55
36	wakaba	1.58	my $input = get_input_document ($http);
37	wakaba	1.53	$out->input ($input);
38
39	wakaba	1.58	require WebHACC::Result;
40	wakaba	1.55	my $result = WebHACC::Result->new;
41			$result->output ($out);
42			$result->{conforming_min} = 1;
43			$result->{conforming_max} = 1;
44	wakaba	1.14
45	wakaba	1.55	$out->html ('<script src="../cc-script.js"></script>');
46	wakaba	1.54
47	wakaba	1.55	check_and_print ($input => $result => $out);
48
49			$result->generate_result_section;
50	wakaba	1.1
51	wakaba	1.53	$out->nav_list;
52	wakaba	1.16
53	wakaba	1.53	exit;
54	wakaba	1.35	}
55	wakaba	1.1
56	wakaba	1.53	sub check_and_print ($$$) {
57			my ($input, $result, $out) = @_;
58			my $original_input = $out->input;
59			$out->input ($input);
60	wakaba	1.31
61	wakaba	1.55	$input->generate_info_section ($result);
62
63	wakaba	1.54	$input->generate_transfer_sections ($result);
64	wakaba	1.31
65	wakaba	1.55	unless (defined $input->{s}) {
66			$result->{conforming_min} = 0;
67			return;
68			}
69	wakaba	1.31
70	wakaba	1.53	my $checker_class = {
71			'text/cache-manifest' => 'WebHACC::Language::CacheManifest',
72			'text/css' => 'WebHACC::Language::CSS',
73			'text/html' => 'WebHACC::Language::HTML',
74			'text/x-webidl' => 'WebHACC::Language::WebIDL',
75
76			'text/xml' => 'WebHACC::Language::XML',
77			'application/atom+xml' => 'WebHACC::Language::XML',
78			'application/rss+xml' => 'WebHACC::Language::XML',
79			'image/svg+xml' => 'WebHACC::Language::XML',
80			'application/xhtml+xml' => 'WebHACC::Language::XML',
81			'application/xml' => 'WebHACC::Language::XML',
82			## TODO: Should we make all XML MIME Types fall
83			## into this category?
84
85			## NOTE: This type has different model from normal XML types.
86			'application/rdf+xml' => 'WebHACC::Language::XML',
87			}->{$input->{media_type}} \|\| 'WebHACC::Language::Default';
88
89			eval qq{ require $checker_class } or die "$0: Loading $checker_class: $@";
90			my $checker = $checker_class->new;
91			$checker->input ($input);
92			$checker->output ($out);
93			$checker->result ($result);
94
95			## TODO: A cache manifest MUST be text/cache-manifest
96			## TODO: WebIDL media type "text/x-webidl"
97
98			$checker->generate_syntax_error_section;
99			$checker->generate_source_string_section;
100
101	wakaba	1.55	my @subdoc;
102	wakaba	1.53	$checker->onsubdoc (sub {
103			push @subdoc, shift;
104			});
105
106			$checker->generate_structure_dump_section;
107			$checker->generate_structure_error_section;
108			$checker->generate_additional_sections;
109
110	wakaba	1.34	my $id_prefix = 0;
111	wakaba	1.53	for my $_subinput (@subdoc) {
112	wakaba	1.55	my $subinput = WebHACC::Input::Subdocument->new (++$id_prefix);
113	wakaba	1.53	$subinput->{$_} = $_subinput->{$_} for keys %$_subinput;
114	wakaba	1.34	$subinput->{base_uri} = $subinput->{container_node}->base_uri
115			unless defined $subinput->{base_uri};
116	wakaba	1.55	$subinput->{parent_input} = $input;
117	wakaba	1.34
118	wakaba	1.55	$subinput->start_section ($result);
119	wakaba	1.53	check_and_print ($subinput => $result => $out);
120	wakaba	1.55	$subinput->end_section ($result);
121	wakaba	1.34	}
122	wakaba	1.53
123			$out->input ($original_input);
124	wakaba	1.31	} # check_and_print
125
126	wakaba	1.58	sub get_input_document ($) {
127			my $http = shift;
128
129			require Message::DOM::DOMImplementation;
130			my $dom = Message::DOM::DOMImplementation->new;
131	wakaba	1.9
132	wakaba	1.54	require Encode;
133			my $request_uri = Encode::decode ('utf-8', $http->get_parameter ('uri'));
134	wakaba	1.53	my $r = WebHACC::Input->new;
135	wakaba	1.9	if (defined $request_uri and length $request_uri) {
136			my $uri = $dom->create_uri_reference ($request_uri);
137			unless ({
138			http => 1,
139			}->{lc $uri->uri_scheme}) {
140	wakaba	1.54	$r = WebHACC::Input::Error->new;
141			$r->{uri} = $request_uri;
142			$r->{request_uri} = $request_uri;
143			$r->{error_status_text} = 'URL scheme not allowed';
144	wakaba	1.9	}
145
146			require Message::Util::HostPermit;
147			my $host_permit = new Message::Util::HostPermit;
148			$host_permit->add_rule (<<EOH);
149			Allow host=suika port=80
150			Deny host=suika
151			Allow host=suika.fam.cx port=80
152			Deny host=suika.fam.cx
153			Deny host=localhost
154			Deny host=*.localdomain
155			Deny ipv4=0.0.0.0/8
156			Deny ipv4=10.0.0.0/8
157			Deny ipv4=127.0.0.0/8
158			Deny ipv4=169.254.0.0/16
159			Deny ipv4=172.0.0.0/11
160			Deny ipv4=192.0.2.0/24
161			Deny ipv4=192.88.99.0/24
162			Deny ipv4=192.168.0.0/16
163			Deny ipv4=198.18.0.0/15
164			Deny ipv4=224.0.0.0/4
165			Deny ipv4=255.255.255.255/32
166			Deny ipv6=0::0/0
167			Allow host=*
168			EOH
169			unless ($host_permit->check ($uri->uri_host, $uri->uri_port \|\| 80)) {
170	wakaba	1.54	my $r = WebHACC::Input::Error->new;
171			$r->{uri} = $request_uri;
172			$r->{request_uri} = $request_uri;
173			$r->{error_status_text} = 'Connection to the host is forbidden';
174			return $r;
175	wakaba	1.9	}
176
177			require LWP::UserAgent;
178			my $ua = WDCC::LWPUA->new;
179			$ua->{wdcc_dom} = $dom;
180			$ua->{wdcc_host_permit} = $host_permit;
181			$ua->agent ('Mozilla'); ## TODO: for now.
182			$ua->parse_head (0);
183			$ua->protocols_allowed ([qw/http/]);
184			$ua->max_size (1000_000);
185			my $req = HTTP::Request->new (GET => $request_uri);
186	wakaba	1.28	$req->header ('Accept-Encoding' => 'identity, *; q=0');
187	wakaba	1.9	my $res = $ua->request ($req);
188	wakaba	1.16	## TODO: 401 sets \|is_success\| true.
189			if ($res->is_success or $http->get_parameter ('error-page')) {
190	wakaba	1.9	$r->{base_uri} = $res->base; ## NOTE: It does check \|Content-Base\|, \|Content-Location\|, and <base>. ## TODO: Use our own code!
191			$r->{uri} = $res->request->uri;
192			$r->{request_uri} = $request_uri;
193
194			## TODO: More strict parsing...
195			my $ct = $res->header ('Content-Type');
196	wakaba	1.22	if (defined $ct and $ct =~ /;\scharset\s=\s*"?([^\s;"]+)"?/i) {
197	wakaba	1.9	$r->{charset} = lc $1;
198			$r->{charset} =~ tr/\\//d;
199	wakaba	1.26	$r->{official_charset} = $r->{charset};
200	wakaba	1.9	}
201
202	wakaba	1.16	my $input_charset = $http->get_parameter ('charset');
203	wakaba	1.9	if (defined $input_charset and length $input_charset) {
204			$r->{charset_overridden}
205			= (not defined $r->{charset} or $r->{charset} ne $input_charset);
206			$r->{charset} = $input_charset;
207	wakaba	1.25	}
208
209			## TODO: Support for HTTP Content-Encoding
210	wakaba	1.9
211			$r->{s} = ''.$res->content;
212	wakaba	1.25
213			require Whatpm::ContentType;
214			($r->{official_type}, $r->{media_type})
215			= Whatpm::ContentType->get_sniffed_type
216			(get_file_head => sub {
217			return substr $r->{s}, 0, shift;
218			},
219			http_content_type_byte => $ct,
220			has_http_content_encoding =>
221			defined $res->header ('Content-Encoding'),
222			supported_image_types => {});
223	wakaba	1.9	} else {
224			$r->{uri} = $res->request->uri;
225			$r->{request_uri} = $request_uri;
226			$r->{error_status_text} = $res->status_line;
227			}
228
229			$r->{header_field} = [];
230			$res->scan (sub {
231			push @{$r->{header_field}}, [$_[0], $_[1]];
232			});
233			$r->{header_status_code} = $res->code;
234			$r->{header_status_text} = $res->message;
235			} else {
236	wakaba	1.16	$r->{s} = ''.$http->get_parameter ('s');
237	wakaba	1.9	$r->{uri} = q<thismessage:/>;
238			$r->{request_uri} = q<thismessage:/>;
239			$r->{base_uri} = q<thismessage:/>;
240	wakaba	1.16	$r->{charset} = ''.$http->get_parameter ('_charset_');
241	wakaba	1.9	$r->{charset} =~ s/\s+//g;
242			$r->{charset} = 'utf-8' if $r->{charset} eq '';
243	wakaba	1.26	$r->{official_charset} = $r->{charset};
244	wakaba	1.9	$r->{header_field} = [];
245	wakaba	1.25
246			require Whatpm::ContentType;
247			($r->{official_type}, $r->{media_type})
248			= Whatpm::ContentType->get_sniffed_type
249			(get_file_head => sub {
250			return substr $r->{s}, 0, shift;
251			},
252			http_content_type_byte => undef,
253			has_http_content_encoding => 0,
254			supported_image_types => {});
255	wakaba	1.9	}
256
257	wakaba	1.16	my $input_format = $http->get_parameter ('i');
258	wakaba	1.9	if (defined $input_format and length $input_format) {
259			$r->{media_type_overridden}
260			= (not defined $r->{media_type} or $input_format ne $r->{media_type});
261			$r->{media_type} = $input_format;
262			}
263			if (defined $r->{s} and not defined $r->{media_type}) {
264			$r->{media_type} = 'text/html';
265			$r->{media_type_overridden} = 1;
266			}
267
268			if ($r->{media_type} eq 'text/xml') {
269			unless (defined $r->{charset}) {
270			$r->{charset} = 'us-ascii';
271	wakaba	1.26	$r->{official_charset} = $r->{charset};
272	wakaba	1.9	} elsif ($r->{charset_overridden} and $r->{charset} eq 'us-ascii') {
273			$r->{charset_overridden} = 0;
274			}
275			}
276
277			if (length $r->{s} > 1000_000) {
278			$r->{error_status_text} = 'Entity-body too large';
279			delete $r->{s};
280			return $r;
281			}
282
283	wakaba	1.35	$r->{inner_html_element} = $http->get_parameter ('e');
284
285	wakaba	1.9	return $r;
286			} # get_input_document
287
288			package WDCC::LWPUA;
289			BEGIN { push our @ISA, 'LWP::UserAgent'; }
290
291			sub redirect_ok {
292			my $ua = shift;
293			unless ($ua->SUPER::redirect_ok (@_)) {
294			return 0;
295			}
296
297			my $uris = $_[1]->header ('Location');
298			return 0 unless $uris;
299			my $uri = $ua->{wdcc_dom}->create_uri_reference ($uris);
300			unless ({
301			http => 1,
302			}->{lc $uri->uri_scheme}) {
303			return 0;
304			}
305			unless ($ua->{wdcc_host_permit}->check ($uri->uri_host, $uri->uri_port \|\| 80)) {
306			return 0;
307			}
308			return 1;
309			} # redirect_ok
310
311	wakaba	1.1	=head1 AUTHOR
312
313			Wakaba <w@suika.fam.cx>.
314
315			=head1 LICENSE
316
317	wakaba	1.35	Copyright 2007-2008 Wakaba <w@suika.fam.cx>
318	wakaba	1.1
319			This library is free software; you can redistribute it
320			and/or modify it under the same terms as Perl itself.
321
322			=cut
323
324	wakaba	1.59	## $Date: 2008/07/21 09:40:59 $