#!/usr/bin/perl
use strict;
use utf8;
use lib qw[/home/httpd/html/www/markup/html/whatpm
/home/wakaba/work/manakai2/lib];
use CGI::Carp qw[fatalsToBrowser];
use Scalar::Util qw[refaddr];
use Time::HiRes qw/time/;
sub htescape ($) {
my $s = $_[0];
$s =~ s/&/&/g;
$s =~ s//>/g;
$s =~ s/"/"/g;
$s =~ s{([\x00-\x09\x0B-\x1F\x7F-\xA0\x{FEFF}\x{FFFC}-\x{FFFF}])}{
sprintf 'U+%04X ', ord $1;
}ge;
return $s;
} # htescape
use Message::CGI::HTTP;
my $http = Message::CGI::HTTP->new;
if ($http->get_meta_variable ('PATH_INFO') ne '/') {
print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400";
exit;
}
binmode STDOUT, ':utf8';
$| = 1;
require Message::DOM::DOMImplementation;
my $dom = Message::DOM::DOMImplementation->new;
load_text_catalog ('en'); ## TODO: conneg
my @nav;
print STDOUT qq[Content-Type: text/html; charset=utf-8
Web Document Conformance Checker (BETA)
Request URI
<@{[htescape $input->{request_uri}]} >Document URI
<@{[htescape $input->{uri}]} >
yet
push @nav, ['#document-info' => 'Information'];
if (defined $input->{s}) {
$char_length = length $input->{s};
print STDOUT qq[
Base URI
<@{[htescape $input->{base_uri}]} >Internet Media Type
@{[htescape $input->{media_type}]}
@{[$input->{media_type_overridden} ? '(overridden) ' : defined $input->{official_type} ? $input->{media_type} eq $input->{official_type} ? '' : '(sniffed; official type is: '.htescape ($input->{official_type}).')' : '(sniffed) ']} Character Encoding
@{[defined $input->{charset} ? ''.htescape ($input->{charset}).'' : '(none)']}
@{[$input->{charset_overridden} ? '(overridden) ' : '']}
Length
$char_length byte@{[$char_length == 1 ? '' : 's']}
];
my $result = {conforming_min => 1, conforming_max => 1};
check_and_print ($input => $result);
print_result_section ($result);
} else {
print STDOUT qq[];
print_result_input_error_section ($input);
}
print STDOUT qq[
];
for (@nav) {
print STDOUT qq[$_->[1]
];
for (qw/decode parse parse_html parse_xml parse_manifest
check check_manifest/) {
next unless defined $time{$_};
open my $file, '>>', ".cc-$_.txt" or die ".cc-$_.txt: $!";
print $file $char_length, "\t", $time{$_}, "\n";
}
exit;
sub add_error ($$$) {
my ($layer, $err, $result) = @_;
if (defined $err->{level}) {
if ($err->{level} eq 's') {
$result->{$layer}->{should}++;
$result->{$layer}->{score_min} -= 2;
$result->{conforming_min} = 0;
} elsif ($err->{level} eq 'w' or $err->{level} eq 'g') {
$result->{$layer}->{warning}++;
} elsif ($err->{level} eq 'u' or $err->{level} eq 'unsupported') {
$result->{$layer}->{unsupported}++;
$result->{unsupported} = 1;
} else {
$result->{$layer}->{must}++;
$result->{$layer}->{score_max} -= 2;
$result->{$layer}->{score_min} -= 2;
$result->{conforming_min} = 0;
$result->{conforming_max} = 0;
}
} else {
$result->{$layer}->{must}++;
$result->{$layer}->{score_max} -= 2;
$result->{$layer}->{score_min} -= 2;
$result->{conforming_min} = 0;
$result->{conforming_max} = 0;
}
} # add_error
sub check_and_print ($$) {
my ($input, $result) = @_;
$input->{id_prefix} = '';
#$input->{nested} = 1/0;
print_http_header_section ($input, $result);
my $doc;
my $el;
my $manifest;
my @subdoc;
if ($input->{media_type} eq 'text/html') {
($doc, $el) = print_syntax_error_html_section ($input, $result);
print_source_string_section
(\($input->{s}), $input->{charset} || $doc->input_encoding);
} elsif ({
'text/xml' => 1,
'application/atom+xml' => 1,
'application/rss+xml' => 1,
'application/svg+xml' => 1,
'application/xhtml+xml' => 1,
'application/xml' => 1,
}->{$input->{media_type}}) {
($doc, $el) = print_syntax_error_xml_section ($input, $result);
print_source_string_section (\($input->{s}), $doc->input_encoding);
} elsif ($input->{media_type} eq 'text/cache-manifest') {
## TODO: MUST be text/cache-manifest
$manifest = print_syntax_error_manifest_section ($input, $result);
print_source_string_section (\($input->{s}), 'utf-8');
} else {
## TODO: Change HTTP status code??
print_result_unknown_type_section ($input, $result);
}
if (defined $doc or defined $el) {
$doc->document_uri ($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, $result, sub {
push @subdoc, shift;
});
print_table_section ($input, $elements->{table}) if @{$elements->{table}};
print_listing_section ({
id => 'identifiers', label => 'IDs', heading => 'Identifiers',
}, $input, $elements->{id}) if keys %{$elements->{id}};
print_listing_section ({
id => 'terms', label => 'Terms', heading => 'Terms',
}, $input, $elements->{term}) if keys %{$elements->{term}};
print_listing_section ({
id => 'classes', label => 'Classes', heading => 'Classes',
}, $input, $elements->{class}) if keys %{$elements->{class}};
} elsif (defined $manifest) {
print_structure_dump_manifest_section ($input, $manifest);
print_structure_error_manifest_section ($input, $manifest, $result);
}
my $id_prefix = 0;
for my $subinput (@subdoc) {
$subinput->{id_prefix} = 'subdoc-' . ++$id_prefix;
$subinput->{nested} = 1;
$subinput->{base_uri} = $subinput->{container_node}->base_uri
unless defined $subinput->{base_uri};
my $ebaseuri = htescape ($subinput->{base_uri});
push @nav, ['#' . $subinput->{id_prefix} => 'Sub #' . $id_prefix];
print STDOUT qq[];
}
} # check_and_print
sub print_http_header_section ($$) {
my ($input, $result) = @_;
return unless defined $input->{header_status_code} or
defined $input->{header_status_text} or
@{$input->{header_field} or []};
push @nav, ['#source-header' => 'HTTP Header'] unless $input->{nested};
print STDOUT qq[];
} # print_http_header_section
sub print_syntax_error_html_section ($$) {
my ($input, $result) = @_;
require Encode;
require Whatpm::HTML;
print STDOUT qq[
];
return ($doc, $el);
} # print_syntax_error_html_section
sub print_syntax_error_xml_section ($$) {
my ($input, $result) = @_;
require Message::DOM::XMLParserTemp;
print STDOUT qq[
];
return ($doc, undef);
} # print_syntax_error_xml_section
sub print_syntax_error_manifest_section ($$) {
my ($input, $result) = @_;
require Whatpm::CacheManifest;
print STDOUT qq[
];
return $manifest;
} # print_syntax_error_manifest_section
sub print_source_string_section ($$) {
require Encode;
my $enc = Encode::find_encoding ($_[1]); ## TODO: charset name -> Perl name
return unless $enc;
my $s = \($enc->decode (${$_[0]}));
my $i = 1;
push @nav, ['#source-string' => 'Source'] unless $input->{nested};
print STDOUT qq[";
} # print_input_string_section
sub print_document_tree ($) {
my $node = shift;
my $r = '';
my @node = ($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;
$r .= qq[] . htescape ($child->tag_name) .
''; ## ISSUE: case
if ($child->has_attributes) {
$r .= '';
for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] }
@{$child->attributes}) {
$r .= qq[] . htescape ($attr->[0]) . ' = '; ## ISSUE: case?
$r .= '' . htescape ($attr->[1]) . ' ';
}
if ($child->has_child_nodes) {
$r .= '';
unshift @node, @{$child->child_nodes}, ' ' . htescape ($child->data) . ' <[CDATA[' . htescape ($child->data) . ' ]]>Document';
$r .= qq[];
my $cp = $child->manakai_charset;
if (defined $cp) {
$r .= qq[charset parameter = ];
$r .= htescape ($cp) . qq[inputEncoding = ];
my $ie = $child->input_encoding;
if (defined $ie) {
$r .= qq[@{[htescape ($ie)]}];
if ($child->manakai_has_bom) {
$r .= qq[ (with BOM null)];
}
$r .= qq[@{[scalar get_text ('manakaiIsHTML:'.($child->manakai_is_html?1:0))]} ];
$r .= qq[@{[scalar get_text ('manakaiCompatMode:'.$child->manakai_compat_mode)]} ];
unless ($child->manakai_is_html) {
$r .= qq[XML version = @{[htescape ($child->xml_version)]} ];
if (defined $child->xml_encoding) {
$r .= qq[XML encoding = @{[htescape ($child->xml_encoding)]} ];
} else {
$r .= qq[XML encoding = (null) ];
}
$r .= qq[XML standalone = @{[$child->xml_standalone ? 'true' : 'false']} ];
}
$r .= qq[ ];
if ($child->has_child_nodes) {
$r .= '';
unshift @node, @{$child->child_nodes}, ' ';
}
} elsif ($nt == $child->DOCUMENT_TYPE_NODE) {
$r .= qq'<!DOCTYPE>';
$r .= qq[Name = @{[htescape ($child->name)]} ];
$r .= qq[Public identifier = @{[htescape ($child->public_id)]} ];
$r .= qq[System identifier = @{[htescape ($child->system_id)]} ];
$r .= ' <?@{[htescape ($child->target)]} @{[htescape ($child->data)]} ?>@{[$child->node_type]} @{[htescape ($child->node_name)]} '; # error
}
}
$r .= ' ';
print STDOUT $r;
} # print_document_tree
sub print_structure_dump_dom_section ($$$) {
my ($input, $doc, $el) = @_;
print STDOUT qq[
Document Tree
];
push @nav, ['#document-tree' => 'Tree'] unless $input->{nested};
print_document_tree ($el || $doc);
print STDOUT qq[];
} # print_structure_dump_dom_section
sub print_structure_dump_manifest_section ($$) {
my ($input, $manifest) = @_;
print STDOUT qq[
];
} # print_structure_dump_manifest_section
sub print_structure_error_dom_section ($$$$$) {
my ($input, $doc, $el, $result, $onsubdoc) = @_;
print STDOUT qq[];
return $elements;
} # print_structure_error_dom_section
sub print_structure_error_manifest_section ($$$) {
my ($input, $manifest, $result) = @_;
print STDOUT qq[];
} # print_structure_error_manifest_section
sub print_table_section ($$) {
my ($input, $tables) = @_;
push @nav, ['#tables' => 'Tables'] unless $input->{nested};
print STDOUT qq[
];
} # print_table_section
sub print_listing_section ($$$) {
my ($opt, $input, $ids) = @_;
push @nav, ['#' . $opt->{id} => $opt->{label}] unless $input->{nested};
print STDOUT qq[
];
} # print_listing_section
sub print_result_section ($) {
my $result = shift;
print STDOUT qq[
Result ];
if ($result->{unsupported} and $result->{conforming_max}) {
print STDOUT qq[
The conformance
checker cannot decide whether the document is conforming or
not, since the document contains one or more unsupported
features. The document might or might not be conforming.
];
} elsif ($result->{conforming_min}) {
print STDOUT qq[
No conformance-error is
found in this document.
];
} elsif ($result->{conforming_max}) {
print STDOUT qq[
This document
is likely non -conforming , but in rare case
it might be conforming.
];
} else {
print STDOUT qq[
This document is
non -conforming
];
}
print STDOUT qq[
MUST ‐level
ErrorsSHOULD ‐level
ErrorsWarnings Score ];
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}) {
$label = qq[$label ];
}
print STDOUT qq[$label $result->{$_->[1]}->{must}$uncertain $result->{$_->[1]}->{should}$uncertain $result->{$_->[1]}->{warning}$uncertain ];
if ($uncertain) {
print qq[−∞..$result->{$_->[1]}->{score_max} ];
} elsif ($result->{$_->[1]}->{score_min} != $result->{$_->[1]}->{score_max}) {
print qq[$result->{$_->[1]}->{score_min}..$result->{$_->[1]}->{score_max} $result->{$_->[1]}->{score_min} ];
}
}
$score_max += $score_base;
print STDOUT qq[
Semantics 0? 0? 0? −∞..$score_base
Total
$must_error?
$should_error?
$warning?
−∞..$score_max
Important : This conformance checking service
is under development . The result above might be wrong .
];
push @nav, ['#result-summary' => 'Result'];
} # print_result_section
sub print_result_unknown_type_section ($$) {
my ($input, $result) = @_;
my $euri = htescape ($input->{uri});
print STDOUT qq[
Errors
<$euri >Not
supported @{[htescape $input->{media_type}]}
is not supported.
];
push @nav, ['#parse-errors' => 'Errors'];
add_error (char => {level => 'u'} => $result);
add_error (syntax => {level => 'u'} => $result);
add_error (structure => {level => 'u'} => $result);
} # print_result_unknown_type_section
sub print_result_input_error_section ($) {
my $input = shift;
print STDOUT qq[
Input Error : @{[htescape ($input->{error_status_text})]}
];
push @nav, ['#result-summary' => 'Result'];
} # print_result_input_error_section
sub get_error_label ($$) {
my ($input, $err) = @_;
my $r = '';
if (defined $err->{line}) {
if ($err->{column} > 0) {
$r = qq[Line $err->{line} column $err->{column}];
} else {
$err->{line} = $err->{line} - 1 || 1;
$r = qq[Line $err->{line} ];
}
}
if (defined $err->{node}) {
$r .= ' ' if length $r;
$r = get_node_link ($input, $err->{node});
}
if (defined $err->{index}) {
$r .= ' ' if length $r;
$r .= 'Index ' . (0+$err->{index});
}
if (defined $err->{value}) {
$r .= ' ' if length $r;
$r .= '' . htescape ($err->{value}) . 'MUST ‐level
errorSHOULD ‐level
errorWarning Not
supported $elevel ] .
htescape (get_node_path ($_[1])) . qq[ ];
} # get_node_link
{
my $Msg = {};
sub load_text_catalog ($) {
my $lang = shift; # MUST be a canonical lang name
open my $file, '<:utf8', "cc-msg.$lang.txt"
or die "$0: cc-msg.$lang.txt: $!";
while (<$file>) {
if (s/^([^;]+);([^;]*);//) {
my ($type, $cls, $msg) = ($1, $2, $_);
$msg =~ tr/\x0D\x0A//d;
$Msg->{$type} = [$cls, $msg];
}
}
} # load_text_catalog
sub get_text ($) {
my ($type, $level, $node) = @_;
$type = $level . ':' . $type if defined $level;
$level = 'm' unless defined $level;
my @arg;
{
if (defined $Msg->{$type}) {
my $msg = $Msg->{$type}->[1];
$msg =~ s{\$([0-9]+) }{
defined $arg[$1] ? htescape ($arg[$1]) : '(undef)';
}ge;
$msg =~ s{{\@([A-Za-z0-9:_.-]+)} }{
UNIVERSAL::can ($node, 'get_attribute_ns')
? htescape ($node->get_attribute_ns (undef, $1)) : ''
}ge;
$msg =~ s{{\@} }{
UNIVERSAL::can ($node, 'value') ? htescape ($node->value) : ''
}ge;
$msg =~ s{{local-name} }{
UNIVERSAL::can ($node, 'manakai_local_name')
? htescape ($node->manakai_local_name) : ''
}ge;
$msg =~ s{{element-local-name} }{
(UNIVERSAL::can ($node, 'owner_element') and
$node->owner_element)
? htescape ($node->owner_element->manakai_local_name)
: ''
}ge;
return ($type, 'level-' . $level . ' ' . $Msg->{$type}->[0], $msg);
} elsif ($type =~ s/:([^:]*)$//) {
unshift @arg, $1;
redo;
}
}
return ($type, 'level-'.$level, htescape ($_[0]));
} # get_text
}
sub get_input_document ($$) {
my ($http, $dom) = @_;
my $request_uri = $http->get_parameter ('uri');
my $r = {};
if (defined $request_uri and length $request_uri) {
my $uri = $dom->create_uri_reference ($request_uri);
unless ({
http => 1,
}->{lc $uri->uri_scheme}) {
return {uri => $request_uri, request_uri => $request_uri,
error_status_text => 'URI scheme not allowed'};
}
require Message::Util::HostPermit;
my $host_permit = new Message::Util::HostPermit;
$host_permit->add_rule (<check ($uri->uri_host, $uri->uri_port || 80)) {
return {uri => $request_uri, request_uri => $request_uri,
error_status_text => 'Connection to the host is forbidden'};
}
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 .
=head1 LICENSE
Copyright 2007 Wakaba
This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.
=cut
## $Date: 2008/02/10 03:11:06 $