test/html-webhacc/cc.cgi

#!/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/&/&amp;/g;
  $s =~ s/</&lt;/g;
  $s =~ s/>/&gt;/g;
  $s =~ s/"/&quot;/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

  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

<!DOCTYPE html>
<html lang="en">
<head>
<title>Web Document Conformance Checker (BETA)</title>
<link rel="stylesheet" href="../cc-style.css" type="text/css">
</head>
<body>
<h1><a href="../cc-interface">Web Document Conformance Checker</a> 
(<em>beta</em>)</h1>
];

  $| = 0;
  my $input = get_input_document ($http, $dom);
  my $inner_html_element = $http->get_parameter ('e');
  my $char_length = 0;
  my %time;

  print qq[
<div id="document-info" class="section">
<dl>
<dt>Request URI</dt>
    <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{request_uri}]}">@{[htescape $input->{request_uri}]}</a>&gt;</code></dd>
<dt>Document URI</dt>
    <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{uri}]}">@{[htescape $input->{uri}]}</a>&gt;</code></dd>
]; # no </dl> yet
  push @nav, ['#document-info' => 'Information'];

if (defined $input->{s}) {
  $char_length = length $input->{s};

  print STDOUT qq[
<dt>Base URI</dt>
    <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input->{base_uri}]}">@{[htescape $input->{base_uri}]}</a>&gt;</code></dd>
<dt>Internet Media Type</dt>
    <dd><code class="MIME" lang="en">@{[htescape $input->{media_type}]}</code>
    @{[$input->{media_type_overridden} ? '<em>(overridden)</em>' : '']}</dd>
<dt>Character Encoding</dt>
    <dd>@{[defined $input->{charset} ? '<code class="charset" lang="en">'.htescape ($input->{charset}).'</code>' : '(none)']}
    @{[$input->{charset_overridden} ? '<em>(overridden)</em>' : '']}</dd>
<dt>Length</dt>
    <dd>$char_length byte@{[$char_length == 1 ? '' : 's']}</dd>
</dl>
</div>
];

  my $result = {conforming_min => 1, conforming_max => 1};
  print_http_header_section ($input, $result);

  my $doc;
  my $el;
  my $manifest;

  if ($input->{media_type} eq 'text/html') {
    ($doc, $el) = print_syntax_error_html_section ($input, $result);
    print_source_string_section (\($input->{s}), $input->{charset});
  } elsif ({
            'text/xml' => 1,
            'application/atom+xml' => 1,
            'application/rss+xml' => 1,
            'application/svg+xml' => 1,
            'application/xhtml+xml' => 1,
            'application/xml' => 1,
           }->{$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);
  }

  if (defined $doc or defined $el) {
    print_structure_dump_dom_section ($doc, $el);
    my $elements = print_structure_error_dom_section ($doc, $el, $result);
    print_table_section ($elements->{table}) if @{$elements->{table}};
    print_id_section ($elements->{id}) if keys %{$elements->{id}};
    print_term_section ($elements->{term}) if keys %{$elements->{term}};
    print_class_section ($elements->{class}) if keys %{$elements->{class}};
  } elsif (defined $manifest) {
    print_structure_dump_manifest_section ($manifest);
    print_structure_error_manifest_section ($manifest, $result);
  }

  print_result_section ($result);
} else {
  print STDOUT qq[</dl></div>];
  print_result_input_error_section ($input);
}

  print STDOUT qq[
<ul class="navigation" id="nav-items">
];
  for (@nav) {
    print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>];
  }
  print STDOUT qq[
</ul>
</body>
</html>
];

  for (qw/decode parse 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 '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 print_http_header_section ($$) {
  my ($input, $result) = @_;
  return unless defined $input->{header_status_code} or
      defined $input->{header_status_text} or
      @{$input->{header_field}};
  
  push @nav, ['#source-header' => 'HTTP Header'];
  print STDOUT qq[<div id="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;

  $input->{charset} ||= 'ISO-8859-1'; ## TODO: for now.
  
  my $time1 = time;
  my $t = Encode::decode ($input->{charset}, $input->{s});
  $time{decode} = time - $time1;

  print STDOUT qq[
<div id="parse-errors" class="section">
<h2>Parse Errors</h2>

<dl>];
  push @nav, ['#parse-errors' => 'Parse Error'];

  my $onerror = sub {
    my (%opt) = @_;
    my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level});
    if ($opt{column} > 0) {
      print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n];
    } else {
      $opt{line} = $opt{line} - 1 || 1;
      print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n];
    }
    $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;
  $time1 = time;
  if (defined $inner_html_element and length $inner_html_element) {
    $el = $doc->create_element_ns
        ('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
    Whatpm::HTML->set_inner_html ($el, $t, $onerror);
  } else {
    Whatpm::HTML->parse_string ($t => $doc, $onerror);
  }
  $time{parse} = time - $time1;

  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="parse-errors" class="section">
<h2>Parse Errors</h2>

<dl>];
  push @nav, ['#parse-errors' => 'Parse Error'];

  my $onerror = sub {
    my $err = shift;
    my $line = $err->location->line_number;
    print STDOUT qq[<dt><a href="#line-$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;

  print STDOUT qq[</dl></div>];

  return ($doc, undef);
} # print_syntax_error_xml_section

sub print_syntax_error_manifest_section ($$) {
  my ($input, $result) = @_;

  require Whatpm::CacheManifest;

  print STDOUT qq[
<div id="parse-errors" class="section">
<h2>Parse Errors</h2>

<dl>];
  push @nav, ['#parse-errors' => 'Parse Error'];

  my $onerror = sub {
    my (%opt) = @_;
    my ($type, $cls, $msg) = get_text ($opt{type}, $opt{level});
    print STDOUT qq[<dt class="$cls">], get_error_label (\%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 ($$) {
  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'];
  print STDOUT qq[<div id="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="line-$i">], htescape $1, "</li>\n";
      $i++;
    }
    if ($$s =~ /\G([^\x0A]+)/gc) {
      print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n";
    }
  } else {
    print STDOUT q[<li id="line-1"></li>];
  }
  print STDOUT "</ol></div>";
} # print_input_string_section

sub print_document_tree ($) {
  my $node = shift;
  my $r = '<ol class="xoxo">';

  my @node = ($node);
  while (@node) {
    my $child = shift @node;
    unless (ref $child) {
      $r .= $child;
      next;
    }

    my $node_id = '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}) {
          $r .= qq[<li id="$attr->[3]" class="tree-attribute"><code title="@{[defined $_->[2] ? $_->[2] : '']}">] . htescape ($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>&lt;[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]&gt;</code></li>';
    } elsif ($nt == $child->COMMENT_NODE) {
      $r .= qq'<li id="$node_id" class="tree-comment"><code>&lt;!--</code><q lang="">' . htescape ($child->data) . '</q><code>--&gt;</code></li>';
    } elsif ($nt == $child->DOCUMENT_NODE) {
      $r .= qq'<li id="$node_id" class="tree-document">Document';
      $r .= qq[<ul class="attributes">];
      $r .= qq[<li>@{[scalar get_text ('manakaiIsHTML:'.($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>&lt;!DOCTYPE&gt;</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>&lt;?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?&gt;</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 ($doc, $el) = @_;

  print STDOUT qq[
<div id="document-tree" class="section">
<h2>Document Tree</h2>
];
  push @nav, ['#document-tree' => 'Tree'];

  print_document_tree ($el || $doc);

  print STDOUT qq[</div>];
} # print_structure_dump_dom_section

sub print_structure_dump_manifest_section ($) {
  my $manifest = shift;

  print STDOUT qq[
<div id="dump-manifest" class="section">
<h2>Cache Manifest</h2>
];
  push @nav, ['#dump-manifest' => 'Caceh Manifest'];

  print STDOUT qq[<dl><dt>Explicit entries</dt>];
  for my $uri (@{$manifest->[0]}) {
    my $euri = htescape ($uri);
    print STDOUT qq[<dd><code class=uri>&lt;<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]->{$uri});
    print STDOUT qq[<tr><td><code class=uri>&lt;<a href="$euri">$euri</a>></code></td>
        <td><code class=uri>&lt;<a href="$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><code class=uri>&lt;<a href="$euri">$euri</a>></code></dd>];
  }

  print STDOUT qq[</dl></div>];
} # print_structure_dump_manifest_section

sub print_structure_error_dom_section ($$$) {
  my ($doc, $el, $result) = @_;

  print STDOUT qq[<div id="document-errors" class="section">
<h2>Document Errors</h2>

<dl>];
  push @nav, ['#document-errors' => 'Document Error'];

  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="$cls">] . get_error_label (\%opt) .
        qq[</dt>\n<dd class="$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);
  } else {
    $elements = Whatpm::ContentChecker->check_document ($doc, $onerror);
  }
  $time{check} = time - $time1;

  print STDOUT qq[</dl></div>];

  return $elements;
} # print_structure_error_dom_section

sub print_structure_error_manifest_section ($$$) {
  my ($manifest, $result) = @_;

  print STDOUT qq[<div id="document-errors" class="section">
<h2>Document Errors</h2>

<dl>];
  push @nav, ['#document-errors' => 'Document Error'];

  require Whatpm::CacheManifest;
  Whatpm::CacheManifest->check_manifest ($manifest, sub {
    my %opt = @_;
    my ($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="$cls">] . get_error_label (\%opt) .
        qq[</dt>\n<dd class="$cls">], $msg, "</dd>\n";
    add_error ('structure', \%opt => $result);
  });

  print STDOUT qq[</div>];
} # print_structure_error_manifest_section

sub print_table_section ($) {
  my $tables = shift;
  
  push @nav, ['#tables' => 'Tables'];
  print STDOUT qq[
<div id="tables" class="section">
<h2>Tables</h2>

<!--[if IE]><script type="text/javascript" src="../excanvas.js"></script><![endif]-->
<script src="../table-script.js" type="text/javascript"></script>
<noscript>
<p><em>Structure of tables are visualized here if scripting is enabled.</em></p>
</noscript>
];
  
  require JSON;
  
  my $i = 0;
  for my $table_el (@$tables) {
    $i++;
    print STDOUT qq[<div class="section" id="table-$i"><h3>] .
        get_node_link ($table_el) . q[</h3>];

    ## TODO: Make |ContentChecker| return |form_table| result
    ## so that this script don't have to run the algorithm twice.
    my $table = Whatpm::HTMLTable->form_table ($table_el);
    
    for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) {
      next unless $_;
      delete $_->{element};
    }
    
    for (@{$table->{row_group}}) {
      next unless $_;
      next unless $_->{element};
      $_->{type} = $_->{element}->manakai_local_name;
      delete $_->{element};
    }
    
    for (@{$table->{cell}}) {
      next unless $_;
      for (@{$_}) {
        next unless $_;
        for (@$_) {
          $_->{id} = refaddr $_->{element} if defined $_->{element};