--- test/html-webhacc/cc.cgi 2007/06/27 13:30:15 1.3 +++ test/html-webhacc/cc.cgi 2007/09/10 11:51:09 1.19 @@ -2,12 +2,10 @@ use strict; use lib qw[/home/httpd/html/www/markup/html/whatpm - /home/wakaba/work/manakai/lib - /home/wakaba/public_html/-temp/wiki/lib]; + /home/wakaba/work/manakai2/lib]; use CGI::Carp qw[fatalsToBrowser]; use Scalar::Util qw[refaddr]; - -use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module +use Time::HiRes qw/time/; sub htescape ($) { my $s = $_[0]; @@ -15,24 +13,28 @@ $s =~ s//>/g; $s =~ s/"/"/g; - $s =~ s!([\x00-\x09\x0B-\x1F\x7F-\x80])!sprintf 'U+%04X', ord $1!ge; + $s =~ s{([\x00-\x09\x0B-\x1F\x7F-\xA0\x{FEFF}\x{FFFC}-\x{FFFF}])}{ + sprintf 'U+%04X', ord $1; + }ge; return $s; } # htescape -my $http = SuikaWiki::Input::HTTP->new; - -## TODO: _charset_ + use Message::CGI::HTTP; + my $http = Message::CGI::HTTP->new; - my $input_format = $http->parameter ('i') || 'text/html'; - my $inner_html_element = $http->parameter ('e'); - my $input_uri = 'thismessage:/'; - - my $s = $http->parameter ('s'); - if (length $s > 1000_000) { - print STDOUT "Status: 400 Document Too Long\nContent-Type: text/plain; charset=us-ascii\n\nToo long"; + if ($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 @@ -43,197 +45,282 @@ -

Web Document Conformance Checker (beta)

+

Web Document Conformance Checker +(beta)

+]; -
+ $| = 0; + my $input = get_input_document ($http, $dom); + my $inner_html_element = $http->get_parameter ('e'); + my $char_length = 0; + my %time; + + print qq[ +
+
Request URI
+
<@{[htescape $input->{request_uri}]}>
Document URI
-
<@{[htescape $input_uri]}>
-
Internet Media Type
-
@{[htescape $input_format]}
+
<@{[htescape $input->{uri}]}>
]; # no
yet push @nav, ['#document-info' => 'Information']; - require Message::DOM::DOMImplementation; - my $dom = Message::DOM::DOMImplementation->____new; - my $doc; - my $el; - - if ($input_format eq 'text/html') { - require Encode; - require Whatpm::HTML; - - $s = Encode::decode ('utf-8', $s); +if (defined $input->{s}) { + $char_length = length $input->{s}; - print STDOUT qq[ + print STDOUT qq[ +
Base URI
+
<@{[htescape $input->{base_uri}]}>
+
Internet Media Type
+
@{[htescape $input->{media_type}]} + @{[$input->{media_type_overridden} ? '(overridden)' : '']}
Character Encoding
-
(none)
+
@{[defined $input->{charset} ? ''.htescape ($input->{charset}).'' : '(none)']} + @{[$input->{charset_overridden} ? '(overridden)' : '']}
+
Length
+
$char_length byte@{[$char_length == 1 ? '' : 's']}
+]; -
-

Document Source

+ my $result = {}; + print_http_header_section ($input, $result); + + my $doc; + my $el; + + if ($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}), $doc->input_encoding); + } else { + ## TODO: Change HTTP status code?? + print_result_unknown_type_section ($input); + } + + if (defined $doc or defined $el) { + print_structure_dump_section ($doc, $el); + my $elements = print_structure_error_section ($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 %{$elements->{class}}; + } + + print_result_section ($result); +} else { + print STDOUT qq[
]; + print_result_input_error_section ($input); +} + + print STDOUT qq[ +
+ for (@nav) { + print STDOUT qq[
  • $_->[1]
  • ]; + } + print STDOUT qq[ + + + +]; + + for (qw/decode parse parse_xml check/) { + 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[
    +

    HTTP Header

    + +

    Note: Due to the limitation of the +network library in use, the content of this section might +not be the real header.

    + + +]; + + if (defined $input->{header_status_code}) { + print STDOUT qq[]; + print STDOUT qq[]; + } + if (defined $input->{header_status_text}) { + print STDOUT qq[]; + print STDOUT qq[]; + } + + for (@{$input->{header_field}}) { + print STDOUT qq[]; + print STDOUT qq[]; + } + + print STDOUT qq[
    Status code@{[htescape ($input->{header_status_code})]}
    Status text@{[htescape ($input->{header_status_text})]}
    @{[htescape ($_->[0])]}@{[htescape ($_->[1])]}
    ]; +} # 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[

    Parse Errors

    - -
    -]; - } elsif ($input_format eq 'application/xhtml+xml') { - require Message::DOM::XMLParserTemp; - require Encode; - - my $t = Encode::decode ('utf-8', $s); - - print STDOUT qq[ -
    Character Encoding
    -
    (none)
    - - + print STDOUT qq[]; -
    -

    Document Source

    -]; - push @nav, ['#source-string' => 'Source']; - print_source_string (\$t); - 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[

    Parse Errors

    - -
    -]; - } else { - print STDOUT qq[ - - -
    -

    Media type @{[htescape $input_format]} is not supported!

    -
    -]; - push @nav, ['#result-summary' => 'Result']; - } - - - if (defined $doc or defined $el) { - print STDOUT qq[ -
    -

    Document Tree

    -]; - push @nav, ['#document-tree' => 'Tree']; - - print_document_tree ($el || $doc); - - print STDOUT qq[ -
    - -
    -

    Document Errors

    - -
    ]; + + return ($doc, undef); +} # print_syntax_error_xml_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[
    +

    Document Source

    +
      \n]; + if (length $$s) { + while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) { + print STDOUT qq[
    1. ], htescape $1, "
    2. \n"; + $i++; } - - print STDOUT qq[ - -
    -]; - } - - ## TODO: Show result - - print STDOUT qq[ - - - -]; - -exit; - -sub print_source_string ($) { - my $s = $_[0]; - my $i = 1; - print STDOUT qq[
      \n]; - while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) { - print STDOUT qq[
    1. ], htescape $1, "
    2. \n"; - $i++; - } - if ($$s =~ /\G([^\x0A]+)/gc) { - print STDOUT qq[
    3. ], htescape $1, "
    4. \n"; + if ($$s =~ /\G([^\x0A]+)/gc) { + print STDOUT qq[
    5. ], htescape $1, "
    6. \n"; + } + } else { + print STDOUT q[
    7. ]; } - print STDOUT "
    "; -} # print_input_string + print STDOUT ""; +} # print_input_string_section sub print_document_tree ($) { my $node = shift; @@ -250,46 +337,61 @@ my $node_id = 'node-'.refaddr $child; my $nt = $child->node_type; if ($nt == $child->ELEMENT_NODE) { - $r .= qq'
  • ' . htescape ($child->tag_name) . + my $child_nsuri = $child->namespace_uri; + $r .= qq[
  • ] . htescape ($child->tag_name) . ''; ## ISSUE: case if ($child->has_attributes) { $r .= ''; } - if ($node->has_child_nodes) { + if ($child->has_child_nodes) { $r .= '
      '; - unshift @node, @{$child->child_nodes}, '
    '; + unshift @node, @{$child->child_nodes}, '
  • '; + } else { + $r .= ''; } } elsif ($nt == $child->TEXT_NODE) { - $r .= qq'
  • ' . htescape ($child->data) . '
  • '; + $r .= qq'
  • ' . htescape ($child->data) . '
  • '; } elsif ($nt == $child->CDATA_SECTION_NODE) { - $r .= qq'
  • <[CDATA[' . htescape ($child->data) . ']]>
  • '; + $r .= qq'
  • <[CDATA[' . htescape ($child->data) . ']]>
  • '; } elsif ($nt == $child->COMMENT_NODE) { - $r .= qq'
  • <!--' . htescape ($child->data) . '-->
  • '; + $r .= qq'
  • <!--' . htescape ($child->data) . '-->
  • '; } elsif ($nt == $child->DOCUMENT_NODE) { - $r .= qq'
  • Document
  • '; + $r .= qq'
  • Document'; + $r .= qq[]; if ($child->has_child_nodes) { - $r .= '
      '; - unshift @node, @{$child->child_nodes}, '
    '; + $r .= '
      '; + unshift @node, @{$child->child_nodes}, '
  • '; } } elsif ($nt == $child->DOCUMENT_TYPE_NODE) { - $r .= qq'
  • <!DOCTYPE>