/[suikacvs]/markup/html/whatpm/Whatpm/HTML.pm.src
Suika

Diff of /markup/html/whatpm/Whatpm/HTML.pm.src

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

<
revision 1.35 by wakaba, Mon Jul 16 03:21:04 2007 UTC revision 1.106 by wakaba, Sun Mar 9 10:31:19 2008 UTC
# Line 1  Line 1 
1  package Whatpm::HTML;  package Whatpm::HTML;
2  use strict;  use strict;
3  our $VERSION=do{my @r=(q$Revision$=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};  our $VERSION=do{my @r=(q$Revision$=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
4    use Error qw(:try);
5    
6  ## ISSUE:  ## ISSUE:
7  ## var doc = implementation.createDocument (null, null, null);  ## var doc = implementation.createDocument (null, null, null);
8  ## doc.write ('');  ## doc.write ('');
9  ## alert (doc.compatMode);  ## alert (doc.compatMode);
10    
11  ## ISSUE: HTML5 revision 967 says that the encoding layer MUST NOT  ## TODO: Control charcters and noncharacters are not allowed (HTML5 revision 1263)
12  ## strip BOM and the HTML layer MUST ignore it.  Whether we can do it  ## TODO: 1252 parse error (revision 1264)
13  ## is not yet clear.  ## TODO: 8859-11 = 874 (revision 1271)
 ## "{U+FEFF}..." in UTF-16BE/UTF-16LE is three or four characters?  
 ## "{U+FEFF}..." in GB18030?  
14    
15  my $permitted_slash_tag_name = {  my $permitted_slash_tag_name = {
16    base => 1,    base => 1,
# Line 19  my $permitted_slash_tag_name = { Line 18  my $permitted_slash_tag_name = {
18    meta => 1,    meta => 1,
19    hr => 1,    hr => 1,
20    br => 1,    br => 1,
21    img=> 1,    img => 1,
22    embed => 1,    embed => 1,
23    param => 1,    param => 1,
24    area => 1,    area => 1,
# Line 75  my $special_category = { Line 74  my $special_category = {
74    textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1,    textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1,
75  };  };
76  my $scoping_category = {  my $scoping_category = {
77    button => 1, caption => 1, html => 1, marquee => 1, object => 1,    applet => 1, button => 1, caption => 1, html => 1, marquee => 1, object => 1,
78    table => 1, td => 1, th => 1,    table => 1, td => 1, th => 1,
79  };  };
80  my $formatting_category = {  my $formatting_category = {
# Line 84  my $formatting_category = { Line 83  my $formatting_category = {
83  };  };
84  # $phrasing_category: all other elements  # $phrasing_category: all other elements
85    
86    sub parse_byte_string ($$$$;$) {
87      my $self = ref $_[0] ? shift : shift->new;
88      my $charset = shift;
89      my $bytes_s = ref $_[0] ? $_[0] : \($_[0]);
90      my $s;
91      
92      if (defined $charset) {