/[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 1.49 by wakaba, Sat Jul 21 10:39:45 2007 UTC revision 1.241 by wakaba, Sun Sep 6 12:53:19 2009 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    use Whatpm::HTML::Tokenizer;
7    
8    ## NOTE: This module don't check all HTML5 parse errors; character
9    ## encoding related parse errors are expected to be handled by relevant
10    ## modules.
11    ## Parse errors for control characters that are not allowed in HTML5
12    ## documents, for surrogate code points, and for noncharacter code
13    ## points, as well as U+FFFD substitions for characters whose code points
14    ## is higher than U+10FFFF may be detected by combining the parser with
15    ## the checker implemented by Whatpm::Charset::UnicodeChecker (for its
16    ## usage example, see |t/HTML-tree.t| in the Whatpm package or the
17    ## WebHACC::Language::HTML module in the WebHACC package).
18    
19  ## ISSUE:  ## ISSUE:
20  ## var doc = implementation.createDocument (null, null, null);  ## var doc = implementation.createDocument (null, null, null);
21  ## doc.write ('');  ## doc.write ('');
22  ## alert (doc.compatMode);  ## alert (doc.compatMode);
23    
24  ## ISSUE: HTML5 revision 967 says that the encoding layer MUST NOT  require IO::Handle;
 ## strip BOM and the HTML layer MUST ignore it.  Whether we can do it  
 ## is not yet clear.  
 ## "{U+FEFF}..." in UTF-16BE/UTF-16LE is three or four characters?  
 ## "{U+FEFF}..." in GB18030?  
   
 my $permitted_slash_tag_name = {  
   base => 1,  
   link => 1,  
   meta => 1,  
   hr => 1,  
   br => 1,  
   img=> 1,  
   embed => 1,  
   param => 1,  
   area => 1,  
   col => 1,  
   input => 1,  
 };  
25    
26  my $c1_entity_char = {  ## Namespace URLs
27    0x80 => 0x20AC,  
28    0x81 => 0xFFFD,  my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
29    0x82 => 0x201A,  my $MML_NS = q<http://www.w3.org/1998/Math/MathML>;
30    0x83 => 0x0192,  my $SVG_NS = q<http://www.w3.org/2000/svg>;
31    0x84 => 0x201E,  my $XLINK_NS = q<http://www.w3.org/1999/xlink>;
32    0x85 => 0x2026,  my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
33    0x86 => 0x2020,  my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;
34    0x87 => 0x2021,  
35    0x88 => 0x02C6,  ## Element categories
36    0x89 => 0x2030,  
37    0x8A => 0x0160,  ## Bits 12-15
38    0x8B => 0x2039,  sub SPECIAL_EL () { 0b1_000000000000000 }
39    0x8C => 0x0152,  sub SCOPING_EL () { 0b1_00000000000000 }
40    0x8D => 0xFFFD,  sub FORMATTING_EL () { 0b1_0000000000000 }
41    0x8E => 0x017D,  sub PHRASING_EL () { 0b1_000000000000 }
42    0x8F => 0xFFFD,  
43    0x90 => 0xFFFD,  ## Bits 10-11
44    0x91 => 0x2018,  #sub FOREIGN_EL () { 0b1_00000000000 } # see Whatpm::HTML::Tokenizer
45    0x92 => 0x2019,  sub FOREIGN_FLOW_CONTENT_EL () { 0b1_0000000000 }
46    0x93 => 0x201C,  
47    0x94 => 0x201D,  ## Bits 6-9
48    0x95 => 0x2022,  sub TABLE_SCOPING_EL () { 0b1_000000000 }
49    0x96 => 0x2013,  sub TABLE_ROWS_SCOPING_EL () { 0b1_00000000 }
50    0x97 => 0x2014,  sub TABLE_ROW_SCOPING_EL () { 0b1_0000000 }
51    0x98 => 0x02DC,  sub TABLE_ROWS_EL () { 0b1_000000 }
52    0x99 => 0x2122,  
53    0x9A => 0x0161,  ## Bit 5
54    0x9B => 0x203A,  sub ADDRESS_DIV_P_EL () { 0b1_00000 }
55    0x9C => 0x0153,  
56    0x9D => 0xFFFD,  ## NOTE: Used in </body> and EOF algorithms.
57    0x9E => 0x017E,  ## Bit 4
58    0x9F => 0x0178,  sub ALL_END_TAG_OPTIONAL_EL () { 0b1_0000 }
59  }; # $c1_entity_char  
60    ## NOTE: Used in "generate implied end tags" algorithm.
61  my $special_category = {  ## NOTE: There is a code where a modified version of
62    address => 1, area => 1, base => 1, basefont => 1, bgsound => 1,  ## END_TAG_OPTIONAL_EL is used in "generate implied end tags"
63    blockquote => 1, body => 1, br => 1, center => 1, col => 1, colgroup => 1,  ## implementation (search for the algorithm name).
64    dd => 1, dir => 1, div => 1, dl => 1, dt => 1, embed => 1, fieldset => 1,  ## Bit 3
65    form => 1, frame => 1, frameset => 1, h1 => 1, h2 => 1, h3 => 1,  sub END_TAG_OPTIONAL_EL () { 0b1_000 }
66    h4 => 1, h5 => 1, h6 => 1, head => 1, hr => 1, iframe => 1, image => 1,  
67    img => 1, input => 1, isindex => 1, li => 1, link => 1, listing => 1,  ## Bits 0-2
68    menu => 1, meta => 1, noembed => 1, noframes => 1, noscript => 1,  
69    ol => 1, optgroup => 1, option => 1, p => 1, param => 1, plaintext => 1,  sub MISC_SPECIAL_EL () { SPECIAL_EL | 0b000 }
70    pre => 1, script => 1, select => 1, spacer => 1, style => 1, tbody => 1,  sub FORM_EL () { SPECIAL_EL | 0b001 }
71    textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1,  sub FRAMESET_EL () { SPECIAL_EL | 0b010 }
72  };  sub HEADING_EL () { SPECIAL_EL | 0b011 }
73  my $scoping_category = {  sub SELECT_EL () { SPECIAL_EL | 0b100 }
74    button => 1, caption => 1, html => 1, marquee => 1, object => 1,  sub SCRIPT_EL () { SPECIAL_EL | 0b101 }
75    table => 1, td => 1, th => 1,  
76    sub ADDRESS_DIV_EL () { SPECIAL_EL | ADDRESS_DIV_P_EL | 0b001 }
77    sub BODY_EL () { SPECIAL_EL | ALL_END_TAG_OPTIONAL_EL | 0b001 }
78    
79    sub DTDD_EL () {
80      SPECIAL_EL |
81      END_TAG_OPTIONAL_EL |
82      ALL_END_TAG_OPTIONAL_EL |
83      0b010
84    }
85    sub LI_EL () {
86      SPECIAL_EL |
87      END_TAG_OPTIONAL_EL |
88      ALL_END_TAG_OPTIONAL_EL |
89      0b100
90    }
91    sub P_EL () {
92      SPECIAL_EL |
93      ADDRESS_DIV_P_EL |
94      END_TAG_OPTIONAL_EL |
95      ALL_END_TAG_OPTIONAL_EL |
96      0b001
97    }
98    
99    sub TABLE_ROW_EL () {
100      SPECIAL_EL |
101      TABLE_ROWS_EL |
102      TABLE_ROW_SCOPING_EL |
103      ALL_END_TAG_OPTIONAL_EL |
104      0b001
105    }
106    sub TABLE_ROW_GROUP_EL () {
107      SPECIAL_EL |
108      TABLE_ROWS_EL |
109      TABLE_ROWS_SCOPING_EL |
110      ALL_END_TAG_OPTIONAL_EL |
111      0b001
112    }
113    
114    sub MISC_SCOPING_EL () { SCOPING_EL | 0b000 }
115    sub BUTTON_EL () { SCOPING_EL | 0b001 }
116    sub CAPTION_EL () { SCOPING_EL | 0b010 }
117    sub HTML_EL () {
118      SCOPING_EL |
119      TABLE_SCOPING_EL |
120      TABLE_ROWS_SCOPING_EL |
121      TABLE_ROW_SCOPING_EL |
122      ALL_END_TAG_OPTIONAL_EL |
123      0b001
124    }
125    sub TABLE_EL () {
126      SCOPING_EL |
127      TABLE_ROWS_EL |
128      TABLE_SCOPING_EL |
129      0b001
130    }
131    sub TABLE_CELL_EL () {
132      SCOPING_EL |
133      TABLE_ROW_SCOPING_EL |
134      ALL_END_TAG_OPTIONAL_EL |
135      0b001
136    }
137    
138    sub MISC_FORMATTING_EL () { FORMATTING_EL | 0b000 }
139    sub A_EL () { FORMATTING_EL | 0b001 }
140    sub NOBR_EL () { FORMATTING_EL | 0b010 }
141    
142    sub RUBY_EL () { PHRASING_EL | 0b001 }
143    
144    ## ISSUE: ALL_END_TAG_OPTIONAL_EL?
145    sub OPTGROUP_EL () { PHRASING_EL | END_TAG_OPTIONAL_EL | 0b001 }
146    sub OPTION_EL () { PHRASING_EL | END_TAG_OPTIONAL_EL | 0b010 }
147    sub RUBY_COMPONENT_EL () { PHRASING_EL | END_TAG_OPTIONAL_EL | 0b100 }
148