/[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 1.29 by wakaba, Mon Jun 25 11:05:57 2007 UTC revision 1.194 by wakaba, Sat Oct 4 05:53:45 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    ## NOTE: This module don't check all HTML5 parse errors; character
7    ## encoding related parse errors are expected to be handled by relevant
8    ## modules.
9    ## Parse errors for control characters that are not allowed in HTML5
10    ## documents, for surrogate code points, and for noncharacter code
11    ## points, as well as U+FFFD substitions for characters whose code points
12    ## is higher than U+10FFFF may be detected by combining the parser with
13    ## the checker implemented by Whatpm::Charset::UnicodeChecker (for its
14    ## usage example, see |t/HTML-tree.t| in the Whatpm package or the
15    ## WebHACC::Language::HTML module in the WebHACC package).
16    
17  ## ISSUE:  ## ISSUE:
18  ## var doc = implementation.createDocument (null, null, null);  ## var doc = implementation.createDocument (null, null, null);
19  ## doc.write ('');  ## doc.write ('');
20  ## alert (doc.compatMode);  ## alert (doc.compatMode);
21    
22  my $permitted_slash_tag_name = {  require IO::Handle;
23    base => 1,  
24    link => 1,  my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
25    meta => 1,  my $MML_NS = q<http://www.w3.org/1998/Math/MathML>;
26    hr => 1,  my $SVG_NS = q<http://www.w3.org/2000/svg>;
27    br => 1,  my $XLINK_NS = q<http://www.w3.org/1999/xlink>;
28    img=> 1,  my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
29    embed => 1,  my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;
30    param => 1,  
31    area => 1,  sub A_EL () { 0b1 }
32    col => 1,  sub ADDRESS_EL () { 0b10 }
33    input => 1,  sub BODY_EL () { 0b100 }
34    sub BUTTON_EL () { 0b1000 }
35    sub CAPTION_EL () { 0b10000 }
36    sub DD_EL () { 0b100000 }
37    sub DIV_EL () { 0b1000000 }
38    sub DT_EL () { 0b10000000 }
39    sub FORM_EL () { 0b100000000 }
40    sub FORMATTING_EL () { 0b1000000000 }
41    sub FRAMESET_EL () { 0b10000000000 }
42    sub HEADING_EL () { 0b100000000000 }
43    sub HTML_EL () { 0b1000000000000 }
44    sub LI_EL () { 0b10000000000000 }
45    sub NOBR_EL () { 0b100000000000000 }
46    sub OPTION_EL () { 0b1000000000000000 }
47    sub OPTGROUP_EL () { 0b10000000000000000 }
48    sub P_EL () { 0b100000000000000000 }
49    sub SELECT_EL () { 0b1000000000000000000 }
50    sub TABLE_EL () { 0b10000000000000000000 }
51    sub TABLE_CELL_EL () { 0b100000000000000000000 }
52    sub TABLE_ROW_EL () { 0b1000000000000000000000 }
53    sub TABLE_ROW_GROUP_EL () { 0b10000000000000000000000 }
54    sub MISC_SCOPING_EL () { 0b100000000000000000000000 }
55    sub MISC_SPECIAL_EL () { 0b1000000000000000000000000 }
56    sub FOREIGN_EL () { 0b10000000000000000000000000 }
57    sub FOREIGN_FLOW_CONTENT_EL () { 0b100000000000000000000000000 }
58    sub MML_AXML_EL () { 0b1000000000000000000000000000 }
59    sub RUBY_EL () { 0b10000000000000000000000000000 }
60    sub RUBY_COMPONENT_EL () { 0b100000000000000000000000000000 }
61    
62    sub TABLE_ROWS_EL () {
63      TABLE_EL |
64      TABLE_ROW_EL |
65      TABLE_ROW_GROUP_EL
66    }
67    
68    ## NOTE: Used in "generate implied end tags" algorithm.
69    ## NOTE: There is a code where a modified version of
70    ## END_TAG_OPTIONAL_EL is used in "generate implied end tags"
71    ## implementation (search for the algorithm name).
72    sub END_TAG_OPTIONAL_EL () {
73      DD_EL |
74      DT_EL |
75      LI_EL |
76      OPTION_EL |
77      OPTGROUP_EL |
78      P_EL |
79      RUBY_COMPONENT_EL
80    }
81    
82    ## NOTE: Used in </body> and EOF algorithms.
83    sub ALL_END_TAG_OPTIONAL_EL () {
84      DD_EL |
85      DT_EL |
86      LI_EL |
87      P_EL |
88    
89      BODY_EL |
90      HTML_EL |
91      TABLE_CELL_EL |
92      TABLE_ROW_EL |
93      TABLE_ROW_GROUP_EL
94    }
95    
96    sub SCOPING_EL () {
97      BUTTON_EL |
98      CAPTION_EL |
99      HTML_EL |
100      TABLE_EL |
101      TABLE_CELL_EL |
102      MISC_SCOPING_EL
103    }
104    
105    sub TABLE_SCOPING_EL () {
106      HTML_EL |
107      TABLE_EL
108    }
109    
110    sub TABLE_ROWS_SCOPING_EL () {
111      HTML_EL |
112      TABLE_ROW_GROUP_EL
113    }
114    
115    sub TABLE_ROW_SCOPING_EL () {
116      HTML_EL |
117      TABLE_ROW_EL
118    }
119    
120    sub SPECIAL_EL () {
121      ADDRESS_EL |
122      BODY_EL |
123      DIV_EL |
124    
125      DD_EL |
126      DT_EL |
127      LI_EL |
128      P_EL |
129    
130      FORM_EL |
131      FRAMESET_EL |
132      HEADING_EL |
133      OPTION_EL |
134      OPTGROUP_EL |
135      SELECT_EL |
136      TABLE_ROW_EL |
137      TABLE_ROW_GROUP_EL |
138      MISC_SPECIAL_EL
139    }
140    
141    my $el_category = {
142      a => A_EL | FORMATTING_EL,
143      address => ADDRESS_EL,
144      applet => MISC_SCOPING_EL,
145      area => MISC_SPECIAL_EL,
146      article => MISC_SPECIAL_EL,
147      aside => MISC_SPECIAL_EL,
148      b => FORMATTING_EL,
149      base => MISC_SPECIAL_EL,
150      basefont => MISC_SPECIAL_EL,
151      bgsound => MISC_SPECIAL_EL,
152      big => FORMATTING_EL,
153      blockquote => MISC_SPECIAL_EL,
154      body => BODY_EL,
155      br => MISC_SPECIAL_EL,
156      button => BUTTON_EL,
157      caption => CAPTION_EL,
158      center => MISC_SPECIAL_EL,
159      col => MISC_SPECIAL_EL,
160      colgroup => MISC_SPECIAL_EL,
161      command => MISC_SPECIAL_EL,
162      datagrid => MISC_SPECIAL_EL,
163      dd => DD_EL,
164      details => MISC_SPECIAL_EL,
165      dialog => MISC_SPECIAL_EL,
166      dir => MISC_SPECIAL_EL,
167      div => DIV_EL,
168      dl => MISC_SPECIAL_EL,
169      dt => DT_EL,
170      em => FORMATTING_EL,
171      embed => MISC_SPECIAL_EL,
172      eventsource => MISC_SPECIAL_EL,
173      fieldset => MISC_SPECIAL_EL,
174      figure => MISC_SPECIAL_EL,
175      font => FORMATTING_EL,
176      footer => MISC_SPECIAL_EL,
177      form => FORM_EL,
178      frame => MISC_SPECIAL_EL,
179      frameset => FRAMESET_EL,
180      h1 => HEADING_EL,
181      h2 => HEADING_EL,
182      h3 => HEADING_EL,
183      h4 => HEADING_EL,
184      h5 => HEADING_EL,
185      h6 => HEADING_EL,
186      head => MISC_SPECIAL_EL,
187      header => MISC_SPECIAL_EL,
188      hr => MISC_SPECIAL_EL,
189      html => HTML_EL,
190      i => FORMATTING_EL,
191  </