Parent Directory
|
Revision Log
|
Patch
| revision 1.4 by wakaba, Fri May 4 09:16:04 2007 UTC | revision 1.155 by wakaba, Sat Aug 30 12:57:05 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 | ## This is an early version of an HTML parser. | ## ISSUE: |
| 7 | ## var doc = implementation.createDocument (null, null, null); | |
| 8 | ## doc.write (''); | |
| 9 | ## alert (doc.compatMode); | |
| 10 | ||
| 11 | require IO::Handle; | |
| 12 | ||
| 13 | my $HTML_NS = q<http://www.w3.org/1999/xhtml>; | |
| 14 | my $MML_NS = q<http://www.w3.org/1998/Math/MathML>; | |
| 15 | my $SVG_NS = q<http://www.w3.org/2000/svg>; | |
| 16 | my $XLINK_NS = q<http://www.w3.org/1999/xlink>; | |
| 17 | my $XML_NS = q<http://www.w3.org/XML/1998/namespace>; | |
| 18 | my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>; | |
| 19 | ||
| 20 | sub A_EL () { 0b1 } | |
| 21 | sub ADDRESS_EL () { 0b10 } | |
| 22 | sub BODY_EL () { 0b100 } | |
| 23 | sub BUTTON_EL () { 0b1000 } | |
| 24 | sub CAPTION_EL () { 0b10000 } | |
| 25 | sub DD_EL () { 0b100000 } | |
| 26 | sub DIV_EL () { 0b1000000 } | |
| 27 | sub DT_EL () { 0b10000000 } | |
| 28 | sub FORM_EL () { 0b100000000 } | |
| 29 | sub FORMATTING_EL () { 0b1000000000 } | |
| 30 | sub FRAMESET_EL () { 0b10000000000 } | |
| 31 | sub HEADING_EL () { 0b100000000000 } | |
| 32 | sub HTML_EL () { 0b1000000000000 } | |
| 33 | sub LI_EL () { 0b10000000000000 } | |
| 34 | sub NOBR_EL () { 0b100000000000000 } | |
| 35 | sub OPTION_EL () { 0b1000000000000000 } | |
| 36 | sub OPTGROUP_EL () { 0b10000000000000000 } | |
| 37 | sub P_EL () { 0b100000000000000000 } | |
| 38 | sub SELECT_EL () { 0b1000000000000000000 } | |
| 39 | sub TABLE_EL () { 0b10000000000000000000 } | |
| 40 | sub TABLE_CELL_EL () { 0b100000000000000000000 } | |
| 41 | sub TABLE_ROW_EL () { 0b1000000000000000000000 } | |
| 42 | sub TABLE_ROW_GROUP_EL () { 0b10000000000000000000000 } | |
| 43 | sub MISC_SCOPING_EL () { 0b100000000000000000000000 } | |
| 44 | sub MISC_SPECIAL_EL () { 0b1000000000000000000000000 } | |
| 45 | sub FOREIGN_EL () { 0b10000000000000000000000000 } | |
| 46 | sub FOREIGN_FLOW_CONTENT_EL () { 0b100000000000000000000000000 } | |
| 47 | sub MML_AXML_EL () { 0b1000000000000000000000000000 } | |
| 48 | sub RUBY_EL () { 0b10000000000000000000000000000 } | |
| 49 | sub RUBY_COMPONENT_EL () { 0b100000000000000000000000000000 } | |
| 50 | ||
| 51 | sub TABLE_ROWS_EL () { | |
| 52 | TABLE_EL | | |
| 53 | TABLE_ROW_EL | | |
| 54 | TABLE_ROW_GROUP_EL | |
| 55 | } | |
| 56 | ||
| 57 | ## NOTE: Used in "generate implied end tags" algorithm. | |
| 58 | ## NOTE: There is a code where a modified version of END_TAG_OPTIONAL_EL | |
| 59 | ## is used in "generate implied end tags" implementation (search for the | |
| 60 | ## function mae). | |
| 61 | sub END_TAG_OPTIONAL_EL () { | |
| 62 | DD_EL | | |
| 63 | DT_EL | | |
| 64 | LI_EL | | |
| 65 | P_EL | | |
| 66 | RUBY_COMPONENT_EL | |
| 67 | } | |
| 68 | ||
| 69 | ## NOTE: Used in </body> and EOF algorithms. | |
| 70 | sub ALL_END_TAG_OPTIONAL_EL () { | |
| 71 | DD_EL | | |
| 72 | DT_EL | | |
| 73 | LI_EL | | |
| 74 | P_EL | | |
| 75 | ||
| 76 | BODY_EL | | |
| 77 | HTML_EL | | |
| 78 | TABLE_CELL_EL | | |
| 79 | TABLE_ROW_EL | | |
| 80 | TABLE_ROW_GROUP_EL | |
| 81 | } | |
| 82 | ||
| 83 | sub SCOPING_EL () { | |
| 84 | BUTTON_EL | | |
| 85 | CAPTION_EL | | |
| 86 | HTML_EL | | |
| 87 | TABLE_EL | | |
| 88 | TABLE_CELL_EL | | |
| 89 | MISC_SCOPING_EL | |
| 90 | } | |
| 91 | ||
| 92 | sub TABLE_SCOPING_EL () { | |
| 93 | HTML_EL | | |
| 94 | TABLE_EL | |
| 95 | } | |
| 96 | ||
| 97 | sub TABLE_ROWS_SCOPING_EL () { | |
| 98 | HTML_EL | | |
| 99 | TABLE_ROW_GROUP_EL | |
| 100 | } | |
| 101 | ||
| 102 | sub TABLE_ROW_SCOPING_EL () { | |
| 103 | HTML_EL | | |
| 104 | TABLE_ROW_EL | |
| 105 | } | |
| 106 | ||
| 107 | sub SPECIAL_EL () { | |
| 108 | ADDRESS_EL | | |
| 109 | BODY_EL | | |
| 110 | DIV_EL | | |
| 111 | ||
| 112 | DD_EL | | |
| 113 | DT_EL | | |
| 114 | LI_EL | | |
| 115 | P_EL | | |
| 116 | ||
| 117 | FORM_EL | | |
| 118 | FRAMESET_EL | | |
| 119 | HEADING_EL | | |
| 120 | OPTION_EL | | |
| 121 | OPTGROUP_EL | | |
| 122 | SELECT_EL | | |
| 123 | TABLE_ROW_EL | | |
| 124 | TABLE_ROW_GROUP_EL | | |
| 125 | MISC_SPECIAL_EL | |
| 126 | } | |
| 127 | ||
| 128 | my $el_category = { | |
| 129 | a => A_EL | FORMATTING_EL, | |
| 130 | address => ADDRESS_EL, | |
| 131 | applet => MISC_SCOPING_EL, | |
| 132 | area => MISC_SPECIAL_EL, | |
| 133 | b => FORMATTING_EL, | |
| 134 | base => MISC_SPECIAL_EL, | |
| 135 | basefont => MISC_SPECIAL_EL, | |
| 136 | bgsound => MISC_SPECIAL_EL, | |
| 137 | big => FORMATTING_EL, | |
| 138 | blockquote => MISC_SPECIAL_EL, | |
| 139 | body => BODY_EL, | |
| 140 | br => MISC_SPECIAL_EL, | |
| 141 | button => BUTTON_EL, | |
| 142 | caption => CAPTION_EL, | |
| 143 | center => MISC_SPECIAL_EL, | |
| 144 | col => MISC_SPECIAL_EL, | |
| 145 | colgroup => MISC_SPECIAL_EL, | |
| 146 | dd => DD_EL, | |
| 147 | dir => MISC_SPECIAL_EL, | |
| 148 | div => DIV_EL, | |
| 149 | dl => MISC_SPECIAL_EL, | |
| 150 | dt => DT_EL, | |
| 151 | em => FORMATTING_EL, | |
| 152 | embed => MISC_SPECIAL_EL, | |
| 153 | fieldset => MISC_SPECIAL_EL, | |
| 154 | font => FORMATTING_EL, | |
| 155 | form => FORM_EL, | |
| 156 | frame => MISC_SPECIAL_EL, | |
| 157 | frameset => FRAMESET_EL, | |
| 158 | h1 => HEADING_EL, | |
| 159 | h2 => HEADING_EL, | |
| 160 | h3 => HEADING_EL, | |
| 161 | h4 => HEADING_EL, | |
| 162 | h5 => HEADING_EL, | |
| 163 | h6 => HEADING_EL, | |
| 164 | head => MISC_SPECIAL_EL, | |
| 165 | hr => MISC_SPECIAL_EL, | |
| 166 | html => HTML_EL, | |
| 167 | i => FORMATTING_EL, | |
| 168 | iframe => MISC_SPECIAL_EL, | |
| 169 | img => MISC_SPECIAL_EL, | |
| 170 | input => MISC_SPECIAL_EL, | |
| 171 | isindex => MISC_SPECIAL_EL, | |
| 172 | li => LI_EL, | |
| 173 | link => MISC_SPECIAL_EL, | |
| 174 | listing => MISC_SPECIAL_EL, | |
| 175 | marquee => MISC_SCOPING_EL, | |
| 176 | menu => MISC_SPECIAL_EL, | |
| 177 | meta => MISC_SPECIAL_EL, | |
| 178 | nobr => NOBR_EL | FORMATTING_EL, | |
| 179 | noembed => MISC_SPECIAL_EL, | |
| 180 | noframes => MISC_SPECIAL_EL, | |
| 181 | noscript => MISC_SPECIAL_EL, | |
| 182 | object => MISC_SCOPING_EL, | |
| 183 | ol => MISC_SPECIAL_EL, | |
| 184 | optgroup => OPTGROUP_EL, | |
| 185 | option => OPTION_EL, | |
| 186 | p => P_EL, | |
| 187 | param => MISC_SPECIAL_EL, | |
| 188 | plaintext => MISC_SPECIAL_EL, | |
| 189 | pre => MISC_SPECIAL_EL, | |
| 190 | rp => RUBY_COMPONENT_EL, | |
| 191 | rt => RUBY_COMPONENT_EL, | |
| 192 | ruby => RUBY_EL, | |
| 193 | s => FORMATTING_EL, | |
| 194 | script => MISC_SPECIAL_EL, | |
| 195 | select => SELECT_EL, | |
| 196 | small => FORMATTING_EL, | |
| 197 | spacer => MISC_SPECIAL_EL, | |
| 198 | strike => FORMATTING_EL, | |
| 199 | strong => FORMATTING_EL, | |
| 200 | style => MISC_SPECIAL_EL, | |
| 201 | table => TABLE_EL, | |
| 202 | tbody => TABLE_ROW_GROUP_EL, | |
| 203 | td => TABLE_CELL_EL, | |
| 204 | textarea => MISC_SPECIAL_EL, | |
| 205 | tfoot => TABLE_ROW_GROUP_EL, | |
| 206 | th => TABLE_CELL_EL, | |
| 207 | thead => TABLE_ROW_GROUP_EL, | |
| 208 | title => MISC_SPECIAL_EL, | |
| 209 | tr => TABLE_ROW_EL, | |
| 210 | tt => FORMATTING_EL, | |
| 211 | u => FORMATTING_EL, | |
| 212 | ul => MISC_SPECIAL_EL, | |
| 213 | wbr => MISC_SPECIAL_EL, | |
| 214 | }; | |
| 215 | ||
| 216 | my $el_category_f = { | |
| 217 | $MML_NS => { | |
| 218 | 'annotation-xml' => MML_AXML_EL, | |
| 219 | mi => FOREIGN_FLOW_CONTENT_EL, | |
| 220 | mo => FOREIGN_FLOW_CONTENT |