| 12 |
## TODO: 1252 parse error (revision 1264) |
## TODO: 1252 parse error (revision 1264) |
| 13 |
## TODO: 8859-11 = 874 (revision 1271) |
## TODO: 8859-11 = 874 (revision 1271) |
| 14 |
|
|
| 15 |
my $permitted_slash_tag_name = { |
my $HTML_NS = q<http://www.w3.org/1999/xhtml>; |
| 16 |
base => 1, |
my $MML_NS = q<http://www.w3.org/1998/Math/MathML>; |
| 17 |
link => 1, |
my $SVG_NS = q<http://www.w3.org/2000/svg>; |
| 18 |
meta => 1, |
my $XLINK_NS = q<http://www.w3.org/1999/xlink>; |
| 19 |
hr => 1, |
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>; |
| 20 |
br => 1, |
my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>; |
| 21 |
img => 1, |
|
| 22 |
embed => 1, |
sub A_EL () { 0b1 } |
| 23 |
param => 1, |
sub ADDRESS_EL () { 0b10 } |
| 24 |
area => 1, |
sub BODY_EL () { 0b100 } |
| 25 |
col => 1, |
sub BUTTON_EL () { 0b1000 } |
| 26 |
input => 1, |
sub CAPTION_EL () { 0b10000 } |
| 27 |
|
sub DD_EL () { 0b100000 } |
| 28 |
|
sub DIV_EL () { 0b1000000 } |
| 29 |
|
sub DT_EL () { 0b10000000 } |
| 30 |
|
sub FORM_EL () { 0b100000000 } |
| 31 |
|
sub FORMATTING_EL () { 0b1000000000 } |
| 32 |
|
sub FRAMESET_EL () { 0b10000000000 } |
| 33 |
|
sub HEADING_EL () { 0b100000000000 } |
| 34 |
|
sub HTML_EL () { 0b1000000000000 } |
| 35 |
|
sub LI_EL () { 0b10000000000000 } |
| 36 |
|
sub NOBR_EL () { 0b100000000000000 } |
| 37 |
|
sub OPTION_EL () { 0b1000000000000000 } |
| 38 |
|
sub OPTGROUP_EL () { 0b10000000000000000 } |
| 39 |
|
sub P_EL () { 0b100000000000000000 } |
| 40 |
|
sub SELECT_EL () { 0b1000000000000000000 } |
| 41 |
|
sub TABLE_EL () { 0b10000000000000000000 } |
| 42 |
|
sub TABLE_CELL_EL () { 0b100000000000000000000 } |
| 43 |
|
sub TABLE_ROW_EL () { 0b1000000000000000000000 } |
| 44 |
|
sub TABLE_ROW_GROUP_EL () { 0b10000000000000000000000 } |
| 45 |
|
sub MISC_SCOPING_EL () { 0b100000000000000000000000 } |
| 46 |
|
sub MISC_SPECIAL_EL () { 0b1000000000000000000000000 } |
| 47 |
|
sub FOREIGN_EL () { 0b10000000000000000000000000 } |
| 48 |
|
sub FOREIGN_FLOW_CONTENT_EL () { 0b100000000000000000000000000 } |
| 49 |
|
sub MML_AXML_EL () { 0b1000000000000000000000000000 } |
| 50 |
|
|
| 51 |
|
sub TABLE_ROWS_EL () { |
| 52 |
|
TABLE_EL | |
| 53 |
|
TABLE_ROW_EL | |
| 54 |
|
TABLE_ROW_GROUP_EL |
| 55 |
|
} |
| 56 |
|
|
| 57 |
|
sub END_TAG_OPTIONAL_EL () { |
| 58 |
|
DD_EL | |
| 59 |
|
DT_EL | |
| 60 |
|
LI_EL | |
| 61 |
|
P_EL |
| 62 |
|
} |
| 63 |
|
|
| 64 |
|
sub ALL_END_TAG_OPTIONAL_EL () { |
| 65 |
|
END_TAG_OPTIONAL_EL | |
| 66 |
|
BODY_EL | |
| 67 |
|
HTML_EL | |
| 68 |
|
TABLE_CELL_EL | |
| 69 |
|
TABLE_ROW_EL | |
| 70 |
|
TABLE_ROW_GROUP_EL |
| 71 |
|
} |
| 72 |
|
|
| 73 |
|
sub SCOPING_EL () { |
| 74 |
|
BUTTON_EL | |
| 75 |
|
CAPTION_EL | |
| 76 |
|
HTML_EL | |
| 77 |
|
TABLE_EL | |
| 78 |
|
TABLE_CELL_EL | |
| 79 |
|
MISC_SCOPING_EL |
| 80 |
|
} |
| 81 |
|
|
| 82 |
|
sub TABLE_SCOPING_EL () { |
| 83 |
|
HTML_EL | |
| 84 |
|
TABLE_EL |
| 85 |
|
} |
| 86 |
|
|
| 87 |
|
sub TABLE_ROWS_SCOPING_EL () { |
| 88 |
|
HTML_EL | |
| 89 |
|
TABLE_ROW_GROUP_EL |
| 90 |
|
} |
| 91 |
|
|
| 92 |
|
sub TABLE_ROW_SCOPING_EL () { |
| 93 |
|
HTML_EL | |
| 94 |
|
TABLE_ROW_EL |
| 95 |
|
} |
| 96 |
|
|
| 97 |
|
sub SPECIAL_EL () { |
| 98 |
|
ADDRESS_EL | |
| 99 |
|
BODY_EL | |
| 100 |
|
DIV_EL | |
| 101 |
|
END_TAG_OPTIONAL_EL | |
| 102 |
|
FORM_EL | |
| 103 |
|
FRAMESET_EL | |
| 104 |
|
HEADING_EL | |
| 105 |
|
OPTION_EL | |
| 106 |
|
OPTGROUP_EL | |
| 107 |
|
SELECT_EL | |
| 108 |
|
TABLE_ROW_EL | |
| 109 |
|
TABLE_ROW_GROUP_EL | |
| 110 |
|
MISC_SPECIAL_EL |
| 111 |
|
} |
| 112 |
|
|
| 113 |
|
my $el_category = { |
| 114 |
|
a => A_EL | FORMATTING_EL, |
| 115 |
|
address => ADDRESS_EL, |
| 116 |
|
applet => MISC_SCOPING_EL, |
| 117 |
|
area => MISC_SPECIAL_EL, |
| 118 |
|
b => FORMATTING_EL, |
| 119 |
|
base => MISC_SPECIAL_EL, |
| 120 |
|
basefont => MISC_SPECIAL_EL, |
| 121 |
|
bgsound => MISC_SPECIAL_EL, |
| 122 |
|
big => FORMATTING_EL, |
| 123 |
|
blockquote => MISC_SPECIAL_EL, |
| 124 |
|
body => BODY_EL, |
| 125 |
|
br => MISC_SPECIAL_EL, |
| 126 |
|
button => BUTTON_EL, |
| 127 |
|
caption => CAPTION_EL, |
| 128 |
|
center => MISC_SPECIAL_EL, |
| 129 |
|
col => MISC_SPECIAL_EL, |
| 130 |
|
colgroup => MISC_SPECIAL_EL, |
| 131 |
|
dd => DD_EL, |
| 132 |
|
dir => MISC_SPECIAL_EL, |
| 133 |
|
div => DIV_EL, |
| 134 |
|
dl => MISC_SPECIAL_EL, |
| 135 |
|
dt => DT_EL, |
| 136 |
|
em => FORMATTING_EL, |
| 137 |
|
embed => MISC_SPECIAL_EL, |
| 138 |
|
fieldset => MISC_SPECIAL_EL, |
| 139 |
|
font => FORMATTING_EL, |
| 140 |
|
form => FORM_EL, |
| 141 |
|
frame => MISC_SPECIAL_EL, |
| 142 |
|
frameset => FRAMESET_EL, |
| 143 |
|
h1 => HEADING_EL, |
| 144 |
|
h2 => HEADING_EL, |
| 145 |
|
h3 => HEADING_EL, |
| 146 |
|
h4 => HEADING_EL, |
| 147 |
|
h5 => HEADING_EL, |
| 148 |
|
h6 => HEADING_EL, |
| 149 |
|
head => MISC_SPECIAL_EL, |
| 150 |
|
hr => MISC_SPECIAL_EL, |
| 151 |
|
html => HTML_EL, |
| 152 |
|
i => FORMATTING_EL, |
| 153 |
|
iframe => MISC_SPECIAL_EL, |
| 154 |
|
img => MISC_SPECIAL_EL, |
| 155 |
|
input => MISC_SPECIAL_EL, |
| 156 |
|
isindex => MISC_SPECIAL_EL, |
| 157 |
|
li => LI_EL, |
| 158 |
|
link => MISC_SPECIAL_EL, |
| 159 |
|
listing => MISC_SPECIAL_EL, |
| 160 |
|
marquee => MISC_SCOPING_EL, |
| 161 |
|
menu => MISC_SPECIAL_EL, |
| 162 |
|
meta => MISC_SPECIAL_EL, |
| 163 |
|
nobr => NOBR_EL | FORMATTING_EL, |
| 164 |
|
noembed => MISC_SPECIAL_EL, |
| 165 |
|
noframes => MISC_SPECIAL_EL, |
| 166 |
|
noscript => MISC_SPECIAL_EL, |
| 167 |
|
object => MISC_SCOPING_EL, |
| 168 |
|
ol => MISC_SPECIAL_EL, |
| 169 |
|
optgroup => OPTGROUP_EL, |
| 170 |
|
option => OPTION_EL, |
| 171 |
|
p => P_EL, |
| 172 |
|
param => MISC_SPECIAL_EL, |
| 173 |
|
plaintext => MISC_SPECIAL_EL, |
| 174 |
|
pre => MISC_SPECIAL_EL, |
| 175 |
|
s => FORMATTING_EL, |
| 176 |
|
script => MISC_SPECIAL_EL, |
| 177 |
|
select => SELECT_EL, |
| 178 |
|
small => FORMATTING_EL, |
| 179 |
|
spacer => MISC_SPECIAL_EL, |
| 180 |
|
strike => FORMATTING_EL, |
| 181 |
|
strong => FORMATTING_EL, |
| 182 |
|
style => MISC_SPECIAL_EL, |
| 183 |
|
table => TABLE_EL, |
| 184 |
|
tbody => TABLE_ROW_GROUP_EL, |
| 185 |
|
td => TABLE_CELL_EL, |
| 186 |
|
textarea => MISC_SPECIAL_EL, |
| 187 |
|
tfoot => TABLE_ROW_GROUP_EL, |
| 188 |
|
th => TABLE_CELL_EL, |
| 189 |
|
thead => TABLE_ROW_GROUP_EL, |
| 190 |
|
title => MISC_SPECIAL_EL, |
| 191 |
|
tr => TABLE_ROW_EL, |
| 192 |
|
tt => FORMATTING_EL, |
| 193 |
|
u => FORMATTING_EL, |
| 194 |
|
ul => MISC_SPECIAL_EL, |
| 195 |
|
wbr => MISC_SPECIAL_EL, |
| 196 |
|
}; |
| 197 |
|
|
| 198 |
|
my $el_category_f = { |
| 199 |
|
$MML_NS => { |
| 200 |
|
'annotation-xml' => MML_AXML_EL, |
| 201 |
|
mi => FOREIGN_FLOW_CONTENT_EL, |
| 202 |
|
mo => FOREIGN_FLOW_CONTENT_EL, |
| 203 |
|
mn => FOREIGN_FLOW_CONTENT_EL, |
| 204 |
|
ms => FOREIGN_FLOW_CONTENT_EL, |
| 205 |
|
mtext => FOREIGN_FLOW_CONTENT_EL, |
| 206 |
|
}, |
| 207 |
|
$SVG_NS => { |
| 208 |
|
foreignobject => FOREIGN_FLOW_CONTENT_EL, ## TODO: case |
| 209 |
|
desc => FOREIGN_FLOW_CONTENT_EL, |
| 210 |
|
title => FOREIGN_FLOW_CONTENT_EL, |
| 211 |
|
}, |
| 212 |
|
## NOTE: In addition, FOREIGN_EL is set to non-HTML elements. |
| 213 |
}; |
}; |
| 214 |
|
|
| 215 |
my $c1_entity_char = { |
my $c1_entity_char = { |
| 247 |
0x9F => 0x0178, |
0x9F => 0x0178, |
| 248 |
}; # $c1_entity_char |
}; # $c1_entity_char |
| 249 |
|
|
|
my $special_category = { |
|
|
address => 1, area => 1, base => 1, basefont => 1, bgsound => 1, |
|
|
blockquote => 1, body => 1, br => 1, center => 1, col => 1, colgroup => 1, |
|
|
dd => 1, dir => 1, div => 1, dl => 1, dt => 1, embed => 1, fieldset => 1, |
|
|
form => 1, frame => 1, frameset => 1, h1 => 1, h2 => 1, h3 => 1, |
|
|
h4 => 1, h5 => 1, h6 => 1, head => 1, hr => 1, iframe => 1, image => 1, |
|
|
img => 1, input => 1, isindex => 1, li => 1, link => 1, listing => 1, |
|
|
menu => 1, meta => 1, noembed => 1, noframes => 1, noscript => 1, |
|
|
ol => 1, optgroup => 1, option => 1, p => 1, param => 1, plaintext => 1, |
|
|
pre => 1, script => 1, select => 1, spacer => 1, style => 1, tbody => 1, |
|
|
textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1, |
|
|
}; |
|
|
my $scoping_category = { |
|
|
button => 1, caption => 1, html => 1, marquee => 1, object => 1, |
|
|
table => 1, td => 1, th => 1, |
|
|
}; |
|
|
my $formatting_category = { |
|
|
a => 1, b => 1, big => 1, em => 1, font => 1, i => 1, nobr => 1, |
|
|
s => 1, small => 1, strile => 1, strong => 1, tt => 1, u => 1, |
|
|
}; |
|
|
# $phrasing_category: all other elements |
|
|
|
|
| 250 |
sub parse_byte_string ($$$$;$) { |
sub parse_byte_string ($$$$;$) { |
| 251 |
my $self = ref $_[0] ? shift : shift->new; |
my $self = ref $_[0] ? shift : shift->new; |
| 252 |
my $charset = shift; |
my $charset = shift; |
| 272 |
$self->{change_encoding} = sub { |
$self->{change_encoding} = sub { |
| 273 |
my $self = shift; |
my $self = shift; |
| 274 |
my $charset = lc shift; |
my $charset = lc shift; |
| 275 |
|
my $token = shift; |
| 276 |
## TODO: if $charset is supported |
## TODO: if $charset is supported |
| 277 |
## TODO: normalize charset name |
## TODO: normalize charset name |
| 278 |
|
|
| 291 |
} |
} |
| 292 |
|
|
| 293 |
!!!parse-error (type => 'charset label detected:'.$self->{input_encoding}. |
!!!parse-error (type => 'charset label detected:'.$self->{input_encoding}. |
| 294 |
':'.$charset, level => 'w'); |
':'.$charset, level => 'w', token => $token); |
| 295 |
|
|
| 296 |
## Step 3 |
## Step 3 |
| 297 |
# if (can) { |
# if (can) { |
| 342 |
if defined $self->{input_encoding}; |
if defined $self->{input_encoding}; |
| 343 |
|
|
| 344 |
my $i = 0; |
my $i = 0; |
| 345 |
my $line = 1; |
$self->{line_prev} = $self->{line} = 1; |
| 346 |
my $column = 0; |
$self->{column_prev} = $self->{column} = 0; |
| 347 |
$self->{set_next_char} = sub { |
$self->{set_next_char} = sub { |
| 348 |
my $self = shift; |
my $self = shift; |
| 349 |
|
|
| 352 |
|
|
| 353 |
$self->{next_char} = -1 and return if $i >= length $$s; |
$self->{next_char} = -1 and return if $i >= length $$s; |
| 354 |
$self->{next_char} = ord substr $$s, $i++, 1; |
$self->{next_char} = ord substr $$s, $i++, 1; |
| 355 |
$column++; |
|
| 356 |
|
($self->{line_prev}, $self->{column_prev}) |
| 357 |
|
= ($self->{line}, $self->{column}); |
| 358 |
|
$self->{column}++; |
| 359 |
|
|
| 360 |
if ($self->{next_char} == 0x000A) { # LF |
if ($self->{next_char} == 0x000A) { # LF |
| 361 |
$line++; |
$self->{line}++; |
| 362 |
$column = 0; |
$self->{column} = 0; |
| 363 |
} elsif ($self->{next_char} == 0x000D) { # CR |
} elsif ($self->{next_char} == 0x000D) { # CR |
| 364 |
$i++ if substr ($$s, $i, 1) eq "\x0A"; |
$i++ if substr ($$s, $i, 1) eq "\x0A"; |
| 365 |
$self->{next_char} = 0x000A; # LF # MUST |
$self->{next_char} = 0x000A; # LF # MUST |
| 366 |
$line++; |
$self->{line}++; |
| 367 |
$column = 0; |
$self->{column} = 0; |
| 368 |
} elsif ($self->{next_char} > 0x10FFFF) { |
} elsif ($self->{next_char} > 0x10FFFF) { |
| 369 |
$self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
$self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
| 370 |
} elsif ($self->{next_char} == 0x0000) { # NULL |
} elsif ($self->{next_char} == 0x0000) { # NULL |
| 377 |
|
|
| 378 |
my $onerror = $_[2] || sub { |
my $onerror = $_[2] || sub { |
| 379 |
my (%opt) = @_; |
my (%opt) = @_; |
| 380 |
warn "Parse error ($opt{type}) at line $opt{line} column $opt{column}\n"; |
my $line = $opt{token} ? $opt{token}->{line} : $opt{line}; |
| 381 |
|
my $column = $opt{token} ? $opt{token}->{column} : $opt{column}; |
| 382 |
|
warn "Parse error ($opt{type}) at line $line column $column\n"; |
| 383 |
}; |
}; |
| 384 |
$self->{parse_error} = sub { |
$self->{parse_error} = sub { |
| 385 |
$onerror->(@_, line => $line, column => $column); |
$onerror->(line => $self->{line}, column => $self->{column}, @_); |
| 386 |
}; |
}; |
| 387 |
|
|
| 388 |
$self->_initialize_tokenizer; |
$self->_initialize_tokenizer; |
| 390 |
$self->_construct_tree; |
$self->_construct_tree; |
| 391 |
$self->_terminate_tree_constructor; |
$self->_terminate_tree_constructor; |
| 392 |
|
|
| 393 |
|
delete $self->{parse_error}; # remove loop |
| 394 |
|
|
| 395 |
return $self->{document}; |
return $self->{document}; |
| 396 |
} # parse_string |
} # parse_string |
| 397 |
|
|
| 459 |
sub AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE () { 31 } |
sub AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE () { 31 } |
| 460 |
sub BOGUS_DOCTYPE_STATE () { 32 } |
sub BOGUS_DOCTYPE_STATE () { 32 } |
| 461 |
sub AFTER_ATTRIBUTE_VALUE_QUOTED_STATE () { 33 } |
sub AFTER_ATTRIBUTE_VALUE_QUOTED_STATE () { 33 } |
| 462 |
|
sub SELF_CLOSING_START_TAG_STATE () { 34 } |
| 463 |
|
sub CDATA_BLOCK_STATE () { 35 } |
| 464 |
|
|
| 465 |
sub DOCTYPE_TOKEN () { 1 } |
sub DOCTYPE_TOKEN () { 1 } |
| 466 |
sub COMMENT_TOKEN () { 2 } |
sub COMMENT_TOKEN () { 2 } |
| 477 |
sub ROW_IMS () { 0b10000000 } |
sub ROW_IMS () { 0b10000000 } |
| 478 |
sub BODY_AFTER_IMS () { 0b100000000 } |
sub BODY_AFTER_IMS () { 0b100000000 } |
| 479 |
sub FRAME_IMS () { 0b1000000000 } |
sub FRAME_IMS () { 0b1000000000 } |
| 480 |
|
sub SELECT_IMS () { 0b10000000000 } |
| 481 |
|
sub IN_FOREIGN_CONTENT_IM () { 0b100000000000 } |
| 482 |
|
## NOTE: "in foreign content" insertion mode is special; it is combined |
| 483 |
|
## with the secondary insertion mode. In this parser, they are stored |
| 484 |
|
## together in the bit-or'ed form. |
| 485 |
|
|
| 486 |
## NOTE: "initial" and "before html" insertion modes have no constants. |
## NOTE: "initial" and "before html" insertion modes have no constants. |
| 487 |
|
|
| 504 |
sub AFTER_BODY_IM () { BODY_AFTER_IMS } |
sub AFTER_BODY_IM () { BODY_AFTER_IMS } |
| 505 |
sub IN_FRAMESET_IM () { FRAME_IMS | 0b01 } |
sub IN_FRAMESET_IM () { FRAME_IMS | 0b01 } |
| 506 |
sub AFTER_FRAMESET_IM () { FRAME_IMS | 0b10 } |
sub AFTER_FRAMESET_IM () { FRAME_IMS | 0b10 } |
| 507 |
sub IN_SELECT_IM () { 0b01 } |
sub IN_SELECT_IM () { SELECT_IMS | 0b01 } |
| 508 |
|
sub IN_SELECT_IN_TABLE_IM () { SELECT_IMS | 0b10 } |
| 509 |
sub IN_COLUMN_GROUP_IM () { 0b10 } |
sub IN_COLUMN_GROUP_IM () { 0b10 } |
| 510 |
|
|
| 511 |
## Implementations MUST act as if state machine in the spec |
## Implementations MUST act as if state machine in the spec |
| 518 |
undef $self->{current_attribute}; |
undef $self->{current_attribute}; |
| 519 |
undef $self->{last_emitted_start_tag_name}; |
undef $self->{last_emitted_start_tag_name}; |
| 520 |
undef $self->{last_attribute_value_state}; |
undef $self->{last_attribute_value_state}; |
| 521 |
|
delete $self->{self_closing}; |
| 522 |
$self->{char} = []; |
$self->{char} = []; |
| 523 |
# $self->{next_char} |
# $self->{next_char} |
| 524 |
!!!next-input-character; |
!!!next-input-character; |
| 539 |
## ->{value} |
## ->{value} |
| 540 |
## ->{has_reference} == 1 or 0 |
## ->{has_reference} == 1 or 0 |
| 541 |
## ->{data} (COMMENT_TOKEN, CHARACTER_TOKEN) |
## ->{data} (COMMENT_TOKEN, CHARACTER_TOKEN) |
| 542 |
|
## NOTE: The "self-closing flag" is hold as |$self->{self_closing}|. |
| 543 |
|
## |->{self_closing}| is used to save the value of |$self->{self_closing}| |
| 544 |
|
## while the token is pushed back to the stack. |
| 545 |
|
|
| 546 |
|
## ISSUE: "When a DOCTYPE token is created, its |
| 547 |
|
## <i>self-closing flag</i> must be unset (its other state is that it |
| 548 |
|
## be set), and its attributes list must be empty.": Wrong subject? |
| 549 |
|
|
| 550 |
## Emitted token MUST immediately be handled by the tree construction state. |
## Emitted token MUST immediately be handled by the tree construction state. |
| 551 |
|
|
| 572 |
|
|
| 573 |
sub _get_next_token ($) { |
sub _get_next_token ($) { |
| 574 |
my $self = shift; |
my $self = shift; |
| 575 |
|
|
| 576 |
|
if ($self->{self_closing}) { |
| 577 |
|
!!!parse-error (type => 'nestc', token => $self->{current_token}); |
| 578 |
|
## NOTE: The |self_closing| flag is only set by start tag token. |
| 579 |
|
## In addition, when a start tag token is emitted, it is always set to |
| 580 |
|
## |current_token|. |
| 581 |
|
delete $self->{self_closing}; |
| 582 |
|
} |
| 583 |
|
|
| 584 |
if (@{$self->{token}}) { |
if (@{$self->{token}}) { |
| 585 |
|
$self->{self_closing} = $self->{token}->[0]->{self_closing}; |
| 586 |
return shift @{$self->{token}}; |
return shift @{$self->{token}}; |
| 587 |
} |
} |
| 588 |
|
|
| 645 |
# |
# |
| 646 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 647 |
!!!cp (11); |
!!!cp (11); |
| 648 |
!!!emit ({type => END_OF_FILE_TOKEN}); |
!!!emit ({type => END_OF_FILE_TOKEN, |
| 649 |
|
line => $self->{line}, column => $self->{column}}); |
| 650 |
last A; ## TODO: ok? |
last A; ## TODO: ok? |
| 651 |
} else { |
} else { |
| 652 |
!!!cp (12); |
!!!cp (12); |
| 653 |
} |
} |
| 654 |
# Anything else |
# Anything else |
| 655 |
my $token = {type => CHARACTER_TOKEN, |
my $token = {type => CHARACTER_TOKEN, |
| 656 |
data => chr $self->{next_char}}; |
data => chr $self->{next_char}, |
| 657 |
|
line => $self->{line}, column => $self->{column}, |
| 658 |
|
}; |
| 659 |
## Stay in the data state |
## Stay in the data state |
| 660 |
!!!next-input-character; |
!!!next-input-character; |
| 661 |
|
|
| 664 |
redo A; |
redo A; |
| 665 |
} elsif ($self->{state} == ENTITY_DATA_STATE) { |
} elsif ($self->{state} == ENTITY_DATA_STATE) { |
| 666 |
## (cannot happen in CDATA state) |
## (cannot happen in CDATA state) |
| 667 |
|
|
| 668 |
|
my ($l, $c) = ($self->{line_prev}, $self->{column_prev}); |
| 669 |
|
|
| 670 |
my $token = $self->_tokenize_attempt_to_consume_an_entity (0, -1); |
my $token = $self->_tokenize_attempt_to_consume_an_entity (0, -1); |
| 671 |
|
|
| 674 |
|
|
| 675 |
unless (defined $token) { |
unless (defined $token) { |
| 676 |
!!!cp (13); |
!!!cp (13); |
| 677 |
!!!emit ({type => CHARACTER_TOKEN, data => '&'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '&', |
| 678 |
|
line => $l, column => $c, |
| 679 |
|
}); |
| 680 |
} else { |
} else { |
| 681 |
!!!cp (14); |
!!!cp (14); |
| 682 |
!!!emit ($token); |
!!!emit ($token); |
| 695 |
## reconsume |
## reconsume |
| 696 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 697 |
|
|
| 698 |
!!!emit ({type => CHARACTER_TOKEN, data => '<'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '<', |
| 699 |
|
line => $self->{line_prev}, |
| 700 |
|
column => $self->{column_prev}, |
| 701 |
|
}); |
| 702 |
|
|
| 703 |
redo A; |
redo A; |
| 704 |
} |
} |
| 718 |
!!!cp (19); |
!!!cp (19); |
| 719 |
$self->{current_token} |
$self->{current_token} |
| 720 |
= {type => START_TAG_TOKEN, |
= {type => START_TAG_TOKEN, |
| 721 |
tag_name => chr ($self->{next_char} + 0x0020)}; |
tag_name => chr ($self->{next_char} + 0x0020), |
| 722 |
|
line => $self->{line_prev}, |
| 723 |
|
column => $self->{column_prev}}; |
| 724 |
$self->{state} = TAG_NAME_STATE; |
$self->{state} = TAG_NAME_STATE; |
| 725 |
!!!next-input-character; |
!!!next-input-character; |
| 726 |
redo A; |
redo A; |
| 728 |
$self->{next_char} <= 0x007A) { # a..z |
$self->{next_char} <= 0x007A) { # a..z |
| 729 |
!!!cp (20); |
!!!cp (20); |
| 730 |
$self->{current_token} = {type => START_TAG_TOKEN, |
$self->{current_token} = {type => START_TAG_TOKEN, |
| 731 |
tag_name => chr ($self->{next_char})}; |
tag_name => chr ($self->{next_char}), |
| 732 |
|
line => $self->{line_prev}, |
| 733 |
|
column => $self->{column_prev}}; |
| 734 |
$self->{state} = TAG_NAME_STATE; |
$self->{state} = TAG_NAME_STATE; |
| 735 |
!!!next-input-character; |
!!!next-input-character; |
| 736 |
redo A; |
redo A; |
| 737 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 738 |
!!!cp (21); |
!!!cp (21); |
| 739 |
!!!parse-error (type => 'empty start tag'); |
!!!parse-error (type => 'empty start tag', |
| 740 |
|
line => $self->{line_prev}, |
| 741 |
|
column => $self->{column_prev}); |
| 742 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 743 |
!!!next-input-character; |
!!!next-input-character; |
| 744 |
|
|
| 745 |
!!!emit ({type => CHARACTER_TOKEN, data => '<>'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '<>', |
| 746 |
|
line => $self->{line_prev}, |
| 747 |
|
column => $self->{column_prev}, |
| 748 |
|
}); |
| 749 |
|
|
| 750 |
redo A; |
redo A; |
| 751 |
} elsif ($self->{next_char} == 0x003F) { # ? |
} elsif ($self->{next_char} == 0x003F) { # ? |
| 752 |
!!!cp (22); |
!!!cp (22); |
| 753 |
!!!parse-error (type => 'pio'); |
!!!parse-error (type => 'pio', |
| 754 |
|
line => $self->{line_prev}, |
| 755 |
|
column => $self->{column_prev}); |
| 756 |
$self->{state} = BOGUS_COMMENT_STATE; |
$self->{state} = BOGUS_COMMENT_STATE; |
| 757 |
|
$self->{current_token} = {type => COMMENT_TOKEN, data => '', |
| 758 |
|
line => $self->{line_prev}, |
| 759 |
|
column => $self->{column_prev}, |
| 760 |
|
}; |
| 761 |
## $self->{next_char} is intentionally left as is |
## $self->{next_char} is intentionally left as is |
| 762 |
redo A; |
redo A; |
| 763 |
} else { |
} else { |
| 766 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 767 |
## reconsume |
## reconsume |
| 768 |
|
|
| 769 |
!!!emit ({type => CHARACTER_TOKEN, data => '<'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '<', |
| 770 |
|
line => $self->{line_prev}, |
| 771 |
|
column => $self->{column_prev}, |
| 772 |
|
}); |
| 773 |
|
|
| 774 |
redo A; |
redo A; |
| 775 |
} |
} |
| 777 |
die "$0: $self->{content_model} in tag open"; |
die "$0: $self->{content_model} in tag open"; |
| 778 |
} |
} |
| 779 |
} elsif ($self->{state} == CLOSE_TAG_OPEN_STATE) { |
} elsif ($self->{state} == CLOSE_TAG_OPEN_STATE) { |
| 780 |
|
my ($l, $c) = ($self->{line_prev}, $self->{column_prev} - 1); # "<"of"</" |
| 781 |
if ($self->{content_model} & CM_LIMITED_MARKUP) { # RCDATA | CDATA |
if ($self->{content_model} & CM_LIMITED_MARKUP) { # RCDATA | CDATA |
| 782 |
if (defined $self->{last_emitted_start_tag_name}) { |
if (defined $self->{last_emitted_start_tag_name}) { |
| 783 |
|
|
| 784 |
## NOTE: <http://krijnhoetmer.nl/irc-logs/whatwg/20070626#l-564> |
## NOTE: <http://krijnhoetmer.nl/irc-logs/whatwg/20070626#l-564> |
| 785 |
my @next_char; |
my @next_char; |
| 786 |
TAGNAME: for (my $i = 0; $i < length $self->{last_emitted_start_tag_name}; $i++) { |
TAGNAME: for (my $i = 0; $i < length $self->{last_emitted_start_tag_name}; $i++) { |
| 797 |
!!!back-next-input-character (@next_char); |
!!!back-next-input-character (@next_char); |
| 798 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 799 |
|
|
| 800 |
!!!emit ({type => CHARACTER_TOKEN, data => '</'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '</', |
| 801 |
|
line => $l, column => $c, |
| 802 |
|
}); |
| 803 |
|
|
| 804 |
redo A; |
redo A; |
| 805 |
} |
} |
| 818 |
$self->{next_char} = shift @next_char; # reconsume |
$self->{next_char} = shift @next_char; # reconsume |
| 819 |
!!!back-next-input-character (@next_char); |
!!!back-next-input-character (@next_char); |
| 820 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 821 |
!!!emit ({type => CHARACTER_TOKEN, data => '</'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '</', |
| 822 |
|
line => $l, column => $c, |
| 823 |
|
}); |
| 824 |
redo A; |
redo A; |
| 825 |
} else { |
} else { |
| 826 |
!!!cp (27); |
!!!cp (27); |
| 833 |
!!!cp (28); |
!!!cp (28); |
| 834 |
# next-input-character is already done |
# next-input-character is already done |
| 835 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 836 |
!!!emit ({type => CHARACTER_TOKEN, data => '</'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '</', |
| 837 |
|
line => $l, column => $c, |
| 838 |
|
}); |
| 839 |
redo A; |
redo A; |
| 840 |
} |
} |
| 841 |
} |
} |
| 843 |
if (0x0041 <= $self->{next_char} and |
if (0x0041 <= $self->{next_char} and |
| 844 |
$self->{next_char} <= 0x005A) { # A..Z |
$self->{next_char} <= 0x005A) { # A..Z |
| 845 |
!!!cp (29); |
!!!cp (29); |
| 846 |
$self->{current_token} = {type => END_TAG_TOKEN, |
$self->{current_token} |
| 847 |
tag_name => chr ($self->{next_char} + 0x0020)}; |
= {type => END_TAG_TOKEN, |
| 848 |
|
tag_name => chr ($self->{next_char} + 0x0020), |
| 849 |
|
line => $l, column => $c}; |
| 850 |
$self->{state} = TAG_NAME_STATE; |
$self->{state} = TAG_NAME_STATE; |
| 851 |
!!!next-input-character; |
!!!next-input-character; |
| 852 |
redo A; |
redo A; |
| 854 |
$self->{next_char} <= 0x007A) { # a..z |
$self->{next_char} <= 0x007A) { # a..z |
| 855 |
!!!cp (30); |
!!!cp (30); |
| 856 |
$self->{current_token} = {type => END_TAG_TOKEN, |
$self->{current_token} = {type => END_TAG_TOKEN, |
| 857 |
tag_name => chr ($self->{next_char})}; |
tag_name => chr ($self->{next_char}), |
| 858 |
|
line => $l, column => $c}; |
| 859 |
$self->{state} = TAG_NAME_STATE; |
$self->{state} = TAG_NAME_STATE; |
| 860 |
!!!next-input-character; |
!!!next-input-character; |
| 861 |
redo A; |
redo A; |
| 862 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 863 |
!!!cp (31); |
!!!cp (31); |
| 864 |
!!!parse-error (type => 'empty end tag'); |
!!!parse-error (type => 'empty end tag', |
| 865 |
|
line => $self->{line_prev}, ## "<" in "</>" |
| 866 |
|
column => $self->{column_prev} - 1); |
| 867 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 868 |
!!!next-input-character; |
!!!next-input-character; |
| 869 |
redo A; |
redo A; |
| 873 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 874 |
# reconsume |
# reconsume |
| 875 |
|
|
| 876 |
!!!emit ({type => CHARACTER_TOKEN, data => '</'}); |
!!!emit ({type => CHARACTER_TOKEN, data => '</', |
| 877 |
|
line => $l, column => $c, |
| 878 |
|
}); |
| 879 |
|
|
| 880 |
redo A; |
redo A; |
| 881 |
} else { |
} else { |
| 882 |
!!!cp (33); |
!!!cp (33); |
| 883 |
!!!parse-error (type => 'bogus end tag'); |
!!!parse-error (type => 'bogus end tag'); |
| 884 |
$self->{state} = BOGUS_COMMENT_STATE; |
$self->{state} = BOGUS_COMMENT_STATE; |
| 885 |
|
$self->{current_token} = {type => COMMENT_TOKEN, data => '', |
| 886 |
|
line => $self->{line_prev}, # "<" of "</" |
| 887 |
|
column => $self->{column_prev} - 1, |
| 888 |
|
}; |
| 889 |
## $self->{next_char} is intentionally left as is |
## $self->{next_char} is intentionally left as is |
| 890 |
redo A; |
redo A; |
| 891 |
} |
} |
| 902 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 903 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 904 |
!!!cp (35); |
!!!cp (35); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 905 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 906 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 907 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 933 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 934 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 935 |
!!!cp (39); |
!!!cp (39); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 936 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 937 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 938 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 953 |
|
|
| 954 |
redo A; |
redo A; |
| 955 |
} elsif ($self->{next_char} == 0x002F) { # / |
} elsif ($self->{next_char} == 0x002F) { # / |
| 956 |
|
!!!cp (42); |
| 957 |
|
$self->{state} = SELF_CLOSING_START_TAG_STATE; |
| 958 |
!!!next-input-character; |
!!!next-input-character; |
|
if ($self->{next_char} == 0x003E and # > |
|
|
$self->{current_token}->{type} == START_TAG_TOKEN and |
|
|
$permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) { |
|
|
# permitted slash |
|
|
!!!cp (42); |
|
|
# |
|
|
} else { |
|
|
!!!cp (43); |
|
|
!!!parse-error (type => 'nestc'); |
|
|
} |
|
|
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
|
|
# next-input-character is already done |
|
| 959 |
redo A; |
redo A; |
| 960 |
} else { |
} else { |
| 961 |
!!!cp (44); |
!!!cp (44); |
| 978 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 979 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 980 |
!!!cp (46); |
!!!cp (46); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 981 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 982 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 983 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 999 |
} elsif (0x0041 <= $self->{next_char} and |
} elsif (0x0041 <= $self->{next_char} and |
| 1000 |
$self->{next_char} <= 0x005A) { # A..Z |
$self->{next_char} <= 0x005A) { # A..Z |
| 1001 |
!!!cp (49); |
!!!cp (49); |
| 1002 |
$self->{current_attribute} = {name => chr ($self->{next_char} + 0x0020), |
$self->{current_attribute} |
| 1003 |
value => ''}; |
= {name => chr ($self->{next_char} + 0x0020), |
| 1004 |
|
value => '', |
| 1005 |
|
line => $self->{line}, column => $self->{column}}; |
| 1006 |
$self->{state} = ATTRIBUTE_NAME_STATE; |
$self->{state} = ATTRIBUTE_NAME_STATE; |
| 1007 |
!!!next-input-character; |
!!!next-input-character; |
| 1008 |
redo A; |
redo A; |
| 1009 |
} elsif ($self->{next_char} == 0x002F) { # / |
} elsif ($self->{next_char} == 0x002F) { # / |
| 1010 |
|
!!!cp (50); |
| 1011 |
|
$self->{state} = SELF_CLOSING_START_TAG_STATE; |
| 1012 |
!!!next-input-character; |
!!!next-input-character; |
|
if ($self->{next_char} == 0x003E and # > |
|
|
$self->{current_token}->{type} == START_TAG_TOKEN and |
|
|
$permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) { |
|
|
# permitted slash |
|
|
!!!cp (50); |
|
|
# |
|
|
} else { |
|
|
!!!cp (51); |
|
|
!!!parse-error (type => 'nestc'); |
|
|
} |
|
|
## Stay in the state |
|
|
# next-input-character is already done |
|
| 1013 |
redo A; |
redo A; |
| 1014 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 1015 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 1016 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1017 |
!!!cp (52); |
!!!cp (52); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1018 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1019 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1020 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1044 |
} else { |
} else { |
| 1045 |
!!!cp (56); |
!!!cp (56); |
| 1046 |
} |
} |
| 1047 |
$self->{current_attribute} = {name => chr ($self->{next_char}), |
$self->{current_attribute} |
| 1048 |
value => ''}; |
= {name => chr ($self->{next_char}), |
| 1049 |
|
value => '', |
| 1050 |
|
line => $self->{line}, column => $self->{column}}; |
| 1051 |
$self->{state} = ATTRIBUTE_NAME_STATE; |
$self->{state} = ATTRIBUTE_NAME_STATE; |
| 1052 |
!!!next-input-character; |
!!!next-input-character; |
| 1053 |
redo A; |
redo A; |
| 1057 |
if (exists $self->{current_token}->{attributes} # start tag or end tag |
if (exists $self->{current_token}->{attributes} # start tag or end tag |
| 1058 |
->{$self->{current_attribute}->{name}}) { # MUST |
->{$self->{current_attribute}->{name}}) { # MUST |
| 1059 |
!!!cp (57); |
!!!cp (57); |
| 1060 |
!!!parse-error (type => 'duplicate attribute:'.$self->{current_attribute}->{name}); |
!!!parse-error (type => 'duplicate attribute:'.$self->{current_attribute}->{name}, line => $self->{current_attribute}->{line}, column => $self->{current_attribute}->{column}); |
| 1061 |
## Discard $self->{current_attribute} # MUST |
## Discard $self->{current_attribute} # MUST |
| 1062 |
} else { |
} else { |
| 1063 |
!!!cp (58); |
!!!cp (58); |
| 1086 |
$before_leave->(); |
$before_leave->(); |
| 1087 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1088 |
!!!cp (61); |
!!!cp (61); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1089 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1090 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1091 |
!!!cp (62); |
!!!cp (62); |
| 1110 |
!!!next-input-character; |
!!!next-input-character; |
| 1111 |
redo A; |
redo A; |
| 1112 |
} elsif ($self->{next_char} == 0x002F) { # / |
} elsif ($self->{next_char} == 0x002F) { # / |
| 1113 |
|
!!!cp (64); |
| 1114 |
$before_leave->(); |
$before_leave->(); |
| 1115 |
|
$self->{state} = SELF_CLOSING_START_TAG_STATE; |
| 1116 |
!!!next-input-character; |
!!!next-input-character; |
|
if ($self->{next_char} == 0x003E and # > |
|
|
$self->{current_token}->{type} == START_TAG_TOKEN and |
|
|
$permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) { |
|
|
# permitted slash |
|
|
!!!cp (64); |
|
|
# |
|
|
} else { |
|
|
!!!cp (65); |
|
|
!!!parse-error (type => 'nestc'); |
|
|
} |
|
|
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
|
|
# next-input-character is already done |
|
| 1117 |
redo A; |
redo A; |
| 1118 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 1119 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 1120 |
$before_leave->(); |
$before_leave->(); |
| 1121 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1122 |
!!!cp (66); |
!!!cp (66); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1123 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1124 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1125 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1170 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 1171 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1172 |
!!!cp (73); |
!!!cp (73); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1173 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1174 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1175 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1192 |
} elsif (0x0041 <= $self->{next_char} and |
} elsif (0x0041 <= $self->{next_char} and |
| 1193 |
$self->{next_char} <= 0x005A) { # A..Z |
$self->{next_char} <= 0x005A) { # A..Z |
| 1194 |
!!!cp (76); |
!!!cp (76); |
| 1195 |
$self->{current_attribute} = {name => chr ($self->{next_char} + 0x0020), |
$self->{current_attribute} |
| 1196 |
value => ''}; |
= {name => chr ($self->{next_char} + 0x0020), |
| 1197 |
|
value => '', |
| 1198 |
|
line => $self->{line}, column => $self->{column}}; |
| 1199 |
$self->{state} = ATTRIBUTE_NAME_STATE; |
$self->{state} = ATTRIBUTE_NAME_STATE; |
| 1200 |
!!!next-input-character; |
!!!next-input-character; |
| 1201 |
redo A; |
redo A; |
| 1202 |
} elsif ($self->{next_char} == 0x002F) { # / |
} elsif ($self->{next_char} == 0x002F) { # / |
| 1203 |
|
!!!cp (77); |
| 1204 |
|
$self->{state} = SELF_CLOSING_START_TAG_STATE; |
| 1205 |
!!!next-input-character; |
!!!next-input-character; |
|
if ($self->{next_char} == 0x003E and # > |
|
|
$self->{current_token}->{type} == START_TAG_TOKEN and |
|
|
$permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) { |
|
|
# permitted slash |
|
|
!!!cp (77); |
|
|
# |
|
|
} else { |
|
|
!!!cp (78); |
|
|
!!!parse-error (type => 'nestc'); |
|
|
## TODO: Different error type for <aa / bb> than <aa/> |
|
|
} |
|
|
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
|
|
# next-input-character is already done |
|
| 1206 |
redo A; |
redo A; |
| 1207 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 1208 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 1209 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1210 |
!!!cp (79); |
!!!cp (79); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1211 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1212 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1213 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1229 |
redo A; |
redo A; |
| 1230 |
} else { |
} else { |
| 1231 |
!!!cp (82); |
!!!cp (82); |
| 1232 |
$self->{current_attribute} = {name => chr ($self->{next_char}), |
$self->{current_attribute} |
| 1233 |
value => ''}; |
= {name => chr ($self->{next_char}), |
| 1234 |
|
value => '', |
| 1235 |
|
line => $self->{line}, column => $self->{column}}; |
| 1236 |
$self->{state} = ATTRIBUTE_NAME_STATE; |
$self->{state} = ATTRIBUTE_NAME_STATE; |
| 1237 |
!!!next-input-character; |
!!!next-input-character; |
| 1238 |
redo A; |
redo A; |
| 1265 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 1266 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1267 |
!!!cp (87); |
!!!cp (87); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1268 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1269 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1270 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1288 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 1289 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1290 |
!!!cp (90); |
!!!cp (90); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1291 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1292 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1293 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1335 |
!!!parse-error (type => 'unclosed attribute value'); |
!!!parse-error (type => 'unclosed attribute value'); |
| 1336 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1337 |
!!!cp (97); |
!!!cp (97); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1338 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1339 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1340 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1377 |
!!!parse-error (type => 'unclosed attribute value'); |
!!!parse-error (type => 'unclosed attribute value'); |
| 1378 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1379 |
!!!cp (103); |
!!!cp (103); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1380 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1381 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1382 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1422 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 1423 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1424 |
!!!cp (109); |
!!!cp (109); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1425 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1426 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1427 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1445 |
!!!parse-error (type => 'unclosed tag'); |
!!!parse-error (type => 'unclosed tag'); |
| 1446 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1447 |
!!!cp (112); |
!!!cp (112); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1448 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1449 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1450 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1515 |
} elsif ($self->{next_char} == 0x003E) { # > |
} elsif ($self->{next_char} == 0x003E) { # > |
| 1516 |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
if ($self->{current_token}->{type} == START_TAG_TOKEN) { |
| 1517 |
!!!cp (119); |
!!!cp (119); |
|
$self->{current_token}->{first_start_tag} |
|
|
= not defined $self->{last_emitted_start_tag_name}; |
|
| 1518 |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
$self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name}; |
| 1519 |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
} elsif ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1520 |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1535 |
|
|
| 1536 |
redo A; |
redo A; |
| 1537 |
} elsif ($self->{next_char} == 0x002F) { # / |
} elsif ($self->{next_char} == 0x002F) { # / |
| 1538 |
|
!!!cp (122); |
| 1539 |
|
$self->{state} = SELF_CLOSING_START_TAG_STATE; |
| 1540 |
!!!next-input-character; |
!!!next-input-character; |
|
if ($self->{next_char} == 0x003E and # > |
|
|
$self->{current_token}->{type} == START_TAG_TOKEN and |
|
|
$permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) { |
|
|
# permitted slash |
|
|
!!!cp (122); |
|
|
# |
|
|
} else { |
|
|
!!!cp (123); |
|
|
!!!parse-error (type => 'nestc'); |
|
|
} |
|
|
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
|
|
# next-input-character is already done |
|
| 1541 |
redo A; |
redo A; |
| 1542 |
} else { |
} else { |
| 1543 |
!!!cp (124); |
!!!cp ('124.1'); |
| 1544 |
!!!parse-error (type => 'no space between attributes'); |
!!!parse-error (type => 'no space between attributes'); |
| 1545 |
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
| 1546 |
## reconsume |
## reconsume |
| 1547 |
redo A; |
redo A; |
| 1548 |
} |
} |
| 1549 |
|
} elsif ($self->{state} == SELF_CLOSING_START_TAG_STATE) { |
| 1550 |
|
if ($self->{next_char} == 0x003E) { # > |
| 1551 |
|
if ($self->{current_token}->{type} == END_TAG_TOKEN) { |
| 1552 |
|
!!!cp ('124.2'); |
| 1553 |
|
!!!parse-error (type => 'nestc', token => $self->{current_token}); |
| 1554 |
|
## TODO: Different type than slash in start tag |
| 1555 |
|
$self->{content_model} = PCDATA_CONTENT_MODEL; # MUST |
| 1556 |
|
if ($self->{current_token}->{attributes}) { |
| 1557 |
|
!!!cp ('124.4'); |
| 1558 |
|
!!!parse-error (type => 'end tag attribute'); |
| 1559 |
|
} else { |
| 1560 |
|
!!!cp ('124.5'); |
| 1561 |
|
} |
| 1562 |
|
## TODO: Test |<title></title/>| |
| 1563 |
|
} else { |
| 1564 |
|
!!!cp ('124.3'); |
| 1565 |
|
$self->{self_closing} = 1; |
| 1566 |
|
} |
| 1567 |
|
|
| 1568 |
|
$self->{state} = DATA_STATE; |
| 1569 |
|
!!!next-input-character; |
| 1570 |
|
|
| 1571 |
|
!!!emit ($self->{current_token}); # start tag or end tag |
| 1572 |
|
|
| 1573 |
|
redo A; |
| 1574 |
|
} else { |
| 1575 |
|
!!!cp ('124.4'); |
| 1576 |
|
!!!parse-error (type => 'nestc'); |
| 1577 |
|
## TODO: This error type is wrong. |
| 1578 |
|
$self->{state} = BEFORE_ATTRIBUTE_NAME_STATE; |
| 1579 |
|
## Reconsume. |
| 1580 |
|
redo A; |
| 1581 |
|
} |
| 1582 |
} elsif ($self->{state} == BOGUS_COMMENT_STATE) { |
} elsif ($self->{state} == BOGUS_COMMENT_STATE) { |
| 1583 |
## (only happen if PCDATA state) |
## (only happen if PCDATA state) |
| 1584 |
|
|
| 1585 |
my $token = {type => COMMENT_TOKEN, data => ''}; |
## NOTE: Set by the previous state |
| 1586 |
|
#my $token = {type => COMMENT_TOKEN, data => ''}; |
| 1587 |
|
|
| 1588 |
BC: { |
BC: { |
| 1589 |
if ($self->{next_char} == 0x003E) { # > |
if ($self->{next_char} == 0x003E) { # > |
| 1591 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 1592 |
!!!next-input-character; |
!!!next-input-character; |
| 1593 |
|
|
| 1594 |
!!!emit ($token); |
!!!emit ($self->{current_token}); # comment |
| 1595 |
|
|
| 1596 |
redo A; |
redo A; |
| 1597 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 1599 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 1600 |
## reconsume |
## reconsume |
| 1601 |
|
|
| 1602 |
!!!emit ($token); |
!!!emit ($self->{current_token}); # comment |
| 1603 |
|
|
| 1604 |
redo A; |
redo A; |
| 1605 |
} else { |
} else { |
| 1606 |
!!!cp (126); |
!!!cp (126); |
| 1607 |
$token->{data} .= chr ($self->{next_char}); |
$self->{current_token}->{data} .= chr ($self->{next_char}); # comment |
| 1608 |
!!!next-input-character; |
!!!next-input-character; |
| 1609 |
redo BC; |
redo BC; |
| 1610 |
} |
} |
| 1614 |
} elsif ($self->{state} == MARKUP_DECLARATION_OPEN_STATE) { |
} elsif ($self->{state} == MARKUP_DECLARATION_OPEN_STATE) { |
| 1615 |
## (only happen if PCDATA state) |
## (only happen if PCDATA state) |
| 1616 |
|
|
| 1617 |
|
my ($l, $c) = ($self->{line_prev}, $self->{column_prev} - 1); |
| 1618 |
|
|
| 1619 |
my @next_char; |
my @next_char; |
| 1620 |
push @next_char, $self->{next_char}; |
push @next_char, $self->{next_char}; |
| 1621 |
|
|
| 1624 |
push @next_char, $self->{next_char}; |
push @next_char, $self->{next_char}; |
| 1625 |
if ($self->{next_char} == 0x002D) { # - |
if ($self->{next_char} == 0x002D) { # - |
| 1626 |
!!!cp (127); |
!!!cp (127); |
| 1627 |
$self->{current_token} = {type => COMMENT_TOKEN, data => ''}; |
$self->{current_token} = {type => COMMENT_TOKEN, data => '', |
| 1628 |
|
line => $l, column => $c, |
| 1629 |
|
}; |
| 1630 |
$self->{state} = COMMENT_START_STATE; |
$self->{state} = COMMENT_START_STATE; |
| 1631 |
!!!next-input-character; |
!!!next-input-character; |
| 1632 |
redo A; |
redo A; |
| 1662 |
!!!cp (129); |
!!!cp (129); |
| 1663 |
## TODO: What a stupid code this is! |
## TODO: What a stupid code this is! |
| 1664 |
$self->{state} = DOCTYPE_STATE; |
$self->{state} = DOCTYPE_STATE; |
| 1665 |
|
$self->{current_token} = {type => DOCTYPE_TOKEN, |
| 1666 |
|
quirks => 1, |
| 1667 |
|
line => $l, column => $c, |
| 1668 |
|
}; |
| 1669 |
!!!next-input-character; |
!!!next-input-character; |
| 1670 |
redo A; |
redo A; |
| 1671 |
} else { |
} else { |
| 1686 |
} else { |
} else { |
| 1687 |
!!!cp (135); |
!!!cp (135); |
| 1688 |
} |
} |
| 1689 |
|
} elsif ($self->{insertion_mode} & IN_FOREIGN_CONTENT_IM and |
| 1690 |
|
$self->{open_elements}->[-1]->[1] & FOREIGN_EL and |
| 1691 |
|
$self->{next_char} == 0x005B) { # [ |
| 1692 |
|
!!!next-input-character; |
| 1693 |
|
push @next_char, $self->{next_char}; |
| 1694 |
|
if ($self->{next_char} == 0x0043) { # C |
| 1695 |
|
!!!next-input-character; |
| 1696 |
|
push @next_char, $self->{next_char}; |
| 1697 |
|
if ($self->{next_char} == 0x0044) { # D |
| 1698 |
|
!!!next-input-character; |
| 1699 |
|
push @next_char, $self->{next_char}; |
| 1700 |
|
if ($self->{next_char} == 0x0041) { # A |
| 1701 |
|
!!!next-input-character; |
| 1702 |
|
push @next_char, $self->{next_char}; |
| 1703 |
|
if ($self->{next_char} == 0x0054) { # T |
| 1704 |
|
!!!next-input-character; |
| 1705 |
|
push @next_char, $self->{next_char}; |
| 1706 |
|
if ($self->{next_char} == 0x0041) { # A |
| 1707 |
|
!!!next-input-character; |
| 1708 |
|
push @next_char, $self->{next_char}; |
| 1709 |
|
if ($self->{next_char} == 0x005B) { # [ |
| 1710 |
|
!!!cp (135.1); |
| 1711 |
|
$self->{state} = CDATA_BLOCK_STATE; |
| 1712 |
|
!!!next-input-character; |
| 1713 |
|
redo A; |
| 1714 |
|
} else { |
| 1715 |
|
!!!cp (135.2); |
| 1716 |
|
} |
| 1717 |
|
} else { |
| 1718 |
|
!!!cp (135.3); |
| 1719 |
|
} |
| 1720 |
|
} else { |
| 1721 |
|
!!!cp (135.4); |
| 1722 |
|
} |
| 1723 |
|
} else { |
| 1724 |
|
!!!cp (135.5); |
| 1725 |
|
} |
| 1726 |
|
} else { |
| 1727 |
|
!!!cp (135.6); |
| 1728 |
|
} |
| 1729 |
|
} else { |
| 1730 |
|
!!!cp (135.7); |
| 1731 |
|
} |
| 1732 |
} else { |
} else { |
| 1733 |
!!!cp (136); |
!!!cp (136); |
| 1734 |
} |
} |
| 1737 |
$self->{next_char} = shift @next_char; |
$self->{next_char} = shift @next_char; |
| 1738 |
!!!back-next-input-character (@next_char); |
!!!back-next-input-character (@next_char); |
| 1739 |
$self->{state} = BOGUS_COMMENT_STATE; |
$self->{state} = BOGUS_COMMENT_STATE; |
| 1740 |
|
$self->{current_token} = {type => COMMENT_TOKEN, data => '', |
| 1741 |
|
line => $l, column => $c, |
| 1742 |
|
}; |
| 1743 |
redo A; |
redo A; |
| 1744 |
|
|
| 1745 |
## ISSUE: typos in spec: chacacters, is is a parse error |
## ISSUE: typos in spec: chacacters, is is a parse error |
| 1863 |
redo A; |
redo A; |
| 1864 |
} elsif ($self->{next_char} == 0x002D) { # - |
} elsif ($self->{next_char} == 0x002D) { # - |
| 1865 |
!!!cp (152); |
!!!cp (152); |
| 1866 |
!!!parse-error (type => 'dash in comment'); |
!!!parse-error (type => 'dash in comment', |
| 1867 |
|
line => $self->{line_prev}, |
| 1868 |
|
column => $self->{column_prev}); |
| 1869 |
$self->{current_token}->{data} .= '-'; # comment |
$self->{current_token}->{data} .= '-'; # comment |
| 1870 |
## Stay in the state |
## Stay in the state |
| 1871 |
!!!next-input-character; |
!!!next-input-character; |
| 1881 |
redo A; |
redo A; |
| 1882 |
} else { |
} else { |
| 1883 |
!!!cp (154); |
!!!cp (154); |
| 1884 |
!!!parse-error (type => 'dash in comment'); |
!!!parse-error (type => 'dash in comment', |
| 1885 |
|
line => $self->{line_prev}, |
| 1886 |
|
column => $self->{column_prev}); |
| 1887 |
$self->{current_token}->{data} .= '--' . chr ($self->{next_char}); # comment |
$self->{current_token}->{data} .= '--' . chr ($self->{next_char}); # comment |
| 1888 |
$self->{state} = COMMENT_STATE; |
$self->{state} = COMMENT_STATE; |
| 1889 |
!!!next-input-character; |
!!!next-input-character; |
| 1922 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 1923 |
!!!next-input-character; |
!!!next-input-character; |
| 1924 |
|
|
| 1925 |
!!!emit ({type => DOCTYPE_TOKEN, quirks => 1}); |
!!!emit ($self->{current_token}); # DOCTYPE (quirks) |
| 1926 |
|
|
| 1927 |
redo A; |
redo A; |
| 1928 |
} elsif ($self->{next_char} == -1) { |
} elsif ($self->{next_char} == -1) { |
| 1931 |
$self->{state} = DATA_STATE; |
$self->{state} = DATA_STATE; |
| 1932 |
## reconsume |
## reconsume |
| 1933 |
|
|
| 1934 |
!!!emit ({type => DOCTYPE_TOKEN, quirks => 1}); |
!!!emit ($self->{current_token}); # DOCTYPE (quirks) |
| 1935 |
|
|
| 1936 |
redo A; |
redo A; |
| 1937 |
} else { |
} else { |
| 1938 |
!!!cp (160); |
!!!cp (160); |
| 1939 |
$self->{current_token} |
$self->{current_token}->{name} = chr $self->{next_char}; |
| 1940 |
= {type => DOCTYPE_TOKEN, |
delete $self->{current_token}->{quirks}; |
|
name => chr ($self->{next_char}), |
|
|
#quirks => 0, |
|
|
}; |
|
| 1941 |
## ISSUE: "Set the token's name name to the" in the spec |
## ISSUE: "Set the token's name name to the" in the spec |
| 1942 |
$self->{state} = DOCTYPE_NAME_STATE; |
$self->{state} = DOCTYPE_NAME_STATE; |
| 1943 |
!!!next-input-character; |
!!!next-input-character; |
| 2453 |
!!!next-input-character; |
!!!next-input-character; |
| 2454 |
redo A; |
redo A; |
| 2455 |
} |
} |
| 2456 |
|
} elsif ($self->{state} == CDATA_BLOCK_STATE) { |
| 2457 |
|
my $s = ''; |
| 2458 |
|
|
| 2459 |
|
my ($l, $c) = ($self->{line}, $self->{column}); |
| 2460 |
|
|
| 2461 |
|
CS: while ($self->{next_char} != -1) { |
| 2462 |
|
if ($self->{next_char} == 0x005D) { # ] |
| 2463 |
|
!!!next-input-character; |
| 2464 |
|
if ($self->{next_char} == 0x005D) { # ] |
| 2465 |
|
!!!next-input-character; |
| 2466 |
|
MDC: { |
| 2467 |
|
if ($self->{next_char} == 0x003E) { # > |
| 2468 |
|
!!!cp (221.1); |
| 2469 |
|
!!!next-input-character; |
| 2470 |
|
last CS; |
| 2471 |
|
} elsif ($self->{next_char} == 0x005D) { # ] |
| 2472 |
|
!!!cp (221.2); |
| 2473 |
|
$s .= ']'; |
| 2474 |
|
!!!next-input-character; |
| 2475 |
|
redo MDC; |
| 2476 |
|
} else { |
| 2477 |
|
!!!cp (221.3); |
| 2478 |
|
$s .= ']]'; |
| 2479 |
|
# |
| 2480 |
|
} |
| 2481 |
|
} # MDC |
| 2482 |
|
} else { |
| 2483 |
|
!!!cp (221.4); |
| 2484 |
|
$s .= ']'; |
| 2485 |
|
# |
| 2486 |
|
} |
| 2487 |
|
} else { |
| 2488 |
|
!!!cp (221.5); |
| 2489 |
|
# |
| 2490 |
|
} |
| 2491 |
|
$s .= chr $self->{next_char}; |
| 2492 |
|
!!!next-input-character; |
| 2493 |
|
} # CS |
| 2494 |
|
|
| 2495 |
|
$self->{state} = DATA_STATE; |
| 2496 |
|
## next-input-character done or EOF, which is reconsumed. |
| 2497 |
|
|
| 2498 |
|
if (length $s) { |
| 2499 |
|
!!!cp (221.6); |
| 2500 |
|
!!!emit ({type => CHARACTER_TOKEN, data => $s, |
| 2501 |
|
line => $l, column => $c}); |
| 2502 |
|
} else { |
| 2503 |
|
!!!cp (221.7); |
| 2504 |
|
} |
| 2505 |
|
|
| 2506 |
|
redo A; |
| 2507 |
|
|
| 2508 |
|
## ISSUE: "text tokens" in spec. |
| 2509 |
|
## TODO: Streaming support |
| 2510 |
} else { |
} else { |
| 2511 |
die "$0: $self->{state}: Unknown state"; |
die "$0: $self->{state}: Unknown state"; |
| 2512 |
} |
} |
| 2518 |
sub _tokenize_attempt_to_consume_an_entity ($$$) { |
sub _tokenize_attempt_to_consume_an_entity ($$$) { |
| 2519 |
my ($self, $in_attr, $additional) = @_; |
my ($self, $in_attr, $additional) = @_; |
| 2520 |
|
|
| 2521 |
|
my ($l, $c) = ($self->{line_prev}, $self->{column_prev}); |
| 2522 |
|
|
| 2523 |
if ({ |
if ({ |
| 2524 |
0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, # HT, LF, VT, FF, |
0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, # HT, LF, VT, FF, |
| 2525 |
0x0020 => 1, 0x003C => 1, 0x0026 => 1, -1 => 1, # SP, <, & # 0x000D # CR |
0x0020 => 1, 0x003C => 1, 0x0026 => 1, -1 => 1, # SP, <, & # 0x000D # CR |
| 2560 |
redo X; |
redo X; |
| 2561 |
} elsif (not defined $code) { # no hexadecimal digit |
} elsif (not defined $code) { # no hexadecimal digit |
| 2562 |
!!!cp (1005); |
!!!cp (1005); |
| 2563 |
!!!parse-error (type => 'bare hcro'); |
!!!parse-error (type => 'bare hcro', line => $l, column => $c); |
| 2564 |
!!!back-next-input-character ($x_char, $self->{next_char}); |
!!!back-next-input-character ($x_char, $self->{next_char}); |
| 2565 |
$self->{next_char} = 0x0023; # # |
$self->{next_char} = 0x0023; # # |
| 2566 |
return undef; |
return undef; |
| 2569 |
!!!next-input-character; |
!!!next-input-character; |
| 2570 |
} else { |
} else { |
| 2571 |
!!!cp (1007); |
!!!cp (1007); |
| 2572 |
!!!parse-error (type => 'no refc'); |
!!!parse-error (type => 'no refc', line => $l, column => $c); |
| 2573 |
} |
} |
| 2574 |
|
|
| 2575 |
if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) { |
if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) { |
| 2576 |
!!!cp (1008); |
!!!cp (1008); |
| 2577 |
!!!parse-error (type => sprintf 'invalid character reference:U+%04X', $code); |
!!!parse-error (type => (sprintf 'invalid character reference:U+%04X', $code), line => $l, column => $c); |
| 2578 |
$code = 0xFFFD; |
$code = 0xFFFD; |
| 2579 |
} elsif ($code > 0x10FFFF) { |
} elsif ($code > 0x10FFFF) { |
| 2580 |
!!!cp (1009); |
!!!cp (1009); |
| 2581 |
!!!parse-error (type => sprintf 'invalid character reference:U-%08X', $code); |
!!!parse-error (type => (sprintf 'invalid character reference:U-%08X', $code), line => $l, column => $c); |
| 2582 |
$code = 0xFFFD; |
$code = 0xFFFD; |
| 2583 |
} elsif ($code == 0x000D) { |
} elsif ($code == 0x000D) { |
| 2584 |
!!!cp (1010); |
!!!cp (1010); |
| 2585 |
!!!parse-error (type => 'CR character reference'); |
!!!parse-error (type => 'CR character reference', line => $l, column => $c); |
| 2586 |
$code = 0x000A; |
$code = 0x000A; |
| 2587 |
} elsif (0x80 <= $code and $code <= 0x9F) { |
} elsif (0x80 <= $code and $code <= 0x9F) { |
| 2588 |
!!!cp (1011); |
!!!cp (1011); |
| 2589 |
!!!parse-error (type => sprintf 'C1 character reference:U+%04X', $code); |
!!!parse-error (type => (sprintf 'C1 character reference:U+%04X', $code), line => $l, column => $c); |
| 2590 |
$code = $c1_entity_char->{$code}; |
$code = $c1_entity_char->{$code}; |
| 2591 |
} |
} |
| 2592 |
|
|
| 2593 |
return {type => CHARACTER_TOKEN, data => chr $code, |
return {type => CHARACTER_TOKEN, data => chr $code, |
| 2594 |
has_reference => 1}; |
has_reference => 1, |
| 2595 |
|
line => $l, column => $c, |
| 2596 |
|
}; |
| 2597 |
} # X |
} # X |
| 2598 |
} elsif (0x0030 <= $self->{next_char} and |
} elsif (0x0030 <= $self->{next_char} and |
| 2599 |
$self->{next_char} <= 0x0039) { # 0..9 |
$self->{next_char} <= 0x0039) { # 0..9 |
| 2614 |
!!!next-input-character; |
!!!next-input-character; |
| 2615 |
} else { |
} else { |
| 2616 |
!!!cp (1014); |
!!!cp (1014); |
| 2617 |
!!!parse-error (type => 'no refc'); |
!!!parse-error (type => 'no refc', line => $l, column => $c); |
| 2618 |
} |
} |
| 2619 |
|
|
| 2620 |
if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) { |
if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) { |
| 2621 |
!!!cp (1015); |
!!!cp (1015); |
| 2622 |
!!!parse-error (type => sprintf 'invalid character reference:U+%04X', $code); |
!!!parse-error (type => (sprintf 'invalid character reference:U+%04X', $code), line => $l, column => $c); |
| 2623 |
$code = 0xFFFD; |
$code = 0xFFFD; |
| 2624 |
} elsif ($code > 0x10FFFF) { |
} elsif ($code > 0x10FFFF) { |
| 2625 |
!!!cp (1016); |
!!!cp (1016); |
| 2626 |
!!!parse-error (type => sprintf 'invalid character reference:U-%08X', $code); |
!!!parse-error (type => (sprintf 'invalid character reference:U-%08X', $code), line => $l, column => $c); |
| 2627 |
$code = 0xFFFD; |
$code = 0xFFFD; |
| 2628 |
} elsif ($code == 0x000D) { |
} elsif ($code == 0x000D) { |
| 2629 |
!!!cp (1017); |
!!!cp (1017); |
| 2630 |
!!!parse-error (type => 'CR character reference'); |
!!!parse-error (type => 'CR character reference', line => $l, column => $c); |
| 2631 |
$code = 0x000A; |
$code = 0x000A; |
| 2632 |
} elsif (0x80 <= $code and $code <= 0x9F) { |
} elsif (0x80 <= $code and $code <= 0x9F) { |
| 2633 |
!!!cp (1018); |
!!!cp (1018); |
| 2634 |
!!!parse-error (type => sprintf 'C1 character reference:U+%04X', $code); |
!!!parse-error (type => (sprintf 'C1 character reference:U+%04X', $code), line => $l, column => $c); |
| 2635 |
$code = $c1_entity_char->{$code}; |
$code = $c1_entity_char->{$code}; |
| 2636 |
} |
} |
| 2637 |
|
|
| 2638 |
return {type => CHARACTER_TOKEN, data => chr $code, has_reference => 1}; |
return {type => CHARACTER_TOKEN, data => chr $code, has_reference => 1, |
| 2639 |
|
line => $l, column => $c, |
| 2640 |
|
}; |
| 2641 |
} else { |
} else { |
| 2642 |
!!!cp (1019); |
!!!cp (1019); |
| 2643 |
!!!parse-error (type => 'bare nero'); |
!!!parse-error (type => 'bare nero', line => $l, column => $c); |
| 2644 |
!!!back-next-input-character ($self->{next_char}); |
!!!back-next-input-character ($self->{next_char}); |
| 2645 |
$self->{next_char} = 0x0023; # # |
$self->{next_char} = 0x0023; # # |
| 2646 |
return undef; |
return undef; |
| 2657 |
require Whatpm::_NamedEntityList; |
require Whatpm::_NamedEntityList; |
| 2658 |
our $EntityChar; |
our $EntityChar; |
| 2659 |
|
|
| 2660 |
while (length $entity_name < 10 and |
while (length $entity_name < 30 and |
| 2661 |
## NOTE: Some number greater than the maximum length of entity name |
## NOTE: Some number greater than the maximum length of entity name |
| 2662 |
((0x0041 <= $self->{next_char} and # a |
((0x0041 <= $self->{next_char} and # a |
| 2663 |
$self->{next_char} <= 0x005A) or # x |
$self->{next_char} <= 0x005A) or # x |
| 2690 |
|
|
| 2691 |
if ($match > 0) { |
if ($match > 0) { |
| 2692 |
!!!cp (1023); |
!!!cp (1023); |
| 2693 |
return {type => CHARACTER_TOKEN, data => $value, has_reference => 1}; |
return {type => CHARACTER_TOKEN, data => $value, has_reference => 1, |
| 2694 |
|
line => $l, column => $c, |
| 2695 |
|
}; |
| 2696 |
} elsif ($match < 0) { |
} elsif ($match < 0) { |
| 2697 |
!!!parse-error (type => 'no refc'); |
!!!parse-error (type => 'no refc', line => $l, column => $c); |
| 2698 |
if ($in_attr and $match < -1) { |
if ($in_attr and $match < -1) { |
| 2699 |
!!!cp (1024); |
!!!cp (1024); |
| 2700 |
return {type => CHARACTER_TOKEN, data => '&'.$entity_name}; |
return {type => CHARACTER_TOKEN, data => '&'.$entity_name, |
| 2701 |
|
line => $l, column => $c, |
| 2702 |
|
}; |
| 2703 |
} else { |
} else { |
| 2704 |
!!!cp (1025); |
!!!cp (1025); |
| 2705 |
return {type => CHARACTER_TOKEN, data => $value, has_reference => 1}; |
return {type => CHARACTER_TOKEN, data => $value, has_reference => 1, |
| 2706 |
|
line => $l, column => $c, |
| 2707 |
|
}; |
| 2708 |
} |
} |
| 2709 |
} else { |
} else { |
| 2710 |
!!!cp (1026); |
!!!cp (1026); |
| 2711 |
!!!parse-error (type => 'bare ero'); |
!!!parse-error (type => 'bare ero', line => $l, column => $c); |
| 2712 |
## NOTE: "No characters are consumed" in the spec. |
## NOTE: "No characters are consumed" in the spec. |
| 2713 |
return {type => CHARACTER_TOKEN, data => '&'.$value}; |
return {type => CHARACTER_TOKEN, data => '&'.$value, |
| 2714 |
|
line => $l, column => $c, |
| 2715 |
|
}; |
| 2716 |
} |
} |
| 2717 |
} else { |
} else { |
| 2718 |
!!!cp (1027); |
!!!cp (1027); |
| 2719 |
## no characters are consumed |
## no characters are consumed |
| 2720 |
!!!parse-error (type => 'bare ero'); |
!!!parse-error (type => 'bare ero', line => $l, column => $c); |
| 2721 |
return undef; |
return undef; |
| 2722 |
} |
} |
| 2723 |
} # _tokenize_attempt_to_consume_an_entity |
} # _tokenize_attempt_to_consume_an_entity |
| 2788 |
defined $token->{public_identifier} or |
defined $token->{public_identifier} or |
| 2789 |
defined $token->{system_identifier}) { |
defined $token->{system_identifier}) { |
| 2790 |
!!!cp ('t1'); |
!!!cp ('t1'); |
| 2791 |
!!!parse-error (type => 'not HTML5'); |
!!!parse-error (type => 'not HTML5', token => $token); |
| 2792 |
} elsif ($doctype_name ne 'HTML') { |
} elsif ($doctype_name ne 'HTML') { |
| 2793 |
!!!cp ('t2'); |
!!!cp ('t2'); |
| 2794 |
## ISSUE: ASCII case-insensitive? (in fact it does not matter) |
## ISSUE: ASCII case-insensitive? (in fact it does not matter) |
| 2795 |
!!!parse-error (type => 'not HTML5'); |
!!!parse-error (type => 'not HTML5', token => $token); |
| 2796 |
} else { |
} else { |
| 2797 |
!!!cp ('t3'); |
!!!cp ('t3'); |
| 2798 |
} |
} |
| 2799 |
|
|
| 2800 |
my $doctype = $self->{document}->create_document_type_definition |
my $doctype = $self->{document}->create_document_type_definition |
| 2801 |
($token->{name}); ## ISSUE: If name is missing (e.g. <!DOCTYPE>)? |
($token->{name}); ## ISSUE: If name is missing (e.g. <!DOCTYPE>)? |
| 2802 |
|
## NOTE: Default value for both |public_id| and |system_id| attributes |
| 2803 |
|
## are empty strings, so that we don't set any value in missing cases. |
| 2804 |
$doctype->public_id ($token->{public_identifier}) |
$doctype->public_id ($token->{public_identifier}) |
| 2805 |
if defined $token->{public_identifier}; |
if defined $token->{public_identifier}; |
| 2806 |
$doctype->system_id ($token->{system_identifier}) |
$doctype->system_id ($token->{system_identifier}) |
| 2933 |
END_OF_FILE_TOKEN, 1, |
END_OF_FILE_TOKEN, 1, |
| 2934 |
}->{$token->{type}}) { |
}->{$token->{type}}) { |
| 2935 |
!!!cp ('t14'); |
!!!cp ('t14'); |
| 2936 |
!!!parse-error (type => 'no DOCTYPE'); |
!!!parse-error (type => 'no DOCTYPE', token => $token); |
| 2937 |
$self->{document}->manakai_compat_mode ('quirks'); |
$self->{document}->manakai_compat_mode ('quirks'); |
| 2938 |
## Go to the "before html" insertion mode. |
## Go to the "before html" insertion mode. |
| 2939 |
## reprocess |
## reprocess |
| 2940 |
|
!!!ack-later; |
| 2941 |
return; |
return; |
| 2942 |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
| 2943 |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { # \x0D |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { # \x0D |
| 2955 |
!!!cp ('t17'); |
!!!cp ('t17'); |
| 2956 |
} |
} |
| 2957 |
|
|
| 2958 |
!!!parse-error (type => 'no DOCTYPE'); |
!!!parse-error (type => 'no DOCTYPE', token => $token); |
| 2959 |
$self->{document}->manakai_compat_mode ('quirks'); |
$self->{document}->manakai_compat_mode ('quirks'); |
| 2960 |
## Go to the "before html" insertion mode. |
## Go to the "before html" insertion mode. |
| 2961 |
## reprocess |
## reprocess |
| 2984 |
B: { |
B: { |
| 2985 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 2986 |
!!!cp ('t19'); |
!!!cp ('t19'); |
| 2987 |
!!!parse-error (type => 'in html:#DOCTYPE'); |
!!!parse-error (type => 'in html:#DOCTYPE', token => $token); |
| 2988 |
## Ignore the token |
## Ignore the token |
| 2989 |
## Stay in the insertion mode. |
## Stay in the insertion mode. |
| 2990 |
!!!next-token; |
!!!next-token; |
| 3018 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 3019 |
if ($token->{tag_name} eq 'html') { |
if ($token->{tag_name} eq 'html') { |
| 3020 |
my $root_element; |
my $root_element; |
| 3021 |
!!!create-element ($root_element, $token->{tag_name}, $token->{attributes}); |
!!!create-element ($root_element, $HTML_NS, $token->{tag_name}, $token->{attributes}, $token); |
| 3022 |
$self->{document}->append_child ($root_element); |
$self->{document}->append_child ($root_element); |
| 3023 |
push @{$self->{open_elements}}, [$root_element, 'html']; |
push @{$self->{open_elements}}, |
| 3024 |
|
[$root_element, $el_category->{html}]; |
| 3025 |
|
|
| 3026 |
if ($token->{attributes}->{manifest}) { |
if ($token->{attributes}->{manifest}) { |
| 3027 |
!!!cp ('t24'); |
!!!cp ('t24'); |
| 3028 |
$self->{application_cache_selection} |
$self->{application_cache_selection} |
| 3029 |
->($token->{attributes}->{manifest}->{value}); |
->($token->{attributes}->{manifest}->{value}); |
| 3030 |
## ISSUE: No relative reference resolution? |
## ISSUE: Spec is unclear on relative references. |
| 3031 |
|
## According to Hixie (#whatwg 2008-03-19), it should be |
| 3032 |
|
## resolved against the base URI of the document in HTML |
| 3033 |
|
## or xml:base of the element in XHTML. |
| 3034 |
} else { |
} else { |
| 3035 |
!!!cp ('t25'); |
!!!cp ('t25'); |
| 3036 |
$self->{application_cache_selection}->(undef); |
$self->{application_cache_selection}->(undef); |
| 3037 |
} |
} |
| 3038 |
|
|
| 3039 |
|
!!!nack ('t25c'); |
| 3040 |
|
|
| 3041 |
!!!next-token; |
!!!next-token; |
| 3042 |
return; ## Go to the "before head" insertion mode. |
return; ## Go to the "before head" insertion mode. |
| 3043 |
} else { |
} else { |
| 3054 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 3055 |
} |
} |
| 3056 |
|
|
| 3057 |
my $root_element; !!!create-element ($root_element, 'html'); |
my $root_element; |
| 3058 |
|
!!!create-element ($root_element, $HTML_NS, 'html',, $token); |
| 3059 |
$self->{document}->append_child ($root_element); |
$self->{document}->append_child ($root_element); |
| 3060 |
push @{$self->{open_elements}}, [$root_element, 'html']; |
push @{$self->{open_elements}}, [$root_element, $el_category->{html}]; |
| 3061 |
|
|
| 3062 |
$self->{application_cache_selection}->(undef); |
$self->{application_cache_selection}->(undef); |
| 3063 |
|
|
| 3064 |
## NOTE: Reprocess the token. |
## NOTE: Reprocess the token. |
| 3065 |
|
!!!ack-later; |
| 3066 |
return; ## Go to the "before head" insertion mode. |
return; ## Go to the "before head" insertion mode. |
| 3067 |
|
|
| 3068 |
## ISSUE: There is an issue in the spec |
## ISSUE: There is an issue in the spec |
| 3083 |
|
|
| 3084 |
## Step 3 |
## Step 3 |
| 3085 |
S3: { |
S3: { |
|
## ISSUE: Oops! "If node is the first node in the stack of open |
|
|
## elements, then set last to true. If the context element of the |
|
|
## HTML fragment parsing algorithm is neither a td element nor a |
|
|
## th element, then set node to the context element. (fragment case)": |
|
|
## The second "if" is in the scope of the first "if"!? |
|
| 3086 |
if ($self->{open_elements}->[0]->[0] eq $node->[0]) { |
if ($self->{open_elements}->[0]->[0] eq $node->[0]) { |
| 3087 |
$last = 1; |
$last = 1; |
| 3088 |
if (defined $self->{inner_html_node}) { |
if (defined $self->{inner_html_node}) { |
| 3089 |
if ($self->{inner_html_node}->[1] eq 'td' or |
if ($self->{inner_html_node}->[1] & TABLE_CELL_EL) { |
|
$self->{inner_html_node}->[1] eq 'th') { |
|
| 3090 |
!!!cp ('t27'); |
!!!cp ('t27'); |
| 3091 |
# |
# |
| 3092 |
} else { |
} else { |
| 3096 |
} |
} |
| 3097 |
} |
} |
| 3098 |
|
|
| 3099 |
## Step 4..13 |
## Step 4..14 |
| 3100 |
my $new_mode = { |
my $new_mode; |
| 3101 |
|
if ($node->[1] & FOREIGN_EL) { |
| 3102 |
|
## NOTE: Strictly spaking, the line below only applies to MathML and |
| 3103 |
|
## SVG elements. Currently the HTML syntax supports only MathML and |
| 3104 |
|
## SVG elements as foreigners. |
| 3105 |
|
$new_mode = $self->{insertion_mode} | IN_FOREIGN_CONTENT_IM; |
| 3106 |
|
## ISSUE: What is set as the secondary insertion mode? |
| 3107 |
|
} else { |
| 3108 |
|
$new_mode = { |
| 3109 |
select => IN_SELECT_IM, |
select => IN_SELECT_IM, |
| 3110 |
## NOTE: |option| and |optgroup| do not set |
## NOTE: |option| and |optgroup| do not set |
| 3111 |
## insertion mode to "in select" by themselves. |
## insertion mode to "in select" by themselves. |
| 3121 |
head => IN_BODY_IM, # not in head! |
head => IN_BODY_IM, # not in head! |
| 3122 |
body => IN_BODY_IM, |
body => IN_BODY_IM, |
| 3123 |
frameset => IN_FRAMESET_IM, |
frameset => IN_FRAMESET_IM, |
| 3124 |
}->{$node->[1]}; |
}->{$node->[0]->manakai_local_name}; |
| 3125 |
$self->{insertion_mode} = $new_mode and return if defined $new_mode; |
} |
| 3126 |
|
$self->{insertion_mode} = $new_mode and return if defined $new_mode; |
| 3127 |
|
|
| 3128 |
## Step 14 |
## Step 15 |
| 3129 |
if ($node->[1] eq 'html') { |
if ($node->[1] & HTML_EL) { |
| 3130 |
unless (defined $self->{head_element}) { |
unless (defined $self->{head_element}) { |
| 3131 |
!!!cp ('t29'); |
!!!cp ('t29'); |
| 3132 |
$self->{insertion_mode} = BEFORE_HEAD_IM; |
$self->{insertion_mode} = BEFORE_HEAD_IM; |
| 3140 |
!!!cp ('t31'); |
!!!cp ('t31'); |
| 3141 |
} |
} |
| 3142 |
|
|
| 3143 |
## Step 15 |
## Step 16 |
| 3144 |
$self->{insertion_mode} = IN_BODY_IM and return if $last; |
$self->{insertion_mode} = IN_BODY_IM and return if $last; |
| 3145 |
|
|
| 3146 |
## Step 16 |
## Step 17 |
| 3147 |
$i--; |
$i--; |
| 3148 |
$node = $self->{open_elements}->[$i]; |
$node = $self->{open_elements}->[$i]; |
| 3149 |
|
|
| 3150 |
## Step 17 |
## Step 18 |
| 3151 |
redo S3; |
redo S3; |
| 3152 |
} # S3 |
} # S3 |
| 3153 |
|
|
| 3251 |
!!!cp ('t39'); |
!!!cp ('t39'); |
| 3252 |
}; # $clear_up_to_marker |
}; # $clear_up_to_marker |
| 3253 |
|
|
| 3254 |
my $parse_rcdata = sub ($$) { |
my $insert; |
| 3255 |
my ($content_model_flag, $insert) = @_; |
|
| 3256 |
|
my $parse_rcdata = sub ($) { |
| 3257 |
|
my ($content_model_flag) = @_; |
| 3258 |
|
|
| 3259 |
## Step 1 |
## Step 1 |
| 3260 |
my $start_tag_name = $token->{tag_name}; |
my $start_tag_name = $token->{tag_name}; |
| 3261 |
my $el; |
my $el; |
| 3262 |
!!!create-element ($el, $start_tag_name, $token->{attributes}); |
!!!create-element ($el, $HTML_NS, $start_tag_name, $token->{attributes}, $token); |
| 3263 |
|
|
| 3264 |
## Step 2 |
## Step 2 |
| 3265 |
$insert->($el); # /context node/->append_child ($el) |
$insert->($el); |
| 3266 |
|
|
| 3267 |
## Step 3 |
## Step 3 |
| 3268 |
$self->{content_model} = $content_model_flag; # CDATA or RCDATA |
$self->{content_model} = $content_model_flag; # CDATA or RCDATA |
| 3270 |
|
|
| 3271 |
## Step 4 |
## Step 4 |
| 3272 |
my $text = ''; |
my $text = ''; |
| 3273 |
|
!!!nack ('t40.1'); |
| 3274 |
!!!next-token; |
!!!next-token; |
| 3275 |
while ($token->{type} == CHARACTER_TOKEN) { # or until stop tokenizing |
while ($token->{type} == CHARACTER_TOKEN) { # or until stop tokenizing |
| 3276 |
!!!cp ('t40'); |
!!!cp ('t40'); |
| 3293 |
$token->{tag_name} eq $start_tag_name) { |
$token->{tag_name} eq $start_tag_name) { |
| 3294 |
!!!cp ('t42'); |
!!!cp ('t42'); |
| 3295 |
## Ignore the token |
## Ignore the token |
|
} elsif ($content_model_flag == CDATA_CONTENT_MODEL) { |
|
|
!!!cp ('t43'); |
|
|
!!!parse-error (type => 'in CDATA:#'.$token->{type}); |
|
|
} elsif ($content_model_flag == RCDATA_CONTENT_MODEL) { |
|
|
!!!cp ('t44'); |
|
|
!!!parse-error (type => 'in RCDATA:#'.$token->{type}); |
|
| 3296 |
} else { |
} else { |
| 3297 |
die "$0: $content_model_flag in parse_rcdata"; |
## NOTE: An end-of-file token. |
| 3298 |
|
if ($content_model_flag == CDATA_CONTENT_MODEL) { |
| 3299 |
|
!!!cp ('t43'); |
| 3300 |
|
!!!parse-error (type => 'in CDATA:#'.$token->{type}, token => $token); |
| 3301 |
|
} elsif ($content_model_flag == RCDATA_CONTENT_MODEL) { |
| 3302 |
|
!!!cp ('t44'); |
| 3303 |
|
!!!parse-error (type => 'in RCDATA:#'.$token->{type}, token => $token); |
| 3304 |
|
} else { |
| 3305 |
|
die "$0: $content_model_flag in parse_rcdata"; |
| 3306 |
|
} |
| 3307 |
} |
} |
| 3308 |
!!!next-token; |
!!!next-token; |
| 3309 |
}; # $parse_rcdata |
}; # $parse_rcdata |
| 3310 |
|
|
| 3311 |
my $script_start_tag = sub ($) { |
my $script_start_tag = sub () { |
|
my $insert = $_[0]; |
|
| 3312 |
my $script_el; |
my $script_el; |
| 3313 |
!!!create-element ($script_el, 'script', $token->{attributes}); |
!!!create-element ($script_el, $HTML_NS, 'script', $token->{attributes}, $token); |
| 3314 |
## TODO: mark as "parser-inserted" |
## TODO: mark as "parser-inserted" |
| 3315 |
|
|
| 3316 |
$self->{content_model} = CDATA_CONTENT_MODEL; |
$self->{content_model} = CDATA_CONTENT_MODEL; |
| 3317 |
delete $self->{escape}; # MUST |
delete $self->{escape}; # MUST |
| 3318 |
|
|
| 3319 |
my $text = ''; |
my $text = ''; |
| 3320 |
|
!!!nack ('t45.1'); |
| 3321 |
!!!next-token; |
!!!next-token; |
| 3322 |
while ($token->{type} == CHARACTER_TOKEN) { |
while ($token->{type} == CHARACTER_TOKEN) { |
| 3323 |
!!!cp ('t45'); |
!!!cp ('t45'); |
| 3337 |
## Ignore the token |
## Ignore the token |
| 3338 |
} else { |
} else { |
| 3339 |
!!!cp ('t48'); |
!!!cp ('t48'); |
| 3340 |
!!!parse-error (type => 'in CDATA:#'.$token->{type}); |
!!!parse-error (type => 'in CDATA:#'.$token->{type}, token => $token); |
| 3341 |
## ISSUE: And ignore? |
## ISSUE: And ignore? |
| 3342 |
## TODO: mark as "already executed" |
## TODO: mark as "already executed" |
| 3343 |
} |
} |
| 3360 |
!!!next-token; |
!!!next-token; |
| 3361 |
}; # $script_start_tag |
}; # $script_start_tag |
| 3362 |
|
|
| 3363 |
|
## NOTE: $open_tables->[-1]->[0] is the "current table" element node. |
| 3364 |
|
## NOTE: $open_tables->[-1]->[1] is the "tainted" flag. |
| 3365 |
|
my $open_tables = [[$self->{open_elements}->[0]->[0]]]; |
| 3366 |
|
|
| 3367 |
my $formatting_end_tag = sub { |
my $formatting_end_tag = sub { |
| 3368 |
my $tag_name = shift; |
my $end_tag_token = shift; |
| 3369 |
|
my $tag_name = $end_tag_token->{tag_name}; |
| 3370 |
|
|
| 3371 |
|
## NOTE: The adoption agency algorithm (AAA). |
| 3372 |
|
|
| 3373 |
FET: { |
FET: { |
| 3374 |
## Step 1 |
## Step 1 |
| 3375 |
my $formatting_element; |
my $formatting_element; |
| 3376 |
my $formatting_element_i_in_active; |
my $formatting_element_i_in_active; |
| 3377 |
AFE: for (reverse 0..$#$active_formatting_elements) { |
AFE: for (reverse 0..$#$active_formatting_elements) { |
| 3378 |
if ($active_formatting_elements->[$_]->[1] eq $tag_name) { |
if ($active_formatting_elements->[$_]->[0] eq '#marker') { |
| 3379 |
|
!!!cp ('t52'); |
| 3380 |
|
last AFE; |
| 3381 |
|
} elsif ($active_formatting_elements->[$_]->[0]->manakai_local_name |
| 3382 |
|
eq $tag_name) { |
| 3383 |
!!!cp ('t51'); |
!!!cp ('t51'); |
| 3384 |
$formatting_element = $active_formatting_elements->[$_]; |
$formatting_element = $active_formatting_elements->[$_]; |
| 3385 |
$formatting_element_i_in_active = $_; |
$formatting_element_i_in_active = $_; |
| 3386 |
last AFE; |
last AFE; |
|
} elsif ($active_formatting_elements->[$_]->[0] eq '#marker') { |
|
|
!!!cp ('t52'); |
|
|
last AFE; |
|
| 3387 |
} |
} |
| 3388 |
} # AFE |
} # AFE |
| 3389 |
unless (defined $formatting_element) { |
unless (defined $formatting_element) { |
| 3390 |
!!!cp ('t53'); |
!!!cp ('t53'); |
| 3391 |
!!!parse-error (type => 'unmatched end tag:'.$tag_name); |
!!!parse-error (type => 'unmatched end tag:'.$tag_name, token => $end_tag_token); |
| 3392 |
## Ignore the token |
## Ignore the token |
| 3393 |
!!!next-token; |
!!!next-token; |
| 3394 |
return; |
return; |
| 3405 |
last INSCOPE; |
last INSCOPE; |
| 3406 |
} else { # in open elements but not in scope |
} else { # in open elements but not in scope |
| 3407 |
!!!cp ('t55'); |
!!!cp ('t55'); |
| 3408 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, |
| 3409 |
|
token => $end_tag_token); |
| 3410 |
## Ignore the token |
## Ignore the token |
| 3411 |
!!!next-token; |
!!!next-token; |
| 3412 |
return; |
return; |
| 3413 |
} |
} |
| 3414 |
} elsif ({ |
} elsif ($node->[1] & SCOPING_EL) { |
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 3415 |
!!!cp ('t56'); |
!!!cp ('t56'); |
| 3416 |
$in_scope = 0; |
$in_scope = 0; |
| 3417 |
} |
} |
| 3418 |
} # INSCOPE |
} # INSCOPE |
| 3419 |
unless (defined $formatting_element_i_in_open) { |
unless (defined $formatting_element_i_in_open) { |
| 3420 |
!!!cp ('t57'); |
!!!cp ('t57'); |
| 3421 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, |
| 3422 |
|
token => $end_tag_token); |
| 3423 |
pop @$active_formatting_elements; # $formatting_element |
pop @$active_formatting_elements; # $formatting_element |
| 3424 |
!!!next-token; ## TODO: ok? |
!!!next-token; ## TODO: ok? |
| 3425 |
return; |
return; |
| 3426 |
} |
} |
| 3427 |
if (not $self->{open_elements}->[-1]->[0] eq $formatting_element->[0]) { |
if (not $self->{open_elements}->[-1]->[0] eq $formatting_element->[0]) { |
| 3428 |
!!!cp ('t58'); |
!!!cp ('t58'); |
| 3429 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 3430 |
|
value => $self->{open_elements}->[-1]->[0] |
| 3431 |
|
->manakai_local_name, |
| 3432 |
|
token => $end_tag_token); |
| 3433 |
} |
} |
| 3434 |
|
|
| 3435 |
## Step 2 |
## Step 2 |
| 3437 |
my $furthest_block_i_in_open; |
my $furthest_block_i_in_open; |
| 3438 |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3439 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 3440 |
if (not $formatting_category->{$node->[1]} and |
if (not ($node->[1] & FORMATTING_EL) and |
| 3441 |
#not $phrasing_category->{$node->[1]} and |
#not $phrasing_category->{$node->[1]} and |
| 3442 |
($special_category->{$node->[1]} or |
($node->[1] & SPECIAL_EL or |
| 3443 |
$scoping_category->{$node->[1]})) { |
$node->[1] & SCOPING_EL)) { ## Scoping is redundant, maybe |
| 3444 |
!!!cp ('t59'); |
!!!cp ('t59'); |
| 3445 |
$furthest_block = $node; |
$furthest_block = $node; |
| 3446 |
$furthest_block_i_in_open = $_; |
$furthest_block_i_in_open = $_; |
| 3526 |
} # S7 |
} # S7 |
| 3527 |
|
|
| 3528 |
## Step 8 |
## Step 8 |
| 3529 |
$common_ancestor_node->[0]->append_child ($last_node->[0]); |
if ($common_ancestor_node->[1] & TABLE_ROWS_EL) { |
| 3530 |
|
my $foster_parent_element; |
| 3531 |
|
my $next_sibling; |
| 3532 |
|
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3533 |
|
if ($self->{open_elements}->[$_]->[1] & TABLE_EL) { |
| 3534 |
|
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
| 3535 |
|
if (defined $parent and $parent->node_type == 1) { |
| 3536 |
|
!!!cp ('t65.1'); |
| 3537 |
|
$foster_parent_element = $parent; |
| 3538 |
|
$next_sibling = $self->{open_elements}->[$_]->[0]; |
| 3539 |
|
} else { |
| 3540 |
|
!!!cp ('t65.2'); |
| 3541 |
|
$foster_parent_element |
| 3542 |
|
= $self->{open_elements}->[$_ - 1]->[0]; |
| 3543 |
|
} |
| 3544 |
|
last OE; |
| 3545 |
|
} |
| 3546 |
|
} # OE |
| 3547 |
|
$foster_parent_element = $self->{open_elements}->[0]->[0] |
| 3548 |
|
unless defined $foster_parent_element; |
| 3549 |
|
$foster_parent_element->insert_before ($last_node->[0], $next_sibling); |
| 3550 |
|
$open_tables->[-1]->[1] = 1; # tainted |
| 3551 |
|
} else { |
| 3552 |
|
!!!cp ('t65.3'); |
| 3553 |
|
$common_ancestor_node->[0]->append_child ($last_node->[0]); |
| 3554 |
|
} |
| 3555 |
|
|
| 3556 |
## Step 9 |
## Step 9 |
| 3557 |
my $clone = [$formatting_element->[0]->clone_node (0), |
my $clone = [$formatting_element->[0]->clone_node (0), |
| 3597 |
} # FET |
} # FET |
| 3598 |
}; # $formatting_end_tag |
}; # $formatting_end_tag |
| 3599 |
|
|
| 3600 |
my $insert_to_current = sub { |
$insert = my $insert_to_current = sub { |
| 3601 |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
| 3602 |
}; # $insert_to_current |
}; # $insert_to_current |
| 3603 |
|
|
| 3604 |
my $insert_to_foster = sub { |
my $insert_to_foster = sub { |
| 3605 |
my $child = shift; |
my $child = shift; |
| 3606 |
if ({ |
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
| 3607 |
table => 1, tbody => 1, tfoot => 1, |
# MUST |
| 3608 |
thead => 1, tr => 1, |
my $foster_parent_element; |
| 3609 |
}->{$self->{open_elements}->[-1]->[1]}) { |
my $next_sibling; |
| 3610 |
# MUST |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3611 |
my $foster_parent_element; |
if ($self->{open_elements}->[$_]->[1] & TABLE_EL) { |
|
my $next_sibling; |
|
|
OE: for (reverse 0..$#{$self->{open_elements}}) { |
|
|
if ($self->{open_elements}->[$_]->[1] eq 'table') { |
|
| 3612 |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
| 3613 |
if (defined $parent and $parent->node_type == 1) { |
if (defined $parent and $parent->node_type == 1) { |
| 3614 |
!!!cp ('t70'); |
!!!cp ('t70'); |
| 3626 |
unless defined $foster_parent_element; |
unless defined $foster_parent_element; |
| 3627 |
$foster_parent_element->insert_before |
$foster_parent_element->insert_before |
| 3628 |
($child, $next_sibling); |
($child, $next_sibling); |
| 3629 |
} else { |
$open_tables->[-1]->[1] = 1; # tainted |
| 3630 |
!!!cp ('t72'); |
} else { |
| 3631 |
$self->{open_elements}->[-1]->[0]->append_child ($child); |
!!!cp ('t72'); |
| 3632 |
} |
$self->{open_elements}->[-1]->[0]->append_child ($child); |
| 3633 |
|
} |
| 3634 |
}; # $insert_to_foster |
}; # $insert_to_foster |
| 3635 |
|
|
| 3636 |
my $insert; |
B: while (1) { |
|
|
|
|
B: { |
|
| 3637 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 3638 |
!!!cp ('t73'); |
!!!cp ('t73'); |
| 3639 |
!!!parse-error (type => 'DOCTYPE in the middle'); |
!!!parse-error (type => 'DOCTYPE in the middle', token => $token); |
| 3640 |
## Ignore the token |
## Ignore the token |
| 3641 |
## Stay in the phase |
## Stay in the phase |
| 3642 |
!!!next-token; |
!!!next-token; |
| 3643 |
redo B; |
next B; |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
|
|
if ($self->{insertion_mode} & AFTER_HTML_IMS) { |
|
|
!!!cp ('t74'); |
|
|
# |
|
|
} else { |
|
|
## Generate implied end tags |
|
|
while ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t75'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} |
|
|
|
|
|
if (@{$self->{open_elements}} > 2 or |
|
|
(@{$self->{open_elements}} == 2 and $self->{open_elements}->[1]->[1] ne 'body')) { |
|
|
!!!cp ('t76'); |
|
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
|
} elsif (defined $self->{inner_html_node} and |
|
|
@{$self->{open_elements}} > 1 and |
|
|
$self->{open_elements}->[1]->[1] ne 'body') { |
|
|
## ISSUE: This case is never reached. |
|
|
!!!cp ('t77'); |
|
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
|
} else { |
|
|
!!!cp ('t78'); |
|
|
} |
|
|
|
|
|
## ISSUE: There is an issue in the spec. |
|
|
} |
|
|
|
|
|
## Stop parsing |
|
|
last B; |
|
| 3644 |
} elsif ($token->{type} == START_TAG_TOKEN and |
} elsif ($token->{type} == START_TAG_TOKEN and |
| 3645 |
$token->{tag_name} eq 'html') { |
$token->{tag_name} eq 'html') { |
| 3646 |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
| 3647 |
!!!cp ('t79'); |
!!!cp ('t79'); |
| 3648 |
!!!parse-error (type => 'after html:html'); |
!!!parse-error (type => 'after html:html', token => $token); |
| 3649 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 3650 |
} elsif ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
} elsif ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
| 3651 |
!!!cp ('t80'); |
!!!cp ('t80'); |
| 3652 |
!!!parse-error (type => 'after html:html'); |
!!!parse-error (type => 'after html:html', token => $token); |
| 3653 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 3654 |
} else { |
} else { |
| 3655 |
!!!cp ('t81'); |
!!!cp ('t81'); |
| 3656 |
} |
} |
| 3657 |
|
|
| 3658 |
!!!cp ('t82'); |
!!!cp ('t82'); |
| 3659 |
!!!parse-error (type => 'not first start tag'); |
!!!parse-error (type => 'not first start tag', token => $token); |
| 3660 |
my $top_el = $self->{open_elements}->[0]->[0]; |
my $top_el = $self->{open_elements}->[0]->[0]; |
| 3661 |
for my $attr_name (keys %{$token->{attributes}}) { |
for my $attr_name (keys %{$token->{attributes}}) { |
| 3662 |
unless ($top_el->has_attribute_ns (undef, $attr_name)) { |
unless ($top_el->has_attribute_ns (undef, $attr_name)) { |
| 3666 |
$token->{attributes}->{$attr_name}->{value}); |
$token->{attributes}->{$attr_name}->{value}); |
| 3667 |
} |
} |
| 3668 |
} |
} |
| 3669 |
|
!!!nack ('t84.1'); |
| 3670 |
!!!next-token; |
!!!next-token; |
| 3671 |
redo B; |
next B; |
| 3672 |
} elsif ($token->{type} == COMMENT_TOKEN) { |
} elsif ($token->{type} == COMMENT_TOKEN) { |
| 3673 |
my $comment = $self->{document}->create_comment ($token->{data}); |
my $comment = $self->{document}->create_comment ($token->{data}); |
| 3674 |
if ($self->{insertion_mode} & AFTER_HTML_IMS) { |
if ($self->{insertion_mode} & AFTER_HTML_IMS) { |
| 3682 |
$self->{open_elements}->[-1]->[0]->append_child ($comment); |
$self->{open_elements}->[-1]->[0]->append_child ($comment); |
| 3683 |
} |
} |
| 3684 |
!!!next-token; |
!!!next-token; |
| 3685 |
redo B; |
next B; |
| 3686 |
} elsif ($self->{insertion_mode} & HEAD_IMS) { |
} elsif ($self->{insertion_mode} & IN_FOREIGN_CONTENT_IM) { |
| 3687 |
|
if ($token->{type} == CHARACTER_TOKEN) { |
| 3688 |
|
!!!cp ('t87.1'); |
| 3689 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 3690 |
|
!!!next-token; |
| 3691 |
|
next B; |
| 3692 |
|
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 3693 |
|
if ((not {mglyph => 1, malignmark => 1}->{$token->{tag_name}} and |
| 3694 |
|
$self->{open_elements}->[-1]->[1] & FOREIGN_FLOW_CONTENT_EL) or |
| 3695 |
|
not ($self->{open_elements}->[-1]->[1] & FOREIGN_EL) or |
| 3696 |
|
($token->{tag_name} eq 'svg' and |
| 3697 |
|
$self->{open_elements}->[-1]->[1] & MML_AXML_EL)) { |
| 3698 |
|
## NOTE: "using the rules for secondary insertion mode"then"continue" |
| 3699 |
|
!!!cp ('t87.2'); |
| 3700 |
|
# |
| 3701 |
|
} elsif ({ |
| 3702 |
|
b => 1, big => 1, blockquote => 1, body => 1, br => 1, |
| 3703 |
|
center => 1, code => 1, dd => 1, div => 1, dl => 1, em => 1, |
| 3704 |
|
embed => 1, font => 1, h1 => 1, h2 => 1, h3 => 1, ## No h4! |
| 3705 |
|
h5 => 1, h6 => 1, head => 1, hr => 1, i => 1, img => 1, |
| 3706 |
|
li => 1, menu => 1, meta => 1, nobr => 1, p => 1, pre => 1, |
| 3707 |
|
ruby => 1, s => 1, small => 1, span => 1, strong => 1, |
| 3708 |
|
sub => 1, sup => 1, table => 1, tt => 1, u => 1, ul => 1, |
| 3709 |
|
var => 1, |
| 3710 |
|
}->{$token->{tag_name}}) { |
| 3711 |
|
!!!cp ('t87.2'); |
| 3712 |
|
!!!parse-error (type => 'not closed', |
| 3713 |
|
value => $self->{open_elements}->[-1]->[0] |
| 3714 |
|
->manakai_local_name, |
| 3715 |
|
token => $token); |
| 3716 |
|
|
| 3717 |
|
pop @{$self->{open_elements}} |
| 3718 |
|
while $self->{open_elements}->[-1]->[1] & FOREIGN_EL; |
| 3719 |
|
|
| 3720 |
|
$self->{insertion_mode} &= ~ IN_FOREIGN_CONTENT_IM; |
| 3721 |
|
## Reprocess. |
| 3722 |
|
next B; |
| 3723 |
|
} else { |
| 3724 |
|
## TODO: case fixup |
| 3725 |
|
|
| 3726 |
|
!!!insert-element-f ($self->{open_elements}->[-1]->[0]->namespace_uri, $token); |
| 3727 |
|
|
| 3728 |
|
if ($self->{self_closing}) { |
| 3729 |
|
pop @{$self->{open_elements}}; |
| 3730 |
|
!!!ack ('t87.3'); |
| 3731 |
|
} else { |
| 3732 |
|
!!!cp ('t87.4'); |
| 3733 |
|
} |
| 3734 |
|
|
| 3735 |
|
!!!next-token; |
| 3736 |
|
next B; |
| 3737 |
|
} |
| 3738 |
|
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 3739 |
|
## NOTE: "using the rules for secondary insertion mode" then "continue" |
| 3740 |
|
!!!cp ('t87.5'); |
| 3741 |
|
# |
| 3742 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 3743 |
|
## NOTE: "using the rules for secondary insertion mode" then "continue" |
| 3744 |
|
!!!cp ('t87.6'); |
| 3745 |
|
# |
| 3746 |
|
## TODO: ... |
| 3747 |
|
} else { |
| 3748 |
|
die "$0: $token->{type}: Unknown token type"; |
| 3749 |
|
} |
| 3750 |
|
} |
| 3751 |
|
|
| 3752 |
|
if ($self->{insertion_mode} & HEAD_IMS) { |
| 3753 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 3754 |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
| 3755 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
unless ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 3756 |
|
!!!cp ('t88.2'); |
| 3757 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
| 3758 |
|
} else { |
| 3759 |
|
!!!cp ('t88.1'); |
| 3760 |
|
## Ignore the token. |
| 3761 |
|
!!!next-token; |
| 3762 |
|
next B; |
| 3763 |
|
} |
| 3764 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 3765 |
!!!cp ('t88'); |
!!!cp ('t88'); |
| 3766 |
!!!next-token; |
!!!next-token; |
| 3767 |
redo B; |
next B; |
| 3768 |
} |
} |
| 3769 |
} |
} |
| 3770 |
|
|
| 3771 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 3772 |
!!!cp ('t89'); |
!!!cp ('t89'); |
| 3773 |
## As if <head> |
## As if <head> |
| 3774 |
!!!create-element ($self->{head_element}, 'head'); |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 3775 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 3776 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3777 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3778 |
|
|
| 3779 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 3780 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 3784 |
!!!cp ('t90'); |
!!!cp ('t90'); |
| 3785 |
## As if </noscript> |
## As if </noscript> |
| 3786 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 3787 |
!!!parse-error (type => 'in noscript:#character'); |
!!!parse-error (type => 'in noscript:#character', token => $token); |
| 3788 |
|
|
| 3789 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 3790 |
## As if </head> |
## As if </head> |
| 3800 |
!!!cp ('t92'); |
!!!cp ('t92'); |
| 3801 |
} |
} |
| 3802 |
|
|
| 3803 |
## "after head" insertion mode |
## "after head" insertion mode |
| 3804 |
## As if <body> |
## As if <body> |
| 3805 |
!!!insert-element ('body'); |
!!!insert-element ('body',, $token); |
| 3806 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 3807 |
## reprocess |
## reprocess |
| 3808 |
redo B; |
next B; |
| 3809 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 3810 |
if ($token->{tag_name} eq 'head') { |
if ($token->{tag_name} eq 'head') { |
| 3811 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 3812 |
!!!cp ('t93'); |
!!!cp ('t93'); |
| 3813 |
!!!create-element ($self->{head_element}, $token->{tag_name}, $token->{attributes}); |
!!!create-element ($self->{head_element}, $HTML_NS, $token->{tag_name}, $token->{attributes}, $token); |
| 3814 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child |
| 3815 |
push @{$self->{open_elements}}, [$self->{head_element}, $token->{tag_name}]; |
($self->{head_element}); |
| 3816 |
$self->{insertion_mode} = IN_HEAD_IM; |
push @{$self->{open_elements}}, |
| 3817 |
!!!next-token; |
[$self->{head_element}, $el_category->{head}]; |
| 3818 |
redo B; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 3819 |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
!!!nack ('t93.1'); |
| 3820 |
!!!cp ('t94'); |
!!!next-token; |
| 3821 |
# |
next B; |
| 3822 |
} else { |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3823 |
!!!cp ('t95'); |
!!!cp ('t94'); |
| 3824 |
!!!parse-error (type => 'in head:head'); # or in head noscript |
# |
| 3825 |
## Ignore the token |
} else { |
| 3826 |
!!!next-token; |
!!!cp ('t95'); |
| 3827 |
redo B; |
!!!parse-error (type => 'in head:head', token => $token); # or in head noscript |
| 3828 |
} |
## Ignore the token |
| 3829 |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
!!!nack ('t95.1'); |
| 3830 |
!!!cp ('t96'); |
!!!next-token; |
| 3831 |
## As if <head> |
next B; |
| 3832 |
!!!create-element ($self->{head_element}, 'head'); |
} |
| 3833 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 3834 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
!!!cp ('t96'); |
| 3835 |
|
## As if <head> |
| 3836 |
|
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 3837 |
|
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 3838 |
|
push @{$self->{open_elements}}, |
| 3839 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3840 |
|
|
| 3841 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 3842 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 3843 |
} else { |
} else { |
| 3844 |
!!!cp ('t97'); |
!!!cp ('t97'); |
| 3845 |
} |
} |
| 3846 |
|
|
| 3847 |
if ($token->{tag_name} eq 'base') { |
if ($token->{tag_name} eq 'base') { |
| 3848 |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 3849 |
!!!cp ('t98'); |
!!!cp ('t98'); |
| 3850 |
## As if </noscript> |
## As if </noscript> |
| 3851 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 3852 |
!!!parse-error (type => 'in noscript:base'); |
!!!parse-error (type => 'in noscript:base', token => $token); |
| 3853 |
|
|
| 3854 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 3855 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 3860 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3861 |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3862 |
!!!cp ('t100'); |
!!!cp ('t100'); |
| 3863 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 3864 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3865 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3866 |
} else { |
} else { |
| 3867 |
!!!cp ('t101'); |
!!!cp ('t101'); |
| 3868 |
} |
} |
| 3869 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 3870 |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 3871 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3872 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3873 |
|
!!!nack ('t101.1'); |
| 3874 |
!!!next-token; |
!!!next-token; |
| 3875 |
redo B; |
next B; |
| 3876 |
} elsif ($token->{tag_name} eq 'link') { |
} elsif ($token->{tag_name} eq 'link') { |
| 3877 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3878 |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3879 |
!!!cp ('t102'); |
!!!cp ('t102'); |
| 3880 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 3881 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3882 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3883 |
} else { |
} else { |
| 3884 |
!!!cp ('t103'); |
!!!cp ('t103'); |
| 3885 |
} |
} |
| 3886 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 3887 |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 3888 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3889 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3890 |
|
!!!ack ('t103.1'); |
| 3891 |
!!!next-token; |
!!!next-token; |
| 3892 |
redo B; |
next B; |
| 3893 |
} elsif ($token->{tag_name} eq 'meta') { |
} elsif ($token->{tag_name} eq 'meta') { |
| 3894 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3895 |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3896 |
!!!cp ('t104'); |
!!!cp ('t104'); |
| 3897 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 3898 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3899 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3900 |
} else { |
} else { |
| 3901 |
!!!cp ('t105'); |
!!!cp ('t105'); |
| 3902 |
} |
} |
| 3903 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 3904 |
my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 3905 |
|
|
| 3906 |
unless ($self->{confident}) { |
unless ($self->{confident}) { |
| 3907 |
if ($token->{attributes}->{charset}) { ## TODO: And if supported |
if ($token->{attributes}->{charset}) { ## TODO: And if supported |
| 3908 |
!!!cp ('t106'); |
!!!cp ('t106'); |
| 3909 |
$self->{change_encoding} |
$self->{change_encoding} |
| 3910 |
->($self, $token->{attributes}->{charset}->{value}); |
->($self, $token->{attributes}->{charset}->{value}, |
| 3911 |
|
$token); |
| 3912 |
|
|
| 3913 |
$meta_el->[0]->get_attribute_node_ns (undef, 'charset') |
$meta_el->[0]->get_attribute_node_ns (undef, 'charset') |
| 3914 |
->set_user_data (manakai_has_reference => |
->set_user_data (manakai_has_reference => |
| 3923 |
([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) { |
([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) { |
| 3924 |
!!!cp ('t107'); |
!!!cp ('t107'); |
| 3925 |
$self->{change_encoding} |
$self->{change_encoding} |
| 3926 |
->($self, defined $1 ? $1 : defined $2 ? $2 : $3); |
->($self, defined $1 ? $1 : defined $2 ? $2 : $3, |
| 3927 |
|
$token); |
| 3928 |
$meta_el->[0]->get_attribute_node_ns (undef, 'content') |
$meta_el->[0]->get_attribute_node_ns (undef, 'content') |
| 3929 |
->set_user_data (manakai_has_reference => |
->set_user_data (manakai_has_reference => |
| 3930 |
$token->{attributes}->{content} |
$token->{attributes}->{content} |
| 3950 |
} |
} |
| 3951 |
} |
} |
| 3952 |
|
|
| 3953 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3954 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3955 |
|
!!!ack ('t110.1'); |
| 3956 |
!!!next-token; |
!!!next-token; |
| 3957 |
redo B; |
next B; |
| 3958 |
} elsif ($token->{tag_name} eq 'title') { |
} elsif ($token->{tag_name} eq 'title') { |
| 3959 |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 3960 |
!!!cp ('t111'); |
!!!cp ('t111'); |
| 3961 |
## As if </noscript> |
## As if </noscript> |
| 3962 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 3963 |
!!!parse-error (type => 'in noscript:title'); |
!!!parse-error (type => 'in noscript:title', token => $token); |
| 3964 |
|
|
| 3965 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 3966 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 3967 |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3968 |
!!!cp ('t112'); |
!!!cp ('t112'); |
| 3969 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 3970 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3971 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3972 |
} else { |
} else { |
| 3973 |
!!!cp ('t113'); |
!!!cp ('t113'); |
| 3974 |
} |
} |
| 3976 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3977 |
my $parent = defined $self->{head_element} ? $self->{head_element} |
my $parent = defined $self->{head_element} ? $self->{head_element} |
| 3978 |
: $self->{open_elements}->[-1]->[0]; |
: $self->{open_elements}->[-1]->[0]; |
| 3979 |
$parse_rcdata->(RCDATA_CONTENT_MODEL, |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
| 3980 |
sub { $parent->append_child ($_[0]) }); |
pop @{$self->{open_elements}} # <head> |
|
pop @{$self->{open_elements}} |
|
| 3981 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3982 |
redo B; |
next B; |
| 3983 |
} elsif ($token->{tag_name} eq 'style') { |
} elsif ($token->{tag_name} eq 'style') { |
| 3984 |
## NOTE: Or (scripting is enabled and tag_name eq 'noscript' and |
## NOTE: Or (scripting is enabled and tag_name eq 'noscript' and |
| 3985 |
## insertion mode IN_HEAD_IM) |
## insertion mode IN_HEAD_IM) |
| 3986 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3987 |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
if ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 3988 |
!!!cp ('t114'); |
!!!cp ('t114'); |
| 3989 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 3990 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 3991 |
|
[$self->{head_element}, $el_category->{head}]; |
| 3992 |
} else { |
} else { |
| 3993 |
!!!cp ('t115'); |
!!!cp ('t115'); |
| 3994 |
} |
} |
| 3995 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert_to_current); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 3996 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3997 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3998 |
redo B; |
next B; |
| 3999 |
} elsif ($token->{tag_name} eq 'noscript') { |
} elsif ($token->{tag_name} eq 'noscript') { |
| 4000 |
if ($self->{insertion_mode} == IN_HEAD_IM) { |
if ($self->{insertion_mode} == IN_HEAD_IM) { |
| 4001 |
!!!cp ('t116'); |
!!!cp ('t116'); |
| 4002 |
## NOTE: and scripting is disalbed |
## NOTE: and scripting is disalbed |
| 4003 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 4004 |
$self->{insertion_mode} = IN_HEAD_NOSCRIPT_IM; |
$self->{insertion_mode} = IN_HEAD_NOSCRIPT_IM; |
| 4005 |
|
!!!nack ('t116.1'); |
| 4006 |
!!!next-token; |
!!!next-token; |
| 4007 |
redo B; |
next B; |
| 4008 |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 4009 |
!!!cp ('t117'); |
!!!cp ('t117'); |
| 4010 |
!!!parse-error (type => 'in noscript:noscript'); |
!!!parse-error (type => 'in noscript:noscript', token => $token); |
| 4011 |
## Ignore the token |
## Ignore the token |
| 4012 |
|
!!!nack ('t117.1'); |
| 4013 |
!!!next-token; |
!!!next-token; |
| 4014 |
redo B; |
next B; |
| 4015 |
} else { |
} else { |
| 4016 |
!!!cp ('t118'); |
!!!cp ('t118'); |
| 4017 |
# |
# |
| 4021 |
!!!cp ('t119'); |
!!!cp ('t119'); |
| 4022 |
## As if </noscript> |
## As if </noscript> |
| 4023 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4024 |
!!!parse-error (type => 'in noscript:script'); |
!!!parse-error (type => 'in noscript:script', token => $token); |
| 4025 |
|
|
| 4026 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 4027 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4028 |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 4029 |
!!!cp ('t120'); |
!!!cp ('t120'); |
| 4030 |
!!!parse-error (type => 'after head:'.$token->{tag_name}); |
!!!parse-error (type => 'after head:'.$token->{tag_name}, token => $token); |
| 4031 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 4032 |
|
[$self->{head_element}, $el_category->{head}]; |
| 4033 |
} else { |
} else { |
| 4034 |
!!!cp ('t121'); |
!!!cp ('t121'); |
| 4035 |
} |
} |
| 4036 |
|
|
| 4037 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 4038 |
$script_start_tag->($insert_to_current); |
$script_start_tag->(); |
| 4039 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 4040 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 4041 |
redo B; |
next B; |
| 4042 |
} elsif ($token->{tag_name} eq 'body' or |
} elsif ($token->{tag_name} eq 'body' or |
| 4043 |
$token->{tag_name} eq 'frameset') { |
$token->{tag_name} eq 'frameset') { |
| 4044 |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 4045 |
!!!cp ('t122'); |
!!!cp ('t122'); |
| 4046 |
## As if </noscript> |
## As if </noscript> |
| 4047 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4048 |
!!!parse-error (type => 'in noscript:'.$token->{tag_name}); |
!!!parse-error (type => 'in noscript:'.$token->{tag_name}, token => $token); |
| 4049 |
|
|
| 4050 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4051 |
## As if </head> |
## As if </head> |
| 4062 |
} |
} |
| 4063 |
|
|
| 4064 |
## "after head" insertion mode |
## "after head" insertion mode |
| 4065 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 4066 |
if ($token->{tag_name} eq 'body') { |
if ($token->{tag_name} eq 'body') { |
| 4067 |
!!!cp ('t126'); |
!!!cp ('t126'); |
| 4068 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 4072 |
} else { |
} else { |
| 4073 |
die "$0: tag name: $self->{tag_name}"; |
die "$0: tag name: $self->{tag_name}"; |
| 4074 |
} |
} |
| 4075 |
|
!!!nack ('t127.1'); |
| 4076 |
!!!next-token; |
!!!next-token; |
| 4077 |
redo B; |
next B; |
| 4078 |
} else { |
} else { |
| 4079 |
!!!cp ('t128'); |
!!!cp ('t128'); |
| 4080 |
# |
# |
| 4084 |
!!!cp ('t129'); |
!!!cp ('t129'); |
| 4085 |
## As if </noscript> |
## As if </noscript> |
| 4086 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4087 |
!!!parse-error (type => 'in noscript:/'.$token->{tag_name}); |
!!!parse-error (type => 'in noscript:/'.$token->{tag_name}, token => $token); |
| 4088 |
|
|
| 4089 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4090 |
## As if </head> |
## As if </head> |
| 4103 |
|
|
| 4104 |
## "after head" insertion mode |
## "after head" insertion mode |
| 4105 |
## As if <body> |
## As if <body> |
| 4106 |
!!!insert-element ('body'); |
!!!insert-element ('body',, $token); |
| 4107 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 4108 |
## reprocess |
## reprocess |
| 4109 |
redo B; |
!!!ack-later; |
| 4110 |
|
next B; |
| 4111 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 4112 |
if ($token->{tag_name} eq 'head') { |
if ($token->{tag_name} eq 'head') { |
| 4113 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4114 |
!!!cp ('t132'); |
!!!cp ('t132'); |
| 4115 |
## As if <head> |
## As if <head> |
| 4116 |
!!!create-element ($self->{head_element}, 'head'); |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 4117 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 4118 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 4119 |
|
[$self->{head_element}, $el_category->{head}]; |
| 4120 |
|
|
| 4121 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4122 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4123 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 4124 |
!!!next-token; |
!!!next-token; |
| 4125 |
redo B; |
next B; |
| 4126 |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 4127 |
!!!cp ('t133'); |
!!!cp ('t133'); |
| 4128 |
## As if </noscript> |
## As if </noscript> |
| 4129 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4130 |
!!!parse-error (type => 'in noscript:/head'); |
!!!parse-error (type => 'in noscript:/head', token => $token); |
| 4131 |
|
|
| 4132 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4133 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4134 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 4135 |
!!!next-token; |
!!!next-token; |
| 4136 |
redo B; |
next B; |
| 4137 |
} elsif ($self->{insertion_mode} == IN_HEAD_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_IM) { |
| 4138 |
!!!cp ('t134'); |
!!!cp ('t134'); |
| 4139 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4140 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 4141 |
!!!next-token; |
!!!next-token; |
| 4142 |
redo B; |
next B; |
| 4143 |
} else { |
} else { |
| 4144 |
!!!cp ('t135'); |
!!!cp ('t135'); |
| 4145 |
# |
# |
| 4150 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4151 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 4152 |
!!!next-token; |
!!!next-token; |
| 4153 |
redo B; |
next B; |
| 4154 |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4155 |
!!!cp ('t137'); |
!!!cp ('t137'); |
| 4156 |
!!!parse-error (type => 'unmatched end tag:noscript'); |
!!!parse-error (type => 'unmatched end tag:noscript', token => $token); |
| 4157 |
## Ignore the token ## ISSUE: An issue in the spec. |
## Ignore the token ## ISSUE: An issue in the spec. |
| 4158 |
!!!next-token; |
!!!next-token; |
| 4159 |
redo B; |
next B; |
| 4160 |
} else { |
} else { |
| 4161 |
!!!cp ('t138'); |
!!!cp ('t138'); |
| 4162 |
# |
# |
| 4167 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4168 |
!!!cp ('t139'); |
!!!cp ('t139'); |
| 4169 |
## As if <head> |
## As if <head> |
| 4170 |
!!!create-element ($self->{head_element}, 'head'); |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 4171 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 4172 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 4173 |
|
[$self->{head_element}, $el_category->{head}]; |
| 4174 |
|
|
| 4175 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 4176 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4177 |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 4178 |
!!!cp ('t140'); |
!!!cp ('t140'); |
| 4179 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4180 |
## Ignore the token |
## Ignore the token |
| 4181 |
!!!next-token; |
!!!next-token; |
| 4182 |
redo B; |
next B; |
| 4183 |
} else { |
} else { |
| 4184 |
!!!cp ('t141'); |
!!!cp ('t141'); |
| 4185 |
} |
} |
| 4191 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4192 |
!!!cp ('t142'); |
!!!cp ('t142'); |
| 4193 |
## As if <head> |
## As if <head> |
| 4194 |
!!!create-element ($self->{head_element}, 'head'); |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 4195 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 4196 |
push @{$self->{open_elements}}, [$self->{head_element}, 'head']; |
push @{$self->{open_elements}}, |
| 4197 |
|
[$self->{head_element}, $el_category->{head}]; |
| 4198 |
|
|
| 4199 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 4200 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4209 |
# |
# |
| 4210 |
} else { |
} else { |
| 4211 |
!!!cp ('t145'); |
!!!cp ('t145'); |
| 4212 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4213 |
## Ignore the token |
## Ignore the token |
| 4214 |
!!!next-token; |
!!!next-token; |
| 4215 |
redo B; |
next B; |
| 4216 |
} |
} |
| 4217 |
} |
} |
| 4218 |
|
|
| 4220 |
!!!cp ('t146'); |
!!!cp ('t146'); |
| 4221 |
## As if </noscript> |
## As if </noscript> |
| 4222 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4223 |
!!!parse-error (type => 'in noscript:/'.$token->{tag_name}); |
!!!parse-error (type => 'in noscript:/'.$token->{tag_name}, token => $token); |
| 4224 |
|
|
| 4225 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 4226 |
## As if </head> |
## As if </head> |
| 4236 |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
} elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4237 |
## ISSUE: This case cannot be reached? |
## ISSUE: This case cannot be reached? |
| 4238 |
!!!cp ('t148'); |
!!!cp ('t148'); |
| 4239 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4240 |
## Ignore the token ## ISSUE: An issue in the spec. |
## Ignore the token ## ISSUE: An issue in the spec. |
| 4241 |
!!!next-token; |
!!!next-token; |
| 4242 |
redo B; |
next B; |
| 4243 |
} else { |
} else { |
| 4244 |
!!!cp ('t149'); |
!!!cp ('t149'); |
| 4245 |
} |
} |
| 4246 |
|
|
| 4247 |
## "after head" insertion mode |
## "after head" insertion mode |
| 4248 |
## As if <body> |
## As if <body> |
| 4249 |
!!!insert-element ('body'); |
!!!insert-element ('body',, $token); |
| 4250 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 4251 |
## reprocess |
## reprocess |
| 4252 |
redo B; |
next B; |
| 4253 |
} else { |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 4254 |
die "$0: $token->{type}: Unknown token type"; |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 4255 |
} |
!!!cp ('t149.1'); |
| 4256 |
|
|
| 4257 |
|
## NOTE: As if <head> |
| 4258 |
|
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 4259 |
|
$self->{open_elements}->[-1]->[0]->append_child |
| 4260 |
|
($self->{head_element}); |
| 4261 |
|
#push @{$self->{open_elements}}, |
| 4262 |
|
# [$self->{head_element}, $el_category->{head}]; |
| 4263 |
|
#$self->{insertion_mode} = IN_HEAD_IM; |
| 4264 |
|
## NOTE: Reprocess. |
| 4265 |
|
|
| 4266 |
|
## NOTE: As if </head> |
| 4267 |
|
#pop @{$self->{open_elements}}; |
| 4268 |
|
#$self->{insertion_mode} = IN_AFTER_HEAD_IM; |
| 4269 |
|
## NOTE: Reprocess. |
| 4270 |
|
|
| 4271 |
|
# |
| 4272 |
|
} elsif ($self->{insertion_mode} == IN_HEAD_IM) { |
| 4273 |
|
!!!cp ('t149.2'); |
| 4274 |
|
|
| 4275 |
|
## NOTE: As if </head> |
| 4276 |
|
pop @{$self->{open_elements}}; |
| 4277 |
|
#$self->{insertion_mode} = IN_AFTER_HEAD_IM; |
| 4278 |
|
## NOTE: Reprocess. |
| 4279 |
|
|
| 4280 |
|
# |
| 4281 |
|
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 4282 |
|
!!!cp ('t149.3'); |
| 4283 |
|
|
| 4284 |
|
!!!parse-error (type => 'in noscript:#eof', token => $token); |
| 4285 |
|
|
| 4286 |
|
## As if </noscript> |
| 4287 |
|
pop @{$self->{open_elements}}; |
| 4288 |
|
#$self->{insertion_mode} = IN_HEAD_IM; |
| 4289 |
|
## NOTE: Reprocess. |
| 4290 |
|
|
| 4291 |
|
## NOTE: As if </head> |
| 4292 |
|
pop @{$self->{open_elements}}; |
| 4293 |
|
#$self->{insertion_mode} = IN_AFTER_HEAD_IM; |
| 4294 |
|
## NOTE: Reprocess. |
| 4295 |
|
|
| 4296 |
|
# |
| 4297 |
|
} else { |
| 4298 |
|
!!!cp ('t149.4'); |
| 4299 |
|
# |
| 4300 |
|
} |
| 4301 |
|
|
| 4302 |
|
## NOTE: As if <body> |
| 4303 |
|
!!!insert-element ('body',, $token); |
| 4304 |
|
$self->{insertion_mode} = IN_BODY_IM; |
| 4305 |
|
## NOTE: Reprocess. |
| 4306 |
|
next B; |
| 4307 |
|
} else { |
| 4308 |
|
die "$0: $token->{type}: Unknown token type"; |
| 4309 |
|
} |
| 4310 |
|
|
| 4311 |
## ISSUE: An issue in the spec. |
## ISSUE: An issue in the spec. |
| 4312 |
} elsif ($self->{insertion_mode} & BODY_IMS) { |
} elsif ($self->{insertion_mode} & BODY_IMS) { |
| 4318 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 4319 |
|
|
| 4320 |
!!!next-token; |
!!!next-token; |
| 4321 |
redo B; |
next B; |
| 4322 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 4323 |
if ({ |
if ({ |
| 4324 |
caption => 1, col => 1, colgroup => 1, tbody => 1, |
caption => 1, col => 1, colgroup => 1, tbody => 1, |
| 4326 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4327 |
if ($self->{insertion_mode} == IN_CELL_IM) { |
if ($self->{insertion_mode} == IN_CELL_IM) { |
| 4328 |
## have an element in table scope |
## have an element in table scope |
| 4329 |
my $tn; |
for (reverse 0..$#{$self->{open_elements}}) { |
|
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
|
| 4330 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4331 |
if ($node->[1] eq 'td' or $node->[1] eq 'th') { |
if ($node->[1] & TABLE_CELL_EL) { |
| 4332 |
!!!cp ('t151'); |
!!!cp ('t151'); |
| 4333 |
$tn = $node->[1]; |
|
| 4334 |
last INSCOPE; |
## Close the cell |
| 4335 |
} elsif ({ |
!!!back-token; # <x> |
| 4336 |
table => 1, html => 1, |
$token = {type => END_TAG_TOKEN, |
| 4337 |
}->{$node->[1]}) { |
tag_name => $node->[0]->manakai_local_name, |
| 4338 |
|
line => $token->{line}, |
| 4339 |
|
column => $token->{column}}; |
| 4340 |
|
next B; |
| 4341 |
|
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 4342 |
!!!cp ('t152'); |
!!!cp ('t152'); |
| 4343 |
last INSCOPE; |
## ISSUE: This case can never be reached, maybe. |
| 4344 |
|
last; |
| 4345 |
} |
} |
| 4346 |
} # INSCOPE |
} |
| 4347 |
unless (defined $tn) { |
|
| 4348 |
!!!cp ('t153'); |
!!!cp ('t153'); |
| 4349 |
## TODO: This error type is wrong. |
!!!parse-error (type => 'start tag not allowed', |
| 4350 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
value => $token->{tag_name}, token => $token); |
| 4351 |
## Ignore the token |
## Ignore the token |
| 4352 |
!!!next-token; |
!!!nack ('t153.1'); |
| 4353 |
redo B; |
!!!next-token; |
| 4354 |
} |
next B; |
|
|
|
|
!!!cp ('t154'); |
|
|
## Close the cell |
|
|
!!!back-token; # <?> |
|
|
$token = {type => END_TAG_TOKEN, tag_name => $tn}; |
|
|
redo B; |
|
| 4355 |
} elsif ($self->{insertion_mode} == IN_CAPTION_IM) { |
} elsif ($self->{insertion_mode} == IN_CAPTION_IM) { |
| 4356 |
!!!parse-error (type => 'not closed:caption'); |
!!!parse-error (type => 'not closed:caption', token => $token); |
| 4357 |
|
|
| 4358 |
## As if </caption> |
## NOTE: As if </caption>. |
| 4359 |
## have a table element in table scope |
## have a table element in table scope |
| 4360 |
my $i; |
my $i; |
| 4361 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: { |
| 4362 |
my $node = $self->{open_elements}->[$_]; |
for (reverse 0..$#{$self->{open_elements}}) { |
| 4363 |
if ($node->[1] eq 'caption') { |
my $node = $self->{open_elements}->[$_]; |
| 4364 |
!!!cp ('t155'); |
if ($node->[1] & CAPTION_EL) { |
| 4365 |
$i = $_; |
!!!cp ('t155'); |
| 4366 |
last INSCOPE; |
$i = $_; |
| 4367 |
} elsif ({ |
last INSCOPE; |
| 4368 |
table => 1, html => 1, |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 4369 |
}->{$node->[1]}) { |
!!!cp ('t156'); |
| 4370 |
!!!cp ('t156'); |
last; |
| 4371 |
last INSCOPE; |
} |
| 4372 |
} |
} |
| 4373 |
|
|
| 4374 |
|
!!!cp ('t157'); |
| 4375 |
|
!!!parse-error (type => 'start tag not allowed', |
| 4376 |
|
value => $token->{tag_name}, token => $token); |
| 4377 |
|
## Ignore the token |
| 4378 |
|
!!!nack ('t157.1'); |
| 4379 |
|
!!!next-token; |
| 4380 |
|
next B; |
| 4381 |
} # INSCOPE |
} # INSCOPE |
|
unless (defined $i) { |
|
|
!!!cp ('t157'); |
|
|
## TODO: this type is wrong. |
|
|
!!!parse-error (type => 'unmatched end tag:caption'); |
|
|
## Ignore the token |
|
|
!!!next-token; |
|
|
redo B; |
|
|
} |
|
| 4382 |
|
|
| 4383 |
## generate implied end tags |
## generate implied end tags |
| 4384 |
while ({ |
while ($self->{open_elements}->[-1]->[1] |
| 4385 |
dd => 1, dt => 1, li => 1, p => 1, |
& END_TAG_OPTIONAL_EL) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4386 |
!!!cp ('t158'); |
!!!cp ('t158'); |
| 4387 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4388 |
} |
} |
| 4389 |
|
|
| 4390 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
unless ($self->{open_elements}->[-1]->[1] & CAPTION_EL) { |
| 4391 |
!!!cp ('t159'); |
!!!cp ('t159'); |
| 4392 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4393 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4394 |
|
->manakai_local_name, |
| 4395 |
|
token => $token); |
| 4396 |
} else { |
} else { |
| 4397 |
!!!cp ('t160'); |
!!!cp ('t160'); |
| 4398 |
} |
} |
| 4404 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 4405 |
|
|
| 4406 |
## reprocess |
## reprocess |
| 4407 |
redo B; |
!!!ack-later; |
| 4408 |
|
next B; |
| 4409 |
} else { |
} else { |
| 4410 |
!!!cp ('t161'); |
!!!cp ('t161'); |
| 4411 |
# |
# |
| 4421 |
my $i; |
my $i; |
| 4422 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4423 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4424 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 4425 |
!!!cp ('t163'); |
!!!cp ('t163'); |
| 4426 |
$i = $_; |
$i = $_; |
| 4427 |
last INSCOPE; |
last INSCOPE; |
| 4428 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 4429 |
!!!cp ('t164'); |
!!!cp ('t164'); |
| 4430 |
last INSCOPE; |
last INSCOPE; |
| 4431 |
} |
} |
| 4432 |
} # INSCOPE |
} # INSCOPE |
| 4433 |
unless (defined $i) { |
unless (defined $i) { |
| 4434 |
!!!cp ('t165'); |
!!!cp ('t165'); |
| 4435 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4436 |
## Ignore the token |
## Ignore the token |
| 4437 |
!!!next-token; |
!!!next-token; |
| 4438 |
redo B; |
next B; |
| 4439 |
} |
} |
| 4440 |
|
|
| 4441 |
## generate implied end tags |
## generate implied end tags |
| 4442 |
while ({ |
while ($self->{open_elements}->[-1]->[1] |
| 4443 |
dd => 1, dt => 1, li => 1, p => 1, |
& END_TAG_OPTIONAL_EL) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4444 |
!!!cp ('t166'); |
!!!cp ('t166'); |
| 4445 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4446 |
} |
} |
| 4447 |
|
|
| 4448 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 4449 |
|
ne $token->{tag_name}) { |
| 4450 |
!!!cp ('t167'); |
!!!cp ('t167'); |
| 4451 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4452 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4453 |
|
->manakai_local_name, |
| 4454 |
|
token => $token); |
| 4455 |
} else { |
} else { |
| 4456 |
!!!cp ('t168'); |
!!!cp ('t168'); |
| 4457 |
} |
} |
| 4463 |
$self->{insertion_mode} = IN_ROW_IM; |
$self->{insertion_mode} = IN_ROW_IM; |
| 4464 |
|
|
| 4465 |
!!!next-token; |
!!!next-token; |
| 4466 |
redo B; |
next B; |
| 4467 |
} elsif ($self->{insertion_mode} == IN_CAPTION_IM) { |
} elsif ($self->{insertion_mode} == IN_CAPTION_IM) { |
| 4468 |
!!!cp ('t169'); |
!!!cp ('t169'); |
| 4469 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4470 |
## Ignore the token |
## Ignore the token |
| 4471 |
!!!next-token; |
!!!next-token; |
| 4472 |
redo B; |
next B; |
| 4473 |
} else { |
} else { |
| 4474 |
!!!cp ('t170'); |
!!!cp ('t170'); |
| 4475 |
# |
# |
| 4478 |
if ($self->{insertion_mode} == IN_CAPTION_IM) { |
if ($self->{insertion_mode} == IN_CAPTION_IM) { |
| 4479 |
## have a table element in table scope |
## have a table element in table scope |
| 4480 |
my $i; |
my $i; |
| 4481 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: { |
| 4482 |
my $node = $self->{open_elements}->[$_]; |
for (reverse 0..$#{$self->{open_elements}}) { |
| 4483 |
if ($node->[1] eq $token->{tag_name}) { |
my $node = $self->{open_elements}->[$_]; |
| 4484 |
!!!cp ('t171'); |
if ($node->[1] & CAPTION_EL) { |
| 4485 |
$i = $_; |
!!!cp ('t171'); |
| 4486 |
last INSCOPE; |
$i = $_; |
| 4487 |
} elsif ({ |
last INSCOPE; |
| 4488 |
table => 1, html => 1, |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 4489 |
}->{$node->[1]}) { |
!!!cp ('t172'); |
| 4490 |
!!!cp ('t172'); |
last; |
| 4491 |
last INSCOPE; |
} |
| 4492 |
} |
} |
| 4493 |
|
|
| 4494 |
|
!!!cp ('t173'); |
| 4495 |
|
!!!parse-error (type => 'unmatched end tag', |
| 4496 |
|
value => $token->{tag_name}, token => $token); |
| 4497 |
|
## Ignore the token |
| 4498 |
|
!!!next-token; |
| 4499 |
|
next B; |
| 4500 |
} # INSCOPE |
} # INSCOPE |
|
unless (defined $i) { |
|
|
!!!cp ('t173'); |
|
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
|
|
## Ignore the token |
|
|
!!!next-token; |
|
|
redo B; |
|
|
} |
|
| 4501 |
|
|
| 4502 |
## generate implied end tags |
## generate implied end tags |
| 4503 |
while ({ |
while ($self->{open_elements}->[-1]->[1] |
| 4504 |
dd => 1, dt => 1, li => 1, p => 1, |
& END_TAG_OPTIONAL_EL) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4505 |
!!!cp ('t174'); |
!!!cp ('t174'); |
| 4506 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4507 |
} |
} |
| 4508 |
|
|
| 4509 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
unless ($self->{open_elements}->[-1]->[1] & CAPTION_EL) { |
| 4510 |
!!!cp ('t175'); |
!!!cp ('t175'); |
| 4511 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4512 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4513 |
|
->manakai_local_name, |
| 4514 |
|
token => $token); |
| 4515 |
} else { |
} else { |
| 4516 |
!!!cp ('t176'); |
!!!cp ('t176'); |
| 4517 |
} |
} |
| 4523 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 4524 |
|
|
| 4525 |
!!!next-token; |
!!!next-token; |
| 4526 |
redo B; |
next B; |
| 4527 |
} elsif ($self->{insertion_mode} == IN_CELL_IM) { |
} elsif ($self->{insertion_mode} == IN_CELL_IM) { |
| 4528 |
!!!cp ('t177'); |
!!!cp ('t177'); |
| 4529 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4530 |
## Ignore the token |
## Ignore the token |
| 4531 |
!!!next-token; |
!!!next-token; |
| 4532 |
redo B; |
next B; |
| 4533 |
} else { |
} else { |
| 4534 |
!!!cp ('t178'); |
!!!cp ('t178'); |
| 4535 |
# |
# |
| 4542 |
## have an element in table scope |
## have an element in table scope |
| 4543 |
my $i; |
my $i; |
| 4544 |
my $tn; |
my $tn; |
| 4545 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: { |
| 4546 |
my $node = $self->{open_elements}->[$_]; |
for (reverse 0..$#{$self->{open_elements}}) { |
| 4547 |
if ($node->[1] eq $token->{tag_name}) { |
my $node = $self->{open_elements}->[$_]; |
| 4548 |
!!!cp ('t179'); |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 4549 |
$i = $_; |
!!!cp ('t179'); |
| 4550 |
last INSCOPE; |
$i = $_; |
| 4551 |
} elsif ($node->[1] eq 'td' or $node->[1] eq 'th') { |
|
| 4552 |
!!!cp ('t180'); |
## Close the cell |
| 4553 |
$tn = $node->[1]; |
!!!back-token; # </x> |
| 4554 |
## NOTE: There is exactly one |td| or |th| element |
$token = {type => END_TAG_TOKEN, tag_name => $tn, |
| 4555 |
## in scope in the stack of open elements by definition. |
line => $token->{line}, |
| 4556 |
} elsif ({ |
column => $token->{column}}; |
| 4557 |
table => 1, html => 1, |
next B; |
| 4558 |
}->{$node->[1]}) { |
} elsif ($node->[1] & TABLE_CELL_EL) { |
| 4559 |
!!!cp ('t181'); |
!!!cp ('t180'); |
| 4560 |
last INSCOPE; |
$tn = $node->[0]->manakai_local_name; |
| 4561 |
|
## NOTE: There is exactly one |td| or |th| element |
| 4562 |
|
## in scope in the stack of open elements by definition. |
| 4563 |
|
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 4564 |
|
## ISSUE: Can this be reached? |
| 4565 |
|
!!!cp ('t181'); |
| 4566 |
|
last; |
| 4567 |
|
} |
| 4568 |
} |
} |
| 4569 |
} # INSCOPE |
|
|
unless (defined $i) { |
|
| 4570 |
!!!cp ('t182'); |
!!!cp ('t182'); |
| 4571 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag', |
| 4572 |
|
value => $token->{tag_name}, token => $token); |
| 4573 |
## Ignore the token |
## Ignore the token |
| 4574 |
!!!next-token; |
!!!next-token; |
| 4575 |
redo B; |
next B; |
| 4576 |
} else { |
} # INSCOPE |
|
!!!cp ('t183'); |
|
|
} |
|
|
|
|
|
## Close the cell |
|
|
!!!back-token; # </?> |
|
|
$token = {type => END_TAG_TOKEN, tag_name => $tn}; |
|
|
redo B; |
|
| 4577 |
} elsif ($token->{tag_name} eq 'table' and |
} elsif ($token->{tag_name} eq 'table' and |
| 4578 |
$self->{insertion_mode} == IN_CAPTION_IM) { |
$self->{insertion_mode} == IN_CAPTION_IM) { |
| 4579 |
!!!parse-error (type => 'not closed:caption'); |
!!!parse-error (type => 'not closed:caption', token => $token); |
| 4580 |
|
|
| 4581 |
## As if </caption> |
## As if </caption> |
| 4582 |
## have a table element in table scope |
## have a table element in table scope |
| 4583 |
my $i; |
my $i; |
| 4584 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4585 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4586 |
if ($node->[1] eq 'caption') { |
if ($node->[1] & CAPTION_EL) { |
| 4587 |
!!!cp ('t184'); |
!!!cp ('t184'); |
| 4588 |
$i = $_; |
$i = $_; |
| 4589 |
last INSCOPE; |
last INSCOPE; |
| 4590 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 4591 |
!!!cp ('t185'); |
!!!cp ('t185'); |
| 4592 |
last INSCOPE; |
last INSCOPE; |
| 4593 |
} |
} |
| 4594 |
} # INSCOPE |
} # INSCOPE |
| 4595 |
unless (defined $i) { |
unless (defined $i) { |
| 4596 |
!!!cp ('t186'); |
!!!cp ('t186'); |
| 4597 |
!!!parse-error (type => 'unmatched end tag:caption'); |
!!!parse-error (type => 'unmatched end tag:caption', token => $token); |
| 4598 |
## Ignore the token |
## Ignore the token |
| 4599 |
!!!next-token; |
!!!next-token; |
| 4600 |
redo B; |
next B; |
| 4601 |
} |
} |
| 4602 |
|
|
| 4603 |
## generate implied end tags |
## generate implied end tags |
| 4604 |
while ({ |
while ($self->{open_elements}->[-1]->[1] & END_TAG_OPTIONAL_EL) { |
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4605 |
!!!cp ('t187'); |
!!!cp ('t187'); |
| 4606 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4607 |
} |
} |
| 4608 |
|
|
| 4609 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
unless ($self->{open_elements}->[-1]->[1] & CAPTION_EL) { |
| 4610 |
!!!cp ('t188'); |
!!!cp ('t188'); |
| 4611 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4612 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4613 |
|
->manakai_local_name, |
| 4614 |
|
token => $token); |
| 4615 |
} else { |
} else { |
| 4616 |
!!!cp ('t189'); |
!!!cp ('t189'); |
| 4617 |
} |
} |
| 4623 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 4624 |
|
|
| 4625 |
## reprocess |
## reprocess |
| 4626 |
redo B; |
next B; |
| 4627 |
} elsif ({ |
} elsif ({ |
| 4628 |
body => 1, col => 1, colgroup => 1, html => 1, |
body => 1, col => 1, colgroup => 1, html => 1, |
| 4629 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4630 |
if ($self->{insertion_mode} & BODY_TABLE_IMS) { |
if ($self->{insertion_mode} & BODY_TABLE_IMS) { |
| 4631 |
!!!cp ('t190'); |
!!!cp ('t190'); |
| 4632 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4633 |
## Ignore the token |
## Ignore the token |
| 4634 |
!!!next-token; |
!!!next-token; |
| 4635 |
redo B; |
next B; |
| 4636 |
} else { |
} else { |
| 4637 |
!!!cp ('t191'); |
!!!cp ('t191'); |
| 4638 |
# |
# |
| 4643 |
}->{$token->{tag_name}} and |
}->{$token->{tag_name}} and |
| 4644 |
$self->{insertion_mode} == IN_CAPTION_IM) { |
$self->{insertion_mode} == IN_CAPTION_IM) { |
| 4645 |
!!!cp ('t192'); |
!!!cp ('t192'); |
| 4646 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4647 |
## Ignore the token |
## Ignore the token |
| 4648 |
!!!next-token; |
!!!next-token; |
| 4649 |
redo B; |
next B; |
| 4650 |
} else { |
} else { |
| 4651 |
!!!cp ('t193'); |
!!!cp ('t193'); |
| 4652 |
# |
# |
| 4653 |
} |
} |
| 4654 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 4655 |
|
for my $entry (@{$self->{open_elements}}) { |
| 4656 |
|
unless ($entry->[1] & ALL_END_TAG_OPTIONAL_EL) { |
| 4657 |
|
!!!cp ('t75'); |
| 4658 |
|
!!!parse-error (type => 'in body:#eof', token => $token); |
| 4659 |
|
last; |
| 4660 |
|
} |
| 4661 |
|
} |
| 4662 |
|
|
| 4663 |
|
## Stop parsing. |
| 4664 |
|
last B; |
| 4665 |
} else { |
} else { |
| 4666 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 4667 |
} |
} |
| 4670 |
# |
# |
| 4671 |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
| 4672 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 4673 |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
if (not $open_tables->[-1]->[1] and # tainted |
| 4674 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
$token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
| 4675 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
| 4676 |
|
|
| 4677 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 4678 |
!!!cp ('t194'); |
!!!cp ('t194'); |
| 4679 |
!!!next-token; |
!!!next-token; |
| 4680 |
redo B; |
next B; |
| 4681 |
} else { |
} else { |
| 4682 |
!!!cp ('t195'); |
!!!cp ('t195'); |
| 4683 |
} |
} |
| 4684 |
} |
} |
| 4685 |
|
|
| 4686 |
!!!parse-error (type => 'in table:#character'); |
!!!parse-error (type => 'in table:#character', token => $token); |
| 4687 |
|
|
| 4688 |
## As if in body, but insert into foster parent element |
## As if in body, but insert into foster parent element |
| 4689 |
## ISSUE: Spec says that "whenever a node would be inserted |
## ISSUE: Spec says that "whenever a node would be inserted |
| 4691 |
## result in a new Text node. |
## result in a new Text node. |
| 4692 |
$reconstruct_active_formatting_elements->($insert_to_foster); |
$reconstruct_active_formatting_elements->($insert_to_foster); |
| 4693 |
|
|
| 4694 |
if ({ |
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
|
table => 1, tbody => 1, tfoot => 1, |
|
|
thead => 1, tr => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4695 |
# MUST |
# MUST |
| 4696 |
my $foster_parent_element; |
my $foster_parent_element; |
| 4697 |
my $next_sibling; |
my $next_sibling; |
| 4698 |
my $prev_sibling; |
my $prev_sibling; |
| 4699 |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4700 |
if ($self->{open_elements}->[$_]->[1] eq 'table') { |
if ($self->{open_elements}->[$_]->[1] & TABLE_EL) { |
| 4701 |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
| 4702 |
if (defined $parent and $parent->node_type == 1) { |
if (defined $parent and $parent->node_type == 1) { |
| 4703 |
!!!cp ('t196'); |
!!!cp ('t196'); |
| 4725 |
($self->{document}->create_text_node ($token->{data}), |
($self->{document}->create_text_node ($token->{data}), |
| 4726 |
$next_sibling); |
$next_sibling); |
| 4727 |
} |
} |
| 4728 |
} else { |
$open_tables->[-1]->[1] = 1; # tainted |
| 4729 |
!!!cp ('t200'); |
} else { |
| 4730 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
!!!cp ('t200'); |
| 4731 |
} |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 4732 |
|
} |
| 4733 |
|
|
| 4734 |
!!!next-token; |
!!!next-token; |
| 4735 |
redo B; |
next B; |
| 4736 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 4737 |
if ({ |
if ({ |
| 4738 |
tr => ($self->{insertion_mode} != IN_ROW_IM), |
tr => ($self->{insertion_mode} != IN_ROW_IM), |
| 4740 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4741 |
if ($self->{insertion_mode} == IN_TABLE_IM) { |
if ($self->{insertion_mode} == IN_TABLE_IM) { |
| 4742 |
## Clear back to table context |
## Clear back to table context |
| 4743 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while (not ($self->{open_elements}->[-1]->[1] |
| 4744 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
& TABLE_SCOPING_EL)) { |
| 4745 |
!!!cp ('t201'); |
!!!cp ('t201'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4746 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4747 |
} |
} |
| 4748 |
|
|
| 4749 |
!!!insert-element ('tbody'); |
!!!insert-element ('tbody',, $token); |
| 4750 |
$self->{insertion_mode} = IN_TABLE_BODY_IM; |
$self->{insertion_mode} = IN_TABLE_BODY_IM; |
| 4751 |
## reprocess in the "in table body" insertion mode... |
## reprocess in the "in table body" insertion mode... |
| 4752 |
} |
} |
| 4754 |
if ($self->{insertion_mode} == IN_TABLE_BODY_IM) { |
if ($self->{insertion_mode} == IN_TABLE_BODY_IM) { |
| 4755 |
unless ($token->{tag_name} eq 'tr') { |
unless ($token->{tag_name} eq 'tr') { |
| 4756 |
!!!cp ('t202'); |
!!!cp ('t202'); |
| 4757 |
!!!parse-error (type => 'missing start tag:tr'); |
!!!parse-error (type => 'missing start tag:tr', token => $token); |
| 4758 |
} |
} |
| 4759 |
|
|
| 4760 |
## Clear back to table body context |
## Clear back to table body context |
| 4761 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 4762 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
& TABLE_ROWS_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4763 |
!!!cp ('t203'); |
!!!cp ('t203'); |
| 4764 |
## ISSUE: Can this case be reached? |
## ISSUE: Can this case be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4765 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4766 |
} |
} |
| 4767 |
|
|
| 4768 |
$self->{insertion_mode} = IN_ROW_IM; |
$self->{insertion_mode} = IN_ROW_IM; |
| 4769 |
if ($token->{tag_name} eq 'tr') { |
if ($token->{tag_name} eq 'tr') { |
| 4770 |
!!!cp ('t204'); |
!!!cp ('t204'); |
| 4771 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 4772 |
|
!!!nack ('t204'); |
| 4773 |
!!!next-token; |
!!!next-token; |
| 4774 |
redo B; |
next B; |
| 4775 |
} else { |
} else { |
| 4776 |
!!!cp ('t205'); |
!!!cp ('t205'); |
| 4777 |
!!!insert-element ('tr'); |
!!!insert-element ('tr',, $token); |
| 4778 |
## reprocess in the "in row" insertion mode |
## reprocess in the "in row" insertion mode |
| 4779 |
} |
} |
| 4780 |
} else { |
} else { |
| 4782 |
} |
} |
| 4783 |
|
|
| 4784 |
## Clear back to table row context |
## Clear back to table row context |
| 4785 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 4786 |
tr => 1, html => 1, |
& TABLE_ROW_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4787 |
!!!cp ('t207'); |
!!!cp ('t207'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4788 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4789 |
} |
} |
| 4790 |
|
|
| 4791 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 4792 |
$self->{insertion_mode} = IN_CELL_IM; |
$self->{insertion_mode} = IN_CELL_IM; |
| 4793 |
|
|
| 4794 |
push @$active_formatting_elements, ['#marker', '']; |
push @$active_formatting_elements, ['#marker', '']; |
| 4795 |
|
|
| 4796 |
|
!!!nack ('t207.1'); |
| 4797 |
!!!next-token; |
!!!next-token; |
| 4798 |
redo B; |
next B; |
| 4799 |
} elsif ({ |
} elsif ({ |
| 4800 |
caption => 1, col => 1, colgroup => 1, |
caption => 1, col => 1, colgroup => 1, |
| 4801 |
tbody => 1, tfoot => 1, thead => 1, |
tbody => 1, tfoot => 1, thead => 1, |
| 4807 |
my $i; |
my $i; |
| 4808 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4809 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4810 |
if ($node->[1] eq 'tr') { |
if ($node->[1] & TABLE_ROW_EL) { |
| 4811 |
!!!cp ('t208'); |
!!!cp ('t208'); |
| 4812 |
$i = $_; |
$i = $_; |
| 4813 |
last INSCOPE; |
last INSCOPE; |
| 4814 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
html => 1, |
|
|
|
|
|
## NOTE: This element does not appear here, maybe. |
|
|
table => 1, |
|
|
}->{$node->[1]}) { |
|
| 4815 |
!!!cp ('t209'); |
!!!cp ('t209'); |
| 4816 |
last INSCOPE; |
last INSCOPE; |
| 4817 |
} |
} |
| 4818 |
} # INSCOPE |
} # INSCOPE |
| 4819 |
unless (defined $i) { |
unless (defined $i) { |
| 4820 |
!!!cp ('t210'); |
!!!cp ('t210'); |
| 4821 |
## TODO: This type is wrong. |
## TODO: This type is wrong. |
| 4822 |
!!!parse-error (type => 'unmacthed end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmacthed end tag:'.$token->{tag_name}, token => $token); |
| 4823 |
## Ignore the token |
## Ignore the token |
| 4824 |
|
!!!nack ('t210.1'); |
| 4825 |
!!!next-token; |
!!!next-token; |
| 4826 |
redo B; |
next B; |
| 4827 |
} |
} |
| 4828 |
|
|
| 4829 |
## Clear back to table row context |
## Clear back to table row context |
| 4830 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 4831 |
tr => 1, html => 1, |
& TABLE_ROW_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4832 |
!!!cp ('t211'); |
!!!cp ('t211'); |
| 4833 |
## ISSUE: Can this case be reached? |
## ISSUE: Can this case be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4834 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4835 |
} |
} |
| 4836 |
|
|
| 4839 |
if ($token->{tag_name} eq 'tr') { |
if ($token->{tag_name} eq 'tr') { |
| 4840 |
!!!cp ('t212'); |
!!!cp ('t212'); |
| 4841 |
## reprocess |
## reprocess |
| 4842 |
redo B; |
!!!ack-later; |
| 4843 |
|
next B; |
| 4844 |
} else { |
} else { |
| 4845 |
!!!cp ('t213'); |
!!!cp ('t213'); |
| 4846 |
## reprocess in the "in table body" insertion mode... |
## reprocess in the "in table body" insertion mode... |
| 4852 |
my $i; |
my $i; |
| 4853 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4854 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4855 |
if ({ |
if ($node->[1] & TABLE_ROW_GROUP_EL) { |
|
tbody => 1, thead => 1, tfoot => 1, |
|
|
}->{$node->[1]}) { |
|
| 4856 |
!!!cp ('t214'); |
!!!cp ('t214'); |
| 4857 |
$i = $_; |
$i = $_; |
| 4858 |
last INSCOPE; |
last INSCOPE; |
| 4859 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 4860 |
!!!cp ('t215'); |
!!!cp ('t215'); |
| 4861 |
last INSCOPE; |
last INSCOPE; |
| 4862 |
} |
} |
| 4864 |
unless (defined $i) { |
unless (defined $i) { |
| 4865 |
!!!cp ('t216'); |
!!!cp ('t216'); |
| 4866 |
## TODO: This erorr type ios wrong. |
## TODO: This erorr type ios wrong. |
| 4867 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 4868 |
## Ignore the token |
## Ignore the token |
| 4869 |
|
!!!nack ('t216.1'); |
| 4870 |
!!!next-token; |
!!!next-token; |
| 4871 |
redo B; |
next B; |
| 4872 |
} |
} |
| 4873 |
|
|
| 4874 |
## Clear back to table body context |
## Clear back to table body context |
| 4875 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 4876 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
& TABLE_ROWS_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4877 |
!!!cp ('t217'); |
!!!cp ('t217'); |
| 4878 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4879 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4880 |
} |
} |
| 4881 |
|
|
| 4895 |
|
|
| 4896 |
if ($token->{tag_name} eq 'col') { |
if ($token->{tag_name} eq 'col') { |
| 4897 |
## Clear back to table context |
## Clear back to table context |
| 4898 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while (not ($self->{open_elements}->[-1]->[1] |
| 4899 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
& TABLE_SCOPING_EL)) { |
| 4900 |
!!!cp ('t219'); |
!!!cp ('t219'); |
| 4901 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4902 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4903 |
} |
} |
| 4904 |
|
|
| 4905 |
!!!insert-element ('colgroup'); |
!!!insert-element ('colgroup',, $token); |
| 4906 |
$self->{insertion_mode} = IN_COLUMN_GROUP_IM; |
$self->{insertion_mode} = IN_COLUMN_GROUP_IM; |
| 4907 |
## reprocess |
## reprocess |
| 4908 |
redo B; |
!!!ack-later; |
| 4909 |
|
next B; |
| 4910 |
} elsif ({ |
} elsif ({ |
| 4911 |
caption => 1, |
caption => 1, |
| 4912 |
colgroup => 1, |
colgroup => 1, |
| 4913 |
tbody => 1, tfoot => 1, thead => 1, |
tbody => 1, tfoot => 1, thead => 1, |
| 4914 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4915 |
## Clear back to table context |
## Clear back to table context |
| 4916 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while (not ($self->{open_elements}->[-1]->[1] |
| 4917 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
& TABLE_SCOPING_EL)) { |
| 4918 |
!!!cp ('t220'); |
!!!cp ('t220'); |
| 4919 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4920 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4921 |
} |
} |
| 4922 |
|
|
| 4923 |
push @$active_formatting_elements, ['#marker', ''] |
push @$active_formatting_elements, ['#marker', ''] |
| 4924 |
if $token->{tag_name} eq 'caption'; |
if $token->{tag_name} eq 'caption'; |
| 4925 |
|
|
| 4926 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 4927 |
$self->{insertion_mode} = { |
$self->{insertion_mode} = { |
| 4928 |
caption => IN_CAPTION_IM, |
caption => IN_CAPTION_IM, |
| 4929 |
colgroup => IN_COLUMN_GROUP_IM, |
colgroup => IN_COLUMN_GROUP_IM, |
| 4932 |
thead => IN_TABLE_BODY_IM, |
thead => IN_TABLE_BODY_IM, |
| 4933 |
}->{$token->{tag_name}}; |
}->{$token->{tag_name}}; |
| 4934 |
!!!next-token; |
!!!next-token; |
| 4935 |
redo B; |
!!!nack ('t220.1'); |
| 4936 |
|
next B; |
| 4937 |
} else { |
} else { |
| 4938 |
die "$0: in table: <>: $token->{tag_name}"; |
die "$0: in table: <>: $token->{tag_name}"; |
| 4939 |
} |
} |
| 4940 |
} elsif ($token->{tag_name} eq 'table') { |
} elsif ($token->{tag_name} eq 'table') { |
| 4941 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4942 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4943 |
|
->manakai_local_name, |
| 4944 |
|
token => $token); |
| 4945 |
|
|
| 4946 |
## As if </table> |
## As if </table> |
| 4947 |
## have a table element in table scope |
## have a table element in table scope |
| 4948 |
my $i; |
my $i; |
| 4949 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4950 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4951 |
if ($node->[1] eq 'table') { |
if ($node->[1] & TABLE_EL) { |
| 4952 |
!!!cp ('t221'); |
!!!cp ('t221'); |
| 4953 |
$i = $_; |
$i = $_; |
| 4954 |
last INSCOPE; |
last INSCOPE; |
| 4955 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
#table => 1, |
|
|
html => 1, |
|
|
}->{$node->[1]}) { |
|
| 4956 |
!!!cp ('t222'); |
!!!cp ('t222'); |
| 4957 |
last INSCOPE; |
last INSCOPE; |
| 4958 |
} |
} |
| 4960 |
unless (defined $i) { |
unless (defined $i) { |
| 4961 |
!!!cp ('t223'); |
!!!cp ('t223'); |
| 4962 |
## TODO: The following is wrong, maybe. |
## TODO: The following is wrong, maybe. |
| 4963 |
!!!parse-error (type => 'unmatched end tag:table'); |
!!!parse-error (type => 'unmatched end tag:table', token => $token); |
| 4964 |
## Ignore tokens </table><table> |
## Ignore tokens </table><table> |
| 4965 |
|
!!!nack ('t223.1'); |
| 4966 |
!!!next-token; |
!!!next-token; |
| 4967 |
redo B; |
next B; |
| 4968 |
} |
} |
| 4969 |
|
|
| 4970 |
|
## TODO: Followings are removed from the latest spec. |
| 4971 |
## generate implied end tags |
## generate implied end tags |
| 4972 |
while ({ |
while ($self->{open_elements}->[-1]->[1] & END_TAG_OPTIONAL_EL) { |
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4973 |
!!!cp ('t224'); |
!!!cp ('t224'); |
| 4974 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4975 |
} |
} |
| 4976 |
|
|
| 4977 |
if ($self->{open_elements}->[-1]->[1] ne 'table') { |
unless ($self->{open_elements}->[-1]->[1] & TABLE_EL) { |
| 4978 |
!!!cp ('t225'); |
!!!cp ('t225'); |
| 4979 |
## ISSUE: Can this case be reached? |
## NOTE: |<table><tr><table>| |
| 4980 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 4981 |
|
value => $self->{open_elements}->[-1]->[0] |
| 4982 |
|
->manakai_local_name, |
| 4983 |
|
token => $token); |
| 4984 |
} else { |
} else { |
| 4985 |
!!!cp ('t226'); |
!!!cp ('t226'); |
| 4986 |
} |
} |
| 4987 |
|
|
| 4988 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 4989 |
|
pop @{$open_tables}; |
| 4990 |
|
|
| 4991 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 4992 |
|
|
| 4993 |
## reprocess |
## reprocess |
| 4994 |
redo B; |
!!!ack-later; |
| 4995 |
|
next B; |
| 4996 |
|
} elsif ($token->{tag_name} eq 'style') { |
| 4997 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 4998 |
|
!!!cp ('t227.8'); |
| 4999 |
|
## NOTE: This is a "as if in head" code clone. |
| 5000 |
|
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5001 |
|
next B; |
| 5002 |
|
} else { |
| 5003 |
|
!!!cp ('t227.7'); |
| 5004 |
|
# |
| 5005 |
|
} |
| 5006 |
|
} elsif ($token->{tag_name} eq 'script') { |
| 5007 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 5008 |
|
!!!cp ('t227.6'); |
| 5009 |
|
## NOTE: This is a "as if in head" code clone. |
| 5010 |
|
$script_start_tag->(); |
| 5011 |
|
next B; |
| 5012 |
|
} else { |
| 5013 |
|
!!!cp ('t227.5'); |
| 5014 |
|
# |
| 5015 |
|
} |
| 5016 |
|
} elsif ($token->{tag_name} eq 'input') { |
| 5017 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 5018 |
|
if ($token->{attributes}->{type}) { ## TODO: case |
| 5019 |
|
my $type = lc $token->{attributes}->{type}->{value}; |
| 5020 |
|
if ($type eq 'hidden') { |
| 5021 |
|
!!!cp ('t227.3'); |
| 5022 |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}, token => $token); |
| 5023 |
|
|
| 5024 |
|
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5025 |
|
|
| 5026 |
|
## TODO: form element pointer |
| 5027 |
|
|
| 5028 |
|
pop @{$self->{open_elements}}; |
| 5029 |
|
|
| 5030 |
|
!!!next-token; |
| 5031 |
|
!!!ack ('t227.2.1'); |
| 5032 |
|
next B; |
| 5033 |
|
} else { |
| 5034 |
|
!!!cp ('t227.2'); |
| 5035 |
|
# |
| 5036 |
|
} |
| 5037 |
|
} else { |
| 5038 |
|
!!!cp ('t227.1'); |
| 5039 |
|
# |
| 5040 |
|
} |
| 5041 |
|
} else { |
| 5042 |
|
!!!cp ('t227.4'); |
| 5043 |
|
# |
| 5044 |
|
} |
| 5045 |
} else { |
} else { |
| 5046 |
!!!cp ('t227'); |
!!!cp ('t227'); |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}); |
|
|
|
|
|
$insert = $insert_to_foster; |
|
| 5047 |
# |
# |
| 5048 |
} |
} |
| 5049 |
|
|
| 5050 |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}, token => $token); |
| 5051 |
|
|
| 5052 |
|
$insert = $insert_to_foster; |
| 5053 |
|
# |
| 5054 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5055 |
if ($token->{tag_name} eq 'tr' and |
if ($token->{tag_name} eq 'tr' and |
| 5056 |
$self->{insertion_mode} == IN_ROW_IM) { |
$self->{insertion_mode} == IN_ROW_IM) { |
| 5058 |
my $i; |
my $i; |
| 5059 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5060 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5061 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] & TABLE_ROW_EL) { |
| 5062 |
!!!cp ('t228'); |
!!!cp ('t228'); |
| 5063 |
$i = $_; |
$i = $_; |
| 5064 |
last INSCOPE; |
last INSCOPE; |
| 5065 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5066 |
!!!cp ('t229'); |
!!!cp ('t229'); |
| 5067 |
last INSCOPE; |
last INSCOPE; |
| 5068 |
} |
} |
| 5069 |
} # INSCOPE |
} # INSCOPE |
| 5070 |
unless (defined $i) { |
unless (defined $i) { |
| 5071 |
!!!cp ('t230'); |
!!!cp ('t230'); |
| 5072 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5073 |
## Ignore the token |
## Ignore the token |
| 5074 |
|
!!!nack ('t230.1'); |
| 5075 |
!!!next-token; |
!!!next-token; |
| 5076 |
redo B; |
next B; |
| 5077 |
} else { |
} else { |
| 5078 |
!!!cp ('t232'); |
!!!cp ('t232'); |
| 5079 |
} |
} |
| 5080 |
|
|
| 5081 |
## Clear back to table row context |
## Clear back to table row context |
| 5082 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 5083 |
tr => 1, html => 1, |
& TABLE_ROW_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 5084 |
!!!cp ('t231'); |
!!!cp ('t231'); |
| 5085 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 5086 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5087 |
} |
} |
| 5088 |
|
|
| 5089 |
pop @{$self->{open_elements}}; # tr |
pop @{$self->{open_elements}}; # tr |
| 5090 |
$self->{insertion_mode} = IN_TABLE_BODY_IM; |
$self->{insertion_mode} = IN_TABLE_BODY_IM; |
| 5091 |
!!!next-token; |
!!!next-token; |
| 5092 |
redo B; |
!!!nack ('t231.1'); |
| 5093 |
|
next B; |
| 5094 |
} elsif ($token->{tag_name} eq 'table') { |
} elsif ($token->{tag_name} eq 'table') { |
| 5095 |
if ($self->{insertion_mode} == IN_ROW_IM) { |
if ($self->{insertion_mode} == IN_ROW_IM) { |
| 5096 |
## As if </tr> |
## As if </tr> |
| 5098 |
my $i; |
my $i; |
| 5099 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5100 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5101 |
if ($node->[1] eq 'tr') { |
if ($node->[1] & TABLE_ROW_EL) { |
| 5102 |
!!!cp ('t233'); |
!!!cp ('t233'); |
| 5103 |
$i = $_; |
$i = $_; |
| 5104 |
last INSCOPE; |
last INSCOPE; |
| 5105 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5106 |
!!!cp ('t234'); |
!!!cp ('t234'); |
| 5107 |
last INSCOPE; |
last INSCOPE; |
| 5108 |
} |
} |
| 5110 |
unless (defined $i) { |
unless (defined $i) { |
| 5111 |
!!!cp ('t235'); |
!!!cp ('t235'); |
| 5112 |
## TODO: The following is wrong. |
## TODO: The following is wrong. |
| 5113 |
!!!parse-error (type => 'unmatched end tag:'.$token->{type}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{type}, token => $token); |
| 5114 |
## Ignore the token |
## Ignore the token |
| 5115 |
|
!!!nack ('t236.1'); |
| 5116 |
!!!next-token; |
!!!next-token; |
| 5117 |
redo B; |
next B; |
| 5118 |
} |
} |
| 5119 |
|
|
| 5120 |
## Clear back to table row context |
## Clear back to table row context |
| 5121 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 5122 |
tr => 1, html => 1, |
& TABLE_ROW_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 5123 |
!!!cp ('t236'); |
!!!cp ('t236'); |
| 5124 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 5125 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5126 |
} |
} |
| 5127 |
|
|
| 5135 |
my $i; |
my $i; |
| 5136 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5137 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5138 |
if ({ |
if ($node->[1] & TABLE_ROW_GROUP_EL) { |
|
tbody => 1, thead => 1, tfoot => 1, |
|
|
}->{$node->[1]}) { |
|
| 5139 |
!!!cp ('t237'); |
!!!cp ('t237'); |
| 5140 |
$i = $_; |
$i = $_; |
| 5141 |
last INSCOPE; |
last INSCOPE; |
| 5142 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5143 |
!!!cp ('t238'); |
!!!cp ('t238'); |
| 5144 |
last INSCOPE; |
last INSCOPE; |
| 5145 |
} |
} |
| 5146 |
} # INSCOPE |
} # INSCOPE |
| 5147 |
unless (defined $i) { |
unless (defined $i) { |
| 5148 |
!!!cp ('t239'); |
!!!cp ('t239'); |
| 5149 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5150 |
## Ignore the token |
## Ignore the token |
| 5151 |
|
!!!nack ('t239.1'); |
| 5152 |
!!!next-token; |
!!!next-token; |
| 5153 |
redo B; |
next B; |
| 5154 |
} |
} |
| 5155 |
|
|
| 5156 |
## Clear back to table body context |
## Clear back to table body context |
| 5157 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 5158 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
& TABLE_ROWS_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 5159 |
!!!cp ('t240'); |
!!!cp ('t240'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 5160 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5161 |
} |
} |
| 5162 |
|
|
| 5172 |
## reprocess in the "in table" insertion mode... |
## reprocess in the "in table" insertion mode... |
| 5173 |
} |
} |
| 5174 |
|
|
| 5175 |
|
## NOTE: </table> in the "in table" insertion mode. |
| 5176 |
|
## When you edit the code fragment below, please ensure that |
| 5177 |
|
## the code for <table> in the "in table" insertion mode |
| 5178 |
|
## is synced with it. |
| 5179 |
|
|
| 5180 |
## have a table element in table scope |
## have a table element in table scope |
| 5181 |
my $i; |
my $i; |
| 5182 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5183 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5184 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] & TABLE_EL) { |
| 5185 |
!!!cp ('t241'); |
!!!cp ('t241'); |
| 5186 |
$i = $_; |
$i = $_; |
| 5187 |
last INSCOPE; |
last INSCOPE; |
| 5188 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5189 |
!!!cp ('t242'); |
!!!cp ('t242'); |
| 5190 |
last INSCOPE; |
last INSCOPE; |
| 5191 |
} |
} |
| 5192 |
} # INSCOPE |
} # INSCOPE |
| 5193 |
unless (defined $i) { |
unless (defined $i) { |
| 5194 |
!!!cp ('t243'); |
!!!cp ('t243'); |
| 5195 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5196 |
## Ignore the token |
## Ignore the token |
| 5197 |
|
!!!nack ('t243.1'); |
| 5198 |
!!!next-token; |
!!!next-token; |
| 5199 |
redo B; |
next B; |
|
} |
|
|
|
|
|
## generate implied end tags |
|
|
while ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t244'); |
|
|
## ISSUE: Can this case be reached? |
|
|
pop @{$self->{open_elements}}; |
|
|
} |
|
|
|
|
|
if ($self->{open_elements}->[-1]->[1] ne 'table') { |
|
|
!!!cp ('t245'); |
|
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
|
} else { |
|
|
!!!cp ('t246'); |
|
| 5200 |
} |
} |
| 5201 |
|
|
| 5202 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 5203 |
|
pop @{$open_tables}; |
| 5204 |
|
|
| 5205 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 5206 |
|
|
| 5207 |
!!!next-token; |
!!!next-token; |
| 5208 |
redo B; |
next B; |
| 5209 |
} elsif ({ |
} elsif ({ |
| 5210 |
tbody => 1, tfoot => 1, thead => 1, |
tbody => 1, tfoot => 1, thead => 1, |
| 5211 |
}->{$token->{tag_name}} and |
}->{$token->{tag_name}} and |
| 5215 |
my $i; |
my $i; |
| 5216 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5217 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5218 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 5219 |
!!!cp ('t247'); |
!!!cp ('t247'); |
| 5220 |
$i = $_; |
$i = $_; |
| 5221 |
last INSCOPE; |
last INSCOPE; |
| 5222 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5223 |
!!!cp ('t248'); |
!!!cp ('t248'); |
| 5224 |
last INSCOPE; |
last INSCOPE; |
| 5225 |
} |
} |
| 5226 |
} # INSCOPE |
} # INSCOPE |
| 5227 |
unless (defined $i) { |
unless (defined $i) { |
| 5228 |
!!!cp ('t249'); |
!!!cp ('t249'); |
| 5229 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5230 |
## Ignore the token |
## Ignore the token |
| 5231 |
|
!!!nack ('t249.1'); |
| 5232 |
!!!next-token; |
!!!next-token; |
| 5233 |
redo B; |
next B; |
| 5234 |
} |
} |
| 5235 |
|
|
| 5236 |
## As if </tr> |
## As if </tr> |
| 5238 |
my $i; |
my $i; |
| 5239 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5240 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5241 |
if ($node->[1] eq 'tr') { |
if ($node->[1] & TABLE_ROW_EL) { |
| 5242 |
!!!cp ('t250'); |
!!!cp ('t250'); |
| 5243 |
$i = $_; |
$i = $_; |
| 5244 |
last INSCOPE; |
last INSCOPE; |
| 5245 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5246 |
!!!cp ('t251'); |
!!!cp ('t251'); |
| 5247 |
last INSCOPE; |
last INSCOPE; |
| 5248 |
} |
} |
| 5249 |
} # INSCOPE |
} # INSCOPE |
| 5250 |
unless (defined $i) { |
unless (defined $i) { |
| 5251 |
!!!cp ('t252'); |
!!!cp ('t252'); |
| 5252 |
!!!parse-error (type => 'unmatched end tag:tr'); |
!!!parse-error (type => 'unmatched end tag:tr', token => $token); |
| 5253 |
## Ignore the token |
## Ignore the token |
| 5254 |
|
!!!nack ('t252.1'); |
| 5255 |
!!!next-token; |
!!!next-token; |
| 5256 |
redo B; |
next B; |
| 5257 |
} |
} |
| 5258 |
|
|
| 5259 |
## Clear back to table row context |
## Clear back to table row context |
| 5260 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 5261 |
tr => 1, html => 1, |
& TABLE_ROW_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 5262 |
!!!cp ('t253'); |
!!!cp ('t253'); |
| 5263 |
## ISSUE: Can this case be reached? |
## ISSUE: Can this case be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 5264 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5265 |
} |
} |
| 5266 |
|
|
| 5273 |
my $i; |
my $i; |
| 5274 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5275 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5276 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 5277 |
!!!cp ('t254'); |
!!!cp ('t254'); |
| 5278 |
$i = $_; |
$i = $_; |
| 5279 |
last INSCOPE; |
last INSCOPE; |
| 5280 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5281 |
!!!cp ('t255'); |
!!!cp ('t255'); |
| 5282 |
last INSCOPE; |
last INSCOPE; |
| 5283 |
} |
} |
| 5284 |
} # INSCOPE |
} # INSCOPE |
| 5285 |
unless (defined $i) { |
unless (defined $i) { |
| 5286 |
!!!cp ('t256'); |
!!!cp ('t256'); |
| 5287 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5288 |
## Ignore the token |
## Ignore the token |
| 5289 |
|
!!!nack ('t256.1'); |
| 5290 |
!!!next-token; |
!!!next-token; |
| 5291 |
redo B; |
next B; |
| 5292 |
} |
} |
| 5293 |
|
|
| 5294 |
## Clear back to table body context |
## Clear back to table body context |
| 5295 |
while (not { |
while (not ($self->{open_elements}->[-1]->[1] |
| 5296 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
& TABLE_ROWS_SCOPING_EL)) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 5297 |
!!!cp ('t257'); |
!!!cp ('t257'); |
| 5298 |
## ISSUE: Can this case be reached? |
## ISSUE: Can this case be reached? |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 5299 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5300 |
} |
} |
| 5301 |
|
|
| 5302 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5303 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 5304 |
|
!!!nack ('t257.1'); |
| 5305 |
!!!next-token; |
!!!next-token; |
| 5306 |
redo B; |
next B; |
| 5307 |
} elsif ({ |
} elsif ({ |
| 5308 |
body => 1, caption => 1, col => 1, colgroup => 1, |
body => 1, caption => 1, col => 1, colgroup => 1, |
| 5309 |
html => 1, td => 1, th => 1, |
html => 1, td => 1, th => 1, |
| 5310 |
tr => 1, # $self->{insertion_mode} == IN_ROW_IM |
tr => 1, # $self->{insertion_mode} == IN_ROW_IM |
| 5311 |
tbody => 1, tfoot => 1, thead => 1, # $self->{insertion_mode} == IN_TABLE_IM |
tbody => 1, tfoot => 1, thead => 1, # $self->{insertion_mode} == IN_TABLE_IM |
| 5312 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 5313 |
!!!cp ('t258'); |
!!!cp ('t258'); |
| 5314 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5315 |
## Ignore the token |
## Ignore the token |
| 5316 |
!!!next-token; |
!!!nack ('t258.1'); |
| 5317 |
redo B; |
!!!next-token; |
| 5318 |
|
next B; |
| 5319 |
} else { |
} else { |
| 5320 |
!!!cp ('t259'); |
!!!cp ('t259'); |
| 5321 |
!!!parse-error (type => 'in table:/'.$token->{tag_name}); |
!!!parse-error (type => 'in table:/'.$token->{tag_name}, token => $token); |
| 5322 |
|
|
| 5323 |
$insert = $insert_to_foster; |
$insert = $insert_to_foster; |
| 5324 |
# |
# |
| 5325 |
} |
} |
| 5326 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 5327 |
|
unless ($self->{open_elements}->[-1]->[1] & HTML_EL and |
| 5328 |
|
@{$self->{open_elements}} == 1) { # redundant, maybe |
| 5329 |
|
!!!parse-error (type => 'in body:#eof', token => $token); |
| 5330 |
|
!!!cp ('t259.1'); |
| 5331 |
|
# |
| 5332 |
|
} else { |
| 5333 |
|
!!!cp ('t259.2'); |
| 5334 |
|
# |
| 5335 |
|
} |
| 5336 |
|
|
| 5337 |
|
## Stop parsing |
| 5338 |
|
last B; |
| 5339 |
} else { |
} else { |
| 5340 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 5341 |
} |
} |
| 5346 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 5347 |
!!!cp ('t260'); |
!!!cp ('t260'); |
| 5348 |
!!!next-token; |
!!!next-token; |
| 5349 |
redo B; |
next B; |
| 5350 |
} |
} |
| 5351 |
} |
} |
| 5352 |
|
|
| 5355 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 5356 |
if ($token->{tag_name} eq 'col') { |
if ($token->{tag_name} eq 'col') { |
| 5357 |
!!!cp ('t262'); |
!!!cp ('t262'); |
| 5358 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5359 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5360 |
|
!!!ack ('t262.1'); |
| 5361 |
!!!next-token; |
!!!next-token; |
| 5362 |
redo B; |
next B; |
| 5363 |
} else { |
} else { |
| 5364 |
!!!cp ('t263'); |
!!!cp ('t263'); |
| 5365 |
# |
# |
| 5366 |
} |
} |
| 5367 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5368 |
if ($token->{tag_name} eq 'colgroup') { |
if ($token->{tag_name} eq 'colgroup') { |
| 5369 |
if ($self->{open_elements}->[-1]->[1] eq 'html') { |
if ($self->{open_elements}->[-1]->[1] & HTML_EL) { |
| 5370 |
!!!cp ('t264'); |
!!!cp ('t264'); |
| 5371 |
!!!parse-error (type => 'unmatched end tag:colgroup'); |
!!!parse-error (type => 'unmatched end tag:colgroup', token => $token); |
| 5372 |
## Ignore the token |
## Ignore the token |
| 5373 |
!!!next-token; |
!!!next-token; |
| 5374 |
redo B; |
next B; |
| 5375 |
} else { |
} else { |
| 5376 |
!!!cp ('t265'); |
!!!cp ('t265'); |
| 5377 |
pop @{$self->{open_elements}}; # colgroup |
pop @{$self->{open_elements}}; # colgroup |
| 5378 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 5379 |
!!!next-token; |
!!!next-token; |
| 5380 |
redo B; |
next B; |
| 5381 |
} |
} |
| 5382 |
} elsif ($token->{tag_name} eq 'col') { |
} elsif ($token->{tag_name} eq 'col') { |
| 5383 |
!!!cp ('t266'); |
!!!cp ('t266'); |
| 5384 |
!!!parse-error (type => 'unmatched end tag:col'); |
!!!parse-error (type => 'unmatched end tag:col', token => $token); |
| 5385 |
## Ignore the token |
## Ignore the token |
| 5386 |
!!!next-token; |
!!!next-token; |
| 5387 |
redo B; |
next B; |
| 5388 |
} else { |
} else { |
| 5389 |
!!!cp ('t267'); |
!!!cp ('t267'); |
| 5390 |
# |
# |
| 5391 |
} |
} |
| 5392 |
} else { |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 5393 |
die "$0: $token->{type}: Unknown token type"; |
if ($self->{open_elements}->[-1]->[1] & HTML_EL and |
| 5394 |
} |
@{$self->{open_elements}} == 1) { # redundant, maybe |
| 5395 |
|
!!!cp ('t270.2'); |
| 5396 |
|
## Stop parsing. |
| 5397 |
|
last B; |
| 5398 |
|
} else { |
| 5399 |
|
## NOTE: As if </colgroup>. |
| 5400 |
|
!!!cp ('t270.1'); |
| 5401 |
|
pop @{$self->{open_elements}}; # colgroup |
| 5402 |
|
$self->{insertion_mode} = IN_TABLE_IM; |
| 5403 |
|
## Reprocess. |
| 5404 |
|
next B; |
| 5405 |
|
} |
| 5406 |
|
} else { |
| 5407 |
|
die "$0: $token->{type}: Unknown token type"; |
| 5408 |
|
} |
| 5409 |
|
|
| 5410 |
## As if </colgroup> |
## As if </colgroup> |
| 5411 |
if ($self->{open_elements}->[-1]->[1] eq 'html') { |
if ($self->{open_elements}->[-1]->[1] & HTML_EL) { |
| 5412 |
!!!cp ('t269'); |
!!!cp ('t269'); |
| 5413 |
!!!parse-error (type => 'unmatched end tag:colgroup'); |
## TODO: Wrong error type? |
| 5414 |
|
!!!parse-error (type => 'unmatched end tag:colgroup', token => $token); |
| 5415 |
## Ignore the token |
## Ignore the token |
| 5416 |
|
!!!nack ('t269.1'); |
| 5417 |
!!!next-token; |
!!!next-token; |
| 5418 |
redo B; |
next B; |
| 5419 |
} else { |
} else { |
| 5420 |
!!!cp ('t270'); |
!!!cp ('t270'); |
| 5421 |
pop @{$self->{open_elements}}; # colgroup |
pop @{$self->{open_elements}}; # colgroup |
| 5422 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 5423 |
|
!!!ack-later; |
| 5424 |
## reprocess |
## reprocess |
| 5425 |
redo B; |
next B; |
| 5426 |
} |
} |
| 5427 |
} elsif ($self->{insertion_mode} == IN_SELECT_IM) { |
} elsif ($self->{insertion_mode} & SELECT_IMS) { |
| 5428 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 5429 |
!!!cp ('t271'); |
!!!cp ('t271'); |
| 5430 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 5431 |
!!!next-token; |
!!!next-token; |
| 5432 |
redo B; |
next B; |
| 5433 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 5434 |
if ($token->{tag_name} eq 'option') { |
if ($token->{tag_name} eq 'option') { |
| 5435 |
if ($self->{open_elements}->[-1]->[1] eq 'option') { |
if ($self->{open_elements}->[-1]->[1] & OPTION_EL) { |
| 5436 |
!!!cp ('t272'); |
!!!cp ('t272'); |
| 5437 |
## As if </option> |
## As if </option> |
| 5438 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5439 |
} else { |
} else { |
| 5440 |
!!!cp ('t273'); |
!!!cp ('t273'); |
| 5441 |
} |
} |
| 5442 |
|
|
| 5443 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5444 |
!!!next-token; |
!!!nack ('t273.1'); |
| 5445 |
redo B; |
!!!next-token; |
| 5446 |
} elsif ($token->{tag_name} eq 'optgroup') { |
next B; |
| 5447 |
if ($self->{open_elements}->[-1]->[1] eq 'option') { |
} elsif ($token->{tag_name} eq 'optgroup') { |
| 5448 |
!!!cp ('t274'); |
if ($self->{open_elements}->[-1]->[1] & OPTION_EL) { |
| 5449 |
## As if </option> |
!!!cp ('t274'); |
| 5450 |
pop @{$self->{open_elements}}; |
## As if </option> |
| 5451 |
} else { |
pop @{$self->{open_elements}}; |
| 5452 |
!!!cp ('t275'); |
} else { |
| 5453 |
} |
!!!cp ('t275'); |
| 5454 |
|
} |
| 5455 |
|
|
| 5456 |
if ($self->{open_elements}->[-1]->[1] eq 'optgroup') { |
if ($self->{open_elements}->[-1]->[1] & OPTGROUP_EL) { |
| 5457 |
!!!cp ('t276'); |
!!!cp ('t276'); |
| 5458 |
## As if </optgroup> |
## As if </optgroup> |
| 5459 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5460 |
} else { |
} else { |
| 5461 |
!!!cp ('t277'); |
!!!cp ('t277'); |
| 5462 |
} |
} |
| 5463 |
|
|
| 5464 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5465 |
!!!next-token; |
!!!nack ('t277.1'); |
| 5466 |
redo B; |
!!!next-token; |
| 5467 |
} elsif ($token->{tag_name} eq 'select') { |
next B; |
| 5468 |
## TODO: The type below is not good - <select> is replaced by </select> |
} elsif ($token->{tag_name} eq 'select' or |
| 5469 |
!!!parse-error (type => 'not closed:select'); |
$token->{tag_name} eq 'input' or |
| 5470 |
## As if </select> instead |
($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and |
| 5471 |
## have an element in table scope |
{ |
| 5472 |
my $i; |
caption => 1, table => 1, |
| 5473 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
tbody => 1, tfoot => 1, thead => 1, |
| 5474 |
my $node = $self->{open_elements}->[$_]; |
tr => 1, td => 1, th => 1, |
| 5475 |
if ($node->[1] eq $token->{tag_name}) { |
}->{$token->{tag_name}})) { |
| 5476 |
!!!cp ('t278'); |
## TODO: The type below is not good - <select> is replaced by </select> |
| 5477 |
$i = $_; |
!!!parse-error (type => 'not closed:select', token => $token); |
| 5478 |
last INSCOPE; |
## NOTE: As if the token were </select> (<select> case) or |
| 5479 |
} elsif ({ |
## as if there were </select> (otherwise). |
| 5480 |
table => 1, html => 1, |
## have an element in table scope |
| 5481 |
}->{$node->[1]}) { |
my $i; |
| 5482 |
!!!cp ('t279'); |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5483 |
last INSCOPE; |
my $node = $self->{open_elements}->[$_]; |
| 5484 |
} |
if ($node->[1] & SELECT_EL) { |
| 5485 |
} # INSCOPE |
!!!cp ('t278'); |
| 5486 |
unless (defined $i) { |
$i = $_; |
| 5487 |
!!!cp ('t280'); |
last INSCOPE; |
| 5488 |
!!!parse-error (type => 'unmatched end tag:select'); |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 5489 |
## Ignore the token |
!!!cp ('t279'); |
| 5490 |
!!!next-token; |
last INSCOPE; |
| 5491 |
redo B; |
} |
| 5492 |
} |
} # INSCOPE |
| 5493 |
|
unless (defined $i) { |
| 5494 |
|
!!!cp ('t280'); |
| 5495 |
|
!!!parse-error (type => 'unmatched end tag:select', token => $token); |
| 5496 |
|
## Ignore the token |
| 5497 |
|
!!!nack ('t280.1'); |
| 5498 |
|
!!!next-token; |
| 5499 |
|
next B; |
| 5500 |
|
} |
| 5501 |
|
|
| 5502 |
!!!cp ('t281'); |
!!!cp ('t281'); |
| 5503 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 5504 |
|
|
| 5505 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 5506 |
|
|
| 5507 |
!!!next-token; |
if ($token->{tag_name} eq 'select') { |
| 5508 |
redo B; |
!!!nack ('t281.2'); |
| 5509 |
|
!!!next-token; |
| 5510 |
|
next B; |
| 5511 |
|
} else { |
| 5512 |
|
!!!cp ('t281.1'); |
| 5513 |
|
!!!ack-later; |
| 5514 |
|
## Reprocess the token. |
| 5515 |
|
next B; |
| 5516 |
|
} |
| 5517 |
} else { |
} else { |
| 5518 |
!!!cp ('t282'); |
!!!cp ('t282'); |
| 5519 |
!!!parse-error (type => 'in select:'.$token->{tag_name}); |
!!!parse-error (type => 'in select:'.$token->{tag_name}, token => $token); |
| 5520 |
## Ignore the token |
## Ignore the token |
| 5521 |
|
!!!nack ('t282.1'); |
| 5522 |
!!!next-token; |
!!!next-token; |
| 5523 |
redo B; |
next B; |
| 5524 |
} |
} |
| 5525 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5526 |
if ($token->{tag_name} eq 'optgroup') { |
if ($token->{tag_name} eq 'optgroup') { |
| 5527 |
if ($self->{open_elements}->[-1]->[1] eq 'option' and |
if ($self->{open_elements}->[-1]->[1] & OPTION_EL and |
| 5528 |
$self->{open_elements}->[-2]->[1] eq 'optgroup') { |
$self->{open_elements}->[-2]->[1] & OPTGROUP_EL) { |
| 5529 |
!!!cp ('t283'); |
!!!cp ('t283'); |
| 5530 |
## As if </option> |
## As if </option> |
| 5531 |
splice @{$self->{open_elements}}, -2; |
splice @{$self->{open_elements}}, -2; |
| 5532 |
} elsif ($self->{open_elements}->[-1]->[1] eq 'optgroup') { |
} elsif ($self->{open_elements}->[-1]->[1] & OPTGROUP_EL) { |
| 5533 |
!!!cp ('t284'); |
!!!cp ('t284'); |
| 5534 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5535 |
} else { |
} else { |
| 5536 |
!!!cp ('t285'); |
!!!cp ('t285'); |
| 5537 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5538 |
## Ignore the token |
## Ignore the token |
| 5539 |
} |
} |
| 5540 |
!!!next-token; |
!!!nack ('t285.1'); |
| 5541 |
redo B; |
!!!next-token; |
| 5542 |
} elsif ($token->{tag_name} eq 'option') { |
next B; |
| 5543 |
if ($self->{open_elements}->[-1]->[1] eq 'option') { |
} elsif ($token->{tag_name} eq 'option') { |
| 5544 |
!!!cp ('t286'); |
if ($self->{open_elements}->[-1]->[1] & OPTION_EL) { |
| 5545 |
pop @{$self->{open_elements}}; |
!!!cp ('t286'); |
| 5546 |
} else { |
pop @{$self->{open_elements}}; |
| 5547 |
!!!cp ('t287'); |
} else { |
| 5548 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!cp ('t287'); |
| 5549 |
## Ignore the token |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5550 |
} |
## Ignore the token |
| 5551 |
!!!next-token; |
} |
| 5552 |
redo B; |
!!!nack ('t287.1'); |
| 5553 |
} elsif ($token->{tag_name} eq 'select') { |
!!!next-token; |
| 5554 |
## have an element in table scope |
next B; |
| 5555 |
my $i; |
} elsif ($token->{tag_name} eq 'select') { |
| 5556 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
## have an element in table scope |
| 5557 |
my $node = $self->{open_elements}->[$_]; |
my $i; |
| 5558 |
if ($node->[1] eq $token->{tag_name}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5559 |
!!!cp ('t288'); |
my $node = $self->{open_elements}->[$_]; |
| 5560 |
$i = $_; |
if ($node->[1] & SELECT_EL) { |
| 5561 |
last INSCOPE; |
!!!cp ('t288'); |
| 5562 |
} elsif ({ |
$i = $_; |
| 5563 |
table => 1, html => 1, |
last INSCOPE; |
| 5564 |
}->{$node->[1]}) { |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 5565 |
!!!cp ('t289'); |
!!!cp ('t289'); |
| 5566 |
last INSCOPE; |
last INSCOPE; |
| 5567 |
} |
} |
| 5568 |
} # INSCOPE |
} # INSCOPE |
| 5569 |
unless (defined $i) { |
unless (defined $i) { |
| 5570 |
!!!cp ('t290'); |
!!!cp ('t290'); |
| 5571 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5572 |
## Ignore the token |
## Ignore the token |
| 5573 |
!!!next-token; |
!!!nack ('t290.1'); |
| 5574 |
redo B; |
!!!next-token; |
| 5575 |
} |
next B; |
| 5576 |
|
} |
| 5577 |
|
|
| 5578 |
!!!cp ('t291'); |
!!!cp ('t291'); |
| 5579 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 5580 |
|
|
| 5581 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 5582 |
|
|
| 5583 |
!!!next-token; |
!!!nack ('t291.1'); |
| 5584 |
redo B; |
!!!next-token; |
| 5585 |
} elsif ({ |
next B; |
| 5586 |
caption => 1, table => 1, tbody => 1, |
} elsif ($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and |
| 5587 |
tfoot => 1, thead => 1, tr => 1, td => 1, th => 1, |
{ |
| 5588 |
}->{$token->{tag_name}}) { |
caption => 1, table => 1, tbody => 1, |
| 5589 |
|
tfoot => 1, thead => 1, tr => 1, td => 1, th => 1, |
| 5590 |
|
}->{$token->{tag_name}}) { |
| 5591 |
## TODO: The following is wrong? |
## TODO: The following is wrong? |
| 5592 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5593 |
|
|
| 5594 |
## have an element in table scope |
## have an element in table scope |
| 5595 |
my $i; |
my $i; |
| 5596 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5597 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5598 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 5599 |
!!!cp ('t292'); |
!!!cp ('t292'); |
| 5600 |
$i = $_; |
$i = $_; |
| 5601 |
last INSCOPE; |
last INSCOPE; |
| 5602 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
| 5603 |
table => 1, html => 1, |
!!!cp ('t293'); |
| 5604 |
}->{$node->[1]}) { |
last INSCOPE; |
| 5605 |
!!!cp ('t293'); |
} |
| 5606 |
last INSCOPE; |
} # INSCOPE |
| 5607 |
} |
unless (defined $i) { |
| 5608 |
} # INSCOPE |
!!!cp ('t294'); |
| 5609 |
unless (defined $i) { |
## Ignore the token |
| 5610 |
!!!cp ('t294'); |
!!!nack ('t294.1'); |
| 5611 |
## Ignore the token |
!!!next-token; |
| 5612 |
!!!next-token; |
next B; |
| 5613 |
redo B; |
} |
|
} |
|
| 5614 |
|
|
| 5615 |
## As if </select> |
## As if </select> |
| 5616 |
## have an element in table scope |
## have an element in table scope |
| 5617 |
undef $i; |
undef $i; |
| 5618 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 5619 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 5620 |
if ($node->[1] eq 'select') { |
if ($node->[1] & SELECT_EL) { |
| 5621 |
!!!cp ('t295'); |
!!!cp ('t295'); |
| 5622 |
$i = $_; |
$i = $_; |
| 5623 |
last INSCOPE; |
last INSCOPE; |
| 5624 |
} elsif ({ |
} elsif ($node->[1] & TABLE_SCOPING_EL) { |
|
table => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 5625 |
## ISSUE: Can this state be reached? |
## ISSUE: Can this state be reached? |
| 5626 |
!!!cp ('t296'); |
!!!cp ('t296'); |
| 5627 |
last INSCOPE; |
last INSCOPE; |
| 5628 |
} |
} |
| 5629 |
} # INSCOPE |
} # INSCOPE |
| 5630 |
unless (defined $i) { |
unless (defined $i) { |
| 5631 |
!!!cp ('t297'); |
!!!cp ('t297'); |
| 5632 |
## TODO: The following error type is correct? |
## TODO: The following error type is correct? |
| 5633 |
!!!parse-error (type => 'unmatched end tag:select'); |
!!!parse-error (type => 'unmatched end tag:select', token => $token); |
| 5634 |
## Ignore the </select> token |
## Ignore the </select> token |
| 5635 |
!!!next-token; ## TODO: ok? |
!!!nack ('t297.1'); |
| 5636 |
redo B; |
!!!next-token; ## TODO: ok? |
| 5637 |
} |
next B; |
| 5638 |
|
} |
| 5639 |
|
|
| 5640 |
!!!cp ('t298'); |
!!!cp ('t298'); |
| 5641 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 5642 |
|
|
| 5643 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 5644 |
|
|
| 5645 |
## reprocess |
!!!ack-later; |
| 5646 |
redo B; |
## reprocess |
| 5647 |
|
next B; |
| 5648 |
} else { |
} else { |
| 5649 |
!!!cp ('t299'); |
!!!cp ('t299'); |
| 5650 |
!!!parse-error (type => 'in select:/'.$token->{tag_name}); |
!!!parse-error (type => 'in select:/'.$token->{tag_name}, token => $token); |
| 5651 |
## Ignore the token |
## Ignore the token |
| 5652 |
|
!!!nack ('t299.3'); |
| 5653 |
!!!next-token; |
!!!next-token; |
| 5654 |
redo B; |
next B; |
| 5655 |
|
} |
| 5656 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 5657 |
|
unless ($self->{open_elements}->[-1]->[1] & HTML_EL and |
| 5658 |
|
@{$self->{open_elements}} == 1) { # redundant, maybe |
| 5659 |
|
!!!cp ('t299.1'); |
| 5660 |
|
!!!parse-error (type => 'in body:#eof', token => $token); |
| 5661 |
|
} else { |
| 5662 |
|
!!!cp ('t299.2'); |
| 5663 |
} |
} |
| 5664 |
|
|
| 5665 |
|
## Stop parsing. |
| 5666 |
|
last B; |
| 5667 |
} else { |
} else { |
| 5668 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 5669 |
} |
} |
| 5679 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 5680 |
!!!cp ('t300'); |
!!!cp ('t300'); |
| 5681 |
!!!next-token; |
!!!next-token; |
| 5682 |
redo B; |
next B; |
| 5683 |
} |
} |
| 5684 |
} |
} |
| 5685 |
|
|
| 5686 |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
| 5687 |
!!!cp ('t301'); |
!!!cp ('t301'); |
| 5688 |
!!!parse-error (type => 'after html:#character'); |
!!!parse-error (type => 'after html:#character', token => $token); |
| 5689 |
|
|
| 5690 |
## Reprocess in the "after body" insertion mode. |
## Reprocess in the "after body" insertion mode. |
| 5691 |
} else { |
} else { |
| 5693 |
} |
} |
| 5694 |
|
|
| 5695 |
## "after body" insertion mode |
## "after body" insertion mode |
| 5696 |
!!!parse-error (type => 'after body:#character'); |
!!!parse-error (type => 'after body:#character', token => $token); |
| 5697 |
|
|
| 5698 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 5699 |
## reprocess |
## reprocess |
| 5700 |
redo B; |
next B; |
| 5701 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 5702 |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
| 5703 |
!!!cp ('t303'); |
!!!cp ('t303'); |
| 5704 |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
!!!parse-error (type => 'after html:'.$token->{tag_name}, token => $token); |
| 5705 |
|
|
| 5706 |
## Reprocess in the "after body" insertion mode. |
## Reprocess in the "after body" insertion mode. |
| 5707 |
} else { |
} else { |
| 5709 |
} |
} |
| 5710 |
|
|
| 5711 |
## "after body" insertion mode |
## "after body" insertion mode |
| 5712 |
!!!parse-error (type => 'after body:'.$token->{tag_name}); |
!!!parse-error (type => 'after body:'.$token->{tag_name}, token => $token); |
| 5713 |
|
|
| 5714 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 5715 |
|
!!!ack-later; |
| 5716 |
## reprocess |
## reprocess |
| 5717 |
redo B; |
next B; |
| 5718 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5719 |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
| 5720 |
!!!cp ('t305'); |
!!!cp ('t305'); |
| 5721 |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
!!!parse-error (type => 'after html:/'.$token->{tag_name}, token => $token); |
| 5722 |
|
|
| 5723 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 5724 |
## Reprocess in the "after body" insertion mode. |
## Reprocess in the "after body" insertion mode. |
| 5730 |
if ($token->{tag_name} eq 'html') { |
if ($token->{tag_name} eq 'html') { |
| 5731 |
if (defined $self->{inner_html_node}) { |
if (defined $self->{inner_html_node}) { |
| 5732 |
!!!cp ('t307'); |
!!!cp ('t307'); |
| 5733 |
!!!parse-error (type => 'unmatched end tag:html'); |
!!!parse-error (type => 'unmatched end tag:html', token => $token); |
| 5734 |
## Ignore the token |
## Ignore the token |
| 5735 |
!!!next-token; |
!!!next-token; |
| 5736 |
redo B; |
next B; |
| 5737 |
} else { |
} else { |
| 5738 |
!!!cp ('t308'); |
!!!cp ('t308'); |
| 5739 |
$self->{insertion_mode} = AFTER_HTML_BODY_IM; |
$self->{insertion_mode} = AFTER_HTML_BODY_IM; |
| 5740 |
!!!next-token; |
!!!next-token; |
| 5741 |
redo B; |
next B; |
| 5742 |
} |
} |
| 5743 |
} else { |
} else { |
| 5744 |
!!!cp ('t309'); |
!!!cp ('t309'); |
| 5745 |
!!!parse-error (type => 'after body:/'.$token->{tag_name}); |
!!!parse-error (type => 'after body:/'.$token->{tag_name}, token => $token); |
| 5746 |
|
|
| 5747 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 5748 |
## reprocess |
## reprocess |
| 5749 |
redo B; |
next B; |
| 5750 |
} |
} |
| 5751 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 5752 |
|
!!!cp ('t309.2'); |
| 5753 |
|
## Stop parsing |
| 5754 |
|
last B; |
| 5755 |
} else { |
} else { |
| 5756 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 5757 |
} |
} |
| 5763 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 5764 |
!!!cp ('t310'); |
!!!cp ('t310'); |
| 5765 |
!!!next-token; |
!!!next-token; |
| 5766 |
redo B; |
next B; |
| 5767 |
} |
} |
| 5768 |
} |
} |
| 5769 |
|
|
| 5770 |
if ($token->{data} =~ s/^[^\x09\x0A\x0B\x0C\x20]+//) { |
if ($token->{data} =~ s/^[^\x09\x0A\x0B\x0C\x20]+//) { |
| 5771 |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5772 |
!!!cp ('t311'); |
!!!cp ('t311'); |
| 5773 |
!!!parse-error (type => 'in frameset:#character'); |
!!!parse-error (type => 'in frameset:#character', token => $token); |
| 5774 |
} elsif ($self->{insertion_mode} == AFTER_FRAMESET_IM) { |
} elsif ($self->{insertion_mode} == AFTER_FRAMESET_IM) { |
| 5775 |
!!!cp ('t312'); |
!!!cp ('t312'); |
| 5776 |
!!!parse-error (type => 'after frameset:#character'); |
!!!parse-error (type => 'after frameset:#character', token => $token); |
| 5777 |
} else { # "after html frameset" |
} else { # "after html frameset" |
| 5778 |
!!!cp ('t313'); |
!!!cp ('t313'); |
| 5779 |
!!!parse-error (type => 'after html:#character'); |
!!!parse-error (type => 'after html:#character', token => $token); |
| 5780 |
|
|
| 5781 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5782 |
## Reprocess in the "after frameset" insertion mode. |
## Reprocess in the "after frameset" insertion mode. |
| 5783 |
!!!parse-error (type => 'after frameset:#character'); |
!!!parse-error (type => 'after frameset:#character', token => $token); |
| 5784 |
} |
} |
| 5785 |
|
|
| 5786 |
## Ignore the token. |
## Ignore the token. |
| 5791 |
!!!cp ('t315'); |
!!!cp ('t315'); |
| 5792 |
!!!next-token; |
!!!next-token; |
| 5793 |
} |
} |
| 5794 |
redo B; |
next B; |
| 5795 |
} |
} |
| 5796 |
|
|
| 5797 |
die qq[$0: Character "$token->{data}"]; |
die qq[$0: Character "$token->{data}"]; |
| 5798 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 5799 |
if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
| 5800 |
!!!cp ('t316'); |
!!!cp ('t316'); |
| 5801 |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
!!!parse-error (type => 'after html:'.$token->{tag_name}, token => $token); |
| 5802 |
|
|
| 5803 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5804 |
## Process in the "after frameset" insertion mode. |
## Process in the "after frameset" insertion mode. |
| 5809 |
if ($token->{tag_name} eq 'frameset' and |
if ($token->{tag_name} eq 'frameset' and |
| 5810 |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5811 |
!!!cp ('t318'); |
!!!cp ('t318'); |
| 5812 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5813 |
|
!!!nack ('t318.1'); |
| 5814 |
!!!next-token; |
!!!next-token; |
| 5815 |
redo B; |
next B; |
| 5816 |
} elsif ($token->{tag_name} eq 'frame' and |
} elsif ($token->{tag_name} eq 'frame' and |
| 5817 |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5818 |
!!!cp ('t319'); |
!!!cp ('t319'); |
| 5819 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 5820 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5821 |
|
!!!ack ('t319.1'); |
| 5822 |
!!!next-token; |
!!!next-token; |
| 5823 |
redo B; |
next B; |
| 5824 |
} elsif ($token->{tag_name} eq 'noframes') { |
} elsif ($token->{tag_name} eq 'noframes') { |
| 5825 |
!!!cp ('t320'); |
!!!cp ('t320'); |
| 5826 |
## NOTE: As if in body. |
## NOTE: As if in body. |
| 5827 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert_to_current); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5828 |
redo B; |
next B; |
| 5829 |
} else { |
} else { |
| 5830 |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5831 |
!!!cp ('t321'); |
!!!cp ('t321'); |
| 5832 |
!!!parse-error (type => 'in frameset:'.$token->{tag_name}); |
!!!parse-error (type => 'in frameset:'.$token->{tag_name}, token => $token); |
| 5833 |
} else { |
} else { |
| 5834 |
!!!cp ('t322'); |
!!!cp ('t322'); |
| 5835 |
!!!parse-error (type => 'after frameset:'.$token->{tag_name}); |
!!!parse-error (type => 'after frameset:'.$token->{tag_name}, token => $token); |
| 5836 |
} |
} |
| 5837 |
## Ignore the token |
## Ignore the token |
| 5838 |
|
!!!nack ('t322.1'); |
| 5839 |
!!!next-token; |
!!!next-token; |
| 5840 |
redo B; |
next B; |
| 5841 |
} |
} |
| 5842 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5843 |
if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
| 5844 |
!!!cp ('t323'); |
!!!cp ('t323'); |
| 5845 |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
!!!parse-error (type => 'after html:/'.$token->{tag_name}, token => $token); |
| 5846 |
|
|
| 5847 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5848 |
## Process in the "after frameset" insertion mode. |
## Process in the "after frameset" insertion mode. |
| 5852 |
|
|
| 5853 |
if ($token->{tag_name} eq 'frameset' and |
if ($token->{tag_name} eq 'frameset' and |
| 5854 |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
$self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5855 |
if ($self->{open_elements}->[-1]->[1] eq 'html' and |
if ($self->{open_elements}->[-1]->[1] & HTML_EL and |
| 5856 |
@{$self->{open_elements}} == 1) { |
@{$self->{open_elements}} == 1) { |
| 5857 |
!!!cp ('t325'); |
!!!cp ('t325'); |
| 5858 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 5859 |
## Ignore the token |
## Ignore the token |
| 5860 |
!!!next-token; |
!!!next-token; |
| 5861 |
} else { |
} else { |
| 5865 |
} |
} |
| 5866 |
|
|
| 5867 |
if (not defined $self->{inner_html_node} and |
if (not defined $self->{inner_html_node} and |
| 5868 |
$self->{open_elements}->[-1]->[1] ne 'frameset') { |
not ($self->{open_elements}->[-1]->[1] & FRAMESET_EL)) { |
| 5869 |
!!!cp ('t327'); |
!!!cp ('t327'); |
| 5870 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5871 |
} else { |
} else { |
| 5872 |
!!!cp ('t328'); |
!!!cp ('t328'); |
| 5873 |
} |
} |
| 5874 |
redo B; |
next B; |
| 5875 |
} elsif ($token->{tag_name} eq 'html' and |
} elsif ($token->{tag_name} eq 'html' and |
| 5876 |
$self->{insertion_mode} == AFTER_FRAMESET_IM) { |
$self->{insertion_mode} == AFTER_FRAMESET_IM) { |
| 5877 |
!!!cp ('t329'); |
!!!cp ('t329'); |
| 5878 |
$self->{insertion_mode} = AFTER_HTML_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_HTML_FRAMESET_IM; |
| 5879 |
!!!next-token; |
!!!next-token; |
| 5880 |
redo B; |
next B; |
| 5881 |
} else { |
} else { |
| 5882 |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5883 |
!!!cp ('t330'); |
!!!cp ('t330'); |
| 5884 |
!!!parse-error (type => 'in frameset:/'.$token->{tag_name}); |
!!!parse-error (type => 'in frameset:/'.$token->{tag_name}, token => $token); |
| 5885 |
} else { |
} else { |
| 5886 |
!!!cp ('t331'); |
!!!cp ('t331'); |
| 5887 |
!!!parse-error (type => 'after frameset:/'.$token->{tag_name}); |
!!!parse-error (type => 'after frameset:/'.$token->{tag_name}, token => $token); |
| 5888 |
} |
} |
| 5889 |
## Ignore the token |
## Ignore the token |
| 5890 |
!!!next-token; |
!!!next-token; |
| 5891 |
redo B; |
next B; |
| 5892 |
|
} |
| 5893 |
|
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 5894 |
|
unless ($self->{open_elements}->[-1]->[1] & HTML_EL and |
| 5895 |
|
@{$self->{open_elements}} == 1) { # redundant, maybe |
| 5896 |
|
!!!cp ('t331.1'); |
| 5897 |
|
!!!parse-error (type => 'in body:#eof', token => $token); |
| 5898 |
|
} else { |
| 5899 |
|
!!!cp ('t331.2'); |
| 5900 |
} |
} |
| 5901 |
|
|
| 5902 |
|
## Stop parsing |
| 5903 |
|
last B; |
| 5904 |
} else { |
} else { |
| 5905 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 5906 |
} |
} |
| 5915 |
if ($token->{tag_name} eq 'script') { |
if ($token->{tag_name} eq 'script') { |
| 5916 |
!!!cp ('t332'); |
!!!cp ('t332'); |
| 5917 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5918 |
$script_start_tag->($insert); |
$script_start_tag->(); |
| 5919 |
redo B; |
next B; |
| 5920 |
} elsif ($token->{tag_name} eq 'style') { |
} elsif ($token->{tag_name} eq 'style') { |
| 5921 |
!!!cp ('t333'); |
!!!cp ('t333'); |
| 5922 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5923 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5924 |
redo B; |
next B; |
| 5925 |
} elsif ({ |
} elsif ({ |
| 5926 |
base => 1, link => 1, |
base => 1, link => 1, |
| 5927 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 5928 |
!!!cp ('t334'); |
!!!cp ('t334'); |
| 5929 |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
| 5930 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 5931 |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 5932 |
|
!!!ack ('t334.1'); |
| 5933 |
!!!next-token; |
!!!next-token; |
| 5934 |
redo B; |
next B; |
| 5935 |
} elsif ($token->{tag_name} eq 'meta') { |
} elsif ($token->{tag_name} eq 'meta') { |
| 5936 |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
| 5937 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 5938 |
my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 5939 |
|
|
| 5940 |
unless ($self->{confident}) { |
unless ($self->{confident}) { |
| 5941 |
if ($token->{attributes}->{charset}) { ## TODO: And if supported |
if ($token->{attributes}->{charset}) { ## TODO: And if supported |
| 5942 |
!!!cp ('t335'); |
!!!cp ('t335'); |
| 5943 |
$self->{change_encoding} |
$self->{change_encoding} |
| 5944 |
->($self, $token->{attributes}->{charset}->{value}); |
->($self, $token->{attributes}->{charset}->{value}, $token); |
| 5945 |
|
|
| 5946 |
$meta_el->[0]->get_attribute_node_ns (undef, 'charset') |
$meta_el->[0]->get_attribute_node_ns (undef, 'charset') |
| 5947 |
->set_user_data (manakai_has_reference => |
->set_user_data (manakai_has_reference => |
| 5956 |
([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) { |
([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) { |
| 5957 |
!!!cp ('t336'); |
!!!cp ('t336'); |
| 5958 |
$self->{change_encoding} |
$self->{change_encoding} |
| 5959 |
->($self, defined $1 ? $1 : defined $2 ? $2 : $3); |
->($self, defined $1 ? $1 : defined $2 ? $2 : $3, $token); |
| 5960 |
$meta_el->[0]->get_attribute_node_ns (undef, 'content') |
$meta_el->[0]->get_attribute_node_ns (undef, 'content') |
| 5961 |
->set_user_data (manakai_has_reference => |
->set_user_data (manakai_has_reference => |
| 5962 |
$token->{attributes}->{content} |
$token->{attributes}->{content} |
| 5980 |
} |
} |
| 5981 |
} |
} |
| 5982 |
|
|
| 5983 |
|
!!!ack ('t338.1'); |
| 5984 |
!!!next-token; |
!!!next-token; |
| 5985 |
redo B; |
next B; |
| 5986 |
} elsif ($token->{tag_name} eq 'title') { |
} elsif ($token->{tag_name} eq 'title') { |
| 5987 |
!!!cp ('t341'); |
!!!cp ('t341'); |
|
!!!parse-error (type => 'in body:title'); |
|
| 5988 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5989 |
$parse_rcdata->(RCDATA_CONTENT_MODEL, sub { |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
| 5990 |
if (defined $self->{head_element}) { |
next B; |
|
!!!cp ('t339'); |
|
|
$self->{head_element}->append_child ($_[0]); |
|
|
} else { |
|
|
!!!cp ('t340'); |
|
|
$insert->($_[0]); |
|
|
} |
|
|
}); |
|
|
redo B; |
|
| 5991 |
} elsif ($token->{tag_name} eq 'body') { |
} elsif ($token->{tag_name} eq 'body') { |
| 5992 |
!!!parse-error (type => 'in body:body'); |
!!!parse-error (type => 'in body:body', token => $token); |
| 5993 |
|
|
| 5994 |
if (@{$self->{open_elements}} == 1 or |
if (@{$self->{open_elements}} == 1 or |
| 5995 |
$self->{open_elements}->[1]->[1] ne 'body') { |
not ($self->{open_elements}->[1]->[1] & BODY_EL)) { |
| 5996 |
!!!cp ('t342'); |
!!!cp ('t342'); |
| 5997 |
## Ignore the token |
## Ignore the token |
| 5998 |
} else { |
} else { |
| 6006 |
} |
} |
| 6007 |
} |
} |
| 6008 |
} |
} |
| 6009 |
|
!!!nack ('t343.1'); |
| 6010 |
!!!next-token; |
!!!next-token; |
| 6011 |
redo B; |
next B; |
| 6012 |
} elsif ({ |
} elsif ({ |
| 6013 |
address => 1, blockquote => 1, center => 1, dir => 1, |
address => 1, blockquote => 1, center => 1, dir => 1, |
| 6014 |
div => 1, dl => 1, fieldset => 1, |
div => 1, dl => 1, fieldset => 1, |
| 6015 |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
| 6016 |
listing => 1, menu => 1, ol => 1, p => 1, ul => 1, |
menu => 1, ol => 1, p => 1, ul => 1, |
| 6017 |
pre => 1, |
pre => 1, listing => 1, |
| 6018 |
|
form => 1, |
| 6019 |
|
table => 1, |
| 6020 |
|
hr => 1, |
| 6021 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6022 |
|
if ($token->{tag_name} eq 'form' and defined $self->{form_element}) { |
| 6023 |
|
!!!cp ('t350'); |
| 6024 |
|
!!!parse-error (type => 'in form:form', token => $token); |
| 6025 |
|
## Ignore the token |
| 6026 |
|
!!!nack ('t350.1'); |
| 6027 |
|
!!!next-token; |
| 6028 |
|
next B; |
| 6029 |
|
} |
| 6030 |
|
|
| 6031 |
## has a p element in scope |
## has a p element in scope |
| 6032 |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
| 6033 |
if ($_->[1] eq 'p') { |
if ($_->[1] & P_EL) { |
| 6034 |
!!!cp ('t344'); |
!!!cp ('t344'); |
| 6035 |
!!!back-token; |
!!!back-token; # <form> |
| 6036 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p', |
| 6037 |
redo B; |
line => $token->{line}, column => $token->{column}}; |
| 6038 |
} elsif ({ |
next B; |
| 6039 |
table => 1, caption => 1, td => 1, th => 1, |
} elsif ($_->[1] & SCOPING_EL) { |
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
| 6040 |
!!!cp ('t345'); |
!!!cp ('t345'); |
| 6041 |
last INSCOPE; |
last INSCOPE; |
| 6042 |
} |
} |
| 6043 |
} # INSCOPE |
} # INSCOPE |
| 6044 |
|
|
| 6045 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6046 |
if ($token->{tag_name} eq 'pre') { |
if ($token->{tag_name} eq 'pre' or $token->{tag_name} eq 'listing') { |
| 6047 |
|
!!!nack ('t346.1'); |
| 6048 |
!!!next-token; |
!!!next-token; |
| 6049 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 6050 |
$token->{data} =~ s/^\x0A//; |
$token->{data} =~ s/^\x0A//; |
| 6057 |
} else { |
} else { |
| 6058 |
!!!cp ('t348'); |
!!!cp ('t348'); |
| 6059 |
} |
} |
| 6060 |
} else { |
} elsif ($token->{tag_name} eq 'form') { |
| 6061 |
!!!cp ('t347'); |
!!!cp ('t347.1'); |
| 6062 |
|
$self->{form_element} = $self->{open_elements}->[-1]->[0]; |
| 6063 |
|
|
| 6064 |
|
!!!nack ('t347.2'); |
| 6065 |
!!!next-token; |
!!!next-token; |
| 6066 |
} |
} elsif ($token->{tag_name} eq 'table') { |
| 6067 |
redo B; |
!!!cp ('t382'); |
| 6068 |
} elsif ($token->{tag_name} eq 'form') { |
push @{$open_tables}, [$self->{open_elements}->[-1]->[0]]; |
| 6069 |
if (defined $self->{form_element}) { |
|
| 6070 |
!!!cp ('t350'); |
$self->{insertion_mode} = IN_TABLE_IM; |
| 6071 |
!!!parse-error (type => 'in form:form'); |
|
| 6072 |
## Ignore the token |
!!!nack ('t382.1'); |
| 6073 |
|
!!!next-token; |
| 6074 |
|
} elsif ($token->{tag_name} eq 'hr') { |
| 6075 |
|
!!!cp ('t386'); |
| 6076 |
|
pop @{$self->{open_elements}}; |
| 6077 |
|
|
| 6078 |
|
!!!nack ('t386.1'); |
| 6079 |
!!!next-token; |
!!!next-token; |
|
redo B; |
|
| 6080 |
} else { |
} else { |
| 6081 |
## has a p element in scope |
!!!nack ('t347.1'); |
|
INSCOPE: for (reverse @{$self->{open_elements}}) { |
|
|
if ($_->[1] eq 'p') { |
|
|
!!!cp ('t351'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
|
|
redo B; |
|
|
} elsif ({ |
|
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
|
!!!cp ('t352'); |
|
|
last INSCOPE; |
|
|
} |
|
|
} # INSCOPE |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
$self->{form_element} = $self->{open_elements}->[-1]->[0]; |
|
| 6082 |
!!!next-token; |
!!!next-token; |
|
redo B; |
|
| 6083 |
} |
} |
| 6084 |
} elsif ($token->{tag_name} eq 'li') { |
next B; |
| 6085 |
|
} elsif ({li => 1, dt => 1, dd => 1}->{$token->{tag_name}}) { |
| 6086 |
## has a p element in scope |
## has a p element in scope |
| 6087 |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
| 6088 |
if ($_->[1] eq 'p') { |
if ($_->[1] & P_EL) { |
| 6089 |
!!!cp ('t353'); |
!!!cp ('t353'); |
| 6090 |
!!!back-token; |
!!!back-token; # <x> |
| 6091 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p', |
| 6092 |
redo B; |
line => $token->{line}, column => $token->{column}}; |
| 6093 |
} elsif ({ |
next B; |
| 6094 |
table => 1, caption => 1, td => 1, th => 1, |
} elsif ($_->[1] & SCOPING_EL) { |
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
| 6095 |
!!!cp ('t354'); |
!!!cp ('t354'); |
| 6096 |
last INSCOPE; |
last INSCOPE; |
| 6097 |
} |
} |
| 6100 |
## Step 1 |
## Step 1 |
| 6101 |
my $i = -1; |
my $i = -1; |
| 6102 |
my $node = $self->{open_elements}->[$i]; |
my $node = $self->{open_elements}->[$i]; |
| 6103 |
|
my $li_or_dtdd = {li => {li => 1}, |
| 6104 |
|
dt => {dt => 1, dd => 1}, |
| 6105 |
|
dd => {dt => 1, dd => 1}}->{$token->{tag_name}}; |
| 6106 |
LI: { |
LI: { |
| 6107 |
## Step 2 |
## Step 2 |
| 6108 |
if ($node->[1] eq 'li') { |
if ($li_or_dtdd->{$node->[0]->manakai_local_name}) { |
| 6109 |
if ($i != -1) { |
if ($i != -1) { |
| 6110 |
!!!cp ('t355'); |
!!!cp ('t355'); |
| 6111 |
!!!parse-error (type => 'end tag missing:'. |
!!!parse-error (type => 'not closed', |
| 6112 |
$self->{open_elements}->[-1]->[1]); |
value => $self->{open_elements}->[-1]->[0] |
| 6113 |
|
->manakai_local_name, |
| 6114 |
|
token => $token); |
| 6115 |
} else { |
} else { |
| 6116 |
!!!cp ('t356'); |
!!!cp ('t356'); |
| 6117 |
} |
} |
| 6122 |
} |
} |
| 6123 |
|
|
| 6124 |
## Step 3 |
## Step 3 |
| 6125 |
if (not $formatting_category->{$node->[1]} and |
if (not ($node->[1] & FORMATTING_EL) and |
| 6126 |
#not $phrasing_category->{$node->[1]} and |
#not $phrasing_category->{$node->[1]} and |
| 6127 |
($special_category->{$node->[1]} or |
($node->[1] & SPECIAL_EL or |
| 6128 |
$scoping_category->{$node->[1]}) and |
$node->[1] & SCOPING_EL) and |
| 6129 |
$node->[1] ne 'address' and $node->[1] ne 'div') { |
not ($node->[1] & ADDRESS_EL) and |
| 6130 |
|
not ($node->[1] & DIV_EL)) { |
| 6131 |
!!!cp ('t358'); |
!!!cp ('t358'); |
| 6132 |
last LI; |
last LI; |
| 6133 |
} |
} |
| 6139 |
redo LI; |
redo LI; |
| 6140 |
} # LI |
} # LI |
| 6141 |
|
|
| 6142 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6143 |
|
!!!nack ('t359.1'); |
| 6144 |
!!!next-token; |
!!!next-token; |
| 6145 |
redo B; |
next B; |
|
} elsif ($token->{tag_name} eq 'dd' or $token->{tag_name} eq 'dt') { |
|
|
## has a p element in scope |
|
|
INSCOPE: for (reverse @{$self->{open_elements}}) { |
|
|
if ($_->[1] eq 'p') { |
|
|
!!!cp ('t360'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
|
|
redo B; |
|
|
} elsif ({ |
|
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
|
!!!cp ('t361'); |
|
|
last INSCOPE; |
|
|
} |
|
|
} # INSCOPE |
|
|
|
|
|
## Step 1 |
|
|
my $i = -1; |
|
|
my $node = $self->{open_elements}->[$i]; |
|
|
LI: { |
|
|
## Step 2 |
|
|
if ($node->[1] eq 'dt' or $node->[1] eq 'dd') { |
|
|
if ($i != -1) { |
|
|
!!!cp ('t362'); |
|
|
!!!parse-error (type => 'end tag missing:'. |
|
|
$self->{open_elements}->[-1]->[1]); |
|
|
} else { |
|
|
!!!cp ('t363'); |
|
|
} |
|
|
splice @{$self->{open_elements}}, $i; |
|
|
last LI; |
|
|
} else { |
|
|
!!!cp ('t364'); |
|
|
} |
|
|
|
|
|
## Step 3 |
|
|
if (not $formatting_category->{$node->[1]} and |
|
|
#not $phrasing_category->{$node->[1]} and |
|
|
($special_category->{$node->[1]} or |
|
|
$scoping_category->{$node->[1]}) and |
|
|
$node->[1] ne 'address' and $node->[1] ne 'div') { |
|
|
!!!cp ('t365'); |
|
|
last LI; |
|
|
} |
|
|
|
|
|
!!!cp ('t366'); |
|
|
## Step 4 |
|
|
$i--; |
|
|
$node = $self->{open_elements}->[$i]; |
|
|
redo LI; |
|
|
} # LI |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
!!!next-token; |
|
|
redo B; |
|
| 6146 |
} elsif ($token->{tag_name} eq 'plaintext') { |
} elsif ($token->{tag_name} eq 'plaintext') { |
| 6147 |
## has a p element in scope |
## has a p element in scope |
| 6148 |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
| 6149 |
if ($_->[1] eq 'p') { |
if ($_->[1] & P_EL) { |
| 6150 |
!!!cp ('t367'); |
!!!cp ('t367'); |
| 6151 |
!!!back-token; |
!!!back-token; # <plaintext> |
| 6152 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p', |
| 6153 |
redo B; |
line => $token->{line}, column => $token->{column}}; |
| 6154 |
} elsif ({ |
next B; |
| 6155 |
table => 1, caption => 1, td => 1, th => 1, |
} elsif ($_->[1] & SCOPING_EL) { |
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
| 6156 |
!!!cp ('t368'); |
!!!cp ('t368'); |
| 6157 |
last INSCOPE; |
last INSCOPE; |
| 6158 |
} |
} |
| 6159 |
} # INSCOPE |
} # INSCOPE |
| 6160 |
|
|
| 6161 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6162 |
|
|
| 6163 |
$self->{content_model} = PLAINTEXT_CONTENT_MODEL; |
$self->{content_model} = PLAINTEXT_CONTENT_MODEL; |
| 6164 |
|
|
| 6165 |
|
!!!nack ('t368.1'); |
| 6166 |
!!!next-token; |
!!!next-token; |
| 6167 |
redo B; |
next B; |
| 6168 |
} elsif ($token->{tag_name} eq 'a') { |
} elsif ($token->{tag_name} eq 'a') { |
| 6169 |
AFE: for my $i (reverse 0..$#$active_formatting_elements) { |
AFE: for my $i (reverse 0..$#$active_formatting_elements) { |
| 6170 |
my $node = $active_formatting_elements->[$i]; |
my $node = $active_formatting_elements->[$i]; |
| 6171 |
if ($node->[1] eq 'a') { |
if ($node->[1] & A_EL) { |
| 6172 |
!!!cp ('t371'); |
!!!cp ('t371'); |
| 6173 |
!!!parse-error (type => 'in a:a'); |
!!!parse-error (type => 'in a:a', token => $token); |
| 6174 |
|
|
| 6175 |
!!!back-token; |
!!!back-token; # <a> |
| 6176 |
$token = {type => END_TAG_TOKEN, tag_name => 'a'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'a', |
| 6177 |
$formatting_end_tag->($token->{tag_name}); |
line => $token->{line}, column => $token->{column}}; |
| 6178 |
|
$formatting_end_tag->($token); |
| 6179 |
|
|
| 6180 |
AFE2: for (reverse 0..$#$active_formatting_elements) { |
AFE2: for (reverse 0..$#$active_formatting_elements) { |
| 6181 |
if ($active_formatting_elements->[$_]->[0] eq $node->[0]) { |
if ($active_formatting_elements->[$_]->[0] eq $node->[0]) { |
| 6200 |
|
|
| 6201 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6202 |
|
|
| 6203 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6204 |
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
| 6205 |
|
|
| 6206 |
|
!!!nack ('t374.1'); |
| 6207 |
!!!next-token; |
!!!next-token; |
| 6208 |
redo B; |
next B; |
|
} elsif ({ |
|
|
b => 1, big => 1, em => 1, font => 1, i => 1, |
|
|
s => 1, small => 1, strile => 1, |
|
|
strong => 1, tt => 1, u => 1, |
|
|
}->{$token->{tag_name}}) { |
|
|
!!!cp ('t375'); |
|
|
$reconstruct_active_formatting_elements->($insert_to_current); |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
| 6209 |
} elsif ($token->{tag_name} eq 'nobr') { |
} elsif ($token->{tag_name} eq 'nobr') { |
| 6210 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6211 |
|
|
| 6212 |
## has a |nobr| element in scope |
## has a |nobr| element in scope |
| 6213 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6214 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6215 |
if ($node->[1] eq 'nobr') { |
if ($node->[1] & NOBR_EL) { |
| 6216 |
!!!cp ('t376'); |
!!!cp ('t376'); |
| 6217 |
!!!parse-error (type => 'in nobr:nobr'); |
!!!parse-error (type => 'in nobr:nobr', token => $token); |
| 6218 |
!!!back-token; |
!!!back-token; # <nobr> |
| 6219 |
$token = {type => END_TAG_TOKEN, tag_name => 'nobr'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'nobr', |
| 6220 |
redo B; |
line => $token->{line}, column => $token->{column}}; |
| 6221 |
} elsif ({ |
next B; |
| 6222 |
table => 1, caption => 1, td => 1, th => 1, |
} elsif ($node->[1] & SCOPING_EL) { |
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 6223 |
!!!cp ('t377'); |
!!!cp ('t377'); |
| 6224 |
last INSCOPE; |
last INSCOPE; |
| 6225 |
} |
} |
| 6226 |
} # INSCOPE |
} # INSCOPE |
| 6227 |
|
|
| 6228 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6229 |
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
| 6230 |
|
|
| 6231 |
|
!!!nack ('t377.1'); |
| 6232 |
!!!next-token; |
!!!next-token; |
| 6233 |
redo B; |
next B; |
| 6234 |
} elsif ($token->{tag_name} eq 'button') { |
} elsif ($token->{tag_name} eq 'button') { |
| 6235 |
## has a button element in scope |
## has a button element in scope |
| 6236 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6237 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6238 |
if ($node->[1] eq 'button') { |
if ($node->[1] & BUTTON_EL) { |
| 6239 |
!!!cp ('t378'); |
!!!cp ('t378'); |
| 6240 |
!!!parse-error (type => 'in button:button'); |
!!!parse-error (type => 'in button:button', token => $token); |
| 6241 |
!!!back-token; |
!!!back-token; # <button> |
| 6242 |
$token = {type => END_TAG_TOKEN, tag_name => 'button'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'button', |
| 6243 |
redo B; |
line => $token->{line}, column => $token->{column}}; |
| 6244 |
} elsif ({ |
next B; |
| 6245 |
table => 1, caption => 1, td => 1, th => 1, |
} elsif ($node->[1] & SCOPING_EL) { |
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 6246 |
!!!cp ('t379'); |
!!!cp ('t379'); |
| 6247 |
last INSCOPE; |
last INSCOPE; |
| 6248 |
} |
} |
| 6250 |
|
|
| 6251 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6252 |
|
|
| 6253 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6254 |
|
|
| 6255 |
## TODO: associate with $self->{form_element} if defined |
## TODO: associate with $self->{form_element} if defined |
| 6256 |
|
|
| 6257 |
push @$active_formatting_elements, ['#marker', '']; |
push @$active_formatting_elements, ['#marker', '']; |
| 6258 |
|
|
| 6259 |
|
!!!nack ('t379.1'); |
| 6260 |
!!!next-token; |
!!!next-token; |
| 6261 |
redo B; |
next B; |
|
} elsif ($token->{tag_name} eq 'marquee' or |
|
|
$token->{tag_name} eq 'object') { |
|
|
!!!cp ('t380'); |
|
|
$reconstruct_active_formatting_elements->($insert_to_current); |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
push @$active_formatting_elements, ['#marker', '']; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
|
} elsif ($token->{tag_name} eq 'xmp') { |
|
|
!!!cp ('t381'); |
|
|
$reconstruct_active_formatting_elements->($insert_to_current); |
|
|
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
|
|
redo B; |
|
|
} elsif ($token->{tag_name} eq 'table') { |
|
|
## has a p element in scope |
|
|
INSCOPE: for (reverse @{$self->{open_elements}}) { |
|
|
if ($_->[1] eq 'p') { |
|
|
!!!cp ('t382'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
|
|
redo B; |
|
|
} elsif ({ |
|
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
|
!!!cp ('t383'); |
|
|
last INSCOPE; |
|
|
} |
|
|
} # INSCOPE |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
|
|
|
$self->{insertion_mode} = IN_TABLE_IM; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
| 6262 |
} elsif ({ |
} elsif ({ |
| 6263 |
area => 1, basefont => 1, bgsound => 1, br => 1, |
xmp => 1, |
| 6264 |
embed => 1, img => 1, param => 1, spacer => 1, wbr => 1, |
iframe => 1, |
| 6265 |
image => 1, |
noembed => 1, |
| 6266 |
|
noframes => 1, |
| 6267 |
|
noscript => 0, ## TODO: 1 if scripting is enabled |
| 6268 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6269 |
if ($token->{tag_name} eq 'image') { |
if ($token->{tag_name} eq 'xmp') { |
| 6270 |
!!!cp ('t384'); |
!!!cp ('t381'); |
| 6271 |
!!!parse-error (type => 'image'); |
$reconstruct_active_formatting_elements->($insert_to_current); |
|
$token->{tag_name} = 'img'; |
|
| 6272 |
} else { |
} else { |
| 6273 |
!!!cp ('t385'); |
!!!cp ('t399'); |
| 6274 |
} |
} |
| 6275 |
|
## NOTE: There is an "as if in body" code clone. |
| 6276 |
## NOTE: There is an "as if <br>" code clone. |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 6277 |
$reconstruct_active_formatting_elements->($insert_to_current); |
next B; |
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
pop @{$self->{open_elements}}; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
|
} elsif ($token->{tag_name} eq 'hr') { |
|
|
## has a p element in scope |
|
|
INSCOPE: for (reverse @{$self->{open_elements}}) { |
|
|
if ($_->[1] eq 'p') { |
|
|
!!!cp ('t386'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
|
|
redo B; |
|
|
} elsif ({ |
|
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$_->[1]}) { |
|
|
!!!cp ('t387'); |
|
|
last INSCOPE; |
|
|
} |
|
|
} # INSCOPE |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
pop @{$self->{open_elements}}; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
|
} elsif ($token->{tag_name} eq 'input') { |
|
|
!!!cp ('t388'); |
|
|
$reconstruct_active_formatting_elements->($insert_to_current); |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
## TODO: associate with $self->{form_element} if defined |
|
|
pop @{$self->{open_elements}}; |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
| 6278 |
} elsif ($token->{tag_name} eq 'isindex') { |
} elsif ($token->{tag_name} eq 'isindex') { |
| 6279 |
!!!parse-error (type => 'isindex'); |
!!!parse-error (type => 'isindex', token => $token); |
| 6280 |
|
|
| 6281 |
if (defined $self->{form_element}) { |
if (defined $self->{form_element}) { |
| 6282 |
!!!cp ('t389'); |
!!!cp ('t389'); |
| 6283 |
## Ignore the token |
## Ignore the token |
| 6284 |
|
!!!nack ('t389'); ## NOTE: Not acknowledged. |
| 6285 |
!!!next-token; |
!!!next-token; |
| 6286 |
redo B; |
next B; |
| 6287 |
} else { |
} else { |
| 6288 |
my $at = $token->{attributes}; |
my $at = $token->{attributes}; |
| 6289 |
my $form_attrs; |
my $form_attrs; |
| 6294 |
delete $at->{prompt}; |
delete $at->{prompt}; |
| 6295 |
my @tokens = ( |
my @tokens = ( |
| 6296 |
{type => START_TAG_TOKEN, tag_name => 'form', |
{type => START_TAG_TOKEN, tag_name => 'form', |
| 6297 |
attributes => $form_attrs}, |
attributes => $form_attrs, |
| 6298 |
{type => START_TAG_TOKEN, tag_name => 'hr'}, |
line => $token->{line}, column => $token->{column}}, |
| 6299 |
{type => START_TAG_TOKEN, tag_name => 'p'}, |
{type => START_TAG_TOKEN, tag_name => 'hr', |
| 6300 |
{type => START_TAG_TOKEN, tag_name => 'label'}, |
line => $token->{line}, column => $token->{column}}, |
| 6301 |
|
{type => START_TAG_TOKEN, tag_name => 'p', |
| 6302 |
|
line => $token->{line}, column => $token->{column}}, |
| 6303 |
|
{type => START_TAG_TOKEN, tag_name => 'label', |
| 6304 |
|
line => $token->{line}, column => $token->{column}}, |
| 6305 |
); |
); |
| 6306 |
if ($prompt_attr) { |
if ($prompt_attr) { |
| 6307 |
!!!cp ('t390'); |
!!!cp ('t390'); |
| 6308 |
push @tokens, {type => CHARACTER_TOKEN, data => $prompt_attr->{value}}; |
push @tokens, {type => CHARACTER_TOKEN, data => $prompt_attr->{value}, |
| 6309 |
|
#line => $token->{line}, column => $token->{column}, |
| 6310 |
|
}; |
| 6311 |
} else { |
} else { |
| 6312 |
!!!cp ('t391'); |
!!!cp ('t391'); |
| 6313 |
push @tokens, {type => CHARACTER_TOKEN, |
push @tokens, {type => CHARACTER_TOKEN, |
| 6314 |
data => 'This is a searchable index. Insert your search keywords here: '}; # SHOULD |
data => 'This is a searchable index. Insert your search keywords here: ', |
| 6315 |
|
#line => $token->{line}, column => $token->{column}, |
| 6316 |
|
}; # SHOULD |
| 6317 |
## TODO: make this configurable |
## TODO: make this configurable |
| 6318 |
} |
} |
| 6319 |
push @tokens, |
push @tokens, |
| 6320 |
{type => START_TAG_TOKEN, tag_name => 'input', attributes => $at}, |
{type => START_TAG_TOKEN, tag_name => 'input', attributes => $at, |
| 6321 |
|
line => $token->{line}, column => $token->{column}}, |
| 6322 |
#{type => CHARACTER_TOKEN, data => ''}, # SHOULD |
#{type => CHARACTER_TOKEN, data => ''}, # SHOULD |
| 6323 |
{type => END_TAG_TOKEN, tag_name => 'label'}, |
{type => END_TAG_TOKEN, tag_name => 'label', |
| 6324 |
{type => END_TAG_TOKEN, tag_name => 'p'}, |
line => $token->{line}, column => $token->{column}}, |
| 6325 |
{type => START_TAG_TOKEN, tag_name => 'hr'}, |
{type => END_TAG_TOKEN, tag_name => 'p', |
| 6326 |
{type => END_TAG_TOKEN, tag_name => 'form'}; |
line => $token->{line}, column => $token->{column}}, |
| 6327 |
$token = shift @tokens; |
{type => START_TAG_TOKEN, tag_name => 'hr', |
| 6328 |
|
line => $token->{line}, column => $token->{column}}, |
| 6329 |
|
{type => END_TAG_TOKEN, tag_name => 'form', |
| 6330 |
|
line => $token->{line}, column => $token->{column}}; |
| 6331 |
|
!!!nack ('t391.1'); ## NOTE: Not acknowledged. |
| 6332 |
!!!back-token (@tokens); |
!!!back-token (@tokens); |
| 6333 |
redo B; |
!!!next-token; |
| 6334 |
|
next B; |
| 6335 |
} |
} |
| 6336 |
} elsif ($token->{tag_name} eq 'textarea') { |
} elsif ($token->{tag_name} eq 'textarea') { |
| 6337 |
my $tag_name = $token->{tag_name}; |
my $tag_name = $token->{tag_name}; |
| 6338 |
my $el; |
my $el; |
| 6339 |
!!!create-element ($el, $token->{tag_name}, $token->{attributes}); |
!!!create-element ($el, $HTML_NS, $token->{tag_name}, $token->{attributes}, $token); |
| 6340 |
|
|
| 6341 |
## TODO: $self->{form_element} if defined |
## TODO: $self->{form_element} if defined |
| 6342 |
$self->{content_model} = RCDATA_CONTENT_MODEL; |
$self->{content_model} = RCDATA_CONTENT_MODEL; |
| 6345 |
$insert->($el); |
$insert->($el); |
| 6346 |
|
|
| 6347 |
my $text = ''; |
my $text = ''; |
| 6348 |
|
!!!nack ('t392.1'); |
| 6349 |
!!!next-token; |
!!!next-token; |
| 6350 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 6351 |
$token->{data} =~ s/^\x0A//; |
$token->{data} =~ s/^\x0A//; |
| 6376 |
## Ignore the token |
## Ignore the token |
| 6377 |
} else { |
} else { |
| 6378 |
!!!cp ('t398'); |
!!!cp ('t398'); |
| 6379 |
!!!parse-error (type => 'in RCDATA:#'.$token->{type}); |
!!!parse-error (type => 'in RCDATA:#'.$token->{type}, token => $token); |
| 6380 |
} |
} |
| 6381 |
!!!next-token; |
!!!next-token; |
| 6382 |
redo B; |
next B; |
| 6383 |
} elsif ({ |
} elsif ($token->{tag_name} eq 'math' or |
| 6384 |
iframe => 1, |
$token->{tag_name} eq 'svg') { |
|
noembed => 1, |
|
|
noframes => 1, |
|
|
noscript => 0, ## TODO: 1 if scripting is enabled |
|
|
}->{$token->{tag_name}}) { |
|
|
!!!cp ('t399'); |
|
|
## NOTE: There is an "as if in body" code clone. |
|
|
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
|
|
redo B; |
|
|
} elsif ($token->{tag_name} eq 'select') { |
|
|
!!!cp ('t400'); |
|
| 6385 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6386 |
|
|
| 6387 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-f ($token->{tag_name} eq 'math' ? $MML_NS : $SVG_NS, $token); |
|
|
|
|
## TODO: associate with $self->{form_element} if defined |
|
| 6388 |
|
|
| 6389 |
$self->{insertion_mode} = IN_SELECT_IM; |
if ($self->{self_closing}) { |
| 6390 |
|
pop @{$self->{open_elements}}; |
| 6391 |
|
!!!ack ('t398.1'); |
| 6392 |
|
} else { |
| 6393 |
|
!!!cp ('t398.2'); |
| 6394 |
|
$self->{insertion_mode} |= IN_FOREIGN_CONTENT_IM; |
| 6395 |
|
## NOTE: |<body><math><mi><svg>| -> "in foreign content" insertion |
| 6396 |
|
## mode, "in body" (not "in foreign content") secondary insertion |
| 6397 |
|
## mode, maybe. |
| 6398 |
|
} |
| 6399 |
|
|
| 6400 |
!!!next-token; |
!!!next-token; |
| 6401 |
redo B; |
next B; |
| 6402 |
} elsif ({ |
} elsif ({ |
| 6403 |
caption => 1, col => 1, colgroup => 1, frame => 1, |
caption => 1, col => 1, colgroup => 1, frame => 1, |
| 6404 |
frameset => 1, head => 1, option => 1, optgroup => 1, |
frameset => 1, head => 1, option => 1, optgroup => 1, |
| 6406 |
thead => 1, tr => 1, |
thead => 1, tr => 1, |
| 6407 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6408 |
!!!cp ('t401'); |
!!!cp ('t401'); |
| 6409 |
!!!parse-error (type => 'in body:'.$token->{tag_name}); |
!!!parse-error (type => 'in body:'.$token->{tag_name}, token => $token); |
| 6410 |
## Ignore the token |
## Ignore the token |
| 6411 |
|
!!!nack ('t401.1'); ## NOTE: |<col/>| or |<frame/>| here is an error. |
| 6412 |
!!!next-token; |
!!!next-token; |
| 6413 |
redo B; |
next B; |
| 6414 |
|
|
| 6415 |
## ISSUE: An issue on HTML5 new elements in the spec. |
## ISSUE: An issue on HTML5 new elements in the spec. |
| 6416 |
} else { |
} else { |
| 6417 |
!!!cp ('t402'); |
if ($token->{tag_name} eq 'image') { |
| 6418 |
|
!!!cp ('t384'); |
| 6419 |
|
!!!parse-error (type => 'image', token => $token); |
| 6420 |
|
$token->{tag_name} = 'img'; |
| 6421 |
|
} else { |
| 6422 |
|
!!!cp ('t385'); |
| 6423 |
|
} |
| 6424 |
|
|
| 6425 |
|
## NOTE: There is an "as if <br>" code clone. |
| 6426 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6427 |
|
|
| 6428 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 6429 |
|
|
| 6430 |
|
if ({ |
| 6431 |
|
applet => 1, marquee => 1, object => 1, |
| 6432 |
|
}->{$token->{tag_name}}) { |
| 6433 |
|
!!!cp ('t380'); |
| 6434 |
|
push @$active_formatting_elements, ['#marker', '']; |
| 6435 |
|
!!!nack ('t380.1'); |
| 6436 |
|
} elsif ({ |
| 6437 |
|
b => 1, big => 1, em => 1, font => 1, i => 1, |
| 6438 |
|
s => 1, small => 1, strile => 1, |
| 6439 |
|
strong => 1, tt => 1, u => 1, |
| 6440 |
|
}->{$token->{tag_name}}) { |
| 6441 |
|
!!!cp ('t375'); |
| 6442 |
|
push @$active_formatting_elements, $self->{open_elements}->[-1]; |
| 6443 |
|
!!!nack ('t375.1'); |
| 6444 |
|
} elsif ($token->{tag_name} eq 'input') { |
| 6445 |
|
!!!cp ('t388'); |
| 6446 |
|
## TODO: associate with $self->{form_element} if defined |
| 6447 |
|
pop @{$self->{open_elements}}; |
| 6448 |
|
!!!ack ('t388.2'); |
| 6449 |
|
} elsif ({ |
| 6450 |
|
area => 1, basefont => 1, bgsound => 1, br => 1, |
| 6451 |
|
embed => 1, img => 1, param => 1, spacer => 1, wbr => 1, |
| 6452 |
|
#image => 1, |
| 6453 |
|
}->{$token->{tag_name}}) { |
| 6454 |
|
!!!cp ('t388.1'); |
| 6455 |
|
pop @{$self->{open_elements}}; |
| 6456 |
|
!!!ack ('t388.3'); |
| 6457 |
|
} elsif ($token->{tag_name} eq 'select') { |
| 6458 |
|
## TODO: associate with $self->{form_element} if defined |
| 6459 |
|
|
| 6460 |
|
if ($self->{insertion_mode} & TABLE_IMS or |
| 6461 |
|
$self->{insertion_mode} & BODY_TABLE_IMS or |
| 6462 |
|
$self->{insertion_mode} == IN_COLUMN_GROUP_IM) { |
| 6463 |
|
!!!cp ('t400.1'); |
| 6464 |
|
$self->{insertion_mode} = IN_SELECT_IN_TABLE_IM; |
| 6465 |
|
} else { |
| 6466 |
|
!!!cp ('t400.2'); |
| 6467 |
|
$self->{insertion_mode} = IN_SELECT_IM; |
| 6468 |
|
} |
| 6469 |
|
!!!nack ('t400.3'); |
| 6470 |
|
} else { |
| 6471 |
|
!!!nack ('t402'); |
| 6472 |
|
} |
| 6473 |
|
|
| 6474 |
!!!next-token; |
!!!next-token; |
| 6475 |
redo B; |
next B; |
| 6476 |
} |
} |
| 6477 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 6478 |
if ($token->{tag_name} eq 'body') { |
if ($token->{tag_name} eq 'body') { |
| 6479 |
if (@{$self->{open_elements}} > 1 and |
## has a |body| element in scope |
| 6480 |
$self->{open_elements}->[1]->[1] eq 'body') { |
my $i; |
| 6481 |
for (@{$self->{open_elements}}) { |
INSCOPE: { |
| 6482 |
unless ({ |
for (reverse @{$self->{open_elements}}) { |
| 6483 |
dd => 1, dt => 1, li => 1, p => 1, td => 1, |
if ($_->[1] & BODY_EL) { |
| 6484 |
th => 1, tr => 1, body => 1, html => 1, |
!!!cp ('t405'); |
| 6485 |
tbody => 1, tfoot => 1, thead => 1, |
$i = $_; |
| 6486 |
}->{$_->[1]}) { |
last INSCOPE; |
| 6487 |
!!!cp ('t403'); |
} elsif ($_->[1] & SCOPING_EL) { |
| 6488 |
!!!parse-error (type => 'not closed:'.$_->[1]); |
!!!cp ('t405.1'); |
| 6489 |
} else { |
last; |
|
!!!cp ('t404'); |
|
| 6490 |
} |
} |
| 6491 |
} |
} |
| 6492 |
|
|
| 6493 |
$self->{insertion_mode} = AFTER_BODY_IM; |
!!!parse-error (type => 'start tag not allowed', |
| 6494 |
|
value => $token->{tag_name}, token => $token); |
| 6495 |
|
## NOTE: Ignore the token. |
| 6496 |
!!!next-token; |
!!!next-token; |
| 6497 |
redo B; |
next B; |
| 6498 |
} else { |
} # INSCOPE |
| 6499 |
!!!cp ('t405'); |
|
| 6500 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
for (@{$self->{open_elements}}) { |
| 6501 |
## Ignore the token |
unless ($_->[1] & ALL_END_TAG_OPTIONAL_EL) { |
| 6502 |
!!!next-token; |
!!!cp ('t403'); |
| 6503 |
redo B; |
!!!parse-error (type => 'not closed', |
| 6504 |
|
value => $_->[0]->manakai_local_name, |
| 6505 |
|
token => $token); |
| 6506 |
|
last; |
| 6507 |
|
} else { |
| 6508 |
|
!!!cp ('t404'); |
| 6509 |
|
} |
| 6510 |
} |
} |
| 6511 |
|
|
| 6512 |
|
$self->{insertion_mode} = AFTER_BODY_IM; |
| 6513 |
|
!!!next-token; |
| 6514 |
|
next B; |
| 6515 |
} elsif ($token->{tag_name} eq 'html') { |
} elsif ($token->{tag_name} eq 'html') { |
| 6516 |
if (@{$self->{open_elements}} > 1 and $self->{open_elements}->[1]->[1] eq 'body') { |
## TODO: Update this code. It seems that the code below is not |
| 6517 |
|
## up-to-date, though it has same effect as speced. |
| 6518 |
|
if (@{$self->{open_elements}} > 1 and |
| 6519 |
|
$self->{open_elements}->[1]->[1] & BODY_EL) { |
| 6520 |
## ISSUE: There is an issue in the spec. |
## ISSUE: There is an issue in the spec. |
| 6521 |
if ($self->{open_elements}->[-1]->[1] ne 'body') { |
unless ($self->{open_elements}->[-1]->[1] & BODY_EL) { |
| 6522 |
!!!cp ('t406'); |
!!!cp ('t406'); |
| 6523 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[1]->[1]); |
!!!parse-error (type => 'not closed', |
| 6524 |
|
value => $self->{open_elements}->[1]->[0] |
| 6525 |
|
->manakai_local_name, |
| 6526 |
|
token => $token); |
| 6527 |
} else { |
} else { |
| 6528 |
!!!cp ('t407'); |
!!!cp ('t407'); |
| 6529 |
} |
} |
| 6530 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 6531 |
## reprocess |
## reprocess |
| 6532 |
redo B; |
next B; |
| 6533 |
} else { |
} else { |
| 6534 |
!!!cp ('t408'); |
!!!cp ('t408'); |
| 6535 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6536 |
## Ignore the token |
## Ignore the token |
| 6537 |
!!!next-token; |
!!!next-token; |
| 6538 |
redo B; |
next B; |
| 6539 |
} |
} |
| 6540 |
} elsif ({ |
} elsif ({ |
| 6541 |
address => 1, blockquote => 1, center => 1, dir => 1, |
address => 1, blockquote => 1, center => 1, dir => 1, |
| 6542 |
div => 1, dl => 1, fieldset => 1, listing => 1, |
div => 1, dl => 1, fieldset => 1, listing => 1, |
| 6543 |
menu => 1, ol => 1, pre => 1, ul => 1, |
menu => 1, ol => 1, pre => 1, ul => 1, |
|
p => 1, |
|
| 6544 |
dd => 1, dt => 1, li => 1, |
dd => 1, dt => 1, li => 1, |
| 6545 |
button => 1, marquee => 1, object => 1, |
applet => 1, button => 1, marquee => 1, object => 1, |
| 6546 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6547 |
## has an element in scope |
## has an element in scope |
| 6548 |
my $i; |
my $i; |
| 6549 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6550 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6551 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
|
## generate implied end tags |
|
|
while ({ |
|
|
dd => ($token->{tag_name} ne 'dd'), |
|
|
dt => ($token->{tag_name} ne 'dt'), |
|
|
li => ($token->{tag_name} ne 'li'), |
|
|
p => ($token->{tag_name} ne 'p'), |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t409'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} |
|
|
|
|
| 6552 |
!!!cp ('t410'); |
!!!cp ('t410'); |
| 6553 |
$i = $_; |
$i = $_; |
| 6554 |
last INSCOPE unless $token->{tag_name} eq 'p'; |
last INSCOPE; |
| 6555 |
} elsif ({ |
} elsif ($node->[1] & SCOPING_EL) { |
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 6556 |
!!!cp ('t411'); |
!!!cp ('t411'); |
| 6557 |
last INSCOPE; |
last INSCOPE; |
| 6558 |
} |
} |
| 6559 |
} # INSCOPE |
} # INSCOPE |
| 6560 |
|
|
| 6561 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
unless (defined $i) { # has an element in scope |
| 6562 |
if (defined $i) { |
!!!cp ('t413'); |
| 6563 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6564 |
|
} else { |
| 6565 |
|
## Step 1. generate implied end tags |
| 6566 |
|
while ({ |
| 6567 |
|
dd => ($token->{tag_name} ne 'dd'), |
| 6568 |
|
dt => ($token->{tag_name} ne 'dt'), |
| 6569 |
|
li => ($token->{tag_name} ne 'li'), |
| 6570 |
|
p => 1, |
| 6571 |
|
}->{$self->{open_elements}->[-1]->[0]->manakai_local_name}) { |
| 6572 |
|
!!!cp ('t409'); |
| 6573 |
|
pop @{$self->{open_elements}}; |
| 6574 |
|
} |
| 6575 |
|
|
| 6576 |
|
## Step 2. |
| 6577 |
|
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 6578 |
|
ne $token->{tag_name}) { |
| 6579 |
!!!cp ('t412'); |
!!!cp ('t412'); |
| 6580 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 6581 |
|
value => $self->{open_elements}->[-1]->[0] |
| 6582 |
|
->manakai_local_name, |
| 6583 |
|
token => $token); |
| 6584 |
} else { |
} else { |
| 6585 |
!!!cp ('t413'); |
!!!cp ('t414'); |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
|
| 6586 |
} |
} |
| 6587 |
} |
|
| 6588 |
|
## Step 3. |
|
if (defined $i) { |
|
|
!!!cp ('t414'); |
|
| 6589 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 6590 |
} elsif ($token->{tag_name} eq 'p') { |
|
| 6591 |
!!!cp ('t415'); |
## Step 4. |
| 6592 |
## As if <p>, then reprocess the current token |
$clear_up_to_marker->() |
| 6593 |
my $el; |
if { |
| 6594 |
!!!create-element ($el, 'p'); |
applet => 1, button => 1, marquee => 1, object => 1, |
| 6595 |
$insert->($el); |
}->{$token->{tag_name}}; |
|
} else { |
|
|
!!!cp ('t416'); |
|
| 6596 |
} |
} |
|
$clear_up_to_marker->() |
|
|
if { |
|
|
button => 1, marquee => 1, object => 1, |
|
|
}->{$token->{tag_name}}; |
|
| 6597 |
!!!next-token; |
!!!next-token; |
| 6598 |
redo B; |
next B; |
| 6599 |
} elsif ($token->{tag_name} eq 'form') { |
} elsif ($token->{tag_name} eq 'form') { |
| 6600 |
|
undef $self->{form_element}; |
| 6601 |
|
|
| 6602 |
## has an element in scope |
## has an element in scope |
| 6603 |
|
my $i; |
| 6604 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6605 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6606 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] & FORM_EL) { |
|
## generate implied end tags |
|
|
while ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t417'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} |
|
|
|
|
| 6607 |
!!!cp ('t418'); |
!!!cp ('t418'); |
| 6608 |
|
$i = $_; |
| 6609 |
last INSCOPE; |
last INSCOPE; |
| 6610 |
} elsif ({ |
} elsif ($node->[1] & SCOPING_EL) { |
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 6611 |
!!!cp ('t419'); |
!!!cp ('t419'); |
| 6612 |
last INSCOPE; |
last INSCOPE; |
| 6613 |
} |
} |
| 6614 |
} # INSCOPE |
} # INSCOPE |
| 6615 |
|
|
| 6616 |
if ($self->{open_elements}->[-1]->[1] eq $token->{tag_name}) { |
unless (defined $i) { # has an element in scope |
|
!!!cp ('t420'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} else { |
|
| 6617 |
!!!cp ('t421'); |
!!!cp ('t421'); |
| 6618 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6619 |
|
} else { |
| 6620 |
|
## Step 1. generate implied end tags |
| 6621 |
|
while ($self->{open_elements}->[-1]->[1] & END_TAG_OPTIONAL_EL) { |
| 6622 |
|
!!!cp ('t417'); |
| 6623 |
|
pop @{$self->{open_elements}}; |
| 6624 |
|
} |
| 6625 |
|
|
| 6626 |
|
## Step 2. |
| 6627 |
|
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 6628 |
|
ne $token->{tag_name}) { |
| 6629 |
|
!!!cp ('t417.1'); |
| 6630 |
|
!!!parse-error (type => 'not closed', |
| 6631 |
|
value => $self->{open_elements}->[-1]->[0] |
| 6632 |
|
->manakai_local_name, |
| 6633 |
|
token => $token); |
| 6634 |
|
} else { |
| 6635 |
|
!!!cp ('t420'); |
| 6636 |
|
} |
| 6637 |
|
|
| 6638 |
|
## Step 3. |
| 6639 |
|
splice @{$self->{open_elements}}, $i; |
| 6640 |
} |
} |
| 6641 |
|
|
|
undef $self->{form_element}; |
|
| 6642 |
!!!next-token; |
!!!next-token; |
| 6643 |
redo B; |
next B; |
| 6644 |
} elsif ({ |
} elsif ({ |
| 6645 |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
| 6646 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6648 |
my $i; |
my $i; |
| 6649 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6650 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6651 |
if ({ |
if ($node->[1] & HEADING_EL) { |
|
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
|
|
}->{$node->[1]}) { |
|
|
## generate implied end tags |
|
|
while ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t422'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} |
|
|
|
|
| 6652 |
!!!cp ('t423'); |
!!!cp ('t423'); |
| 6653 |
$i = $_; |
$i = $_; |
| 6654 |
last INSCOPE; |
last INSCOPE; |
| 6655 |
} elsif ({ |
} elsif ($node->[1] & SCOPING_EL) { |
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
| 6656 |
!!!cp ('t424'); |
!!!cp ('t424'); |
| 6657 |
last INSCOPE; |
last INSCOPE; |
| 6658 |
} |
} |
| 6659 |
} # INSCOPE |
} # INSCOPE |
| 6660 |
|
|
| 6661 |
|
unless (defined $i) { # has an element in scope |
| 6662 |
|
!!!cp ('t425.1'); |
| 6663 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6664 |
|
} else { |
| 6665 |
|
## Step 1. generate implied end tags |
| 6666 |
|
while ($self->{open_elements}->[-1]->[1] & END_TAG_OPTIONAL_EL) { |
| 6667 |
|
!!!cp ('t422'); |
| 6668 |
|
pop @{$self->{open_elements}}; |
| 6669 |
|
} |
| 6670 |
|
|
| 6671 |
|
## Step 2. |
| 6672 |
|
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 6673 |
|
ne $token->{tag_name}) { |
| 6674 |
|
!!!cp ('t425'); |
| 6675 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6676 |
|
} else { |
| 6677 |
|
!!!cp ('t426'); |
| 6678 |
|
} |
| 6679 |
|
|
| 6680 |
|
## Step 3. |
| 6681 |
|
splice @{$self->{open_elements}}, $i; |
| 6682 |
|
} |
| 6683 |
|
|
| 6684 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
!!!next-token; |
| 6685 |
!!!cp ('t425'); |
next B; |
| 6686 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
} elsif ($token->{tag_name} eq 'p') { |
| 6687 |
|
## has an element in scope |
| 6688 |
|
my $i; |
| 6689 |
|
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6690 |
|
my $node = $self->{open_elements}->[$_]; |
| 6691 |
|
if ($node->[1] & P_EL) { |
| 6692 |
|
!!!cp ('t410.1'); |
| 6693 |
|
$i = $_; |
| 6694 |
|
last INSCOPE; |
| 6695 |
|
} elsif ($node->[1] & SCOPING_EL) { |
| 6696 |
|
!!!cp ('t411.1'); |
| 6697 |
|
last INSCOPE; |
| 6698 |
|
} |
| 6699 |
|
} # INSCOPE |
| 6700 |
|
|
| 6701 |
|
if (defined $i) { |
| 6702 |
|
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 6703 |
|
ne $token->{tag_name}) { |
| 6704 |
|
!!!cp ('t412.1'); |
| 6705 |
|
!!!parse-error (type => 'not closed', |
| 6706 |
|
value => $self->{open_elements}->[-1]->[0] |
| 6707 |
|
->manakai_local_name, |
| 6708 |
|
token => $token); |
| 6709 |
|
} else { |
| 6710 |
|
!!!cp ('t414.1'); |
| 6711 |
|
} |
| 6712 |
|
|
| 6713 |
|
splice @{$self->{open_elements}}, $i; |
| 6714 |
} else { |
} else { |
| 6715 |
!!!cp ('t426'); |
!!!cp ('t413.1'); |
| 6716 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6717 |
|
|
| 6718 |
|
!!!cp ('t415.1'); |
| 6719 |
|
## As if <p>, then reprocess the current token |
| 6720 |
|
my $el; |
| 6721 |
|
!!!create-element ($el, $HTML_NS, 'p',, $token); |
| 6722 |
|
$insert->($el); |
| 6723 |
|
## NOTE: Not inserted into |$self->{open_elements}|. |
| 6724 |
} |
} |
| 6725 |
|
|
|
splice @{$self->{open_elements}}, $i if defined $i; |
|
| 6726 |
!!!next-token; |
!!!next-token; |
| 6727 |
redo B; |
next B; |
| 6728 |
} elsif ({ |
} elsif ({ |
| 6729 |
a => 1, |
a => 1, |
| 6730 |
b => 1, big => 1, em => 1, font => 1, i => 1, |
b => 1, big => 1, em => 1, font => 1, i => 1, |
| 6732 |
strong => 1, tt => 1, u => 1, |
strong => 1, tt => 1, u => 1, |
| 6733 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6734 |
!!!cp ('t427'); |
!!!cp ('t427'); |
| 6735 |
$formatting_end_tag->($token->{tag_name}); |
$formatting_end_tag->($token); |
| 6736 |
redo B; |
next B; |
| 6737 |
} elsif ($token->{tag_name} eq 'br') { |
} elsif ($token->{tag_name} eq 'br') { |
| 6738 |
!!!cp ('t428'); |
!!!cp ('t428'); |
| 6739 |
!!!parse-error (type => 'unmatched end tag:br'); |
!!!parse-error (type => 'unmatched end tag:br', token => $token); |
| 6740 |
|
|
| 6741 |
## As if <br> |
## As if <br> |
| 6742 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 6743 |
|
|
| 6744 |
my $el; |
my $el; |
| 6745 |
!!!create-element ($el, 'br'); |
!!!create-element ($el, $HTML_NS, 'br',, $token); |
| 6746 |
$insert->($el); |
$insert->($el); |
| 6747 |
|
|
| 6748 |
## Ignore the token. |
## Ignore the token. |
| 6749 |
!!!next-token; |
!!!next-token; |
| 6750 |
redo B; |
next B; |
| 6751 |
} elsif ({ |
} elsif ({ |
| 6752 |
caption => 1, col => 1, colgroup => 1, frame => 1, |
caption => 1, col => 1, colgroup => 1, frame => 1, |
| 6753 |
frameset => 1, head => 1, option => 1, optgroup => 1, |
frameset => 1, head => 1, option => 1, optgroup => 1, |
| 6761 |
noscript => 0, ## TODO: if scripting is enabled |
noscript => 0, ## TODO: if scripting is enabled |
| 6762 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6763 |
!!!cp ('t429'); |
!!!cp ('t429'); |
| 6764 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6765 |
## Ignore the token |
## Ignore the token |
| 6766 |
!!!next-token; |
!!!next-token; |
| 6767 |
redo B; |
next B; |
| 6768 |
|
|
| 6769 |
## ISSUE: Issue on HTML5 new elements in spec |
## ISSUE: Issue on HTML5 new elements in spec |
| 6770 |
|
|
| 6775 |
|
|
| 6776 |
## Step 2 |
## Step 2 |
| 6777 |
S2: { |
S2: { |
| 6778 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[0]->manakai_local_name eq $token->{tag_name}) { |
| 6779 |
## Step 1 |
## Step 1 |
| 6780 |
## generate implied end tags |
## generate implied end tags |
| 6781 |
while ({ |
while ($self->{open_elements}->[-1]->[1] & END_TAG_OPTIONAL_EL) { |
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 6782 |
!!!cp ('t430'); |
!!!cp ('t430'); |
| 6783 |
## ISSUE: Can this case be reached? |
## ISSUE: Can this case be reached? |
| 6784 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 6785 |
} |
} |
| 6786 |
|
|
| 6787 |
## Step 2 |
## Step 2 |
| 6788 |
if ($token->{tag_name} ne $self->{open_elements}->[-1]->[1]) { |
if ($self->{open_elements}->[-1]->[0]->manakai_local_name |
| 6789 |
|
ne $token->{tag_name}) { |
| 6790 |
!!!cp ('t431'); |
!!!cp ('t431'); |
| 6791 |
## NOTE: <x><y></x> |
## NOTE: <x><y></x> |
| 6792 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed', |
| 6793 |
|
value => $self->{open_elements}->[-1]->[0] |
| 6794 |
|
->manakai_local_name, |
| 6795 |
|
token => $token); |
| 6796 |
} else { |
} else { |
| 6797 |
!!!cp ('t432'); |
!!!cp ('t432'); |
| 6798 |
} |
} |
| 6804 |
last S2; |
last S2; |
| 6805 |
} else { |
} else { |
| 6806 |
## Step 3 |
## Step 3 |
| 6807 |
if (not $formatting_category->{$node->[1]} and |
if (not ($node->[1] & FORMATTING_EL) and |
| 6808 |
#not $phrasing_category->{$node->[1]} and |
#not $phrasing_category->{$node->[1]} and |
| 6809 |
($special_category->{$node->[1]} or |
($node->[1] & SPECIAL_EL or |
| 6810 |
$scoping_category->{$node->[1]})) { |
$node->[1] & SCOPING_EL)) { |
| 6811 |
!!!cp ('t433'); |
!!!cp ('t433'); |
| 6812 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}, token => $token); |
| 6813 |
## Ignore the token |
## Ignore the token |
| 6814 |
!!!next-token; |
!!!next-token; |
| 6815 |
last S2; |
last S2; |
| 6825 |
## Step 5; |
## Step 5; |
| 6826 |
redo S2; |
redo S2; |
| 6827 |
} # S2 |
} # S2 |
| 6828 |
redo B; |
next B; |
| 6829 |
} |
} |
| 6830 |
} |
} |
| 6831 |
redo B; |
next B; |
| 6832 |
|
} continue { # B |
| 6833 |
|
if ($self->{insertion_mode} & IN_FOREIGN_CONTENT_IM) { |
| 6834 |
|
## NOTE: The code below is executed in cases where it does not have |
| 6835 |
|
## to be, but it it is harmless even in those cases. |
| 6836 |
|
## has an element in scope |
| 6837 |
|
INSCOPE: { |
| 6838 |
|
for (reverse 0..$#{$self->{open_elements}}) { |
| 6839 |
|
my $node = $self->{open_elements}->[$_]; |
| 6840 |
|
if ($node->[1] & FOREIGN_EL) { |
| 6841 |
|
last INSCOPE; |
| 6842 |
|
} elsif ($node->[1] & SCOPING_EL) { |
| 6843 |
|
last; |
| 6844 |
|
} |
| 6845 |
|
} |
| 6846 |
|
|
| 6847 |
|
## NOTE: No foreign element in scope. |
| 6848 |
|
$self->{insertion_mode} &= ~ IN_FOREIGN_CONTENT_IM; |
| 6849 |
|
} # INSCOPE |
| 6850 |
|
} |
| 6851 |
} # B |
} # B |
| 6852 |
|
|
| 6853 |
## Stop parsing # MUST |
## Stop parsing # MUST |
| 6893 |
|
|
| 6894 |
## Step 8 # MUST |
## Step 8 # MUST |
| 6895 |
my $i = 0; |
my $i = 0; |
| 6896 |
my $line = 1; |
$p->{line_prev} = $p->{line} = 1; |
| 6897 |
my $column = 0; |
$p->{column_prev} = $p->{column} = 0; |
| 6898 |
$p->{set_next_char} = sub { |
$p->{set_next_char} = sub { |
| 6899 |
my $self = shift; |
my $self = shift; |
| 6900 |
|
|
| 6903 |
|
|
| 6904 |
$self->{next_char} = -1 and return if $i >= length $$s; |
$self->{next_char} = -1 and return if $i >= length $$s; |
| 6905 |
$self->{next_char} = ord substr $$s, $i++, 1; |
$self->{next_char} = ord substr $$s, $i++, 1; |
| 6906 |
$column++; |
|
| 6907 |
|
($p->{line_prev}, $p->{column_prev}) = ($p->{line}, $p->{column}); |
| 6908 |
|
$p->{column}++; |
| 6909 |
|
|
| 6910 |
if ($self->{next_char} == 0x000A) { # LF |
if ($self->{next_char} == 0x000A) { # LF |
| 6911 |
$line++; |
$p->{line}++; |
| 6912 |
$column = 0; |
$p->{column} = 0; |
| 6913 |
!!!cp ('i1'); |
!!!cp ('i1'); |
| 6914 |
} elsif ($self->{next_char} == 0x000D) { # CR |
} elsif ($self->{next_char} == 0x000D) { # CR |
| 6915 |
$i++ if substr ($$s, $i, 1) eq "\x0A"; |
$i++ if substr ($$s, $i, 1) eq "\x0A"; |
| 6916 |
$self->{next_char} = 0x000A; # LF # MUST |
$self->{next_char} = 0x000A; # LF # MUST |
| 6917 |
$line++; |
$p->{line}++; |
| 6918 |
$column = 0; |
$p->{column} = 0; |
| 6919 |
!!!cp ('i2'); |
!!!cp ('i2'); |
| 6920 |
} elsif ($self->{next_char} > 0x10FFFF) { |
} elsif ($self->{next_char} > 0x10FFFF) { |
| 6921 |
$self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
$self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
| 6931 |
|
|
| 6932 |
my $ponerror = $onerror || sub { |
my $ponerror = $onerror || sub { |
| 6933 |
my (%opt) = @_; |
my (%opt) = @_; |
| 6934 |
warn "Parse error ($opt{type}) at line $opt{line} column $opt{column}\n"; |
my $line = $opt{line}; |
| 6935 |
|
my $column = $opt{column}; |
| 6936 |
|
if (defined $opt{token} and defined $opt{token}->{line}) { |
| 6937 |
|
$line = $opt{token}->{line}; |
| 6938 |
|
$column = $opt{token}->{column}; |
| 6939 |
|
} |
| 6940 |
|
warn "Parse error ($opt{type}) at line $line column $column\n"; |
| 6941 |
}; |
}; |
| 6942 |
$p->{parse_error} = sub { |
$p->{parse_error} = sub { |
| 6943 |
$ponerror->(@_, line => $line, column => $column); |
$ponerror->(line => $p->{line}, column => $p->{column}, @_); |
| 6944 |
}; |
}; |
| 6945 |
|
|
| 6946 |
$p->_initialize_tokenizer; |
$p->_initialize_tokenizer; |
| 6964 |
unless defined $p->{content_model}; |
unless defined $p->{content_model}; |
| 6965 |
## ISSUE: What is "the name of the element"? local name? |
## ISSUE: What is "the name of the element"? local name? |
| 6966 |
|
|
| 6967 |
$p->{inner_html_node} = [$node, $node_ln]; |
$p->{inner_html_node} = [$node, $el_category->{$node_ln}]; |
| 6968 |
|
## TODO: Foreign element OK? |
| 6969 |
|
|
| 6970 |
## Step 3 |
## Step 3 |
| 6971 |
my $root = $doc->create_element_ns |
my $root = $doc->create_element_ns |
| 6975 |
$doc->append_child ($root); |
$doc->append_child ($root); |
| 6976 |
|
|
| 6977 |
## Step 5 # MUST |
## Step 5 # MUST |
| 6978 |
push @{$p->{open_elements}}, [$root, 'html']; |
push @{$p->{open_elements}}, [$root, $el_category->{html}]; |
| 6979 |
|
|
| 6980 |
undef $p->{head_element}; |
undef $p->{head_element}; |
| 6981 |
|
|
| 7021 |
## ISSUE: mutation events? |
## ISSUE: mutation events? |
| 7022 |
|
|
| 7023 |
$p->_terminate_tree_constructor; |
$p->_terminate_tree_constructor; |
| 7024 |
|
|
| 7025 |
|
delete $p->{parse_error}; # delete loop |
| 7026 |
} else { |
} else { |
| 7027 |
die "$0: |set_inner_html| is not defined for node of type $nt"; |
die "$0: |set_inner_html| is not defined for node of type $nt"; |
| 7028 |
} |
} |