| 74 |
textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1, |
textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1, |
| 75 |
}; |
}; |
| 76 |
my $scoping_category = { |
my $scoping_category = { |
| 77 |
button => 1, caption => 1, html => 1, marquee => 1, object => 1, |
applet => 1, button => 1, caption => 1, html => 1, marquee => 1, object => 1, |
| 78 |
table => 1, td => 1, th => 1, |
table => 1, td => 1, th => 1, |
| 79 |
}; |
}; |
| 80 |
my $formatting_category = { |
my $formatting_category = { |
| 303 |
sub ROW_IMS () { 0b10000000 } |
sub ROW_IMS () { 0b10000000 } |
| 304 |
sub BODY_AFTER_IMS () { 0b100000000 } |
sub BODY_AFTER_IMS () { 0b100000000 } |
| 305 |
sub FRAME_IMS () { 0b1000000000 } |
sub FRAME_IMS () { 0b1000000000 } |
| 306 |
|
sub SELECT_IMS () { 0b10000000000 } |
| 307 |
|
|
| 308 |
|
## NOTE: "initial" and "before html" insertion modes have no constants. |
| 309 |
|
|
| 310 |
|
## NOTE: "after after body" insertion mode. |
| 311 |
sub AFTER_HTML_BODY_IM () { AFTER_HTML_IMS | BODY_AFTER_IMS } |
sub AFTER_HTML_BODY_IM () { AFTER_HTML_IMS | BODY_AFTER_IMS } |
| 312 |
|
|
| 313 |
|
## NOTE: "after after frameset" insertion mode. |
| 314 |
sub AFTER_HTML_FRAMESET_IM () { AFTER_HTML_IMS | FRAME_IMS } |
sub AFTER_HTML_FRAMESET_IM () { AFTER_HTML_IMS | FRAME_IMS } |
| 315 |
|
|
| 316 |
sub IN_HEAD_IM () { HEAD_IMS | 0b00 } |
sub IN_HEAD_IM () { HEAD_IMS | 0b00 } |
| 317 |
sub IN_HEAD_NOSCRIPT_IM () { HEAD_IMS | 0b01 } |
sub IN_HEAD_NOSCRIPT_IM () { HEAD_IMS | 0b01 } |
| 318 |
sub AFTER_HEAD_IM () { HEAD_IMS | 0b10 } |
sub AFTER_HEAD_IM () { HEAD_IMS | 0b10 } |
| 326 |
sub AFTER_BODY_IM () { BODY_AFTER_IMS } |
sub AFTER_BODY_IM () { BODY_AFTER_IMS } |
| 327 |
sub IN_FRAMESET_IM () { FRAME_IMS | 0b01 } |
sub IN_FRAMESET_IM () { FRAME_IMS | 0b01 } |
| 328 |
sub AFTER_FRAMESET_IM () { FRAME_IMS | 0b10 } |
sub AFTER_FRAMESET_IM () { FRAME_IMS | 0b10 } |
| 329 |
sub IN_SELECT_IM () { 0b01 } |
sub IN_SELECT_IM () { SELECT_IMS | 0b01 } |
| 330 |
|
sub IN_SELECT_IN_TABLE_IM () { SELECT_IMS | 0b10 } |
| 331 |
sub IN_COLUMN_GROUP_IM () { 0b10 } |
sub IN_COLUMN_GROUP_IM () { 0b10 } |
| 332 |
|
|
| 333 |
## Implementations MUST act as if state machine in the spec |
## Implementations MUST act as if state machine in the spec |
| 2428 |
|
|
| 2429 |
!!!next-token; |
!!!next-token; |
| 2430 |
|
|
|
$self->{insertion_mode} = BEFORE_HEAD_IM; |
|
| 2431 |
undef $self->{form_element}; |
undef $self->{form_element}; |
| 2432 |
undef $self->{head_element}; |
undef $self->{head_element}; |
| 2433 |
$self->{open_elements} = []; |
$self->{open_elements} = []; |
| 2434 |
undef $self->{inner_html_node}; |
undef $self->{inner_html_node}; |
| 2435 |
|
|
| 2436 |
|
## NOTE: The "initial" insertion mode. |
| 2437 |
$self->_tree_construction_initial; # MUST |
$self->_tree_construction_initial; # MUST |
| 2438 |
|
|
| 2439 |
|
## NOTE: The "before html" insertion mode. |
| 2440 |
$self->_tree_construction_root_element; |
$self->_tree_construction_root_element; |
| 2441 |
|
$self->{insertion_mode} = BEFORE_HEAD_IM; |
| 2442 |
|
|
| 2443 |
|
## NOTE: The "before head" insertion mode and so on. |
| 2444 |
$self->_tree_construction_main; |
$self->_tree_construction_main; |
| 2445 |
} # _construct_tree |
} # _construct_tree |
| 2446 |
|
|
| 2447 |
sub _tree_construction_initial ($) { |
sub _tree_construction_initial ($) { |
| 2448 |
my $self = shift; |
my $self = shift; |
| 2449 |
|
|
| 2450 |
|
## NOTE: "initial" insertion mode |
| 2451 |
|
|
| 2452 |
INITIAL: { |
INITIAL: { |
| 2453 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 2454 |
## NOTE: Conformance checkers MAY, instead of reporting "not HTML5" |
## NOTE: Conformance checkers MAY, instead of reporting "not HTML5" |
| 2595 |
!!!cp ('t13'); |
!!!cp ('t13'); |
| 2596 |
} |
} |
| 2597 |
|
|
| 2598 |
## Go to the root element phase. |
## Go to the "before html" insertion mode. |
| 2599 |
!!!next-token; |
!!!next-token; |
| 2600 |
return; |
return; |
| 2601 |
} elsif ({ |
} elsif ({ |
| 2606 |
!!!cp ('t14'); |
!!!cp ('t14'); |
| 2607 |
!!!parse-error (type => 'no DOCTYPE'); |
!!!parse-error (type => 'no DOCTYPE'); |
| 2608 |
$self->{document}->manakai_compat_mode ('quirks'); |
$self->{document}->manakai_compat_mode ('quirks'); |
| 2609 |
## Go to the root element phase |
## Go to the "before html" insertion mode. |
| 2610 |
## reprocess |
## reprocess |
| 2611 |
return; |
return; |
| 2612 |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
| 2615 |
|
|
| 2616 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 2617 |
!!!cp ('t15'); |
!!!cp ('t15'); |
| 2618 |
## Stay in the phase |
## Stay in the insertion mode. |
| 2619 |
!!!next-token; |
!!!next-token; |
| 2620 |
redo INITIAL; |
redo INITIAL; |
| 2621 |
} else { |
} else { |
| 2627 |
|
|
| 2628 |
!!!parse-error (type => 'no DOCTYPE'); |
!!!parse-error (type => 'no DOCTYPE'); |
| 2629 |
$self->{document}->manakai_compat_mode ('quirks'); |
$self->{document}->manakai_compat_mode ('quirks'); |
| 2630 |
## Go to the root element phase |
## Go to the "before html" insertion mode. |
| 2631 |
## reprocess |
## reprocess |
| 2632 |
return; |
return; |
| 2633 |
} elsif ($token->{type} == COMMENT_TOKEN) { |
} elsif ($token->{type} == COMMENT_TOKEN) { |
| 2635 |
my $comment = $self->{document}->create_comment ($token->{data}); |
my $comment = $self->{document}->create_comment ($token->{data}); |
| 2636 |
$self->{document}->append_child ($comment); |
$self->{document}->append_child ($comment); |
| 2637 |
|
|
| 2638 |
## Stay in the phase. |
## Stay in the insertion mode. |
| 2639 |
!!!next-token; |
!!!next-token; |
| 2640 |
redo INITIAL; |
redo INITIAL; |
| 2641 |
} else { |
} else { |
| 2648 |
|
|
| 2649 |
sub _tree_construction_root_element ($) { |
sub _tree_construction_root_element ($) { |
| 2650 |
my $self = shift; |
my $self = shift; |
| 2651 |
|
|
| 2652 |
|
## NOTE: "before html" insertion mode. |
| 2653 |
|
|
| 2654 |
B: { |
B: { |
| 2655 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 2656 |
!!!cp ('t19'); |
!!!cp ('t19'); |
| 2657 |
!!!parse-error (type => 'in html:#DOCTYPE'); |
!!!parse-error (type => 'in html:#DOCTYPE'); |
| 2658 |
## Ignore the token |
## Ignore the token |
| 2659 |
## Stay in the phase |
## Stay in the insertion mode. |
| 2660 |
!!!next-token; |
!!!next-token; |
| 2661 |
redo B; |
redo B; |
| 2662 |
} elsif ($token->{type} == COMMENT_TOKEN) { |
} elsif ($token->{type} == COMMENT_TOKEN) { |
| 2663 |
!!!cp ('t20'); |
!!!cp ('t20'); |
| 2664 |
my $comment = $self->{document}->create_comment ($token->{data}); |
my $comment = $self->{document}->create_comment ($token->{data}); |
| 2665 |
$self->{document}->append_child ($comment); |
$self->{document}->append_child ($comment); |
| 2666 |
## Stay in the phase |
## Stay in the insertion mode. |
| 2667 |
!!!next-token; |
!!!next-token; |
| 2668 |
redo B; |
redo B; |
| 2669 |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
} elsif ($token->{type} == CHARACTER_TOKEN) { |
| 2672 |
|
|
| 2673 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 2674 |
!!!cp ('t21'); |
!!!cp ('t21'); |
| 2675 |
## Stay in the phase |
## Stay in the insertion mode. |
| 2676 |
!!!next-token; |
!!!next-token; |
| 2677 |
redo B; |
redo B; |
| 2678 |
} else { |
} else { |
| 2686 |
|
|
| 2687 |
# |
# |
| 2688 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 2689 |
if ($token->{tag_name} eq 'html' and |
if ($token->{tag_name} eq 'html') { |
| 2690 |
$token->{attributes}->{manifest}) { |
my $root_element; |
| 2691 |
!!!cp ('t24'); |
!!!create-element ($root_element, $token->{tag_name}, $token->{attributes}); |
| 2692 |
$self->{application_cache_selection} |
$self->{document}->append_child ($root_element); |
| 2693 |
->($token->{attributes}->{manifest}->{value}); |
push @{$self->{open_elements}}, [$root_element, 'html']; |
| 2694 |
## ISSUE: No relative reference resolution? |
|
| 2695 |
|
if ($token->{attributes}->{manifest}) { |
| 2696 |
|
!!!cp ('t24'); |
| 2697 |
|
$self->{application_cache_selection} |
| 2698 |
|
->($token->{attributes}->{manifest}->{value}); |
| 2699 |
|
## ISSUE: No relative reference resolution? |
| 2700 |
|
} else { |
| 2701 |
|
!!!cp ('t25'); |
| 2702 |
|
$self->{application_cache_selection}->(undef); |
| 2703 |
|
} |
| 2704 |
|
|
| 2705 |
|
!!!next-token; |
| 2706 |
|
return; ## Go to the "before head" insertion mode. |
| 2707 |
} else { |
} else { |
| 2708 |
!!!cp ('t25'); |
!!!cp ('t25.1'); |
| 2709 |
$self->{application_cache_selection}->(undef); |
# |
| 2710 |
} |
} |
|
|
|
|
## ISSUE: There is an issue in the spec |
|
|
# |
|
| 2711 |
} elsif ({ |
} elsif ({ |
| 2712 |
END_TAG_TOKEN, 1, |
END_TAG_TOKEN, 1, |
| 2713 |
END_OF_FILE_TOKEN, 1, |
END_OF_FILE_TOKEN, 1, |
| 2714 |
}->{$token->{type}}) { |
}->{$token->{type}}) { |
| 2715 |
!!!cp ('t26'); |
!!!cp ('t26'); |
|
$self->{application_cache_selection}->(undef); |
|
|
|
|
|
## ISSUE: There is an issue in the spec |
|
| 2716 |
# |
# |
| 2717 |
} else { |
} else { |
| 2718 |
die "$0: $token->{type}: Unknown token type"; |
die "$0: $token->{type}: Unknown token type"; |
| 2719 |
} |
} |
| 2720 |
|
|
| 2721 |
my $root_element; !!!create-element ($root_element, 'html'); |
my $root_element; !!!create-element ($root_element, 'html'); |
| 2722 |
$self->{document}->append_child ($root_element); |
$self->{document}->append_child ($root_element); |
| 2723 |
push @{$self->{open_elements}}, [$root_element, 'html']; |
push @{$self->{open_elements}}, [$root_element, 'html']; |
| 2724 |
## reprocess |
|
| 2725 |
#redo B; |
$self->{application_cache_selection}->(undef); |
| 2726 |
return; ## Go to the main phase. |
|
| 2727 |
|
## NOTE: Reprocess the token. |
| 2728 |
|
return; ## Go to the "before head" insertion mode. |
| 2729 |
|
|
| 2730 |
|
## ISSUE: There is an issue in the spec |
| 2731 |
} # B |
} # B |
| 2732 |
|
|
| 2733 |
die "$0: _tree_construction_root_element: This should never be reached"; |
die "$0: _tree_construction_root_element: This should never be reached"; |
| 2745 |
|
|
| 2746 |
## Step 3 |
## Step 3 |
| 2747 |
S3: { |
S3: { |
|
## ISSUE: Oops! "If node is the first node in the stack of open |
|
|
## elements, then set last to true. If the context element of the |
|
|
## HTML fragment parsing algorithm is neither a td element nor a |
|
|
## th element, then set node to the context element. (fragment case)": |
|
|
## The second "if" is in the scope of the first "if"!? |
|
| 2748 |
if ($self->{open_elements}->[0]->[0] eq $node->[0]) { |
if ($self->{open_elements}->[0]->[0] eq $node->[0]) { |
| 2749 |
$last = 1; |
$last = 1; |
| 2750 |
if (defined $self->{inner_html_node}) { |
if (defined $self->{inner_html_node}) { |
| 2762 |
## Step 4..13 |
## Step 4..13 |
| 2763 |
my $new_mode = { |
my $new_mode = { |
| 2764 |
select => IN_SELECT_IM, |
select => IN_SELECT_IM, |
| 2765 |
|
## NOTE: |option| and |optgroup| do not set |
| 2766 |
|
## insertion mode to "in select" by themselves. |
| 2767 |
td => IN_CELL_IM, |
td => IN_CELL_IM, |
| 2768 |
th => IN_CELL_IM, |
th => IN_CELL_IM, |
| 2769 |
tr => IN_ROW_IM, |
tr => IN_ROW_IM, |
| 2905 |
!!!cp ('t39'); |
!!!cp ('t39'); |
| 2906 |
}; # $clear_up_to_marker |
}; # $clear_up_to_marker |
| 2907 |
|
|
| 2908 |
my $parse_rcdata = sub ($$) { |
my $insert; |
| 2909 |
my ($content_model_flag, $insert) = @_; |
|
| 2910 |
|
my $parse_rcdata = sub ($) { |
| 2911 |
|
my ($content_model_flag) = @_; |
| 2912 |
|
|
| 2913 |
## Step 1 |
## Step 1 |
| 2914 |
my $start_tag_name = $token->{tag_name}; |
my $start_tag_name = $token->{tag_name}; |
| 2916 |
!!!create-element ($el, $start_tag_name, $token->{attributes}); |
!!!create-element ($el, $start_tag_name, $token->{attributes}); |
| 2917 |
|
|
| 2918 |
## Step 2 |
## Step 2 |
| 2919 |
$insert->($el); # /context node/->append_child ($el) |
$insert->($el); |
| 2920 |
|
|
| 2921 |
## Step 3 |
## Step 3 |
| 2922 |
$self->{content_model} = $content_model_flag; # CDATA or RCDATA |
$self->{content_model} = $content_model_flag; # CDATA or RCDATA |
| 2946 |
$token->{tag_name} eq $start_tag_name) { |
$token->{tag_name} eq $start_tag_name) { |
| 2947 |
!!!cp ('t42'); |
!!!cp ('t42'); |
| 2948 |
## Ignore the token |
## Ignore the token |
|
} elsif ($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}); |
|
| 2949 |
} else { |
} else { |
| 2950 |
die "$0: $content_model_flag in parse_rcdata"; |
## NOTE: An end-of-file token. |
| 2951 |
|
if ($content_model_flag == CDATA_CONTENT_MODEL) { |
| 2952 |
|
!!!cp ('t43'); |
| 2953 |
|
!!!parse-error (type => 'in CDATA:#'.$token->{type}); |
| 2954 |
|
} elsif ($content_model_flag == RCDATA_CONTENT_MODEL) { |
| 2955 |
|
!!!cp ('t44'); |
| 2956 |
|
!!!parse-error (type => 'in RCDATA:#'.$token->{type}); |
| 2957 |
|
} else { |
| 2958 |
|
die "$0: $content_model_flag in parse_rcdata"; |
| 2959 |
|
} |
| 2960 |
} |
} |
| 2961 |
!!!next-token; |
!!!next-token; |
| 2962 |
}; # $parse_rcdata |
}; # $parse_rcdata |
| 2963 |
|
|
| 2964 |
my $script_start_tag = sub ($) { |
my $script_start_tag = sub () { |
|
my $insert = $_[0]; |
|
| 2965 |
my $script_el; |
my $script_el; |
| 2966 |
!!!create-element ($script_el, 'script', $token->{attributes}); |
!!!create-element ($script_el, 'script', $token->{attributes}); |
| 2967 |
## TODO: mark as "parser-inserted" |
## TODO: mark as "parser-inserted" |
| 3012 |
!!!next-token; |
!!!next-token; |
| 3013 |
}; # $script_start_tag |
}; # $script_start_tag |
| 3014 |
|
|
| 3015 |
|
## NOTE: $open_tables->[-1]->[0] is the "current table" element node. |
| 3016 |
|
## NOTE: $open_tables->[-1]->[1] is the "tainted" flag. |
| 3017 |
|
my $open_tables = [[$self->{open_elements}->[0]->[0]]]; |
| 3018 |
|
|
| 3019 |
my $formatting_end_tag = sub { |
my $formatting_end_tag = sub { |
| 3020 |
my $tag_name = shift; |
my $tag_name = shift; |
| 3021 |
|
|
| 3022 |
|
## NOTE: The adoption agency algorithm (AAA). |
| 3023 |
|
|
| 3024 |
FET: { |
FET: { |
| 3025 |
## Step 1 |
## Step 1 |
| 3026 |
my $formatting_element; |
my $formatting_element; |
| 3061 |
return; |
return; |
| 3062 |
} |
} |
| 3063 |
} elsif ({ |
} elsif ({ |
| 3064 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 3065 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 3066 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 3067 |
!!!cp ('t56'); |
!!!cp ('t56'); |
| 3088 |
if (not $formatting_category->{$node->[1]} and |
if (not $formatting_category->{$node->[1]} and |
| 3089 |
#not $phrasing_category->{$node->[1]} and |
#not $phrasing_category->{$node->[1]} and |
| 3090 |
($special_category->{$node->[1]} or |
($special_category->{$node->[1]} or |
| 3091 |
$scoping_category->{$node->[1]})) { |
$scoping_category->{$node->[1]})) { ## Scoping is redundant, maybe |
| 3092 |
!!!cp ('t59'); |
!!!cp ('t59'); |
| 3093 |
$furthest_block = $node; |
$furthest_block = $node; |
| 3094 |
$furthest_block_i_in_open = $_; |
$furthest_block_i_in_open = $_; |
| 3174 |
} # S7 |
} # S7 |
| 3175 |
|
|
| 3176 |
## Step 8 |
## Step 8 |
| 3177 |
$common_ancestor_node->[0]->append_child ($last_node->[0]); |
if ({ |
| 3178 |
|
table => 1, tbody => 1, tfoot => 1, thead => 1, tr => 1, |
| 3179 |
|
}->{$common_ancestor_node->[1]}) { |
| 3180 |
|
my $foster_parent_element; |
| 3181 |
|
my $next_sibling; |
| 3182 |
|
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3183 |
|
if ($self->{open_elements}->[$_]->[1] eq 'table') { |
| 3184 |
|
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
| 3185 |
|
if (defined $parent and $parent->node_type == 1) { |
| 3186 |
|
!!!cp ('t65.1'); |
| 3187 |
|
$foster_parent_element = $parent; |
| 3188 |
|
$next_sibling = $self->{open_elements}->[$_]->[0]; |
| 3189 |
|
} else { |
| 3190 |
|
!!!cp ('t65.2'); |
| 3191 |
|
$foster_parent_element |
| 3192 |
|
= $self->{open_elements}->[$_ - 1]->[0]; |
| 3193 |
|
} |
| 3194 |
|
last OE; |
| 3195 |
|
} |
| 3196 |
|
} # OE |
| 3197 |
|
$foster_parent_element = $self->{open_elements}->[0]->[0] |
| 3198 |
|
unless defined $foster_parent_element; |
| 3199 |
|
$foster_parent_element->insert_before ($last_node->[0], $next_sibling); |
| 3200 |
|
$open_tables->[-1]->[1] = 1; # tainted |
| 3201 |
|
} else { |
| 3202 |
|
!!!cp ('t65.3'); |
| 3203 |
|
$common_ancestor_node->[0]->append_child ($last_node->[0]); |
| 3204 |
|
} |
| 3205 |
|
|
| 3206 |
## Step 9 |
## Step 9 |
| 3207 |
my $clone = [$formatting_element->[0]->clone_node (0), |
my $clone = [$formatting_element->[0]->clone_node (0), |
| 3247 |
} # FET |
} # FET |
| 3248 |
}; # $formatting_end_tag |
}; # $formatting_end_tag |
| 3249 |
|
|
| 3250 |
my $insert_to_current = sub { |
$insert = my $insert_to_current = sub { |
| 3251 |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
| 3252 |
}; # $insert_to_current |
}; # $insert_to_current |
| 3253 |
|
|
| 3254 |
my $insert_to_foster = sub { |
my $insert_to_foster = sub { |
| 3255 |
my $child = shift; |
my $child = shift; |
| 3256 |
if ({ |
if ({ |
| 3257 |
table => 1, tbody => 1, tfoot => 1, |
table => 1, tbody => 1, tfoot => 1, thead => 1, tr => 1, |
| 3258 |
thead => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 3259 |
}->{$self->{open_elements}->[-1]->[1]}) { |
# MUST |
| 3260 |
# MUST |
my $foster_parent_element; |
| 3261 |
my $foster_parent_element; |
my $next_sibling; |
|
my $next_sibling; |
|
| 3262 |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3263 |
if ($self->{open_elements}->[$_]->[1] eq 'table') { |
if ($self->{open_elements}->[$_]->[1] eq 'table') { |
| 3264 |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
| 3278 |
unless defined $foster_parent_element; |
unless defined $foster_parent_element; |
| 3279 |
$foster_parent_element->insert_before |
$foster_parent_element->insert_before |
| 3280 |
($child, $next_sibling); |
($child, $next_sibling); |
| 3281 |
} else { |
$open_tables->[-1]->[1] = 1; # tainted |
| 3282 |
!!!cp ('t72'); |
} else { |
| 3283 |
$self->{open_elements}->[-1]->[0]->append_child ($child); |
!!!cp ('t72'); |
| 3284 |
} |
$self->{open_elements}->[-1]->[0]->append_child ($child); |
| 3285 |
|
} |
| 3286 |
}; # $insert_to_foster |
}; # $insert_to_foster |
| 3287 |
|
|
|
my $insert; |
|
|
|
|
| 3288 |
B: { |
B: { |
| 3289 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 3290 |
!!!cp ('t73'); |
!!!cp ('t73'); |
| 3299 |
# |
# |
| 3300 |
} else { |
} else { |
| 3301 |
## Generate implied end tags |
## Generate implied end tags |
| 3302 |
if ({ |
while ({ |
| 3303 |
dd => 1, dt => 1, li => 1, p => 1, td => 1, th => 1, tr => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 3304 |
tbody => 1, tfoot=> 1, thead => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 3305 |
!!!cp ('t75'); |
!!!cp ('t75'); |
| 3306 |
!!!back-token; |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, tag_name => $self->{open_elements}->[-1]->[1]}; |
|
|
redo B; |
|
| 3307 |
} |
} |
| 3308 |
|
|
| 3309 |
if (@{$self->{open_elements}} > 2 or |
if (@{$self->{open_elements}} > 2 or |
| 3329 |
$token->{tag_name} eq 'html') { |
$token->{tag_name} eq 'html') { |
| 3330 |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) { |
| 3331 |
!!!cp ('t79'); |
!!!cp ('t79'); |
|
## Turn into the main phase |
|
| 3332 |
!!!parse-error (type => 'after html:html'); |
!!!parse-error (type => 'after html:html'); |
| 3333 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 3334 |
} elsif ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
} elsif ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) { |
| 3335 |
!!!cp ('t80'); |
!!!cp ('t80'); |
|
## Turn into the main phase |
|
| 3336 |
!!!parse-error (type => 'after html:html'); |
!!!parse-error (type => 'after html:html'); |
| 3337 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 3338 |
} else { |
} else { |
| 3339 |
!!!cp ('t81'); |
!!!cp ('t81'); |
| 3340 |
} |
} |
| 3341 |
|
|
| 3342 |
## ISSUE: "aa<html>" is not a parse error. |
!!!cp ('t82'); |
| 3343 |
## ISSUE: "<html>" in fragment is not a parse error. |
!!!parse-error (type => 'not first start tag'); |
|
unless ($token->{first_start_tag}) { |
|
|
!!!cp ('t82'); |
|
|
!!!parse-error (type => 'not first start tag'); |
|
|
} else { |
|
|
!!!cp ('t83'); |
|
|
} |
|
| 3344 |
my $top_el = $self->{open_elements}->[0]->[0]; |
my $top_el = $self->{open_elements}->[0]->[0]; |
| 3345 |
for my $attr_name (keys %{$token->{attributes}}) { |
for my $attr_name (keys %{$token->{attributes}}) { |
| 3346 |
unless ($top_el->has_attribute_ns (undef, $attr_name)) { |
unless ($top_el->has_attribute_ns (undef, $attr_name)) { |
| 3369 |
} elsif ($self->{insertion_mode} & HEAD_IMS) { |
} elsif ($self->{insertion_mode} & HEAD_IMS) { |
| 3370 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 3371 |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
| 3372 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
unless ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 3373 |
|
!!!cp ('t88.2'); |
| 3374 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
| 3375 |
|
} else { |
| 3376 |
|
!!!cp ('t88.1'); |
| 3377 |
|
## Ignore the token. |
| 3378 |
|
!!!next-token; |
| 3379 |
|
redo B; |
| 3380 |
|
} |
| 3381 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 3382 |
!!!cp ('t88'); |
!!!cp ('t88'); |
| 3383 |
!!!next-token; |
!!!next-token; |
| 3478 |
} |
} |
| 3479 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
| 3480 |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 3481 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3482 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3483 |
!!!next-token; |
!!!next-token; |
| 3484 |
redo B; |
redo B; |
| 3493 |
} |
} |
| 3494 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
| 3495 |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec. |
| 3496 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3497 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3498 |
!!!next-token; |
!!!next-token; |
| 3499 |
redo B; |
redo B; |
| 3554 |
} |
} |
| 3555 |
} |
} |
| 3556 |
|
|
| 3557 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3558 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3559 |
!!!next-token; |
!!!next-token; |
| 3560 |
redo B; |
redo B; |
| 3578 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3579 |
my $parent = defined $self->{head_element} ? $self->{head_element} |
my $parent = defined $self->{head_element} ? $self->{head_element} |
| 3580 |
: $self->{open_elements}->[-1]->[0]; |
: $self->{open_elements}->[-1]->[0]; |
| 3581 |
$parse_rcdata->(RCDATA_CONTENT_MODEL, |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
| 3582 |
sub { $parent->append_child ($_[0]) }); |
pop @{$self->{open_elements}} # <head> |
|
pop @{$self->{open_elements}} |
|
| 3583 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3584 |
redo B; |
redo B; |
| 3585 |
} elsif ($token->{tag_name} eq 'style') { |
} elsif ($token->{tag_name} eq 'style') { |
| 3593 |
} else { |
} else { |
| 3594 |
!!!cp ('t115'); |
!!!cp ('t115'); |
| 3595 |
} |
} |
| 3596 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert_to_current); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 3597 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3598 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3599 |
redo B; |
redo B; |
| 3600 |
} elsif ($token->{tag_name} eq 'noscript') { |
} elsif ($token->{tag_name} eq 'noscript') { |
| 3633 |
} |
} |
| 3634 |
|
|
| 3635 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 3636 |
$script_start_tag->($insert_to_current); |
$script_start_tag->(); |
| 3637 |
pop @{$self->{open_elements}} |
pop @{$self->{open_elements}} # <head> |
| 3638 |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
if $self->{insertion_mode} == AFTER_HEAD_IM; |
| 3639 |
redo B; |
redo B; |
| 3640 |
} elsif ($token->{tag_name} eq 'body' or |
} elsif ($token->{tag_name} eq 'body' or |
| 3914 |
} # INSCOPE |
} # INSCOPE |
| 3915 |
unless (defined $i) { |
unless (defined $i) { |
| 3916 |
!!!cp ('t157'); |
!!!cp ('t157'); |
| 3917 |
|
## TODO: this type is wrong. |
| 3918 |
!!!parse-error (type => 'unmatched end tag:caption'); |
!!!parse-error (type => 'unmatched end tag:caption'); |
| 3919 |
## Ignore the token |
## Ignore the token |
| 3920 |
!!!next-token; |
!!!next-token; |
| 3922 |
} |
} |
| 3923 |
|
|
| 3924 |
## generate implied end tags |
## generate implied end tags |
| 3925 |
if ({ |
while ({ |
| 3926 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 3927 |
td => 1, th => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 3928 |
!!!cp ('t158'); |
!!!cp ('t158'); |
| 3929 |
!!!back-token; # <?> |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, tag_name => 'caption'}; |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 3930 |
} |
} |
| 3931 |
|
|
| 3932 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
| 3979 |
} |
} |
| 3980 |
|
|
| 3981 |
## generate implied end tags |
## generate implied end tags |
| 3982 |
if ({ |
while ({ |
| 3983 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 3984 |
td => ($token->{tag_name} eq 'th'), |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
th => ($token->{tag_name} eq 'td'), |
|
|
tr => 1, |
|
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 3985 |
!!!cp ('t166'); |
!!!cp ('t166'); |
| 3986 |
!!!back-token; |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 3987 |
} |
} |
| 3988 |
|
|
| 3989 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
| 3990 |
!!!cp ('t167'); |
!!!cp ('t167'); |
| 3991 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
| 4037 |
} |
} |
| 4038 |
|
|
| 4039 |
## generate implied end tags |
## generate implied end tags |
| 4040 |
if ({ |
while ({ |
| 4041 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 4042 |
td => 1, th => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4043 |
!!!cp ('t174'); |
!!!cp ('t174'); |
| 4044 |
!!!back-token; |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 4045 |
} |
} |
| 4046 |
|
|
| 4047 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
| 4138 |
} |
} |
| 4139 |
|
|
| 4140 |
## generate implied end tags |
## generate implied end tags |
| 4141 |
if ({ |
while ({ |
| 4142 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 4143 |
td => 1, th => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4144 |
!!!cp ('t187'); |
!!!cp ('t187'); |
| 4145 |
!!!back-token; # </table> |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, tag_name => 'caption'}; |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 4146 |
} |
} |
| 4147 |
|
|
| 4148 |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
if ($self->{open_elements}->[-1]->[1] ne 'caption') { |
| 4195 |
# |
# |
| 4196 |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
| 4197 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 4198 |
if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
if (not $open_tables->[-1]->[1] and # tainted |
| 4199 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
$token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { |
| 4200 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
| 4201 |
|
|
| 4202 |
unless (length $token->{data}) { |
unless (length $token->{data}) { |
| 4203 |
!!!cp ('t194'); |
!!!cp ('t194'); |
| 4204 |
!!!next-token; |
!!!next-token; |
| 4205 |
redo B; |
redo B; |
| 4206 |
} else { |
} else { |
| 4207 |
!!!cp ('t195'); |
!!!cp ('t195'); |
| 4208 |
} |
} |
| 4209 |
} |
} |
| 4210 |
|
|
| 4211 |
!!!parse-error (type => 'in table:#character'); |
!!!parse-error (type => 'in table:#character'); |
| 4212 |
|
|
| 4253 |
($self->{document}->create_text_node ($token->{data}), |
($self->{document}->create_text_node ($token->{data}), |
| 4254 |
$next_sibling); |
$next_sibling); |
| 4255 |
} |
} |
| 4256 |
} else { |
$open_tables->[-1]->[1] = 1; # tainted |
| 4257 |
!!!cp ('t200'); |
} else { |
| 4258 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
!!!cp ('t200'); |
| 4259 |
} |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 4260 |
|
} |
| 4261 |
|
|
| 4262 |
!!!next-token; |
!!!next-token; |
| 4263 |
redo B; |
redo B; |
| 4264 |
} elsif ($token->{type} == START_TAG_TOKEN) { |
} elsif ($token->{type} == START_TAG_TOKEN) { |
| 4265 |
if ({ |
if ({ |
| 4266 |
tr => ($self->{insertion_mode} != IN_ROW_IM), |
tr => ($self->{insertion_mode} != IN_ROW_IM), |
| 4271 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
| 4272 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
$self->{open_elements}->[-1]->[1] ne 'html') { |
| 4273 |
!!!cp ('t201'); |
!!!cp ('t201'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4274 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4275 |
} |
} |
| 4276 |
|
|
| 4290 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
| 4291 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4292 |
!!!cp ('t203'); |
!!!cp ('t203'); |
| 4293 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this case be reached? |
| 4294 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4295 |
} |
} |
| 4296 |
|
|
| 4314 |
tr => 1, html => 1, |
tr => 1, html => 1, |
| 4315 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4316 |
!!!cp ('t207'); |
!!!cp ('t207'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4317 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4318 |
} |
} |
| 4319 |
|
|
| 4340 |
$i = $_; |
$i = $_; |
| 4341 |
last INSCOPE; |
last INSCOPE; |
| 4342 |
} elsif ({ |
} elsif ({ |
| 4343 |
table => 1, html => 1, |
html => 1, |
| 4344 |
|
|
| 4345 |
|
## NOTE: This element does not appear here, maybe. |
| 4346 |
|
table => 1, |
| 4347 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 4348 |
!!!cp ('t209'); |
!!!cp ('t209'); |
| 4349 |
last INSCOPE; |
last INSCOPE; |
| 4351 |
} # INSCOPE |
} # INSCOPE |
| 4352 |
unless (defined $i) { |
unless (defined $i) { |
| 4353 |
!!!cp ('t210'); |
!!!cp ('t210'); |
| 4354 |
|
## TODO: This type is wrong. |
| 4355 |
!!!parse-error (type => 'unmacthed end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmacthed end tag:'.$token->{tag_name}); |
| 4356 |
## Ignore the token |
## Ignore the token |
| 4357 |
!!!next-token; |
!!!next-token; |
| 4363 |
tr => 1, html => 1, |
tr => 1, html => 1, |
| 4364 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4365 |
!!!cp ('t211'); |
!!!cp ('t211'); |
| 4366 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this case be reached? |
| 4367 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4368 |
} |
} |
| 4369 |
|
|
| 4411 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
| 4412 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4413 |
!!!cp ('t217'); |
!!!cp ('t217'); |
| 4414 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this state be reached? |
| 4415 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4416 |
} |
} |
| 4417 |
|
|
| 4434 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
| 4435 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
$self->{open_elements}->[-1]->[1] ne 'html') { |
| 4436 |
!!!cp ('t219'); |
!!!cp ('t219'); |
| 4437 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this state be reached? |
| 4438 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4439 |
} |
} |
| 4440 |
|
|
| 4451 |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
while ($self->{open_elements}->[-1]->[1] ne 'table' and |
| 4452 |
$self->{open_elements}->[-1]->[1] ne 'html') { |
$self->{open_elements}->[-1]->[1] ne 'html') { |
| 4453 |
!!!cp ('t220'); |
!!!cp ('t220'); |
| 4454 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this state be reached? |
| 4455 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4456 |
} |
} |
| 4457 |
|
|
| 4484 |
$i = $_; |
$i = $_; |
| 4485 |
last INSCOPE; |
last INSCOPE; |
| 4486 |
} elsif ({ |
} elsif ({ |
| 4487 |
table => 1, html => 1, |
#table => 1, |
| 4488 |
|
html => 1, |
| 4489 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 4490 |
!!!cp ('t222'); |
!!!cp ('t222'); |
| 4491 |
last INSCOPE; |
last INSCOPE; |
| 4493 |
} # INSCOPE |
} # INSCOPE |
| 4494 |
unless (defined $i) { |
unless (defined $i) { |
| 4495 |
!!!cp ('t223'); |
!!!cp ('t223'); |
| 4496 |
|
## TODO: The following is wrong, maybe. |
| 4497 |
!!!parse-error (type => 'unmatched end tag:table'); |
!!!parse-error (type => 'unmatched end tag:table'); |
| 4498 |
## Ignore tokens </table><table> |
## Ignore tokens </table><table> |
| 4499 |
!!!next-token; |
!!!next-token; |
| 4501 |
} |
} |
| 4502 |
|
|
| 4503 |
## generate implied end tags |
## generate implied end tags |
| 4504 |
if ({ |
while ({ |
| 4505 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 4506 |
td => 1, th => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 4507 |
!!!cp ('t224'); |
!!!cp ('t224'); |
| 4508 |
!!!back-token; # <table> |
pop @{$self->{open_elements}}; |
|
$token = {type => END_TAG_TOKEN, tag_name => 'table'}; |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 4509 |
} |
} |
| 4510 |
|
|
| 4511 |
if ($self->{open_elements}->[-1]->[1] ne 'table') { |
if ($self->{open_elements}->[-1]->[1] ne 'table') { |
| 4512 |
!!!cp ('t225'); |
!!!cp ('t225'); |
| 4513 |
|
## ISSUE: Can this case be reached? |
| 4514 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
| 4515 |
} else { |
} else { |
| 4516 |
!!!cp ('t226'); |
!!!cp ('t226'); |
| 4517 |
} |
} |
| 4518 |
|
|
| 4519 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 4520 |
|
pop @{$open_tables}; |
| 4521 |
|
|
| 4522 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 4523 |
|
|
| 4524 |
## reprocess |
## reprocess |
| 4525 |
redo B; |
redo B; |
| 4526 |
|
} elsif ($token->{tag_name} eq 'style') { |
| 4527 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 4528 |
|
!!!cp ('t227.8'); |
| 4529 |
|
## NOTE: This is a "as if in head" code clone. |
| 4530 |
|
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 4531 |
|
redo B; |
| 4532 |
|
} else { |
| 4533 |
|
!!!cp ('t227.7'); |
| 4534 |
|
# |
| 4535 |
|
} |
| 4536 |
|
} elsif ($token->{tag_name} eq 'script') { |
| 4537 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 4538 |
|
!!!cp ('t227.6'); |
| 4539 |
|
## NOTE: This is a "as if in head" code clone. |
| 4540 |
|
$script_start_tag->(); |
| 4541 |
|
redo B; |
| 4542 |
|
} else { |
| 4543 |
|
!!!cp ('t227.5'); |
| 4544 |
|
# |
| 4545 |
|
} |
| 4546 |
|
} elsif ($token->{tag_name} eq 'input') { |
| 4547 |
|
if (not $open_tables->[-1]->[1]) { # tainted |
| 4548 |
|
if ($token->{attributes}->{type}) { ## TODO: case |
| 4549 |
|
my $type = lc $token->{attributes}->{type}->{value}; |
| 4550 |
|
if ($type eq 'hidden') { |
| 4551 |
|
!!!cp ('t227.3'); |
| 4552 |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}); |
| 4553 |
|
|
| 4554 |
|
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
| 4555 |
|
|
| 4556 |
|
## TODO: form element pointer |
| 4557 |
|
|
| 4558 |
|
pop @{$self->{open_elements}}; |
| 4559 |
|
|
| 4560 |
|
!!!next-token; |
| 4561 |
|
redo B; |
| 4562 |
|
} else { |
| 4563 |
|
!!!cp ('t227.2'); |
| 4564 |
|
# |
| 4565 |
|
} |
| 4566 |
|
} else { |
| 4567 |
|
!!!cp ('t227.1'); |
| 4568 |
|
# |
| 4569 |
|
} |
| 4570 |
|
} else { |
| 4571 |
|
!!!cp ('t227.4'); |
| 4572 |
|
# |
| 4573 |
|
} |
| 4574 |
} else { |
} else { |
| 4575 |
!!!cp ('t227'); |
!!!cp ('t227'); |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}); |
|
|
|
|
|
$insert = $insert_to_foster; |
|
| 4576 |
# |
# |
| 4577 |
} |
} |
| 4578 |
|
|
| 4579 |
|
!!!parse-error (type => 'in table:'.$token->{tag_name}); |
| 4580 |
|
|
| 4581 |
|
$insert = $insert_to_foster; |
| 4582 |
|
# |
| 4583 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 4584 |
if ($token->{tag_name} eq 'tr' and |
if ($token->{tag_name} eq 'tr' and |
| 4585 |
$self->{insertion_mode} == IN_ROW_IM) { |
$self->{insertion_mode} == IN_ROW_IM) { |
| 4613 |
tr => 1, html => 1, |
tr => 1, html => 1, |
| 4614 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4615 |
!!!cp ('t231'); |
!!!cp ('t231'); |
| 4616 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this state be reached? |
| 4617 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4618 |
} |
} |
| 4619 |
|
|
| 4641 |
} # INSCOPE |
} # INSCOPE |
| 4642 |
unless (defined $i) { |
unless (defined $i) { |
| 4643 |
!!!cp ('t235'); |
!!!cp ('t235'); |
| 4644 |
|
## TODO: The following is wrong. |
| 4645 |
!!!parse-error (type => 'unmatched end tag:'.$token->{type}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{type}); |
| 4646 |
## Ignore the token |
## Ignore the token |
| 4647 |
!!!next-token; |
!!!next-token; |
| 4653 |
tr => 1, html => 1, |
tr => 1, html => 1, |
| 4654 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4655 |
!!!cp ('t236'); |
!!!cp ('t236'); |
| 4656 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this state be reached? |
| 4657 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4658 |
} |
} |
| 4659 |
|
|
| 4693 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
| 4694 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4695 |
!!!cp ('t240'); |
!!!cp ('t240'); |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
| 4696 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4697 |
} |
} |
| 4698 |
|
|
| 4708 |
## reprocess in the "in table" insertion mode... |
## reprocess in the "in table" insertion mode... |
| 4709 |
} |
} |
| 4710 |
|
|
| 4711 |
|
## NOTE: </table> in the "in table" insertion mode. |
| 4712 |
|
## When you edit the code fragment below, please ensure that |
| 4713 |
|
## the code for <table> in the "in table" insertion mode |
| 4714 |
|
## is synced with it. |
| 4715 |
|
|
| 4716 |
## have a table element in table scope |
## have a table element in table scope |
| 4717 |
my $i; |
my $i; |
| 4718 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4735 |
!!!next-token; |
!!!next-token; |
| 4736 |
redo B; |
redo B; |
| 4737 |
} |
} |
|
|
|
|
## generate implied end tags |
|
|
if ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
td => 1, th => 1, tr => 1, |
|
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t244'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
|
} |
|
|
|
|
|
if ($self->{open_elements}->[-1]->[1] ne 'table') { |
|
|
!!!cp ('t245'); |
|
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
|
|
} else { |
|
|
!!!cp ('t246'); |
|
|
} |
|
| 4738 |
|
|
| 4739 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 4740 |
|
pop @{$open_tables}; |
| 4741 |
|
|
| 4742 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 4743 |
|
|
| 4800 |
tr => 1, html => 1, |
tr => 1, html => 1, |
| 4801 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4802 |
!!!cp ('t253'); |
!!!cp ('t253'); |
| 4803 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this case be reached? |
| 4804 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4805 |
} |
} |
| 4806 |
|
|
| 4837 |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
tbody => 1, tfoot => 1, thead => 1, html => 1, |
| 4838 |
}->{$self->{open_elements}->[-1]->[1]}) { |
}->{$self->{open_elements}->[-1]->[1]}) { |
| 4839 |
!!!cp ('t257'); |
!!!cp ('t257'); |
| 4840 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
## ISSUE: Can this case be reached? |
| 4841 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 4842 |
} |
} |
| 4843 |
|
|
| 4916 |
# |
# |
| 4917 |
} |
} |
| 4918 |
} else { |
} else { |
| 4919 |
!!!cp ('t268'); |
die "$0: $token->{type}: Unknown token type"; |
|
# |
|
| 4920 |
} |
} |
| 4921 |
|
|
| 4922 |
## As if </colgroup> |
## As if </colgroup> |
| 4933 |
## reprocess |
## reprocess |
| 4934 |
redo B; |
redo B; |
| 4935 |
} |
} |
| 4936 |
} elsif ($self->{insertion_mode} == IN_SELECT_IM) { |
} elsif ($self->{insertion_mode} & SELECT_IMS) { |
| 4937 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 4938 |
!!!cp ('t271'); |
!!!cp ('t271'); |
| 4939 |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
$self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data}); |
| 4972 |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
!!!insert-element ($token->{tag_name}, $token->{attributes}); |
| 4973 |
!!!next-token; |
!!!next-token; |
| 4974 |
redo B; |
redo B; |
| 4975 |
} elsif ($token->{tag_name} eq 'select') { |
} elsif ($token->{tag_name} eq 'select' or |
| 4976 |
!!!parse-error (type => 'not closed:select'); |
$token->{tag_name} eq 'input' or |
| 4977 |
## As if </select> instead |
($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and |
| 4978 |
|
{ |
| 4979 |
|
caption => 1, table => 1, |
| 4980 |
|
tbody => 1, tfoot => 1, thead => 1, |
| 4981 |
|
tr => 1, td => 1, th => 1, |
| 4982 |
|
}->{$token->{tag_name}})) { |
| 4983 |
|
## TODO: The type below is not good - <select> is replaced by </select> |
| 4984 |
|
!!!parse-error (type => 'not closed:select'); |
| 4985 |
|
## NOTE: As if the token were </select> (<select> case) or |
| 4986 |
|
## as if there were </select> (otherwise). |
| 4987 |
## have an element in table scope |
## have an element in table scope |
| 4988 |
my $i; |
my $i; |
| 4989 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 4990 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 4991 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] eq 'select') { |
| 4992 |
!!!cp ('t278'); |
!!!cp ('t278'); |
| 4993 |
$i = $_; |
$i = $_; |
| 4994 |
last INSCOPE; |
last INSCOPE; |
| 5012 |
|
|
| 5013 |
$self->_reset_insertion_mode; |
$self->_reset_insertion_mode; |
| 5014 |
|
|
| 5015 |
!!!next-token; |
if ($token->{tag_name} eq 'select') { |
| 5016 |
redo B; |
!!!cp ('t281.2'); |
| 5017 |
|
!!!next-token; |
| 5018 |
|
redo B; |
| 5019 |
|
} else { |
| 5020 |
|
!!!cp ('t281.1'); |
| 5021 |
|
## Reprocess the token. |
| 5022 |
|
redo B; |
| 5023 |
|
} |
| 5024 |
} else { |
} else { |
| 5025 |
!!!cp ('t282'); |
!!!cp ('t282'); |
| 5026 |
!!!parse-error (type => 'in select:'.$token->{tag_name}); |
!!!parse-error (type => 'in select:'.$token->{tag_name}); |
| 5087 |
|
|
| 5088 |
!!!next-token; |
!!!next-token; |
| 5089 |
redo B; |
redo B; |
| 5090 |
} elsif ({ |
} elsif ($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and |
| 5091 |
caption => 1, table => 1, tbody => 1, |
{ |
| 5092 |
tfoot => 1, thead => 1, tr => 1, td => 1, th => 1, |
caption => 1, table => 1, tbody => 1, |
| 5093 |
}->{$token->{tag_name}}) { |
tfoot => 1, thead => 1, tr => 1, td => 1, th => 1, |
| 5094 |
|
}->{$token->{tag_name}}) { |
| 5095 |
|
## TODO: The following is wrong? |
| 5096 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 5097 |
|
|
| 5098 |
## have an element in table scope |
## have an element in table scope |
| 5129 |
} elsif ({ |
} elsif ({ |
| 5130 |
table => 1, html => 1, |
table => 1, html => 1, |
| 5131 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 5132 |
|
## ISSUE: Can this state be reached? |
| 5133 |
!!!cp ('t296'); |
!!!cp ('t296'); |
| 5134 |
last INSCOPE; |
last INSCOPE; |
| 5135 |
} |
} |
| 5136 |
} # INSCOPE |
} # INSCOPE |
| 5137 |
unless (defined $i) { |
unless (defined $i) { |
| 5138 |
!!!cp ('t297'); |
!!!cp ('t297'); |
| 5139 |
|
## TODO: The following error type is correct? |
| 5140 |
!!!parse-error (type => 'unmatched end tag:select'); |
!!!parse-error (type => 'unmatched end tag:select'); |
| 5141 |
## Ignore the </select> token |
## Ignore the </select> token |
| 5142 |
!!!next-token; ## TODO: ok? |
!!!next-token; ## TODO: ok? |
| 5180 |
!!!cp ('t301'); |
!!!cp ('t301'); |
| 5181 |
!!!parse-error (type => 'after html:#character'); |
!!!parse-error (type => 'after html:#character'); |
| 5182 |
|
|
| 5183 |
## Reprocess in the "main" phase, "after body" insertion mode... |
## Reprocess in the "after body" insertion mode. |
| 5184 |
} else { |
} else { |
| 5185 |
!!!cp ('t302'); |
!!!cp ('t302'); |
| 5186 |
} |
} |
| 5196 |
!!!cp ('t303'); |
!!!cp ('t303'); |
| 5197 |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
| 5198 |
|
|
| 5199 |
## Reprocess in the "main" phase, "after body" insertion mode... |
## Reprocess in the "after body" insertion mode. |
| 5200 |
} else { |
} else { |
| 5201 |
!!!cp ('t304'); |
!!!cp ('t304'); |
| 5202 |
} |
} |
| 5213 |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
| 5214 |
|
|
| 5215 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 5216 |
## Reprocess in the "main" phase, "after body" insertion mode... |
## Reprocess in the "after body" insertion mode. |
| 5217 |
} else { |
} else { |
| 5218 |
!!!cp ('t306'); |
!!!cp ('t306'); |
| 5219 |
} |
} |
| 5267 |
!!!parse-error (type => 'after html:#character'); |
!!!parse-error (type => 'after html:#character'); |
| 5268 |
|
|
| 5269 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5270 |
## Reprocess in the "main" phase, "after frameset"... |
## Reprocess in the "after frameset" insertion mode. |
| 5271 |
!!!parse-error (type => 'after frameset:#character'); |
!!!parse-error (type => 'after frameset:#character'); |
| 5272 |
} |
} |
| 5273 |
|
|
| 5289 |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
!!!parse-error (type => 'after html:'.$token->{tag_name}); |
| 5290 |
|
|
| 5291 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5292 |
## Process in the "main" phase, "after frameset" insertion mode... |
## Process in the "after frameset" insertion mode. |
| 5293 |
} else { |
} else { |
| 5294 |
!!!cp ('t317'); |
!!!cp ('t317'); |
| 5295 |
} |
} |
| 5310 |
} elsif ($token->{tag_name} eq 'noframes') { |
} elsif ($token->{tag_name} eq 'noframes') { |
| 5311 |
!!!cp ('t320'); |
!!!cp ('t320'); |
| 5312 |
## NOTE: As if in body. |
## NOTE: As if in body. |
| 5313 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert_to_current); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5314 |
redo B; |
redo B; |
| 5315 |
} else { |
} else { |
| 5316 |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
if ($self->{insertion_mode} == IN_FRAMESET_IM) { |
| 5330 |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
!!!parse-error (type => 'after html:/'.$token->{tag_name}); |
| 5331 |
|
|
| 5332 |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
$self->{insertion_mode} = AFTER_FRAMESET_IM; |
| 5333 |
## Process in the "main" phase, "after frameset" insertion mode... |
## Process in the "after frameset" insertion mode. |
| 5334 |
} else { |
} else { |
| 5335 |
!!!cp ('t324'); |
!!!cp ('t324'); |
| 5336 |
} |
} |
| 5389 |
if ($token->{tag_name} eq 'script') { |
if ($token->{tag_name} eq 'script') { |
| 5390 |
!!!cp ('t332'); |
!!!cp ('t332'); |
| 5391 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5392 |
$script_start_tag->($insert); |
$script_start_tag->(); |
| 5393 |
redo B; |
redo B; |
| 5394 |
} elsif ($token->{tag_name} eq 'style') { |
} elsif ($token->{tag_name} eq 'style') { |
| 5395 |
!!!cp ('t333'); |
!!!cp ('t333'); |
| 5396 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5397 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5398 |
redo B; |
redo B; |
| 5399 |
} elsif ({ |
} elsif ({ |
| 5400 |
base => 1, link => 1, |
base => 1, link => 1, |
| 5457 |
redo B; |
redo B; |
| 5458 |
} elsif ($token->{tag_name} eq 'title') { |
} elsif ($token->{tag_name} eq 'title') { |
| 5459 |
!!!cp ('t341'); |
!!!cp ('t341'); |
|
!!!parse-error (type => 'in body:title'); |
|
| 5460 |
## NOTE: This is an "as if in head" code clone |
## NOTE: This is an "as if in head" code clone |
| 5461 |
$parse_rcdata->(RCDATA_CONTENT_MODEL, sub { |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
|
if (defined $self->{head_element}) { |
|
|
!!!cp ('t339'); |
|
|
$self->{head_element}->append_child ($_[0]); |
|
|
} else { |
|
|
!!!cp ('t340'); |
|
|
$insert->($_[0]); |
|
|
} |
|
|
}); |
|
| 5462 |
redo B; |
redo B; |
| 5463 |
} elsif ($token->{tag_name} eq 'body') { |
} elsif ($token->{tag_name} eq 'body') { |
| 5464 |
!!!parse-error (type => 'in body:body'); |
!!!parse-error (type => 'in body:body'); |
| 5482 |
redo B; |
redo B; |
| 5483 |
} elsif ({ |
} elsif ({ |
| 5484 |
address => 1, blockquote => 1, center => 1, dir => 1, |
address => 1, blockquote => 1, center => 1, dir => 1, |
| 5485 |
div => 1, dl => 1, fieldset => 1, listing => 1, |
div => 1, dl => 1, fieldset => 1, |
| 5486 |
|
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
| 5487 |
menu => 1, ol => 1, p => 1, ul => 1, |
menu => 1, ol => 1, p => 1, ul => 1, |
| 5488 |
pre => 1, |
pre => 1, listing => 1, |
| 5489 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 5490 |
## has a p element in scope |
## has a p element in scope |
| 5491 |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
| 5495 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5496 |
redo B; |
redo B; |
| 5497 |
} elsif ({ |
} elsif ({ |
| 5498 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5499 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5500 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5501 |
!!!cp ('t345'); |
!!!cp ('t345'); |
| 5504 |
} # INSCOPE |
} # INSCOPE |
| 5505 |
|
|
| 5506 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
| 5507 |
if ($token->{tag_name} eq 'pre') { |
if ($token->{tag_name} eq 'pre' or $token->{tag_name} eq 'listing') { |
| 5508 |
!!!next-token; |
!!!next-token; |
| 5509 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == CHARACTER_TOKEN) { |
| 5510 |
$token->{data} =~ s/^\x0A//; |
$token->{data} =~ s/^\x0A//; |
| 5538 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5539 |
redo B; |
redo B; |
| 5540 |
} elsif ({ |
} elsif ({ |
| 5541 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5542 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5543 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5544 |
!!!cp ('t352'); |
!!!cp ('t352'); |
| 5560 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5561 |
redo B; |
redo B; |
| 5562 |
} elsif ({ |
} elsif ({ |
| 5563 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5564 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5565 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5566 |
!!!cp ('t354'); |
!!!cp ('t354'); |
| 5616 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5617 |
redo B; |
redo B; |
| 5618 |
} elsif ({ |
} elsif ({ |
| 5619 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5620 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5621 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5622 |
!!!cp ('t361'); |
!!!cp ('t361'); |
| 5672 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5673 |
redo B; |
redo B; |
| 5674 |
} elsif ({ |
} elsif ({ |
| 5675 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5676 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5677 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5678 |
!!!cp ('t368'); |
!!!cp ('t368'); |
| 5686 |
|
|
| 5687 |
!!!next-token; |
!!!next-token; |
| 5688 |
redo B; |
redo B; |
|
} elsif ({ |
|
|
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
|
|
}->{$token->{tag_name}}) { |
|
|
## has a p element in scope |
|
|
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
|
|
my $node = $self->{open_elements}->[$_]; |
|
|
if ($node->[1] eq 'p') { |
|
|
!!!cp ('t369'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
|
|
redo B; |
|
|
} elsif ({ |
|
|
table => 1, caption => 1, td => 1, th => 1, |
|
|
button => 1, marquee => 1, object => 1, html => 1, |
|
|
}->{$node->[1]}) { |
|
|
!!!cp ('t370'); |
|
|
last INSCOPE; |
|
|
} |
|
|
} # INSCOPE |
|
|
|
|
|
## NOTE: See <http://html5.org/tools/web-apps-tracker?from=925&to=926> |
|
|
## has an element in scope |
|
|
#my $i; |
|
|
#INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
|
|
# my $node = $self->{open_elements}->[$_]; |
|
|
# if ({ |
|
|
# h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
|
|
# }->{$node->[1]}) { |
|
|
# $i = $_; |
|
|
# last INSCOPE; |
|
|
# } elsif ({ |
|
|
# table => 1, caption => 1, td => 1, th => 1, |
|
|
# button => 1, marquee => 1, object => 1, html => 1, |
|
|
# }->{$node->[1]}) { |
|
|
# last INSCOPE; |
|
|
# } |
|
|
#} # INSCOPE |
|
|
# |
|
|
#if (defined $i) { |
|
|
# !!! parse-error (type => 'in hn:hn'); |
|
|
# splice @{$self->{open_elements}}, $i; |
|
|
#} |
|
|
|
|
|
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
|
|
|
|
|
!!!next-token; |
|
|
redo B; |
|
| 5689 |
} elsif ($token->{tag_name} eq 'a') { |
} elsif ($token->{tag_name} eq 'a') { |
| 5690 |
AFE: for my $i (reverse 0..$#$active_formatting_elements) { |
AFE: for my $i (reverse 0..$#$active_formatting_elements) { |
| 5691 |
my $node = $active_formatting_elements->[$i]; |
my $node = $active_formatting_elements->[$i]; |
| 5751 |
$token = {type => END_TAG_TOKEN, tag_name => 'nobr'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'nobr'}; |
| 5752 |
redo B; |
redo B; |
| 5753 |
} elsif ({ |
} elsif ({ |
| 5754 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5755 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5756 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 5757 |
!!!cp ('t377'); |
!!!cp ('t377'); |
| 5775 |
$token = {type => END_TAG_TOKEN, tag_name => 'button'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'button'}; |
| 5776 |
redo B; |
redo B; |
| 5777 |
} elsif ({ |
} elsif ({ |
| 5778 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5779 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5780 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 5781 |
!!!cp ('t379'); |
!!!cp ('t379'); |
| 5786 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 5787 |
|
|
| 5788 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
| 5789 |
|
|
| 5790 |
|
## TODO: associate with $self->{form_element} if defined |
| 5791 |
|
|
| 5792 |
push @$active_formatting_elements, ['#marker', '']; |
push @$active_formatting_elements, ['#marker', '']; |
| 5793 |
|
|
| 5794 |
!!!next-token; |
!!!next-token; |
| 5795 |
redo B; |
redo B; |
| 5796 |
} elsif ($token->{tag_name} eq 'marquee' or |
} elsif ({ |
| 5797 |
$token->{tag_name} eq 'object') { |
applet => 1, marquee => 1, object => 1, |
| 5798 |
|
}->{$token->{tag_name}}) { |
| 5799 |
!!!cp ('t380'); |
!!!cp ('t380'); |
| 5800 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 5801 |
|
|
| 5807 |
} elsif ($token->{tag_name} eq 'xmp') { |
} elsif ($token->{tag_name} eq 'xmp') { |
| 5808 |
!!!cp ('t381'); |
!!!cp ('t381'); |
| 5809 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 5810 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5811 |
redo B; |
redo B; |
| 5812 |
} elsif ($token->{tag_name} eq 'table') { |
} elsif ($token->{tag_name} eq 'table') { |
| 5813 |
## has a p element in scope |
## has a p element in scope |
| 5818 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5819 |
redo B; |
redo B; |
| 5820 |
} elsif ({ |
} elsif ({ |
| 5821 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5822 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5823 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5824 |
!!!cp ('t383'); |
!!!cp ('t383'); |
| 5827 |
} # INSCOPE |
} # INSCOPE |
| 5828 |
|
|
| 5829 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
| 5830 |
|
push @{$open_tables}, [$self->{open_elements}->[-1]->[0]]; |
| 5831 |
|
|
| 5832 |
$self->{insertion_mode} = IN_TABLE_IM; |
$self->{insertion_mode} = IN_TABLE_IM; |
| 5833 |
|
|
| 5834 |
!!!next-token; |
!!!next-token; |
| 5863 |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
$token = {type => END_TAG_TOKEN, tag_name => 'p'}; |
| 5864 |
redo B; |
redo B; |
| 5865 |
} elsif ({ |
} elsif ({ |
| 5866 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 5867 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 5868 |
}->{$_->[1]}) { |
}->{$_->[1]}) { |
| 5869 |
!!!cp ('t387'); |
!!!cp ('t387'); |
| 5983 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 5984 |
!!!cp ('t399'); |
!!!cp ('t399'); |
| 5985 |
## NOTE: There is an "as if in body" code clone. |
## NOTE: There is an "as if in body" code clone. |
| 5986 |
$parse_rcdata->(CDATA_CONTENT_MODEL, $insert); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 5987 |
redo B; |
redo B; |
| 5988 |
} elsif ($token->{tag_name} eq 'select') { |
} elsif ($token->{tag_name} eq 'select') { |
| 5989 |
!!!cp ('t400'); |
!!!cp ('t400'); |
| 5990 |
$reconstruct_active_formatting_elements->($insert_to_current); |
$reconstruct_active_formatting_elements->($insert_to_current); |
| 5991 |
|
|
| 5992 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}); |
| 5993 |
|
|
| 5994 |
|
## TODO: associate with $self->{form_element} if defined |
| 5995 |
|
|
| 5996 |
$self->{insertion_mode} = IN_SELECT_IM; |
if ($self->{insertion_mode} & TABLE_IMS or |
| 5997 |
|
$self->{insertion_mode} & BODY_TABLE_IMS or |
| 5998 |
|
$self->{insertion_mode} == IN_COLUMN_GROUP_IM) { |
| 5999 |
|
!!!cp ('t400.1'); |
| 6000 |
|
$self->{insertion_mode} = IN_SELECT_IN_TABLE_IM; |
| 6001 |
|
} else { |
| 6002 |
|
!!!cp ('t400.2'); |
| 6003 |
|
$self->{insertion_mode} = IN_SELECT_IM; |
| 6004 |
|
} |
| 6005 |
!!!next-token; |
!!!next-token; |
| 6006 |
redo B; |
redo B; |
| 6007 |
} elsif ({ |
} elsif ({ |
| 6076 |
address => 1, blockquote => 1, center => 1, dir => 1, |
address => 1, blockquote => 1, center => 1, dir => 1, |
| 6077 |
div => 1, dl => 1, fieldset => 1, listing => 1, |
div => 1, dl => 1, fieldset => 1, listing => 1, |
| 6078 |
menu => 1, ol => 1, pre => 1, ul => 1, |
menu => 1, ol => 1, pre => 1, ul => 1, |
|
p => 1, |
|
| 6079 |
dd => 1, dt => 1, li => 1, |
dd => 1, dt => 1, li => 1, |
| 6080 |
button => 1, marquee => 1, object => 1, |
applet => 1, button => 1, marquee => 1, object => 1, |
| 6081 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 6082 |
## has an element in scope |
## has an element in scope |
| 6083 |
my $i; |
my $i; |
| 6084 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6085 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6086 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] eq $token->{tag_name}) { |
|
## generate implied end tags |
|
|
if ({ |
|
|
dd => ($token->{tag_name} ne 'dd'), |
|
|
dt => ($token->{tag_name} ne 'dt'), |
|
|
li => ($token->{tag_name} ne 'li'), |
|
|
p => ($token->{tag_name} ne 'p'), |
|
|
td => 1, th => 1, tr => 1, |
|
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t409'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
|
} |
|
|
|
|
| 6087 |
!!!cp ('t410'); |
!!!cp ('t410'); |
| 6088 |
$i = $_; |
$i = $_; |
| 6089 |
last INSCOPE unless $token->{tag_name} eq 'p'; |
last INSCOPE; |
| 6090 |
} elsif ({ |
} elsif ({ |
| 6091 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 6092 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 6093 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 6094 |
!!!cp ('t411'); |
!!!cp ('t411'); |
| 6095 |
last INSCOPE; |
last INSCOPE; |
| 6096 |
} |
} |
| 6097 |
} # INSCOPE |
} # INSCOPE |
| 6098 |
|
|
| 6099 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
unless (defined $i) { # has an element in scope |
| 6100 |
if (defined $i) { |
!!!cp ('t413'); |
| 6101 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 6102 |
|
} else { |
| 6103 |
|
## Step 1. generate implied end tags |
| 6104 |
|
while ({ |
| 6105 |
|
dd => ($token->{tag_name} ne 'dd'), |
| 6106 |
|
dt => ($token->{tag_name} ne 'dt'), |
| 6107 |
|
li => ($token->{tag_name} ne 'li'), |
| 6108 |
|
p => 1, |
| 6109 |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
| 6110 |
|
!!!cp ('t409'); |
| 6111 |
|
pop @{$self->{open_elements}}; |
| 6112 |
|
} |
| 6113 |
|
|
| 6114 |
|
## Step 2. |
| 6115 |
|
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
| 6116 |
!!!cp ('t412'); |
!!!cp ('t412'); |
| 6117 |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
| 6118 |
} else { |
} else { |
| 6119 |
!!!cp ('t413'); |
!!!cp ('t414'); |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
|
| 6120 |
} |
} |
| 6121 |
} |
|
| 6122 |
|
## Step 3. |
|
if (defined $i) { |
|
|
!!!cp ('t414'); |
|
| 6123 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 6124 |
} elsif ($token->{tag_name} eq 'p') { |
|
| 6125 |
!!!cp ('t415'); |
## Step 4. |
| 6126 |
## As if <p>, then reprocess the current token |
$clear_up_to_marker->() |
| 6127 |
my $el; |
if { |
| 6128 |
!!!create-element ($el, 'p'); |
applet => 1, button => 1, marquee => 1, object => 1, |
| 6129 |
$insert->($el); |
}->{$token->{tag_name}}; |
|
} else { |
|
|
!!!cp ('t416'); |
|
| 6130 |
} |
} |
|
$clear_up_to_marker->() |
|
|
if { |
|
|
button => 1, marquee => 1, object => 1, |
|
|
}->{$token->{tag_name}}; |
|
| 6131 |
!!!next-token; |
!!!next-token; |
| 6132 |
redo B; |
redo B; |
| 6133 |
} elsif ($token->{tag_name} eq 'form') { |
} elsif ($token->{tag_name} eq 'form') { |
| 6134 |
|
undef $self->{form_element}; |
| 6135 |
|
|
| 6136 |
## has an element in scope |
## has an element in scope |
| 6137 |
|
my $i; |
| 6138 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6139 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 6140 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] eq $token->{tag_name}) { |
|
## generate implied end tags |
|
|
if ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
td => 1, th => 1, tr => 1, |
|
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t417'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
|
} |
|
|
|
|
| 6141 |
!!!cp ('t418'); |
!!!cp ('t418'); |
| 6142 |
|
$i = $_; |
| 6143 |
last INSCOPE; |
last INSCOPE; |
| 6144 |
} elsif ({ |
} elsif ({ |
| 6145 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 6146 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 6147 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 6148 |
!!!cp ('t419'); |
!!!cp ('t419'); |
| 6149 |
last INSCOPE; |
last INSCOPE; |
| 6150 |
} |
} |
| 6151 |
} # INSCOPE |
} # INSCOPE |
| 6152 |
|
|
| 6153 |
if ($self->{open_elements}->[-1]->[1] eq $token->{tag_name}) { |
unless (defined $i) { # has an element in scope |
|
!!!cp ('t420'); |
|
|
pop @{$self->{open_elements}}; |
|
|
} else { |
|
| 6154 |
!!!cp ('t421'); |
!!!cp ('t421'); |
| 6155 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 6156 |
|
} else { |
| 6157 |
|
## Step 1. generate implied end tags |
| 6158 |
|
while ({ |
| 6159 |
|
dd => 1, dt => 1, li => 1, p => 1, |
| 6160 |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
| 6161 |
|
!!!cp ('t417'); |
| 6162 |
|
pop @{$self->{open_elements}}; |
| 6163 |
|
} |
| 6164 |
|
|
| 6165 |
|
## Step 2. |
| 6166 |
|
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
| 6167 |
|
!!!cp ('t417.1'); |
| 6168 |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
| 6169 |
|
} else { |
| 6170 |
|
!!!cp ('t420'); |
| 6171 |
|
} |
| 6172 |
|
|
| 6173 |
|
## Step 3. |
| 6174 |
|
splice @{$self->{open_elements}}, $i; |
| 6175 |
} |
} |
| 6176 |
|
|
|
undef $self->{form_element}; |
|
| 6177 |
!!!next-token; |
!!!next-token; |
| 6178 |
redo B; |
redo B; |
| 6179 |
} elsif ({ |
} elsif ({ |
| 6186 |
if ({ |
if ({ |
| 6187 |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, |
| 6188 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
|
## generate implied end tags |
|
|
if ({ |
|
|
dd => 1, dt => 1, li => 1, p => 1, |
|
|
td => 1, th => 1, tr => 1, |
|
|
tbody => 1, tfoot=> 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
|
!!!cp ('t422'); |
|
|
!!!back-token; |
|
|
$token = {type => END_TAG_TOKEN, |
|
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
|
} |
|
|
|
|
| 6189 |
!!!cp ('t423'); |
!!!cp ('t423'); |
| 6190 |
$i = $_; |
$i = $_; |
| 6191 |
last INSCOPE; |
last INSCOPE; |
| 6192 |
} elsif ({ |
} elsif ({ |
| 6193 |
table => 1, caption => 1, td => 1, th => 1, |
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 6194 |
button => 1, marquee => 1, object => 1, html => 1, |
button => 1, marquee => 1, object => 1, html => 1, |
| 6195 |
}->{$node->[1]}) { |
}->{$node->[1]}) { |
| 6196 |
!!!cp ('t424'); |
!!!cp ('t424'); |
| 6197 |
last INSCOPE; |
last INSCOPE; |
| 6198 |
} |
} |
| 6199 |
} # INSCOPE |
} # INSCOPE |
| 6200 |
|
|
| 6201 |
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
unless (defined $i) { # has an element in scope |
| 6202 |
!!!cp ('t425'); |
!!!cp ('t425.1'); |
| 6203 |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 6204 |
} else { |
} else { |
| 6205 |
!!!cp ('t426'); |
## Step 1. generate implied end tags |
| 6206 |
|
while ({ |
| 6207 |
|
dd => 1, dt => 1, li => 1, p => 1, |
| 6208 |
|
}->{$self->{open_elements}->[-1]->[1]}) { |
| 6209 |
|
!!!cp ('t422'); |
| 6210 |
|
pop @{$self->{open_elements}}; |
| 6211 |
|
} |
| 6212 |
|
|
| 6213 |
|
## Step 2. |
| 6214 |
|
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
| 6215 |
|
!!!cp ('t425'); |
| 6216 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 6217 |
|
} else { |
| 6218 |
|
!!!cp ('t426'); |
| 6219 |
|
} |
| 6220 |
|
|
| 6221 |
|
## Step 3. |
| 6222 |
|
splice @{$self->{open_elements}}, $i; |
| 6223 |
} |
} |
| 6224 |
|
|
| 6225 |
splice @{$self->{open_elements}}, $i if defined $i; |
!!!next-token; |
| 6226 |
|
redo B; |
| 6227 |
|
} elsif ($token->{tag_name} eq 'p') { |
| 6228 |
|
## has an element in scope |
| 6229 |
|
my $i; |
| 6230 |
|
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 6231 |
|
my $node = $self->{open_elements}->[$_]; |
| 6232 |
|
if ($node->[1] eq $token->{tag_name}) { |
| 6233 |
|
!!!cp ('t410.1'); |
| 6234 |
|
$i = $_; |
| 6235 |
|
last INSCOPE; |
| 6236 |
|
} elsif ({ |
| 6237 |
|
applet => 1, table => 1, caption => 1, td => 1, th => 1, |
| 6238 |
|
button => 1, marquee => 1, object => 1, html => 1, |
| 6239 |
|
}->{$node->[1]}) { |
| 6240 |
|
!!!cp ('t411.1'); |
| 6241 |
|
last INSCOPE; |
| 6242 |
|
} |
| 6243 |
|
} # INSCOPE |
| 6244 |
|
|
| 6245 |
|
if (defined $i) { |
| 6246 |
|
if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) { |
| 6247 |
|
!!!cp ('t412.1'); |
| 6248 |
|
!!!parse-error (type => 'not closed:'.$self->{open_elements}->[-1]->[1]); |
| 6249 |
|
} else { |
| 6250 |
|
!!!cp ('t414.1'); |
| 6251 |
|
} |
| 6252 |
|
|
| 6253 |
|
splice @{$self->{open_elements}}, $i; |
| 6254 |
|
} else { |
| 6255 |
|
!!!cp ('t413.1'); |
| 6256 |
|
!!!parse-error (type => 'unmatched end tag:'.$token->{tag_name}); |
| 6257 |
|
|
| 6258 |
|
!!!cp ('t415.1'); |
| 6259 |
|
## As if <p>, then reprocess the current token |
| 6260 |
|
my $el; |
| 6261 |
|
!!!create-element ($el, 'p'); |
| 6262 |
|
$insert->($el); |
| 6263 |
|
## NOTE: Not inserted into |$self->{open_elements}|. |
| 6264 |
|
} |
| 6265 |
|
|
| 6266 |
!!!next-token; |
!!!next-token; |
| 6267 |
redo B; |
redo B; |
| 6268 |
} elsif ({ |
} elsif ({ |
| 6318 |
if ($node->[1] eq $token->{tag_name}) { |
if ($node->[1] eq $token->{tag_name}) { |
| 6319 |
## Step 1 |
## Step 1 |
| 6320 |
## generate implied end tags |
## generate implied end tags |
| 6321 |
if ({ |
while ({ |
| 6322 |
dd => 1, dt => 1, li => 1, p => 1, |
dd => 1, dt => 1, li => 1, p => 1, |
| 6323 |
td => 1, th => 1, tr => 1, |
}->{$self->{open_elements}->[-1]->[1]}) { |
|
tbody => 1, tfoot => 1, thead => 1, |
|
|
}->{$self->{open_elements}->[-1]->[1]}) { |
|
| 6324 |
!!!cp ('t430'); |
!!!cp ('t430'); |
| 6325 |
!!!back-token; |
## ISSUE: Can this case be reached? |
| 6326 |
$token = {type => END_TAG_TOKEN, |
pop @{$self->{open_elements}}; |
|
tag_name => $self->{open_elements}->[-1]->[1]}; # MUST |
|
|
redo B; |
|
| 6327 |
} |
} |
| 6328 |
|
|
| 6329 |
## Step 2 |
## Step 2 |
| 6369 |
redo B; |
redo B; |
| 6370 |
} # B |
} # B |
| 6371 |
|
|
|
## NOTE: The "trailing end" phase in HTML5 is split into |
|
|
## two insertion modes: "after html body" and "after html frameset". |
|
|
## NOTE: States in the main stage is preserved while |
|
|
## the parser stays in the trailing end phase. # MUST |
|
|
|
|
| 6372 |
## Stop parsing # MUST |
## Stop parsing # MUST |
| 6373 |
|
|
| 6374 |
## TODO: script stuffs |
## TODO: script stuffs |
| 6410 |
my $p = $class->new; |
my $p = $class->new; |
| 6411 |
$p->{document} = $doc; |
$p->{document} = $doc; |
| 6412 |
|
|
| 6413 |
## Step 9 # MUST |
## Step 8 # MUST |
| 6414 |
my $i = 0; |
my $i = 0; |
| 6415 |
my $line = 1; |
my $line = 1; |
| 6416 |
my $column = 0; |
my $column = 0; |
| 6477 |
|
|
| 6478 |
$p->{inner_html_node} = [$node, $node_ln]; |
$p->{inner_html_node} = [$node, $node_ln]; |
| 6479 |
|
|
| 6480 |
## Step 4 |
## Step 3 |
| 6481 |
my $root = $doc->create_element_ns |
my $root = $doc->create_element_ns |
| 6482 |
('http://www.w3.org/1999/xhtml', [undef, 'html']); |
('http://www.w3.org/1999/xhtml', [undef, 'html']); |
| 6483 |
|
|
| 6484 |
## Step 5 # MUST |
## Step 4 # MUST |
| 6485 |
$doc->append_child ($root); |
$doc->append_child ($root); |
| 6486 |
|
|
| 6487 |
## Step 6 # MUST |
## Step 5 # MUST |
| 6488 |
push @{$p->{open_elements}}, [$root, 'html']; |
push @{$p->{open_elements}}, [$root, 'html']; |
| 6489 |
|
|
| 6490 |
undef $p->{head_element}; |
undef $p->{head_element}; |
| 6491 |
|
|
| 6492 |
## Step 7 # MUST |
## Step 6 # MUST |
| 6493 |
$p->_reset_insertion_mode; |
$p->_reset_insertion_mode; |
| 6494 |
|
|
| 6495 |
## Step 8 # MUST |
## Step 7 # MUST |
| 6496 |
my $anode = $node; |
my $anode = $node; |
| 6497 |
AN: while (defined $anode) { |
AN: while (defined $anode) { |
| 6498 |
if ($anode->node_type == 1) { |
if ($anode->node_type == 1) { |
| 6508 |
$anode = $anode->parent_node; |
$anode = $anode->parent_node; |
| 6509 |
} # AN |
} # AN |
| 6510 |
|
|
| 6511 |
## Step 3 # MUST |
## Step 9 # MUST |
|
## Step 10 # MUST |
|
| 6512 |
{ |
{ |
| 6513 |
my $self = $p; |
my $self = $p; |
| 6514 |
!!!next-token; |
!!!next-token; |
| 6515 |
} |
} |
| 6516 |
$p->_tree_construction_main; |
$p->_tree_construction_main; |
| 6517 |
|
|
| 6518 |
## Step 11 # MUST |
## Step 10 # MUST |
| 6519 |
my @cn = @{$node->child_nodes}; |
my @cn = @{$node->child_nodes}; |
| 6520 |
for (@cn) { |
for (@cn) { |
| 6521 |
$node->remove_child ($_); |
$node->remove_child ($_); |
| 6522 |
} |
} |
| 6523 |
## ISSUE: mutation events? read-only? |
## ISSUE: mutation events? read-only? |
| 6524 |
|
|
| 6525 |
## Step 12 # MUST |
## Step 11 # MUST |
| 6526 |
@cn = @{$root->child_nodes}; |
@cn = @{$root->child_nodes}; |
| 6527 |
for (@cn) { |
for (@cn) { |
| 6528 |
$this_doc->adopt_node ($_); |
$this_doc->adopt_node ($_); |