| 179 |
dt => DTDD_EL, |
dt => DTDD_EL, |
| 180 |
em => FORMATTING_EL, |
em => FORMATTING_EL, |
| 181 |
embed => MISC_SPECIAL_EL, |
embed => MISC_SPECIAL_EL, |
|
eventsource => MISC_SPECIAL_EL, |
|
| 182 |
fieldset => MISC_SPECIAL_EL, |
fieldset => MISC_SPECIAL_EL, |
| 183 |
figure => MISC_SPECIAL_EL, |
figure => MISC_SPECIAL_EL, |
| 184 |
font => FORMATTING_EL, |
font => FORMATTING_EL, |
| 202 |
#image => MISC_SPECIAL_EL, ## NOTE: Commented out in the spec. |
#image => MISC_SPECIAL_EL, ## NOTE: Commented out in the spec. |
| 203 |
input => MISC_SPECIAL_EL, |
input => MISC_SPECIAL_EL, |
| 204 |
isindex => MISC_SPECIAL_EL, |
isindex => MISC_SPECIAL_EL, |
| 205 |
|
## XXX keygen? (Whether a void element is in Special or not does not |
| 206 |
|
## affect to the processing, however.) |
| 207 |
li => LI_EL, |
li => LI_EL, |
| 208 |
link => MISC_SPECIAL_EL, |
link => MISC_SPECIAL_EL, |
| 209 |
listing => MISC_SPECIAL_EL, |
listing => MISC_SPECIAL_EL, |
| 248 |
u => FORMATTING_EL, |
u => FORMATTING_EL, |
| 249 |
ul => MISC_SPECIAL_EL, |
ul => MISC_SPECIAL_EL, |
| 250 |
wbr => MISC_SPECIAL_EL, |
wbr => MISC_SPECIAL_EL, |
| 251 |
|
xmp => MISC_SPECIAL_EL, |
| 252 |
}; |
}; |
| 253 |
|
|
| 254 |
my $el_category_f = { |
my $el_category_f = { |
| 649 |
|
|
| 650 |
## NOTE: |set_inner_html| copies most of this method's code |
## NOTE: |set_inner_html| copies most of this method's code |
| 651 |
|
|
| 652 |
|
## Confidence: irrelevant. |
| 653 |
$self->{confident} = 1 unless exists $self->{confident}; |
$self->{confident} = 1 unless exists $self->{confident}; |
| 654 |
|
|
| 655 |
$self->{document}->input_encoding ($self->{input_encoding}) |
$self->{document}->input_encoding ($self->{input_encoding}) |
| 656 |
if defined $self->{input_encoding}; |
if defined $self->{input_encoding}; |
| 657 |
## TODO: |{input_encoding}| is needless? |
## TODO: |{input_encoding}| is needless? |
| 915 |
|
|
| 916 |
INITIAL: { |
INITIAL: { |
| 917 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 918 |
## NOTE: Conformance checkers MAY, instead of reporting "not HTML5" |
## NOTE: Conformance checkers MAY, instead of reporting "not |
| 919 |
## error, switch to a conformance checking mode for another |
## HTML5" error, switch to a conformance checking mode for |
| 920 |
## language. |
## another language. (We don't support such mode switchings; it |
| 921 |
|
## is nonsense to do anything different from what browsers do.) |
| 922 |
my $doctype_name = $token->{name}; |
my $doctype_name = $token->{name}; |
| 923 |
$doctype_name = '' unless defined $doctype_name; |
$doctype_name = '' unless defined $doctype_name; |
| 924 |
$doctype_name =~ tr/a-z/A-Z/; # ASCII case-insensitive |
my $doctype = $self->{document}->create_document_type_definition |
| 925 |
if (not defined $token->{name} or # <!DOCTYPE> |
($doctype_name); |
| 926 |
defined $token->{sysid}) { |
|
| 927 |
|
$doctype_name =~ tr/A-Z/a-z/; # ASCII case-insensitive |
| 928 |
|
if ($doctype_name ne 'html') { |
| 929 |
!!!cp ('t1'); |
!!!cp ('t1'); |
| 930 |
!!!parse-error (type => 'not HTML5', token => $token); |
!!!parse-error (type => 'not HTML5', token => $token); |
| 931 |
} elsif ($doctype_name ne 'HTML') { |
} elsif (defined $token->{pubid}) { |
| 932 |
!!!cp ('t2'); |
!!!cp ('t2'); |
| 933 |
|
## XXX Obsolete permitted DOCTYPEs |
| 934 |
!!!parse-error (type => 'not HTML5', token => $token); |
!!!parse-error (type => 'not HTML5', token => $token); |
| 935 |
} elsif (defined $token->{pubid}) { |
} elsif (defined $token->{sysid}) { |
| 936 |
if ($token->{pubid} eq 'XSLT-compat') { |
if ($token->{sysid} eq 'about:legacy-compat') { |
| 937 |
!!!cp ('t1.2'); |
!!!cp ('t1.2'); ## <!DOCTYPE HTML SYSTEM "about:legacy-compat"> |
| 938 |
!!!parse-error (type => 'XSLT-compat', token => $token, |
!!!parse-error (type => 'XSLT-compat', token => $token, |
| 939 |
level => $self->{level}->{should}); |
level => $self->{level}->{should}); |
| 940 |
} else { |
} else { |
| 941 |
!!!parse-error (type => 'not HTML5', token => $token); |
!!!parse-error (type => 'not HTML5', token => $token); |
| 942 |
} |
} |
| 943 |
} else { |
} else { ## <!DOCTYPE HTML> |
| 944 |
!!!cp ('t3'); |
!!!cp ('t3'); |
| 945 |
# |
# |
| 946 |
} |
} |
| 947 |
|
|
|
my $doctype = $self->{document}->create_document_type_definition |
|
|
($token->{name}); ## ISSUE: If name is missing (e.g. <!DOCTYPE>)? |
|
| 948 |
## NOTE: Default value for both |public_id| and |system_id| attributes |
## NOTE: Default value for both |public_id| and |system_id| attributes |
| 949 |
## are empty strings, so that we don't set any value in missing cases. |
## are empty strings, so that we don't set any value in missing cases. |
| 950 |
$doctype->public_id ($token->{pubid}) if defined $token->{pubid}; |
$doctype->public_id ($token->{pubid}) if defined $token->{pubid}; |
| 951 |
$doctype->system_id ($token->{sysid}) if defined $token->{sysid}; |
$doctype->system_id ($token->{sysid}) if defined $token->{sysid}; |
| 952 |
|
|
| 953 |
## NOTE: Other DocumentType attributes are null or empty lists. |
## NOTE: Other DocumentType attributes are null or empty lists. |
| 954 |
## In Firefox3, |internalSubset| attribute is set to the empty |
## In Firefox3, |internalSubset| attribute is set to the empty |
| 955 |
## string, while |null| is an allowed value for the attribute |
## string, while |null| is an allowed value for the attribute |
| 956 |
## according to DOM3 Core. |
## according to DOM3 Core. |
| 957 |
$self->{document}->append_child ($doctype); |
$self->{document}->append_child ($doctype); |
| 958 |
|
|
| 959 |
if ($token->{quirks} or $doctype_name ne 'HTML') { |
if ($token->{quirks} or $doctype_name ne 'html') { |
| 960 |
!!!cp ('t4'); |
!!!cp ('t4'); |
| 961 |
$self->{document}->manakai_compat_mode ('quirks'); |
$self->{document}->manakai_compat_mode ('quirks'); |
| 962 |
} elsif (defined $token->{pubid}) { |
} elsif (defined $token->{pubid}) { |
| 1426 |
## Step 3 |
## Step 3 |
| 1427 |
## TODO: Mark as "already executed", if ... |
## TODO: Mark as "already executed", if ... |
| 1428 |
|
|
| 1429 |
## Step 4 |
## Step 4 (HTML5 revision 2702) |
| 1430 |
$insert->($script_el); |
$insert->($script_el); |
|
|
|
|
## ISSUE: $script_el is not put into the stack |
|
| 1431 |
push @{$self->{open_elements}}, [$script_el, $el_category->{script}]; |
push @{$self->{open_elements}}, [$script_el, $el_category->{script}]; |
| 1432 |
|
|
| 1433 |
## Step 5 |
## Step 5 |
| 1442 |
}; # $script_start_tag |
}; # $script_start_tag |
| 1443 |
|
|
| 1444 |
## NOTE: $open_tables->[-1]->[0] is the "current table" element node. |
## NOTE: $open_tables->[-1]->[0] is the "current table" element node. |
| 1445 |
## NOTE: $open_tables->[-1]->[1] is the "tainted" flag. |
## NOTE: $open_tables->[-1]->[1] is the "tainted" flag (OBSOLETE; unused). |
| 1446 |
## NOTE: $open_tables->[-1]->[2] is set false when non-Text node inserted. |
## NOTE: $open_tables->[-1]->[2] is set false when non-Text node inserted. |
| 1447 |
my $open_tables = [[$self->{open_elements}->[0]->[0]]]; |
my $open_tables = [[$self->{open_elements}->[0]->[0]]]; |
| 1448 |
|
|
| 1612 |
|
|
| 1613 |
## Step 8 |
## Step 8 |
| 1614 |
if ($common_ancestor_node->[1] & TABLE_ROWS_EL) { |
if ($common_ancestor_node->[1] & TABLE_ROWS_EL) { |
| 1615 |
|
## Foster parenting. |
| 1616 |
my $foster_parent_element; |
my $foster_parent_element; |
| 1617 |
my $next_sibling; |
my $next_sibling; |
| 1618 |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 1619 |
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
| 1620 |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
!!!cp ('t65.2'); |
| 1621 |
if (defined $parent and $parent->node_type == 1) { |
$foster_parent_element = $self->{open_elements}->[$_ - 1]->[0]; |
| 1622 |
!!!cp ('t65.1'); |
$next_sibling = $self->{open_elements}->[$_]->[0]; |
| 1623 |
$foster_parent_element = $parent; |
undef $next_sibling |
| 1624 |
$next_sibling = $self->{open_elements}->[$_]->[0]; |
unless $next_sibling->parent_node eq $foster_parent_element; |
| 1625 |
} else { |
last OE; |
| 1626 |
!!!cp ('t65.2'); |
} |
| 1627 |
$foster_parent_element |
} # OE |
| 1628 |
= $self->{open_elements}->[$_ - 1]->[0]; |
$foster_parent_element ||= $self->{open_elements}->[0]->[0]; |
| 1629 |
} |
|
|
last OE; |
|
|
} |
|
|
} # OE |
|
|
$foster_parent_element = $self->{open_elements}->[0]->[0] |
|
|
unless defined $foster_parent_element; |
|
| 1630 |
$foster_parent_element->insert_before ($last_node->[0], $next_sibling); |
$foster_parent_element->insert_before ($last_node->[0], $next_sibling); |
| 1631 |
$open_tables->[-1]->[1] = 1; # tainted |
$open_tables->[-1]->[1] = 1; # tainted |
| 1632 |
} else { |
} else { |
| 1682 |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
$self->{open_elements}->[-1]->[0]->append_child ($_[0]); |
| 1683 |
}; # $insert_to_current |
}; # $insert_to_current |
| 1684 |
|
|
| 1685 |
|
## Foster parenting. Note that there are three "foster parenting" |
| 1686 |
|
## code in the parser: for elements (this one), for texts, and for |
| 1687 |
|
## elements in the AAA code. |
| 1688 |
my $insert_to_foster = sub { |
my $insert_to_foster = sub { |
| 1689 |
my $child = shift; |
my $child = shift; |
| 1690 |
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
| 1693 |
my $next_sibling; |
my $next_sibling; |
| 1694 |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 1695 |
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
| 1696 |
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
!!!cp ('t71'); |
| 1697 |
if (defined $parent and $parent->node_type == 1) { |
$foster_parent_element = $self->{open_elements}->[$_ - 1]->[0]; |
| 1698 |
!!!cp ('t70'); |
$next_sibling = $self->{open_elements}->[$_]->[0]; |
| 1699 |
$foster_parent_element = $parent; |
undef $next_sibling |
| 1700 |
$next_sibling = $self->{open_elements}->[$_]->[0]; |
unless $next_sibling->parent_node eq $foster_parent_element; |
| 1701 |
} else { |
last OE; |
| 1702 |
!!!cp ('t71'); |
} |
| 1703 |
$foster_parent_element |
} # OE |
| 1704 |
= $self->{open_elements}->[$_ - 1]->[0]; |
$foster_parent_element ||= $self->{open_elements}->[0]->[0]; |
| 1705 |
} |
|
| 1706 |
last OE; |
$foster_parent_element->insert_before ($child, $next_sibling); |
|
} |
|
|
} # OE |
|
|
$foster_parent_element = $self->{open_elements}->[0]->[0] |
|
|
unless defined $foster_parent_element; |
|
|
$foster_parent_element->insert_before |
|
|
($child, $next_sibling); |
|
| 1707 |
$open_tables->[-1]->[1] = 1; # tainted |
$open_tables->[-1]->[1] = 1; # tainted |
| 1708 |
} else { |
} else { |
| 1709 |
!!!cp ('t72'); |
!!!cp ('t72'); |
| 1731 |
## document.write ("b")</script>| |
## document.write ("b")</script>| |
| 1732 |
|
|
| 1733 |
B: while (1) { |
B: while (1) { |
| 1734 |
|
|
| 1735 |
|
## The "in table text" insertion mode. |
| 1736 |
|
if ($self->{insertion_mode} & TABLE_IMS and |
| 1737 |
|
not $self->{insertion_mode} & IN_FOREIGN_CONTENT_IM and |
| 1738 |
|
not $self->{insertion_mode} & IN_CDATA_RCDATA_IM) { |
| 1739 |
|
C: { |
| 1740 |
|
my $s; |
| 1741 |
|
if ($token->{type} == CHARACTER_TOKEN) { |
| 1742 |
|
!!!cp ('t194'); |
| 1743 |
|
$self->{pending_chars} ||= []; |
| 1744 |
|
push @{$self->{pending_chars}}, $token; |
| 1745 |
|
!!!next-token; |
| 1746 |
|
next B; |
| 1747 |
|
} else { |
| 1748 |
|
if ($self->{pending_chars}) { |
| 1749 |
|
$s = join '', map { $_->{data} } @{$self->{pending_chars}}; |
| 1750 |
|
delete $self->{pending_chars}; |
| 1751 |
|
if ($s =~ /[^\x09\x0A\x0C\x0D\x20]/) { |
| 1752 |
|
!!!cp ('t195'); |
| 1753 |
|
# |
| 1754 |
|
} else { |
| 1755 |
|
!!!cp ('t195.1'); |
| 1756 |
|
#$self->{open_elements}->[-1]->[0]->manakai_append_text ($s); |
| 1757 |
|
$self->{open_elements}->[-1]->[0]->append_child |
| 1758 |
|
($self->{document}->create_text_node ($s)); |
| 1759 |
|
last C; |
| 1760 |
|
} |
| 1761 |
|
} else { |
| 1762 |
|
!!!cp ('t195.2'); |
| 1763 |
|
last C; |
| 1764 |
|
} |
| 1765 |
|
} |
| 1766 |
|
|
| 1767 |
|
## Foster parenting. |
| 1768 |
|
!!!parse-error (type => 'in table:#text', token => $token); |
| 1769 |
|
|
| 1770 |
|
## NOTE: As if in body, but insert into the foster parent element. |
| 1771 |
|
$reconstruct_active_formatting_elements->($insert_to_foster); |
| 1772 |
|
|
| 1773 |
|
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
| 1774 |
|
# MUST |
| 1775 |
|
my $foster_parent_element; |
| 1776 |
|
my $next_sibling; |
| 1777 |
|
OE: for (reverse 0..$#{$self->{open_elements}}) { |
| 1778 |
|
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
| 1779 |
|
!!!cp ('t197'); |
| 1780 |
|
$foster_parent_element = $self->{open_elements}->[$_ - 1]->[0]; |
| 1781 |
|
$next_sibling = $self->{open_elements}->[$_]->[0]; |
| 1782 |
|
undef $next_sibling |
| 1783 |
|
unless $next_sibling->parent_node eq $foster_parent_element; |
| 1784 |
|
last OE; |
| 1785 |
|
} |
| 1786 |
|
} # OE |
| 1787 |
|
$foster_parent_element ||= $self->{open_elements}->[0]->[0]; |
| 1788 |
|
|
| 1789 |
|
!!!cp ('t199'); |
| 1790 |
|
$foster_parent_element->insert_before |
| 1791 |
|
($self->{document}->create_text_node ($s), $next_sibling); |
| 1792 |
|
|
| 1793 |
|
$open_tables->[-1]->[1] = 1; # tainted |
| 1794 |
|
$open_tables->[-1]->[2] = 1; # ~node inserted |
| 1795 |
|
} else { |
| 1796 |
|
## NOTE: Fragment case or in a foster parent'ed element |
| 1797 |
|
## (e.g. |<table><span>a|). In fragment case, whether the |
| 1798 |
|
## character is appended to existing node or a new node is |
| 1799 |
|
## created is irrelevant, since the foster parent'ed nodes |
| 1800 |
|
## are discarded and fragment parsing does not invoke any |
| 1801 |
|
## script. |
| 1802 |
|
!!!cp ('t200'); |
| 1803 |
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($s); |
| 1804 |
|
} |
| 1805 |
|
} # C |
| 1806 |
|
} # TABLE_IMS |
| 1807 |
|
|
| 1808 |
if ($token->{type} == DOCTYPE_TOKEN) { |
if ($token->{type} == DOCTYPE_TOKEN) { |
| 1809 |
!!!cp ('t73'); |
!!!cp ('t73'); |
| 1810 |
!!!parse-error (type => 'in html:#DOCTYPE', token => $token); |
!!!parse-error (type => 'in html:#DOCTYPE', token => $token); |
| 1946 |
} elsif ({ |
} elsif ({ |
| 1947 |
b => 1, big => 1, blockquote => 1, body => 1, br => 1, |
b => 1, big => 1, blockquote => 1, body => 1, br => 1, |
| 1948 |
center => 1, code => 1, dd => 1, div => 1, dl => 1, dt => 1, |
center => 1, code => 1, dd => 1, div => 1, dl => 1, dt => 1, |
| 1949 |
em => 1, embed => 1, font => 1, h1 => 1, h2 => 1, h3 => 1, |
em => 1, embed => 1, h1 => 1, h2 => 1, h3 => 1, |
| 1950 |
h4 => 1, h5 => 1, h6 => 1, head => 1, hr => 1, i => 1, |
h4 => 1, h5 => 1, h6 => 1, head => 1, hr => 1, i => 1, |
| 1951 |
img => 1, li => 1, listing => 1, menu => 1, meta => 1, |
img => 1, li => 1, listing => 1, menu => 1, meta => 1, |
| 1952 |
nobr => 1, ol => 1, p => 1, pre => 1, ruby => 1, s => 1, |
nobr => 1, ol => 1, p => 1, pre => 1, ruby => 1, s => 1, |
| 1953 |
small => 1, span => 1, strong => 1, strike => 1, sub => 1, |
small => 1, span => 1, strong => 1, strike => 1, sub => 1, |
| 1954 |
sup => 1, table => 1, tt => 1, u => 1, ul => 1, var => 1, |
sup => 1, table => 1, tt => 1, u => 1, ul => 1, var => 1, |
| 1955 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}} or |
| 1956 |
|
($token->{tag_name} eq 'font' and |
| 1957 |
|
($token->{attributes}->{color} or |
| 1958 |
|
$token->{attributes}->{face} or |
| 1959 |
|
$token->{attributes}->{size}))) { |
| 1960 |
!!!cp ('t87.2'); |
!!!cp ('t87.2'); |
| 1961 |
!!!parse-error (type => 'not closed', |
!!!parse-error (type => 'not closed', |
| 1962 |
text => $self->{open_elements}->[-1]->[0] |
text => $self->{open_elements}->[-1]->[0] |
| 2032 |
} |
} |
| 2033 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 2034 |
## NOTE: "using the rules for secondary insertion mode" then "continue" |
## NOTE: "using the rules for secondary insertion mode" then "continue" |
| 2035 |
!!!cp ('t87.5'); |
if ($token->{tag_name} eq 'script') { |
| 2036 |
# |
!!!cp ('t87.41'); |
| 2037 |
|
# |
| 2038 |
|
## XXXscript: Execute script here. |
| 2039 |
|
} else { |
| 2040 |
|
!!!cp ('t87.5'); |
| 2041 |
|
# |
| 2042 |
|
} |
| 2043 |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 2044 |
!!!cp ('t87.6'); |
!!!cp ('t87.6'); |
| 2045 |
!!!parse-error (type => 'not closed', |
!!!parse-error (type => 'not closed', |
| 2223 |
!!!ack ('t103.1'); |
!!!ack ('t103.1'); |
| 2224 |
!!!next-token; |
!!!next-token; |
| 2225 |
next B; |
next B; |
| 2226 |
} elsif ($token->{tag_name} eq 'command' or |
} elsif ($token->{tag_name} eq 'command') { |
|
$token->{tag_name} eq 'eventsource') { |
|
| 2227 |
if ($self->{insertion_mode} == IN_HEAD_IM) { |
if ($self->{insertion_mode} == IN_HEAD_IM) { |
| 2228 |
## NOTE: If the insertion mode at the time of the emission |
## NOTE: If the insertion mode at the time of the emission |
| 2229 |
## of the token was "before head", $self->{insertion_mode} |
## of the token was "before head", $self->{insertion_mode} |
| 2338 |
|
|
| 2339 |
## NOTE: There is a "as if in head" code clone. |
## NOTE: There is a "as if in head" code clone. |
| 2340 |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
$parse_rcdata->(RCDATA_CONTENT_MODEL); |
| 2341 |
## ISSUE: A spec bug [Bug 6038] |
|
| 2342 |
|
## NOTE: At this point the stack of open elements contain |
| 2343 |
|
## the |head| element (index == -2) and the |script| element |
| 2344 |
|
## (index == -1). In the "after head" insertion mode the |
| 2345 |
|
## |head| element is inserted only for the purpose of |
| 2346 |
|
## providing the context for the |script| element, and |
| 2347 |
|
## therefore we can now and have to remove the element from |
| 2348 |
|
## the stack. |
| 2349 |
splice @{$self->{open_elements}}, -2, 1, () # <head> |
splice @{$self->{open_elements}}, -2, 1, () # <head> |
| 2350 |
if ($self->{insertion_mode} & IM_MASK) == AFTER_HEAD_IM; |
if ($self->{insertion_mode} & IM_MASK) == AFTER_HEAD_IM; |
| 2351 |
next B; |
next B; |
| 2580 |
## Ignore the token |
## Ignore the token |
| 2581 |
!!!next-token; |
!!!next-token; |
| 2582 |
next B; |
next B; |
| 2583 |
} elsif ($token->{tag_name} eq 'br') { |
} elsif ($token->{tag_name} eq 'br') { |
| 2584 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 2585 |
!!!cp ('t142.2'); |
!!!cp ('t142.2'); |
| 2586 |
## (before head) as if <head>, (in head) as if </head> |
## (before head) as if <head>, (in head) as if </head> |
| 2587 |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
!!!create-element ($self->{head_element}, $HTML_NS, 'head',, $token); |
| 2588 |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
$self->{open_elements}->[-1]->[0]->append_child ($self->{head_element}); |
| 2589 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 2590 |
|
|
| 2591 |
## Reprocess in the "after head" insertion mode... |
## Reprocess in the "after head" insertion mode... |
| 2592 |
} elsif ($self->{insertion_mode} == IN_HEAD_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_IM) { |
| 2593 |
!!!cp ('t143.2'); |
!!!cp ('t143.2'); |
| 2594 |
## As if </head> |
## As if </head> |
| 2595 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 2596 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 2597 |
|
|
| 2598 |
## Reprocess in the "after head" insertion mode... |
## Reprocess in the "after head" insertion mode... |
| 2599 |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
} elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) { |
| 2600 |
!!!cp ('t143.3'); |
!!!cp ('t143.3'); |
| 2601 |
## NOTE: Two parse errors for <head><noscript></br> |
## NOTE: Two parse errors for <head><noscript></br> |
| 2602 |
!!!parse-error (type => 'unmatched end tag', |
!!!parse-error (type => 'unmatched end tag', |
| 2603 |
text => 'br', token => $token); |
text => 'br', token => $token); |
| 2604 |
## As if </noscript> |
## As if </noscript> |
| 2605 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 2606 |
$self->{insertion_mode} = IN_HEAD_IM; |
$self->{insertion_mode} = IN_HEAD_IM; |
| 2607 |
|
|
| 2608 |
## Reprocess in the "in head" insertion mode... |
## Reprocess in the "in head" insertion mode... |
| 2609 |
## As if </head> |
## As if </head> |
| 2610 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 2611 |
$self->{insertion_mode} = AFTER_HEAD_IM; |
$self->{insertion_mode} = AFTER_HEAD_IM; |
| 2612 |
|
|
| 2613 |
## Reprocess in the "after head" insertion mode... |
## Reprocess in the "after head" insertion mode... |
| 2614 |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
} elsif ($self->{insertion_mode} == AFTER_HEAD_IM) { |
| 2615 |
!!!cp ('t143.4'); |
!!!cp ('t143.4'); |
| 2616 |
# |
# |
| 2617 |
} else { |
} else { |
| 2618 |
die "$0: $self->{insertion_mode}: Unknown insertion mode"; |
die "$0: $self->{insertion_mode}: Unknown insertion mode"; |
| 2619 |
} |
} |
| 2620 |
|
|
| 2621 |
## ISSUE: does not agree with IE7 - it doesn't ignore </br>. |
# |
| 2622 |
!!!parse-error (type => 'unmatched end tag', |
} else { ## Other end tags |
|
text => 'br', token => $token); |
|
|
## Ignore the token |
|
|
!!!next-token; |
|
|
next B; |
|
|
} else { |
|
| 2623 |
!!!cp ('t145'); |
!!!cp ('t145'); |
| 2624 |
!!!parse-error (type => 'unmatched end tag', |
!!!parse-error (type => 'unmatched end tag', |
| 2625 |
text => $token->{tag_name}, token => $token); |
text => $token->{tag_name}, token => $token); |
| 2663 |
!!!insert-element ('body',, $token); |
!!!insert-element ('body',, $token); |
| 2664 |
$self->{insertion_mode} = IN_BODY_IM; |
$self->{insertion_mode} = IN_BODY_IM; |
| 2665 |
## reprocess |
## reprocess |
| 2666 |
next B; |
next B; |
| 2667 |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 2668 |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
if ($self->{insertion_mode} == BEFORE_HEAD_IM) { |
| 2669 |
!!!cp ('t149.1'); |
!!!cp ('t149.1'); |
| 3091 |
$insert = $insert_to_current; |
$insert = $insert_to_current; |
| 3092 |
# |
# |
| 3093 |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
} elsif ($self->{insertion_mode} & TABLE_IMS) { |
| 3094 |
if ($token->{type} == CHARACTER_TOKEN) { |
if ($token->{type} == START_TAG_TOKEN) { |
|
if (not $open_tables->[-1]->[1] and # tainted |
|
|
$token->{data} =~ s/^([\x09\x0A\x0C\x20]+)//) { |
|
|
$self->{open_elements}->[-1]->[0]->manakai_append_text ($1); |
|
|
|
|
|
unless (length $token->{data}) { |
|
|
!!!cp ('t194'); |
|
|
!!!next-token; |
|
|
next B; |
|
|
} else { |
|
|
!!!cp ('t195'); |
|
|
} |
|
|
} |
|
|
|
|
|
!!!parse-error (type => 'in table:#text', token => $token); |
|
|
|
|
|
## NOTE: As if in body, but insert into the foster parent element. |
|
|
$reconstruct_active_formatting_elements->($insert_to_foster); |
|
|
|
|
|
if ($self->{open_elements}->[-1]->[1] & TABLE_ROWS_EL) { |
|
|
# MUST |
|
|
my $foster_parent_element; |
|
|
my $next_sibling; |
|
|
my $prev_sibling; |
|
|
OE: for (reverse 0..$#{$self->{open_elements}}) { |
|
|
if ($self->{open_elements}->[$_]->[1] == TABLE_EL) { |
|
|
my $parent = $self->{open_elements}->[$_]->[0]->parent_node; |
|
|
if (defined $parent and $parent->node_type == 1) { |
|
|
$foster_parent_element = $parent; |
|
|
!!!cp ('t196'); |
|
|
$next_sibling = $self->{open_elements}->[$_]->[0]; |
|
|
$prev_sibling = $next_sibling->previous_sibling; |
|
|
# |
|
|
} else { |
|
|
!!!cp ('t197'); |
|
|
$foster_parent_element = $self->{open_elements}->[$_ - 1]->[0]; |
|
|
$prev_sibling = $foster_parent_element->last_child; |
|
|
# |
|
|
} |
|
|
last OE; |
|
|
} |
|
|
} # OE |
|
|
$foster_parent_element = $self->{open_elements}->[0]->[0] and |
|
|
$prev_sibling = $foster_parent_element->last_child |
|
|
unless defined $foster_parent_element; |
|
|
undef $prev_sibling unless $open_tables->[-1]->[2]; # ~node inserted |
|
|
if (defined $prev_sibling and |
|
|
$prev_sibling->node_type == 3) { |
|
|
!!!cp ('t198'); |
|
|
$prev_sibling->manakai_append_text ($token->{data}); |
|
|
} else { |
|
|
!!!cp ('t199'); |
|
|
$foster_parent_element->insert_before |
|
|
($self->{document}->create_text_node ($token->{data}), |
|
|
$next_sibling); |
|
|
} |
|
|
$open_tables->[-1]->[1] = 1; # tainted |
|
|
$open_tables->[-1]->[2] = 1; # ~node inserted |
|
|
} else { |
|
|
## NOTE: Fragment case or in a foster parent'ed element |
|
|
## (e.g. |<table><span>a|). In fragment case, whether the |
|
|
## character is appended to existing node or a new node is |
|
|
## created is irrelevant, since the foster parent'ed nodes |
|
|
## are discarded and fragment parsing does not invoke any |
|
|
## script. |
|
|
!!!cp ('t200'); |
|
|
$self->{open_elements}->[-1]->[0]->manakai_append_text |
|
|
($token->{data}); |
|
|
} |
|
|
|
|
|
!!!next-token; |
|
|
next B; |
|
|
} elsif ($token->{type} == START_TAG_TOKEN) { |
|
| 3095 |
if ({ |
if ({ |
| 3096 |
tr => (($self->{insertion_mode} & IM_MASK) != IN_ROW_IM), |
tr => (($self->{insertion_mode} & IM_MASK) != IN_ROW_IM), |
| 3097 |
th => 1, td => 1, |
th => 1, td => 1, |
| 3359 |
!!!ack-later; |
!!!ack-later; |
| 3360 |
next B; |
next B; |
| 3361 |
} elsif ($token->{tag_name} eq 'style') { |
} elsif ($token->{tag_name} eq 'style') { |
| 3362 |
if (not $open_tables->[-1]->[1]) { # tainted |
!!!cp ('t227.8'); |
| 3363 |
!!!cp ('t227.8'); |
## NOTE: This is a "as if in head" code clone. |
| 3364 |
## NOTE: This is a "as if in head" code clone. |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 3365 |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
| 3366 |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
next B; |
|
next B; |
|
|
} else { |
|
|
!!!cp ('t227.7'); |
|
|
# |
|
|
} |
|
| 3367 |
} elsif ($token->{tag_name} eq 'script') { |
} elsif ($token->{tag_name} eq 'script') { |
| 3368 |
if (not $open_tables->[-1]->[1]) { # tainted |
!!!cp ('t227.6'); |
| 3369 |
!!!cp ('t227.6'); |
## NOTE: This is a "as if in head" code clone. |
| 3370 |
## NOTE: This is a "as if in head" code clone. |
$script_start_tag->(); |
| 3371 |
$script_start_tag->(); |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
| 3372 |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
next B; |
|
next B; |
|
|
} else { |
|
|
!!!cp ('t227.5'); |
|
|
# |
|
|
} |
|
| 3373 |
} elsif ($token->{tag_name} eq 'input') { |
} elsif ($token->{tag_name} eq 'input') { |
| 3374 |
if (not $open_tables->[-1]->[1]) { # tainted |
if ($token->{attributes}->{type}) { |
| 3375 |
if ($token->{attributes}->{type}) { ## TODO: case |
my $type = $token->{attributes}->{type}->{value}; |
| 3376 |
my $type = lc $token->{attributes}->{type}->{value}; |
$type =~ tr/A-Z/a-z/; ## ASCII case-insensitive. |
| 3377 |
if ($type eq 'hidden') { |
if ($type eq 'hidden') { |
| 3378 |
!!!cp ('t227.3'); |
!!!cp ('t227.3'); |
| 3379 |
!!!parse-error (type => 'in table', |
!!!parse-error (type => 'in table', |
| 3380 |
text => $token->{tag_name}, token => $token); |
text => $token->{tag_name}, token => $token); |
| 3381 |
|
|
| 3382 |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
!!!insert-element ($token->{tag_name}, $token->{attributes}, $token); |
| 3383 |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
$open_tables->[-1]->[2] = 0 if @$open_tables; # ~node inserted |
| 3384 |
|
|
| 3385 |
## TODO: form element pointer |
## TODO: form element pointer |
| 3386 |
|
|
| 3387 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 3388 |
|
|
| 3389 |
!!!next-token; |
!!!next-token; |
| 3390 |
!!!ack ('t227.2.1'); |
!!!ack ('t227.2.1'); |
| 3391 |
next B; |
next B; |
|
} else { |
|
|
!!!cp ('t227.2'); |
|
|
# |
|
|
} |
|
| 3392 |
} else { |
} else { |
| 3393 |
!!!cp ('t227.1'); |
!!!cp ('t227.1'); |
| 3394 |
# |
# |
| 3834 |
!!!next-token; |
!!!next-token; |
| 3835 |
next B; |
next B; |
| 3836 |
} elsif ({ |
} elsif ({ |
| 3837 |
select => 1, input => 1, textarea => 1, |
select => 1, input => 1, textarea => 1, keygen => 1, |
| 3838 |
}->{$token->{tag_name}} or |
}->{$token->{tag_name}} or |
| 3839 |
(($self->{insertion_mode} & IM_MASK) |
(($self->{insertion_mode} & IM_MASK) |
| 3840 |
== IN_SELECT_IN_TABLE_IM and |
== IN_SELECT_IN_TABLE_IM and |
| 3843 |
tbody => 1, tfoot => 1, thead => 1, |
tbody => 1, tfoot => 1, thead => 1, |
| 3844 |
tr => 1, td => 1, th => 1, |
tr => 1, td => 1, th => 1, |
| 3845 |
}->{$token->{tag_name}})) { |
}->{$token->{tag_name}})) { |
| 3846 |
## TODO: The type below is not good - <select> is replaced by </select> |
|
| 3847 |
!!!parse-error (type => 'not closed', text => 'select', |
## 1. Parse error. |
| 3848 |
token => $token); |
if ($token->{tag_name} eq 'select') { |
| 3849 |
## NOTE: As if the token were </select> (<select> case) or |
!!!parse-error (type => 'select in select', ## XXX: documentation |
| 3850 |
## as if there were </select> (otherwise). |
token => $token); |
| 3851 |
## have an element in table scope |
} else { |
| 3852 |
|
!!!parse-error (type => 'not closed', text => 'select', |
| 3853 |
|
token => $token); |
| 3854 |
|
} |
| 3855 |
|
|
| 3856 |
|
## 2./<select>-1. Unless "have an element in table scope" (select): |
| 3857 |
my $i; |
my $i; |
| 3858 |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
INSCOPE: for (reverse 0..$#{$self->{open_elements}}) { |
| 3859 |
my $node = $self->{open_elements}->[$_]; |
my $node = $self->{open_elements}->[$_]; |
| 3868 |
} # INSCOPE |
} # INSCOPE |
| 3869 |
unless (defined $i) { |
unless (defined $i) { |
| 3870 |
!!!cp ('t280'); |
!!!cp ('t280'); |
| 3871 |
!!!parse-error (type => 'unmatched end tag', |
if ($token->{tag_name} eq 'select') { |
| 3872 |
text => 'select', token => $token); |
## NOTE: This error would be raised when |
| 3873 |
## Ignore the token |
## |select.innerHTML = '<select>'| is executed; in this |
| 3874 |
|
## case two errors, "select in select" and "unmatched |
| 3875 |
|
## end tags" are reported to the user, the latter might |
| 3876 |
|
## be confusing but this is what the spec requires. |
| 3877 |
|
!!!parse-error (type => 'unmatched end tag', |
| 3878 |
|
text => 'select', |
| 3879 |
|
token => $token); |
| 3880 |
|
} |
| 3881 |
|
## Ignore the token. |
| 3882 |
!!!nack ('t280.1'); |
!!!nack ('t280.1'); |
| 3883 |
!!!next-token; |
!!!next-token; |
| 3884 |
next B; |
next B; |
| 3885 |
} |
} |
| 3886 |
|
|
| 3887 |
|
## 3. Otherwise, as if there were <select>: |
| 3888 |
|
|
| 3889 |
!!!cp ('t281'); |
!!!cp ('t281'); |
| 3890 |
splice @{$self->{open_elements}}, $i; |
splice @{$self->{open_elements}}, $i; |
| 3901 |
## Reprocess the token. |
## Reprocess the token. |
| 3902 |
next B; |
next B; |
| 3903 |
} |
} |
| 3904 |
|
} elsif ($token->{tag_name} eq 'script') { |
| 3905 |
|
!!!cp ('t281.3'); |
| 3906 |
|
## NOTE: This is an "as if in head" code clone |
| 3907 |
|
$script_start_tag->(); |
| 3908 |
|
next B; |
| 3909 |
} else { |
} else { |
| 3910 |
!!!cp ('t282'); |
!!!cp ('t282'); |
| 3911 |
!!!parse-error (type => 'in select', |
!!!parse-error (type => 'in select', |
| 4317 |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
$parse_rcdata->(CDATA_CONTENT_MODEL); |
| 4318 |
next B; |
next B; |
| 4319 |
} elsif ({ |
} elsif ({ |
| 4320 |
base => 1, command => 1, eventsource => 1, link => 1, |
base => 1, command => 1, link => 1, |
| 4321 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4322 |
!!!cp ('t334'); |
!!!cp ('t334'); |
| 4323 |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
## NOTE: This is an "as if in head" code clone, only "-t" differs |
| 4425 |
table => 1, |
table => 1, |
| 4426 |
hr => 1, |
hr => 1, |
| 4427 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 4428 |
|
|
| 4429 |
|
## 1. When there is an opening |form| element: |
| 4430 |
if ($token->{tag_name} eq 'form' and defined $self->{form_element}) { |
if ($token->{tag_name} eq 'form' and defined $self->{form_element}) { |
| 4431 |
!!!cp ('t350'); |
!!!cp ('t350'); |
| 4432 |
!!!parse-error (type => 'in form:form', token => $token); |
!!!parse-error (type => 'in form:form', token => $token); |
| 4436 |
next B; |
next B; |
| 4437 |
} |
} |
| 4438 |
|
|
| 4439 |
## has a p element in scope |
## 2. Close the |p| element, if any. |
| 4440 |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
if ($token->{tag_name} ne 'table' or # The Hixie Quirk |
| 4441 |
if ($_->[1] == P_EL) { |
$self->{document}->manakai_compat_mode ne 'quirks') { |
| 4442 |
!!!cp ('t344'); |
## has a p element in scope |
| 4443 |
!!!back-token; # <form> |
INSCOPE: for (reverse @{$self->{open_elements}}) { |
| 4444 |
$token = {type => END_TAG_TOKEN, tag_name => 'p', |
if ($_->[1] == P_EL) { |
| 4445 |
line => $token->{line}, column => $token->{column}}; |
!!!cp ('t344'); |
| 4446 |
next B; |
!!!back-token; # <form> |
| 4447 |
} elsif ($_->[1] & SCOPING_EL) { |
$token = {type => END_TAG_TOKEN, tag_name => 'p', |
| 4448 |
!!!cp ('t345'); |
line => $token->{line}, column => $token->{column}}; |
| 4449 |
last INSCOPE; |
next B; |
| 4450 |
|
} elsif ($_->[1] & SCOPING_EL) { |
| 4451 |
|
!!!cp ('t345'); |
| 4452 |
|
last INSCOPE; |
| 4453 |
|
} |
| 4454 |
|
} # INSCOPE |
| 4455 |
|
} |
| 4456 |
|
|
| 4457 |
|
## 3. Close the opening <hn> element, if any. |
| 4458 |
|
if ({h1 => 1, h2 => 1, h3 => 1, |
| 4459 |
|
h4 => 1, h5 => 1, h6 => 1}->{$token->{tag_name}}) { |
| 4460 |
|
if ($self->{open_elements}->[-1]->[1] == HEADING_EL) { |
| 4461 |
|
!!!parse-error (type => 'not closed', |
| 4462 |
|
text => $self->{open_elements}->[-1]->[0]->manakai_local_name, |
| 4463 |
|
token => $token); |
| 4464 |
|
pop @{$self->{open_elements}}; |
| 4465 |
} |
} |
| 4466 |
} # INSCOPE |
} |
| 4467 |
|
|
| 4468 |
|
## 4. Insertion. |
| 4469 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 4470 |
if ($token->{tag_name} eq 'pre' or $token->{tag_name} eq 'listing') { |
if ($token->{tag_name} eq 'pre' or $token->{tag_name} eq 'listing') { |
| 4471 |
!!!nack ('t346.1'); |
!!!nack ('t346.1'); |
| 4509 |
} elsif ($token->{tag_name} eq 'li') { |
} elsif ($token->{tag_name} eq 'li') { |
| 4510 |
## NOTE: As normal, but imply </li> when there's another <li> ... |
## NOTE: As normal, but imply </li> when there's another <li> ... |
| 4511 |
|
|
| 4512 |
## NOTE: Special, Scope (<li><foo><li> == <li><foo><li/></foo></li>) |
## NOTE: Special, Scope (<li><foo><li> == <li><foo><li/></foo></li>):: |
| 4513 |
## Interpreted as <li><foo/></li><li/> (non-conforming) |
## Interpreted as <li><foo/></li><li/> (non-conforming): |
| 4514 |
## blockquote (O9.27), center (O), dd (Fx3, O, S3.1.2, IE7), |
## blockquote (O9.27), center (O), dd (Fx3, O, S3.1.2, IE7), |
| 4515 |
## dt (Fx, O, S, IE), dl (O), fieldset (O, S, IE), form (Fx, O, S), |
## dt (Fx, O, S, IE), dl (O), fieldset (O, S, IE), form (Fx, O, S), |
| 4516 |
## hn (O), pre (O), applet (O, S), button (O, S), marquee (Fx, O, S), |
## hn (O), pre (O), applet (O, S), button (O, S), marquee (Fx, O, S), |
| 4517 |
## object (Fx) |
## object (Fx) |
| 4518 |
## Generate non-tree (non-conforming) |
## Generate non-tree (non-conforming): |
| 4519 |
## basefont (IE7 (where basefont is non-void)), center (IE), |
## basefont (IE7 (where basefont is non-void)), center (IE), |
| 4520 |
## form (IE), hn (IE) |
## form (IE), hn (IE) |
| 4521 |
## address, div, p (<li><foo><li> == <li><foo/></li><li/>) |
## address, div, p (<li><foo><li> == <li><foo/></li><li/>):: |
| 4522 |
## Interpreted as <li><foo><li/></foo></li> (non-conforming) |
## Interpreted as <li><foo><li/></foo></li> (non-conforming): |
| 4523 |
## div (Fx, S) |
## div (Fx, S) |
| 4524 |
|
|
| 4525 |
my $non_optional; |
my $non_optional; |
| 4866 |
next B; |
next B; |
| 4867 |
} |
} |
| 4868 |
} elsif ($token->{tag_name} eq 'textarea') { |
} elsif ($token->{tag_name} eq 'textarea') { |
| 4869 |
## Step 1 |
## 1. Insert |
| 4870 |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
!!!insert-element-t ($token->{tag_name}, $token->{attributes}, $token); |
| 4871 |
|
|
| 4872 |
## Step 2 |
## Step 2 # XXX |
| 4873 |
## TODO: $self->{form_element} if defined |
## TODO: $self->{form_element} if defined |
| 4874 |
|
|
| 4875 |
## Step 3 |
## 2. Drop U+000A LINE FEED |
| 4876 |
$self->{ignore_newline} = 1; |
$self->{ignore_newline} = 1; |
| 4877 |
|
|
| 4878 |
## Step 4 |
## 3. RCDATA |
|
## ISSUE: This step is wrong. (r2302 enbugged) |
|
|
|
|
|
## Step 5 |
|
| 4879 |
$self->{content_model} = RCDATA_CONTENT_MODEL; |
$self->{content_model} = RCDATA_CONTENT_MODEL; |
| 4880 |
delete $self->{escape}; # MUST |
delete $self->{escape}; # MUST |
| 4881 |
|
|
| 4882 |
## Step 6-7 |
## 4., 6. Insertion mode |
| 4883 |
$self->{insertion_mode} |= IN_CDATA_RCDATA_IM; |
$self->{insertion_mode} |= IN_CDATA_RCDATA_IM; |
| 4884 |
|
|
| 4885 |
|
## XXX: 5. frameset-ok flag |
| 4886 |
|
|
| 4887 |
!!!nack ('t392.1'); |
!!!nack ('t392.1'); |
| 4888 |
!!!next-token; |
!!!next-token; |
| 4889 |
next B; |
next B; |
| 5029 |
} elsif ({ |
} elsif ({ |
| 5030 |
area => 1, basefont => 1, bgsound => 1, br => 1, |
area => 1, basefont => 1, bgsound => 1, br => 1, |
| 5031 |
embed => 1, img => 1, spacer => 1, wbr => 1, |
embed => 1, img => 1, spacer => 1, wbr => 1, |
| 5032 |
|
keygen => 1, |
| 5033 |
}->{$token->{tag_name}}) { |
}->{$token->{tag_name}}) { |
| 5034 |
!!!cp ('t388.1'); |
!!!cp ('t388.1'); |
| 5035 |
pop @{$self->{open_elements}}; |
pop @{$self->{open_elements}}; |
| 5056 |
} |
} |
| 5057 |
} elsif ($token->{type} == END_TAG_TOKEN) { |
} elsif ($token->{type} == END_TAG_TOKEN) { |
| 5058 |
if ($token->{tag_name} eq 'body') { |
if ($token->{tag_name} eq 'body') { |
| 5059 |
## has a |body| element in scope |
|
| 5060 |
|
## 1. If not "have an element in scope": |
| 5061 |
|
## "has a |body| element in scope" |
| 5062 |
my $i; |
my $i; |
| 5063 |
INSCOPE: { |
INSCOPE: { |
| 5064 |
for (reverse @{$self->{open_elements}}) { |
for (reverse @{$self->{open_elements}}) { |
| 5081 |
next B; |
next B; |
| 5082 |
} # INSCOPE |
} # INSCOPE |
| 5083 |
|
|
| 5084 |
|
## 2. If unclosed elements: |
| 5085 |
for (@{$self->{open_elements}}) { |
for (@{$self->{open_elements}}) { |
| 5086 |
unless ($_->[1] & ALL_END_TAG_OPTIONAL_EL) { |
unless ($_->[1] & ALL_END_TAG_OPTIONAL_EL || |
| 5087 |
|
$_->[1] == OPTGROUP_EL || |
| 5088 |
|
$_->[1] == OPTION_EL || |
| 5089 |
|
$_->[1] == RUBY_COMPONENT_EL) { |
| 5090 |
!!!cp ('t403'); |
!!!cp ('t403'); |
| 5091 |
!!!parse-error (type => 'not closed', |
!!!parse-error (type => 'not closed', |
| 5092 |
text => $_->[0]->manakai_local_name, |
text => $_->[0]->manakai_local_name, |
| 5097 |
} |
} |
| 5098 |
} |
} |
| 5099 |
|
|
| 5100 |
|
## 3. Switch the insertion mode. |
| 5101 |
$self->{insertion_mode} = AFTER_BODY_IM; |
$self->{insertion_mode} = AFTER_BODY_IM; |
| 5102 |
!!!next-token; |
!!!next-token; |
| 5103 |
next B; |
next B; |
| 5484 |
## TODO: script stuffs |
## TODO: script stuffs |
| 5485 |
} # _tree_construct_main |
} # _tree_construct_main |
| 5486 |
|
|
| 5487 |
|
## XXX: How this method is organized is somewhat out of date, although |
| 5488 |
|
## it still does what the current spec documents. |
| 5489 |
sub set_inner_html ($$$$;$) { |
sub set_inner_html ($$$$;$) { |
| 5490 |
my $class = shift; |
my $class = shift; |
| 5491 |
my $node = shift; |
my $node = shift; # /context/ |
| 5492 |
#my $s = \$_[0]; |
#my $s = \$_[0]; |
| 5493 |
my $onerror = $_[1]; |
my $onerror = $_[1]; |
| 5494 |
my $get_wrapper = $_[2] || sub ($) { return $_[0] }; |
my $get_wrapper = $_[2] || sub ($) { return $_[0] }; |
| 5495 |
|
|
|
## ISSUE: Should {confident} be true? |
|
|
|
|
| 5496 |
my $nt = $node->node_type; |
my $nt = $node->node_type; |
| 5497 |
if ($nt == 9) { |
if ($nt == 9) { # Document (invoke the algorithm with no /context/ element) |
| 5498 |
# MUST |
# MUST |
| 5499 |
|
|
| 5500 |
## Step 1 # MUST |
## Step 1 # MUST |
| 5509 |
|
|
| 5510 |
## Step 3, 4, 5 # MUST |
## Step 3, 4, 5 # MUST |
| 5511 |
$class->parse_char_string ($_[0] => $node, $onerror, $get_wrapper); |
$class->parse_char_string ($_[0] => $node, $onerror, $get_wrapper); |
| 5512 |
} elsif ($nt == 1) { |
} elsif ($nt == 1) { # Element (invoke the algorithm with /context/ element) |
| 5513 |
## TODO: If non-html element |
## TODO: If non-html element |
| 5514 |
|
|
| 5515 |
## NOTE: Most of this code is copied from |parse_string| |
## NOTE: Most of this code is copied from |parse_string| |
| 5516 |
|
|
| 5517 |
## TODO: Support for $get_wrapper |
## TODO: Support for $get_wrapper |
| 5518 |
|
|
| 5519 |
## Step 1 # MUST |
## F1. Create an HTML document. |
| 5520 |
my $this_doc = $node->owner_document; |
my $this_doc = $node->owner_document; |
| 5521 |
my $doc = $this_doc->implementation->create_document; |
my $doc = $this_doc->implementation->create_document; |
| 5522 |
$doc->manakai_is_html (1); |
$doc->manakai_is_html (1); |
| 5523 |
|
|
| 5524 |
|
## F2. Propagate quirkness flag |
| 5525 |
|
my $node_doc = $node->owner_document; |
| 5526 |
|
$doc->manakai_compat_mode ($node_doc->manakai_compat_mode); |
| 5527 |
|
|
| 5528 |
|
## F3. Create an HTML parser |
| 5529 |
my $p = $class->new; |
my $p = $class->new; |
| 5530 |
$p->{document} = $doc; |
$p->{document} = $doc; |
| 5531 |
|
|
| 5653 |
$p->_initialize_tokenizer; |
$p->_initialize_tokenizer; |
| 5654 |
$p->_initialize_tree_constructor; |
$p->_initialize_tree_constructor; |
| 5655 |
|
|
| 5656 |
## Step 2 |
## F4. If /context/ is not undef... |
| 5657 |
|
|
| 5658 |
|
## F4.1. content model flag |
| 5659 |
my $node_ln = $node->manakai_local_name; |
my $node_ln = $node->manakai_local_name; |
| 5660 |
$p->{content_model} = { |
$p->{content_model} = { |
| 5661 |
title => RCDATA_CONTENT_MODEL, |
title => RCDATA_CONTENT_MODEL, |
| 5675 |
$p->{inner_html_node} = [$node, $el_category->{$node_ln}]; |
$p->{inner_html_node} = [$node, $el_category->{$node_ln}]; |
| 5676 |
## TODO: Foreign element OK? |
## TODO: Foreign element OK? |
| 5677 |
|
|
| 5678 |
## Step 3 |
## F4.2. Root |html| element |
| 5679 |
my $root = $doc->create_element_ns |
my $root = $doc->create_element_ns |
| 5680 |
('http://www.w3.org/1999/xhtml', [undef, 'html']); |
('http://www.w3.org/1999/xhtml', [undef, 'html']); |
| 5681 |
|
|
| 5682 |
## Step 4 # MUST |
## F4.3. |
| 5683 |
$doc->append_child ($root); |
$doc->append_child ($root); |
| 5684 |
|
|
| 5685 |
## Step 5 # MUST |
## F4.4. |
| 5686 |
push @{$p->{open_elements}}, [$root, $el_category->{html}]; |
push @{$p->{open_elements}}, [$root, $el_category->{html}]; |
| 5687 |
|
|
| 5688 |
undef $p->{head_element}; |
undef $p->{head_element}; |
| 5689 |
undef $p->{head_element_inserted}; |
undef $p->{head_element_inserted}; |
| 5690 |
|
|
| 5691 |
## Step 6 # MUST |
## F4.5. |
| 5692 |
$p->_reset_insertion_mode; |
$p->_reset_insertion_mode; |
| 5693 |
|
|
| 5694 |
## Step 7 # MUST |
## F4.6. |
| 5695 |
my $anode = $node; |
my $anode = $node; |
| 5696 |
AN: while (defined $anode) { |
AN: while (defined $anode) { |
| 5697 |
if ($anode->node_type == 1) { |
if ($anode->node_type == 1) { |
| 5706 |
} |
} |
| 5707 |
$anode = $anode->parent_node; |
$anode = $anode->parent_node; |
| 5708 |
} # AN |
} # AN |
| 5709 |
|
|
| 5710 |
## Step 9 # MUST |
## F.5. Set the input stream. |
| 5711 |
|
$p->{confident} = 1; ## Confident: irrelevant. |
| 5712 |
|
|
| 5713 |
|
## F.6. Start the parser. |
| 5714 |
{ |
{ |
| 5715 |
my $self = $p; |
my $self = $p; |
| 5716 |
!!!next-token; |
!!!next-token; |
| 5717 |
} |
} |
| 5718 |
$p->_tree_construction_main; |
$p->_tree_construction_main; |
| 5719 |
|
|
| 5720 |
## Step 10 # MUST |
## F.7. |
| 5721 |
my @cn = @{$node->child_nodes}; |
my @cn = @{$node->child_nodes}; |
| 5722 |
for (@cn) { |
for (@cn) { |
| 5723 |
$node->remove_child ($_); |
$node->remove_child ($_); |