nokogumbo 2.0.1 → 2.0.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/ext/nokogumbo/extconf.rb +1 -1
- data/ext/nokogumbo/nokogumbo.c +1 -0
- data/gumbo-parser/src/error.c +17 -8
- data/gumbo-parser/src/gumbo.h +8 -0
- data/gumbo-parser/src/parser.c +473 -480
- data/gumbo-parser/src/tokenizer.c +12 -25
- data/gumbo-parser/src/tokenizer.h +2 -13
- data/lib/nokogumbo/version.rb +1 -1
- metadata +3 -4
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: dbc23d3b7a9665b48d0516f523756407ca46733286e89edaf2b5b01b05820ffd
|
4
|
+
data.tar.gz: 9d777f65fe5170fe66fde53dd9fa0d2904e2ca2d73d25d6fd240fc383cb20804
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 602ecadc3afd998eb380a9d4fed204cdd7fd48e5ee8539e4ae5713289fa65a923d27c4b67835546c048522f1a04a165d87f257f1389c591e41e127be06e98109
|
7
|
+
data.tar.gz: 88a714a552e5cc6f11b65da15bd9816d21bfc383dd5c68e1b6e8ab82ce842adf3159ad56579428b87b7c1dba002b9e38758494865153285f2563aeae9edeba54
|
data/ext/nokogumbo/extconf.rb
CHANGED
@@ -1,3 +1,4 @@
|
|
1
|
+
require 'rubygems'
|
1
2
|
require 'fileutils'
|
2
3
|
require 'mkmf'
|
3
4
|
require 'nokogiri'
|
@@ -23,7 +24,6 @@ def download_headers
|
|
23
24
|
return nil if dep_index.nil?
|
24
25
|
requirement = NG_SPEC.dependencies[dep_index].requirement.to_s
|
25
26
|
|
26
|
-
require 'rubygems'
|
27
27
|
gem 'mini_portile2', requirement
|
28
28
|
require 'mini_portile2'
|
29
29
|
p = MiniPortile::new('libxml2', version).tap do |r|
|
data/ext/nokogumbo/nokogumbo.c
CHANGED
data/gumbo-parser/src/error.c
CHANGED
@@ -365,11 +365,14 @@ static void handle_parser_error (
|
|
365
365
|
// pointer to the beginning of the string if this is the first line.
|
366
366
|
static const char* find_prev_newline (
|
367
367
|
const char* source_text,
|
368
|
+
size_t source_length,
|
368
369
|
const char* error_location
|
369
370
|
) {
|
371
|
+
const char* source_end = source_text + source_length;
|
370
372
|
assert(error_location >= source_text);
|
373
|
+
assert(error_location <= source_end);
|
371
374
|
const char* c = error_location;
|
372
|
-
if (*c == '\n'
|
375
|
+
if (c != source_text && (error_location == source_end || *c == '\n'))
|
373
376
|
--c;
|
374
377
|
while (c != source_text && *c != '\n')
|
375
378
|
--c;
|
@@ -377,20 +380,25 @@ static const char* find_prev_newline (
|
|
377
380
|
}
|
378
381
|
|
379
382
|
// Finds the next newline in the original source buffer from a given byte
|
380
|
-
// location. Returns a character pointer to that newline, or a pointer to
|
381
|
-
//
|
383
|
+
// location. Returns a character pointer to that newline, or a pointer to
|
384
|
+
// source_text + source_length if this is the last line.
|
382
385
|
static const char* find_next_newline(
|
383
|
-
const char*
|
386
|
+
const char* source_text,
|
387
|
+
size_t source_length,
|
384
388
|
const char* error_location
|
385
389
|
) {
|
386
|
-
|
390
|
+
const char* source_end = source_text + source_length;
|
391
|
+
assert(error_location >= source_text);
|
392
|
+
assert(error_location <= source_end);
|
387
393
|
const char* c = error_location;
|
388
|
-
while (c !=
|
394
|
+
while (c != source_end && *c != '\n')
|
389
395
|
++c;
|
390
396
|
return c;
|
391
397
|
}
|
392
398
|
|
393
399
|
GumboError* gumbo_add_error(GumboParser* parser) {
|
400
|
+
parser->_output->document_error = true;
|
401
|
+
|
394
402
|
int max_errors = parser->_options->max_errors;
|
395
403
|
if (max_errors >= 0 && parser->_output->errors.length >= (unsigned int) max_errors) {
|
396
404
|
return NULL;
|
@@ -547,8 +555,9 @@ void caret_diagnostic_to_string (
|
|
547
555
|
) {
|
548
556
|
error_to_string(error, output);
|
549
557
|
|
550
|
-
const char*
|
551
|
-
const char*
|
558
|
+
const char* error_text = error->original_text.data;
|
559
|
+
const char* line_start = find_prev_newline(source_text, source_length, error_text);
|
560
|
+
const char* line_end = find_next_newline(source_text, source_length, error_text);
|
552
561
|
GumboStringPiece original_line;
|
553
562
|
original_line.data = line_start;
|
554
563
|
original_line.length = line_end - line_start;
|
data/gumbo-parser/src/gumbo.h
CHANGED
@@ -820,6 +820,14 @@ typedef struct GumboInternalOutput {
|
|
820
820
|
*/
|
821
821
|
GumboVector /* GumboError */ errors;
|
822
822
|
|
823
|
+
/**
|
824
|
+
* True if the parser encounted an error.
|
825
|
+
*
|
826
|
+
* This can be true and `errors` an empty `GumboVector` if the `max_errors`
|
827
|
+
* option was set to 0.
|
828
|
+
*/
|
829
|
+
bool document_error;
|
830
|
+
|
823
831
|
/**
|
824
832
|
* A status code indicating whether parsing finished successfully or was
|
825
833
|
* stopped mid-document due to exceptional circumstances.
|
data/gumbo-parser/src/parser.c
CHANGED
@@ -336,6 +336,7 @@ static void output_init(GumboParser* parser) {
|
|
336
336
|
GumboOutput* output = gumbo_alloc(sizeof(GumboOutput));
|
337
337
|
output->root = NULL;
|
338
338
|
output->document = new_document_node();
|
339
|
+
output->document_error = false;
|
339
340
|
output->status = GUMBO_STATUS_OK;
|
340
341
|
parser->_output = output;
|
341
342
|
gumbo_init_errors(parser);
|
@@ -608,6 +609,14 @@ static bool node_qualified_tagname_is (
|
|
608
609
|
return !gumbo_ascii_strcasecmp(element_name, name);
|
609
610
|
}
|
610
611
|
|
612
|
+
static bool node_html_tagname_is (
|
613
|
+
const GumboNode* node,
|
614
|
+
GumboTag tag,
|
615
|
+
const char *name
|
616
|
+
) {
|
617
|
+
return node_qualified_tagname_is(node, GUMBO_NAMESPACE_HTML, tag, name);
|
618
|
+
}
|
619
|
+
|
611
620
|
static bool node_tagname_is (
|
612
621
|
const GumboNode* node,
|
613
622
|
GumboTag tag,
|
@@ -633,7 +642,6 @@ static bool node_qualified_tag_is (
|
|
633
642
|
|
634
643
|
// Like node_tag_in, but for the single-tag case in the HTML namespace
|
635
644
|
static bool node_html_tag_is(const GumboNode* node, GumboTag tag) {
|
636
|
-
assert(tag != GUMBO_TAG_UNKNOWN);
|
637
645
|
return node_qualified_tag_is(node, GUMBO_NAMESPACE_HTML, tag);
|
638
646
|
}
|
639
647
|
|
@@ -1675,14 +1683,18 @@ static bool has_an_element_in_select_scope(const GumboParser* parser, GumboTag t
|
|
1675
1683
|
// https://html.spec.whatwg.org/multipage/parsing.html#generate-implied-end-tags
|
1676
1684
|
// "exception" is the "element to exclude from the process" listed in the spec.
|
1677
1685
|
// Pass GUMBO_TAG_LAST to not exclude any of them.
|
1678
|
-
static void generate_implied_end_tags(
|
1686
|
+
static void generate_implied_end_tags (
|
1687
|
+
GumboParser* parser,
|
1688
|
+
GumboTag exception,
|
1689
|
+
const char* exception_name
|
1690
|
+
) {
|
1679
1691
|
static const TagSet tags = {
|
1680
1692
|
TAG(DD), TAG(DT), TAG(LI), TAG(OPTGROUP), TAG(OPTION),
|
1681
1693
|
TAG(P), TAG(RB), TAG(RP), TAG(RT), TAG(RTC)
|
1682
1694
|
};
|
1683
1695
|
while (
|
1684
1696
|
node_tag_in_set(get_current_node(parser), &tags)
|
1685
|
-
&& !
|
1697
|
+
&& !node_html_tagname_is(get_current_node(parser), exception, exception_name)
|
1686
1698
|
) {
|
1687
1699
|
pop_current_node(parser);
|
1688
1700
|
}
|
@@ -1741,30 +1753,26 @@ static bool close_table(GumboParser* parser) {
|
|
1741
1753
|
|
1742
1754
|
// This factors out the clauses relating to "act as if an end tag token with tag
|
1743
1755
|
// name `cell_tag` had been seen".
|
1744
|
-
static
|
1756
|
+
static void close_table_cell (
|
1745
1757
|
GumboParser* parser,
|
1746
1758
|
const GumboToken* token,
|
1747
1759
|
GumboTag cell_tag
|
1748
1760
|
) {
|
1749
|
-
|
1750
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
1761
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
1751
1762
|
const GumboNode* node = get_current_node(parser);
|
1752
|
-
if (!node_html_tag_is(node, cell_tag))
|
1763
|
+
if (!node_html_tag_is(node, cell_tag))
|
1753
1764
|
parser_add_parse_error(parser, token);
|
1754
|
-
result = false;
|
1755
|
-
}
|
1756
1765
|
do {
|
1757
1766
|
node = pop_current_node(parser);
|
1758
1767
|
} while (!node_html_tag_is(node, cell_tag));
|
1759
1768
|
|
1760
1769
|
clear_active_formatting_elements(parser);
|
1761
1770
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_ROW);
|
1762
|
-
return result;
|
1763
1771
|
}
|
1764
1772
|
|
1765
1773
|
// https://html.spec.whatwg.org/multipage/parsing.html#close-the-cell
|
1766
1774
|
// This holds the logic to determine whether we should close a <td> or a <th>.
|
1767
|
-
static
|
1775
|
+
static void close_current_cell(GumboParser* parser, const GumboToken* token) {
|
1768
1776
|
GumboTag cell_tag;
|
1769
1777
|
if (has_an_element_in_table_scope(parser, GUMBO_TAG_TD)) {
|
1770
1778
|
assert(!has_an_element_in_table_scope(parser, GUMBO_TAG_TH));
|
@@ -1773,7 +1781,7 @@ static bool close_current_cell(GumboParser* parser, const GumboToken* token) {
|
|
1773
1781
|
assert(has_an_element_in_table_scope(parser, GUMBO_TAG_TH));
|
1774
1782
|
cell_tag = GUMBO_TAG_TH;
|
1775
1783
|
}
|
1776
|
-
|
1784
|
+
close_table_cell(parser, token, cell_tag);
|
1777
1785
|
}
|
1778
1786
|
|
1779
1787
|
// This factors out the "act as if an end tag of tag name 'select' had been
|
@@ -1830,14 +1838,14 @@ static bool is_special_node(const GumboNode* node) {
|
|
1830
1838
|
// specified qualified name. If the elements closed are in the set handled by
|
1831
1839
|
// generate_implied_end_tags, this is normal operation and this function returns
|
1832
1840
|
// true. Otherwise, a parse error is recorded and this function returns false.
|
1833
|
-
static
|
1841
|
+
static void implicitly_close_tags (
|
1834
1842
|
GumboParser* parser,
|
1835
1843
|
GumboToken* token,
|
1836
1844
|
GumboNamespaceEnum target_ns,
|
1837
1845
|
GumboTag target
|
1838
1846
|
) {
|
1839
|
-
|
1840
|
-
generate_implied_end_tags(parser, target);
|
1847
|
+
assert(target != GUMBO_TAG_UNKNOWN);
|
1848
|
+
generate_implied_end_tags(parser, target, NULL);
|
1841
1849
|
if (!node_qualified_tag_is(get_current_node(parser), target_ns, target)) {
|
1842
1850
|
parser_add_parse_error(parser, token);
|
1843
1851
|
while (
|
@@ -1845,35 +1853,32 @@ static bool implicitly_close_tags (
|
|
1845
1853
|
) {
|
1846
1854
|
pop_current_node(parser);
|
1847
1855
|
}
|
1848
|
-
result = false;
|
1849
1856
|
}
|
1850
1857
|
assert(node_qualified_tag_is(get_current_node(parser), target_ns, target));
|
1851
1858
|
pop_current_node(parser);
|
1852
|
-
return result;
|
1853
1859
|
}
|
1854
1860
|
|
1855
1861
|
// If the stack of open elements has a <p> tag in button scope, this acts as if
|
1856
1862
|
// a </p> tag was encountered, implicitly closing tags. Returns false if a
|
1857
1863
|
// parse error occurs. This is a convenience function because this particular
|
1858
1864
|
// clause appears several times in the spec.
|
1859
|
-
static
|
1865
|
+
static void maybe_implicitly_close_p_tag (
|
1860
1866
|
GumboParser* parser,
|
1861
1867
|
GumboToken* token
|
1862
1868
|
) {
|
1863
1869
|
if (has_an_element_in_button_scope(parser, GUMBO_TAG_P)) {
|
1864
|
-
|
1870
|
+
implicitly_close_tags (
|
1865
1871
|
parser,
|
1866
1872
|
token,
|
1867
1873
|
GUMBO_NAMESPACE_HTML,
|
1868
1874
|
GUMBO_TAG_P
|
1869
1875
|
);
|
1870
1876
|
}
|
1871
|
-
return true;
|
1872
1877
|
}
|
1873
1878
|
|
1874
1879
|
// Convenience function to encapsulate the logic for closing <li> or <dd>/<dt>
|
1875
1880
|
// tags. Pass true to is_li for handling <li> tags, false for <dd> and <dt>.
|
1876
|
-
static
|
1881
|
+
static void maybe_implicitly_close_list_tag (
|
1877
1882
|
GumboParser* parser,
|
1878
1883
|
GumboToken* token,
|
1879
1884
|
bool is_li
|
@@ -1887,21 +1892,22 @@ static bool maybe_implicitly_close_list_tag (
|
|
1887
1892
|
: node_tag_in_set(node, &dd_dt_tags)
|
1888
1893
|
;
|
1889
1894
|
if (is_list_tag) {
|
1890
|
-
|
1895
|
+
implicitly_close_tags (
|
1891
1896
|
parser,
|
1892
1897
|
token,
|
1893
1898
|
node->v.element.tag_namespace,
|
1894
1899
|
node->v.element.tag
|
1895
1900
|
);
|
1901
|
+
return;
|
1896
1902
|
}
|
1903
|
+
|
1897
1904
|
if (
|
1898
1905
|
is_special_node(node)
|
1899
1906
|
&& !node_tag_in_set(node, &(const TagSet){TAG(ADDRESS), TAG(DIV), TAG(P)})
|
1900
1907
|
) {
|
1901
|
-
return
|
1908
|
+
return;
|
1902
1909
|
}
|
1903
1910
|
}
|
1904
|
-
return true;
|
1905
1911
|
}
|
1906
1912
|
|
1907
1913
|
static void merge_attributes (
|
@@ -2020,7 +2026,7 @@ static void adjust_mathml_attributes(GumboToken* token) {
|
|
2020
2026
|
attr->name = gumbo_strdup("definitionURL");
|
2021
2027
|
}
|
2022
2028
|
|
2023
|
-
static
|
2029
|
+
static void maybe_add_doctype_error (
|
2024
2030
|
GumboParser* parser,
|
2025
2031
|
const GumboToken* token
|
2026
2032
|
) {
|
@@ -2032,9 +2038,7 @@ static bool maybe_add_doctype_error (
|
|
2032
2038
|
&& strcmp(doctype->system_identifier, "about:legacy-compat"))
|
2033
2039
|
) {
|
2034
2040
|
parser_add_parse_error(parser, token);
|
2035
|
-
return false;
|
2036
2041
|
}
|
2037
|
-
return true;
|
2038
2042
|
}
|
2039
2043
|
|
2040
2044
|
static void remove_from_parent(GumboNode* node) {
|
@@ -2059,30 +2063,103 @@ static void remove_from_parent(GumboNode* node) {
|
|
2059
2063
|
}
|
2060
2064
|
}
|
2061
2065
|
|
2066
|
+
// This is here to clean up memory when the spec says "Ignore current token."
|
2067
|
+
static void ignore_token(GumboParser* parser) {
|
2068
|
+
GumboToken* token = parser->_parser_state->_current_token;
|
2069
|
+
// Ownership of the token's internal buffers are normally transferred to the
|
2070
|
+
// element, but if no element is emitted (as happens in non-verbatim-mode
|
2071
|
+
// when a token is ignored), we need to free it here to prevent a memory
|
2072
|
+
// leak.
|
2073
|
+
gumbo_token_destroy(token);
|
2074
|
+
#ifndef NDEBUG
|
2075
|
+
if (token->type == GUMBO_TOKEN_START_TAG) {
|
2076
|
+
// Mark this sentinel so the assertion in the main loop knows it's been
|
2077
|
+
// destroyed.
|
2078
|
+
token->v.start_tag.attributes = kGumboEmptyVector;
|
2079
|
+
token->v.start_tag.name = NULL;
|
2080
|
+
}
|
2081
|
+
#endif
|
2082
|
+
}
|
2083
|
+
|
2084
|
+
// The token is usually an end tag; however, the adoption agency algorithm may
|
2085
|
+
// invoke this for an 'a' or 'nobr' start tag.
|
2086
|
+
// Returns false if there was an error.
|
2087
|
+
static void in_body_any_other_end_tag(GumboParser* parser, GumboToken* token)
|
2088
|
+
{
|
2089
|
+
GumboParserState* state = parser->_parser_state;
|
2090
|
+
GumboTag tag;
|
2091
|
+
const char* tagname;
|
2092
|
+
|
2093
|
+
if (token->type == GUMBO_TOKEN_END_TAG) {
|
2094
|
+
tag = token->v.end_tag.tag;
|
2095
|
+
tagname = token->v.end_tag.name;
|
2096
|
+
} else {
|
2097
|
+
assert(token->type == GUMBO_TOKEN_START_TAG);
|
2098
|
+
tag = token->v.start_tag.tag;
|
2099
|
+
tagname = token->v.start_tag.name;
|
2100
|
+
}
|
2101
|
+
|
2102
|
+
assert(state->_open_elements.length > 0);
|
2103
|
+
assert(node_html_tag_is(state->_open_elements.data[0], GUMBO_TAG_HTML));
|
2104
|
+
// Walk up the stack of open elements until we find one that either:
|
2105
|
+
// a) Matches the tag name we saw
|
2106
|
+
// b) Is in the "special" category.
|
2107
|
+
// If we see a), implicitly close everything up to and including it. If we
|
2108
|
+
// see b), then record a parse error, don't close anything (except the
|
2109
|
+
// implied end tags) and ignore the end tag token.
|
2110
|
+
for (int i = state->_open_elements.length; --i >= 0;) {
|
2111
|
+
const GumboNode* node = state->_open_elements.data[i];
|
2112
|
+
if (node_qualified_tagname_is(node, GUMBO_NAMESPACE_HTML, tag, tagname)) {
|
2113
|
+
generate_implied_end_tags(parser, tag, tagname);
|
2114
|
+
// <!DOCTYPE><body><sarcasm><foo></sarcasm> is an example of an error.
|
2115
|
+
// foo is the "current node" but sarcasm is node.
|
2116
|
+
// XXX: Write a test for this.
|
2117
|
+
if (node != get_current_node(parser)) {
|
2118
|
+
parser_add_parse_error(parser, token);
|
2119
|
+
}
|
2120
|
+
while (node != pop_current_node(parser))
|
2121
|
+
; // Pop everything.
|
2122
|
+
return;
|
2123
|
+
} else if (is_special_node(node)) {
|
2124
|
+
parser_add_parse_error(parser, token);
|
2125
|
+
ignore_token(parser);
|
2126
|
+
return;
|
2127
|
+
}
|
2128
|
+
}
|
2129
|
+
// <html> is in the special category, so we should never get here.
|
2130
|
+
assert(0 && "unreachable");
|
2131
|
+
}
|
2132
|
+
|
2062
2133
|
// https://html.spec.whatwg.org/multipage/parsing.html#an-introduction-to-error-handling-and-strange-cases-in-the-parser
|
2063
2134
|
// Also described in the "in body" handling for end formatting tags.
|
2064
|
-
// Returns
|
2065
|
-
|
2066
|
-
|
2067
|
-
GumboParser* parser,
|
2068
|
-
GumboToken* token,
|
2069
|
-
GumboTag subject
|
2070
|
-
) {
|
2135
|
+
// Returns false if there was an error.
|
2136
|
+
static void adoption_agency_algorithm(GumboParser* parser, GumboToken* token)
|
2137
|
+
{
|
2071
2138
|
GumboParserState* state = parser->_parser_state;
|
2072
2139
|
gumbo_debug("Entering adoption agency algorithm.\n");
|
2140
|
+
// Step 1.
|
2141
|
+
GumboTag subject;
|
2142
|
+
if (token->type == GUMBO_TOKEN_START_TAG) {
|
2143
|
+
subject = token->v.start_tag.tag;
|
2144
|
+
} else {
|
2145
|
+
assert(token->type == GUMBO_TOKEN_END_TAG);
|
2146
|
+
subject = token->v.end_tag.tag;
|
2147
|
+
}
|
2148
|
+
assert(subject != GUMBO_TAG_UNKNOWN);
|
2149
|
+
|
2073
2150
|
// Step 2.
|
2074
2151
|
GumboNode* current_node = get_current_node(parser);
|
2075
2152
|
if (
|
2076
|
-
current_node
|
2077
|
-
&& current_node->v.element.tag == subject
|
2153
|
+
node_html_tag_is(current_node, subject)
|
2078
2154
|
&& -1 == gumbo_vector_index_of (
|
2079
2155
|
&state->_active_formatting_elements,
|
2080
2156
|
current_node
|
2081
2157
|
)
|
2082
2158
|
) {
|
2083
2159
|
pop_current_node(parser);
|
2084
|
-
return
|
2160
|
+
return;
|
2085
2161
|
}
|
2162
|
+
|
2086
2163
|
// Steps 3-5 & 21:
|
2087
2164
|
for (unsigned int i = 0; i < 8; ++i) {
|
2088
2165
|
// Step 6.
|
@@ -2093,8 +2170,8 @@ static bool adoption_agency_algorithm (
|
|
2093
2170
|
if (current_node == &kActiveFormattingScopeMarker) {
|
2094
2171
|
gumbo_debug("Broke on scope marker; aborting.\n");
|
2095
2172
|
// Last scope marker; abort the algorithm and handle according to "any
|
2096
|
-
// other end tag.
|
2097
|
-
|
2173
|
+
// other end tag" (below).
|
2174
|
+
break;
|
2098
2175
|
}
|
2099
2176
|
if (node_html_tag_is(current_node, subject)) {
|
2100
2177
|
// Found it.
|
@@ -2116,7 +2193,8 @@ static bool adoption_agency_algorithm (
|
|
2116
2193
|
// "any other end tag" clause (which may potentially add a parse error,
|
2117
2194
|
// but not always).
|
2118
2195
|
gumbo_debug("No active formatting elements; aborting.\n");
|
2119
|
-
|
2196
|
+
in_body_any_other_end_tag(parser, token);
|
2197
|
+
return;
|
2120
2198
|
}
|
2121
2199
|
|
2122
2200
|
// Step 7
|
@@ -2127,20 +2205,19 @@ static bool adoption_agency_algorithm (
|
|
2127
2205
|
formatting_node,
|
2128
2206
|
&state->_active_formatting_elements
|
2129
2207
|
);
|
2130
|
-
return
|
2208
|
+
return;
|
2131
2209
|
}
|
2132
2210
|
|
2133
2211
|
// Step 8
|
2134
2212
|
if (!has_an_element_in_scope(parser, formatting_node->v.element.tag)) {
|
2135
2213
|
parser_add_parse_error(parser, token);
|
2136
2214
|
gumbo_debug("Element not in scope.\n");
|
2137
|
-
return
|
2215
|
+
return;
|
2138
2216
|
}
|
2139
2217
|
|
2140
2218
|
// Step 9
|
2141
|
-
if (formatting_node != get_current_node(parser))
|
2219
|
+
if (formatting_node != get_current_node(parser))
|
2142
2220
|
parser_add_parse_error(parser, token); // But continue onwards.
|
2143
|
-
}
|
2144
2221
|
assert(formatting_node);
|
2145
2222
|
assert(!node_html_tag_is(formatting_node, GUMBO_TAG_HTML));
|
2146
2223
|
assert(!node_html_tag_is(formatting_node, GUMBO_TAG_BODY));
|
@@ -2167,7 +2244,7 @@ static bool adoption_agency_algorithm (
|
|
2167
2244
|
formatting_node,
|
2168
2245
|
&state->_active_formatting_elements
|
2169
2246
|
);
|
2170
|
-
return
|
2247
|
+
return;
|
2171
2248
|
}
|
2172
2249
|
assert(!node_html_tag_is(furthest_block, GUMBO_TAG_HTML));
|
2173
2250
|
|
@@ -2348,25 +2425,6 @@ static bool adoption_agency_algorithm (
|
|
2348
2425
|
&state->_open_elements
|
2349
2426
|
);
|
2350
2427
|
} // Step 21.
|
2351
|
-
return true;
|
2352
|
-
}
|
2353
|
-
|
2354
|
-
// This is here to clean up memory when the spec says "Ignore current token."
|
2355
|
-
static void ignore_token(GumboParser* parser) {
|
2356
|
-
GumboToken* token = parser->_parser_state->_current_token;
|
2357
|
-
// Ownership of the token's internal buffers are normally transferred to the
|
2358
|
-
// element, but if no element is emitted (as happens in non-verbatim-mode
|
2359
|
-
// when a token is ignored), we need to free it here to prevent a memory
|
2360
|
-
// leak.
|
2361
|
-
gumbo_token_destroy(token);
|
2362
|
-
#ifndef NDEBUG
|
2363
|
-
if (token->type == GUMBO_TOKEN_START_TAG) {
|
2364
|
-
// Mark this sentinel so the assertion in the main loop knows it's been
|
2365
|
-
// destroyed.
|
2366
|
-
token->v.start_tag.attributes = kGumboEmptyVector;
|
2367
|
-
token->v.start_tag.name = NULL;
|
2368
|
-
}
|
2369
|
-
#endif
|
2370
2428
|
}
|
2371
2429
|
|
2372
2430
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-end
|
@@ -2391,15 +2449,15 @@ static void finish_parsing(GumboParser* parser) {
|
|
2391
2449
|
; // Pop them all.
|
2392
2450
|
}
|
2393
2451
|
|
2394
|
-
static
|
2452
|
+
static void handle_initial(GumboParser* parser, GumboToken* token) {
|
2395
2453
|
GumboDocument* document = &get_document_node(parser)->v.document;
|
2396
2454
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2397
2455
|
ignore_token(parser);
|
2398
|
-
return
|
2456
|
+
return;
|
2399
2457
|
}
|
2400
2458
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2401
2459
|
append_comment_node(parser, get_document_node(parser), token);
|
2402
|
-
return
|
2460
|
+
return;
|
2403
2461
|
}
|
2404
2462
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2405
2463
|
document->has_doctype = true;
|
@@ -2408,35 +2466,35 @@ static bool handle_initial(GumboParser* parser, GumboToken* token) {
|
|
2408
2466
|
document->system_identifier = token->v.doc_type.system_identifier;
|
2409
2467
|
document->doc_type_quirks_mode = compute_quirks_mode(&token->v.doc_type);
|
2410
2468
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_BEFORE_HTML);
|
2411
|
-
|
2469
|
+
maybe_add_doctype_error(parser, token);
|
2470
|
+
return;
|
2412
2471
|
}
|
2413
2472
|
parser_add_parse_error(parser, token);
|
2414
2473
|
document->doc_type_quirks_mode = GUMBO_DOCTYPE_QUIRKS;
|
2415
2474
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_BEFORE_HTML);
|
2416
2475
|
parser->_parser_state->_reprocess_current_token = true;
|
2417
|
-
return true;
|
2418
2476
|
}
|
2419
2477
|
|
2420
2478
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-before-html-insertion-mode
|
2421
|
-
static
|
2479
|
+
static void handle_before_html(GumboParser* parser, GumboToken* token) {
|
2422
2480
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2423
2481
|
parser_add_parse_error(parser, token);
|
2424
2482
|
ignore_token(parser);
|
2425
|
-
return
|
2483
|
+
return;
|
2426
2484
|
}
|
2427
2485
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2428
2486
|
append_comment_node(parser, get_document_node(parser), token);
|
2429
|
-
return
|
2487
|
+
return;
|
2430
2488
|
}
|
2431
2489
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2432
2490
|
ignore_token(parser);
|
2433
|
-
return
|
2491
|
+
return;
|
2434
2492
|
}
|
2435
2493
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2436
2494
|
GumboNode* html_node = insert_element_from_token(parser, token);
|
2437
2495
|
parser->_output->root = html_node;
|
2438
2496
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_BEFORE_HEAD);
|
2439
|
-
return
|
2497
|
+
return;
|
2440
2498
|
}
|
2441
2499
|
if (
|
2442
2500
|
token->type == GUMBO_TOKEN_END_TAG
|
@@ -2444,7 +2502,7 @@ static bool handle_before_html(GumboParser* parser, GumboToken* token) {
|
|
2444
2502
|
) {
|
2445
2503
|
parser_add_parse_error(parser, token);
|
2446
2504
|
ignore_token(parser);
|
2447
|
-
return
|
2505
|
+
return;
|
2448
2506
|
}
|
2449
2507
|
GumboNode* html_node = insert_element_of_tag_type (
|
2450
2508
|
parser,
|
@@ -2455,37 +2513,37 @@ static bool handle_before_html(GumboParser* parser, GumboToken* token) {
|
|
2455
2513
|
parser->_output->root = html_node;
|
2456
2514
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_BEFORE_HEAD);
|
2457
2515
|
parser->_parser_state->_reprocess_current_token = true;
|
2458
|
-
return true;
|
2459
2516
|
}
|
2460
2517
|
|
2461
2518
|
// Forward declarations because of mutual dependencies.
|
2462
|
-
static
|
2463
|
-
static
|
2464
|
-
static
|
2519
|
+
static void handle_token(GumboParser* parser, GumboToken* token);
|
2520
|
+
static void handle_in_body(GumboParser* parser, GumboToken* token);
|
2521
|
+
static void handle_in_template(GumboParser* parser, GumboToken* token);
|
2465
2522
|
|
2466
2523
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-before-head-insertion-mode
|
2467
|
-
static
|
2524
|
+
static void handle_before_head(GumboParser* parser, GumboToken* token) {
|
2468
2525
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2469
2526
|
ignore_token(parser);
|
2470
|
-
return
|
2527
|
+
return;
|
2471
2528
|
}
|
2472
2529
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2473
2530
|
append_comment_node(parser, get_current_node(parser), token);
|
2474
|
-
return
|
2531
|
+
return;
|
2475
2532
|
}
|
2476
2533
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2477
2534
|
parser_add_parse_error(parser, token);
|
2478
2535
|
ignore_token(parser);
|
2479
|
-
return
|
2536
|
+
return;
|
2480
2537
|
}
|
2481
2538
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2482
|
-
|
2539
|
+
handle_in_body(parser, token);
|
2540
|
+
return;
|
2483
2541
|
}
|
2484
2542
|
if (tag_is(token, kStartTag, GUMBO_TAG_HEAD)) {
|
2485
2543
|
GumboNode* node = insert_element_from_token(parser, token);
|
2486
2544
|
parser->_parser_state->_head_element = node;
|
2487
2545
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_HEAD);
|
2488
|
-
return
|
2546
|
+
return;
|
2489
2547
|
}
|
2490
2548
|
if (
|
2491
2549
|
token->type == GUMBO_TOKEN_END_TAG
|
@@ -2493,7 +2551,7 @@ static bool handle_before_head(GumboParser* parser, GumboToken* token) {
|
|
2493
2551
|
) {
|
2494
2552
|
parser_add_parse_error(parser, token);
|
2495
2553
|
ignore_token(parser);
|
2496
|
-
return
|
2554
|
+
return;
|
2497
2555
|
}
|
2498
2556
|
GumboNode* node = insert_element_of_tag_type (
|
2499
2557
|
parser,
|
@@ -2503,23 +2561,22 @@ static bool handle_before_head(GumboParser* parser, GumboToken* token) {
|
|
2503
2561
|
parser->_parser_state->_head_element = node;
|
2504
2562
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_HEAD);
|
2505
2563
|
parser->_parser_state->_reprocess_current_token = true;
|
2506
|
-
return true;
|
2507
2564
|
}
|
2508
2565
|
|
2509
2566
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inhead
|
2510
|
-
static
|
2567
|
+
static void handle_in_head(GumboParser* parser, GumboToken* token) {
|
2511
2568
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2512
2569
|
insert_text_token(parser, token);
|
2513
|
-
return
|
2570
|
+
return;
|
2514
2571
|
}
|
2515
2572
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2516
2573
|
append_comment_node(parser, get_current_node(parser), token);
|
2517
|
-
return
|
2574
|
+
return;
|
2518
2575
|
}
|
2519
2576
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2520
2577
|
parser_add_parse_error(parser, token);
|
2521
2578
|
ignore_token(parser);
|
2522
|
-
return
|
2579
|
+
return;
|
2523
2580
|
}
|
2524
2581
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2525
2582
|
return handle_in_body(parser, token);
|
@@ -2532,7 +2589,7 @@ static bool handle_in_head(GumboParser* parser, GumboToken* token) {
|
|
2532
2589
|
insert_element_from_token(parser, token);
|
2533
2590
|
pop_current_node(parser);
|
2534
2591
|
acknowledge_self_closing_tag(parser);
|
2535
|
-
return
|
2592
|
+
return;
|
2536
2593
|
}
|
2537
2594
|
if (tag_is(token, kStartTag, GUMBO_TAG_META)) {
|
2538
2595
|
insert_element_from_token(parser, token);
|
@@ -2542,33 +2599,33 @@ static bool handle_in_head(GumboParser* parser, GumboToken* token) {
|
|
2542
2599
|
// spec doesn't apply. If clients want to handle meta-tag re-encoding, they
|
2543
2600
|
// should specifically look for that string in the document and re-encode it
|
2544
2601
|
// before passing to Gumbo.
|
2545
|
-
return
|
2602
|
+
return;
|
2546
2603
|
}
|
2547
2604
|
if (tag_is(token, kStartTag, GUMBO_TAG_TITLE)) {
|
2548
2605
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RCDATA);
|
2549
|
-
return
|
2606
|
+
return;
|
2550
2607
|
}
|
2551
2608
|
if (
|
2552
2609
|
tag_in(token, kStartTag, &(const TagSet){TAG(NOFRAMES), TAG(STYLE)})
|
2553
2610
|
) {
|
2554
2611
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RAWTEXT);
|
2555
|
-
return
|
2612
|
+
return;
|
2556
2613
|
}
|
2557
2614
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOSCRIPT)) {
|
2558
2615
|
insert_element_from_token(parser, token);
|
2559
2616
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_HEAD_NOSCRIPT);
|
2560
|
-
return
|
2617
|
+
return;
|
2561
2618
|
}
|
2562
2619
|
if (tag_is(token, kStartTag, GUMBO_TAG_SCRIPT)) {
|
2563
2620
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_SCRIPT_DATA);
|
2564
|
-
return
|
2621
|
+
return;
|
2565
2622
|
}
|
2566
2623
|
if (tag_is(token, kEndTag, GUMBO_TAG_HEAD)) {
|
2567
2624
|
GumboNode* head = pop_current_node(parser);
|
2568
2625
|
UNUSED_IF_NDEBUG(head);
|
2569
2626
|
assert(node_html_tag_is(head, GUMBO_TAG_HEAD));
|
2570
2627
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_HEAD);
|
2571
|
-
return
|
2628
|
+
return;
|
2572
2629
|
}
|
2573
2630
|
if (
|
2574
2631
|
tag_in(token, kEndTag, &(const TagSet){TAG(BODY), TAG(HTML), TAG(BR)})
|
@@ -2576,7 +2633,7 @@ static bool handle_in_head(GumboParser* parser, GumboToken* token) {
|
|
2576
2633
|
pop_current_node(parser);
|
2577
2634
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_HEAD);
|
2578
2635
|
parser->_parser_state->_reprocess_current_token = true;
|
2579
|
-
return
|
2636
|
+
return;
|
2580
2637
|
}
|
2581
2638
|
if (tag_is(token, kStartTag, GUMBO_TAG_TEMPLATE)) {
|
2582
2639
|
insert_element_from_token(parser, token);
|
@@ -2584,26 +2641,23 @@ static bool handle_in_head(GumboParser* parser, GumboToken* token) {
|
|
2584
2641
|
set_frameset_not_ok(parser);
|
2585
2642
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TEMPLATE);
|
2586
2643
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TEMPLATE);
|
2587
|
-
return
|
2644
|
+
return;
|
2588
2645
|
}
|
2589
2646
|
if (tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)) {
|
2590
2647
|
if (!has_open_element(parser, GUMBO_TAG_TEMPLATE)) {
|
2591
2648
|
parser_add_parse_error(parser, token);
|
2592
2649
|
ignore_token(parser);
|
2593
|
-
return
|
2650
|
+
return;
|
2594
2651
|
}
|
2595
2652
|
generate_all_implied_end_tags_thoroughly(parser);
|
2596
|
-
|
2597
|
-
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_TEMPLATE)) {
|
2653
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_TEMPLATE))
|
2598
2654
|
parser_add_parse_error(parser, token);
|
2599
|
-
success = false;
|
2600
|
-
}
|
2601
2655
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_TEMPLATE))
|
2602
2656
|
;
|
2603
2657
|
clear_active_formatting_elements(parser);
|
2604
2658
|
pop_template_insertion_mode(parser);
|
2605
2659
|
reset_insertion_mode_appropriately(parser);
|
2606
|
-
return
|
2660
|
+
return;
|
2607
2661
|
}
|
2608
2662
|
if (
|
2609
2663
|
tag_is(token, kStartTag, GUMBO_TAG_HEAD)
|
@@ -2611,29 +2665,30 @@ static bool handle_in_head(GumboParser* parser, GumboToken* token) {
|
|
2611
2665
|
) {
|
2612
2666
|
parser_add_parse_error(parser, token);
|
2613
2667
|
ignore_token(parser);
|
2614
|
-
return
|
2668
|
+
return;
|
2615
2669
|
}
|
2616
2670
|
pop_current_node(parser);
|
2617
2671
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_HEAD);
|
2618
2672
|
parser->_parser_state->_reprocess_current_token = true;
|
2619
|
-
return
|
2673
|
+
return;
|
2620
2674
|
}
|
2621
2675
|
|
2622
2676
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inheadnoscript
|
2623
|
-
static
|
2677
|
+
static void handle_in_head_noscript(GumboParser* parser, GumboToken* token) {
|
2624
2678
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2625
2679
|
parser_add_parse_error(parser, token);
|
2626
|
-
return
|
2680
|
+
return;
|
2627
2681
|
}
|
2628
2682
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2629
|
-
|
2683
|
+
handle_in_body(parser, token);
|
2684
|
+
return;
|
2630
2685
|
}
|
2631
2686
|
if (tag_is(token, kEndTag, GUMBO_TAG_NOSCRIPT)) {
|
2632
2687
|
const GumboNode* node = pop_current_node(parser);
|
2633
2688
|
assert(node_html_tag_is(node, GUMBO_TAG_NOSCRIPT));
|
2634
2689
|
UNUSED_IF_NDEBUG(node);
|
2635
2690
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_HEAD);
|
2636
|
-
return
|
2691
|
+
return;
|
2637
2692
|
}
|
2638
2693
|
if (
|
2639
2694
|
token->type == GUMBO_TOKEN_WHITESPACE
|
@@ -2643,7 +2698,8 @@ static bool handle_in_head_noscript(GumboParser* parser, GumboToken* token) {
|
|
2643
2698
|
TAG(META), TAG(NOFRAMES), TAG(STYLE)
|
2644
2699
|
})
|
2645
2700
|
) {
|
2646
|
-
|
2701
|
+
handle_in_head(parser, token);
|
2702
|
+
return;
|
2647
2703
|
}
|
2648
2704
|
if (
|
2649
2705
|
tag_in(token, kStartTag, &(const TagSet){TAG(HEAD), TAG(NOSCRIPT)})
|
@@ -2654,7 +2710,7 @@ static bool handle_in_head_noscript(GumboParser* parser, GumboToken* token) {
|
|
2654
2710
|
) {
|
2655
2711
|
parser_add_parse_error(parser, token);
|
2656
2712
|
ignore_token(parser);
|
2657
|
-
return
|
2713
|
+
return;
|
2658
2714
|
}
|
2659
2715
|
parser_add_parse_error(parser, token);
|
2660
2716
|
const GumboNode* node = pop_current_node(parser);
|
@@ -2662,38 +2718,38 @@ static bool handle_in_head_noscript(GumboParser* parser, GumboToken* token) {
|
|
2662
2718
|
UNUSED_IF_NDEBUG(node);
|
2663
2719
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_HEAD);
|
2664
2720
|
parser->_parser_state->_reprocess_current_token = true;
|
2665
|
-
return false;
|
2666
2721
|
}
|
2667
2722
|
|
2668
2723
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-after-head-insertion-mode
|
2669
|
-
static
|
2724
|
+
static void handle_after_head(GumboParser* parser, GumboToken* token) {
|
2670
2725
|
GumboParserState* state = parser->_parser_state;
|
2671
2726
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2672
2727
|
insert_text_token(parser, token);
|
2673
|
-
return
|
2728
|
+
return;
|
2674
2729
|
}
|
2675
2730
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2676
2731
|
append_comment_node(parser, get_current_node(parser), token);
|
2677
|
-
return
|
2732
|
+
return;
|
2678
2733
|
}
|
2679
2734
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2680
2735
|
parser_add_parse_error(parser, token);
|
2681
2736
|
ignore_token(parser);
|
2682
|
-
return
|
2737
|
+
return;
|
2683
2738
|
}
|
2684
2739
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2685
|
-
|
2740
|
+
handle_in_body(parser, token);
|
2741
|
+
return;
|
2686
2742
|
}
|
2687
2743
|
if (tag_is(token, kStartTag, GUMBO_TAG_BODY)) {
|
2688
2744
|
insert_element_from_token(parser, token);
|
2689
2745
|
set_frameset_not_ok(parser);
|
2690
2746
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
2691
|
-
return
|
2747
|
+
return;
|
2692
2748
|
}
|
2693
2749
|
if (tag_is(token, kStartTag, GUMBO_TAG_FRAMESET)) {
|
2694
2750
|
insert_element_from_token(parser, token);
|
2695
2751
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_FRAMESET);
|
2696
|
-
return
|
2752
|
+
return;
|
2697
2753
|
}
|
2698
2754
|
if (
|
2699
2755
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -2709,10 +2765,11 @@ static bool handle_after_head(GumboParser* parser, GumboToken* token) {
|
|
2709
2765
|
gumbo_vector_add(state->_head_element, &state->_open_elements);
|
2710
2766
|
handle_in_head(parser, token);
|
2711
2767
|
gumbo_vector_remove(state->_head_element, &state->_open_elements);
|
2712
|
-
return
|
2768
|
+
return;
|
2713
2769
|
}
|
2714
2770
|
if (tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)) {
|
2715
|
-
|
2771
|
+
handle_in_head(parser, token);
|
2772
|
+
return;
|
2716
2773
|
}
|
2717
2774
|
if (
|
2718
2775
|
tag_is(token, kStartTag, GUMBO_TAG_HEAD)
|
@@ -2723,27 +2780,26 @@ static bool handle_after_head(GumboParser* parser, GumboToken* token) {
|
|
2723
2780
|
) {
|
2724
2781
|
parser_add_parse_error(parser, token);
|
2725
2782
|
ignore_token(parser);
|
2726
|
-
return
|
2783
|
+
return;
|
2727
2784
|
}
|
2728
2785
|
insert_element_of_tag_type(parser, GUMBO_TAG_BODY, GUMBO_INSERTION_IMPLIED);
|
2729
2786
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
2730
2787
|
state->_reprocess_current_token = true;
|
2731
|
-
return true;
|
2732
2788
|
}
|
2733
2789
|
|
2734
2790
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inbody
|
2735
|
-
static
|
2791
|
+
static void handle_in_body(GumboParser* parser, GumboToken* token) {
|
2736
2792
|
GumboParserState* state = parser->_parser_state;
|
2737
2793
|
assert(state->_open_elements.length > 0);
|
2738
2794
|
if (token->type == GUMBO_TOKEN_NULL) {
|
2739
2795
|
parser_add_parse_error(parser, token);
|
2740
2796
|
ignore_token(parser);
|
2741
|
-
return
|
2797
|
+
return;
|
2742
2798
|
}
|
2743
2799
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
2744
2800
|
reconstruct_active_formatting_elements(parser);
|
2745
2801
|
insert_text_token(parser, token);
|
2746
|
-
return
|
2802
|
+
return;
|
2747
2803
|
}
|
2748
2804
|
if (
|
2749
2805
|
token->type == GUMBO_TOKEN_CHARACTER
|
@@ -2752,27 +2808,27 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2752
2808
|
reconstruct_active_formatting_elements(parser);
|
2753
2809
|
insert_text_token(parser, token);
|
2754
2810
|
set_frameset_not_ok(parser);
|
2755
|
-
return
|
2811
|
+
return;
|
2756
2812
|
}
|
2757
2813
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
2758
2814
|
append_comment_node(parser, get_current_node(parser), token);
|
2759
|
-
return
|
2815
|
+
return;
|
2760
2816
|
}
|
2761
2817
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
2762
2818
|
parser_add_parse_error(parser, token);
|
2763
2819
|
ignore_token(parser);
|
2764
|
-
return
|
2820
|
+
return;
|
2765
2821
|
}
|
2766
2822
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
2767
2823
|
parser_add_parse_error(parser, token);
|
2768
2824
|
if (has_open_element(parser, GUMBO_TAG_TEMPLATE)) {
|
2769
2825
|
ignore_token(parser);
|
2770
|
-
return
|
2826
|
+
return;
|
2771
2827
|
}
|
2772
2828
|
assert(parser->_output->root != NULL);
|
2773
2829
|
assert(parser->_output->root->type == GUMBO_NODE_ELEMENT);
|
2774
2830
|
merge_attributes(token, parser->_output->root);
|
2775
|
-
return
|
2831
|
+
return;
|
2776
2832
|
}
|
2777
2833
|
if (
|
2778
2834
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -2782,7 +2838,8 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2782
2838
|
})
|
2783
2839
|
|| tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)
|
2784
2840
|
) {
|
2785
|
-
|
2841
|
+
handle_in_head(parser, token);
|
2842
|
+
return;
|
2786
2843
|
}
|
2787
2844
|
if (tag_is(token, kStartTag, GUMBO_TAG_BODY)) {
|
2788
2845
|
parser_add_parse_error(parser, token);
|
@@ -2796,7 +2853,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2796
2853
|
set_frameset_not_ok(parser);
|
2797
2854
|
merge_attributes(token, state->_open_elements.data[1]);
|
2798
2855
|
}
|
2799
|
-
return
|
2856
|
+
return;
|
2800
2857
|
}
|
2801
2858
|
if (tag_is(token, kStartTag, GUMBO_TAG_FRAMESET)) {
|
2802
2859
|
parser_add_parse_error(parser, token);
|
@@ -2806,7 +2863,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2806
2863
|
|| !state->_frameset_ok
|
2807
2864
|
) {
|
2808
2865
|
ignore_token(parser);
|
2809
|
-
return
|
2866
|
+
return;
|
2810
2867
|
}
|
2811
2868
|
// Save the body node for later removal.
|
2812
2869
|
GumboNode* body_node = state->_open_elements.data[1];
|
@@ -2838,50 +2895,43 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2838
2895
|
// Insert the <frameset>, and switch the insertion mode.
|
2839
2896
|
insert_element_from_token(parser, token);
|
2840
2897
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_FRAMESET);
|
2841
|
-
return
|
2898
|
+
return;
|
2842
2899
|
}
|
2843
2900
|
if (token->type == GUMBO_TOKEN_EOF) {
|
2844
2901
|
if (get_current_template_insertion_mode(parser) !=
|
2845
2902
|
GUMBO_INSERTION_MODE_INITIAL) {
|
2846
|
-
|
2903
|
+
handle_in_template(parser, token);
|
2904
|
+
return;
|
2847
2905
|
}
|
2848
|
-
if (stack_contains_nonclosable_element(parser))
|
2906
|
+
if (stack_contains_nonclosable_element(parser))
|
2849
2907
|
parser_add_parse_error(parser, token);
|
2850
|
-
|
2851
|
-
}
|
2852
|
-
return true;
|
2908
|
+
return;
|
2853
2909
|
}
|
2854
2910
|
if (tag_is(token, kEndTag, GUMBO_TAG_BODY)) {
|
2855
2911
|
if (!has_an_element_in_scope(parser, GUMBO_TAG_BODY)) {
|
2856
2912
|
parser_add_parse_error(parser, token);
|
2857
2913
|
ignore_token(parser);
|
2858
|
-
return
|
2914
|
+
return;
|
2859
2915
|
}
|
2860
|
-
|
2861
|
-
if (stack_contains_nonclosable_element(parser)) {
|
2916
|
+
if (stack_contains_nonclosable_element(parser))
|
2862
2917
|
parser_add_parse_error(parser, token);
|
2863
|
-
success = false;
|
2864
|
-
}
|
2865
2918
|
GumboNode* body = state->_open_elements.data[1];
|
2866
2919
|
assert(node_html_tag_is(body, GUMBO_TAG_BODY));
|
2867
2920
|
record_end_of_element(state->_current_token, &body->v.element);
|
2868
2921
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_BODY);
|
2869
|
-
return
|
2922
|
+
return;
|
2870
2923
|
}
|
2871
2924
|
if (tag_is(token, kEndTag, GUMBO_TAG_HTML)) {
|
2872
2925
|
if (!has_an_element_in_scope(parser, GUMBO_TAG_BODY)) {
|
2873
2926
|
parser_add_parse_error(parser, token);
|
2874
2927
|
ignore_token(parser);
|
2875
|
-
return
|
2928
|
+
return;
|
2876
2929
|
}
|
2877
|
-
|
2878
|
-
if (stack_contains_nonclosable_element(parser)) {
|
2930
|
+
if (stack_contains_nonclosable_element(parser))
|
2879
2931
|
parser_add_parse_error(parser, token);
|
2880
|
-
success = false;
|
2881
|
-
}
|
2882
2932
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_BODY);
|
2883
2933
|
parser->_parser_state->_reprocess_current_token = true;
|
2884
|
-
return
|
2934
|
+
return;
|
2885
2935
|
}
|
2886
2936
|
if (
|
2887
2937
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -2892,26 +2942,25 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2892
2942
|
TAG(SUMMARY), TAG(UL)
|
2893
2943
|
})
|
2894
2944
|
) {
|
2895
|
-
|
2945
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2896
2946
|
insert_element_from_token(parser, token);
|
2897
|
-
return
|
2947
|
+
return;
|
2898
2948
|
}
|
2899
2949
|
if (tag_in(token, kStartTag, &heading_tags)) {
|
2900
|
-
|
2950
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2901
2951
|
if (node_tag_in_set(get_current_node(parser), &heading_tags)) {
|
2902
2952
|
parser_add_parse_error(parser, token);
|
2903
2953
|
pop_current_node(parser);
|
2904
|
-
result = false;
|
2905
2954
|
}
|
2906
2955
|
insert_element_from_token(parser, token);
|
2907
|
-
return
|
2956
|
+
return;
|
2908
2957
|
}
|
2909
2958
|
if (tag_in(token, kStartTag, &(const TagSet){TAG(PRE), TAG(LISTING)})) {
|
2910
|
-
|
2959
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2911
2960
|
insert_element_from_token(parser, token);
|
2912
2961
|
state->_ignore_next_linefeed = true;
|
2913
2962
|
set_frameset_not_ok(parser);
|
2914
|
-
return
|
2963
|
+
return;
|
2915
2964
|
}
|
2916
2965
|
if (tag_is(token, kStartTag, GUMBO_TAG_FORM)) {
|
2917
2966
|
if (
|
@@ -2921,48 +2970,46 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2921
2970
|
gumbo_debug("Ignoring nested form.\n");
|
2922
2971
|
parser_add_parse_error(parser, token);
|
2923
2972
|
ignore_token(parser);
|
2924
|
-
return
|
2973
|
+
return;
|
2925
2974
|
}
|
2926
|
-
|
2975
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2927
2976
|
GumboNode* form_element = insert_element_from_token(parser, token);
|
2928
2977
|
if (!has_open_element(parser, GUMBO_TAG_TEMPLATE)) {
|
2929
2978
|
state->_form_element = form_element;
|
2930
2979
|
}
|
2931
|
-
return
|
2980
|
+
return;
|
2932
2981
|
}
|
2933
2982
|
if (tag_is(token, kStartTag, GUMBO_TAG_LI)) {
|
2934
|
-
|
2935
|
-
|
2983
|
+
maybe_implicitly_close_list_tag(parser, token, true);
|
2984
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2936
2985
|
insert_element_from_token(parser, token);
|
2937
|
-
return
|
2986
|
+
return;
|
2938
2987
|
}
|
2939
2988
|
if (tag_in(token, kStartTag, &dd_dt_tags)) {
|
2940
|
-
|
2941
|
-
|
2989
|
+
maybe_implicitly_close_list_tag(parser, token, false);
|
2990
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2942
2991
|
insert_element_from_token(parser, token);
|
2943
|
-
return
|
2992
|
+
return;
|
2944
2993
|
}
|
2945
2994
|
if (tag_is(token, kStartTag, GUMBO_TAG_PLAINTEXT)) {
|
2946
|
-
|
2995
|
+
maybe_implicitly_close_p_tag(parser, token);
|
2947
2996
|
insert_element_from_token(parser, token);
|
2948
2997
|
gumbo_tokenizer_set_state(parser, GUMBO_LEX_PLAINTEXT);
|
2949
|
-
return
|
2998
|
+
return;
|
2950
2999
|
}
|
2951
3000
|
if (tag_is(token, kStartTag, GUMBO_TAG_BUTTON)) {
|
2952
|
-
bool success = true;
|
2953
3001
|
if (has_an_element_in_scope(parser, GUMBO_TAG_BUTTON)) {
|
2954
3002
|
parser_add_parse_error(parser, token);
|
2955
|
-
success = false;
|
2956
3003
|
// We don't want to use implicitly_close_tags here because it may add an
|
2957
3004
|
// error and we've already added the only error the standard specifies.
|
2958
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3005
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
2959
3006
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_BUTTON))
|
2960
3007
|
;
|
2961
3008
|
}
|
2962
3009
|
reconstruct_active_formatting_elements(parser);
|
2963
3010
|
insert_element_from_token(parser, token);
|
2964
3011
|
set_frameset_not_ok(parser);
|
2965
|
-
return
|
3012
|
+
return;
|
2966
3013
|
}
|
2967
3014
|
if (
|
2968
3015
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -2977,7 +3024,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2977
3024
|
if (!has_an_element_in_scope(parser, tag)) {
|
2978
3025
|
parser_add_parse_error(parser, token);
|
2979
3026
|
ignore_token(parser);
|
2980
|
-
return
|
3027
|
+
return;
|
2981
3028
|
}
|
2982
3029
|
return implicitly_close_tags (
|
2983
3030
|
parser,
|
@@ -2991,19 +3038,15 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
2991
3038
|
if (!has_an_element_in_scope(parser, GUMBO_TAG_FORM)) {
|
2992
3039
|
parser_add_parse_error(parser, token);
|
2993
3040
|
ignore_token(parser);
|
2994
|
-
return
|
3041
|
+
return;
|
2995
3042
|
}
|
2996
|
-
|
2997
|
-
|
2998
|
-
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_FORM)) {
|
3043
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3044
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_FORM))
|
2999
3045
|
parser_add_parse_error(parser, token);
|
3000
|
-
success = false;
|
3001
|
-
}
|
3002
3046
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_FORM))
|
3003
3047
|
;
|
3004
|
-
return
|
3048
|
+
return;
|
3005
3049
|
} else {
|
3006
|
-
bool result = true;
|
3007
3050
|
GumboNode* node = state->_form_element;
|
3008
3051
|
assert(!node || node->type == GUMBO_NODE_ELEMENT);
|
3009
3052
|
state->_form_element = NULL;
|
@@ -3011,27 +3054,24 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3011
3054
|
gumbo_debug("Closing an unopened form.\n");
|
3012
3055
|
parser_add_parse_error(parser, token);
|
3013
3056
|
ignore_token(parser);
|
3014
|
-
return
|
3057
|
+
return;
|
3015
3058
|
}
|
3016
3059
|
// This differs from implicitly_close_tags because we remove *only* the
|
3017
3060
|
// <form> element; other nodes are left in scope.
|
3018
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3019
|
-
if (get_current_node(parser) != node)
|
3061
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3062
|
+
if (get_current_node(parser) != node)
|
3020
3063
|
parser_add_parse_error(parser, token);
|
3021
|
-
|
3022
|
-
} else {
|
3064
|
+
else
|
3023
3065
|
record_end_of_element(token, &node->v.element);
|
3024
|
-
}
|
3025
3066
|
|
3026
3067
|
GumboVector* open_elements = &state->_open_elements;
|
3027
3068
|
int index = gumbo_vector_index_of(open_elements, node);
|
3028
3069
|
assert(index >= 0);
|
3029
3070
|
gumbo_vector_remove_at(index, open_elements);
|
3030
|
-
return
|
3071
|
+
return;
|
3031
3072
|
}
|
3032
3073
|
}
|
3033
3074
|
if (tag_is(token, kEndTag, GUMBO_TAG_P)) {
|
3034
|
-
bool success = true;
|
3035
3075
|
if (!has_an_element_in_button_scope(parser, GUMBO_TAG_P)) {
|
3036
3076
|
parser_add_parse_error(parser, token);
|
3037
3077
|
// reconstruct_active_formatting_elements(parser);
|
@@ -3040,41 +3080,43 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3040
3080
|
GUMBO_TAG_P,
|
3041
3081
|
GUMBO_INSERTION_CONVERTED_FROM_END_TAG
|
3042
3082
|
);
|
3043
|
-
success = false;
|
3044
3083
|
}
|
3045
|
-
|
3084
|
+
implicitly_close_tags (
|
3046
3085
|
parser,
|
3047
3086
|
token,
|
3048
3087
|
GUMBO_NAMESPACE_HTML,
|
3049
3088
|
GUMBO_TAG_P
|
3050
|
-
)
|
3089
|
+
);
|
3090
|
+
return;
|
3051
3091
|
}
|
3052
3092
|
if (tag_is(token, kEndTag, GUMBO_TAG_LI)) {
|
3053
3093
|
if (!has_an_element_in_list_scope(parser, GUMBO_TAG_LI)) {
|
3054
3094
|
parser_add_parse_error(parser, token);
|
3055
3095
|
ignore_token(parser);
|
3056
|
-
return
|
3096
|
+
return;
|
3057
3097
|
}
|
3058
|
-
|
3098
|
+
implicitly_close_tags (
|
3059
3099
|
parser,
|
3060
3100
|
token,
|
3061
3101
|
GUMBO_NAMESPACE_HTML,
|
3062
3102
|
GUMBO_TAG_LI
|
3063
3103
|
);
|
3104
|
+
return;
|
3064
3105
|
}
|
3065
3106
|
if (tag_in(token, kEndTag, &dd_dt_tags)) {
|
3066
3107
|
GumboTag token_tag = token->v.end_tag.tag;
|
3067
3108
|
if (!has_an_element_in_scope(parser, token_tag)) {
|
3068
3109
|
parser_add_parse_error(parser, token);
|
3069
3110
|
ignore_token(parser);
|
3070
|
-
return
|
3111
|
+
return;
|
3071
3112
|
}
|
3072
|
-
|
3113
|
+
implicitly_close_tags (
|
3073
3114
|
parser,
|
3074
3115
|
token,
|
3075
3116
|
GUMBO_NAMESPACE_HTML,
|
3076
3117
|
token_tag
|
3077
3118
|
);
|
3119
|
+
return;
|
3078
3120
|
}
|
3079
3121
|
if (tag_in(token, kEndTag, &heading_tags)) {
|
3080
3122
|
if (
|
@@ -3086,12 +3128,11 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3086
3128
|
// No heading open; ignore the token entirely.
|
3087
3129
|
parser_add_parse_error(parser, token);
|
3088
3130
|
ignore_token(parser);
|
3089
|
-
return
|
3131
|
+
return;
|
3090
3132
|
}
|
3091
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3133
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3092
3134
|
const GumboNode* current_node = get_current_node(parser);
|
3093
|
-
|
3094
|
-
if (!success) {
|
3135
|
+
if (!node_html_tag_is(current_node, token->v.end_tag.tag)) {
|
3095
3136
|
// There're children of the heading currently open; close them below and
|
3096
3137
|
// record a parse error.
|
3097
3138
|
// TODO(jdtang): Add a way to distinguish this error case from the one
|
@@ -3101,17 +3142,15 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3101
3142
|
do {
|
3102
3143
|
current_node = pop_current_node(parser);
|
3103
3144
|
} while (!node_tag_in_set(current_node, &heading_tags));
|
3104
|
-
return
|
3145
|
+
return;
|
3105
3146
|
}
|
3106
3147
|
if (tag_is(token, kStartTag, GUMBO_TAG_A)) {
|
3107
|
-
bool success = true;
|
3108
3148
|
int last_a;
|
3109
3149
|
int has_matching_a = find_last_anchor_index(parser, &last_a);
|
3110
3150
|
if (has_matching_a) {
|
3111
3151
|
assert(has_matching_a == 1);
|
3112
3152
|
parser_add_parse_error(parser, token);
|
3113
|
-
|
3114
|
-
assert(handled);
|
3153
|
+
(void)adoption_agency_algorithm(parser, token);
|
3115
3154
|
// The adoption agency algorithm usually removes all instances of <a>
|
3116
3155
|
// from the list of active formatting elements, but in case it doesn't,
|
3117
3156
|
// we're supposed to do this. (The conditions where it might not are
|
@@ -3123,11 +3162,10 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3123
3162
|
);
|
3124
3163
|
gumbo_vector_remove(last_element, &state->_open_elements);
|
3125
3164
|
}
|
3126
|
-
success = false;
|
3127
3165
|
}
|
3128
3166
|
reconstruct_active_formatting_elements(parser);
|
3129
3167
|
add_formatting_element(parser, insert_element_from_token(parser, token));
|
3130
|
-
return
|
3168
|
+
return;
|
3131
3169
|
}
|
3132
3170
|
if (
|
3133
3171
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3137,21 +3175,18 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3137
3175
|
) {
|
3138
3176
|
reconstruct_active_formatting_elements(parser);
|
3139
3177
|
add_formatting_element(parser, insert_element_from_token(parser, token));
|
3140
|
-
return
|
3178
|
+
return;
|
3141
3179
|
}
|
3142
3180
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOBR)) {
|
3143
|
-
bool result = true;
|
3144
3181
|
reconstruct_active_formatting_elements(parser);
|
3145
3182
|
if (has_an_element_in_scope(parser, GUMBO_TAG_NOBR)) {
|
3146
|
-
result = false;
|
3147
3183
|
parser_add_parse_error(parser, token);
|
3148
|
-
|
3149
|
-
assert(handled);
|
3184
|
+
adoption_agency_algorithm(parser, token);
|
3150
3185
|
reconstruct_active_formatting_elements(parser);
|
3151
3186
|
}
|
3152
3187
|
insert_element_from_token(parser, token);
|
3153
3188
|
add_formatting_element(parser, get_current_node(parser));
|
3154
|
-
return
|
3189
|
+
return;
|
3155
3190
|
}
|
3156
3191
|
if (
|
3157
3192
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3160,9 +3195,8 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3160
3195
|
TAG(U)
|
3161
3196
|
})
|
3162
3197
|
) {
|
3163
|
-
|
3164
|
-
|
3165
|
-
return true;
|
3198
|
+
adoption_agency_algorithm(parser, token);
|
3199
|
+
return;
|
3166
3200
|
}
|
3167
3201
|
if (
|
3168
3202
|
tag_in(token, kStartTag, &(const TagSet){TAG(APPLET), TAG(MARQUEE), TAG(OBJECT)})
|
@@ -3171,7 +3205,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3171
3205
|
insert_element_from_token(parser, token);
|
3172
3206
|
add_formatting_element(parser, &kActiveFormattingScopeMarker);
|
3173
3207
|
set_frameset_not_ok(parser);
|
3174
|
-
return
|
3208
|
+
return;
|
3175
3209
|
}
|
3176
3210
|
if (
|
3177
3211
|
tag_in(token, kEndTag, &(const TagSet){TAG(APPLET), TAG(MARQUEE), TAG(OBJECT)})
|
@@ -3180,11 +3214,11 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3180
3214
|
if (!has_an_element_in_scope(parser, token_tag)) {
|
3181
3215
|
parser_add_parse_error(parser, token);
|
3182
3216
|
ignore_token(parser);
|
3183
|
-
return
|
3217
|
+
return;
|
3184
3218
|
}
|
3185
|
-
|
3219
|
+
implicitly_close_tags(parser, token, GUMBO_NAMESPACE_HTML, token_tag);
|
3186
3220
|
clear_active_formatting_elements(parser);
|
3187
|
-
return
|
3221
|
+
return;
|
3188
3222
|
}
|
3189
3223
|
if (tag_is(token, kStartTag, GUMBO_TAG_TABLE)) {
|
3190
3224
|
if (
|
@@ -3196,7 +3230,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3196
3230
|
insert_element_from_token(parser, token);
|
3197
3231
|
set_frameset_not_ok(parser);
|
3198
3232
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3199
|
-
return
|
3233
|
+
return;
|
3200
3234
|
}
|
3201
3235
|
if (tag_is(token, kEndTag, GUMBO_TAG_BR)) {
|
3202
3236
|
parser_add_parse_error(parser, token);
|
@@ -3209,7 +3243,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3209
3243
|
pop_current_node(parser);
|
3210
3244
|
acknowledge_self_closing_tag(parser);
|
3211
3245
|
set_frameset_not_ok(parser);
|
3212
|
-
return
|
3246
|
+
return;
|
3213
3247
|
}
|
3214
3248
|
if (
|
3215
3249
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3229,7 +3263,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3229
3263
|
pop_current_node(parser);
|
3230
3264
|
acknowledge_self_closing_tag(parser);
|
3231
3265
|
set_frameset_not_ok(parser);
|
3232
|
-
return
|
3266
|
+
return;
|
3233
3267
|
}
|
3234
3268
|
if (tag_is(token, kStartTag, GUMBO_TAG_INPUT)) {
|
3235
3269
|
reconstruct_active_formatting_elements(parser);
|
@@ -3238,7 +3272,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3238
3272
|
acknowledge_self_closing_tag(parser);
|
3239
3273
|
if (!attribute_matches(&input->v.element.attributes, "type", "hidden"))
|
3240
3274
|
set_frameset_not_ok(parser);
|
3241
|
-
return
|
3275
|
+
return;
|
3242
3276
|
}
|
3243
3277
|
if (
|
3244
3278
|
tag_in(token, kStartTag, &(const TagSet){TAG(PARAM), TAG(SOURCE), TAG(TRACK)})
|
@@ -3246,37 +3280,37 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3246
3280
|
insert_element_from_token(parser, token);
|
3247
3281
|
pop_current_node(parser);
|
3248
3282
|
acknowledge_self_closing_tag(parser);
|
3249
|
-
return
|
3283
|
+
return;
|
3250
3284
|
}
|
3251
3285
|
if (tag_is(token, kStartTag, GUMBO_TAG_HR)) {
|
3252
|
-
|
3286
|
+
maybe_implicitly_close_p_tag(parser, token);
|
3253
3287
|
insert_element_from_token(parser, token);
|
3254
3288
|
pop_current_node(parser);
|
3255
3289
|
acknowledge_self_closing_tag(parser);
|
3256
3290
|
set_frameset_not_ok(parser);
|
3257
|
-
return
|
3291
|
+
return;
|
3258
3292
|
}
|
3259
3293
|
if (tag_is(token, kStartTag, GUMBO_TAG_TEXTAREA)) {
|
3260
3294
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RCDATA);
|
3261
3295
|
parser->_parser_state->_ignore_next_linefeed = true;
|
3262
3296
|
set_frameset_not_ok(parser);
|
3263
|
-
return
|
3297
|
+
return;
|
3264
3298
|
}
|
3265
3299
|
if (tag_is(token, kStartTag, GUMBO_TAG_XMP)) {
|
3266
|
-
|
3300
|
+
maybe_implicitly_close_p_tag(parser, token);
|
3267
3301
|
reconstruct_active_formatting_elements(parser);
|
3268
3302
|
set_frameset_not_ok(parser);
|
3269
3303
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RAWTEXT);
|
3270
|
-
return
|
3304
|
+
return;
|
3271
3305
|
}
|
3272
3306
|
if (tag_is(token, kStartTag, GUMBO_TAG_IFRAME)) {
|
3273
3307
|
set_frameset_not_ok(parser);
|
3274
3308
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RAWTEXT);
|
3275
|
-
return
|
3309
|
+
return;
|
3276
3310
|
}
|
3277
3311
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOEMBED)) {
|
3278
3312
|
run_generic_parsing_algorithm(parser, token, GUMBO_LEX_RAWTEXT);
|
3279
|
-
return
|
3313
|
+
return;
|
3280
3314
|
}
|
3281
3315
|
if (tag_is(token, kStartTag, GUMBO_TAG_SELECT)) {
|
3282
3316
|
reconstruct_active_formatting_elements(parser);
|
@@ -3294,7 +3328,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3294
3328
|
} else {
|
3295
3329
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_SELECT);
|
3296
3330
|
}
|
3297
|
-
return
|
3331
|
+
return;
|
3298
3332
|
}
|
3299
3333
|
if (
|
3300
3334
|
tag_in(token, kStartTag, &(const TagSet){TAG(OPTGROUP), TAG(OPTION)})
|
@@ -3304,33 +3338,28 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3304
3338
|
}
|
3305
3339
|
reconstruct_active_formatting_elements(parser);
|
3306
3340
|
insert_element_from_token(parser, token);
|
3307
|
-
return
|
3341
|
+
return;
|
3308
3342
|
}
|
3309
3343
|
if (tag_in(token, kStartTag, &(const TagSet){TAG(RB), TAG(RTC)})) {
|
3310
|
-
bool success = true;
|
3311
3344
|
if (has_an_element_in_scope(parser, GUMBO_TAG_RUBY)) {
|
3312
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3313
|
-
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_RUBY))
|
3345
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3346
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_RUBY))
|
3314
3347
|
parser_add_parse_error(parser, token);
|
3315
|
-
success = false;
|
3316
|
-
}
|
3317
3348
|
}
|
3318
3349
|
insert_element_from_token(parser, token);
|
3319
|
-
return
|
3350
|
+
return;
|
3320
3351
|
}
|
3321
3352
|
if (tag_in(token, kStartTag, &(const TagSet){TAG(RP), TAG(RT)})) {
|
3322
|
-
bool success = true;
|
3323
3353
|
if (has_an_element_in_scope(parser, GUMBO_TAG_RUBY)) {
|
3324
|
-
generate_implied_end_tags(parser, GUMBO_TAG_RTC);
|
3354
|
+
generate_implied_end_tags(parser, GUMBO_TAG_RTC, NULL);
|
3325
3355
|
GumboNode* current = get_current_node(parser);
|
3326
3356
|
if (!node_html_tag_is(current, GUMBO_TAG_RUBY) &&
|
3327
3357
|
!node_html_tag_is(current, GUMBO_TAG_RTC)) {
|
3328
3358
|
parser_add_parse_error(parser, token);
|
3329
|
-
success = false;
|
3330
3359
|
}
|
3331
3360
|
}
|
3332
3361
|
insert_element_from_token(parser, token);
|
3333
|
-
return
|
3362
|
+
return;
|
3334
3363
|
}
|
3335
3364
|
if (tag_is(token, kStartTag, GUMBO_TAG_MATH)) {
|
3336
3365
|
reconstruct_active_formatting_elements(parser);
|
@@ -3341,7 +3370,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3341
3370
|
pop_current_node(parser);
|
3342
3371
|
acknowledge_self_closing_tag(parser);
|
3343
3372
|
}
|
3344
|
-
return
|
3373
|
+
return;
|
3345
3374
|
}
|
3346
3375
|
if (tag_is(token, kStartTag, GUMBO_TAG_SVG)) {
|
3347
3376
|
reconstruct_active_formatting_elements(parser);
|
@@ -3352,7 +3381,7 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3352
3381
|
pop_current_node(parser);
|
3353
3382
|
acknowledge_self_closing_tag(parser);
|
3354
3383
|
}
|
3355
|
-
return
|
3384
|
+
return;
|
3356
3385
|
}
|
3357
3386
|
if (
|
3358
3387
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3362,60 +3391,24 @@ static bool handle_in_body(GumboParser* parser, GumboToken* token) {
|
|
3362
3391
|
) {
|
3363
3392
|
parser_add_parse_error(parser, token);
|
3364
3393
|
ignore_token(parser);
|
3365
|
-
return
|
3394
|
+
return;
|
3366
3395
|
}
|
3367
3396
|
if (token->type == GUMBO_TOKEN_START_TAG) {
|
3368
3397
|
reconstruct_active_formatting_elements(parser);
|
3369
3398
|
insert_element_from_token(parser, token);
|
3370
|
-
return
|
3371
|
-
}
|
3372
|
-
any_other_end_tag:
|
3373
|
-
assert(token->type == GUMBO_TOKEN_END_TAG);
|
3374
|
-
GumboTag end_tag = token->v.end_tag.tag;
|
3375
|
-
const char *end_tagname = token->v.end_tag.name;
|
3376
|
-
assert(state->_open_elements.length > 0);
|
3377
|
-
assert(node_html_tag_is(state->_open_elements.data[0], GUMBO_TAG_HTML));
|
3378
|
-
// Walk up the stack of open elements until we find one that either:
|
3379
|
-
// a) Matches the tag name we saw
|
3380
|
-
// b) Is in the "special" category.
|
3381
|
-
// If we see a), implicitly close everything up to and including it. If we
|
3382
|
-
// see b), then record a parse error, don't close anything (except the
|
3383
|
-
// implied end tags) and ignore the end tag token.
|
3384
|
-
for (int i = state->_open_elements.length; --i >= 0;) {
|
3385
|
-
const GumboNode* node = state->_open_elements.data[i];
|
3386
|
-
if (node_qualified_tagname_is(node, GUMBO_NAMESPACE_HTML, end_tag, end_tagname)) {
|
3387
|
-
generate_implied_end_tags(parser, end_tag);
|
3388
|
-
// TODO(jdtang): Do I need to add a parse error here? The condition in
|
3389
|
-
// the spec seems like it's the inverse of the loop condition above, and
|
3390
|
-
// so would never fire.
|
3391
|
-
// sfc: Yes, an error is needed here.
|
3392
|
-
// <!DOCTYPE><body><sarcasm><foo></sarcasm> is an example.
|
3393
|
-
// foo is the "current node" but sarcasm is node.
|
3394
|
-
// XXX: Write a test for this.
|
3395
|
-
if (node != get_current_node(parser))
|
3396
|
-
parser_add_parse_error(parser, token);
|
3397
|
-
while (node != pop_current_node(parser))
|
3398
|
-
; // Pop everything.
|
3399
|
-
return true;
|
3400
|
-
} else if (is_special_node(node)) {
|
3401
|
-
parser_add_parse_error(parser, token);
|
3402
|
-
ignore_token(parser);
|
3403
|
-
return false;
|
3404
|
-
}
|
3399
|
+
return;
|
3405
3400
|
}
|
3406
|
-
|
3407
|
-
assert(0);
|
3408
|
-
return false;
|
3401
|
+
in_body_any_other_end_tag(parser, token);
|
3409
3402
|
}
|
3410
3403
|
|
3411
3404
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-incdata
|
3412
|
-
static
|
3405
|
+
static void handle_text(GumboParser* parser, GumboToken* token) {
|
3413
3406
|
if (
|
3414
3407
|
token->type == GUMBO_TOKEN_CHARACTER
|
3415
3408
|
|| token->type == GUMBO_TOKEN_WHITESPACE
|
3416
3409
|
) {
|
3417
3410
|
insert_text_token(parser, token);
|
3418
|
-
return
|
3411
|
+
return;
|
3419
3412
|
}
|
3420
3413
|
// We provide only bare-bones script handling that doesn't involve any of
|
3421
3414
|
// the parser-pause/already-started/script-nesting flags or re-entrant
|
@@ -3424,19 +3417,16 @@ static bool handle_text(GumboParser* parser, GumboToken* token) {
|
|
3424
3417
|
// provide the script body as a text-node child of the <script> element.
|
3425
3418
|
// This behavior doesn't support document.write of partial HTML elements,
|
3426
3419
|
// but should be adequate for almost all other scripting support.
|
3427
|
-
bool success = true;
|
3428
3420
|
if (token->type == GUMBO_TOKEN_EOF) {
|
3429
3421
|
parser_add_parse_error(parser, token);
|
3430
|
-
success = false;
|
3431
3422
|
parser->_parser_state->_reprocess_current_token = true;
|
3432
3423
|
}
|
3433
3424
|
pop_current_node(parser);
|
3434
3425
|
set_insertion_mode(parser, parser->_parser_state->_original_insertion_mode);
|
3435
|
-
return success;
|
3436
3426
|
}
|
3437
3427
|
|
3438
3428
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intable
|
3439
|
-
static
|
3429
|
+
static void handle_in_table(GumboParser* parser, GumboToken* token) {
|
3440
3430
|
GumboParserState* state = parser->_parser_state;
|
3441
3431
|
if (
|
3442
3432
|
(token->type == GUMBO_TOKEN_CHARACTER
|
@@ -3456,29 +3446,29 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3456
3446
|
state->_original_insertion_mode = state->_insertion_mode;
|
3457
3447
|
state->_reprocess_current_token = true;
|
3458
3448
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_TEXT);
|
3459
|
-
return
|
3449
|
+
return;
|
3460
3450
|
}
|
3461
3451
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
3462
3452
|
append_comment_node(parser, get_current_node(parser), token);
|
3463
|
-
return
|
3453
|
+
return;
|
3464
3454
|
}
|
3465
3455
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
3466
3456
|
parser_add_parse_error(parser, token);
|
3467
3457
|
ignore_token(parser);
|
3468
|
-
return
|
3458
|
+
return;
|
3469
3459
|
}
|
3470
3460
|
if (tag_is(token, kStartTag, GUMBO_TAG_CAPTION)) {
|
3471
3461
|
clear_stack_to_table_context(parser);
|
3472
3462
|
add_formatting_element(parser, &kActiveFormattingScopeMarker);
|
3473
3463
|
insert_element_from_token(parser, token);
|
3474
3464
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_CAPTION);
|
3475
|
-
return
|
3465
|
+
return;
|
3476
3466
|
}
|
3477
3467
|
if (tag_is(token, kStartTag, GUMBO_TAG_COLGROUP)) {
|
3478
3468
|
clear_stack_to_table_context(parser);
|
3479
3469
|
insert_element_from_token(parser, token);
|
3480
3470
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_COLUMN_GROUP);
|
3481
|
-
return
|
3471
|
+
return;
|
3482
3472
|
}
|
3483
3473
|
if (tag_is(token, kStartTag, GUMBO_TAG_COL)) {
|
3484
3474
|
clear_stack_to_table_context(parser);
|
@@ -3489,7 +3479,7 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3489
3479
|
);
|
3490
3480
|
state->_reprocess_current_token = true;
|
3491
3481
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_COLUMN_GROUP);
|
3492
|
-
return
|
3482
|
+
return;
|
3493
3483
|
}
|
3494
3484
|
if (
|
3495
3485
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3499,7 +3489,7 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3499
3489
|
clear_stack_to_table_context(parser);
|
3500
3490
|
insert_element_from_token(parser, token);
|
3501
3491
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
3502
|
-
return
|
3492
|
+
return;
|
3503
3493
|
}
|
3504
3494
|
if (
|
3505
3495
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3514,7 +3504,7 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3514
3504
|
);
|
3515
3505
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
3516
3506
|
state->_reprocess_current_token = true;
|
3517
|
-
return
|
3507
|
+
return;
|
3518
3508
|
}
|
3519
3509
|
if (tag_is(token, kStartTag, GUMBO_TAG_TABLE)) {
|
3520
3510
|
parser_add_parse_error(parser, token);
|
@@ -3523,14 +3513,14 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3523
3513
|
} else {
|
3524
3514
|
ignore_token(parser);
|
3525
3515
|
}
|
3526
|
-
return
|
3516
|
+
return;
|
3527
3517
|
}
|
3528
3518
|
if (tag_is(token, kEndTag, GUMBO_TAG_TABLE)) {
|
3529
3519
|
if (!close_table(parser)) {
|
3530
3520
|
parser_add_parse_error(parser, token);
|
3531
|
-
return
|
3521
|
+
return;
|
3532
3522
|
}
|
3533
|
-
return
|
3523
|
+
return;
|
3534
3524
|
}
|
3535
3525
|
if (
|
3536
3526
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3540,13 +3530,14 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3540
3530
|
) {
|
3541
3531
|
parser_add_parse_error(parser, token);
|
3542
3532
|
ignore_token(parser);
|
3543
|
-
return
|
3533
|
+
return;
|
3544
3534
|
}
|
3545
3535
|
if (
|
3546
3536
|
tag_in(token, kStartTag, &(const TagSet){TAG(STYLE), TAG(SCRIPT), TAG(TEMPLATE)})
|
3547
3537
|
|| (tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE))
|
3548
3538
|
) {
|
3549
|
-
|
3539
|
+
handle_in_head(parser, token);
|
3540
|
+
return;
|
3550
3541
|
}
|
3551
3542
|
if (
|
3552
3543
|
tag_is(token, kStartTag, GUMBO_TAG_INPUT)
|
@@ -3556,35 +3547,35 @@ static bool handle_in_table(GumboParser* parser, GumboToken* token) {
|
|
3556
3547
|
insert_element_from_token(parser, token);
|
3557
3548
|
pop_current_node(parser);
|
3558
3549
|
acknowledge_self_closing_tag(parser);
|
3559
|
-
return
|
3550
|
+
return;
|
3560
3551
|
}
|
3561
3552
|
if (tag_is(token, kStartTag, GUMBO_TAG_FORM)) {
|
3562
3553
|
parser_add_parse_error(parser, token);
|
3563
3554
|
if (state->_form_element || has_open_element(parser, GUMBO_TAG_TEMPLATE)) {
|
3564
3555
|
ignore_token(parser);
|
3565
|
-
return
|
3556
|
+
return;
|
3566
3557
|
}
|
3567
3558
|
state->_form_element = insert_element_from_token(parser, token);
|
3568
3559
|
pop_current_node(parser);
|
3569
|
-
return
|
3560
|
+
return;
|
3570
3561
|
}
|
3571
3562
|
if (token->type == GUMBO_TOKEN_EOF) {
|
3572
|
-
|
3563
|
+
handle_in_body(parser, token);
|
3564
|
+
return;
|
3573
3565
|
}
|
3574
3566
|
// foster-parenting-start-tag or foster-parenting-end-tag error
|
3575
3567
|
parser_add_parse_error(parser, token);
|
3576
3568
|
state->_foster_parent_insertions = true;
|
3577
|
-
|
3569
|
+
handle_in_body(parser, token);
|
3578
3570
|
state->_foster_parent_insertions = false;
|
3579
|
-
return result;
|
3580
3571
|
}
|
3581
3572
|
|
3582
3573
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intabletext
|
3583
|
-
static
|
3574
|
+
static void handle_in_table_text(GumboParser* parser, GumboToken* token) {
|
3584
3575
|
if (token->type == GUMBO_TOKEN_NULL) {
|
3585
3576
|
parser_add_parse_error(parser, token);
|
3586
3577
|
ignore_token(parser);
|
3587
|
-
return
|
3578
|
+
return;
|
3588
3579
|
}
|
3589
3580
|
GumboParserState* state = parser->_parser_state;
|
3590
3581
|
// Non-whitespace tokens will cause parse errors later.
|
@@ -3594,7 +3585,7 @@ static bool handle_in_table_text(GumboParser* parser, GumboToken* token) {
|
|
3594
3585
|
|| token->type == GUMBO_TOKEN_CHARACTER) {
|
3595
3586
|
insert_text_token(parser, token);
|
3596
3587
|
gumbo_character_token_buffer_append(token, &state->_table_character_tokens);
|
3597
|
-
return
|
3588
|
+
return;
|
3598
3589
|
}
|
3599
3590
|
|
3600
3591
|
GumboCharacterTokenBuffer* buffer = &state->_table_character_tokens;
|
@@ -3616,26 +3607,24 @@ static bool handle_in_table_text(GumboParser* parser, GumboToken* token) {
|
|
3616
3607
|
state->_foster_parent_insertions = false;
|
3617
3608
|
state->_reprocess_current_token = true;
|
3618
3609
|
state->_insertion_mode = state->_original_insertion_mode;
|
3619
|
-
return true;
|
3620
3610
|
}
|
3621
3611
|
|
3622
3612
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-incaption
|
3623
|
-
static
|
3613
|
+
static void handle_in_caption(GumboParser* parser, GumboToken* token) {
|
3624
3614
|
if (tag_is(token, kEndTag, GUMBO_TAG_CAPTION)) {
|
3625
3615
|
if (!has_an_element_in_table_scope(parser, GUMBO_TAG_CAPTION)) {
|
3626
3616
|
parser_add_parse_error(parser, token);
|
3627
3617
|
ignore_token(parser);
|
3628
|
-
return
|
3618
|
+
return;
|
3629
3619
|
}
|
3630
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3631
|
-
|
3632
|
-
if (!result)
|
3620
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3621
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_CAPTION))
|
3633
3622
|
parser_add_parse_error(parser, token);
|
3634
3623
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_CAPTION))
|
3635
3624
|
;
|
3636
3625
|
clear_active_formatting_elements(parser);
|
3637
3626
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3638
|
-
return
|
3627
|
+
return;
|
3639
3628
|
}
|
3640
3629
|
if (
|
3641
3630
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3647,18 +3636,17 @@ static bool handle_in_caption(GumboParser* parser, GumboToken* token) {
|
|
3647
3636
|
if (!has_an_element_in_table_scope(parser, GUMBO_TAG_CAPTION)) {
|
3648
3637
|
parser_add_parse_error(parser, token);
|
3649
3638
|
ignore_token(parser);
|
3650
|
-
return
|
3639
|
+
return;
|
3651
3640
|
}
|
3652
|
-
generate_implied_end_tags(parser, GUMBO_TAG_LAST);
|
3653
|
-
|
3654
|
-
if (!result)
|
3641
|
+
generate_implied_end_tags(parser, GUMBO_TAG_LAST, NULL);
|
3642
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_CAPTION))
|
3655
3643
|
parser_add_parse_error(parser, token);
|
3656
3644
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_CAPTION))
|
3657
3645
|
;
|
3658
3646
|
clear_active_formatting_elements(parser);
|
3659
3647
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3660
3648
|
parser->_parser_state->_reprocess_current_token = true;
|
3661
|
-
return
|
3649
|
+
return;
|
3662
3650
|
}
|
3663
3651
|
if (
|
3664
3652
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3668,77 +3656,79 @@ static bool handle_in_caption(GumboParser* parser, GumboToken* token) {
|
|
3668
3656
|
) {
|
3669
3657
|
parser_add_parse_error(parser, token);
|
3670
3658
|
ignore_token(parser);
|
3671
|
-
return
|
3659
|
+
return;
|
3672
3660
|
}
|
3673
|
-
|
3661
|
+
handle_in_body(parser, token);
|
3674
3662
|
}
|
3675
3663
|
|
3676
3664
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-incolgroup
|
3677
|
-
static
|
3665
|
+
static void handle_in_column_group(GumboParser* parser, GumboToken* token) {
|
3678
3666
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
3679
3667
|
insert_text_token(parser, token);
|
3680
|
-
return
|
3668
|
+
return;
|
3681
3669
|
}
|
3682
3670
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
3683
3671
|
append_comment_node(parser, get_current_node(parser), token);
|
3684
|
-
return
|
3672
|
+
return;
|
3685
3673
|
}
|
3686
3674
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
3687
3675
|
parser_add_parse_error(parser, token);
|
3688
3676
|
ignore_token(parser);
|
3689
|
-
return
|
3677
|
+
return;
|
3690
3678
|
}
|
3691
3679
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
3692
|
-
|
3680
|
+
handle_in_body(parser, token);
|
3681
|
+
return;
|
3693
3682
|
}
|
3694
3683
|
if (tag_is(token, kStartTag, GUMBO_TAG_COL)) {
|
3695
3684
|
insert_element_from_token(parser, token);
|
3696
3685
|
pop_current_node(parser);
|
3697
3686
|
acknowledge_self_closing_tag(parser);
|
3698
|
-
return
|
3687
|
+
return;
|
3699
3688
|
}
|
3700
3689
|
if (tag_is(token, kEndTag, GUMBO_TAG_COLGROUP)) {
|
3701
3690
|
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_COLGROUP)) {
|
3702
3691
|
parser_add_parse_error(parser, token);
|
3703
3692
|
ignore_token(parser);
|
3704
|
-
return
|
3693
|
+
return;
|
3705
3694
|
}
|
3706
3695
|
pop_current_node(parser);
|
3707
3696
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3708
|
-
return
|
3697
|
+
return;
|
3709
3698
|
}
|
3710
3699
|
if (tag_is(token, kEndTag, GUMBO_TAG_COL)) {
|
3711
3700
|
parser_add_parse_error(parser, token);
|
3712
3701
|
ignore_token(parser);
|
3713
|
-
return
|
3702
|
+
return;
|
3714
3703
|
}
|
3715
3704
|
if (
|
3716
3705
|
tag_is(token, kStartTag, GUMBO_TAG_TEMPLATE)
|
3717
3706
|
|| tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)
|
3718
3707
|
) {
|
3719
|
-
|
3708
|
+
handle_in_head(parser, token);
|
3709
|
+
return;
|
3720
3710
|
}
|
3721
3711
|
if (token->type == GUMBO_TOKEN_EOF) {
|
3722
|
-
|
3712
|
+
handle_in_body(parser, token);
|
3713
|
+
return;
|
3723
3714
|
}
|
3724
3715
|
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_COLGROUP)) {
|
3725
3716
|
parser_add_parse_error(parser, token);
|
3726
3717
|
ignore_token(parser);
|
3727
|
-
return
|
3718
|
+
return;
|
3728
3719
|
}
|
3729
3720
|
pop_current_node(parser);
|
3730
3721
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3731
3722
|
parser->_parser_state->_reprocess_current_token = true;
|
3732
|
-
return true;
|
3733
3723
|
}
|
3734
3724
|
|
3735
3725
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intbody
|
3736
|
-
static
|
3726
|
+
static void handle_in_table_body(GumboParser* parser, GumboToken* token) {
|
3737
3727
|
if (tag_is(token, kStartTag, GUMBO_TAG_TR)) {
|
3738
3728
|
clear_stack_to_table_body_context(parser);
|
3739
3729
|
insert_element_from_token(parser, token);
|
3740
3730
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_ROW);
|
3741
|
-
return
|
3731
|
+
return;
|
3742
3732
|
}
|
3743
3733
|
if (tag_in(token, kStartTag, &td_th_tags)) {
|
3744
3734
|
parser_add_parse_error(parser, token);
|
@@ -3746,7 +3736,7 @@ static bool handle_in_table_body(GumboParser* parser, GumboToken* token) {
|
|
3746
3736
|
insert_element_of_tag_type(parser, GUMBO_TAG_TR, GUMBO_INSERTION_IMPLIED);
|
3747
3737
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_ROW);
|
3748
3738
|
parser->_parser_state->_reprocess_current_token = true;
|
3749
|
-
return
|
3739
|
+
return;
|
3750
3740
|
}
|
3751
3741
|
if (
|
3752
3742
|
tag_in(token, kEndTag, &(const TagSet){TAG(TBODY), TAG(TFOOT), TAG(THEAD)})
|
@@ -3754,12 +3744,12 @@ static bool handle_in_table_body(GumboParser* parser, GumboToken* token) {
|
|
3754
3744
|
if (!has_an_element_in_table_scope(parser, token->v.end_tag.tag)) {
|
3755
3745
|
parser_add_parse_error(parser, token);
|
3756
3746
|
ignore_token(parser);
|
3757
|
-
return
|
3747
|
+
return;
|
3758
3748
|
}
|
3759
3749
|
clear_stack_to_table_body_context(parser);
|
3760
3750
|
pop_current_node(parser);
|
3761
3751
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3762
|
-
return
|
3752
|
+
return;
|
3763
3753
|
}
|
3764
3754
|
if (
|
3765
3755
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3777,13 +3767,13 @@ static bool handle_in_table_body(GumboParser* parser, GumboToken* token) {
|
|
3777
3767
|
) {
|
3778
3768
|
parser_add_parse_error(parser, token);
|
3779
3769
|
ignore_token(parser);
|
3780
|
-
return
|
3770
|
+
return;
|
3781
3771
|
}
|
3782
3772
|
clear_stack_to_table_body_context(parser);
|
3783
3773
|
pop_current_node(parser);
|
3784
3774
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
3785
3775
|
parser->_parser_state->_reprocess_current_token = true;
|
3786
|
-
return
|
3776
|
+
return;
|
3787
3777
|
}
|
3788
3778
|
if (
|
3789
3779
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3793,30 +3783,30 @@ static bool handle_in_table_body(GumboParser* parser, GumboToken* token) {
|
|
3793
3783
|
) {
|
3794
3784
|
parser_add_parse_error(parser, token);
|
3795
3785
|
ignore_token(parser);
|
3796
|
-
return
|
3786
|
+
return;
|
3797
3787
|
}
|
3798
|
-
|
3788
|
+
handle_in_table(parser, token);
|
3799
3789
|
}
|
3800
3790
|
|
3801
3791
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intr
|
3802
|
-
static
|
3792
|
+
static void handle_in_row(GumboParser* parser, GumboToken* token) {
|
3803
3793
|
if (tag_in(token, kStartTag, &td_th_tags)) {
|
3804
3794
|
clear_stack_to_table_row_context(parser);
|
3805
3795
|
insert_element_from_token(parser, token);
|
3806
3796
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_CELL);
|
3807
3797
|
add_formatting_element(parser, &kActiveFormattingScopeMarker);
|
3808
|
-
return
|
3798
|
+
return;
|
3809
3799
|
}
|
3810
3800
|
if (tag_is(token, kEndTag, GUMBO_TAG_TR)) {
|
3811
3801
|
if (!has_an_element_in_table_scope(parser, GUMBO_TAG_TR)) {
|
3812
3802
|
parser_add_parse_error(parser, token);
|
3813
3803
|
ignore_token(parser);
|
3814
|
-
return
|
3804
|
+
return;
|
3815
3805
|
}
|
3816
3806
|
clear_stack_to_table_row_context(parser);
|
3817
3807
|
pop_current_node(parser);
|
3818
3808
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
3819
|
-
return
|
3809
|
+
return;
|
3820
3810
|
}
|
3821
3811
|
if (
|
3822
3812
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3828,13 +3818,13 @@ static bool handle_in_row(GumboParser* parser, GumboToken* token) {
|
|
3828
3818
|
if (!has_an_element_in_table_scope(parser, GUMBO_TAG_TR)) {
|
3829
3819
|
parser_add_parse_error(parser, token);
|
3830
3820
|
ignore_token(parser);
|
3831
|
-
return
|
3821
|
+
return;
|
3832
3822
|
}
|
3833
3823
|
clear_stack_to_table_row_context(parser);
|
3834
3824
|
pop_current_node(parser);
|
3835
3825
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
3836
3826
|
parser->_parser_state->_reprocess_current_token = true;
|
3837
|
-
return
|
3827
|
+
return;
|
3838
3828
|
}
|
3839
3829
|
if (
|
3840
3830
|
tag_in(token, kEndTag, &(const TagSet) {TAG(TBODY), TAG(TFOOT), TAG(THEAD)})
|
@@ -3842,17 +3832,17 @@ static bool handle_in_row(GumboParser* parser, GumboToken* token) {
|
|
3842
3832
|
if (!has_an_element_in_table_scope(parser, token->v.end_tag.tag)) {
|
3843
3833
|
parser_add_parse_error(parser, token);
|
3844
3834
|
ignore_token(parser);
|
3845
|
-
return
|
3835
|
+
return;
|
3846
3836
|
}
|
3847
3837
|
if (!has_an_element_in_table_scope(parser, GUMBO_TAG_TR)) {
|
3848
3838
|
ignore_token(parser);
|
3849
|
-
return
|
3839
|
+
return;
|
3850
3840
|
}
|
3851
3841
|
clear_stack_to_table_row_context(parser);
|
3852
3842
|
pop_current_node(parser);
|
3853
3843
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
3854
3844
|
parser->_parser_state->_reprocess_current_token = true;
|
3855
|
-
return
|
3845
|
+
return;
|
3856
3846
|
}
|
3857
3847
|
if (
|
3858
3848
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3862,21 +3852,22 @@ static bool handle_in_row(GumboParser* parser, GumboToken* token) {
|
|
3862
3852
|
) {
|
3863
3853
|
parser_add_parse_error(parser, token);
|
3864
3854
|
ignore_token(parser);
|
3865
|
-
return
|
3855
|
+
return;
|
3866
3856
|
}
|
3867
|
-
|
3857
|
+
handle_in_table(parser, token);
|
3868
3858
|
}
|
3869
3859
|
|
3870
3860
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intd
|
3871
|
-
static
|
3861
|
+
static void handle_in_cell(GumboParser* parser, GumboToken* token) {
|
3872
3862
|
if (tag_in(token, kEndTag, &td_th_tags)) {
|
3873
3863
|
GumboTag token_tag = token->v.end_tag.tag;
|
3874
3864
|
if (!has_an_element_in_table_scope(parser, token_tag)) {
|
3875
3865
|
parser_add_parse_error(parser, token);
|
3876
3866
|
ignore_token(parser);
|
3877
|
-
return
|
3867
|
+
return;
|
3878
3868
|
}
|
3879
|
-
|
3869
|
+
close_table_cell(parser, token, token_tag);
|
3870
|
+
return;
|
3880
3871
|
}
|
3881
3872
|
if (
|
3882
3873
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -3892,10 +3883,11 @@ static bool handle_in_cell(GumboParser* parser, GumboToken* token) {
|
|
3892
3883
|
gumbo_debug("Bailing out because there's no <td> or <th> in scope.\n");
|
3893
3884
|
parser_add_parse_error(parser, token);
|
3894
3885
|
ignore_token(parser);
|
3895
|
-
return
|
3886
|
+
return;
|
3896
3887
|
}
|
3897
3888
|
parser->_parser_state->_reprocess_current_token = true;
|
3898
|
-
|
3889
|
+
close_current_cell(parser, token);
|
3890
|
+
return;
|
3899
3891
|
}
|
3900
3892
|
if (
|
3901
3893
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3904,7 +3896,7 @@ static bool handle_in_cell(GumboParser* parser, GumboToken* token) {
|
|
3904
3896
|
) {
|
3905
3897
|
parser_add_parse_error(parser, token);
|
3906
3898
|
ignore_token(parser);
|
3907
|
-
return
|
3899
|
+
return;
|
3908
3900
|
}
|
3909
3901
|
if (
|
3910
3902
|
tag_in(token, kEndTag, &(const TagSet) {
|
@@ -3914,46 +3906,48 @@ static bool handle_in_cell(GumboParser* parser, GumboToken* token) {
|
|
3914
3906
|
if (!has_an_element_in_table_scope(parser, token->v.end_tag.tag)) {
|
3915
3907
|
parser_add_parse_error(parser, token);
|
3916
3908
|
ignore_token(parser);
|
3917
|
-
return
|
3909
|
+
return;
|
3918
3910
|
}
|
3919
3911
|
parser->_parser_state->_reprocess_current_token = true;
|
3920
|
-
|
3912
|
+
close_current_cell(parser, token);
|
3913
|
+
return;
|
3921
3914
|
}
|
3922
|
-
|
3915
|
+
handle_in_body(parser, token);
|
3923
3916
|
}
|
3924
3917
|
|
3925
3918
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inselect
|
3926
|
-
static
|
3919
|
+
static void handle_in_select(GumboParser* parser, GumboToken* token) {
|
3927
3920
|
if (token->type == GUMBO_TOKEN_NULL) {
|
3928
3921
|
parser_add_parse_error(parser, token);
|
3929
3922
|
ignore_token(parser);
|
3930
|
-
return
|
3923
|
+
return;
|
3931
3924
|
}
|
3932
3925
|
if (
|
3933
3926
|
token->type == GUMBO_TOKEN_CHARACTER
|
3934
3927
|
|| token->type == GUMBO_TOKEN_WHITESPACE
|
3935
3928
|
) {
|
3936
3929
|
insert_text_token(parser, token);
|
3937
|
-
return
|
3930
|
+
return;
|
3938
3931
|
}
|
3939
3932
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
3940
3933
|
append_comment_node(parser, get_current_node(parser), token);
|
3941
|
-
return
|
3934
|
+
return;
|
3942
3935
|
}
|
3943
3936
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
3944
3937
|
parser_add_parse_error(parser, token);
|
3945
3938
|
ignore_token(parser);
|
3946
|
-
return
|
3939
|
+
return;
|
3947
3940
|
}
|
3948
3941
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
3949
|
-
|
3942
|
+
handle_in_body(parser, token);
|
3943
|
+
return;
|
3950
3944
|
}
|
3951
3945
|
if (tag_is(token, kStartTag, GUMBO_TAG_OPTION)) {
|
3952
3946
|
if (node_html_tag_is(get_current_node(parser), GUMBO_TAG_OPTION)) {
|
3953
3947
|
pop_current_node(parser);
|
3954
3948
|
}
|
3955
3949
|
insert_element_from_token(parser, token);
|
3956
|
-
return
|
3950
|
+
return;
|
3957
3951
|
}
|
3958
3952
|
if (tag_is(token, kStartTag, GUMBO_TAG_OPTGROUP)) {
|
3959
3953
|
if (node_html_tag_is(get_current_node(parser), GUMBO_TAG_OPTION)) {
|
@@ -3963,7 +3957,7 @@ static bool handle_in_select(GumboParser* parser, GumboToken* token) {
|
|
3963
3957
|
pop_current_node(parser);
|
3964
3958
|
}
|
3965
3959
|
insert_element_from_token(parser, token);
|
3966
|
-
return
|
3960
|
+
return;
|
3967
3961
|
}
|
3968
3962
|
if (tag_is(token, kEndTag, GUMBO_TAG_OPTGROUP)) {
|
3969
3963
|
GumboVector* open_elements = &parser->_parser_state->_open_elements;
|
@@ -3978,29 +3972,29 @@ static bool handle_in_select(GumboParser* parser, GumboToken* token) {
|
|
3978
3972
|
}
|
3979
3973
|
if (node_html_tag_is(get_current_node(parser), GUMBO_TAG_OPTGROUP)) {
|
3980
3974
|
pop_current_node(parser);
|
3981
|
-
return
|
3975
|
+
return;
|
3982
3976
|
}
|
3983
3977
|
parser_add_parse_error(parser, token);
|
3984
3978
|
ignore_token(parser);
|
3985
|
-
return
|
3979
|
+
return;
|
3986
3980
|
}
|
3987
3981
|
if (tag_is(token, kEndTag, GUMBO_TAG_OPTION)) {
|
3988
3982
|
if (node_html_tag_is(get_current_node(parser), GUMBO_TAG_OPTION)) {
|
3989
3983
|
pop_current_node(parser);
|
3990
|
-
return
|
3984
|
+
return;
|
3991
3985
|
}
|
3992
3986
|
parser_add_parse_error(parser, token);
|
3993
3987
|
ignore_token(parser);
|
3994
|
-
return
|
3988
|
+
return;
|
3995
3989
|
}
|
3996
3990
|
if (tag_is(token, kEndTag, GUMBO_TAG_SELECT)) {
|
3997
3991
|
if (!has_an_element_in_select_scope(parser, GUMBO_TAG_SELECT)) {
|
3998
3992
|
parser_add_parse_error(parser, token);
|
3999
3993
|
ignore_token(parser);
|
4000
|
-
return
|
3994
|
+
return;
|
4001
3995
|
}
|
4002
3996
|
close_current_select(parser);
|
4003
|
-
return
|
3997
|
+
return;
|
4004
3998
|
}
|
4005
3999
|
if (tag_is(token, kStartTag, GUMBO_TAG_SELECT)) {
|
4006
4000
|
parser_add_parse_error(parser, token);
|
@@ -4008,7 +4002,7 @@ static bool handle_in_select(GumboParser* parser, GumboToken* token) {
|
|
4008
4002
|
if (has_an_element_in_select_scope(parser, GUMBO_TAG_SELECT)) {
|
4009
4003
|
close_current_select(parser);
|
4010
4004
|
}
|
4011
|
-
return
|
4005
|
+
return;
|
4012
4006
|
}
|
4013
4007
|
if (
|
4014
4008
|
tag_in(token, kStartTag, &(const TagSet) {TAG(INPUT), TAG(KEYGEN), TAG(TEXTAREA)})
|
@@ -4020,23 +4014,25 @@ static bool handle_in_select(GumboParser* parser, GumboToken* token) {
|
|
4020
4014
|
close_current_select(parser);
|
4021
4015
|
parser->_parser_state->_reprocess_current_token = true;
|
4022
4016
|
}
|
4023
|
-
return
|
4017
|
+
return;
|
4024
4018
|
}
|
4025
4019
|
if (
|
4026
4020
|
tag_in(token, kStartTag, &(const TagSet){TAG(SCRIPT), TAG(TEMPLATE)})
|
4027
4021
|
|| tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)
|
4028
4022
|
) {
|
4029
|
-
|
4023
|
+
handle_in_head(parser, token);
|
4024
|
+
return;
|
4025
|
+
}
|
4026
|
+
if (token->type == GUMBO_TOKEN_EOF) {
|
4027
|
+
handle_in_body(parser, token);
|
4028
|
+
return;
|
4030
4029
|
}
|
4031
|
-
if (token->type == GUMBO_TOKEN_EOF)
|
4032
|
-
return handle_in_body(parser, token);
|
4033
4030
|
parser_add_parse_error(parser, token);
|
4034
4031
|
ignore_token(parser);
|
4035
|
-
return false;
|
4036
4032
|
}
|
4037
4033
|
|
4038
4034
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inselectintable
|
4039
|
-
static
|
4035
|
+
static void handle_in_select_in_table(GumboParser* parser, GumboToken* token) {
|
4040
4036
|
static const TagSet tags = {
|
4041
4037
|
TAG(CAPTION), TAG(TABLE), TAG(TBODY), TAG(TFOOT), TAG(THEAD),
|
4042
4038
|
TAG(TR), TAG(TD), TAG(TH)
|
@@ -4045,23 +4041,23 @@ static bool handle_in_select_in_table(GumboParser* parser, GumboToken* token) {
|
|
4045
4041
|
parser_add_parse_error(parser, token);
|
4046
4042
|
close_current_select(parser);
|
4047
4043
|
parser->_parser_state->_reprocess_current_token = true;
|
4048
|
-
return
|
4044
|
+
return;
|
4049
4045
|
}
|
4050
4046
|
if (tag_in(token, kEndTag, &tags)) {
|
4051
4047
|
parser_add_parse_error(parser, token);
|
4052
4048
|
if (!has_an_element_in_table_scope(parser, token->v.end_tag.tag)) {
|
4053
4049
|
ignore_token(parser);
|
4054
|
-
return
|
4050
|
+
return;
|
4055
4051
|
}
|
4056
4052
|
close_current_select(parser);
|
4057
4053
|
parser->_parser_state->_reprocess_current_token = true;
|
4058
|
-
return
|
4054
|
+
return;
|
4059
4055
|
}
|
4060
|
-
|
4056
|
+
handle_in_select(parser, token);
|
4061
4057
|
}
|
4062
4058
|
|
4063
4059
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-intemplate
|
4064
|
-
static
|
4060
|
+
static void handle_in_template(GumboParser* parser, GumboToken* token) {
|
4065
4061
|
GumboParserState* state = parser->_parser_state;
|
4066
4062
|
switch (token->type) {
|
4067
4063
|
case GUMBO_TOKEN_WHITESPACE:
|
@@ -4069,7 +4065,8 @@ static bool handle_in_template(GumboParser* parser, GumboToken* token) {
|
|
4069
4065
|
case GUMBO_TOKEN_COMMENT:
|
4070
4066
|
case GUMBO_TOKEN_NULL:
|
4071
4067
|
case GUMBO_TOKEN_DOCTYPE:
|
4072
|
-
|
4068
|
+
handle_in_body(parser, token);
|
4069
|
+
return;
|
4073
4070
|
default:
|
4074
4071
|
break;
|
4075
4072
|
}
|
@@ -4080,7 +4077,8 @@ static bool handle_in_template(GumboParser* parser, GumboToken* token) {
|
|
4080
4077
|
})
|
4081
4078
|
|| tag_is(token, kEndTag, GUMBO_TAG_TEMPLATE)
|
4082
4079
|
) {
|
4083
|
-
|
4080
|
+
handle_in_head(parser, token);
|
4081
|
+
return;
|
4084
4082
|
}
|
4085
4083
|
if (
|
4086
4084
|
tag_in(token, kStartTag, &(const TagSet) {
|
@@ -4091,45 +4089,45 @@ static bool handle_in_template(GumboParser* parser, GumboToken* token) {
|
|
4091
4089
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
4092
4090
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE);
|
4093
4091
|
state->_reprocess_current_token = true;
|
4094
|
-
return
|
4092
|
+
return;
|
4095
4093
|
}
|
4096
4094
|
if (tag_is(token, kStartTag, GUMBO_TAG_COL)) {
|
4097
4095
|
pop_template_insertion_mode(parser);
|
4098
4096
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_COLUMN_GROUP);
|
4099
4097
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_COLUMN_GROUP);
|
4100
4098
|
state->_reprocess_current_token = true;
|
4101
|
-
return
|
4099
|
+
return;
|
4102
4100
|
}
|
4103
4101
|
if (tag_is(token, kStartTag, GUMBO_TAG_TR)) {
|
4104
4102
|
pop_template_insertion_mode(parser);
|
4105
4103
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
4106
4104
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_TABLE_BODY);
|
4107
4105
|
state->_reprocess_current_token = true;
|
4108
|
-
return
|
4106
|
+
return;
|
4109
4107
|
}
|
4110
4108
|
if (tag_in(token, kStartTag, &td_th_tags)) {
|
4111
4109
|
pop_template_insertion_mode(parser);
|
4112
4110
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_ROW);
|
4113
4111
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_ROW);
|
4114
4112
|
state->_reprocess_current_token = true;
|
4115
|
-
return
|
4113
|
+
return;
|
4116
4114
|
}
|
4117
4115
|
if (token->type == GUMBO_TOKEN_START_TAG) {
|
4118
4116
|
pop_template_insertion_mode(parser);
|
4119
4117
|
push_template_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
4120
4118
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
4121
4119
|
state->_reprocess_current_token = true;
|
4122
|
-
return
|
4120
|
+
return;
|
4123
4121
|
}
|
4124
4122
|
if (token->type == GUMBO_TOKEN_END_TAG) {
|
4125
4123
|
parser_add_parse_error(parser, token);
|
4126
4124
|
ignore_token(parser);
|
4127
|
-
return
|
4125
|
+
return;
|
4128
4126
|
}
|
4129
4127
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4130
4128
|
if (!has_open_element(parser, GUMBO_TAG_TEMPLATE)) {
|
4131
4129
|
// Stop parsing.
|
4132
|
-
return
|
4130
|
+
return;
|
4133
4131
|
}
|
4134
4132
|
parser_add_parse_error(parser, token);
|
4135
4133
|
while (!node_html_tag_is(pop_current_node(parser), GUMBO_TAG_TEMPLATE))
|
@@ -4138,40 +4136,41 @@ static bool handle_in_template(GumboParser* parser, GumboToken* token) {
|
|
4138
4136
|
pop_template_insertion_mode(parser);
|
4139
4137
|
reset_insertion_mode_appropriately(parser);
|
4140
4138
|
state->_reprocess_current_token = true;
|
4141
|
-
return
|
4139
|
+
return;
|
4142
4140
|
}
|
4143
4141
|
assert(0 && "unreachable");
|
4144
|
-
return false;
|
4145
4142
|
}
|
4146
4143
|
|
4147
4144
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-afterbody
|
4148
|
-
static
|
4145
|
+
static void handle_after_body(GumboParser* parser, GumboToken* token) {
|
4149
4146
|
if (
|
4150
4147
|
token->type == GUMBO_TOKEN_WHITESPACE
|
4151
4148
|
|| tag_is(token, kStartTag, GUMBO_TAG_HTML)
|
4152
4149
|
) {
|
4153
|
-
|
4150
|
+
handle_in_body(parser, token);
|
4151
|
+
return;
|
4154
4152
|
}
|
4155
4153
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
4156
4154
|
GumboNode* html_node = parser->_output->root;
|
4157
4155
|
assert(html_node != NULL);
|
4158
4156
|
append_comment_node(parser, html_node, token);
|
4159
|
-
return
|
4157
|
+
return;
|
4160
4158
|
}
|
4161
4159
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
4162
4160
|
parser_add_parse_error(parser, token);
|
4163
4161
|
ignore_token(parser);
|
4164
|
-
return
|
4162
|
+
return;
|
4165
4163
|
}
|
4166
4164
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
4167
|
-
|
4165
|
+
handle_in_body(parser, token);
|
4166
|
+
return;
|
4168
4167
|
}
|
4169
4168
|
if (tag_is(token, kEndTag, GUMBO_TAG_HTML)) {
|
4170
4169
|
/* fragment case: ignore the closing HTML token */
|
4171
4170
|
if (is_fragment_parser(parser)) {
|
4172
4171
|
parser_add_parse_error(parser, token);
|
4173
4172
|
ignore_token(parser);
|
4174
|
-
return
|
4173
|
+
return;
|
4175
4174
|
}
|
4176
4175
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_AFTER_BODY);
|
4177
4176
|
GumboNode* html = parser->_parser_state->_open_elements.data[0];
|
@@ -4180,44 +4179,44 @@ static bool handle_after_body(GumboParser* parser, GumboToken* token) {
|
|
4180
4179
|
parser->_parser_state->_current_token,
|
4181
4180
|
&html->v.element
|
4182
4181
|
);
|
4183
|
-
return
|
4182
|
+
return;
|
4184
4183
|
}
|
4185
4184
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4186
|
-
return
|
4185
|
+
return;
|
4187
4186
|
}
|
4188
4187
|
parser_add_parse_error(parser, token);
|
4189
4188
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
4190
4189
|
parser->_parser_state->_reprocess_current_token = true;
|
4191
|
-
return false;
|
4192
4190
|
}
|
4193
4191
|
|
4194
4192
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inframeset
|
4195
|
-
static
|
4193
|
+
static void handle_in_frameset(GumboParser* parser, GumboToken* token) {
|
4196
4194
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
4197
4195
|
insert_text_token(parser, token);
|
4198
|
-
return
|
4196
|
+
return;
|
4199
4197
|
}
|
4200
4198
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
4201
4199
|
append_comment_node(parser, get_current_node(parser), token);
|
4202
|
-
return
|
4200
|
+
return;
|
4203
4201
|
}
|
4204
4202
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
4205
4203
|
parser_add_parse_error(parser, token);
|
4206
4204
|
ignore_token(parser);
|
4207
|
-
return
|
4205
|
+
return;
|
4208
4206
|
}
|
4209
4207
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
4210
|
-
|
4208
|
+
handle_in_body(parser, token);
|
4209
|
+
return;
|
4211
4210
|
}
|
4212
4211
|
if (tag_is(token, kStartTag, GUMBO_TAG_FRAMESET)) {
|
4213
4212
|
insert_element_from_token(parser, token);
|
4214
|
-
return
|
4213
|
+
return;
|
4215
4214
|
}
|
4216
4215
|
if (tag_is(token, kEndTag, GUMBO_TAG_FRAMESET)) {
|
4217
4216
|
if (node_html_tag_is(get_current_node(parser), GUMBO_TAG_HTML)) {
|
4218
4217
|
parser_add_parse_error(parser, token);
|
4219
4218
|
ignore_token(parser);
|
4220
|
-
return
|
4219
|
+
return;
|
4221
4220
|
}
|
4222
4221
|
pop_current_node(parser);
|
4223
4222
|
if (
|
@@ -4226,46 +4225,45 @@ static bool handle_in_frameset(GumboParser* parser, GumboToken* token) {
|
|
4226
4225
|
) {
|
4227
4226
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_FRAMESET);
|
4228
4227
|
}
|
4229
|
-
return
|
4228
|
+
return;
|
4230
4229
|
}
|
4231
4230
|
if (tag_is(token, kStartTag, GUMBO_TAG_FRAME)) {
|
4232
4231
|
insert_element_from_token(parser, token);
|
4233
4232
|
pop_current_node(parser);
|
4234
4233
|
acknowledge_self_closing_tag(parser);
|
4235
|
-
return
|
4234
|
+
return;
|
4236
4235
|
}
|
4237
4236
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOFRAMES)) {
|
4238
|
-
|
4237
|
+
handle_in_head(parser, token);
|
4238
|
+
return;
|
4239
4239
|
}
|
4240
4240
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4241
|
-
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_HTML))
|
4241
|
+
if (!node_html_tag_is(get_current_node(parser), GUMBO_TAG_HTML))
|
4242
4242
|
parser_add_parse_error(parser, token);
|
4243
|
-
|
4244
|
-
}
|
4245
|
-
return true;
|
4243
|
+
return;
|
4246
4244
|
}
|
4247
4245
|
parser_add_parse_error(parser, token);
|
4248
4246
|
ignore_token(parser);
|
4249
|
-
return false;
|
4250
4247
|
}
|
4251
4248
|
|
4252
4249
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-afterframeset
|
4253
|
-
static
|
4250
|
+
static void handle_after_frameset(GumboParser* parser, GumboToken* token) {
|
4254
4251
|
if (token->type == GUMBO_TOKEN_WHITESPACE) {
|
4255
4252
|
insert_text_token(parser, token);
|
4256
|
-
return
|
4253
|
+
return;
|
4257
4254
|
}
|
4258
4255
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
4259
4256
|
append_comment_node(parser, get_current_node(parser), token);
|
4260
|
-
return
|
4257
|
+
return;
|
4261
4258
|
}
|
4262
4259
|
if (token->type == GUMBO_TOKEN_DOCTYPE) {
|
4263
4260
|
parser_add_parse_error(parser, token);
|
4264
4261
|
ignore_token(parser);
|
4265
|
-
return
|
4262
|
+
return;
|
4266
4263
|
}
|
4267
4264
|
if (tag_is(token, kStartTag, GUMBO_TAG_HTML)) {
|
4268
|
-
|
4265
|
+
handle_in_body(parser, token);
|
4266
|
+
return;
|
4269
4267
|
}
|
4270
4268
|
if (tag_is(token, kEndTag, GUMBO_TAG_HTML)) {
|
4271
4269
|
GumboNode* html = parser->_parser_state->_open_elements.data[0];
|
@@ -4275,71 +4273,71 @@ static bool handle_after_frameset(GumboParser* parser, GumboToken* token) {
|
|
4275
4273
|
&html->v.element
|
4276
4274
|
);
|
4277
4275
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_AFTER_AFTER_FRAMESET);
|
4278
|
-
return
|
4276
|
+
return;
|
4279
4277
|
}
|
4280
4278
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOFRAMES)) {
|
4281
4279
|
return handle_in_head(parser, token);
|
4282
4280
|
}
|
4283
4281
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4284
|
-
return
|
4282
|
+
return;
|
4285
4283
|
}
|
4286
4284
|
parser_add_parse_error(parser, token);
|
4287
4285
|
ignore_token(parser);
|
4288
|
-
return false;
|
4289
4286
|
}
|
4290
4287
|
|
4291
4288
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-after-after-body-insertion-mode
|
4292
|
-
static
|
4289
|
+
static void handle_after_after_body(GumboParser* parser, GumboToken* token) {
|
4293
4290
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
4294
4291
|
append_comment_node(parser, get_document_node(parser), token);
|
4295
|
-
return
|
4292
|
+
return;
|
4296
4293
|
}
|
4297
4294
|
if (
|
4298
4295
|
token->type == GUMBO_TOKEN_DOCTYPE
|
4299
4296
|
|| token->type == GUMBO_TOKEN_WHITESPACE
|
4300
4297
|
|| tag_is(token, kStartTag, GUMBO_TAG_HTML)
|
4301
4298
|
) {
|
4302
|
-
|
4299
|
+
handle_in_body(parser, token);
|
4300
|
+
return;
|
4303
4301
|
}
|
4304
4302
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4305
|
-
return
|
4303
|
+
return;
|
4306
4304
|
}
|
4307
4305
|
parser_add_parse_error(parser, token);
|
4308
4306
|
set_insertion_mode(parser, GUMBO_INSERTION_MODE_IN_BODY);
|
4309
4307
|
parser->_parser_state->_reprocess_current_token = true;
|
4310
|
-
return false;
|
4311
4308
|
}
|
4312
4309
|
|
4313
4310
|
// https://html.spec.whatwg.org/multipage/parsing.html#the-after-after-frameset-insertion-mode
|
4314
|
-
static
|
4311
|
+
static void handle_after_after_frameset (
|
4315
4312
|
GumboParser* parser,
|
4316
4313
|
GumboToken* token
|
4317
4314
|
) {
|
4318
4315
|
if (token->type == GUMBO_TOKEN_COMMENT) {
|
4319
4316
|
append_comment_node(parser, get_document_node(parser), token);
|
4320
|
-
return
|
4317
|
+
return;
|
4321
4318
|
}
|
4322
4319
|
if (
|
4323
4320
|
token->type == GUMBO_TOKEN_DOCTYPE
|
4324
4321
|
|| token->type == GUMBO_TOKEN_WHITESPACE
|
4325
4322
|
|| tag_is(token, kStartTag, GUMBO_TAG_HTML)
|
4326
4323
|
) {
|
4327
|
-
|
4324
|
+
handle_in_body(parser, token);
|
4325
|
+
return;
|
4328
4326
|
}
|
4329
4327
|
if (token->type == GUMBO_TOKEN_EOF) {
|
4330
|
-
return
|
4328
|
+
return;
|
4331
4329
|
}
|
4332
4330
|
if (tag_is(token, kStartTag, GUMBO_TAG_NOFRAMES)) {
|
4333
|
-
|
4331
|
+
handle_in_head(parser, token);
|
4332
|
+
return;
|
4334
4333
|
}
|
4335
4334
|
parser_add_parse_error(parser, token);
|
4336
4335
|
ignore_token(parser);
|
4337
|
-
return false;
|
4338
4336
|
}
|
4339
4337
|
|
4340
4338
|
// Function pointers for each insertion mode.
|
4341
4339
|
// Keep in sync with insertion_mode.h.
|
4342
|
-
typedef
|
4340
|
+
typedef void (*TokenHandler)(GumboParser* parser, GumboToken* token);
|
4343
4341
|
static const TokenHandler kTokenHandlers[] = {
|
4344
4342
|
handle_initial,
|
4345
4343
|
handle_before_html,
|
@@ -4366,36 +4364,36 @@ static const TokenHandler kTokenHandlers[] = {
|
|
4366
4364
|
handle_after_after_frameset
|
4367
4365
|
};
|
4368
4366
|
|
4369
|
-
static
|
4367
|
+
static void handle_html_content(GumboParser* parser, GumboToken* token) {
|
4370
4368
|
const GumboInsertionMode mode = parser->_parser_state->_insertion_mode;
|
4371
4369
|
const TokenHandler handler = kTokenHandlers[mode];
|
4372
|
-
|
4370
|
+
handler(parser, token);
|
4373
4371
|
}
|
4374
4372
|
|
4375
4373
|
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inforeign
|
4376
|
-
static
|
4374
|
+
static void handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
4377
4375
|
gumbo_debug("Handling foreign content");
|
4378
4376
|
switch (token->type) {
|
4379
4377
|
case GUMBO_TOKEN_NULL:
|
4380
4378
|
parser_add_parse_error(parser, token);
|
4381
4379
|
token->v.character = kUtf8ReplacementChar;
|
4382
4380
|
insert_text_token(parser, token);
|
4383
|
-
return
|
4381
|
+
return;
|
4384
4382
|
case GUMBO_TOKEN_WHITESPACE:
|
4385
4383
|
insert_text_token(parser, token);
|
4386
|
-
return
|
4384
|
+
return;
|
4387
4385
|
case GUMBO_TOKEN_CDATA:
|
4388
4386
|
case GUMBO_TOKEN_CHARACTER:
|
4389
4387
|
insert_text_token(parser, token);
|
4390
4388
|
set_frameset_not_ok(parser);
|
4391
|
-
return
|
4389
|
+
return;
|
4392
4390
|
case GUMBO_TOKEN_COMMENT:
|
4393
4391
|
append_comment_node(parser, get_current_node(parser), token);
|
4394
|
-
return
|
4392
|
+
return;
|
4395
4393
|
case GUMBO_TOKEN_DOCTYPE:
|
4396
4394
|
parser_add_parse_error(parser, token);
|
4397
4395
|
ignore_token(parser);
|
4398
|
-
return
|
4396
|
+
return;
|
4399
4397
|
default:
|
4400
4398
|
// Fall through to the if-statements below.
|
4401
4399
|
break;
|
@@ -4439,7 +4437,7 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
|
4439
4437
|
)
|
4440
4438
|
);
|
4441
4439
|
parser->_parser_state->_reprocess_current_token = true;
|
4442
|
-
return
|
4440
|
+
return;
|
4443
4441
|
}
|
4444
4442
|
// This is a start tag so the next if's then branch will be taken.
|
4445
4443
|
}
|
@@ -4460,7 +4458,7 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
|
4460
4458
|
pop_current_node(parser);
|
4461
4459
|
acknowledge_self_closing_tag(parser);
|
4462
4460
|
}
|
4463
|
-
return
|
4461
|
+
return;
|
4464
4462
|
// </script> tags are handled like any other end tag, putting the script's
|
4465
4463
|
// text into a text node child and closing the current node.
|
4466
4464
|
}
|
@@ -4470,11 +4468,8 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
|
4470
4468
|
const char* name = token->v.end_tag.name;
|
4471
4469
|
assert(node != NULL);
|
4472
4470
|
|
4473
|
-
|
4474
|
-
if (!node_tagname_is(node, tag, name)) {
|
4471
|
+
if (!node_tagname_is(node, tag, name))
|
4475
4472
|
parser_add_parse_error(parser, token);
|
4476
|
-
is_success = false;
|
4477
|
-
}
|
4478
4473
|
int i = parser->_parser_state->_open_elements.length;
|
4479
4474
|
for (--i; i > 0;) {
|
4480
4475
|
// Here we move up the stack until we find an HTML element (in which
|
@@ -4489,7 +4484,7 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
|
4489
4484
|
// be an element on the stack of open elements (set below), so
|
4490
4485
|
// this loop is guaranteed to terminate.
|
4491
4486
|
}
|
4492
|
-
return
|
4487
|
+
return;
|
4493
4488
|
}
|
4494
4489
|
--i;
|
4495
4490
|
node = parser->_parser_state->_open_elements.data[i];
|
@@ -4500,22 +4495,22 @@ static bool handle_in_foreign_content(GumboParser* parser, GumboToken* token) {
|
|
4500
4495
|
}
|
4501
4496
|
assert(node->v.element.tag_namespace == GUMBO_NAMESPACE_HTML);
|
4502
4497
|
if (i == 0)
|
4503
|
-
return
|
4498
|
+
return;
|
4504
4499
|
// We can't call handle_token directly because the current node is still in
|
4505
4500
|
// a foriegn namespace, so it would re-enter this and result in infinite
|
4506
4501
|
// recursion.
|
4507
|
-
|
4502
|
+
handle_html_content(parser, token);
|
4508
4503
|
}
|
4509
4504
|
|
4510
4505
|
// https://html.spec.whatwg.org/multipage/parsing.html#tree-construction
|
4511
|
-
static
|
4506
|
+
static void handle_token(GumboParser* parser, GumboToken* token) {
|
4512
4507
|
if (
|
4513
4508
|
parser->_parser_state->_ignore_next_linefeed
|
4514
4509
|
&& token->type == GUMBO_TOKEN_WHITESPACE && token->v.character == '\n'
|
4515
4510
|
) {
|
4516
4511
|
parser->_parser_state->_ignore_next_linefeed = false;
|
4517
4512
|
ignore_token(parser);
|
4518
|
-
return
|
4513
|
+
return;
|
4519
4514
|
}
|
4520
4515
|
// This needs to be reset both here and in the conditional above to catch both
|
4521
4516
|
// the case where the next token is not whitespace (so we don't ignore
|
@@ -4557,9 +4552,9 @@ static bool handle_token(GumboParser* parser, GumboToken* token) {
|
|
4557
4552
|
token->type == GUMBO_TOKEN_NULL ||
|
4558
4553
|
token->type == GUMBO_TOKEN_WHITESPACE)) ||
|
4559
4554
|
token->type == GUMBO_TOKEN_EOF) {
|
4560
|
-
|
4555
|
+
handle_html_content(parser, token);
|
4561
4556
|
} else {
|
4562
|
-
|
4557
|
+
handle_in_foreign_content(parser, token);
|
4563
4558
|
}
|
4564
4559
|
}
|
4565
4560
|
|
@@ -4746,7 +4741,6 @@ GumboOutput* gumbo_parse_with_options (
|
|
4746
4741
|
|
4747
4742
|
const unsigned int max_tree_depth = options->max_tree_depth;
|
4748
4743
|
GumboToken token;
|
4749
|
-
bool has_error = false;
|
4750
4744
|
|
4751
4745
|
do {
|
4752
4746
|
if (state->_reprocess_current_token) {
|
@@ -4758,7 +4752,7 @@ GumboOutput* gumbo_parse_with_options (
|
|
4758
4752
|
adjusted_current_node &&
|
4759
4753
|
adjusted_current_node->v.element.tag_namespace != GUMBO_NAMESPACE_HTML
|
4760
4754
|
);
|
4761
|
-
|
4755
|
+
gumbo_lex(&parser, &token);
|
4762
4756
|
}
|
4763
4757
|
|
4764
4758
|
const char* token_type = "text";
|
@@ -4792,7 +4786,7 @@ GumboOutput* gumbo_parse_with_options (
|
|
4792
4786
|
state->_current_token = &token;
|
4793
4787
|
state->_self_closing_flag_acknowledged = false;
|
4794
4788
|
|
4795
|
-
|
4789
|
+
handle_token(&parser, &token);
|
4796
4790
|
|
4797
4791
|
// Check for memory leaks when ownership is transferred from start tag
|
4798
4792
|
// tokens to nodes.
|
@@ -4809,7 +4803,6 @@ GumboOutput* gumbo_parse_with_options (
|
|
4809
4803
|
if (token.type == GUMBO_TOKEN_START_TAG &&
|
4810
4804
|
token.v.start_tag.is_self_closing &&
|
4811
4805
|
!state->_self_closing_flag_acknowledged) {
|
4812
|
-
has_error = true;
|
4813
4806
|
GumboError* error = gumbo_add_error(&parser);
|
4814
4807
|
if (error) {
|
4815
4808
|
// This is essentially a tokenizer error that's only caught during
|
@@ -4837,7 +4830,7 @@ GumboOutput* gumbo_parse_with_options (
|
|
4837
4830
|
|
4838
4831
|
} while (
|
4839
4832
|
(token.type != GUMBO_TOKEN_EOF || state->_reprocess_current_token)
|
4840
|
-
&& !(options->stop_on_first_error &&
|
4833
|
+
&& !(options->stop_on_first_error && parser._output->document_error)
|
4841
4834
|
);
|
4842
4835
|
|
4843
4836
|
finish_parsing(&parser);
|