| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
/* |
|
2
|
|
|
|
|
|
|
Copyright (C) 2015-2017 Alexander Borisov |
|
3
|
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
This library is free software; you can redistribute it and/or |
|
5
|
|
|
|
|
|
|
modify it under the terms of the GNU Lesser General Public |
|
6
|
|
|
|
|
|
|
License as published by the Free Software Foundation; either |
|
7
|
|
|
|
|
|
|
version 2.1 of the License, or (at your option) any later version. |
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
This library is distributed in the hope that it will be useful, |
|
10
|
|
|
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11
|
|
|
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
12
|
|
|
|
|
|
|
Lesser General Public License for more details. |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
You should have received a copy of the GNU Lesser General Public |
|
15
|
|
|
|
|
|
|
License along with this library; if not, write to the Free Software |
|
16
|
|
|
|
|
|
|
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
17
|
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
Author: lex.borisov@gmail.com (Alexander Borisov) |
|
19
|
|
|
|
|
|
|
*/ |
|
20
|
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
#include "myhtml/tokenizer_end.h" |
|
22
|
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
|
|
24
|
144
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_data(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
25
|
|
|
|
|
|
|
{ |
|
26
|
144
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
27
|
|
|
|
|
|
|
|
|
28
|
144
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_size + tree->global_offset), MyHTML_TOKEN_TYPE_DATA); |
|
29
|
144
|
|
|
|
|
|
return html_offset; |
|
30
|
|
|
|
|
|
|
} |
|
31
|
|
|
|
|
|
|
|
|
32
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
33
|
|
|
|
|
|
|
{ |
|
34
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
35
|
|
|
|
|
|
|
|
|
36
|
0
|
0
|
|
|
|
|
if(token_node->raw_begin < (html_size + tree->global_offset)) { |
|
37
|
0
|
0
|
|
|
|
|
if(token_node->raw_begin) { |
|
38
|
0
|
|
|
|
|
|
token_node->raw_length = (html_offset + tree->global_offset) - token_node->raw_begin; |
|
39
|
0
|
|
|
|
|
|
myhtml_check_tag_parser(tree, token_node, html, html_offset); |
|
40
|
|
|
|
|
|
|
|
|
41
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
42
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
43
|
0
|
|
|
|
|
|
return 0; |
|
44
|
|
|
|
|
|
|
} |
|
45
|
|
|
|
|
|
|
} |
|
46
|
|
|
|
|
|
|
else { |
|
47
|
0
|
|
|
|
|
|
token_node->type ^= (token_node->type & MyHTML_TOKEN_TYPE_WHITESPACE); |
|
48
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_size + tree->global_offset), MyHTML_TOKEN_TYPE_DATA); |
|
49
|
|
|
|
|
|
|
} |
|
50
|
|
|
|
|
|
|
} |
|
51
|
|
|
|
|
|
|
|
|
52
|
0
|
|
|
|
|
|
return html_offset; |
|
53
|
|
|
|
|
|
|
} |
|
54
|
|
|
|
|
|
|
|
|
55
|
1
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_tag_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
56
|
|
|
|
|
|
|
{ |
|
57
|
1
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
58
|
|
|
|
|
|
|
|
|
59
|
1
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_size + tree->global_offset), MyHTML_TOKEN_TYPE_DATA); |
|
60
|
1
|
|
|
|
|
|
return html_offset; |
|
61
|
|
|
|
|
|
|
} |
|
62
|
|
|
|
|
|
|
|
|
63
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_end_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
64
|
|
|
|
|
|
|
{ |
|
65
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
66
|
|
|
|
|
|
|
|
|
67
|
0
|
0
|
|
|
|
|
if(token_node->raw_begin < (html_size + tree->global_offset)) |
|
68
|
|
|
|
|
|
|
{ |
|
69
|
0
|
|
|
|
|
|
token_node->raw_length = (html_offset + tree->global_offset) - token_node->raw_begin; |
|
70
|
0
|
|
|
|
|
|
token_node->type ^= (token_node->type & MyHTML_TOKEN_TYPE_WHITESPACE); |
|
71
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_size + tree->global_offset), MyHTML_TOKEN_TYPE_DATA); |
|
72
|
|
|
|
|
|
|
} |
|
73
|
|
|
|
|
|
|
|
|
74
|
0
|
|
|
|
|
|
return html_offset; |
|
75
|
|
|
|
|
|
|
} |
|
76
|
|
|
|
|
|
|
|
|
77
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_self_closing_start_tag(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
78
|
|
|
|
|
|
|
{ |
|
79
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
80
|
|
|
|
|
|
|
|
|
81
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_size + tree->global_offset), MyHTML_TOKEN_TYPE_DATA); |
|
82
|
0
|
|
|
|
|
|
return html_offset; |
|
83
|
|
|
|
|
|
|
} |
|
84
|
|
|
|
|
|
|
|
|
85
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_markup_declaration_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
86
|
|
|
|
|
|
|
{ |
|
87
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
88
|
|
|
|
|
|
|
|
|
89
|
0
|
0
|
|
|
|
|
if(token_node->raw_begin > 1) { |
|
90
|
0
|
|
|
|
|
|
tree->incoming_buf->length = myhtml_tokenizer_state_bogus_comment(tree, token_node, html, token_node->raw_begin, html_size); |
|
91
|
|
|
|
|
|
|
|
|
92
|
0
|
0
|
|
|
|
|
if(token_node != tree->current_token_node) |
|
93
|
|
|
|
|
|
|
{ |
|
94
|
0
|
|
|
|
|
|
token_node = tree->current_token_node; |
|
95
|
0
|
|
|
|
|
|
token_node->raw_length = (html_size + tree->global_offset) - token_node->raw_begin; |
|
96
|
|
|
|
|
|
|
|
|
97
|
0
|
0
|
|
|
|
|
if(token_node->raw_length) |
|
98
|
|
|
|
|
|
|
{ |
|
99
|
0
|
|
|
|
|
|
token_node->type ^= (token_node->type & MyHTML_TOKEN_TYPE_WHITESPACE); |
|
100
|
0
|
|
|
|
|
|
token_node->tag_id = MyHTML_TAG__TEXT; |
|
101
|
0
|
|
|
|
|
|
token_node->type |= MyHTML_TOKEN_TYPE_DATA; |
|
102
|
|
|
|
|
|
|
|
|
103
|
0
|
|
|
|
|
|
token_node->raw_length = (html_size + tree->global_offset) - token_node->raw_begin; |
|
104
|
|
|
|
|
|
|
|
|
105
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
106
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
107
|
0
|
|
|
|
|
|
return 0; |
|
108
|
|
|
|
|
|
|
} |
|
109
|
|
|
|
|
|
|
} |
|
110
|
|
|
|
|
|
|
} |
|
111
|
|
|
|
|
|
|
else { |
|
112
|
0
|
|
|
|
|
|
token_node->type ^= (token_node->type & MyHTML_TOKEN_TYPE_WHITESPACE); |
|
113
|
0
|
|
|
|
|
|
token_node->tag_id = MyHTML_TAG__COMMENT; |
|
114
|
0
|
|
|
|
|
|
token_node->type |= MyHTML_TOKEN_TYPE_COMMENT; |
|
115
|
|
|
|
|
|
|
|
|
116
|
0
|
|
|
|
|
|
token_node->raw_length = (html_size + tree->global_offset) - token_node->raw_begin; |
|
117
|
|
|
|
|
|
|
|
|
118
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
119
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
120
|
0
|
|
|
|
|
|
return 0; |
|
121
|
|
|
|
|
|
|
} |
|
122
|
|
|
|
|
|
|
} |
|
123
|
|
|
|
|
|
|
} |
|
124
|
|
|
|
|
|
|
|
|
125
|
0
|
|
|
|
|
|
return html_offset; |
|
126
|
|
|
|
|
|
|
} |
|
127
|
|
|
|
|
|
|
|
|
128
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_before_attribute_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
129
|
|
|
|
|
|
|
{ |
|
130
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
131
|
|
|
|
|
|
|
|
|
132
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
133
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
134
|
0
|
|
|
|
|
|
return 0; |
|
135
|
|
|
|
|
|
|
} |
|
136
|
|
|
|
|
|
|
|
|
137
|
0
|
|
|
|
|
|
return html_offset; |
|
138
|
|
|
|
|
|
|
} |
|
139
|
|
|
|
|
|
|
|
|
140
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_attribute_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
141
|
|
|
|
|
|
|
{ |
|
142
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
143
|
|
|
|
|
|
|
|
|
144
|
0
|
|
|
|
|
|
tree->attr_current->raw_key_length = (html_offset + tree->global_offset) - tree->attr_current->raw_key_begin; |
|
145
|
|
|
|
|
|
|
|
|
146
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
147
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
148
|
0
|
|
|
|
|
|
return 0; |
|
149
|
|
|
|
|
|
|
} |
|
150
|
|
|
|
|
|
|
|
|
151
|
0
|
|
|
|
|
|
return html_offset; |
|
152
|
|
|
|
|
|
|
} |
|
153
|
|
|
|
|
|
|
|
|
154
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_after_attribute_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
155
|
|
|
|
|
|
|
{ |
|
156
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
157
|
|
|
|
|
|
|
|
|
158
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
159
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
160
|
0
|
|
|
|
|
|
return 0; |
|
161
|
|
|
|
|
|
|
} |
|
162
|
|
|
|
|
|
|
|
|
163
|
0
|
|
|
|
|
|
return html_offset; |
|
164
|
|
|
|
|
|
|
} |
|
165
|
|
|
|
|
|
|
|
|
166
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_before_attribute_value(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
167
|
|
|
|
|
|
|
{ |
|
168
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
169
|
|
|
|
|
|
|
|
|
170
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
171
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
172
|
0
|
|
|
|
|
|
return 0; |
|
173
|
|
|
|
|
|
|
} |
|
174
|
|
|
|
|
|
|
|
|
175
|
0
|
|
|
|
|
|
tree->attr_current = myhtml_token_attr_create(tree->token, tree->token->mcasync_attr_id); |
|
176
|
0
|
0
|
|
|
|
|
if(tree->attr_current == NULL) { |
|
177
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
178
|
0
|
|
|
|
|
|
return 0; |
|
179
|
|
|
|
|
|
|
} |
|
180
|
|
|
|
|
|
|
|
|
181
|
0
|
|
|
|
|
|
return html_offset; |
|
182
|
|
|
|
|
|
|
} |
|
183
|
|
|
|
|
|
|
|
|
184
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_attribute_value_double_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
185
|
|
|
|
|
|
|
{ |
|
186
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
187
|
|
|
|
|
|
|
|
|
188
|
0
|
|
|
|
|
|
return html_offset; |
|
189
|
|
|
|
|
|
|
} |
|
190
|
|
|
|
|
|
|
|
|
191
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_attribute_value_single_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
192
|
|
|
|
|
|
|
{ |
|
193
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
194
|
|
|
|
|
|
|
|
|
195
|
0
|
|
|
|
|
|
return html_offset; |
|
196
|
|
|
|
|
|
|
} |
|
197
|
|
|
|
|
|
|
|
|
198
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_attribute_value_unquoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
199
|
|
|
|
|
|
|
{ |
|
200
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
201
|
|
|
|
|
|
|
|
|
202
|
0
|
|
|
|
|
|
tree->attr_current->raw_value_length = (html_offset + tree->global_offset) - tree->attr_current->raw_value_begin; |
|
203
|
|
|
|
|
|
|
|
|
204
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
205
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
206
|
0
|
|
|
|
|
|
return 0; |
|
207
|
|
|
|
|
|
|
} |
|
208
|
|
|
|
|
|
|
|
|
209
|
0
|
|
|
|
|
|
tree->attr_current = myhtml_token_attr_create(tree->token, tree->token->mcasync_attr_id); |
|
210
|
0
|
0
|
|
|
|
|
if(tree->attr_current == NULL) { |
|
211
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
212
|
0
|
|
|
|
|
|
return 0; |
|
213
|
|
|
|
|
|
|
} |
|
214
|
|
|
|
|
|
|
|
|
215
|
0
|
|
|
|
|
|
return html_offset; |
|
216
|
|
|
|
|
|
|
} |
|
217
|
|
|
|
|
|
|
|
|
218
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_after_attribute_value_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
219
|
|
|
|
|
|
|
{ |
|
220
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
221
|
|
|
|
|
|
|
|
|
222
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
223
|
|
|
|
|
|
|
|
|
224
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
225
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
226
|
0
|
|
|
|
|
|
return 0; |
|
227
|
|
|
|
|
|
|
} |
|
228
|
|
|
|
|
|
|
|
|
229
|
0
|
|
|
|
|
|
return html_offset; |
|
230
|
|
|
|
|
|
|
} |
|
231
|
|
|
|
|
|
|
|
|
232
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment_start(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
233
|
|
|
|
|
|
|
{ |
|
234
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
235
|
|
|
|
|
|
|
|
|
236
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
237
|
|
|
|
|
|
|
|
|
238
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
239
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
240
|
0
|
|
|
|
|
|
return 0; |
|
241
|
|
|
|
|
|
|
} |
|
242
|
|
|
|
|
|
|
|
|
243
|
0
|
|
|
|
|
|
return html_offset; |
|
244
|
|
|
|
|
|
|
} |
|
245
|
|
|
|
|
|
|
|
|
246
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment_start_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
247
|
|
|
|
|
|
|
{ |
|
248
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
249
|
|
|
|
|
|
|
|
|
250
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
251
|
|
|
|
|
|
|
|
|
252
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
253
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
254
|
0
|
|
|
|
|
|
return 0; |
|
255
|
|
|
|
|
|
|
} |
|
256
|
|
|
|
|
|
|
|
|
257
|
0
|
|
|
|
|
|
return html_offset; |
|
258
|
|
|
|
|
|
|
} |
|
259
|
|
|
|
|
|
|
|
|
260
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
261
|
|
|
|
|
|
|
{ |
|
262
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
263
|
|
|
|
|
|
|
|
|
264
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
265
|
|
|
|
|
|
|
|
|
266
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
267
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
268
|
0
|
|
|
|
|
|
return 0; |
|
269
|
|
|
|
|
|
|
} |
|
270
|
|
|
|
|
|
|
|
|
271
|
0
|
|
|
|
|
|
return html_offset; |
|
272
|
|
|
|
|
|
|
} |
|
273
|
|
|
|
|
|
|
|
|
274
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment_end(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
275
|
|
|
|
|
|
|
{ |
|
276
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
277
|
|
|
|
|
|
|
|
|
278
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
279
|
|
|
|
|
|
|
|
|
280
|
0
|
0
|
|
|
|
|
if(token_node->raw_length > 2) { |
|
281
|
0
|
|
|
|
|
|
token_node->raw_length -= 2; |
|
282
|
|
|
|
|
|
|
|
|
283
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
284
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
285
|
0
|
|
|
|
|
|
return 0; |
|
286
|
|
|
|
|
|
|
} |
|
287
|
|
|
|
|
|
|
} |
|
288
|
|
|
|
|
|
|
|
|
289
|
0
|
|
|
|
|
|
return html_offset; |
|
290
|
|
|
|
|
|
|
} |
|
291
|
|
|
|
|
|
|
|
|
292
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment_end_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
293
|
|
|
|
|
|
|
{ |
|
294
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
295
|
|
|
|
|
|
|
|
|
296
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
297
|
|
|
|
|
|
|
|
|
298
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
299
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
300
|
0
|
|
|
|
|
|
return 0; |
|
301
|
|
|
|
|
|
|
} |
|
302
|
|
|
|
|
|
|
|
|
303
|
0
|
|
|
|
|
|
return html_offset; |
|
304
|
|
|
|
|
|
|
} |
|
305
|
|
|
|
|
|
|
|
|
306
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_comment_end_bang(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
307
|
|
|
|
|
|
|
{ |
|
308
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
309
|
|
|
|
|
|
|
|
|
310
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
311
|
|
|
|
|
|
|
|
|
312
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
313
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
314
|
0
|
|
|
|
|
|
return 0; |
|
315
|
|
|
|
|
|
|
} |
|
316
|
|
|
|
|
|
|
|
|
317
|
0
|
|
|
|
|
|
return html_offset; |
|
318
|
|
|
|
|
|
|
} |
|
319
|
|
|
|
|
|
|
|
|
320
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_bogus_comment(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
321
|
|
|
|
|
|
|
{ |
|
322
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
323
|
|
|
|
|
|
|
|
|
324
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
325
|
|
|
|
|
|
|
|
|
326
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
327
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
328
|
0
|
|
|
|
|
|
return 0; |
|
329
|
|
|
|
|
|
|
} |
|
330
|
|
|
|
|
|
|
|
|
331
|
0
|
|
|
|
|
|
return html_offset; |
|
332
|
|
|
|
|
|
|
} |
|
333
|
|
|
|
|
|
|
|
|
334
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_cdata_section(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
335
|
|
|
|
|
|
|
{ |
|
336
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
337
|
|
|
|
|
|
|
|
|
338
|
0
|
|
|
|
|
|
token_node->raw_length = ((html_offset + tree->global_offset) - token_node->raw_begin); |
|
339
|
|
|
|
|
|
|
|
|
340
|
0
|
0
|
|
|
|
|
if(token_node->raw_length) { |
|
341
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
342
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
343
|
0
|
|
|
|
|
|
return 0; |
|
344
|
|
|
|
|
|
|
} |
|
345
|
|
|
|
|
|
|
} |
|
346
|
|
|
|
|
|
|
|
|
347
|
0
|
|
|
|
|
|
return html_offset; |
|
348
|
|
|
|
|
|
|
} |
|
349
|
|
|
|
|
|
|
|
|
350
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rcdata(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
351
|
|
|
|
|
|
|
{ |
|
352
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
353
|
|
|
|
|
|
|
|
|
354
|
0
|
0
|
|
|
|
|
if(token_node->raw_begin < (html_size + tree->global_offset)) { |
|
355
|
0
|
|
|
|
|
|
token_node->type |= MyHTML_TOKEN_TYPE_RCDATA; |
|
356
|
0
|
|
|
|
|
|
token_node->tag_id = MyHTML_TAG__TEXT; |
|
357
|
0
|
|
|
|
|
|
token_node->raw_length = (html_size + tree->global_offset) - token_node->raw_begin; |
|
358
|
|
|
|
|
|
|
|
|
359
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, 0, token_node) != MyHTML_STATUS_OK) { |
|
360
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
361
|
0
|
|
|
|
|
|
return 0; |
|
362
|
|
|
|
|
|
|
} |
|
363
|
|
|
|
|
|
|
} |
|
364
|
|
|
|
|
|
|
|
|
365
|
0
|
|
|
|
|
|
return html_offset; |
|
366
|
|
|
|
|
|
|
} |
|
367
|
|
|
|
|
|
|
|
|
368
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rcdata_less_than_sign(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
369
|
|
|
|
|
|
|
{ |
|
370
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
371
|
|
|
|
|
|
|
|
|
372
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RCDATA); |
|
373
|
0
|
|
|
|
|
|
return html_offset; |
|
374
|
|
|
|
|
|
|
} |
|
375
|
|
|
|
|
|
|
|
|
376
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rcdata_end_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
377
|
|
|
|
|
|
|
{ |
|
378
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
379
|
|
|
|
|
|
|
|
|
380
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RCDATA); |
|
381
|
0
|
|
|
|
|
|
return html_offset; |
|
382
|
|
|
|
|
|
|
} |
|
383
|
|
|
|
|
|
|
|
|
384
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rcdata_end_tag_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
385
|
|
|
|
|
|
|
{ |
|
386
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
387
|
|
|
|
|
|
|
|
|
388
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RCDATA); |
|
389
|
0
|
|
|
|
|
|
return html_offset; |
|
390
|
|
|
|
|
|
|
} |
|
391
|
|
|
|
|
|
|
|
|
392
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rawtext(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
393
|
|
|
|
|
|
|
{ |
|
394
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
395
|
|
|
|
|
|
|
|
|
396
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RAWTEXT); |
|
397
|
0
|
|
|
|
|
|
return html_offset; |
|
398
|
|
|
|
|
|
|
} |
|
399
|
|
|
|
|
|
|
|
|
400
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rawtext_less_than_sign(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
401
|
|
|
|
|
|
|
{ |
|
402
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
403
|
|
|
|
|
|
|
|
|
404
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RAWTEXT); |
|
405
|
0
|
|
|
|
|
|
return html_offset; |
|
406
|
|
|
|
|
|
|
} |
|
407
|
|
|
|
|
|
|
|
|
408
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rawtext_end_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
409
|
|
|
|
|
|
|
{ |
|
410
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
411
|
|
|
|
|
|
|
|
|
412
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RAWTEXT); |
|
413
|
0
|
|
|
|
|
|
return html_offset; |
|
414
|
|
|
|
|
|
|
} |
|
415
|
|
|
|
|
|
|
|
|
416
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_rawtext_end_tag_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
417
|
|
|
|
|
|
|
{ |
|
418
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
419
|
|
|
|
|
|
|
|
|
420
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RAWTEXT); |
|
421
|
0
|
|
|
|
|
|
return html_offset; |
|
422
|
|
|
|
|
|
|
} |
|
423
|
|
|
|
|
|
|
|
|
424
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_plaintext(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
425
|
|
|
|
|
|
|
{ |
|
426
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
427
|
|
|
|
|
|
|
|
|
428
|
|
|
|
|
|
|
// all we need inside myhtml_tokenizer_state_plaintext |
|
429
|
0
|
|
|
|
|
|
return html_offset; |
|
430
|
|
|
|
|
|
|
} |
|
431
|
|
|
|
|
|
|
|
|
432
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
433
|
|
|
|
|
|
|
{ |
|
434
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
435
|
|
|
|
|
|
|
|
|
436
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_RAWTEXT); |
|
437
|
0
|
|
|
|
|
|
return html_offset; |
|
438
|
|
|
|
|
|
|
} |
|
439
|
|
|
|
|
|
|
|
|
440
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_before_doctype_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
441
|
|
|
|
|
|
|
{ |
|
442
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
443
|
|
|
|
|
|
|
|
|
444
|
0
|
|
|
|
|
|
tree->compat_mode = MyHTML_TREE_COMPAT_MODE_QUIRKS; |
|
445
|
|
|
|
|
|
|
|
|
446
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
447
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
448
|
0
|
|
|
|
|
|
return 0; |
|
449
|
|
|
|
|
|
|
} |
|
450
|
|
|
|
|
|
|
|
|
451
|
0
|
|
|
|
|
|
return html_offset; |
|
452
|
|
|
|
|
|
|
} |
|
453
|
|
|
|
|
|
|
|
|
454
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
455
|
|
|
|
|
|
|
{ |
|
456
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
457
|
|
|
|
|
|
|
|
|
458
|
0
|
|
|
|
|
|
tree->attr_current->raw_key_length = (html_offset + tree->global_offset) - tree->attr_current->raw_key_begin; |
|
459
|
|
|
|
|
|
|
|
|
460
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
461
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
462
|
0
|
|
|
|
|
|
return 0; |
|
463
|
|
|
|
|
|
|
} |
|
464
|
|
|
|
|
|
|
|
|
465
|
0
|
|
|
|
|
|
return html_offset; |
|
466
|
|
|
|
|
|
|
} |
|
467
|
|
|
|
|
|
|
|
|
468
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_after_doctype_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
469
|
|
|
|
|
|
|
{ |
|
470
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
471
|
|
|
|
|
|
|
|
|
472
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
473
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
474
|
0
|
|
|
|
|
|
return 0; |
|
475
|
|
|
|
|
|
|
} |
|
476
|
|
|
|
|
|
|
|
|
477
|
0
|
|
|
|
|
|
return html_offset; |
|
478
|
|
|
|
|
|
|
} |
|
479
|
|
|
|
|
|
|
|
|
480
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_custom_after_doctype_name_a_z(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
481
|
|
|
|
|
|
|
{ |
|
482
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
483
|
|
|
|
|
|
|
|
|
484
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
485
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
486
|
0
|
|
|
|
|
|
return 0; |
|
487
|
|
|
|
|
|
|
} |
|
488
|
|
|
|
|
|
|
|
|
489
|
0
|
|
|
|
|
|
return html_offset; |
|
490
|
|
|
|
|
|
|
} |
|
491
|
|
|
|
|
|
|
|
|
492
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_before_doctype_public_identifier(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
493
|
|
|
|
|
|
|
{ |
|
494
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
495
|
|
|
|
|
|
|
|
|
496
|
0
|
|
|
|
|
|
tree->compat_mode = MyHTML_TREE_COMPAT_MODE_QUIRKS; |
|
497
|
|
|
|
|
|
|
|
|
498
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
499
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
500
|
0
|
|
|
|
|
|
return 0; |
|
501
|
|
|
|
|
|
|
} |
|
502
|
|
|
|
|
|
|
|
|
503
|
0
|
|
|
|
|
|
return html_offset; |
|
504
|
|
|
|
|
|
|
} |
|
505
|
|
|
|
|
|
|
|
|
506
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype_public_identifier_double_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
507
|
|
|
|
|
|
|
{ |
|
508
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
509
|
|
|
|
|
|
|
|
|
510
|
0
|
|
|
|
|
|
tree->compat_mode = MyHTML_TREE_COMPAT_MODE_QUIRKS; |
|
511
|
|
|
|
|
|
|
|
|
512
|
0
|
0
|
|
|
|
|
if(tree->attr_current->raw_key_begin && html_size) { |
|
|
|
0
|
|
|
|
|
|
|
513
|
0
|
|
|
|
|
|
tree->attr_current->raw_key_length = (html_offset + tree->global_offset) - tree->attr_current->raw_key_begin; |
|
514
|
|
|
|
|
|
|
} |
|
515
|
|
|
|
|
|
|
|
|
516
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
517
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
518
|
0
|
|
|
|
|
|
return 0; |
|
519
|
|
|
|
|
|
|
} |
|
520
|
|
|
|
|
|
|
|
|
521
|
0
|
|
|
|
|
|
return html_offset; |
|
522
|
|
|
|
|
|
|
} |
|
523
|
|
|
|
|
|
|
|
|
524
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype_public_identifier_single_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
525
|
|
|
|
|
|
|
{ |
|
526
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
527
|
|
|
|
|
|
|
|
|
528
|
0
|
|
|
|
|
|
myhtml_tokenizer_end_state_doctype_public_identifier_double_quoted(tree, token_node, html, (html_offset + tree->global_offset), html_size); |
|
529
|
0
|
|
|
|
|
|
return html_offset; |
|
530
|
|
|
|
|
|
|
} |
|
531
|
|
|
|
|
|
|
|
|
532
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_after_doctype_public_identifier(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
533
|
|
|
|
|
|
|
{ |
|
534
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
535
|
|
|
|
|
|
|
|
|
536
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
537
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
538
|
0
|
|
|
|
|
|
return 0; |
|
539
|
|
|
|
|
|
|
} |
|
540
|
|
|
|
|
|
|
|
|
541
|
0
|
|
|
|
|
|
return html_offset; |
|
542
|
|
|
|
|
|
|
} |
|
543
|
|
|
|
|
|
|
|
|
544
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype_system_identifier_double_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
545
|
|
|
|
|
|
|
{ |
|
546
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
547
|
|
|
|
|
|
|
|
|
548
|
0
|
|
|
|
|
|
tree->compat_mode = MyHTML_TREE_COMPAT_MODE_QUIRKS; |
|
549
|
|
|
|
|
|
|
|
|
550
|
0
|
0
|
|
|
|
|
if(tree->attr_current->raw_key_begin && html_size) { |
|
|
|
0
|
|
|
|
|
|
|
551
|
0
|
|
|
|
|
|
tree->attr_current->raw_key_length = (html_offset + tree->global_offset) - tree->attr_current->raw_key_begin; |
|
552
|
|
|
|
|
|
|
} |
|
553
|
|
|
|
|
|
|
|
|
554
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
555
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
556
|
0
|
|
|
|
|
|
return 0; |
|
557
|
|
|
|
|
|
|
} |
|
558
|
|
|
|
|
|
|
|
|
559
|
0
|
|
|
|
|
|
return html_offset; |
|
560
|
|
|
|
|
|
|
} |
|
561
|
|
|
|
|
|
|
|
|
562
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_doctype_system_identifier_single_quoted(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
563
|
|
|
|
|
|
|
{ |
|
564
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
565
|
|
|
|
|
|
|
|
|
566
|
0
|
|
|
|
|
|
myhtml_tokenizer_end_state_doctype_system_identifier_double_quoted(tree, token_node, html, (html_offset + tree->global_offset), html_size); |
|
567
|
0
|
|
|
|
|
|
return html_offset; |
|
568
|
|
|
|
|
|
|
} |
|
569
|
|
|
|
|
|
|
|
|
570
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_after_doctype_system_identifier(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
571
|
|
|
|
|
|
|
{ |
|
572
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
573
|
|
|
|
|
|
|
|
|
574
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
575
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
576
|
0
|
|
|
|
|
|
return 0; |
|
577
|
|
|
|
|
|
|
} |
|
578
|
|
|
|
|
|
|
|
|
579
|
0
|
|
|
|
|
|
return html_offset; |
|
580
|
|
|
|
|
|
|
} |
|
581
|
|
|
|
|
|
|
|
|
582
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_bogus_doctype(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
583
|
|
|
|
|
|
|
{ |
|
584
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
585
|
|
|
|
|
|
|
|
|
586
|
0
|
0
|
|
|
|
|
if(myhtml_queue_add(tree, html_offset, token_node) != MyHTML_STATUS_OK) { |
|
587
|
0
|
|
|
|
|
|
myhtml_tokenizer_state_set(tree) = MyHTML_TOKENIZER_STATE_PARSE_ERROR_STOP; |
|
588
|
0
|
|
|
|
|
|
return 0; |
|
589
|
|
|
|
|
|
|
} |
|
590
|
|
|
|
|
|
|
|
|
591
|
0
|
|
|
|
|
|
return html_offset; |
|
592
|
|
|
|
|
|
|
} |
|
593
|
|
|
|
|
|
|
|
|
594
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
595
|
|
|
|
|
|
|
{ |
|
596
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
597
|
|
|
|
|
|
|
|
|
598
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
599
|
0
|
|
|
|
|
|
return html_offset; |
|
600
|
|
|
|
|
|
|
} |
|
601
|
|
|
|
|
|
|
|
|
602
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_less_than_sign(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
603
|
|
|
|
|
|
|
{ |
|
604
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
605
|
|
|
|
|
|
|
|
|
606
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
607
|
0
|
|
|
|
|
|
return html_offset; |
|
608
|
|
|
|
|
|
|
} |
|
609
|
|
|
|
|
|
|
|
|
610
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_end_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
611
|
|
|
|
|
|
|
{ |
|
612
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
613
|
|
|
|
|
|
|
|
|
614
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
615
|
0
|
|
|
|
|
|
return html_offset; |
|
616
|
|
|
|
|
|
|
} |
|
617
|
|
|
|
|
|
|
|
|
618
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_end_tag_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
619
|
|
|
|
|
|
|
{ |
|
620
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
621
|
|
|
|
|
|
|
|
|
622
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
623
|
0
|
|
|
|
|
|
return html_offset; |
|
624
|
|
|
|
|
|
|
} |
|
625
|
|
|
|
|
|
|
|
|
626
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escape_start(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
627
|
|
|
|
|
|
|
{ |
|
628
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
629
|
|
|
|
|
|
|
|
|
630
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
631
|
0
|
|
|
|
|
|
return html_offset; |
|
632
|
|
|
|
|
|
|
} |
|
633
|
|
|
|
|
|
|
|
|
634
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escape_start_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
635
|
|
|
|
|
|
|
{ |
|
636
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
637
|
|
|
|
|
|
|
|
|
638
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
639
|
0
|
|
|
|
|
|
return html_offset; |
|
640
|
|
|
|
|
|
|
} |
|
641
|
|
|
|
|
|
|
|
|
642
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
643
|
|
|
|
|
|
|
{ |
|
644
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
645
|
|
|
|
|
|
|
|
|
646
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
647
|
0
|
|
|
|
|
|
return html_offset; |
|
648
|
|
|
|
|
|
|
} |
|
649
|
|
|
|
|
|
|
|
|
650
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
651
|
|
|
|
|
|
|
{ |
|
652
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
653
|
|
|
|
|
|
|
|
|
654
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
655
|
0
|
|
|
|
|
|
return html_offset; |
|
656
|
|
|
|
|
|
|
} |
|
657
|
|
|
|
|
|
|
|
|
658
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped_dash_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
659
|
|
|
|
|
|
|
{ |
|
660
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
661
|
|
|
|
|
|
|
|
|
662
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
663
|
0
|
|
|
|
|
|
return html_offset; |
|
664
|
|
|
|
|
|
|
} |
|
665
|
|
|
|
|
|
|
|
|
666
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped_less_than_sign(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
667
|
|
|
|
|
|
|
{ |
|
668
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
669
|
|
|
|
|
|
|
|
|
670
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
671
|
0
|
|
|
|
|
|
return html_offset; |
|
672
|
|
|
|
|
|
|
} |
|
673
|
|
|
|
|
|
|
|
|
674
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped_end_tag_open(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
675
|
|
|
|
|
|
|
{ |
|
676
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
677
|
|
|
|
|
|
|
|
|
678
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
679
|
0
|
|
|
|
|
|
return html_offset; |
|
680
|
|
|
|
|
|
|
} |
|
681
|
|
|
|
|
|
|
|
|
682
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_escaped_end_tag_name(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
683
|
|
|
|
|
|
|
{ |
|
684
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
685
|
|
|
|
|
|
|
|
|
686
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
687
|
0
|
|
|
|
|
|
return html_offset; |
|
688
|
|
|
|
|
|
|
} |
|
689
|
|
|
|
|
|
|
|
|
690
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escape_start(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
691
|
|
|
|
|
|
|
{ |
|
692
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
693
|
|
|
|
|
|
|
|
|
694
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
695
|
0
|
|
|
|
|
|
return html_offset; |
|
696
|
|
|
|
|
|
|
} |
|
697
|
|
|
|
|
|
|
|
|
698
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escaped(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
699
|
|
|
|
|
|
|
{ |
|
700
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
701
|
|
|
|
|
|
|
|
|
702
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
703
|
0
|
|
|
|
|
|
return html_offset; |
|
704
|
|
|
|
|
|
|
} |
|
705
|
|
|
|
|
|
|
|
|
706
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escaped_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
707
|
|
|
|
|
|
|
{ |
|
708
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
709
|
|
|
|
|
|
|
|
|
710
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
711
|
0
|
|
|
|
|
|
return html_offset; |
|
712
|
|
|
|
|
|
|
} |
|
713
|
|
|
|
|
|
|
|
|
714
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escaped_dash_dash(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
715
|
|
|
|
|
|
|
{ |
|
716
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
717
|
|
|
|
|
|
|
|
|
718
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
719
|
0
|
|
|
|
|
|
return html_offset; |
|
720
|
|
|
|
|
|
|
} |
|
721
|
|
|
|
|
|
|
|
|
722
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escaped_less_than_sign(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
723
|
|
|
|
|
|
|
{ |
|
724
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
725
|
|
|
|
|
|
|
|
|
726
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
727
|
0
|
|
|
|
|
|
return html_offset; |
|
728
|
|
|
|
|
|
|
} |
|
729
|
|
|
|
|
|
|
|
|
730
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_script_data_double_escape_end(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
731
|
|
|
|
|
|
|
{ |
|
732
|
0
|
|
|
|
|
|
token_node->element_length = (tree->global_offset + html_size) - token_node->element_begin; |
|
733
|
|
|
|
|
|
|
|
|
734
|
0
|
|
|
|
|
|
myhtml_tokenizer_queue_create_text_node_if_need(tree, token_node, html, (html_offset + tree->global_offset), MyHTML_TOKEN_TYPE_SCRIPT); |
|
735
|
0
|
|
|
|
|
|
return html_offset; |
|
736
|
|
|
|
|
|
|
} |
|
737
|
|
|
|
|
|
|
|
|
738
|
0
|
|
|
|
|
|
size_t myhtml_tokenizer_end_state_parse_error_stop(myhtml_tree_t* tree, myhtml_token_node_t* token_node, const char* html, size_t html_offset, size_t html_size) |
|
739
|
|
|
|
|
|
|
{ |
|
740
|
|
|
|
|
|
|
|
|
741
|
|
|
|
|
|
|
|
|
742
|
0
|
|
|
|
|
|
return html_size; |
|
743
|
|
|
|
|
|
|
} |