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/token.h" |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
// all key size == value size |
24
|
|
|
|
|
|
|
static const myhtml_token_replacement_entry_t myhtml_token_attr_svg_replacement[] = { |
25
|
|
|
|
|
|
|
{"attributename", 13, "attributeName", 13}, |
26
|
|
|
|
|
|
|
{"attributetype", 13, "attributeType", 13}, |
27
|
|
|
|
|
|
|
{"basefrequency", 13, "baseFrequency", 13}, |
28
|
|
|
|
|
|
|
{"baseprofile", 11, "baseProfile", 11}, |
29
|
|
|
|
|
|
|
{"calcmode", 8, "calcMode", 8}, |
30
|
|
|
|
|
|
|
{"clippathunits", 13, "clipPathUnits", 13}, |
31
|
|
|
|
|
|
|
{"diffuseconstant", 15, "diffuseConstant", 15}, |
32
|
|
|
|
|
|
|
{"edgemode", 8, "edgeMode", 8}, |
33
|
|
|
|
|
|
|
{"filterunits", 11, "filterUnits", 11}, |
34
|
|
|
|
|
|
|
{"glyphref", 8, "glyphRef", 8}, |
35
|
|
|
|
|
|
|
{"gradienttransform", 17, "gradientTransform", 17}, |
36
|
|
|
|
|
|
|
{"gradientunits", 13, "gradientUnits", 13}, |
37
|
|
|
|
|
|
|
{"kernelmatrix", 12, "kernelMatrix", 12}, |
38
|
|
|
|
|
|
|
{"kernelunitlength", 16, "kernelUnitLength", 16}, |
39
|
|
|
|
|
|
|
{"keypoints", 9, "keyPoints", 9}, |
40
|
|
|
|
|
|
|
{"keysplines", 10, "keySplines", 10}, |
41
|
|
|
|
|
|
|
{"keytimes", 8, "keyTimes", 8}, |
42
|
|
|
|
|
|
|
{"lengthadjust", 12, "lengthAdjust", 12}, |
43
|
|
|
|
|
|
|
{"limitingconeangle", 17, "limitingConeAngle", 17}, |
44
|
|
|
|
|
|
|
{"markerheight", 12, "markerHeight", 12}, |
45
|
|
|
|
|
|
|
{"markerunits", 11, "markerUnits", 11}, |
46
|
|
|
|
|
|
|
{"markerwidth", 11, "markerWidth", 11}, |
47
|
|
|
|
|
|
|
{"maskcontentunits", 16, "maskContentUnits", 16}, |
48
|
|
|
|
|
|
|
{"maskunits", 9, "maskUnits", 9}, |
49
|
|
|
|
|
|
|
{"numoctaves", 10, "numOctaves", 10}, |
50
|
|
|
|
|
|
|
{"pathlength", 10, "pathLength", 10}, |
51
|
|
|
|
|
|
|
{"patterncontentunits", 19, "patternContentUnits", 19}, |
52
|
|
|
|
|
|
|
{"patterntransform", 16, "patternTransform", 16}, |
53
|
|
|
|
|
|
|
{"patternunits", 12, "patternUnits", 12}, |
54
|
|
|
|
|
|
|
{"pointsatx", 9, "pointsAtX", 9}, |
55
|
|
|
|
|
|
|
{"pointsaty", 9, "pointsAtY", 9}, |
56
|
|
|
|
|
|
|
{"pointsatz", 9, "pointsAtZ", 9}, |
57
|
|
|
|
|
|
|
{"preservealpha", 13, "preserveAlpha", 13}, |
58
|
|
|
|
|
|
|
{"preserveaspectratio", 19, "preserveAspectRatio", 19}, |
59
|
|
|
|
|
|
|
{"primitiveunits", 14, "primitiveUnits", 14}, |
60
|
|
|
|
|
|
|
{"refx", 4, "refX", 4}, |
61
|
|
|
|
|
|
|
{"refy", 4, "refY", 4}, |
62
|
|
|
|
|
|
|
{"repeatcount", 11, "repeatCount", 11}, |
63
|
|
|
|
|
|
|
{"repeatdur", 9, "repeatDur", 9}, |
64
|
|
|
|
|
|
|
{"requiredextensions", 18, "requiredExtensions", 18}, |
65
|
|
|
|
|
|
|
{"requiredfeatures", 16, "requiredFeatures", 16}, |
66
|
|
|
|
|
|
|
{"specularconstant", 16, "specularConstant", 16}, |
67
|
|
|
|
|
|
|
{"specularexponent", 16, "specularExponent", 16}, |
68
|
|
|
|
|
|
|
{"spreadmethod", 12, "spreadMethod", 12}, |
69
|
|
|
|
|
|
|
{"startoffset", 11, "startOffset", 11}, |
70
|
|
|
|
|
|
|
{"stddeviation", 12, "stdDeviation", 12}, |
71
|
|
|
|
|
|
|
{"stitchtiles", 11, "stitchTiles", 11}, |
72
|
|
|
|
|
|
|
{"surfacescale", 12, "surfaceScale", 12}, |
73
|
|
|
|
|
|
|
{"systemlanguage", 14, "systemLanguage", 14}, |
74
|
|
|
|
|
|
|
{"tablevalues", 11, "tableValues", 11}, |
75
|
|
|
|
|
|
|
{"targetx", 7, "targetX", 7}, |
76
|
|
|
|
|
|
|
{"targety", 7, "targetY", 7}, |
77
|
|
|
|
|
|
|
{"textlength", 10, "textLength", 10}, |
78
|
|
|
|
|
|
|
{"viewbox", 7, "viewBox", 7}, |
79
|
|
|
|
|
|
|
{"viewtarget", 10, "viewTarget", 10}, |
80
|
|
|
|
|
|
|
{"xchannelselector", 16, "xChannelSelector", 16}, |
81
|
|
|
|
|
|
|
{"ychannelselector", 16, "yChannelSelector", 16}, |
82
|
|
|
|
|
|
|
{"zoomandpan", 10, "zoomAndPan", 10} |
83
|
|
|
|
|
|
|
}; |
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
// all key size > value size |
86
|
|
|
|
|
|
|
static const myhtml_token_namespace_replacement_t myhtml_token_attr_namespace_replacement[] = { |
87
|
|
|
|
|
|
|
{"xlink:actuate", 13, "actuate", 7, MyHTML_NAMESPACE_XLINK}, |
88
|
|
|
|
|
|
|
{"xlink:arcrole", 13, "arcrole", 7, MyHTML_NAMESPACE_XLINK}, |
89
|
|
|
|
|
|
|
{"xlink:href", 10, "href", 4, MyHTML_NAMESPACE_XLINK}, |
90
|
|
|
|
|
|
|
{"xlink:role", 10, "role", 4, MyHTML_NAMESPACE_XLINK}, |
91
|
|
|
|
|
|
|
{"xlink:show", 10, "show", 4, MyHTML_NAMESPACE_XLINK}, |
92
|
|
|
|
|
|
|
{"xlink:title", 11, "title", 5, MyHTML_NAMESPACE_XLINK}, |
93
|
|
|
|
|
|
|
{"xlink:type", 10, "type", 4, MyHTML_NAMESPACE_XLINK}, |
94
|
|
|
|
|
|
|
{"xml:lang", 8, "lang", 4, MyHTML_NAMESPACE_XML}, |
95
|
|
|
|
|
|
|
{"xml:space", 9, "space", 5, MyHTML_NAMESPACE_XML}, |
96
|
|
|
|
|
|
|
{"xmlns", 5, "xmlns", 5, MyHTML_NAMESPACE_XMLNS}, |
97
|
|
|
|
|
|
|
{"xmlns:xlink", 11, "xlink", 5, MyHTML_NAMESPACE_XMLNS} |
98
|
|
|
|
|
|
|
}; |
99
|
|
|
|
|
|
|
|
100
|
145
|
|
|
|
|
|
myhtml_token_t * myhtml_token_create(myhtml_tree_t* tree, size_t size) |
101
|
|
|
|
|
|
|
{ |
102
|
145
|
50
|
|
|
|
|
if(size == 0) |
103
|
0
|
|
|
|
|
|
size = 4096; |
104
|
|
|
|
|
|
|
|
105
|
145
|
|
|
|
|
|
myhtml_token_t* token = (myhtml_token_t*)mycore_malloc(sizeof(myhtml_token_t)); |
106
|
|
|
|
|
|
|
|
107
|
145
|
50
|
|
|
|
|
if(token == NULL) |
108
|
0
|
|
|
|
|
|
return NULL; |
109
|
|
|
|
|
|
|
|
110
|
145
|
|
|
|
|
|
token->nodes_obj = mcobject_async_create(); |
111
|
|
|
|
|
|
|
|
112
|
145
|
50
|
|
|
|
|
if(token->nodes_obj == NULL) { |
113
|
0
|
|
|
|
|
|
mycore_free(token); |
114
|
0
|
|
|
|
|
|
return NULL; |
115
|
|
|
|
|
|
|
} |
116
|
|
|
|
|
|
|
|
117
|
145
|
|
|
|
|
|
token->attr_obj = mcobject_async_create(); |
118
|
|
|
|
|
|
|
|
119
|
145
|
50
|
|
|
|
|
if(token->attr_obj == NULL) { |
120
|
0
|
|
|
|
|
|
mycore_free(token->nodes_obj); |
121
|
0
|
|
|
|
|
|
mycore_free(token); |
122
|
|
|
|
|
|
|
|
123
|
0
|
|
|
|
|
|
return NULL; |
124
|
|
|
|
|
|
|
} |
125
|
|
|
|
|
|
|
|
126
|
145
|
|
|
|
|
|
mcobject_async_init(token->nodes_obj, 128, size, sizeof(myhtml_token_node_t)); |
127
|
145
|
|
|
|
|
|
mcobject_async_init(token->attr_obj, 128, size, sizeof(myhtml_token_attr_t)); |
128
|
|
|
|
|
|
|
|
129
|
145
|
|
|
|
|
|
token->mcasync_token_id = mcobject_async_node_add(token->nodes_obj, NULL); |
130
|
145
|
|
|
|
|
|
token->mcasync_attr_id = mcobject_async_node_add(token->attr_obj, NULL); |
131
|
|
|
|
|
|
|
|
132
|
145
|
|
|
|
|
|
token->tree = tree; |
133
|
|
|
|
|
|
|
|
134
|
145
|
|
|
|
|
|
return token; |
135
|
|
|
|
|
|
|
} |
136
|
|
|
|
|
|
|
|
137
|
145
|
|
|
|
|
|
void myhtml_token_clean(myhtml_token_t* token) |
138
|
|
|
|
|
|
|
{ |
139
|
145
|
|
|
|
|
|
mcobject_async_node_clean(token->nodes_obj, token->mcasync_token_id); |
140
|
145
|
|
|
|
|
|
mcobject_async_node_clean(token->attr_obj, token->mcasync_attr_id); |
141
|
145
|
|
|
|
|
|
} |
142
|
|
|
|
|
|
|
|
143
|
0
|
|
|
|
|
|
void myhtml_token_clean_all(myhtml_token_t* token) |
144
|
|
|
|
|
|
|
{ |
145
|
0
|
|
|
|
|
|
mcobject_async_clean(token->nodes_obj); |
146
|
0
|
|
|
|
|
|
mcobject_async_clean(token->attr_obj); |
147
|
0
|
|
|
|
|
|
} |
148
|
|
|
|
|
|
|
|
149
|
144
|
|
|
|
|
|
myhtml_token_t * myhtml_token_destroy(myhtml_token_t* token) |
150
|
|
|
|
|
|
|
{ |
151
|
144
|
50
|
|
|
|
|
if(token == NULL) |
152
|
0
|
|
|
|
|
|
return NULL; |
153
|
|
|
|
|
|
|
|
154
|
144
|
50
|
|
|
|
|
if(token->nodes_obj) |
155
|
144
|
|
|
|
|
|
token->nodes_obj = mcobject_async_destroy(token->nodes_obj, 1); |
156
|
|
|
|
|
|
|
|
157
|
144
|
50
|
|
|
|
|
if(token->attr_obj) |
158
|
144
|
|
|
|
|
|
token->attr_obj = mcobject_async_destroy(token->attr_obj, 1); |
159
|
|
|
|
|
|
|
|
160
|
144
|
|
|
|
|
|
mycore_free(token); |
161
|
|
|
|
|
|
|
|
162
|
144
|
|
|
|
|
|
return NULL; |
163
|
|
|
|
|
|
|
} |
164
|
|
|
|
|
|
|
|
165
|
2016
|
|
|
|
|
|
myhtml_token_node_t * myhtml_token_node_create(myhtml_token_t* token, size_t async_node_id) |
166
|
|
|
|
|
|
|
{ |
167
|
2016
|
|
|
|
|
|
myhtml_token_node_t *token_node = (myhtml_token_node_t*)mcobject_async_malloc(token->nodes_obj, async_node_id, NULL); |
168
|
2016
|
50
|
|
|
|
|
if(token_node == NULL) |
169
|
0
|
|
|
|
|
|
return NULL; |
170
|
|
|
|
|
|
|
|
171
|
2016
|
|
|
|
|
|
myhtml_token_node_clean(token_node); |
172
|
2016
|
|
|
|
|
|
return token_node; |
173
|
|
|
|
|
|
|
} |
174
|
|
|
|
|
|
|
|
175
|
2016
|
|
|
|
|
|
void myhtml_token_node_clean(myhtml_token_node_t* node) |
176
|
|
|
|
|
|
|
{ |
177
|
2016
|
|
|
|
|
|
memset(node, 0, sizeof(myhtml_token_node_t)); |
178
|
2016
|
|
|
|
|
|
node->type = MyHTML_TOKEN_TYPE_OPEN|MyHTML_TOKEN_TYPE_WHITESPACE; |
179
|
|
|
|
|
|
|
|
180
|
2016
|
|
|
|
|
|
mycore_string_clean_all(&node->str); |
181
|
2016
|
|
|
|
|
|
} |
182
|
|
|
|
|
|
|
|
183
|
386
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_create(myhtml_token_t* token, size_t async_node_id) |
184
|
|
|
|
|
|
|
{ |
185
|
386
|
|
|
|
|
|
myhtml_token_attr_t *attr_node = mcobject_async_malloc(token->attr_obj, async_node_id, NULL); |
186
|
386
|
50
|
|
|
|
|
if(attr_node == NULL) |
187
|
0
|
|
|
|
|
|
return NULL; |
188
|
|
|
|
|
|
|
|
189
|
386
|
|
|
|
|
|
myhtml_token_attr_clean(attr_node); |
190
|
386
|
|
|
|
|
|
return attr_node; |
191
|
|
|
|
|
|
|
} |
192
|
|
|
|
|
|
|
|
193
|
386
|
|
|
|
|
|
void myhtml_token_attr_clean(myhtml_token_attr_t* attr) |
194
|
|
|
|
|
|
|
{ |
195
|
386
|
|
|
|
|
|
memset(attr, 0, sizeof(myhtml_token_attr_t)); |
196
|
386
|
|
|
|
|
|
attr->ns = MyHTML_NAMESPACE_HTML; |
197
|
|
|
|
|
|
|
|
198
|
386
|
|
|
|
|
|
mycore_string_clean_all(&attr->key); |
199
|
386
|
|
|
|
|
|
mycore_string_clean_all(&attr->value); |
200
|
386
|
|
|
|
|
|
} |
201
|
|
|
|
|
|
|
|
202
|
0
|
|
|
|
|
|
myhtml_tag_id_t myhtml_token_node_tag_id(myhtml_token_node_t *token_node) |
203
|
|
|
|
|
|
|
{ |
204
|
0
|
|
|
|
|
|
return token_node->tag_id; |
205
|
|
|
|
|
|
|
} |
206
|
|
|
|
|
|
|
|
207
|
0
|
|
|
|
|
|
myhtml_position_t myhtml_token_node_raw_position(myhtml_token_node_t *token_node) |
208
|
|
|
|
|
|
|
{ |
209
|
0
|
0
|
|
|
|
|
if(token_node) |
210
|
0
|
|
|
|
|
|
return (myhtml_position_t){token_node->raw_begin, token_node->raw_length}; |
211
|
|
|
|
|
|
|
|
212
|
0
|
|
|
|
|
|
return (myhtml_position_t){0, 0}; |
213
|
|
|
|
|
|
|
} |
214
|
|
|
|
|
|
|
|
215
|
0
|
|
|
|
|
|
myhtml_position_t myhtml_token_node_element_position(myhtml_token_node_t *token_node) |
216
|
|
|
|
|
|
|
{ |
217
|
0
|
0
|
|
|
|
|
if(token_node) |
218
|
0
|
|
|
|
|
|
return (myhtml_position_t){token_node->element_begin, token_node->element_length}; |
219
|
|
|
|
|
|
|
|
220
|
0
|
|
|
|
|
|
return (myhtml_position_t){0, 0}; |
221
|
|
|
|
|
|
|
} |
222
|
|
|
|
|
|
|
|
223
|
0
|
|
|
|
|
|
myhtml_tree_attr_t * myhtml_token_node_attribute_first(myhtml_token_node_t *token_node) |
224
|
|
|
|
|
|
|
{ |
225
|
0
|
|
|
|
|
|
return token_node->attr_first; |
226
|
|
|
|
|
|
|
} |
227
|
|
|
|
|
|
|
|
228
|
0
|
|
|
|
|
|
myhtml_tree_attr_t * myhtml_token_node_attribute_last(myhtml_token_node_t *token_node) |
229
|
|
|
|
|
|
|
{ |
230
|
0
|
|
|
|
|
|
return token_node->attr_last; |
231
|
|
|
|
|
|
|
} |
232
|
|
|
|
|
|
|
|
233
|
0
|
|
|
|
|
|
const char * myhtml_token_node_text(myhtml_token_node_t *token_node, size_t *length) |
234
|
|
|
|
|
|
|
{ |
235
|
0
|
0
|
|
|
|
|
if(length) |
236
|
0
|
|
|
|
|
|
*length = token_node->str.length; |
237
|
|
|
|
|
|
|
|
238
|
0
|
|
|
|
|
|
return token_node->str.data; |
239
|
|
|
|
|
|
|
} |
240
|
|
|
|
|
|
|
|
241
|
0
|
|
|
|
|
|
mycore_string_t * myhtml_token_node_string(myhtml_token_node_t *token_node) |
242
|
|
|
|
|
|
|
{ |
243
|
0
|
|
|
|
|
|
return &token_node->str; |
244
|
|
|
|
|
|
|
} |
245
|
|
|
|
|
|
|
|
246
|
0
|
|
|
|
|
|
bool myhtml_token_node_is_close(myhtml_token_node_t *token_node) |
247
|
|
|
|
|
|
|
{ |
248
|
0
|
|
|
|
|
|
return (token_node->type & MyHTML_TOKEN_TYPE_CLOSE); |
249
|
|
|
|
|
|
|
} |
250
|
|
|
|
|
|
|
|
251
|
0
|
|
|
|
|
|
bool myhtml_token_node_is_close_self(myhtml_token_node_t *token_node) |
252
|
|
|
|
|
|
|
{ |
253
|
0
|
|
|
|
|
|
return (token_node->type & MyHTML_TOKEN_TYPE_CLOSE_SELF); |
254
|
|
|
|
|
|
|
} |
255
|
|
|
|
|
|
|
|
256
|
254
|
|
|
|
|
|
void myhtml_token_node_wait_for_done(myhtml_token_t* token, myhtml_token_node_t* node) |
257
|
|
|
|
|
|
|
{ |
258
|
|
|
|
|
|
|
#ifndef MyCORE_BUILD_WITHOUT_THREADS |
259
|
254
|
50
|
|
|
|
|
while((node->type & MyHTML_TOKEN_TYPE_DONE) == 0) {mythread_nanosleep_sleep(token->tree->myhtml->thread_stream->timespec);} |
260
|
|
|
|
|
|
|
#endif |
261
|
254
|
|
|
|
|
|
} |
262
|
|
|
|
|
|
|
|
263
|
17
|
|
|
|
|
|
void myhtml_token_set_done(myhtml_token_node_t* node) |
264
|
|
|
|
|
|
|
{ |
265
|
17
|
|
|
|
|
|
node->type |= MyHTML_TOKEN_TYPE_DONE; |
266
|
17
|
|
|
|
|
|
} |
267
|
|
|
|
|
|
|
|
268
|
0
|
|
|
|
|
|
myhtml_token_node_t * myhtml_token_node_clone(myhtml_token_t* token, myhtml_token_node_t* node, size_t token_thread_idx, size_t attr_thread_idx) |
269
|
|
|
|
|
|
|
{ |
270
|
0
|
0
|
|
|
|
|
if(node == NULL) |
271
|
0
|
|
|
|
|
|
return NULL; |
272
|
|
|
|
|
|
|
|
273
|
0
|
|
|
|
|
|
myhtml_tree_t* tree = token->tree; |
274
|
0
|
|
|
|
|
|
myhtml_token_node_t* new_node = myhtml_token_node_create(token, token_thread_idx); |
275
|
|
|
|
|
|
|
|
276
|
0
|
0
|
|
|
|
|
if(new_node == NULL) |
277
|
0
|
|
|
|
|
|
return NULL; |
278
|
|
|
|
|
|
|
|
279
|
0
|
|
|
|
|
|
new_node->tag_id = node->tag_id; |
280
|
0
|
|
|
|
|
|
new_node->type = node->type; |
281
|
0
|
|
|
|
|
|
new_node->attr_first = NULL; |
282
|
0
|
|
|
|
|
|
new_node->attr_last = NULL; |
283
|
0
|
|
|
|
|
|
new_node->raw_begin = node->raw_begin; |
284
|
0
|
|
|
|
|
|
new_node->raw_length = node->raw_length; |
285
|
0
|
|
|
|
|
|
new_node->element_begin = node->element_begin; |
286
|
0
|
|
|
|
|
|
new_node->element_length = node->element_length; |
287
|
|
|
|
|
|
|
|
288
|
0
|
0
|
|
|
|
|
if(node->str.length) { |
289
|
0
|
|
|
|
|
|
mycore_string_init(tree->mchar, tree->mchar_node_id, &new_node->str, node->str.length + 1); |
290
|
0
|
|
|
|
|
|
mycore_string_append(&new_node->str, node->str.data, node->str.length); |
291
|
|
|
|
|
|
|
} else { |
292
|
0
|
|
|
|
|
|
mycore_string_clean_all(&new_node->str); |
293
|
|
|
|
|
|
|
} |
294
|
|
|
|
|
|
|
|
295
|
0
|
|
|
|
|
|
myhtml_token_node_attr_copy(token, node, new_node, attr_thread_idx); |
296
|
|
|
|
|
|
|
|
297
|
0
|
|
|
|
|
|
return new_node; |
298
|
|
|
|
|
|
|
} |
299
|
|
|
|
|
|
|
|
300
|
0
|
|
|
|
|
|
void myhtml_token_node_text_append(myhtml_token_t* token, myhtml_token_node_t* dest, const char* text, size_t text_len) |
301
|
|
|
|
|
|
|
{ |
302
|
0
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &dest->str, (text_len + 2)); |
303
|
|
|
|
|
|
|
|
304
|
0
|
|
|
|
|
|
mycore_string_t* string = &dest->str; |
305
|
0
|
|
|
|
|
|
mycore_string_append(string, text, text_len); |
306
|
0
|
|
|
|
|
|
} |
307
|
|
|
|
|
|
|
|
308
|
0
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_node_attr_append(myhtml_token_t* token, myhtml_token_node_t* dest, |
309
|
|
|
|
|
|
|
const char* key, size_t key_len, |
310
|
|
|
|
|
|
|
const char* value, size_t value_len, size_t thread_idx) |
311
|
|
|
|
|
|
|
{ |
312
|
0
|
|
|
|
|
|
myhtml_token_attr_t* new_attr = mcobject_async_malloc(token->attr_obj, thread_idx, NULL); |
313
|
0
|
|
|
|
|
|
new_attr->next = 0; |
314
|
|
|
|
|
|
|
|
315
|
0
|
0
|
|
|
|
|
if(key_len) { |
316
|
0
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->key, (key_len + 1)); |
317
|
0
|
|
|
|
|
|
mycore_string_append_lowercase(&new_attr->key, key, key_len); |
318
|
|
|
|
|
|
|
} |
319
|
|
|
|
|
|
|
else |
320
|
0
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->key); |
321
|
|
|
|
|
|
|
|
322
|
0
|
0
|
|
|
|
|
if(value_len) { |
323
|
0
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->value, (value_len + 1)); |
324
|
0
|
|
|
|
|
|
mycore_string_append(&new_attr->value, value, value_len); |
325
|
|
|
|
|
|
|
} |
326
|
|
|
|
|
|
|
else |
327
|
0
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->value); |
328
|
|
|
|
|
|
|
|
329
|
0
|
0
|
|
|
|
|
if(dest->attr_first == NULL) { |
330
|
0
|
|
|
|
|
|
new_attr->prev = 0; |
331
|
|
|
|
|
|
|
|
332
|
0
|
|
|
|
|
|
dest->attr_first = new_attr; |
333
|
0
|
|
|
|
|
|
dest->attr_last = new_attr; |
334
|
|
|
|
|
|
|
} |
335
|
|
|
|
|
|
|
else { |
336
|
0
|
|
|
|
|
|
dest->attr_last->next = new_attr; |
337
|
0
|
|
|
|
|
|
new_attr->prev = dest->attr_last; |
338
|
|
|
|
|
|
|
|
339
|
0
|
|
|
|
|
|
dest->attr_last = new_attr; |
340
|
|
|
|
|
|
|
} |
341
|
|
|
|
|
|
|
|
342
|
0
|
|
|
|
|
|
new_attr->ns = MyHTML_NAMESPACE_HTML; |
343
|
|
|
|
|
|
|
|
344
|
0
|
|
|
|
|
|
return new_attr; |
345
|
|
|
|
|
|
|
} |
346
|
|
|
|
|
|
|
|
347
|
63
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_node_attr_append_with_convert_encoding(myhtml_token_t* token, myhtml_token_node_t* dest, |
348
|
|
|
|
|
|
|
const char* key, size_t key_len, |
349
|
|
|
|
|
|
|
const char* value, size_t value_len, |
350
|
|
|
|
|
|
|
size_t thread_idx, myencoding_t encoding) |
351
|
|
|
|
|
|
|
{ |
352
|
63
|
|
|
|
|
|
myhtml_token_attr_t* new_attr = mcobject_async_malloc(token->attr_obj, thread_idx, NULL); |
353
|
63
|
|
|
|
|
|
new_attr->next = 0; |
354
|
|
|
|
|
|
|
|
355
|
63
|
100
|
|
|
|
|
if(key_len) { |
356
|
26
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->key, (key_len + 1)); |
357
|
|
|
|
|
|
|
|
358
|
26
|
50
|
|
|
|
|
if(encoding == MyENCODING_UTF_8) |
359
|
26
|
|
|
|
|
|
mycore_string_append_lowercase(&new_attr->key, key, key_len); |
360
|
|
|
|
|
|
|
else |
361
|
26
|
|
|
|
|
|
myencoding_string_append_lowercase_ascii(&new_attr->key, key, key_len, encoding); |
362
|
|
|
|
|
|
|
} |
363
|
|
|
|
|
|
|
else |
364
|
37
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->key); |
365
|
|
|
|
|
|
|
|
366
|
63
|
100
|
|
|
|
|
if(value_len) { |
367
|
40
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->value, (value_len + 1)); |
368
|
|
|
|
|
|
|
|
369
|
40
|
50
|
|
|
|
|
if(encoding == MyENCODING_UTF_8) |
370
|
40
|
|
|
|
|
|
mycore_string_append(&new_attr->value, value, value_len); |
371
|
|
|
|
|
|
|
else |
372
|
40
|
|
|
|
|
|
myencoding_string_append(&new_attr->value, value, value_len, encoding); |
373
|
|
|
|
|
|
|
} |
374
|
|
|
|
|
|
|
else |
375
|
23
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->value); |
376
|
|
|
|
|
|
|
|
377
|
63
|
100
|
|
|
|
|
if(dest->attr_first == NULL) { |
378
|
5
|
|
|
|
|
|
new_attr->prev = 0; |
379
|
|
|
|
|
|
|
|
380
|
5
|
|
|
|
|
|
dest->attr_first = new_attr; |
381
|
5
|
|
|
|
|
|
dest->attr_last = new_attr; |
382
|
|
|
|
|
|
|
} |
383
|
|
|
|
|
|
|
else { |
384
|
58
|
|
|
|
|
|
dest->attr_last->next = new_attr; |
385
|
58
|
|
|
|
|
|
new_attr->prev = dest->attr_last; |
386
|
|
|
|
|
|
|
|
387
|
58
|
|
|
|
|
|
dest->attr_last = new_attr; |
388
|
|
|
|
|
|
|
} |
389
|
|
|
|
|
|
|
|
390
|
63
|
|
|
|
|
|
new_attr->ns = MyHTML_NAMESPACE_HTML; |
391
|
|
|
|
|
|
|
|
392
|
63
|
|
|
|
|
|
return new_attr; |
393
|
|
|
|
|
|
|
} |
394
|
|
|
|
|
|
|
|
395
|
0
|
|
|
|
|
|
void myhtml_token_node_attr_copy_with_check(myhtml_token_t* token, myhtml_token_node_t* target, myhtml_token_node_t* dest, size_t thread_idx) |
396
|
|
|
|
|
|
|
{ |
397
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = target->attr_first; |
398
|
|
|
|
|
|
|
|
399
|
0
|
0
|
|
|
|
|
while (attr) |
400
|
|
|
|
|
|
|
{ |
401
|
0
|
0
|
|
|
|
|
if(attr->key.length && myhtml_token_attr_by_name(dest, attr->key.data, attr->key.length) == NULL) { |
|
|
0
|
|
|
|
|
|
402
|
0
|
|
|
|
|
|
myhtml_token_attr_copy(token, attr, dest, thread_idx); |
403
|
|
|
|
|
|
|
} |
404
|
|
|
|
|
|
|
|
405
|
0
|
|
|
|
|
|
attr = attr->next; |
406
|
|
|
|
|
|
|
} |
407
|
0
|
|
|
|
|
|
} |
408
|
|
|
|
|
|
|
|
409
|
0
|
|
|
|
|
|
void myhtml_token_node_attr_copy(myhtml_token_t* token, myhtml_token_node_t* target, myhtml_token_node_t* dest, size_t thread_idx) |
410
|
|
|
|
|
|
|
{ |
411
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = target->attr_first; |
412
|
|
|
|
|
|
|
|
413
|
0
|
0
|
|
|
|
|
while (attr) |
414
|
|
|
|
|
|
|
{ |
415
|
0
|
|
|
|
|
|
myhtml_token_attr_copy(token, attr, dest, thread_idx); |
416
|
0
|
|
|
|
|
|
attr = attr->next; |
417
|
|
|
|
|
|
|
} |
418
|
0
|
|
|
|
|
|
} |
419
|
|
|
|
|
|
|
|
420
|
15
|
|
|
|
|
|
bool myhtml_token_attr_copy(myhtml_token_t* token, myhtml_token_attr_t* attr, myhtml_token_node_t* dest, size_t thread_idx) |
421
|
|
|
|
|
|
|
{ |
422
|
15
|
|
|
|
|
|
myhtml_token_attr_t* new_attr = mcobject_async_malloc(token->attr_obj, thread_idx, NULL); |
423
|
15
|
|
|
|
|
|
new_attr->next = 0; |
424
|
|
|
|
|
|
|
|
425
|
15
|
50
|
|
|
|
|
if(attr->key.length) { |
426
|
15
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->key, (attr->key.length + 1)); |
427
|
15
|
|
|
|
|
|
mycore_string_append_lowercase(&new_attr->key, attr->key.data, attr->key.length); |
428
|
|
|
|
|
|
|
} |
429
|
|
|
|
|
|
|
else |
430
|
0
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->key); |
431
|
|
|
|
|
|
|
|
432
|
15
|
50
|
|
|
|
|
if(attr->value.length) { |
433
|
15
|
|
|
|
|
|
mycore_string_init(token->tree->mchar, token->tree->mchar_node_id, &new_attr->value, (attr->value.length + 1)); |
434
|
15
|
|
|
|
|
|
mycore_string_append(&new_attr->value, attr->value.data, attr->value.length); |
435
|
|
|
|
|
|
|
} |
436
|
|
|
|
|
|
|
else |
437
|
0
|
|
|
|
|
|
mycore_string_clean_all(&new_attr->value); |
438
|
|
|
|
|
|
|
|
439
|
15
|
50
|
|
|
|
|
if(dest->attr_first == NULL) { |
440
|
15
|
|
|
|
|
|
new_attr->prev = 0; |
441
|
|
|
|
|
|
|
|
442
|
15
|
|
|
|
|
|
dest->attr_first = new_attr; |
443
|
15
|
|
|
|
|
|
dest->attr_last = new_attr; |
444
|
|
|
|
|
|
|
} |
445
|
|
|
|
|
|
|
else { |
446
|
0
|
|
|
|
|
|
dest->attr_last->next = new_attr; |
447
|
0
|
|
|
|
|
|
new_attr->prev = dest->attr_last; |
448
|
|
|
|
|
|
|
|
449
|
0
|
|
|
|
|
|
dest->attr_last = new_attr; |
450
|
|
|
|
|
|
|
} |
451
|
|
|
|
|
|
|
|
452
|
15
|
|
|
|
|
|
new_attr->ns = attr->ns; |
453
|
|
|
|
|
|
|
|
454
|
15
|
|
|
|
|
|
return true; |
455
|
|
|
|
|
|
|
} |
456
|
|
|
|
|
|
|
|
457
|
0
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_match(myhtml_token_t* token, myhtml_token_node_t* target, |
458
|
|
|
|
|
|
|
const char* key, size_t key_size, const char* value, size_t value_size) |
459
|
|
|
|
|
|
|
{ |
460
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = target->attr_first; |
461
|
|
|
|
|
|
|
|
462
|
0
|
0
|
|
|
|
|
while (attr) |
463
|
|
|
|
|
|
|
{ |
464
|
0
|
0
|
|
|
|
|
if(attr->key.length == key_size && attr->value.length == value_size) |
|
|
0
|
|
|
|
|
|
465
|
|
|
|
|
|
|
{ |
466
|
0
|
0
|
|
|
|
|
if((mycore_strcmp(attr->key.data, key) == 0)) { |
467
|
0
|
0
|
|
|
|
|
if((mycore_strcmp(attr->value.data, value) == 0)) |
468
|
0
|
|
|
|
|
|
return attr; |
469
|
|
|
|
|
|
|
else |
470
|
0
|
|
|
|
|
|
return NULL; |
471
|
|
|
|
|
|
|
} |
472
|
|
|
|
|
|
|
} |
473
|
|
|
|
|
|
|
|
474
|
0
|
|
|
|
|
|
attr = attr->next; |
475
|
|
|
|
|
|
|
} |
476
|
|
|
|
|
|
|
|
477
|
0
|
|
|
|
|
|
return NULL; |
478
|
|
|
|
|
|
|
} |
479
|
|
|
|
|
|
|
|
480
|
0
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_match_case(myhtml_token_t* token, myhtml_token_node_t* target, |
481
|
|
|
|
|
|
|
const char* key, size_t key_size, const char* value, size_t value_size) |
482
|
|
|
|
|
|
|
{ |
483
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = target->attr_first; |
484
|
|
|
|
|
|
|
|
485
|
0
|
0
|
|
|
|
|
while (attr) |
486
|
|
|
|
|
|
|
{ |
487
|
0
|
0
|
|
|
|
|
if(attr->key.length == key_size && attr->value.length == value_size) |
|
|
0
|
|
|
|
|
|
488
|
|
|
|
|
|
|
{ |
489
|
0
|
0
|
|
|
|
|
if((mycore_strcmp(attr->key.data, key) == 0)) { |
490
|
0
|
0
|
|
|
|
|
if((mycore_strcasecmp(attr->value.data, value) == 0)) |
491
|
0
|
|
|
|
|
|
return attr; |
492
|
|
|
|
|
|
|
else |
493
|
0
|
|
|
|
|
|
return NULL; |
494
|
|
|
|
|
|
|
} |
495
|
|
|
|
|
|
|
} |
496
|
|
|
|
|
|
|
|
497
|
0
|
|
|
|
|
|
attr = attr->next; |
498
|
|
|
|
|
|
|
} |
499
|
|
|
|
|
|
|
|
500
|
0
|
|
|
|
|
|
return NULL; |
501
|
|
|
|
|
|
|
} |
502
|
|
|
|
|
|
|
|
503
|
0
|
|
|
|
|
|
void myhtml_token_adjust_mathml_attributes(myhtml_token_node_t* target) |
504
|
|
|
|
|
|
|
{ |
505
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = myhtml_token_attr_by_name(target, "definitionurl", 13); |
506
|
|
|
|
|
|
|
|
507
|
0
|
0
|
|
|
|
|
if(attr) { |
508
|
0
|
|
|
|
|
|
memcpy(attr->key.data, "definitionURL", 13); |
509
|
|
|
|
|
|
|
} |
510
|
0
|
|
|
|
|
|
} |
511
|
|
|
|
|
|
|
|
512
|
74
|
|
|
|
|
|
void _myhtml_token_create_copy_srt(myhtml_token_t* token, const char* from, size_t from_size, char** to) |
513
|
|
|
|
|
|
|
{ |
514
|
74
|
|
|
|
|
|
*to = mchar_async_malloc(token->tree->mchar, token->tree->mchar_node_id, (from_size + 2)); |
515
|
74
|
|
|
|
|
|
mycore_string_raw_copy(*to, from, from_size); |
516
|
74
|
|
|
|
|
|
} |
517
|
|
|
|
|
|
|
|
518
|
36
|
|
|
|
|
|
void myhtml_token_strict_doctype_by_token(myhtml_token_t* token, myhtml_token_node_t* target, myhtml_tree_doctype_t* return_doctype) |
519
|
|
|
|
|
|
|
{ |
520
|
36
|
|
|
|
|
|
myhtml_token_attr_t* attr = target->attr_first; |
521
|
|
|
|
|
|
|
|
522
|
36
|
100
|
|
|
|
|
if(attr && attr->key.length) { |
|
|
50
|
|
|
|
|
|
523
|
33
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, attr->key.data, attr->key.length, &return_doctype->attr_name); |
524
|
|
|
|
|
|
|
|
525
|
66
|
100
|
|
|
|
|
if(mycore_strcmp("html", return_doctype->attr_name)) |
526
|
2
|
|
|
|
|
|
return_doctype->is_html = false; |
527
|
|
|
|
|
|
|
else |
528
|
31
|
|
|
|
|
|
return_doctype->is_html = true; |
529
|
|
|
|
|
|
|
} |
530
|
|
|
|
|
|
|
else { |
531
|
3
|
|
|
|
|
|
return_doctype->is_html = false; |
532
|
|
|
|
|
|
|
|
533
|
3
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, "\0", 1, &return_doctype->attr_name); |
534
|
|
|
|
|
|
|
|
535
|
3
|
50
|
|
|
|
|
if(return_doctype->attr_public) |
536
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_public); |
537
|
3
|
|
|
|
|
|
return_doctype->attr_public = NULL; |
538
|
|
|
|
|
|
|
|
539
|
3
|
50
|
|
|
|
|
if(return_doctype->attr_system) |
540
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_system); |
541
|
3
|
|
|
|
|
|
return_doctype->attr_system = NULL; |
542
|
|
|
|
|
|
|
|
543
|
3
|
|
|
|
|
|
return; |
544
|
|
|
|
|
|
|
} |
545
|
|
|
|
|
|
|
|
546
|
33
|
|
|
|
|
|
attr = attr->next; |
547
|
|
|
|
|
|
|
|
548
|
33
|
100
|
|
|
|
|
if(attr && attr->value.length) |
|
|
50
|
|
|
|
|
|
549
|
|
|
|
|
|
|
{ |
550
|
22
|
100
|
|
|
|
|
if(mycore_strcasecmp(attr->value.data, "PUBLIC") == 0) |
551
|
|
|
|
|
|
|
{ |
552
|
|
|
|
|
|
|
// try see public |
553
|
16
|
|
|
|
|
|
attr = attr->next; |
554
|
|
|
|
|
|
|
|
555
|
16
|
100
|
|
|
|
|
if(attr && attr->value.length) { |
|
|
100
|
|
|
|
|
|
556
|
14
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, attr->value.data, attr->value.length, &return_doctype->attr_public); |
557
|
|
|
|
|
|
|
|
558
|
|
|
|
|
|
|
// try see system |
559
|
14
|
|
|
|
|
|
attr = attr->next; |
560
|
|
|
|
|
|
|
|
561
|
28
|
100
|
|
|
|
|
if(attr && attr->value.length) |
|
|
50
|
|
|
|
|
|
562
|
9
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, attr->value.data, attr->value.length, &return_doctype->attr_system); |
563
|
|
|
|
|
|
|
else { |
564
|
5
|
50
|
|
|
|
|
if(return_doctype->attr_system) |
565
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_system); |
566
|
|
|
|
|
|
|
|
567
|
5
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, "\0", 1, &return_doctype->attr_system); |
568
|
|
|
|
|
|
|
} |
569
|
|
|
|
|
|
|
} |
570
|
|
|
|
|
|
|
else { |
571
|
2
|
50
|
|
|
|
|
if(return_doctype->attr_public) |
572
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_public); |
573
|
2
|
|
|
|
|
|
return_doctype->attr_public = NULL; |
574
|
|
|
|
|
|
|
|
575
|
2
|
50
|
|
|
|
|
if(return_doctype->attr_system) |
576
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_system); |
577
|
16
|
|
|
|
|
|
return_doctype->attr_system = NULL; |
578
|
|
|
|
|
|
|
} |
579
|
|
|
|
|
|
|
} |
580
|
6
|
50
|
|
|
|
|
else if(mycore_strncasecmp(attr->value.data, "SYSTEM", attr->value.length) == 0) |
581
|
|
|
|
|
|
|
{ |
582
|
6
|
|
|
|
|
|
attr = attr->next; |
583
|
|
|
|
|
|
|
|
584
|
6
|
100
|
|
|
|
|
if(attr && attr->value.length) { |
|
|
50
|
|
|
|
|
|
585
|
5
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, "\0", 1, &return_doctype->attr_public); |
586
|
5
|
|
|
|
|
|
_myhtml_token_create_copy_srt(token, attr->value.data, attr->value.length, &return_doctype->attr_system); |
587
|
|
|
|
|
|
|
} |
588
|
|
|
|
|
|
|
else { |
589
|
1
|
50
|
|
|
|
|
if(return_doctype->attr_public) |
590
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_public); |
591
|
1
|
|
|
|
|
|
return_doctype->attr_public = NULL; |
592
|
|
|
|
|
|
|
|
593
|
1
|
50
|
|
|
|
|
if(return_doctype->attr_system) |
594
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_system); |
595
|
6
|
|
|
|
|
|
return_doctype->attr_system = NULL; |
596
|
|
|
|
|
|
|
} |
597
|
|
|
|
|
|
|
} |
598
|
|
|
|
|
|
|
else { |
599
|
0
|
0
|
|
|
|
|
if(return_doctype->attr_public) |
600
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_public); |
601
|
0
|
|
|
|
|
|
return_doctype->attr_public = NULL; |
602
|
|
|
|
|
|
|
|
603
|
0
|
0
|
|
|
|
|
if(return_doctype->attr_system) |
604
|
0
|
|
|
|
|
|
mycore_free(return_doctype->attr_system); |
605
|
0
|
|
|
|
|
|
return_doctype->attr_system = NULL; |
606
|
|
|
|
|
|
|
} |
607
|
|
|
|
|
|
|
} |
608
|
|
|
|
|
|
|
} |
609
|
|
|
|
|
|
|
|
610
|
18
|
|
|
|
|
|
bool myhtml_token_doctype_check_html_4_0(myhtml_tree_doctype_t* return_doctype) |
611
|
|
|
|
|
|
|
{ |
612
|
34
|
100
|
|
|
|
|
return mycore_strcmp(return_doctype->attr_public, "-//W3C//DTD HTML 4.0//EN") && |
|
|
50
|
|
|
|
|
|
613
|
16
|
50
|
|
|
|
|
(return_doctype->attr_system == NULL || mycore_strcmp(return_doctype->attr_system, "http://www.w3.org/TR/REC-html40/strict.dtd")); |
614
|
|
|
|
|
|
|
} |
615
|
|
|
|
|
|
|
|
616
|
16
|
|
|
|
|
|
bool myhtml_token_doctype_check_html_4_01(myhtml_tree_doctype_t* return_doctype) |
617
|
|
|
|
|
|
|
{ |
618
|
30
|
100
|
|
|
|
|
return mycore_strcmp(return_doctype->attr_public, "-//W3C//DTD HTML 4.01//EN") && |
|
|
50
|
|
|
|
|
|
619
|
14
|
50
|
|
|
|
|
(return_doctype->attr_system == NULL || mycore_strcmp(return_doctype->attr_system, "http://www.w3.org/TR/html4/strict.dtd")); |
620
|
|
|
|
|
|
|
} |
621
|
|
|
|
|
|
|
|
622
|
14
|
|
|
|
|
|
bool myhtml_token_doctype_check_xhtml_1_0(myhtml_tree_doctype_t* return_doctype) |
623
|
|
|
|
|
|
|
{ |
624
|
14
|
50
|
|
|
|
|
if(return_doctype->attr_system == NULL) |
625
|
0
|
|
|
|
|
|
return true; |
626
|
|
|
|
|
|
|
|
627
|
26
|
|
|
|
|
|
return mycore_strcmp(return_doctype->attr_public, "-//W3C//DTD XHTML 1.0 Strict//EN") && |
628
|
12
|
|
|
|
|
|
mycore_strcmp(return_doctype->attr_system, "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"); |
629
|
|
|
|
|
|
|
} |
630
|
|
|
|
|
|
|
|
631
|
12
|
|
|
|
|
|
bool myhtml_token_doctype_check_xhtml_1_1(myhtml_tree_doctype_t* return_doctype) |
632
|
|
|
|
|
|
|
{ |
633
|
12
|
50
|
|
|
|
|
if(return_doctype->attr_system == NULL) |
634
|
0
|
|
|
|
|
|
return true; |
635
|
|
|
|
|
|
|
|
636
|
23
|
|
|
|
|
|
return mycore_strcmp(return_doctype->attr_public, "-//W3C//DTD XHTML 1.1//EN") && |
637
|
11
|
|
|
|
|
|
mycore_strcmp(return_doctype->attr_system, "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd"); |
638
|
|
|
|
|
|
|
} |
639
|
|
|
|
|
|
|
|
640
|
36
|
|
|
|
|
|
bool myhtml_token_release_and_check_doctype_attributes(myhtml_token_t* token, myhtml_token_node_t* target, myhtml_tree_doctype_t* return_doctype) |
641
|
|
|
|
|
|
|
{ |
642
|
36
|
50
|
|
|
|
|
if(return_doctype == NULL) |
643
|
0
|
|
|
|
|
|
return false; |
644
|
|
|
|
|
|
|
|
645
|
36
|
|
|
|
|
|
myhtml_token_strict_doctype_by_token(token, target, return_doctype); |
646
|
|
|
|
|
|
|
|
647
|
36
|
50
|
|
|
|
|
if(return_doctype->attr_name == NULL) |
648
|
0
|
|
|
|
|
|
return false; |
649
|
|
|
|
|
|
|
|
650
|
36
|
100
|
|
|
|
|
if((return_doctype->is_html || |
|
|
100
|
|
|
|
|
|
651
|
4
|
50
|
|
|
|
|
return_doctype->attr_public || |
652
|
0
|
0
|
|
|
|
|
(return_doctype->attr_system && mycore_strcmp(return_doctype->attr_system, "about:legacy-compat")))) |
653
|
|
|
|
|
|
|
{ |
654
|
32
|
100
|
|
|
|
|
if(return_doctype->attr_public == NULL) |
655
|
13
|
|
|
|
|
|
return false; |
656
|
|
|
|
|
|
|
|
657
|
37
|
|
|
|
|
|
if(return_doctype->is_html && |
658
|
34
|
100
|
|
|
|
|
myhtml_token_doctype_check_html_4_0(return_doctype) && |
659
|
30
|
100
|
|
|
|
|
myhtml_token_doctype_check_html_4_01(return_doctype) && |
660
|
26
|
100
|
|
|
|
|
myhtml_token_doctype_check_xhtml_1_0(return_doctype) && |
661
|
12
|
|
|
|
|
|
myhtml_token_doctype_check_xhtml_1_1(return_doctype)) |
662
|
|
|
|
|
|
|
{ |
663
|
11
|
|
|
|
|
|
return false; |
664
|
|
|
|
|
|
|
} |
665
|
|
|
|
|
|
|
} |
666
|
|
|
|
|
|
|
|
667
|
12
|
|
|
|
|
|
return true; |
668
|
|
|
|
|
|
|
} |
669
|
|
|
|
|
|
|
|
670
|
0
|
|
|
|
|
|
void myhtml_token_adjust_svg_attributes(myhtml_token_node_t* target) |
671
|
|
|
|
|
|
|
{ |
672
|
0
|
|
|
|
|
|
size_t count = sizeof(myhtml_token_attr_svg_replacement) / sizeof(myhtml_token_replacement_entry_t); |
673
|
|
|
|
|
|
|
|
674
|
0
|
0
|
|
|
|
|
for (size_t i = 0; i < count; i++) |
675
|
|
|
|
|
|
|
{ |
676
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = myhtml_token_attr_by_name(target, myhtml_token_attr_svg_replacement[i].from, |
677
|
|
|
|
|
|
|
myhtml_token_attr_svg_replacement[i].from_size); |
678
|
|
|
|
|
|
|
|
679
|
0
|
0
|
|
|
|
|
if(attr) { |
680
|
0
|
|
|
|
|
|
mycore_string_clean(&attr->key); |
681
|
0
|
|
|
|
|
|
mycore_string_append(&attr->key, myhtml_token_attr_svg_replacement[i].to, |
682
|
|
|
|
|
|
|
myhtml_token_attr_svg_replacement[i].to_size); |
683
|
|
|
|
|
|
|
} |
684
|
|
|
|
|
|
|
} |
685
|
0
|
|
|
|
|
|
} |
686
|
|
|
|
|
|
|
|
687
|
0
|
|
|
|
|
|
void myhtml_token_adjust_foreign_attributes(myhtml_token_node_t* target) |
688
|
|
|
|
|
|
|
{ |
689
|
0
|
|
|
|
|
|
size_t count = sizeof(myhtml_token_attr_namespace_replacement) / sizeof(myhtml_token_namespace_replacement_t); |
690
|
|
|
|
|
|
|
|
691
|
0
|
0
|
|
|
|
|
for (size_t i = 0; i < count; i++) |
692
|
|
|
|
|
|
|
{ |
693
|
0
|
|
|
|
|
|
myhtml_token_attr_t* attr = myhtml_token_attr_by_name(target, myhtml_token_attr_namespace_replacement[i].from, |
694
|
|
|
|
|
|
|
myhtml_token_attr_namespace_replacement[i].from_size); |
695
|
|
|
|
|
|
|
|
696
|
0
|
0
|
|
|
|
|
if(attr) { |
697
|
0
|
|
|
|
|
|
mycore_string_clean(&attr->key); |
698
|
0
|
|
|
|
|
|
mycore_string_append(&attr->key, myhtml_token_attr_namespace_replacement[i].to, |
699
|
|
|
|
|
|
|
myhtml_token_attr_namespace_replacement[i].to_size); |
700
|
|
|
|
|
|
|
|
701
|
0
|
|
|
|
|
|
attr->ns = myhtml_token_attr_namespace_replacement[i].ns; |
702
|
|
|
|
|
|
|
} |
703
|
|
|
|
|
|
|
} |
704
|
0
|
|
|
|
|
|
} |
705
|
|
|
|
|
|
|
|
706
|
0
|
|
|
|
|
|
bool myhtml_token_attr_compare(myhtml_token_node_t* target, myhtml_token_node_t* dest) |
707
|
|
|
|
|
|
|
{ |
708
|
0
|
0
|
|
|
|
|
if(target == NULL || dest == NULL) |
|
|
0
|
|
|
|
|
|
709
|
0
|
|
|
|
|
|
return false; |
710
|
|
|
|
|
|
|
|
711
|
0
|
|
|
|
|
|
myhtml_token_attr_t* target_attr = target->attr_first; |
712
|
0
|
|
|
|
|
|
myhtml_token_attr_t* dest_attr = dest->attr_first; |
713
|
|
|
|
|
|
|
|
714
|
0
|
0
|
|
|
|
|
while (target_attr && dest_attr) |
|
|
0
|
|
|
|
|
|
715
|
|
|
|
|
|
|
{ |
716
|
0
|
0
|
|
|
|
|
if(target_attr->key.length == dest_attr->key.length && |
|
|
0
|
|
|
|
|
|
717
|
0
|
|
|
|
|
|
target_attr->value.length == dest_attr->value.length) |
718
|
|
|
|
|
|
|
{ |
719
|
0
|
0
|
|
|
|
|
if(mycore_strcmp(target_attr->key.data, dest_attr->key.data) != 0) |
720
|
0
|
|
|
|
|
|
break; |
721
|
|
|
|
|
|
|
|
722
|
0
|
0
|
|
|
|
|
if(mycore_strcasecmp(target_attr->value.data, dest_attr->value.data) != 0) |
723
|
0
|
|
|
|
|
|
break; |
724
|
|
|
|
|
|
|
} |
725
|
|
|
|
|
|
|
else |
726
|
|
|
|
|
|
|
break; |
727
|
|
|
|
|
|
|
|
728
|
0
|
|
|
|
|
|
target_attr = target_attr->next; |
729
|
0
|
|
|
|
|
|
dest_attr = dest_attr->next; |
730
|
|
|
|
|
|
|
} |
731
|
|
|
|
|
|
|
|
732
|
0
|
0
|
|
|
|
|
if(target_attr == NULL && dest_attr == NULL) |
|
|
0
|
|
|
|
|
|
733
|
0
|
|
|
|
|
|
return true; |
734
|
|
|
|
|
|
|
|
735
|
0
|
|
|
|
|
|
return false; |
736
|
|
|
|
|
|
|
} |
737
|
|
|
|
|
|
|
|
738
|
140
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_by_name(myhtml_token_node_t* node, const char* name, size_t name_length) |
739
|
|
|
|
|
|
|
{ |
740
|
140
|
|
|
|
|
|
myhtml_token_attr_t* attr = node->attr_first; |
741
|
|
|
|
|
|
|
|
742
|
351
|
100
|
|
|
|
|
while (attr) |
743
|
|
|
|
|
|
|
{ |
744
|
313
|
100
|
|
|
|
|
if(name_length == attr->key.length) { |
745
|
104
|
100
|
|
|
|
|
if(mycore_strcmp(attr->key.data, name) == 0) |
746
|
102
|
|
|
|
|
|
break; |
747
|
|
|
|
|
|
|
} |
748
|
|
|
|
|
|
|
|
749
|
211
|
|
|
|
|
|
attr = attr->next; |
750
|
|
|
|
|
|
|
} |
751
|
|
|
|
|
|
|
|
752
|
140
|
|
|
|
|
|
return attr; |
753
|
|
|
|
|
|
|
} |
754
|
|
|
|
|
|
|
|
755
|
138
|
|
|
|
|
|
void myhtml_token_delete(myhtml_token_t* token, myhtml_token_node_t* node) |
756
|
|
|
|
|
|
|
{ |
757
|
138
|
100
|
|
|
|
|
if(node->str.data && node->str.mchar) { |
|
|
50
|
|
|
|
|
|
758
|
87
|
|
|
|
|
|
mchar_async_free(node->str.mchar, node->str.node_idx, node->str.data); |
759
|
|
|
|
|
|
|
} |
760
|
|
|
|
|
|
|
|
761
|
138
|
|
|
|
|
|
mcobject_async_free(token->nodes_obj, node); |
762
|
138
|
|
|
|
|
|
} |
763
|
|
|
|
|
|
|
|
764
|
138
|
|
|
|
|
|
void myhtml_token_attr_delete_all(myhtml_token_t* token, myhtml_token_node_t* node) |
765
|
|
|
|
|
|
|
{ |
766
|
138
|
|
|
|
|
|
myhtml_token_attr_t* attr = node->attr_first; |
767
|
|
|
|
|
|
|
|
768
|
169
|
100
|
|
|
|
|
while (attr) |
769
|
|
|
|
|
|
|
{ |
770
|
31
|
50
|
|
|
|
|
if(attr->key.data && attr->key.mchar) { |
|
|
50
|
|
|
|
|
|
771
|
31
|
|
|
|
|
|
mchar_async_free(attr->key.mchar, attr->key.node_idx, attr->key.data); |
772
|
|
|
|
|
|
|
} |
773
|
|
|
|
|
|
|
|
774
|
31
|
50
|
|
|
|
|
if(attr->value.data && attr->value.mchar) { |
|
|
50
|
|
|
|
|
|
775
|
31
|
|
|
|
|
|
mchar_async_free(attr->value.mchar, attr->value.node_idx, attr->value.data); |
776
|
|
|
|
|
|
|
} |
777
|
|
|
|
|
|
|
|
778
|
31
|
|
|
|
|
|
attr = attr->next; |
779
|
|
|
|
|
|
|
} |
780
|
138
|
|
|
|
|
|
} |
781
|
|
|
|
|
|
|
|
782
|
53
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_remove(myhtml_token_node_t* node, myhtml_token_attr_t* attr) |
783
|
|
|
|
|
|
|
{ |
784
|
53
|
50
|
|
|
|
|
if(attr) |
785
|
|
|
|
|
|
|
{ |
786
|
53
|
100
|
|
|
|
|
if(attr->prev) { |
787
|
15
|
|
|
|
|
|
attr->prev->next = attr->next; |
788
|
|
|
|
|
|
|
} |
789
|
|
|
|
|
|
|
else { |
790
|
38
|
|
|
|
|
|
node->attr_first = attr->next; |
791
|
|
|
|
|
|
|
} |
792
|
|
|
|
|
|
|
|
793
|
53
|
100
|
|
|
|
|
if(attr->next) { |
794
|
41
|
|
|
|
|
|
attr->next->prev = attr->prev; |
795
|
|
|
|
|
|
|
} |
796
|
|
|
|
|
|
|
else { |
797
|
12
|
|
|
|
|
|
node->attr_last = attr->prev; |
798
|
|
|
|
|
|
|
} |
799
|
|
|
|
|
|
|
|
800
|
53
|
|
|
|
|
|
attr->next = NULL; |
801
|
53
|
|
|
|
|
|
attr->prev = NULL; |
802
|
|
|
|
|
|
|
} |
803
|
|
|
|
|
|
|
|
804
|
53
|
|
|
|
|
|
return attr; |
805
|
|
|
|
|
|
|
} |
806
|
|
|
|
|
|
|
|
807
|
20
|
|
|
|
|
|
myhtml_token_attr_t * myhtml_token_attr_remove_by_name(myhtml_token_node_t* node, const char* name, size_t name_length) |
808
|
|
|
|
|
|
|
{ |
809
|
20
|
|
|
|
|
|
return myhtml_token_attr_remove(node, myhtml_token_attr_by_name(node, name, name_length)); |
810
|
|
|
|
|
|
|
} |
811
|
|
|
|
|
|
|
|
812
|
0
|
|
|
|
|
|
myhtml_token_node_t * myhtml_token_merged_two_token_string(myhtml_tree_t* tree, myhtml_token_node_t* token_to, myhtml_token_node_t* token_from, bool cp_reverse) |
813
|
|
|
|
|
|
|
{ |
814
|
0
|
|
|
|
|
|
myhtml_token_node_wait_for_done(tree->token, token_to); |
815
|
0
|
|
|
|
|
|
myhtml_token_node_wait_for_done(tree->token, token_from); |
816
|
|
|
|
|
|
|
|
817
|
0
|
|
|
|
|
|
mycore_string_t *string1 = &token_to->str; |
818
|
0
|
|
|
|
|
|
mycore_string_t *string2 = &token_from->str; |
819
|
|
|
|
|
|
|
|
820
|
0
|
|
|
|
|
|
token_to->raw_begin = 0; |
821
|
0
|
|
|
|
|
|
token_to->raw_length = 0; |
822
|
|
|
|
|
|
|
|
823
|
0
|
0
|
|
|
|
|
if(token_to->str.node_idx == tree->mchar_node_id) |
824
|
|
|
|
|
|
|
{ |
825
|
0
|
0
|
|
|
|
|
if(cp_reverse) { |
826
|
|
|
|
|
|
|
//mycore_string_copy(string2, &string_base); |
827
|
|
|
|
|
|
|
} |
828
|
|
|
|
|
|
|
else { |
829
|
0
|
|
|
|
|
|
mycore_string_copy(string1, string2); |
830
|
|
|
|
|
|
|
} |
831
|
|
|
|
|
|
|
|
832
|
0
|
|
|
|
|
|
return token_to; |
833
|
|
|
|
|
|
|
} |
834
|
0
|
0
|
|
|
|
|
if(token_from->str.node_idx == tree->mchar_node_id) |
835
|
|
|
|
|
|
|
{ |
836
|
0
|
0
|
|
|
|
|
if(cp_reverse) { |
837
|
0
|
|
|
|
|
|
mycore_string_copy(string2, string1); |
838
|
|
|
|
|
|
|
} |
839
|
|
|
|
|
|
|
else { |
840
|
0
|
|
|
|
|
|
mycore_string_copy(string1, string2); |
841
|
|
|
|
|
|
|
} |
842
|
|
|
|
|
|
|
|
843
|
0
|
|
|
|
|
|
return token_from; |
844
|
|
|
|
|
|
|
} |
845
|
|
|
|
|
|
|
else { |
846
|
|
|
|
|
|
|
mycore_string_t string_base; |
847
|
0
|
|
|
|
|
|
mycore_string_init(tree->mchar, tree->mchar_node_id, &string_base, (string1->length + string2->length + 2)); |
848
|
|
|
|
|
|
|
|
849
|
0
|
0
|
|
|
|
|
if(cp_reverse) { |
850
|
0
|
|
|
|
|
|
mycore_string_copy(&string_base, string2); |
851
|
0
|
|
|
|
|
|
mycore_string_copy(&string_base, string1); |
852
|
|
|
|
|
|
|
} |
853
|
|
|
|
|
|
|
else { |
854
|
0
|
|
|
|
|
|
mycore_string_copy(&string_base, string1); |
855
|
0
|
|
|
|
|
|
mycore_string_copy(&string_base, string2); |
856
|
|
|
|
|
|
|
} |
857
|
|
|
|
|
|
|
|
858
|
0
|
|
|
|
|
|
token_to->str = string_base; |
859
|
|
|
|
|
|
|
} |
860
|
|
|
|
|
|
|
|
861
|
0
|
|
|
|
|
|
return token_to; |
862
|
|
|
|
|
|
|
} |
863
|
|
|
|
|
|
|
|
864
|
0
|
|
|
|
|
|
void myhtml_token_set_replacement_character_for_null_token(myhtml_tree_t* tree, myhtml_token_node_t* node) |
865
|
|
|
|
|
|
|
{ |
866
|
0
|
|
|
|
|
|
myhtml_token_node_wait_for_done(tree->token, node); |
867
|
|
|
|
|
|
|
|
868
|
|
|
|
|
|
|
mycore_string_t new_str; |
869
|
0
|
|
|
|
|
|
mycore_string_init(tree->mchar, tree->mchar_node_id, &new_str, (node->str.length + 2)); |
870
|
|
|
|
|
|
|
|
871
|
0
|
|
|
|
|
|
mycore_string_append_with_replacement_null_characters(&new_str, node->str.data, node->str.length); |
872
|
|
|
|
|
|
|
|
873
|
0
|
|
|
|
|
|
node->str = new_str; |
874
|
0
|
|
|
|
|
|
} |
875
|
|
|
|
|
|
|
|