selectolax 0.3.25__cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl → 0.3.27__cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of selectolax might be problematic. Click here for more details.

Files changed (13) hide show
  1. selectolax/__init__.py +1 -1
  2. selectolax/lexbor.c +179 -179
  3. selectolax/lexbor.cpython-313-aarch64-linux-gnu.so +0 -0
  4. selectolax/lexbor.pyi +71 -40
  5. selectolax/parser.c +115 -115
  6. selectolax/parser.cpython-313-aarch64-linux-gnu.so +0 -0
  7. selectolax/parser.pyi +69 -38
  8. selectolax/utils.pxi +1 -1
  9. {selectolax-0.3.25.dist-info → selectolax-0.3.27.dist-info}/METADATA +1 -1
  10. {selectolax-0.3.25.dist-info → selectolax-0.3.27.dist-info}/RECORD +19 -19
  11. {selectolax-0.3.25.dist-info → selectolax-0.3.27.dist-info}/WHEEL +1 -1
  12. {selectolax-0.3.25.dist-info → selectolax-0.3.27.dist-info}/LICENSE +0 -0
  13. {selectolax-0.3.25.dist-info → selectolax-0.3.27.dist-info}/top_level.txt +0 -0
selectolax/__init__.py CHANGED
@@ -3,7 +3,7 @@
3
3
 
4
4
  __author__ = """Artem Golubin"""
5
5
  __email__ = 'me@rushter.com'
6
- __version__ = '0.3.25'
6
+ __version__ = '0.3.27'
7
7
 
8
8
  from . import parser
9
9
  from . import lexbor
selectolax/lexbor.c CHANGED
@@ -30,209 +30,209 @@
30
30
  "name": "selectolax.lexbor",
31
31
  "sources": [
32
32
  "selectolax/lexbor.pyx",
33
- "lexbor/source/lexbor/unicode/unicode.c",
34
- "lexbor/source/lexbor/unicode/idna.c",
35
- "lexbor/source/lexbor/core/serialize.c",
36
- "lexbor/source/lexbor/core/dtoa.c",
37
- "lexbor/source/lexbor/core/avl.c",
38
- "lexbor/source/lexbor/core/array.c",
39
- "lexbor/source/lexbor/core/hash.c",
40
- "lexbor/source/lexbor/core/diyfp.c",
41
- "lexbor/source/lexbor/core/mem.c",
42
- "lexbor/source/lexbor/core/dobject.c",
33
+ "lexbor/source/lexbor/tag/tag.c",
34
+ "lexbor/source/lexbor/ports/posix/lexbor/core/perf.c",
35
+ "lexbor/source/lexbor/ports/posix/lexbor/core/memory.c",
36
+ "lexbor/source/lexbor/ports/posix/lexbor/core/fs.c",
43
37
  "lexbor/source/lexbor/core/utils.c",
44
38
  "lexbor/source/lexbor/core/array_obj.c",
39
+ "lexbor/source/lexbor/core/conv.c",
40
+ "lexbor/source/lexbor/core/hash.c",
41
+ "lexbor/source/lexbor/core/array.c",
42
+ "lexbor/source/lexbor/core/serialize.c",
43
+ "lexbor/source/lexbor/core/shs.c",
44
+ "lexbor/source/lexbor/core/print.c",
45
45
  "lexbor/source/lexbor/core/bst.c",
46
+ "lexbor/source/lexbor/core/dtoa.c",
46
47
  "lexbor/source/lexbor/core/mraw.c",
47
- "lexbor/source/lexbor/core/print.c",
48
- "lexbor/source/lexbor/core/conv.c",
49
48
  "lexbor/source/lexbor/core/plog.c",
50
- "lexbor/source/lexbor/core/shs.c",
51
- "lexbor/source/lexbor/core/str.c",
52
- "lexbor/source/lexbor/core/strtod.c",
49
+ "lexbor/source/lexbor/core/diyfp.c",
53
50
  "lexbor/source/lexbor/core/in.c",
51
+ "lexbor/source/lexbor/core/strtod.c",
54
52
  "lexbor/source/lexbor/core/bst_map.c",
53
+ "lexbor/source/lexbor/core/dobject.c",
54
+ "lexbor/source/lexbor/core/avl.c",
55
+ "lexbor/source/lexbor/core/mem.c",
56
+ "lexbor/source/lexbor/core/str.c",
57
+ "lexbor/source/lexbor/ns/ns.c",
58
+ "lexbor/source/lexbor/dom/exception.c",
59
+ "lexbor/source/lexbor/dom/collection.c",
60
+ "lexbor/source/lexbor/dom/interface.c",
61
+ "lexbor/source/lexbor/dom/interfaces/document.c",
62
+ "lexbor/source/lexbor/dom/interfaces/cdata_section.c",
63
+ "lexbor/source/lexbor/dom/interfaces/attr.c",
64
+ "lexbor/source/lexbor/dom/interfaces/event_target.c",
65
+ "lexbor/source/lexbor/dom/interfaces/processing_instruction.c",
66
+ "lexbor/source/lexbor/dom/interfaces/document_type.c",
67
+ "lexbor/source/lexbor/dom/interfaces/comment.c",
68
+ "lexbor/source/lexbor/dom/interfaces/node.c",
69
+ "lexbor/source/lexbor/dom/interfaces/element.c",
70
+ "lexbor/source/lexbor/dom/interfaces/document_fragment.c",
71
+ "lexbor/source/lexbor/dom/interfaces/text.c",
72
+ "lexbor/source/lexbor/dom/interfaces/character_data.c",
73
+ "lexbor/source/lexbor/dom/interfaces/shadow_root.c",
74
+ "lexbor/source/lexbor/selectors/selectors.c",
75
+ "lexbor/source/lexbor/encoding/jis0208.c",
76
+ "lexbor/source/lexbor/encoding/single.c",
77
+ "lexbor/source/lexbor/encoding/iso_2022_jp_katakana.c",
78
+ "lexbor/source/lexbor/encoding/gb18030.c",
79
+ "lexbor/source/lexbor/encoding/encode.c",
80
+ "lexbor/source/lexbor/encoding/euc_kr.c",
81
+ "lexbor/source/lexbor/encoding/decode.c",
82
+ "lexbor/source/lexbor/encoding/jis0212.c",
83
+ "lexbor/source/lexbor/encoding/res.c",
84
+ "lexbor/source/lexbor/encoding/range.c",
85
+ "lexbor/source/lexbor/encoding/big5.c",
86
+ "lexbor/source/lexbor/encoding/encoding.c",
87
+ "lexbor/source/lexbor/css/stylesheet.c",
88
+ "lexbor/source/lexbor/css/state.c",
89
+ "lexbor/source/lexbor/css/value.c",
90
+ "lexbor/source/lexbor/css/unit.c",
91
+ "lexbor/source/lexbor/css/log.c",
92
+ "lexbor/source/lexbor/css/at_rule.c",
93
+ "lexbor/source/lexbor/css/rule.c",
94
+ "lexbor/source/lexbor/css/property.c",
95
+ "lexbor/source/lexbor/css/css.c",
96
+ "lexbor/source/lexbor/css/parser.c",
97
+ "lexbor/source/lexbor/css/declaration.c",
98
+ "lexbor/source/lexbor/css/selectors/state.c",
99
+ "lexbor/source/lexbor/css/selectors/selector.c",
100
+ "lexbor/source/lexbor/css/selectors/pseudo_state.c",
101
+ "lexbor/source/lexbor/css/selectors/pseudo.c",
102
+ "lexbor/source/lexbor/css/selectors/selectors.c",
103
+ "lexbor/source/lexbor/css/at_rule/state.c",
104
+ "lexbor/source/lexbor/css/property/state.c",
105
+ "lexbor/source/lexbor/css/syntax/state.c",
106
+ "lexbor/source/lexbor/css/syntax/syntax.c",
107
+ "lexbor/source/lexbor/css/syntax/parser.c",
108
+ "lexbor/source/lexbor/css/syntax/token.c",
109
+ "lexbor/source/lexbor/css/syntax/anb.c",
110
+ "lexbor/source/lexbor/css/syntax/tokenizer.c",
111
+ "lexbor/source/lexbor/css/syntax/tokenizer/error.c",
55
112
  "lexbor/source/lexbor/html/serialize.c",
56
- "lexbor/source/lexbor/html/tokenizer.c",
113
+ "lexbor/source/lexbor/html/node.c",
114
+ "lexbor/source/lexbor/html/parser.c",
115
+ "lexbor/source/lexbor/html/token.c",
57
116
  "lexbor/source/lexbor/html/style.c",
58
- "lexbor/source/lexbor/html/tree.c",
117
+ "lexbor/source/lexbor/html/tokenizer.c",
59
118
  "lexbor/source/lexbor/html/interface.c",
60
- "lexbor/source/lexbor/html/token.c",
61
- "lexbor/source/lexbor/html/node.c",
119
+ "lexbor/source/lexbor/html/tree.c",
62
120
  "lexbor/source/lexbor/html/encoding.c",
63
121
  "lexbor/source/lexbor/html/token_attr.c",
64
- "lexbor/source/lexbor/html/parser.c",
65
- "lexbor/source/lexbor/html/tokenizer/state_doctype.c",
66
- "lexbor/source/lexbor/html/tokenizer/state_rawtext.c",
67
122
  "lexbor/source/lexbor/html/tokenizer/state.c",
68
123
  "lexbor/source/lexbor/html/tokenizer/state_script.c",
69
- "lexbor/source/lexbor/html/tokenizer/state_rcdata.c",
70
- "lexbor/source/lexbor/html/tokenizer/error.c",
71
124
  "lexbor/source/lexbor/html/tokenizer/state_comment.c",
72
- "lexbor/source/lexbor/html/tree/active_formatting.c",
73
- "lexbor/source/lexbor/html/tree/open_elements.c",
74
- "lexbor/source/lexbor/html/tree/template_insertion.c",
125
+ "lexbor/source/lexbor/html/tokenizer/error.c",
126
+ "lexbor/source/lexbor/html/tokenizer/state_rawtext.c",
127
+ "lexbor/source/lexbor/html/tokenizer/state_doctype.c",
128
+ "lexbor/source/lexbor/html/tokenizer/state_rcdata.c",
75
129
  "lexbor/source/lexbor/html/tree/error.c",
76
- "lexbor/source/lexbor/html/tree/insertion_mode/initial.c",
130
+ "lexbor/source/lexbor/html/tree/template_insertion.c",
131
+ "lexbor/source/lexbor/html/tree/open_elements.c",
132
+ "lexbor/source/lexbor/html/tree/active_formatting.c",
133
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_cell.c",
77
134
  "lexbor/source/lexbor/html/tree/insertion_mode/before_head.c",
135
+ "lexbor/source/lexbor/html/tree/insertion_mode/initial.c",
78
136
  "lexbor/source/lexbor/html/tree/insertion_mode/before_html.c",
79
- "lexbor/source/lexbor/html/tree/insertion_mode/after_head.c",
137
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_select_in_table.c",
138
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_caption.c",
139
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_select.c",
140
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_body.c",
80
141
  "lexbor/source/lexbor/html/tree/insertion_mode/in_frameset.c",
81
- "lexbor/source/lexbor/html/tree/insertion_mode/in_head_noscript.c",
82
- "lexbor/source/lexbor/html/tree/insertion_mode/in_column_group.c",
142
+ "lexbor/source/lexbor/html/tree/insertion_mode/after_frameset.c",
143
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_row.c",
83
144
  "lexbor/source/lexbor/html/tree/insertion_mode/after_after_body.c",
84
- "lexbor/source/lexbor/html/tree/insertion_mode/in_table_text.c",
145
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_template.c",
146
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_column_group.c",
85
147
  "lexbor/source/lexbor/html/tree/insertion_mode/after_body.c",
86
- "lexbor/source/lexbor/html/tree/insertion_mode/text.c",
87
- "lexbor/source/lexbor/html/tree/insertion_mode/in_table_body.c",
88
- "lexbor/source/lexbor/html/tree/insertion_mode/in_body.c",
89
- "lexbor/source/lexbor/html/tree/insertion_mode/in_select_in_table.c",
90
- "lexbor/source/lexbor/html/tree/insertion_mode/in_row.c",
91
- "lexbor/source/lexbor/html/tree/insertion_mode/in_head.c",
92
148
  "lexbor/source/lexbor/html/tree/insertion_mode/foreign_content.c",
93
- "lexbor/source/lexbor/html/tree/insertion_mode/in_cell.c",
94
- "lexbor/source/lexbor/html/tree/insertion_mode/in_select.c",
95
- "lexbor/source/lexbor/html/tree/insertion_mode/after_after_frameset.c",
96
- "lexbor/source/lexbor/html/tree/insertion_mode/after_frameset.c",
149
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_head_noscript.c",
97
150
  "lexbor/source/lexbor/html/tree/insertion_mode/in_table.c",
98
- "lexbor/source/lexbor/html/tree/insertion_mode/in_template.c",
99
- "lexbor/source/lexbor/html/tree/insertion_mode/in_caption.c",
100
- "lexbor/source/lexbor/html/interfaces/html_element.c",
151
+ "lexbor/source/lexbor/html/tree/insertion_mode/after_after_frameset.c",
152
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_head.c",
153
+ "lexbor/source/lexbor/html/tree/insertion_mode/after_head.c",
154
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_table_body.c",
155
+ "lexbor/source/lexbor/html/tree/insertion_mode/in_table_text.c",
156
+ "lexbor/source/lexbor/html/tree/insertion_mode/text.c",
157
+ "lexbor/source/lexbor/html/interfaces/d_list_element.c",
158
+ "lexbor/source/lexbor/html/interfaces/marquee_element.c",
159
+ "lexbor/source/lexbor/html/interfaces/document.c",
160
+ "lexbor/source/lexbor/html/interfaces/dialog_element.c",
161
+ "lexbor/source/lexbor/html/interfaces/span_element.c",
162
+ "lexbor/source/lexbor/html/interfaces/canvas_element.c",
163
+ "lexbor/source/lexbor/html/interfaces/option_element.c",
164
+ "lexbor/source/lexbor/html/interfaces/table_element.c",
165
+ "lexbor/source/lexbor/html/interfaces/picture_element.c",
101
166
  "lexbor/source/lexbor/html/interfaces/form_element.c",
167
+ "lexbor/source/lexbor/html/interfaces/progress_element.c",
168
+ "lexbor/source/lexbor/html/interfaces/anchor_element.c",
169
+ "lexbor/source/lexbor/html/interfaces/audio_element.c",
170
+ "lexbor/source/lexbor/html/interfaces/data_element.c",
171
+ "lexbor/source/lexbor/html/interfaces/mod_element.c",
172
+ "lexbor/source/lexbor/html/interfaces/hr_element.c",
173
+ "lexbor/source/lexbor/html/interfaces/style_element.c",
174
+ "lexbor/source/lexbor/html/interfaces/video_element.c",
175
+ "lexbor/source/lexbor/html/interfaces/param_element.c",
176
+ "lexbor/source/lexbor/html/interfaces/meter_element.c",
177
+ "lexbor/source/lexbor/html/interfaces/iframe_element.c",
178
+ "lexbor/source/lexbor/html/interfaces/table_cell_element.c",
179
+ "lexbor/source/lexbor/html/interfaces/meta_element.c",
180
+ "lexbor/source/lexbor/html/interfaces/frame_element.c",
181
+ "lexbor/source/lexbor/html/interfaces/image_element.c",
102
182
  "lexbor/source/lexbor/html/interfaces/quote_element.c",
183
+ "lexbor/source/lexbor/html/interfaces/o_list_element.c",
184
+ "lexbor/source/lexbor/html/interfaces/head_element.c",
185
+ "lexbor/source/lexbor/html/interfaces/window.c",
186
+ "lexbor/source/lexbor/html/interfaces/table_caption_element.c",
187
+ "lexbor/source/lexbor/html/interfaces/track_element.c",
188
+ "lexbor/source/lexbor/html/interfaces/table_col_element.c",
189
+ "lexbor/source/lexbor/html/interfaces/table_row_element.c",
190
+ "lexbor/source/lexbor/html/interfaces/select_element.c",
191
+ "lexbor/source/lexbor/html/interfaces/label_element.c",
103
192
  "lexbor/source/lexbor/html/interfaces/base_element.c",
193
+ "lexbor/source/lexbor/html/interfaces/frame_set_element.c",
194
+ "lexbor/source/lexbor/html/interfaces/table_section_element.c",
104
195
  "lexbor/source/lexbor/html/interfaces/script_element.c",
196
+ "lexbor/source/lexbor/html/interfaces/heading_element.c",
197
+ "lexbor/source/lexbor/html/interfaces/element.c",
105
198
  "lexbor/source/lexbor/html/interfaces/paragraph_element.c",
106
- "lexbor/source/lexbor/html/interfaces/output_element.c",
107
- "lexbor/source/lexbor/html/interfaces/li_element.c",
108
- "lexbor/source/lexbor/html/interfaces/button_element.c",
109
- "lexbor/source/lexbor/html/interfaces/window.c",
110
- "lexbor/source/lexbor/html/interfaces/audio_element.c",
199
+ "lexbor/source/lexbor/html/interfaces/br_element.c",
200
+ "lexbor/source/lexbor/html/interfaces/body_element.c",
201
+ "lexbor/source/lexbor/html/interfaces/details_element.c",
202
+ "lexbor/source/lexbor/html/interfaces/embed_element.c",
203
+ "lexbor/source/lexbor/html/interfaces/html_element.c",
204
+ "lexbor/source/lexbor/html/interfaces/field_set_element.c",
205
+ "lexbor/source/lexbor/html/interfaces/title_element.c",
111
206
  "lexbor/source/lexbor/html/interfaces/link_element.c",
112
207
  "lexbor/source/lexbor/html/interfaces/object_element.c",
113
- "lexbor/source/lexbor/html/interfaces/select_element.c",
114
- "lexbor/source/lexbor/html/interfaces/frame_set_element.c",
115
- "lexbor/source/lexbor/html/interfaces/title_element.c",
116
- "lexbor/source/lexbor/html/interfaces/menu_element.c",
117
- "lexbor/source/lexbor/html/interfaces/dialog_element.c",
118
- "lexbor/source/lexbor/html/interfaces/option_element.c",
119
- "lexbor/source/lexbor/html/interfaces/o_list_element.c",
120
- "lexbor/source/lexbor/html/interfaces/table_cell_element.c",
121
- "lexbor/source/lexbor/html/interfaces/text_area_element.c",
122
- "lexbor/source/lexbor/html/interfaces/picture_element.c",
123
- "lexbor/source/lexbor/html/interfaces/meta_element.c",
124
- "lexbor/source/lexbor/html/interfaces/video_element.c",
125
- "lexbor/source/lexbor/html/interfaces/track_element.c",
126
- "lexbor/source/lexbor/html/interfaces/data_list_element.c",
127
208
  "lexbor/source/lexbor/html/interfaces/directory_element.c",
128
- "lexbor/source/lexbor/html/interfaces/table_element.c",
129
- "lexbor/source/lexbor/html/interfaces/table_row_element.c",
130
- "lexbor/source/lexbor/html/interfaces/field_set_element.c",
131
- "lexbor/source/lexbor/html/interfaces/media_element.c",
132
- "lexbor/source/lexbor/html/interfaces/area_element.c",
133
- "lexbor/source/lexbor/html/interfaces/font_element.c",
134
- "lexbor/source/lexbor/html/interfaces/table_caption_element.c",
135
- "lexbor/source/lexbor/html/interfaces/document.c",
136
- "lexbor/source/lexbor/html/interfaces/source_element.c",
137
- "lexbor/source/lexbor/html/interfaces/style_element.c",
138
- "lexbor/source/lexbor/html/interfaces/frame_element.c",
139
- "lexbor/source/lexbor/html/interfaces/d_list_element.c",
140
- "lexbor/source/lexbor/html/interfaces/progress_element.c",
141
- "lexbor/source/lexbor/html/interfaces/time_element.c",
142
209
  "lexbor/source/lexbor/html/interfaces/map_element.c",
143
- "lexbor/source/lexbor/html/interfaces/mod_element.c",
144
- "lexbor/source/lexbor/html/interfaces/pre_element.c",
145
- "lexbor/source/lexbor/html/interfaces/unknown_element.c",
146
- "lexbor/source/lexbor/html/interfaces/table_col_element.c",
147
210
  "lexbor/source/lexbor/html/interfaces/input_element.c",
148
- "lexbor/source/lexbor/html/interfaces/head_element.c",
211
+ "lexbor/source/lexbor/html/interfaces/div_element.c",
212
+ "lexbor/source/lexbor/html/interfaces/time_element.c",
213
+ "lexbor/source/lexbor/html/interfaces/unknown_element.c",
214
+ "lexbor/source/lexbor/html/interfaces/opt_group_element.c",
215
+ "lexbor/source/lexbor/html/interfaces/pre_element.c",
216
+ "lexbor/source/lexbor/html/interfaces/font_element.c",
149
217
  "lexbor/source/lexbor/html/interfaces/slot_element.c",
150
- "lexbor/source/lexbor/html/interfaces/image_element.c",
218
+ "lexbor/source/lexbor/html/interfaces/source_element.c",
219
+ "lexbor/source/lexbor/html/interfaces/li_element.c",
220
+ "lexbor/source/lexbor/html/interfaces/output_element.c",
221
+ "lexbor/source/lexbor/html/interfaces/area_element.c",
222
+ "lexbor/source/lexbor/html/interfaces/text_area_element.c",
223
+ "lexbor/source/lexbor/html/interfaces/legend_element.c",
224
+ "lexbor/source/lexbor/html/interfaces/button_element.c",
225
+ "lexbor/source/lexbor/html/interfaces/menu_element.c",
226
+ "lexbor/source/lexbor/html/interfaces/data_list_element.c",
151
227
  "lexbor/source/lexbor/html/interfaces/u_list_element.c",
152
228
  "lexbor/source/lexbor/html/interfaces/template_element.c",
153
- "lexbor/source/lexbor/html/interfaces/meter_element.c",
154
- "lexbor/source/lexbor/html/interfaces/param_element.c",
155
- "lexbor/source/lexbor/html/interfaces/anchor_element.c",
156
- "lexbor/source/lexbor/html/interfaces/canvas_element.c",
157
- "lexbor/source/lexbor/html/interfaces/opt_group_element.c",
158
- "lexbor/source/lexbor/html/interfaces/label_element.c",
159
- "lexbor/source/lexbor/html/interfaces/details_element.c",
160
- "lexbor/source/lexbor/html/interfaces/span_element.c",
161
- "lexbor/source/lexbor/html/interfaces/legend_element.c",
162
- "lexbor/source/lexbor/html/interfaces/embed_element.c",
163
- "lexbor/source/lexbor/html/interfaces/heading_element.c",
164
- "lexbor/source/lexbor/html/interfaces/hr_element.c",
165
- "lexbor/source/lexbor/html/interfaces/element.c",
166
- "lexbor/source/lexbor/html/interfaces/div_element.c",
167
- "lexbor/source/lexbor/html/interfaces/data_element.c",
168
- "lexbor/source/lexbor/html/interfaces/br_element.c",
169
- "lexbor/source/lexbor/html/interfaces/body_element.c",
170
- "lexbor/source/lexbor/html/interfaces/marquee_element.c",
171
- "lexbor/source/lexbor/html/interfaces/table_section_element.c",
172
- "lexbor/source/lexbor/html/interfaces/iframe_element.c",
173
- "lexbor/source/lexbor/ns/ns.c",
174
- "lexbor/source/lexbor/url/url.c",
229
+ "lexbor/source/lexbor/html/interfaces/media_element.c",
175
230
  "lexbor/source/lexbor/punycode/punycode.c",
176
- "lexbor/source/lexbor/selectors/selectors.c",
177
- "lexbor/source/lexbor/ports/posix/lexbor/core/fs.c",
178
- "lexbor/source/lexbor/ports/posix/lexbor/core/memory.c",
179
- "lexbor/source/lexbor/ports/posix/lexbor/core/perf.c",
180
- "lexbor/source/lexbor/dom/exception.c",
181
- "lexbor/source/lexbor/dom/collection.c",
182
- "lexbor/source/lexbor/dom/interface.c",
183
- "lexbor/source/lexbor/dom/interfaces/comment.c",
184
- "lexbor/source/lexbor/dom/interfaces/document_type.c",
185
- "lexbor/source/lexbor/dom/interfaces/character_data.c",
186
- "lexbor/source/lexbor/dom/interfaces/event_target.c",
187
- "lexbor/source/lexbor/dom/interfaces/attr.c",
188
- "lexbor/source/lexbor/dom/interfaces/text.c",
189
- "lexbor/source/lexbor/dom/interfaces/document.c",
190
- "lexbor/source/lexbor/dom/interfaces/shadow_root.c",
191
- "lexbor/source/lexbor/dom/interfaces/processing_instruction.c",
192
- "lexbor/source/lexbor/dom/interfaces/node.c",
193
- "lexbor/source/lexbor/dom/interfaces/cdata_section.c",
194
- "lexbor/source/lexbor/dom/interfaces/element.c",
195
- "lexbor/source/lexbor/dom/interfaces/document_fragment.c",
196
- "lexbor/source/lexbor/css/rule.c",
197
- "lexbor/source/lexbor/css/property.c",
198
- "lexbor/source/lexbor/css/stylesheet.c",
199
- "lexbor/source/lexbor/css/state.c",
200
- "lexbor/source/lexbor/css/at_rule.c",
201
- "lexbor/source/lexbor/css/value.c",
202
- "lexbor/source/lexbor/css/declaration.c",
203
- "lexbor/source/lexbor/css/unit.c",
204
- "lexbor/source/lexbor/css/parser.c",
205
- "lexbor/source/lexbor/css/css.c",
206
- "lexbor/source/lexbor/css/log.c",
207
- "lexbor/source/lexbor/css/at_rule/state.c",
208
- "lexbor/source/lexbor/css/property/state.c",
209
- "lexbor/source/lexbor/css/selectors/selectors.c",
210
- "lexbor/source/lexbor/css/selectors/pseudo_state.c",
211
- "lexbor/source/lexbor/css/selectors/pseudo.c",
212
- "lexbor/source/lexbor/css/selectors/state.c",
213
- "lexbor/source/lexbor/css/selectors/selector.c",
214
- "lexbor/source/lexbor/css/syntax/syntax.c",
215
- "lexbor/source/lexbor/css/syntax/tokenizer.c",
216
- "lexbor/source/lexbor/css/syntax/token.c",
217
- "lexbor/source/lexbor/css/syntax/state.c",
218
- "lexbor/source/lexbor/css/syntax/anb.c",
219
- "lexbor/source/lexbor/css/syntax/parser.c",
220
- "lexbor/source/lexbor/css/syntax/tokenizer/error.c",
221
- "lexbor/source/lexbor/encoding/res.c",
222
- "lexbor/source/lexbor/encoding/euc_kr.c",
223
- "lexbor/source/lexbor/encoding/range.c",
224
- "lexbor/source/lexbor/encoding/jis0208.c",
225
- "lexbor/source/lexbor/encoding/iso_2022_jp_katakana.c",
226
- "lexbor/source/lexbor/encoding/big5.c",
227
- "lexbor/source/lexbor/encoding/decode.c",
228
- "lexbor/source/lexbor/encoding/encode.c",
229
- "lexbor/source/lexbor/encoding/single.c",
230
- "lexbor/source/lexbor/encoding/jis0212.c",
231
- "lexbor/source/lexbor/encoding/gb18030.c",
232
- "lexbor/source/lexbor/encoding/encoding.c",
233
- "lexbor/source/lexbor/tag/tag.c",
234
231
  "lexbor/source/lexbor/utils/warc.c",
235
- "lexbor/source/lexbor/utils/http.c"
232
+ "lexbor/source/lexbor/utils/http.c",
233
+ "lexbor/source/lexbor/unicode/unicode.c",
234
+ "lexbor/source/lexbor/unicode/idna.c",
235
+ "lexbor/source/lexbor/url/url.c"
236
236
  ]
237
237
  },
238
238
  "module_name": "selectolax.lexbor"
@@ -3494,7 +3494,6 @@ static const char __pyx_k_AssertionError[] = "AssertionError";
3494
3494
  static const char __pyx_k_LexborNode_css[] = "LexborNode.css";
3495
3495
  static const char __pyx_k_LexborSelector[] = "LexborSelector";
3496
3496
  static const char __pyx_k_fused_sigindex[] = "_fused_sigindex";
3497
- static const char __pyx_k_html_body_head[] = "<html|<body|<head";
3498
3497
  static const char __pyx_k_parse_fragment[] = "parse_fragment";
3499
3498
  static const char __pyx_k_LexborNode_iter[] = "LexborNode.iter";
3500
3499
  static const char __pyx_k_LexborNode_text[] = "LexborNode.text";
@@ -3518,6 +3517,7 @@ static const char __pyx_k_LexborNode_unwrap[] = "LexborNode.unwrap";
3518
3517
  static const char __pyx_k_any_text_contains[] = "any_text_contains";
3519
3518
  static const char __pyx_k_do_parse_fragment[] = "do_parse_fragment";
3520
3519
  static const char __pyx_k_get_fragment_type[] = "get_fragment_type";
3520
+ static const char __pyx_k_html_body_head_er[] = "<html|<body|<head(?!er)";
3521
3521
  static const char __pyx_k_selectolax_lexbor[] = "selectolax.lexbor";
3522
3522
  static const char __pyx_k_Can_t_extract_text[] = "Can't extract text";
3523
3523
  static const char __pyx_k_LexborSelector_css[] = "LexborSelector.css";
@@ -4160,7 +4160,7 @@ typedef struct {
4160
4160
  PyObject *__pyx_n_u_head_and_body;
4161
4161
  PyObject *__pyx_n_s_html;
4162
4162
  PyObject *__pyx_kp_u_html_2;
4163
- PyObject *__pyx_kp_u_html_body_head;
4163
+ PyObject *__pyx_kp_u_html_body_head_er;
4164
4164
  PyObject *__pyx_n_s_html_len;
4165
4165
  PyObject *__pyx_n_s_html_re;
4166
4166
  PyObject *__pyx_n_s_i;
@@ -4724,7 +4724,7 @@ static int __pyx_m_clear(PyObject *m) {
4724
4724
  Py_CLEAR(clear_module_state->__pyx_n_u_head_and_body);
4725
4725
  Py_CLEAR(clear_module_state->__pyx_n_s_html);
4726
4726
  Py_CLEAR(clear_module_state->__pyx_kp_u_html_2);
4727
- Py_CLEAR(clear_module_state->__pyx_kp_u_html_body_head);
4727
+ Py_CLEAR(clear_module_state->__pyx_kp_u_html_body_head_er);
4728
4728
  Py_CLEAR(clear_module_state->__pyx_n_s_html_len);
4729
4729
  Py_CLEAR(clear_module_state->__pyx_n_s_html_re);
4730
4730
  Py_CLEAR(clear_module_state->__pyx_n_s_i);
@@ -5266,7 +5266,7 @@ static int __pyx_m_traverse(PyObject *m, visitproc visit, void *arg) {
5266
5266
  Py_VISIT(traverse_module_state->__pyx_n_u_head_and_body);
5267
5267
  Py_VISIT(traverse_module_state->__pyx_n_s_html);
5268
5268
  Py_VISIT(traverse_module_state->__pyx_kp_u_html_2);
5269
- Py_VISIT(traverse_module_state->__pyx_kp_u_html_body_head);
5269
+ Py_VISIT(traverse_module_state->__pyx_kp_u_html_body_head_er);
5270
5270
  Py_VISIT(traverse_module_state->__pyx_n_s_html_len);
5271
5271
  Py_VISIT(traverse_module_state->__pyx_n_s_html_re);
5272
5272
  Py_VISIT(traverse_module_state->__pyx_n_s_i);
@@ -5908,7 +5908,7 @@ static int __pyx_m_traverse(PyObject *m, visitproc visit, void *arg) {
5908
5908
  #define __pyx_n_u_head_and_body __pyx_mstate_global->__pyx_n_u_head_and_body
5909
5909
  #define __pyx_n_s_html __pyx_mstate_global->__pyx_n_s_html
5910
5910
  #define __pyx_kp_u_html_2 __pyx_mstate_global->__pyx_kp_u_html_2
5911
- #define __pyx_kp_u_html_body_head __pyx_mstate_global->__pyx_kp_u_html_body_head
5911
+ #define __pyx_kp_u_html_body_head_er __pyx_mstate_global->__pyx_kp_u_html_body_head_er
5912
5912
  #define __pyx_n_s_html_len __pyx_mstate_global->__pyx_n_s_html_len
5913
5913
  #define __pyx_n_s_html_re __pyx_mstate_global->__pyx_n_s_html_re
5914
5914
  #define __pyx_n_s_i __pyx_mstate_global->__pyx_n_s_i
@@ -7298,7 +7298,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_4get_fragment_type(CYTHON_UNUSED
7298
7298
  * tree = parser_cls(html)
7299
7299
  *
7300
7300
  * import re # <<<<<<<<<<<<<<
7301
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE)
7301
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE)
7302
7302
  *
7303
7303
  */
7304
7304
  __pyx_t_3 = __Pyx_ImportDottedModule(__pyx_n_s_re, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 36, __pyx_L1_error)
@@ -7309,7 +7309,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_4get_fragment_type(CYTHON_UNUSED
7309
7309
  /* "selectolax/utils.pxi":37
7310
7310
  *
7311
7311
  * import re
7312
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE) # <<<<<<<<<<<<<<
7312
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE) # <<<<<<<<<<<<<<
7313
7313
  *
7314
7314
  * has_html = False
7315
7315
  */
@@ -7332,7 +7332,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_4get_fragment_type(CYTHON_UNUSED
7332
7332
  }
7333
7333
  #endif
7334
7334
  {
7335
- PyObject *__pyx_callargs[3] = {__pyx_t_7, __pyx_kp_u_html_body_head, __pyx_t_5};
7335
+ PyObject *__pyx_callargs[3] = {__pyx_t_7, __pyx_kp_u_html_body_head_er, __pyx_t_5};
7336
7336
  __pyx_t_3 = __Pyx_PyObject_FastCall(__pyx_t_4, __pyx_callargs+1-__pyx_t_6, 2+__pyx_t_6);
7337
7337
  __Pyx_XDECREF(__pyx_t_7); __pyx_t_7 = 0;
7338
7338
  __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
@@ -7344,7 +7344,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_4get_fragment_type(CYTHON_UNUSED
7344
7344
  __pyx_t_3 = 0;
7345
7345
 
7346
7346
  /* "selectolax/utils.pxi":39
7347
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE)
7347
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE)
7348
7348
  *
7349
7349
  * has_html = False # <<<<<<<<<<<<<<
7350
7350
  * has_head = False
@@ -33919,7 +33919,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_12get_fragment_type(CYTHON_UNUSED
33919
33919
  * tree = parser_cls(html)
33920
33920
  *
33921
33921
  * import re # <<<<<<<<<<<<<<
33922
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE)
33922
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE)
33923
33923
  *
33924
33924
  */
33925
33925
  __pyx_t_3 = __Pyx_ImportDottedModule(__pyx_n_s_re, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(8, 36, __pyx_L1_error)
@@ -33930,7 +33930,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_12get_fragment_type(CYTHON_UNUSED
33930
33930
  /* "selectolax/utils.pxi":37
33931
33931
  *
33932
33932
  * import re
33933
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE) # <<<<<<<<<<<<<<
33933
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE) # <<<<<<<<<<<<<<
33934
33934
  *
33935
33935
  * has_html = False
33936
33936
  */
@@ -33953,7 +33953,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_12get_fragment_type(CYTHON_UNUSED
33953
33953
  }
33954
33954
  #endif
33955
33955
  {
33956
- PyObject *__pyx_callargs[3] = {__pyx_t_7, __pyx_kp_u_html_body_head, __pyx_t_5};
33956
+ PyObject *__pyx_callargs[3] = {__pyx_t_7, __pyx_kp_u_html_body_head_er, __pyx_t_5};
33957
33957
  __pyx_t_3 = __Pyx_PyObject_FastCall(__pyx_t_4, __pyx_callargs+1-__pyx_t_6, 2+__pyx_t_6);
33958
33958
  __Pyx_XDECREF(__pyx_t_7); __pyx_t_7 = 0;
33959
33959
  __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
@@ -33965,7 +33965,7 @@ static PyObject *__pyx_pf_10selectolax_6lexbor_12get_fragment_type(CYTHON_UNUSED
33965
33965
  __pyx_t_3 = 0;
33966
33966
 
33967
33967
  /* "selectolax/utils.pxi":39
33968
- * html_re = re.compile(r"<html|<body|<head", re.IGNORECASE)
33968
+ * html_re = re.compile(r"<html|<body|<head(?!er)", re.IGNORECASE)
33969
33969
  *
33970
33970
  * has_html = False # <<<<<<<<<<<<<<
33971
33971
  * has_head = False
@@ -43679,7 +43679,7 @@ static int __Pyx_CreateStringTabAndInitStrings(void) {
43679
43679
  {&__pyx_n_u_head_and_body, __pyx_k_head_and_body, sizeof(__pyx_k_head_and_body), 0, 1, 0, 1},
43680
43680
  {&__pyx_n_s_html, __pyx_k_html, sizeof(__pyx_k_html), 0, 0, 1, 1},
43681
43681
  {&__pyx_kp_u_html_2, __pyx_k_html_2, sizeof(__pyx_k_html_2), 0, 1, 0, 0},
43682
- {&__pyx_kp_u_html_body_head, __pyx_k_html_body_head, sizeof(__pyx_k_html_body_head), 0, 1, 0, 0},
43682
+ {&__pyx_kp_u_html_body_head_er, __pyx_k_html_body_head_er, sizeof(__pyx_k_html_body_head_er), 0, 1, 0, 0},
43683
43683
  {&__pyx_n_s_html_len, __pyx_k_html_len, sizeof(__pyx_k_html_len), 0, 0, 1, 1},
43684
43684
  {&__pyx_n_s_html_re, __pyx_k_html_re, sizeof(__pyx_k_html_re), 0, 0, 1, 1},
43685
43685
  {&__pyx_n_s_i, __pyx_k_i, sizeof(__pyx_k_i), 0, 0, 1, 1},