Lines Matching refs:data

34                               const lxb_char_t *data, const lxb_char_t *end);
38 const lxb_char_t *data,
44 const lxb_char_t *data,
49 const lxb_char_t *data,
54 const lxb_char_t *data,
60 const lxb_char_t *data,
65 const lxb_char_t *data,
70 const lxb_char_t *data,
75 const lxb_char_t *data,
80 const lxb_char_t *data,
85 const lxb_char_t *data,
90 const lxb_char_t *data,
95 const lxb_char_t *data,
100 const lxb_char_t *data,
106 const lxb_char_t *data,
111 const lxb_char_t *data,
116 const lxb_char_t *data,
121 const lxb_char_t *data,
127 const lxb_char_t *data,
132 const lxb_char_t *data,
137 const lxb_char_t *data,
142 const lxb_char_t *data,
147 const lxb_char_t *data,
152 const lxb_char_t *data,
157 const lxb_char_t *data,
162 const lxb_char_t *data,
167 const lxb_char_t *data,
172 const lxb_char_t *data,
177 const lxb_char_t *data,
182 const lxb_char_t *data,
187 const lxb_char_t *data,
192 const lxb_char_t *data,
196 lxb_html_tokenizer_state_to_ascii_utf_8(size_t codepoint, lxb_char_t *data);
204 const lxb_char_t *data, in lxb_html_tokenizer_state_data_before() argument
208 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_data_before()
217 return data; in lxb_html_tokenizer_state_data_before()
225 const lxb_char_t *data, const lxb_char_t *end) in lxb_html_tokenizer_state_data() argument
227 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_data()
229 while (data != end) { in lxb_html_tokenizer_state_data()
230 switch (*data) { in lxb_html_tokenizer_state_data()
233 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_data()
234 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_data()
237 return (data + 1); in lxb_html_tokenizer_state_data()
241 lxb_html_tokenizer_state_append_data_m(tkz, data + 1); in lxb_html_tokenizer_state_data()
246 return data + 1; in lxb_html_tokenizer_state_data()
250 if (++data >= end) { in lxb_html_tokenizer_state_data()
251 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_data()
256 return data; in lxb_html_tokenizer_state_data()
259 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_data()
262 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_data()
264 if (*data != 0x0A) { in lxb_html_tokenizer_state_data()
265 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_data()
266 data--; in lxb_html_tokenizer_state_data()
285 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_data()
301 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_data()
306 data++; in lxb_html_tokenizer_state_data()
309 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_data()
311 return data; in lxb_html_tokenizer_state_data()
319 const lxb_char_t *data, in lxb_html_tokenizer_state_plaintext_before() argument
323 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_plaintext_before()
330 return data; in lxb_html_tokenizer_state_plaintext_before()
338 const lxb_char_t *data, in lxb_html_tokenizer_state_plaintext() argument
341 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_plaintext()
343 while (data != end) { in lxb_html_tokenizer_state_plaintext()
344 switch (*data) { in lxb_html_tokenizer_state_plaintext()
347 if (++data >= end) { in lxb_html_tokenizer_state_plaintext()
348 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_plaintext()
353 return data; in lxb_html_tokenizer_state_plaintext()
356 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_plaintext()
359 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_plaintext()
361 if (*data != 0x0A) { in lxb_html_tokenizer_state_plaintext()
362 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_plaintext()
363 data--; in lxb_html_tokenizer_state_plaintext()
373 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_plaintext()
386 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_plaintext()
389 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_plaintext()
394 data++; in lxb_html_tokenizer_state_plaintext()
397 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_plaintext()
399 return data; in lxb_html_tokenizer_state_plaintext()
407 const lxb_char_t *data, const lxb_char_t *end) in lxb_html_tokenizer_state_tag_open() argument
410 if (lexbor_str_res_alpha_character[ *data ] != LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_tag_open()
414 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_tag_open()
416 return data; in lxb_html_tokenizer_state_tag_open()
420 else if (*data == 0x2F) { in lxb_html_tokenizer_state_tag_open()
423 return (data + 1); in lxb_html_tokenizer_state_tag_open()
427 else if (*data == 0x21) { in lxb_html_tokenizer_state_tag_open()
432 return (data + 1); in lxb_html_tokenizer_state_tag_open()
436 else if (*data == 0x3F) { in lxb_html_tokenizer_state_tag_open()
440 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_tag_open()
442 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_tag_open()
445 return data; in lxb_html_tokenizer_state_tag_open()
449 else if (*data == 0x00) { in lxb_html_tokenizer_state_tag_open()
465 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_tag_open()
470 return data; in lxb_html_tokenizer_state_tag_open()
478 const lxb_char_t *data, in lxb_html_tokenizer_state_end_tag_open() argument
482 if (lexbor_str_res_alpha_character[ *data ] != LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_end_tag_open()
486 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_end_tag_open()
490 return data; in lxb_html_tokenizer_state_end_tag_open()
494 else if (*data == 0x3E) { in lxb_html_tokenizer_state_end_tag_open()
497 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_end_tag_open()
500 return (data + 1); in lxb_html_tokenizer_state_end_tag_open()
504 else if (*data == 0x00) { in lxb_html_tokenizer_state_end_tag_open()
520 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_end_tag_open()
524 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_end_tag_open()
526 return data; in lxb_html_tokenizer_state_end_tag_open()
534 const lxb_char_t *data, const lxb_char_t *end) in lxb_html_tokenizer_state_tag_name() argument
536 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_tag_name()
538 while (data != end) { in lxb_html_tokenizer_state_tag_name()
539 switch (*data) { in lxb_html_tokenizer_state_tag_name()
552 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_tag_name()
554 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_tag_name()
557 return (data + 1); in lxb_html_tokenizer_state_tag_name()
561 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_tag_name()
563 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_tag_name()
566 return (data + 1); in lxb_html_tokenizer_state_tag_name()
572 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_tag_name()
574 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_tag_name()
577 return (data + 1); in lxb_html_tokenizer_state_tag_name()
590 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_tag_name()
591 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_tag_name()
594 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_tag_name()
602 data++; in lxb_html_tokenizer_state_tag_name()
605 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_tag_name()
607 return data; in lxb_html_tokenizer_state_tag_name()
615 const lxb_char_t *data, in lxb_html_tokenizer_state_before_attribute_name() argument
620 while (data != end) { in lxb_html_tokenizer_state_before_attribute_name()
621 switch (*data) { in lxb_html_tokenizer_state_before_attribute_name()
639 lxb_html_tokenizer_state_token_attr_set_name_begin(tkz, data); in lxb_html_tokenizer_state_before_attribute_name()
641 lxb_html_tokenizer_state_append_m(tkz, data, 1); in lxb_html_tokenizer_state_before_attribute_name()
643 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_before_attribute_name()
647 return (data + 1); in lxb_html_tokenizer_state_before_attribute_name()
656 return data; in lxb_html_tokenizer_state_before_attribute_name()
662 return data; in lxb_html_tokenizer_state_before_attribute_name()
669 lxb_html_tokenizer_state_token_attr_set_name_begin(tkz, data); in lxb_html_tokenizer_state_before_attribute_name()
672 return data; in lxb_html_tokenizer_state_before_attribute_name()
675 data++; in lxb_html_tokenizer_state_before_attribute_name()
678 return data; in lxb_html_tokenizer_state_before_attribute_name()
686 const lxb_char_t *data, in lxb_html_tokenizer_state_attribute_name() argument
689 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_name()
691 while (data != end) { in lxb_html_tokenizer_state_attribute_name()
692 switch (*data) { in lxb_html_tokenizer_state_attribute_name()
709 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_name()
711 lxb_html_tokenizer_state_token_attr_set_name_end(tkz, data); in lxb_html_tokenizer_state_attribute_name()
714 return data; in lxb_html_tokenizer_state_attribute_name()
725 return data; in lxb_html_tokenizer_state_attribute_name()
728 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_name()
729 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_name()
732 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_attribute_name()
738 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_name()
740 lxb_html_tokenizer_state_token_attr_set_name_end(tkz, data); in lxb_html_tokenizer_state_attribute_name()
743 return (data + 1); in lxb_html_tokenizer_state_attribute_name()
753 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_attribute_name()
761 data++; in lxb_html_tokenizer_state_attribute_name()
764 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_name()
766 return data; in lxb_html_tokenizer_state_attribute_name()
774 const lxb_char_t *data, in lxb_html_tokenizer_state_after_attribute_name() argument
779 while (data != end) { in lxb_html_tokenizer_state_after_attribute_name()
780 switch (*data) { in lxb_html_tokenizer_state_after_attribute_name()
798 return (data + 1); in lxb_html_tokenizer_state_after_attribute_name()
803 return (data + 1); in lxb_html_tokenizer_state_after_attribute_name()
811 return (data + 1); in lxb_html_tokenizer_state_after_attribute_name()
823 lxb_html_tokenizer_state_token_attr_set_name_begin(tkz, data); in lxb_html_tokenizer_state_after_attribute_name()
826 return data; in lxb_html_tokenizer_state_after_attribute_name()
829 data++; in lxb_html_tokenizer_state_after_attribute_name()
832 return data; in lxb_html_tokenizer_state_after_attribute_name()
840 const lxb_char_t *data, in lxb_html_tokenizer_state_before_attribute_value() argument
843 while (data != end) { in lxb_html_tokenizer_state_before_attribute_value()
844 switch (*data) { in lxb_html_tokenizer_state_before_attribute_value()
864 return (data + 1); in lxb_html_tokenizer_state_before_attribute_value()
871 return (data + 1); in lxb_html_tokenizer_state_before_attribute_value()
877 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_before_attribute_value()
882 return (data + 1); in lxb_html_tokenizer_state_before_attribute_value()
886 return data; in lxb_html_tokenizer_state_before_attribute_value()
889 data++; in lxb_html_tokenizer_state_before_attribute_value()
892 return data; in lxb_html_tokenizer_state_before_attribute_value()
900 const lxb_char_t *data, in lxb_html_tokenizer_state_attribute_value_double_quoted() argument
904 lxb_html_tokenizer_state_token_attr_set_value_begin(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
907 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
909 while (data != end) { in lxb_html_tokenizer_state_attribute_value_double_quoted()
910 switch (*data) { in lxb_html_tokenizer_state_attribute_value_double_quoted()
913 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
914 lxb_html_tokenizer_state_token_attr_set_value_end(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
920 return (data + 1); in lxb_html_tokenizer_state_attribute_value_double_quoted()
924 lxb_html_tokenizer_state_append_data_m(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_double_quoted()
929 return data + 1; in lxb_html_tokenizer_state_attribute_value_double_quoted()
933 if (++data >= end) { in lxb_html_tokenizer_state_attribute_value_double_quoted()
934 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_attribute_value_double_quoted()
939 return data; in lxb_html_tokenizer_state_attribute_value_double_quoted()
942 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
945 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_double_quoted()
947 if (*data != 0x0A) { in lxb_html_tokenizer_state_attribute_value_double_quoted()
948 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
949 data--; in lxb_html_tokenizer_state_attribute_value_double_quoted()
969 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
970 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_double_quoted()
973 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_attribute_value_double_quoted()
981 data++; in lxb_html_tokenizer_state_attribute_value_double_quoted()
984 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_double_quoted()
986 return data; in lxb_html_tokenizer_state_attribute_value_double_quoted()
994 const lxb_char_t *data, in lxb_html_tokenizer_state_attribute_value_single_quoted() argument
998 lxb_html_tokenizer_state_token_attr_set_value_begin(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1001 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1003 while (data != end) { in lxb_html_tokenizer_state_attribute_value_single_quoted()
1004 switch (*data) { in lxb_html_tokenizer_state_attribute_value_single_quoted()
1007 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1008 lxb_html_tokenizer_state_token_attr_set_value_end(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1014 return (data + 1); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1018 lxb_html_tokenizer_state_append_data_m(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1023 return data + 1; in lxb_html_tokenizer_state_attribute_value_single_quoted()
1027 if (++data >= end) { in lxb_html_tokenizer_state_attribute_value_single_quoted()
1028 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1033 return data; in lxb_html_tokenizer_state_attribute_value_single_quoted()
1036 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1039 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1041 if (*data != 0x0A) { in lxb_html_tokenizer_state_attribute_value_single_quoted()
1042 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1043 data--; in lxb_html_tokenizer_state_attribute_value_single_quoted()
1063 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1064 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1067 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_attribute_value_single_quoted()
1075 data++; in lxb_html_tokenizer_state_attribute_value_single_quoted()
1078 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_single_quoted()
1080 return data; in lxb_html_tokenizer_state_attribute_value_single_quoted()
1088 const lxb_char_t *data, in lxb_html_tokenizer_state_attribute_value_unquoted() argument
1092 lxb_html_tokenizer_state_token_attr_set_value_begin(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1095 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1097 while (data != end) { in lxb_html_tokenizer_state_attribute_value_unquoted()
1098 switch (*data) { in lxb_html_tokenizer_state_attribute_value_unquoted()
1111 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1112 lxb_html_tokenizer_state_token_attr_set_value_end(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1116 return (data + 1); in lxb_html_tokenizer_state_attribute_value_unquoted()
1120 lxb_html_tokenizer_state_append_data_m(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_unquoted()
1125 return data + 1; in lxb_html_tokenizer_state_attribute_value_unquoted()
1131 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1132 lxb_html_tokenizer_state_token_attr_set_value_end(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1137 return (data + 1); in lxb_html_tokenizer_state_attribute_value_unquoted()
1154 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1155 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_attribute_value_unquoted()
1158 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_attribute_value_unquoted()
1182 data++; in lxb_html_tokenizer_state_attribute_value_unquoted()
1185 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_attribute_value_unquoted()
1187 return data; in lxb_html_tokenizer_state_attribute_value_unquoted()
1195 const lxb_char_t *data, in lxb_html_tokenizer_state_after_attribute_value_quoted() argument
1198 switch (*data) { in lxb_html_tokenizer_state_after_attribute_value_quoted()
1213 return (data + 1); in lxb_html_tokenizer_state_after_attribute_value_quoted()
1219 return (data + 1); in lxb_html_tokenizer_state_after_attribute_value_quoted()
1227 return (data + 1); in lxb_html_tokenizer_state_after_attribute_value_quoted()
1239 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_after_attribute_value_quoted()
1244 return data; in lxb_html_tokenizer_state_after_attribute_value_quoted()
1247 return data; in lxb_html_tokenizer_state_after_attribute_value_quoted()
1252 lxb_html_tokenizer_state_cr(lxb_html_tokenizer_t *tkz, const lxb_char_t *data, in lxb_html_tokenizer_state_cr() argument
1257 if (*data == 0x0A) { in lxb_html_tokenizer_state_cr()
1258 data++; in lxb_html_tokenizer_state_cr()
1263 return data; in lxb_html_tokenizer_state_cr()
1271 const lxb_char_t *data, in lxb_html_tokenizer_state_self_closing_start_tag() argument
1274 switch (*data) { in lxb_html_tokenizer_state_self_closing_start_tag()
1282 return (data + 1); in lxb_html_tokenizer_state_self_closing_start_tag()
1294 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_self_closing_start_tag()
1299 return data; in lxb_html_tokenizer_state_self_closing_start_tag()
1302 return data; in lxb_html_tokenizer_state_self_closing_start_tag()
1310 const lxb_char_t *data, in lxb_html_tokenizer_state_bogus_comment_before() argument
1317 return data; in lxb_html_tokenizer_state_bogus_comment_before()
1325 const lxb_char_t *data, in lxb_html_tokenizer_state_bogus_comment() argument
1328 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1330 while (data != end) { in lxb_html_tokenizer_state_bogus_comment()
1331 switch (*data) { in lxb_html_tokenizer_state_bogus_comment()
1336 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1337 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1341 return (data + 1); in lxb_html_tokenizer_state_bogus_comment()
1345 if (++data >= end) { in lxb_html_tokenizer_state_bogus_comment()
1346 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_bogus_comment()
1351 return data; in lxb_html_tokenizer_state_bogus_comment()
1354 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1357 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_bogus_comment()
1359 if (*data != 0x0A) { in lxb_html_tokenizer_state_bogus_comment()
1360 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1361 data--; in lxb_html_tokenizer_state_bogus_comment()
1371 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1385 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_bogus_comment()
1387 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_bogus_comment()
1392 data++; in lxb_html_tokenizer_state_bogus_comment()
1395 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_bogus_comment()
1397 return data; in lxb_html_tokenizer_state_bogus_comment()
1405 const lxb_char_t *data, in lxb_html_tokenizer_state_markup_declaration_open() argument
1410 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_markup_declaration_open()
1414 if (*data == 0x2D) { in lxb_html_tokenizer_state_markup_declaration_open()
1415 if ((end - data) < 2) { in lxb_html_tokenizer_state_markup_declaration_open()
1417 return (data + 1); in lxb_html_tokenizer_state_markup_declaration_open()
1420 if (data[1] == 0x2D) { in lxb_html_tokenizer_state_markup_declaration_open()
1422 return (data + 2); in lxb_html_tokenizer_state_markup_declaration_open()
1429 else if (*data == 0x44 || *data == 0x64) { in lxb_html_tokenizer_state_markup_declaration_open()
1430 if ((end - data) < 7) { in lxb_html_tokenizer_state_markup_declaration_open()
1434 return data; in lxb_html_tokenizer_state_markup_declaration_open()
1437 if (lexbor_str_data_ncasecmp((lxb_char_t *) "doctype", data, 7)) { in lxb_html_tokenizer_state_markup_declaration_open()
1439 return (data + 7); in lxb_html_tokenizer_state_markup_declaration_open()
1446 else if (*data == 0x5B) { in lxb_html_tokenizer_state_markup_declaration_open()
1447 if ((end - data) < 7) { in lxb_html_tokenizer_state_markup_declaration_open()
1451 return data; in lxb_html_tokenizer_state_markup_declaration_open()
1454 if (lexbor_str_data_ncmp((lxb_char_t *) "[CDATA[", data, 7)) { in lxb_html_tokenizer_state_markup_declaration_open()
1458 data += 7; in lxb_html_tokenizer_state_markup_declaration_open()
1460 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_markup_declaration_open()
1464 return data; in lxb_html_tokenizer_state_markup_declaration_open()
1469 return data; in lxb_html_tokenizer_state_markup_declaration_open()
1479 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_markup_declaration_open()
1484 return data; in lxb_html_tokenizer_state_markup_declaration_open()
1493 const lxb_char_t *data, in lxb_html_tokenizer_state_markup_declaration_comment() argument
1497 if (*data == 0x2D) { in lxb_html_tokenizer_state_markup_declaration_comment()
1499 return (data + 1); in lxb_html_tokenizer_state_markup_declaration_comment()
1502 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_markup_declaration_comment()
1506 return data; in lxb_html_tokenizer_state_markup_declaration_comment()
1515 const lxb_char_t *data, in lxb_html_tokenizer_state_markup_declaration_doctype() argument
1519 pos = lexbor_str_data_ncasecmp_first(tkz->markup, data, (end - data)); in lxb_html_tokenizer_state_markup_declaration_doctype()
1522 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_markup_declaration_doctype()
1526 return data; in lxb_html_tokenizer_state_markup_declaration_doctype()
1530 data = (data + (pos - tkz->markup)); in lxb_html_tokenizer_state_markup_declaration_doctype()
1533 return data; in lxb_html_tokenizer_state_markup_declaration_doctype()
1547 const lxb_char_t *data, in lxb_html_tokenizer_state_markup_declaration_cdata() argument
1551 pos = lexbor_str_data_ncasecmp_first(tkz->markup, data, (end - data)); in lxb_html_tokenizer_state_markup_declaration_cdata()
1554 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_markup_declaration_cdata()
1558 return data; in lxb_html_tokenizer_state_markup_declaration_cdata()
1565 data = (data + (pos - tkz->markup)); in lxb_html_tokenizer_state_markup_declaration_cdata()
1568 return data; in lxb_html_tokenizer_state_markup_declaration_cdata()
1574 return data; in lxb_html_tokenizer_state_markup_declaration_cdata()
1587 const lxb_char_t *data, in lxb_html_tokenizer_state_cdata_section_before() argument
1591 lxb_html_tokenizer_state_token_set_begin(tkz, data); in lxb_html_tokenizer_state_cdata_section_before()
1601 return data; in lxb_html_tokenizer_state_cdata_section_before()
1609 const lxb_char_t *data, in lxb_html_tokenizer_state_cdata_section() argument
1612 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1614 while (data != end) { in lxb_html_tokenizer_state_cdata_section()
1615 switch (*data) { in lxb_html_tokenizer_state_cdata_section()
1618 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1619 lxb_html_tokenizer_state_token_set_end(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1622 return (data + 1); in lxb_html_tokenizer_state_cdata_section()
1626 if (++data >= end) { in lxb_html_tokenizer_state_cdata_section()
1627 lxb_html_tokenizer_state_append_data_m(tkz, data - 1); in lxb_html_tokenizer_state_cdata_section()
1632 return data; in lxb_html_tokenizer_state_cdata_section()
1635 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1638 lxb_html_tokenizer_state_begin_set(tkz, data + 1); in lxb_html_tokenizer_state_cdata_section()
1640 if (*data != 0x0A) { in lxb_html_tokenizer_state_cdata_section()
1641 lxb_html_tokenizer_state_begin_set(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1642 data--; in lxb_html_tokenizer_state_cdata_section()
1654 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1677 data++; in lxb_html_tokenizer_state_cdata_section()
1680 lxb_html_tokenizer_state_append_data_m(tkz, data); in lxb_html_tokenizer_state_cdata_section()
1682 return data; in lxb_html_tokenizer_state_cdata_section()
1690 const lxb_char_t *data, in lxb_html_tokenizer_state_cdata_section_bracket() argument
1694 if (*data == 0x5D) { in lxb_html_tokenizer_state_cdata_section_bracket()
1696 return (data + 1); in lxb_html_tokenizer_state_cdata_section_bracket()
1703 return data; in lxb_html_tokenizer_state_cdata_section_bracket()
1711 const lxb_char_t *data, in lxb_html_tokenizer_state_cdata_section_end() argument
1715 if (*data == 0x5D) { in lxb_html_tokenizer_state_cdata_section_end()
1716 lxb_html_tokenizer_state_append_m(tkz, data, 1); in lxb_html_tokenizer_state_cdata_section_end()
1717 return (data + 1); in lxb_html_tokenizer_state_cdata_section_end()
1720 else if (*data == 0x3E) { in lxb_html_tokenizer_state_cdata_section_end()
1726 return (data + 1); in lxb_html_tokenizer_state_cdata_section_end()
1733 return data; in lxb_html_tokenizer_state_cdata_section_end()
1741 const lxb_char_t *data, const lxb_char_t *end) in lxb_html_tokenizer_state_char_ref() argument
1745 return _lxb_html_tokenizer_state_char_ref(tkz, data, end); in lxb_html_tokenizer_state_char_ref()
1750 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_attr() argument
1755 return _lxb_html_tokenizer_state_char_ref(tkz, data, end); in lxb_html_tokenizer_state_char_ref_attr()
1760 const lxb_char_t *data, in _lxb_html_tokenizer_state_char_ref() argument
1764 if (lexbor_str_res_alphanumeric_character[ *data ] != LEXBOR_STR_RES_SLIP) { in _lxb_html_tokenizer_state_char_ref()
1771 return data; in _lxb_html_tokenizer_state_char_ref()
1774 else if (*data == 0x23) { in _lxb_html_tokenizer_state_char_ref()
1775 tkz->markup = data; in _lxb_html_tokenizer_state_char_ref()
1778 lxb_html_tokenizer_state_append_m(tkz, data, 1); in _lxb_html_tokenizer_state_char_ref()
1782 return (data + 1); in _lxb_html_tokenizer_state_char_ref()
1788 return data; in _lxb_html_tokenizer_state_char_ref()
1801 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_named() argument
1808 const lxb_char_t *begin = data; in lxb_html_tokenizer_state_char_ref_named()
1810 while (data < end) { in lxb_html_tokenizer_state_char_ref_named()
1812 entry, *data); in lxb_html_tokenizer_state_char_ref_named()
1814 lxb_html_tokenizer_state_append_m(tkz, begin, (data - begin)); in lxb_html_tokenizer_state_char_ref_named()
1819 tkz->entity_end = (tkz->pos + (data - begin)) - tkz->start; in lxb_html_tokenizer_state_char_ref_named()
1825 data++; in lxb_html_tokenizer_state_char_ref_named()
1832 return data; in lxb_html_tokenizer_state_char_ref_named()
1840 return data; in lxb_html_tokenizer_state_char_ref_named()
1856 if (*data == 0x3D in lxb_html_tokenizer_state_char_ref_named()
1857 || lexbor_str_res_alphanumeric_character[*data] != LEXBOR_STR_RES_SLIP) in lxb_html_tokenizer_state_char_ref_named()
1859 return data; in lxb_html_tokenizer_state_char_ref_named()
1864 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_char_ref_named()
1889 return data; in lxb_html_tokenizer_state_char_ref_named()
1897 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_ambiguous_ampersand() argument
1904 if (*data == 0x3B) { in lxb_html_tokenizer_state_char_ref_ambiguous_ampersand()
1905 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_char_ref_ambiguous_ampersand()
1911 return data; in lxb_html_tokenizer_state_char_ref_ambiguous_ampersand()
1919 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_numeric() argument
1928 if (*data == 0x78 || *data == 0x58) { in lxb_html_tokenizer_state_char_ref_numeric()
1929 lxb_html_tokenizer_state_append_m(tkz, data, 1); in lxb_html_tokenizer_state_char_ref_numeric()
1933 return (data + 1); in lxb_html_tokenizer_state_char_ref_numeric()
1938 return data; in lxb_html_tokenizer_state_char_ref_numeric()
1946 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_hexademical_start() argument
1950 if (lexbor_str_res_map_hex[ *data ] != LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_char_ref_hexademical_start()
1954 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_char_ref_hexademical_start()
1960 return data; in lxb_html_tokenizer_state_char_ref_hexademical_start()
1968 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_decimal_start() argument
1972 if (lexbor_str_res_map_num[ *data ] != LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_char_ref_decimal_start()
1976 lxb_html_tokenizer_error_add(tkz->parse_errors, data, in lxb_html_tokenizer_state_char_ref_decimal_start()
1982 return data; in lxb_html_tokenizer_state_char_ref_decimal_start()
1990 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_hexademical() argument
1993 while (data != end) { in lxb_html_tokenizer_state_char_ref_hexademical()
1994 if (lexbor_str_res_map_hex[ *data ] == LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_char_ref_hexademical()
1997 if (*data == ';') { in lxb_html_tokenizer_state_char_ref_hexademical()
1998 data++; in lxb_html_tokenizer_state_char_ref_hexademical()
2001 return lxb_html_tokenizer_state_char_ref_numeric_end(tkz, data, end); in lxb_html_tokenizer_state_char_ref_hexademical()
2006 tkz->entity_number |= lexbor_str_res_map_hex[ *data ]; in lxb_html_tokenizer_state_char_ref_hexademical()
2009 data++; in lxb_html_tokenizer_state_char_ref_hexademical()
2012 return data; in lxb_html_tokenizer_state_char_ref_hexademical()
2020 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_decimal() argument
2023 while (data != end) { in lxb_html_tokenizer_state_char_ref_decimal()
2024 if (lexbor_str_res_map_num[ *data ] == LEXBOR_STR_RES_SLIP) { in lxb_html_tokenizer_state_char_ref_decimal()
2027 if (*data == ';') { in lxb_html_tokenizer_state_char_ref_decimal()
2028 data++; in lxb_html_tokenizer_state_char_ref_decimal()
2031 return lxb_html_tokenizer_state_char_ref_numeric_end(tkz, data, end); in lxb_html_tokenizer_state_char_ref_decimal()
2035 tkz->entity_number = lexbor_str_res_map_num[ *data ] in lxb_html_tokenizer_state_char_ref_decimal()
2039 data++; in lxb_html_tokenizer_state_char_ref_decimal()
2042 return data; in lxb_html_tokenizer_state_char_ref_decimal()
2050 const lxb_char_t *data, in lxb_html_tokenizer_state_char_ref_numeric_end() argument
2119 return data; in lxb_html_tokenizer_state_char_ref_numeric_end()
2128 return data; in lxb_html_tokenizer_state_char_ref_numeric_end()
2132 lxb_html_tokenizer_state_to_ascii_utf_8(size_t codepoint, lxb_char_t *data) in lxb_html_tokenizer_state_to_ascii_utf_8() argument
2141 data[0] = (char) codepoint; in lxb_html_tokenizer_state_to_ascii_utf_8()
2147 data[0] = (char) (0xC0 | (codepoint >> 6 )); in lxb_html_tokenizer_state_to_ascii_utf_8()
2148 data[1] = (char) (0x80 | (codepoint & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2154 data[0] = (char) (0xE0 | ((codepoint >> 12))); in lxb_html_tokenizer_state_to_ascii_utf_8()
2155 data[1] = (char) (0x80 | ((codepoint >> 6 ) & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2156 data[2] = (char) (0x80 | ( codepoint & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2162 data[0] = (char) (0xF0 | ( codepoint >> 18)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2163 data[1] = (char) (0x80 | ((codepoint >> 12) & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2164 data[2] = (char) (0x80 | ((codepoint >> 6 ) & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()
2165 data[3] = (char) (0x80 | ( codepoint & 0x3F)); in lxb_html_tokenizer_state_to_ascii_utf_8()