nokolexbor 0.2.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (486) hide show
  1. checksums.yaml +7 -0
  2. data/ext/nokolexbor/config.h +186 -0
  3. data/ext/nokolexbor/extconf.rb +131 -0
  4. data/ext/nokolexbor/libxml/HTMLparser.h +320 -0
  5. data/ext/nokolexbor/libxml/SAX2.h +173 -0
  6. data/ext/nokolexbor/libxml/chvalid.h +230 -0
  7. data/ext/nokolexbor/libxml/debugXML.h +217 -0
  8. data/ext/nokolexbor/libxml/dict.h +81 -0
  9. data/ext/nokolexbor/libxml/encoding.h +232 -0
  10. data/ext/nokolexbor/libxml/entities.h +153 -0
  11. data/ext/nokolexbor/libxml/globals.h +529 -0
  12. data/ext/nokolexbor/libxml/hash.h +236 -0
  13. data/ext/nokolexbor/libxml/list.h +137 -0
  14. data/ext/nokolexbor/libxml/parser.h +1264 -0
  15. data/ext/nokolexbor/libxml/parserInternals.h +641 -0
  16. data/ext/nokolexbor/libxml/pattern.h +100 -0
  17. data/ext/nokolexbor/libxml/threads.h +94 -0
  18. data/ext/nokolexbor/libxml/tree.h +1315 -0
  19. data/ext/nokolexbor/libxml/uri.h +94 -0
  20. data/ext/nokolexbor/libxml/valid.h +448 -0
  21. data/ext/nokolexbor/libxml/xmlIO.h +369 -0
  22. data/ext/nokolexbor/libxml/xmlautomata.h +146 -0
  23. data/ext/nokolexbor/libxml/xmlerror.h +919 -0
  24. data/ext/nokolexbor/libxml/xmlexports.h +79 -0
  25. data/ext/nokolexbor/libxml/xmlmemory.h +226 -0
  26. data/ext/nokolexbor/libxml/xmlregexp.h +222 -0
  27. data/ext/nokolexbor/libxml/xmlstring.h +140 -0
  28. data/ext/nokolexbor/libxml/xmlversion.h +526 -0
  29. data/ext/nokolexbor/libxml/xpath.h +575 -0
  30. data/ext/nokolexbor/libxml/xpathInternals.h +632 -0
  31. data/ext/nokolexbor/libxml/xpointer.h +137 -0
  32. data/ext/nokolexbor/libxml.h +76 -0
  33. data/ext/nokolexbor/memory.c +39 -0
  34. data/ext/nokolexbor/nl_document.c +51 -0
  35. data/ext/nokolexbor/nl_node.c +790 -0
  36. data/ext/nokolexbor/nl_node_set.c +368 -0
  37. data/ext/nokolexbor/nl_xpath_context.c +200 -0
  38. data/ext/nokolexbor/nokolexbor.c +63 -0
  39. data/ext/nokolexbor/nokolexbor.h +37 -0
  40. data/ext/nokolexbor/private/buf.h +70 -0
  41. data/ext/nokolexbor/private/dict.h +11 -0
  42. data/ext/nokolexbor/private/enc.h +17 -0
  43. data/ext/nokolexbor/private/error.h +21 -0
  44. data/ext/nokolexbor/private/globals.h +9 -0
  45. data/ext/nokolexbor/private/memory.h +9 -0
  46. data/ext/nokolexbor/private/parser.h +27 -0
  47. data/ext/nokolexbor/private/string.h +9 -0
  48. data/ext/nokolexbor/private/threads.h +50 -0
  49. data/ext/nokolexbor/private/tree.h +18 -0
  50. data/ext/nokolexbor/private/xpath.h +7 -0
  51. data/ext/nokolexbor/timsort.h +601 -0
  52. data/ext/nokolexbor/xml_SAX2.c +80 -0
  53. data/ext/nokolexbor/xml_buf.c +363 -0
  54. data/ext/nokolexbor/xml_chvalid.c +334 -0
  55. data/ext/nokolexbor/xml_dict.c +1264 -0
  56. data/ext/nokolexbor/xml_encoding.c +124 -0
  57. data/ext/nokolexbor/xml_error.c +134 -0
  58. data/ext/nokolexbor/xml_globals.c +1085 -0
  59. data/ext/nokolexbor/xml_hash.c +1141 -0
  60. data/ext/nokolexbor/xml_memory.c +203 -0
  61. data/ext/nokolexbor/xml_parser.c +127 -0
  62. data/ext/nokolexbor/xml_parserInternals.c +338 -0
  63. data/ext/nokolexbor/xml_pattern.c +2375 -0
  64. data/ext/nokolexbor/xml_string.c +1051 -0
  65. data/ext/nokolexbor/xml_threads.c +881 -0
  66. data/ext/nokolexbor/xml_tree.c +148 -0
  67. data/ext/nokolexbor/xml_xpath.c +14743 -0
  68. data/lib/nokolexbor/attribute.rb +18 -0
  69. data/lib/nokolexbor/document.rb +6 -0
  70. data/lib/nokolexbor/node.rb +264 -0
  71. data/lib/nokolexbor/node_set.rb +124 -0
  72. data/lib/nokolexbor/version.rb +5 -0
  73. data/lib/nokolexbor/xpath_context.rb +14 -0
  74. data/lib/nokolexbor.rb +17 -0
  75. data/patches/0001-lexbor-support-text-pseudo-element.patch +137 -0
  76. data/patches/0002-lexbor-match-id-class-case-sensitive.patch +22 -0
  77. data/patches/0003-lexbor-attach-template-content-to-self.patch +13 -0
  78. data/vendor/lexbor/CMakeLists.txt +331 -0
  79. data/vendor/lexbor/config.cmake +890 -0
  80. data/vendor/lexbor/feature.cmake +134 -0
  81. data/vendor/lexbor/source/lexbor/core/array.c +208 -0
  82. data/vendor/lexbor/source/lexbor/core/array.h +100 -0
  83. data/vendor/lexbor/source/lexbor/core/array_obj.c +216 -0
  84. data/vendor/lexbor/source/lexbor/core/array_obj.h +134 -0
  85. data/vendor/lexbor/source/lexbor/core/avl.c +442 -0
  86. data/vendor/lexbor/source/lexbor/core/avl.h +82 -0
  87. data/vendor/lexbor/source/lexbor/core/base.h +86 -0
  88. data/vendor/lexbor/source/lexbor/core/bst.c +468 -0
  89. data/vendor/lexbor/source/lexbor/core/bst.h +108 -0
  90. data/vendor/lexbor/source/lexbor/core/bst_map.c +238 -0
  91. data/vendor/lexbor/source/lexbor/core/bst_map.h +87 -0
  92. data/vendor/lexbor/source/lexbor/core/config.cmake +12 -0
  93. data/vendor/lexbor/source/lexbor/core/conv.c +203 -0
  94. data/vendor/lexbor/source/lexbor/core/conv.h +53 -0
  95. data/vendor/lexbor/source/lexbor/core/core.h +35 -0
  96. data/vendor/lexbor/source/lexbor/core/def.h +57 -0
  97. data/vendor/lexbor/source/lexbor/core/diyfp.c +153 -0
  98. data/vendor/lexbor/source/lexbor/core/diyfp.h +258 -0
  99. data/vendor/lexbor/source/lexbor/core/dobject.c +187 -0
  100. data/vendor/lexbor/source/lexbor/core/dobject.h +92 -0
  101. data/vendor/lexbor/source/lexbor/core/dtoa.c +404 -0
  102. data/vendor/lexbor/source/lexbor/core/dtoa.h +28 -0
  103. data/vendor/lexbor/source/lexbor/core/fs.h +60 -0
  104. data/vendor/lexbor/source/lexbor/core/hash.c +476 -0
  105. data/vendor/lexbor/source/lexbor/core/hash.h +218 -0
  106. data/vendor/lexbor/source/lexbor/core/in.c +267 -0
  107. data/vendor/lexbor/source/lexbor/core/in.h +172 -0
  108. data/vendor/lexbor/source/lexbor/core/lexbor.h +35 -0
  109. data/vendor/lexbor/source/lexbor/core/mem.c +228 -0
  110. data/vendor/lexbor/source/lexbor/core/mem.h +141 -0
  111. data/vendor/lexbor/source/lexbor/core/mraw.c +428 -0
  112. data/vendor/lexbor/source/lexbor/core/mraw.h +114 -0
  113. data/vendor/lexbor/source/lexbor/core/perf.h +45 -0
  114. data/vendor/lexbor/source/lexbor/core/plog.c +73 -0
  115. data/vendor/lexbor/source/lexbor/core/plog.h +102 -0
  116. data/vendor/lexbor/source/lexbor/core/print.c +168 -0
  117. data/vendor/lexbor/source/lexbor/core/print.h +39 -0
  118. data/vendor/lexbor/source/lexbor/core/sbst.h +59 -0
  119. data/vendor/lexbor/source/lexbor/core/serialize.c +27 -0
  120. data/vendor/lexbor/source/lexbor/core/serialize.h +32 -0
  121. data/vendor/lexbor/source/lexbor/core/shs.c +118 -0
  122. data/vendor/lexbor/source/lexbor/core/shs.h +82 -0
  123. data/vendor/lexbor/source/lexbor/core/str.c +617 -0
  124. data/vendor/lexbor/source/lexbor/core/str.h +247 -0
  125. data/vendor/lexbor/source/lexbor/core/str_res.h +369 -0
  126. data/vendor/lexbor/source/lexbor/core/strtod.c +326 -0
  127. data/vendor/lexbor/source/lexbor/core/strtod.h +28 -0
  128. data/vendor/lexbor/source/lexbor/core/types.h +39 -0
  129. data/vendor/lexbor/source/lexbor/core/utils.c +43 -0
  130. data/vendor/lexbor/source/lexbor/core/utils.h +36 -0
  131. data/vendor/lexbor/source/lexbor/css/base.h +44 -0
  132. data/vendor/lexbor/source/lexbor/css/config.cmake +2 -0
  133. data/vendor/lexbor/source/lexbor/css/css.h +25 -0
  134. data/vendor/lexbor/source/lexbor/css/log.c +336 -0
  135. data/vendor/lexbor/source/lexbor/css/log.h +103 -0
  136. data/vendor/lexbor/source/lexbor/css/node.h +29 -0
  137. data/vendor/lexbor/source/lexbor/css/parser.c +473 -0
  138. data/vendor/lexbor/source/lexbor/css/parser.h +368 -0
  139. data/vendor/lexbor/source/lexbor/css/selectors/base.h +48 -0
  140. data/vendor/lexbor/source/lexbor/css/selectors/pseudo.c +91 -0
  141. data/vendor/lexbor/source/lexbor/css/selectors/pseudo.h +66 -0
  142. data/vendor/lexbor/source/lexbor/css/selectors/pseudo_const.h +109 -0
  143. data/vendor/lexbor/source/lexbor/css/selectors/pseudo_res.h +302 -0
  144. data/vendor/lexbor/source/lexbor/css/selectors/pseudo_state.c +279 -0
  145. data/vendor/lexbor/source/lexbor/css/selectors/pseudo_state.h +85 -0
  146. data/vendor/lexbor/source/lexbor/css/selectors/selector.c +927 -0
  147. data/vendor/lexbor/source/lexbor/css/selectors/selector.h +200 -0
  148. data/vendor/lexbor/source/lexbor/css/selectors/selectors.c +340 -0
  149. data/vendor/lexbor/source/lexbor/css/selectors/selectors.h +137 -0
  150. data/vendor/lexbor/source/lexbor/css/selectors/state.c +1718 -0
  151. data/vendor/lexbor/source/lexbor/css/selectors/state.h +79 -0
  152. data/vendor/lexbor/source/lexbor/css/stylesheet.h +37 -0
  153. data/vendor/lexbor/source/lexbor/css/syntax/anb.c +443 -0
  154. data/vendor/lexbor/source/lexbor/css/syntax/anb.h +45 -0
  155. data/vendor/lexbor/source/lexbor/css/syntax/base.h +33 -0
  156. data/vendor/lexbor/source/lexbor/css/syntax/parser.c +9 -0
  157. data/vendor/lexbor/source/lexbor/css/syntax/parser.h +25 -0
  158. data/vendor/lexbor/source/lexbor/css/syntax/res.h +48 -0
  159. data/vendor/lexbor/source/lexbor/css/syntax/state.c +2603 -0
  160. data/vendor/lexbor/source/lexbor/css/syntax/state.h +140 -0
  161. data/vendor/lexbor/source/lexbor/css/syntax/state_res.h +273 -0
  162. data/vendor/lexbor/source/lexbor/css/syntax/syntax.c +67 -0
  163. data/vendor/lexbor/source/lexbor/css/syntax/token.c +618 -0
  164. data/vendor/lexbor/source/lexbor/css/syntax/token.h +298 -0
  165. data/vendor/lexbor/source/lexbor/css/syntax/token_res.h +68 -0
  166. data/vendor/lexbor/source/lexbor/css/syntax/tokenizer/error.c +30 -0
  167. data/vendor/lexbor/source/lexbor/css/syntax/tokenizer/error.h +58 -0
  168. data/vendor/lexbor/source/lexbor/css/syntax/tokenizer.c +278 -0
  169. data/vendor/lexbor/source/lexbor/css/syntax/tokenizer.h +121 -0
  170. data/vendor/lexbor/source/lexbor/dom/base.h +32 -0
  171. data/vendor/lexbor/source/lexbor/dom/collection.c +97 -0
  172. data/vendor/lexbor/source/lexbor/dom/collection.h +112 -0
  173. data/vendor/lexbor/source/lexbor/dom/config.cmake +3 -0
  174. data/vendor/lexbor/source/lexbor/dom/dom.h +29 -0
  175. data/vendor/lexbor/source/lexbor/dom/exception.c +18 -0
  176. data/vendor/lexbor/source/lexbor/dom/exception.h +73 -0
  177. data/vendor/lexbor/source/lexbor/dom/interface.c +110 -0
  178. data/vendor/lexbor/source/lexbor/dom/interface.h +88 -0
  179. data/vendor/lexbor/source/lexbor/dom/interfaces/attr.c +445 -0
  180. data/vendor/lexbor/source/lexbor/dom/interfaces/attr.h +152 -0
  181. data/vendor/lexbor/source/lexbor/dom/interfaces/attr_const.h +62 -0
  182. data/vendor/lexbor/source/lexbor/dom/interfaces/attr_res.h +143 -0
  183. data/vendor/lexbor/source/lexbor/dom/interfaces/cdata_section.c +55 -0
  184. data/vendor/lexbor/source/lexbor/dom/interfaces/cdata_section.h +38 -0
  185. data/vendor/lexbor/source/lexbor/dom/interfaces/character_data.c +110 -0
  186. data/vendor/lexbor/source/lexbor/dom/interfaces/character_data.h +51 -0
  187. data/vendor/lexbor/source/lexbor/dom/interfaces/comment.c +64 -0
  188. data/vendor/lexbor/source/lexbor/dom/interfaces/comment.h +42 -0
  189. data/vendor/lexbor/source/lexbor/dom/interfaces/document.c +536 -0
  190. data/vendor/lexbor/source/lexbor/dom/interfaces/document.h +243 -0
  191. data/vendor/lexbor/source/lexbor/dom/interfaces/document_fragment.c +36 -0
  192. data/vendor/lexbor/source/lexbor/dom/interfaces/document_fragment.h +36 -0
  193. data/vendor/lexbor/source/lexbor/dom/interfaces/document_type.c +125 -0
  194. data/vendor/lexbor/source/lexbor/dom/interfaces/document_type.h +108 -0
  195. data/vendor/lexbor/source/lexbor/dom/interfaces/element.c +1411 -0
  196. data/vendor/lexbor/source/lexbor/dom/interfaces/element.h +319 -0
  197. data/vendor/lexbor/source/lexbor/dom/interfaces/event_target.c +32 -0
  198. data/vendor/lexbor/source/lexbor/dom/interfaces/event_target.h +34 -0
  199. data/vendor/lexbor/source/lexbor/dom/interfaces/node.c +661 -0
  200. data/vendor/lexbor/source/lexbor/dom/interfaces/node.h +192 -0
  201. data/vendor/lexbor/source/lexbor/dom/interfaces/processing_instruction.c +87 -0
  202. data/vendor/lexbor/source/lexbor/dom/interfaces/processing_instruction.h +66 -0
  203. data/vendor/lexbor/source/lexbor/dom/interfaces/shadow_root.c +36 -0
  204. data/vendor/lexbor/source/lexbor/dom/interfaces/shadow_root.h +44 -0
  205. data/vendor/lexbor/source/lexbor/dom/interfaces/text.c +63 -0
  206. data/vendor/lexbor/source/lexbor/dom/interfaces/text.h +42 -0
  207. data/vendor/lexbor/source/lexbor/encoding/base.h +218 -0
  208. data/vendor/lexbor/source/lexbor/encoding/big5.c +42839 -0
  209. data/vendor/lexbor/source/lexbor/encoding/config.cmake +12 -0
  210. data/vendor/lexbor/source/lexbor/encoding/const.h +65 -0
  211. data/vendor/lexbor/source/lexbor/encoding/decode.c +3193 -0
  212. data/vendor/lexbor/source/lexbor/encoding/decode.h +370 -0
  213. data/vendor/lexbor/source/lexbor/encoding/encode.c +1931 -0
  214. data/vendor/lexbor/source/lexbor/encoding/encode.h +377 -0
  215. data/vendor/lexbor/source/lexbor/encoding/encoding.c +252 -0
  216. data/vendor/lexbor/source/lexbor/encoding/encoding.h +475 -0
  217. data/vendor/lexbor/source/lexbor/encoding/euc_kr.c +53883 -0
  218. data/vendor/lexbor/source/lexbor/encoding/gb18030.c +47905 -0
  219. data/vendor/lexbor/source/lexbor/encoding/iso_2022_jp_katakana.c +159 -0
  220. data/vendor/lexbor/source/lexbor/encoding/jis0208.c +22477 -0
  221. data/vendor/lexbor/source/lexbor/encoding/jis0212.c +15787 -0
  222. data/vendor/lexbor/source/lexbor/encoding/multi.h +53 -0
  223. data/vendor/lexbor/source/lexbor/encoding/range.c +71 -0
  224. data/vendor/lexbor/source/lexbor/encoding/range.h +34 -0
  225. data/vendor/lexbor/source/lexbor/encoding/res.c +222 -0
  226. data/vendor/lexbor/source/lexbor/encoding/res.h +34 -0
  227. data/vendor/lexbor/source/lexbor/encoding/single.c +13748 -0
  228. data/vendor/lexbor/source/lexbor/encoding/single.h +116 -0
  229. data/vendor/lexbor/source/lexbor/html/base.h +44 -0
  230. data/vendor/lexbor/source/lexbor/html/config.cmake +3 -0
  231. data/vendor/lexbor/source/lexbor/html/encoding.c +574 -0
  232. data/vendor/lexbor/source/lexbor/html/encoding.h +106 -0
  233. data/vendor/lexbor/source/lexbor/html/html.h +107 -0
  234. data/vendor/lexbor/source/lexbor/html/interface.c +165 -0
  235. data/vendor/lexbor/source/lexbor/html/interface.h +186 -0
  236. data/vendor/lexbor/source/lexbor/html/interface_res.h +4449 -0
  237. data/vendor/lexbor/source/lexbor/html/interfaces/anchor_element.c +36 -0
  238. data/vendor/lexbor/source/lexbor/html/interfaces/anchor_element.h +34 -0
  239. data/vendor/lexbor/source/lexbor/html/interfaces/area_element.c +36 -0
  240. data/vendor/lexbor/source/lexbor/html/interfaces/area_element.h +34 -0
  241. data/vendor/lexbor/source/lexbor/html/interfaces/audio_element.c +36 -0
  242. data/vendor/lexbor/source/lexbor/html/interfaces/audio_element.h +34 -0
  243. data/vendor/lexbor/source/lexbor/html/interfaces/base_element.c +36 -0
  244. data/vendor/lexbor/source/lexbor/html/interfaces/base_element.h +34 -0
  245. data/vendor/lexbor/source/lexbor/html/interfaces/body_element.c +36 -0
  246. data/vendor/lexbor/source/lexbor/html/interfaces/body_element.h +34 -0
  247. data/vendor/lexbor/source/lexbor/html/interfaces/br_element.c +36 -0
  248. data/vendor/lexbor/source/lexbor/html/interfaces/br_element.h +34 -0
  249. data/vendor/lexbor/source/lexbor/html/interfaces/button_element.c +36 -0
  250. data/vendor/lexbor/source/lexbor/html/interfaces/button_element.h +34 -0
  251. data/vendor/lexbor/source/lexbor/html/interfaces/canvas_element.c +36 -0
  252. data/vendor/lexbor/source/lexbor/html/interfaces/canvas_element.h +34 -0
  253. data/vendor/lexbor/source/lexbor/html/interfaces/d_list_element.c +36 -0
  254. data/vendor/lexbor/source/lexbor/html/interfaces/d_list_element.h +34 -0
  255. data/vendor/lexbor/source/lexbor/html/interfaces/data_element.c +36 -0
  256. data/vendor/lexbor/source/lexbor/html/interfaces/data_element.h +34 -0
  257. data/vendor/lexbor/source/lexbor/html/interfaces/data_list_element.c +36 -0
  258. data/vendor/lexbor/source/lexbor/html/interfaces/data_list_element.h +34 -0
  259. data/vendor/lexbor/source/lexbor/html/interfaces/details_element.c +36 -0
  260. data/vendor/lexbor/source/lexbor/html/interfaces/details_element.h +34 -0
  261. data/vendor/lexbor/source/lexbor/html/interfaces/dialog_element.c +36 -0
  262. data/vendor/lexbor/source/lexbor/html/interfaces/dialog_element.h +34 -0
  263. data/vendor/lexbor/source/lexbor/html/interfaces/directory_element.c +36 -0
  264. data/vendor/lexbor/source/lexbor/html/interfaces/directory_element.h +34 -0
  265. data/vendor/lexbor/source/lexbor/html/interfaces/div_element.c +36 -0
  266. data/vendor/lexbor/source/lexbor/html/interfaces/div_element.h +34 -0
  267. data/vendor/lexbor/source/lexbor/html/interfaces/document.c +444 -0
  268. data/vendor/lexbor/source/lexbor/html/interfaces/document.h +256 -0
  269. data/vendor/lexbor/source/lexbor/html/interfaces/element.c +64 -0
  270. data/vendor/lexbor/source/lexbor/html/interfaces/element.h +54 -0
  271. data/vendor/lexbor/source/lexbor/html/interfaces/embed_element.c +36 -0
  272. data/vendor/lexbor/source/lexbor/html/interfaces/embed_element.h +34 -0
  273. data/vendor/lexbor/source/lexbor/html/interfaces/field_set_element.c +36 -0
  274. data/vendor/lexbor/source/lexbor/html/interfaces/field_set_element.h +34 -0
  275. data/vendor/lexbor/source/lexbor/html/interfaces/font_element.c +36 -0
  276. data/vendor/lexbor/source/lexbor/html/interfaces/font_element.h +34 -0
  277. data/vendor/lexbor/source/lexbor/html/interfaces/form_element.c +36 -0
  278. data/vendor/lexbor/source/lexbor/html/interfaces/form_element.h +34 -0
  279. data/vendor/lexbor/source/lexbor/html/interfaces/frame_element.c +36 -0
  280. data/vendor/lexbor/source/lexbor/html/interfaces/frame_element.h +34 -0
  281. data/vendor/lexbor/source/lexbor/html/interfaces/frame_set_element.c +36 -0
  282. data/vendor/lexbor/source/lexbor/html/interfaces/frame_set_element.h +34 -0
  283. data/vendor/lexbor/source/lexbor/html/interfaces/head_element.c +36 -0
  284. data/vendor/lexbor/source/lexbor/html/interfaces/head_element.h +34 -0
  285. data/vendor/lexbor/source/lexbor/html/interfaces/heading_element.c +36 -0
  286. data/vendor/lexbor/source/lexbor/html/interfaces/heading_element.h +34 -0
  287. data/vendor/lexbor/source/lexbor/html/interfaces/hr_element.c +36 -0
  288. data/vendor/lexbor/source/lexbor/html/interfaces/hr_element.h +34 -0
  289. data/vendor/lexbor/source/lexbor/html/interfaces/html_element.c +36 -0
  290. data/vendor/lexbor/source/lexbor/html/interfaces/html_element.h +34 -0
  291. data/vendor/lexbor/source/lexbor/html/interfaces/iframe_element.c +36 -0
  292. data/vendor/lexbor/source/lexbor/html/interfaces/iframe_element.h +34 -0
  293. data/vendor/lexbor/source/lexbor/html/interfaces/image_element.c +36 -0
  294. data/vendor/lexbor/source/lexbor/html/interfaces/image_element.h +34 -0
  295. data/vendor/lexbor/source/lexbor/html/interfaces/input_element.c +36 -0
  296. data/vendor/lexbor/source/lexbor/html/interfaces/input_element.h +34 -0
  297. data/vendor/lexbor/source/lexbor/html/interfaces/label_element.c +36 -0
  298. data/vendor/lexbor/source/lexbor/html/interfaces/label_element.h +34 -0
  299. data/vendor/lexbor/source/lexbor/html/interfaces/legend_element.c +36 -0
  300. data/vendor/lexbor/source/lexbor/html/interfaces/legend_element.h +34 -0
  301. data/vendor/lexbor/source/lexbor/html/interfaces/li_element.c +36 -0
  302. data/vendor/lexbor/source/lexbor/html/interfaces/li_element.h +34 -0
  303. data/vendor/lexbor/source/lexbor/html/interfaces/link_element.c +36 -0
  304. data/vendor/lexbor/source/lexbor/html/interfaces/link_element.h +34 -0
  305. data/vendor/lexbor/source/lexbor/html/interfaces/map_element.c +36 -0
  306. data/vendor/lexbor/source/lexbor/html/interfaces/map_element.h +34 -0
  307. data/vendor/lexbor/source/lexbor/html/interfaces/marquee_element.c +36 -0
  308. data/vendor/lexbor/source/lexbor/html/interfaces/marquee_element.h +34 -0
  309. data/vendor/lexbor/source/lexbor/html/interfaces/media_element.c +36 -0
  310. data/vendor/lexbor/source/lexbor/html/interfaces/media_element.h +34 -0
  311. data/vendor/lexbor/source/lexbor/html/interfaces/menu_element.c +36 -0
  312. data/vendor/lexbor/source/lexbor/html/interfaces/menu_element.h +34 -0
  313. data/vendor/lexbor/source/lexbor/html/interfaces/meta_element.c +36 -0
  314. data/vendor/lexbor/source/lexbor/html/interfaces/meta_element.h +34 -0
  315. data/vendor/lexbor/source/lexbor/html/interfaces/meter_element.c +36 -0
  316. data/vendor/lexbor/source/lexbor/html/interfaces/meter_element.h +34 -0
  317. data/vendor/lexbor/source/lexbor/html/interfaces/mod_element.c +36 -0
  318. data/vendor/lexbor/source/lexbor/html/interfaces/mod_element.h +34 -0
  319. data/vendor/lexbor/source/lexbor/html/interfaces/o_list_element.c +36 -0
  320. data/vendor/lexbor/source/lexbor/html/interfaces/o_list_element.h +34 -0
  321. data/vendor/lexbor/source/lexbor/html/interfaces/object_element.c +36 -0
  322. data/vendor/lexbor/source/lexbor/html/interfaces/object_element.h +34 -0
  323. data/vendor/lexbor/source/lexbor/html/interfaces/opt_group_element.c +36 -0
  324. data/vendor/lexbor/source/lexbor/html/interfaces/opt_group_element.h +34 -0
  325. data/vendor/lexbor/source/lexbor/html/interfaces/option_element.c +36 -0
  326. data/vendor/lexbor/source/lexbor/html/interfaces/option_element.h +34 -0
  327. data/vendor/lexbor/source/lexbor/html/interfaces/output_element.c +36 -0
  328. data/vendor/lexbor/source/lexbor/html/interfaces/output_element.h +34 -0
  329. data/vendor/lexbor/source/lexbor/html/interfaces/paragraph_element.c +36 -0
  330. data/vendor/lexbor/source/lexbor/html/interfaces/paragraph_element.h +34 -0
  331. data/vendor/lexbor/source/lexbor/html/interfaces/param_element.c +36 -0
  332. data/vendor/lexbor/source/lexbor/html/interfaces/param_element.h +34 -0
  333. data/vendor/lexbor/source/lexbor/html/interfaces/picture_element.c +36 -0
  334. data/vendor/lexbor/source/lexbor/html/interfaces/picture_element.h +34 -0
  335. data/vendor/lexbor/source/lexbor/html/interfaces/pre_element.c +36 -0
  336. data/vendor/lexbor/source/lexbor/html/interfaces/pre_element.h +34 -0
  337. data/vendor/lexbor/source/lexbor/html/interfaces/progress_element.c +36 -0
  338. data/vendor/lexbor/source/lexbor/html/interfaces/progress_element.h +34 -0
  339. data/vendor/lexbor/source/lexbor/html/interfaces/quote_element.c +36 -0
  340. data/vendor/lexbor/source/lexbor/html/interfaces/quote_element.h +34 -0
  341. data/vendor/lexbor/source/lexbor/html/interfaces/script_element.c +36 -0
  342. data/vendor/lexbor/source/lexbor/html/interfaces/script_element.h +34 -0
  343. data/vendor/lexbor/source/lexbor/html/interfaces/select_element.c +36 -0
  344. data/vendor/lexbor/source/lexbor/html/interfaces/select_element.h +34 -0
  345. data/vendor/lexbor/source/lexbor/html/interfaces/slot_element.c +36 -0
  346. data/vendor/lexbor/source/lexbor/html/interfaces/slot_element.h +34 -0
  347. data/vendor/lexbor/source/lexbor/html/interfaces/source_element.c +36 -0
  348. data/vendor/lexbor/source/lexbor/html/interfaces/source_element.h +34 -0
  349. data/vendor/lexbor/source/lexbor/html/interfaces/span_element.c +36 -0
  350. data/vendor/lexbor/source/lexbor/html/interfaces/span_element.h +34 -0
  351. data/vendor/lexbor/source/lexbor/html/interfaces/style_element.c +36 -0
  352. data/vendor/lexbor/source/lexbor/html/interfaces/style_element.h +34 -0
  353. data/vendor/lexbor/source/lexbor/html/interfaces/table_caption_element.c +36 -0
  354. data/vendor/lexbor/source/lexbor/html/interfaces/table_caption_element.h +34 -0
  355. data/vendor/lexbor/source/lexbor/html/interfaces/table_cell_element.c +36 -0
  356. data/vendor/lexbor/source/lexbor/html/interfaces/table_cell_element.h +34 -0
  357. data/vendor/lexbor/source/lexbor/html/interfaces/table_col_element.c +36 -0
  358. data/vendor/lexbor/source/lexbor/html/interfaces/table_col_element.h +34 -0
  359. data/vendor/lexbor/source/lexbor/html/interfaces/table_element.c +36 -0
  360. data/vendor/lexbor/source/lexbor/html/interfaces/table_element.h +34 -0
  361. data/vendor/lexbor/source/lexbor/html/interfaces/table_row_element.c +36 -0
  362. data/vendor/lexbor/source/lexbor/html/interfaces/table_row_element.h +34 -0
  363. data/vendor/lexbor/source/lexbor/html/interfaces/table_section_element.c +36 -0
  364. data/vendor/lexbor/source/lexbor/html/interfaces/table_section_element.h +34 -0
  365. data/vendor/lexbor/source/lexbor/html/interfaces/template_element.c +46 -0
  366. data/vendor/lexbor/source/lexbor/html/interfaces/template_element.h +38 -0
  367. data/vendor/lexbor/source/lexbor/html/interfaces/text_area_element.c +36 -0
  368. data/vendor/lexbor/source/lexbor/html/interfaces/text_area_element.h +34 -0
  369. data/vendor/lexbor/source/lexbor/html/interfaces/time_element.c +36 -0
  370. data/vendor/lexbor/source/lexbor/html/interfaces/time_element.h +34 -0
  371. data/vendor/lexbor/source/lexbor/html/interfaces/title_element.c +133 -0
  372. data/vendor/lexbor/source/lexbor/html/interfaces/title_element.h +42 -0
  373. data/vendor/lexbor/source/lexbor/html/interfaces/track_element.c +36 -0
  374. data/vendor/lexbor/source/lexbor/html/interfaces/track_element.h +34 -0
  375. data/vendor/lexbor/source/lexbor/html/interfaces/u_list_element.c +36 -0
  376. data/vendor/lexbor/source/lexbor/html/interfaces/u_list_element.h +34 -0
  377. data/vendor/lexbor/source/lexbor/html/interfaces/unknown_element.c +36 -0
  378. data/vendor/lexbor/source/lexbor/html/interfaces/unknown_element.h +34 -0
  379. data/vendor/lexbor/source/lexbor/html/interfaces/video_element.c +36 -0
  380. data/vendor/lexbor/source/lexbor/html/interfaces/video_element.h +34 -0
  381. data/vendor/lexbor/source/lexbor/html/interfaces/window.c +36 -0
  382. data/vendor/lexbor/source/lexbor/html/interfaces/window.h +34 -0
  383. data/vendor/lexbor/source/lexbor/html/node.c +14 -0
  384. data/vendor/lexbor/source/lexbor/html/node.h +67 -0
  385. data/vendor/lexbor/source/lexbor/html/parser.c +469 -0
  386. data/vendor/lexbor/source/lexbor/html/parser.h +170 -0
  387. data/vendor/lexbor/source/lexbor/html/serialize.c +1510 -0
  388. data/vendor/lexbor/source/lexbor/html/serialize.h +93 -0
  389. data/vendor/lexbor/source/lexbor/html/tag.h +103 -0
  390. data/vendor/lexbor/source/lexbor/html/tag_res.h +2262 -0
  391. data/vendor/lexbor/source/lexbor/html/token.c +386 -0
  392. data/vendor/lexbor/source/lexbor/html/token.h +130 -0
  393. data/vendor/lexbor/source/lexbor/html/token_attr.c +44 -0
  394. data/vendor/lexbor/source/lexbor/html/token_attr.h +67 -0
  395. data/vendor/lexbor/source/lexbor/html/tokenizer/error.c +28 -0
  396. data/vendor/lexbor/source/lexbor/html/tokenizer/error.h +141 -0
  397. data/vendor/lexbor/source/lexbor/html/tokenizer/res.h +4956 -0
  398. data/vendor/lexbor/source/lexbor/html/tokenizer/state.c +2171 -0
  399. data/vendor/lexbor/source/lexbor/html/tokenizer/state.h +225 -0
  400. data/vendor/lexbor/source/lexbor/html/tokenizer/state_comment.c +489 -0
  401. data/vendor/lexbor/source/lexbor/html/tokenizer/state_comment.h +27 -0
  402. data/vendor/lexbor/source/lexbor/html/tokenizer/state_doctype.c +1654 -0
  403. data/vendor/lexbor/source/lexbor/html/tokenizer/state_doctype.h +27 -0
  404. data/vendor/lexbor/source/lexbor/html/tokenizer/state_rawtext.c +303 -0
  405. data/vendor/lexbor/source/lexbor/html/tokenizer/state_rawtext.h +32 -0
  406. data/vendor/lexbor/source/lexbor/html/tokenizer/state_rcdata.c +311 -0
  407. data/vendor/lexbor/source/lexbor/html/tokenizer/state_rcdata.h +32 -0
  408. data/vendor/lexbor/source/lexbor/html/tokenizer/state_script.c +1209 -0
  409. data/vendor/lexbor/source/lexbor/html/tokenizer/state_script.h +32 -0
  410. data/vendor/lexbor/source/lexbor/html/tokenizer.c +499 -0
  411. data/vendor/lexbor/source/lexbor/html/tokenizer.h +343 -0
  412. data/vendor/lexbor/source/lexbor/html/tree/active_formatting.c +241 -0
  413. data/vendor/lexbor/source/lexbor/html/tree/active_formatting.h +117 -0
  414. data/vendor/lexbor/source/lexbor/html/tree/error.c +26 -0
  415. data/vendor/lexbor/source/lexbor/html/tree/error.h +114 -0
  416. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/after_after_body.c +62 -0
  417. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/after_after_frameset.c +63 -0
  418. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/after_body.c +82 -0
  419. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/after_frameset.c +88 -0
  420. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/after_head.c +222 -0
  421. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/before_head.c +144 -0
  422. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/before_html.c +166 -0
  423. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/foreign_content.c +358 -0
  424. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_body.c +1974 -0
  425. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_caption.c +158 -0
  426. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_cell.c +187 -0
  427. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_column_group.c +194 -0
  428. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_frameset.c +149 -0
  429. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_head.c +374 -0
  430. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_head_noscript.c +121 -0
  431. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_row.c +211 -0
  432. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_select.c +341 -0
  433. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_select_in_table.c +115 -0
  434. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_table.c +451 -0
  435. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_table_body.c +208 -0
  436. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_table_text.c +127 -0
  437. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/in_template.c +189 -0
  438. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/initial.c +411 -0
  439. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode/text.c +61 -0
  440. data/vendor/lexbor/source/lexbor/html/tree/insertion_mode.h +135 -0
  441. data/vendor/lexbor/source/lexbor/html/tree/open_elements.c +251 -0
  442. data/vendor/lexbor/source/lexbor/html/tree/open_elements.h +105 -0
  443. data/vendor/lexbor/source/lexbor/html/tree/template_insertion.c +10 -0
  444. data/vendor/lexbor/source/lexbor/html/tree/template_insertion.h +100 -0
  445. data/vendor/lexbor/source/lexbor/html/tree.c +1726 -0
  446. data/vendor/lexbor/source/lexbor/html/tree.h +431 -0
  447. data/vendor/lexbor/source/lexbor/html/tree_res.h +111 -0
  448. data/vendor/lexbor/source/lexbor/ns/base.h +32 -0
  449. data/vendor/lexbor/source/lexbor/ns/config.cmake +2 -0
  450. data/vendor/lexbor/source/lexbor/ns/const.h +37 -0
  451. data/vendor/lexbor/source/lexbor/ns/ns.c +154 -0
  452. data/vendor/lexbor/source/lexbor/ns/ns.h +66 -0
  453. data/vendor/lexbor/source/lexbor/ns/res.h +97 -0
  454. data/vendor/lexbor/source/lexbor/ports/posix/config.cmake +11 -0
  455. data/vendor/lexbor/source/lexbor/ports/posix/lexbor/core/fs.c +236 -0
  456. data/vendor/lexbor/source/lexbor/ports/posix/lexbor/core/memory.c +33 -0
  457. data/vendor/lexbor/source/lexbor/ports/posix/lexbor/core/perf.c +158 -0
  458. data/vendor/lexbor/source/lexbor/ports/windows_nt/config.cmake +18 -0
  459. data/vendor/lexbor/source/lexbor/ports/windows_nt/lexbor/core/fs.c +239 -0
  460. data/vendor/lexbor/source/lexbor/ports/windows_nt/lexbor/core/memory.c +33 -0
  461. data/vendor/lexbor/source/lexbor/ports/windows_nt/lexbor/core/perf.c +81 -0
  462. data/vendor/lexbor/source/lexbor/selectors/base.h +30 -0
  463. data/vendor/lexbor/source/lexbor/selectors/config.cmake +2 -0
  464. data/vendor/lexbor/source/lexbor/selectors/selectors.c +1591 -0
  465. data/vendor/lexbor/source/lexbor/selectors/selectors.h +71 -0
  466. data/vendor/lexbor/source/lexbor/tag/base.h +32 -0
  467. data/vendor/lexbor/source/lexbor/tag/config.cmake +2 -0
  468. data/vendor/lexbor/source/lexbor/tag/const.h +225 -0
  469. data/vendor/lexbor/source/lexbor/tag/res.h +562 -0
  470. data/vendor/lexbor/source/lexbor/tag/tag.c +144 -0
  471. data/vendor/lexbor/source/lexbor/tag/tag.h +123 -0
  472. data/vendor/lexbor/source/lexbor/utils/base.h +32 -0
  473. data/vendor/lexbor/source/lexbor/utils/config.cmake +2 -0
  474. data/vendor/lexbor/source/lexbor/utils/http.c +534 -0
  475. data/vendor/lexbor/source/lexbor/utils/http.h +90 -0
  476. data/vendor/lexbor/source/lexbor/utils/utils.h +15 -0
  477. data/vendor/lexbor/source/lexbor/utils/warc.c +817 -0
  478. data/vendor/lexbor/source/lexbor/utils/warc.h +126 -0
  479. data/vendor/lexbor/utils/lexbor/css/selectors/pseudo.py +231 -0
  480. data/vendor/lexbor/utils/lexbor/css/selectors/tmp/const.h +21 -0
  481. data/vendor/lexbor/utils/lexbor/css/selectors/tmp/res.h +26 -0
  482. data/vendor/lexbor/utils/lexbor/css/syntax/definitions.py +49 -0
  483. data/vendor/lexbor/utils/lexbor/css/syntax/token_res.py +54 -0
  484. data/vendor/lexbor/utils/lexbor/css/syntax/tokenizer_code_map.py +36 -0
  485. data/vendor/lexbor/version +1 -0
  486. metadata +542 -0
@@ -0,0 +1,374 @@
1
+ /*
2
+ * Copyright (C) 2018-2020 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #include "lexbor/html/tree/insertion_mode.h"
8
+ #include "lexbor/html/tree/open_elements.h"
9
+ #include "lexbor/html/tree/active_formatting.h"
10
+ #include "lexbor/html/tree/template_insertion.h"
11
+ #include "lexbor/html/interfaces/script_element.h"
12
+ #include "lexbor/html/interfaces/template_element.h"
13
+ #include "lexbor/html/tokenizer/state_script.h"
14
+
15
+
16
+ static bool
17
+ lxb_html_tree_insertion_mode_in_head_open(lxb_html_tree_t *tree,
18
+ lxb_html_token_t *token);
19
+
20
+ static bool
21
+ lxb_html_tree_insertion_mode_in_head_closed(lxb_html_tree_t *tree,
22
+ lxb_html_token_t *token);
23
+
24
+ lxb_inline bool
25
+ lxb_html_tree_insertion_mode_in_head_script(lxb_html_tree_t *tree,
26
+ lxb_html_token_t *token);
27
+
28
+ lxb_inline bool
29
+ lxb_html_tree_insertion_mode_in_head_template(lxb_html_tree_t *tree,
30
+ lxb_html_token_t *token);
31
+
32
+ lxb_inline bool
33
+ lxb_html_tree_insertion_mode_in_head_template_closed(lxb_html_tree_t *tree,
34
+ lxb_html_token_t *token);
35
+
36
+ lxb_inline bool
37
+ lxb_html_tree_insertion_mode_in_head_anything_else(lxb_html_tree_t *tree);
38
+
39
+
40
+ bool
41
+ lxb_html_tree_insertion_mode_in_head(lxb_html_tree_t *tree,
42
+ lxb_html_token_t *token)
43
+ {
44
+ if (token->type & LXB_HTML_TOKEN_TYPE_CLOSE) {
45
+ return lxb_html_tree_insertion_mode_in_head_closed(tree, token);;
46
+ }
47
+
48
+ return lxb_html_tree_insertion_mode_in_head_open(tree, token);
49
+ }
50
+
51
+ static bool
52
+ lxb_html_tree_insertion_mode_in_head_open(lxb_html_tree_t *tree,
53
+ lxb_html_token_t *token)
54
+ {
55
+ switch (token->tag_id) {
56
+ case LXB_TAG__EM_COMMENT: {
57
+ lxb_dom_comment_t *comment;
58
+
59
+ comment = lxb_html_tree_insert_comment(tree, token, NULL);
60
+ if (comment == NULL) {
61
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
62
+
63
+ return lxb_html_tree_process_abort(tree);
64
+ }
65
+
66
+ break;
67
+ }
68
+
69
+ case LXB_TAG__EM_DOCTYPE:
70
+ lxb_html_tree_parse_error(tree, token,
71
+ LXB_HTML_RULES_ERROR_DOTOINHEMO);
72
+ break;
73
+
74
+ case LXB_TAG_HTML:
75
+ return lxb_html_tree_insertion_mode_in_body(tree, token);
76
+
77
+ case LXB_TAG_BASE:
78
+ case LXB_TAG_BASEFONT:
79
+ case LXB_TAG_BGSOUND:
80
+ case LXB_TAG_LINK: {
81
+ lxb_html_element_t *element;
82
+
83
+ element = lxb_html_tree_insert_html_element(tree, token);
84
+ if (element == NULL) {
85
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
86
+
87
+ return lxb_html_tree_process_abort(tree);
88
+ }
89
+
90
+ lxb_html_tree_open_elements_pop(tree);
91
+ lxb_html_tree_acknowledge_token_self_closing(tree, token);
92
+
93
+ break;
94
+ }
95
+
96
+ case LXB_TAG_META: {
97
+ lxb_html_element_t *element;
98
+
99
+ element = lxb_html_tree_insert_html_element(tree, token);
100
+ if (element == NULL) {
101
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
102
+
103
+ return lxb_html_tree_process_abort(tree);
104
+ }
105
+
106
+ lxb_html_tree_open_elements_pop(tree);
107
+ lxb_html_tree_acknowledge_token_self_closing(tree, token);
108
+
109
+ /*
110
+ * TODO: Check encoding: charset attribute or http-equiv attribute.
111
+ */
112
+
113
+ break;
114
+ }
115
+
116
+ case LXB_TAG_TITLE: {
117
+ lxb_html_element_t *element;
118
+
119
+ element = lxb_html_tree_generic_rcdata_parsing(tree, token);
120
+ if (element == NULL) {
121
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
122
+
123
+ return lxb_html_tree_process_abort(tree);
124
+ }
125
+
126
+ break;
127
+ }
128
+
129
+ case LXB_TAG_NOSCRIPT: {
130
+ lxb_html_element_t *element;
131
+
132
+ if (tree->document->dom_document.scripting) {
133
+ element = lxb_html_tree_generic_rawtext_parsing(tree, token);
134
+ }
135
+ else {
136
+ element = lxb_html_tree_insert_html_element(tree, token);
137
+ tree->mode = lxb_html_tree_insertion_mode_in_head_noscript;
138
+ }
139
+
140
+ if (element == NULL) {
141
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
142
+
143
+ return lxb_html_tree_process_abort(tree);
144
+ }
145
+
146
+ break;
147
+ }
148
+
149
+ case LXB_TAG_NOFRAMES:
150
+ case LXB_TAG_STYLE: {
151
+ lxb_html_element_t *element;
152
+
153
+ element = lxb_html_tree_generic_rawtext_parsing(tree, token);
154
+ if (element == NULL) {
155
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
156
+
157
+ return lxb_html_tree_process_abort(tree);
158
+ }
159
+
160
+ break;
161
+ }
162
+
163
+ case LXB_TAG_SCRIPT:
164
+ return lxb_html_tree_insertion_mode_in_head_script(tree, token);
165
+
166
+ case LXB_TAG_TEMPLATE:
167
+ return lxb_html_tree_insertion_mode_in_head_template(tree, token);
168
+
169
+ case LXB_TAG_HEAD:
170
+ lxb_html_tree_parse_error(tree, token,
171
+ LXB_HTML_RULES_ERROR_HETOINHEMO);
172
+ break;
173
+
174
+ /*
175
+ * We can create function for this, but...
176
+ *
177
+ * The "in head noscript" insertion mode use this
178
+ * is you change this code, please, change it in in head noscript" mode
179
+ */
180
+ case LXB_TAG__TEXT: {
181
+ lxb_html_token_t ws_token = {0};
182
+
183
+ tree->status = lxb_html_token_data_split_ws_begin(token, &ws_token);
184
+ if (tree->status != LXB_STATUS_OK) {
185
+ return lxb_html_tree_process_abort(tree);
186
+ }
187
+
188
+ if (ws_token.text_start != ws_token.text_end) {
189
+ tree->status = lxb_html_tree_insert_character(tree, &ws_token,
190
+ NULL);
191
+ if (tree->status != LXB_STATUS_OK) {
192
+ return lxb_html_tree_process_abort(tree);
193
+ }
194
+ }
195
+
196
+ if (token->text_start == token->text_end) {
197
+ return true;
198
+ }
199
+ }
200
+ /* fall through */
201
+
202
+ default:
203
+ return lxb_html_tree_insertion_mode_in_head_anything_else(tree);
204
+ }
205
+
206
+ return true;
207
+ }
208
+
209
+ static bool
210
+ lxb_html_tree_insertion_mode_in_head_closed(lxb_html_tree_t *tree,
211
+ lxb_html_token_t *token)
212
+ {
213
+ switch (token->tag_id) {
214
+ case LXB_TAG_HEAD:
215
+ lxb_html_tree_open_elements_pop(tree);
216
+
217
+ tree->mode = lxb_html_tree_insertion_mode_after_head;
218
+
219
+ break;
220
+
221
+ case LXB_TAG_BODY:
222
+ case LXB_TAG_HTML:
223
+ case LXB_TAG_BR:
224
+ return lxb_html_tree_insertion_mode_in_head_anything_else(tree);
225
+
226
+ case LXB_TAG_TEMPLATE:
227
+ return lxb_html_tree_insertion_mode_in_head_template_closed(tree,
228
+ token);
229
+
230
+ default:
231
+ lxb_html_tree_parse_error(tree, token,
232
+ LXB_HTML_RULES_ERROR_UNCLTOINHEMO);
233
+ break;
234
+
235
+ }
236
+
237
+ return true;
238
+ }
239
+
240
+ lxb_inline bool
241
+ lxb_html_tree_insertion_mode_in_head_script(lxb_html_tree_t *tree,
242
+ lxb_html_token_t *token)
243
+ {
244
+ lxb_dom_node_t *ap_node;
245
+ lxb_html_element_t *element;
246
+ lxb_html_tree_insertion_position_t ipos;
247
+
248
+ ap_node = lxb_html_tree_appropriate_place_inserting_node(tree, NULL, &ipos);
249
+ if (ap_node == NULL) {
250
+ tree->status = LXB_STATUS_ERROR;
251
+
252
+ return lxb_html_tree_process_abort(tree);
253
+ }
254
+
255
+ if (ipos == LXB_HTML_TREE_INSERTION_POSITION_CHILD) {
256
+ element = lxb_html_tree_create_element_for_token(tree, token,
257
+ LXB_NS_HTML, ap_node);
258
+ }
259
+ else {
260
+ element = lxb_html_tree_create_element_for_token(tree, token,
261
+ LXB_NS_HTML,
262
+ ap_node->parent);
263
+ }
264
+
265
+ if (element == NULL) {
266
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
267
+
268
+ return lxb_html_tree_process_abort(tree);
269
+ }
270
+
271
+ /* TODO: Need code for set flags for Script Element */
272
+
273
+ tree->status = lxb_html_tree_open_elements_push(tree,
274
+ lxb_dom_interface_node(element));
275
+ if (tree->status != LXB_HTML_STATUS_OK) {
276
+ lxb_html_script_element_interface_destroy(lxb_html_interface_script(element));
277
+
278
+ return lxb_html_tree_process_abort(tree);
279
+ }
280
+
281
+ lxb_html_tree_insert_node(ap_node, lxb_dom_interface_node(element), ipos);
282
+
283
+ /*
284
+ * Need for tokenizer state Script
285
+ * See description for
286
+ * 'lxb_html_tokenizer_state_script_data_before' function
287
+ */
288
+ lxb_html_tokenizer_tmp_tag_id_set(tree->tkz_ref, token->tag_id);
289
+ lxb_html_tokenizer_state_set(tree->tkz_ref,
290
+ lxb_html_tokenizer_state_script_data_before);
291
+
292
+ tree->original_mode = tree->mode;
293
+ tree->mode = lxb_html_tree_insertion_mode_text;
294
+
295
+ return true;
296
+ }
297
+
298
+ lxb_inline bool
299
+ lxb_html_tree_insertion_mode_in_head_template(lxb_html_tree_t *tree,
300
+ lxb_html_token_t *token)
301
+ {
302
+ lxb_html_element_t *element;
303
+
304
+ element = lxb_html_tree_insert_html_element(tree, token);
305
+ if (element == NULL) {
306
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
307
+
308
+ return lxb_html_tree_process_abort(tree);
309
+ }
310
+
311
+ tree->status = lxb_html_tree_active_formatting_push_marker(tree);
312
+ if (tree->status != LXB_STATUS_OK) {
313
+ lxb_html_template_element_interface_destroy(lxb_html_interface_template(element));
314
+
315
+ return lxb_html_tree_process_abort(tree);
316
+ }
317
+
318
+ tree->frameset_ok = false;
319
+ tree->mode = lxb_html_tree_insertion_mode_in_template;
320
+
321
+ tree->status = lxb_html_tree_template_insertion_push(tree,
322
+ lxb_html_tree_insertion_mode_in_template);
323
+ if (tree->status != LXB_STATUS_OK) {
324
+ lxb_html_template_element_interface_destroy(lxb_html_interface_template(element));
325
+
326
+ return lxb_html_tree_process_abort(tree);
327
+ }
328
+
329
+ return true;
330
+ }
331
+
332
+ lxb_inline bool
333
+ lxb_html_tree_insertion_mode_in_head_template_closed(lxb_html_tree_t *tree,
334
+ lxb_html_token_t *token)
335
+ {
336
+ lxb_dom_node_t *temp_node;
337
+
338
+ temp_node = lxb_html_tree_open_elements_find_reverse(tree, LXB_TAG_TEMPLATE,
339
+ LXB_NS_HTML, NULL);
340
+ if (temp_node == NULL) {
341
+ lxb_html_tree_parse_error(tree, token,
342
+ LXB_HTML_RULES_ERROR_TECLTOWIOPINHEMO);
343
+ return true;
344
+ }
345
+
346
+ lxb_html_tree_generate_all_implied_end_tags_thoroughly(tree, LXB_TAG__UNDEF,
347
+ LXB_NS__UNDEF);
348
+
349
+ temp_node = lxb_html_tree_current_node(tree);
350
+
351
+ if (lxb_html_tree_node_is(temp_node, LXB_TAG_TEMPLATE) == false) {
352
+ lxb_html_tree_parse_error(tree, token,
353
+ LXB_HTML_RULES_ERROR_TEELISNOCUINHEMO);
354
+ }
355
+
356
+ lxb_html_tree_open_elements_pop_until_tag_id(tree, LXB_TAG_TEMPLATE,
357
+ LXB_NS_HTML, true);
358
+
359
+ lxb_html_tree_active_formatting_up_to_last_marker(tree);
360
+ lxb_html_tree_template_insertion_pop(tree);
361
+ lxb_html_tree_reset_insertion_mode_appropriately(tree);
362
+
363
+ return true;
364
+ }
365
+
366
+ lxb_inline bool
367
+ lxb_html_tree_insertion_mode_in_head_anything_else(lxb_html_tree_t *tree)
368
+ {
369
+ lxb_html_tree_open_elements_pop(tree);
370
+
371
+ tree->mode = lxb_html_tree_insertion_mode_after_head;
372
+
373
+ return false;
374
+ }
@@ -0,0 +1,121 @@
1
+ /*
2
+ * Copyright (C) 2018-2020 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #include "lexbor/html/tree/insertion_mode.h"
8
+ #include "lexbor/html/tree/open_elements.h"
9
+
10
+
11
+ static bool
12
+ lxb_html_tree_insertion_mode_in_head_noscript_open(lxb_html_tree_t *tree,
13
+ lxb_html_token_t *token);
14
+
15
+ static bool
16
+ lxb_html_tree_insertion_mode_in_head_noscript_closed(lxb_html_tree_t *tree,
17
+ lxb_html_token_t *token);
18
+
19
+
20
+ bool
21
+ lxb_html_tree_insertion_mode_in_head_noscript(lxb_html_tree_t *tree,
22
+ lxb_html_token_t *token)
23
+ {
24
+ if (token->type & LXB_HTML_TOKEN_TYPE_CLOSE) {
25
+ return lxb_html_tree_insertion_mode_in_head_noscript_closed(tree, token);
26
+ }
27
+
28
+ return lxb_html_tree_insertion_mode_in_head_noscript_open(tree, token);
29
+ }
30
+
31
+ lxb_inline bool
32
+ lxb_html_tree_insertion_mode_in_head_noscript_anything_else(lxb_html_tree_t *tree,
33
+ lxb_html_token_t *token);
34
+
35
+
36
+ static bool
37
+ lxb_html_tree_insertion_mode_in_head_noscript_open(lxb_html_tree_t *tree,
38
+ lxb_html_token_t *token)
39
+ {
40
+ switch (token->tag_id) {
41
+ case LXB_TAG__EM_DOCTYPE:
42
+ lxb_html_tree_parse_error(tree, token,
43
+ LXB_HTML_RULES_ERROR_DOTOINHENOMO);
44
+ break;
45
+
46
+ case LXB_TAG_HTML:
47
+ return lxb_html_tree_insertion_mode_in_body(tree, token);
48
+
49
+ case LXB_TAG__EM_COMMENT:
50
+ case LXB_TAG_BASEFONT:
51
+ case LXB_TAG_BGSOUND:
52
+ case LXB_TAG_LINK:
53
+ case LXB_TAG_META:
54
+ case LXB_TAG_NOFRAMES:
55
+ case LXB_TAG_STYLE:
56
+ return lxb_html_tree_insertion_mode_in_head(tree, token);
57
+
58
+ case LXB_TAG_HEAD:
59
+ case LXB_TAG_NOSCRIPT:
60
+ lxb_html_tree_parse_error(tree, token,
61
+ LXB_HTML_RULES_ERROR_UNTO);
62
+ break;
63
+
64
+ /* CopyPast from "in head" insertion mode */
65
+ case LXB_TAG__TEXT: {
66
+ lxb_html_token_t ws_token = {0};
67
+
68
+ tree->status = lxb_html_token_data_split_ws_begin(token, &ws_token);
69
+ if (tree->status != LXB_STATUS_OK) {
70
+ return lxb_html_tree_process_abort(tree);
71
+ }
72
+
73
+ if (ws_token.text_start != ws_token.text_end) {
74
+ tree->status = lxb_html_tree_insert_character(tree, &ws_token,
75
+ NULL);
76
+ if (tree->status != LXB_STATUS_OK) {
77
+ return lxb_html_tree_process_abort(tree);
78
+ }
79
+ }
80
+
81
+ if (token->text_start == token->text_end) {
82
+ return true;
83
+ }
84
+ }
85
+ /* fall through */
86
+
87
+ default:
88
+ return lxb_html_tree_insertion_mode_in_head_noscript_anything_else(tree,
89
+ token);
90
+ }
91
+
92
+ return true;
93
+ }
94
+
95
+ static bool
96
+ lxb_html_tree_insertion_mode_in_head_noscript_closed(lxb_html_tree_t *tree,
97
+ lxb_html_token_t *token)
98
+ {
99
+ if(token->tag_id == LXB_TAG_BR) {
100
+ return lxb_html_tree_insertion_mode_in_head_noscript_anything_else(tree,
101
+ token);
102
+ }
103
+
104
+ lxb_html_tree_parse_error(tree, token, LXB_HTML_RULES_ERROR_UNTO);
105
+
106
+ return true;
107
+ }
108
+
109
+ lxb_inline bool
110
+ lxb_html_tree_insertion_mode_in_head_noscript_anything_else(lxb_html_tree_t *tree,
111
+ lxb_html_token_t *token)
112
+ {
113
+ lxb_html_tree_parse_error(tree, token,
114
+ LXB_HTML_RULES_ERROR_UNTO);
115
+
116
+ lxb_html_tree_open_elements_pop(tree);
117
+
118
+ tree->mode = lxb_html_tree_insertion_mode_in_head;
119
+
120
+ return false;
121
+ }
@@ -0,0 +1,211 @@
1
+ /*
2
+ * Copyright (C) 2018-2019 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #include "lexbor/html/tree/insertion_mode.h"
8
+ #include "lexbor/html/tree/open_elements.h"
9
+ #include "lexbor/html/tree/active_formatting.h"
10
+
11
+
12
+ static void
13
+ lxb_html_tree_clear_stack_back_to_table_row(lxb_html_tree_t *tree)
14
+ {
15
+ lxb_dom_node_t *current = lxb_html_tree_current_node(tree);
16
+
17
+ while ((current->local_name != LXB_TAG_TR
18
+ && current->local_name != LXB_TAG_TEMPLATE
19
+ && current->local_name != LXB_TAG_HTML)
20
+ || current->ns != LXB_NS_HTML)
21
+ {
22
+ lxb_html_tree_open_elements_pop(tree);
23
+ current = lxb_html_tree_current_node(tree);
24
+ }
25
+ }
26
+
27
+ /*
28
+ * "th", "td"
29
+ */
30
+ lxb_inline bool
31
+ lxb_html_tree_insertion_mode_in_row_thtd(lxb_html_tree_t *tree,
32
+ lxb_html_token_t *token)
33
+ {
34
+ lxb_html_element_t *element;
35
+
36
+ lxb_html_tree_clear_stack_back_to_table_row(tree);
37
+
38
+ element = lxb_html_tree_insert_html_element(tree, token);
39
+ if (element == NULL) {
40
+ tree->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
41
+
42
+ return lxb_html_tree_process_abort(tree);
43
+ }
44
+
45
+ tree->mode = lxb_html_tree_insertion_mode_in_cell;
46
+
47
+ tree->status = lxb_html_tree_active_formatting_push_marker(tree);
48
+ if (tree->status != LXB_STATUS_OK) {
49
+ return lxb_html_tree_process_abort(tree);
50
+ }
51
+
52
+ return true;
53
+ }
54
+
55
+ lxb_inline bool
56
+ lxb_html_tree_insertion_mode_in_row_tr_closed(lxb_html_tree_t *tree,
57
+ lxb_html_token_t *token)
58
+ {
59
+ lxb_dom_node_t *node;
60
+
61
+ node = lxb_html_tree_element_in_scope(tree, LXB_TAG_TR, LXB_NS_HTML,
62
+ LXB_HTML_TAG_CATEGORY_SCOPE_TABLE);
63
+ if (node == NULL) {
64
+ lxb_html_tree_parse_error(tree, token, LXB_HTML_RULES_ERROR_UNCLTO);
65
+
66
+ return true;
67
+ }
68
+
69
+ lxb_html_tree_clear_stack_back_to_table_row(tree);
70
+ lxb_html_tree_open_elements_pop(tree);
71
+
72
+ tree->mode = lxb_html_tree_insertion_mode_in_table_body;
73
+
74
+ return true;
75
+ }
76
+
77
+ /*
78
+ * A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody",
79
+ * "tfoot", "thead", "tr"
80
+ * An end tag whose tag name is "table"
81
+ */
82
+ lxb_inline bool
83
+ lxb_html_tree_insertion_mode_in_row_ct_open_closed(lxb_html_tree_t *tree,
84
+ lxb_html_token_t *token)
85
+ {
86
+ lxb_dom_node_t *node;
87
+
88
+ node = lxb_html_tree_element_in_scope(tree, LXB_TAG_TR, LXB_NS_HTML,
89
+ LXB_HTML_TAG_CATEGORY_SCOPE_TABLE);
90
+ if (node == NULL) {
91
+ lxb_html_tree_parse_error(tree, token, LXB_HTML_RULES_ERROR_UNTO);
92
+
93
+ return true;
94
+ }
95
+
96
+ lxb_html_tree_clear_stack_back_to_table_row(tree);
97
+ lxb_html_tree_open_elements_pop(tree);
98
+
99
+ tree->mode = lxb_html_tree_insertion_mode_in_table_body;
100
+
101
+ return false;
102
+ }
103
+
104
+ /*
105
+ * "tbody", "tfoot", "thead"
106
+ */
107
+ lxb_inline bool
108
+ lxb_html_tree_insertion_mode_in_row_tbtfth_closed(lxb_html_tree_t *tree,
109
+ lxb_html_token_t *token)
110
+ {
111
+ lxb_dom_node_t *node;
112
+
113
+ node = lxb_html_tree_element_in_scope(tree, token->tag_id, LXB_NS_HTML,
114
+ LXB_HTML_TAG_CATEGORY_SCOPE_TABLE);
115
+ if (node == NULL) {
116
+ lxb_html_tree_parse_error(tree, token, LXB_HTML_RULES_ERROR_UNCLTO);
117
+
118
+ return true;
119
+ }
120
+
121
+ node = lxb_html_tree_element_in_scope(tree, LXB_TAG_TR, LXB_NS_HTML,
122
+ LXB_HTML_TAG_CATEGORY_SCOPE_TABLE);
123
+ if (node == NULL) {
124
+ return true;
125
+ }
126
+
127
+ lxb_html_tree_clear_stack_back_to_table_row(tree);
128
+ lxb_html_tree_open_elements_pop(tree);
129
+
130
+ tree->mode = lxb_html_tree_insertion_mode_in_table_body;
131
+
132
+ return false;
133
+ }
134
+
135
+ /*
136
+ * "body", "caption", "col", "colgroup", "html", "td", "th"
137
+ */
138
+ lxb_inline bool
139
+ lxb_html_tree_insertion_mode_in_row_bcht_closed(lxb_html_tree_t *tree,
140
+ lxb_html_token_t *token)
141
+ {
142
+ lxb_html_tree_parse_error(tree, token, LXB_HTML_RULES_ERROR_UNCLTO);
143
+
144
+ return true;
145
+ }
146
+
147
+ lxb_inline bool
148
+ lxb_html_tree_insertion_mode_in_row_anything_else(lxb_html_tree_t *tree,
149
+ lxb_html_token_t *token)
150
+ {
151
+ return lxb_html_tree_insertion_mode_in_table(tree, token);
152
+ }
153
+
154
+ lxb_inline bool
155
+ lxb_html_tree_insertion_mode_in_row_anything_else_closed(lxb_html_tree_t *tree,
156
+ lxb_html_token_t *token)
157
+ {
158
+ return lxb_html_tree_insertion_mode_in_row_anything_else(tree, token);
159
+ }
160
+
161
+ bool
162
+ lxb_html_tree_insertion_mode_in_row(lxb_html_tree_t *tree,
163
+ lxb_html_token_t *token)
164
+ {
165
+ if (token->type & LXB_HTML_TOKEN_TYPE_CLOSE) {
166
+ switch (token->tag_id) {
167
+ case LXB_TAG_TR:
168
+ return lxb_html_tree_insertion_mode_in_row_tr_closed(tree,
169
+ token);
170
+ case LXB_TAG_TABLE:
171
+ return lxb_html_tree_insertion_mode_in_row_ct_open_closed(tree,
172
+ token);
173
+ case LXB_TAG_TBODY:
174
+ case LXB_TAG_TFOOT:
175
+ case LXB_TAG_THEAD:
176
+ return lxb_html_tree_insertion_mode_in_row_tbtfth_closed(tree,
177
+ token);
178
+ case LXB_TAG_BODY:
179
+ case LXB_TAG_CAPTION:
180
+ case LXB_TAG_COL:
181
+ case LXB_TAG_COLGROUP:
182
+ case LXB_TAG_HTML:
183
+ case LXB_TAG_TD:
184
+ case LXB_TAG_TH:
185
+ return lxb_html_tree_insertion_mode_in_row_bcht_closed(tree,
186
+ token);
187
+ default:
188
+ return lxb_html_tree_insertion_mode_in_row_anything_else_closed(tree,
189
+ token);
190
+ }
191
+ }
192
+
193
+ switch (token->tag_id) {
194
+ case LXB_TAG_TH:
195
+ case LXB_TAG_TD:
196
+ return lxb_html_tree_insertion_mode_in_row_thtd(tree, token);
197
+
198
+ case LXB_TAG_CAPTION:
199
+ case LXB_TAG_COL:
200
+ case LXB_TAG_COLGROUP:
201
+ case LXB_TAG_TBODY:
202
+ case LXB_TAG_TFOOT:
203
+ case LXB_TAG_THEAD:
204
+ case LXB_TAG_TR:
205
+ return lxb_html_tree_insertion_mode_in_row_ct_open_closed(tree,
206
+ token);
207
+ default:
208
+ return lxb_html_tree_insertion_mode_in_row_anything_else(tree,
209
+ token);
210
+ }
211
+ }