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 +1,343 @@
1
+ /*
2
+ * Copyright (C) 2018-2020 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #ifndef LEXBOR_HTML_TOKENIZER_H
8
+ #define LEXBOR_HTML_TOKENIZER_H
9
+
10
+ #ifdef __cplusplus
11
+ extern "C" {
12
+ #endif
13
+
14
+ #include "lexbor/core/sbst.h"
15
+ #include "lexbor/core/array_obj.h"
16
+
17
+ #include "lexbor/html/base.h"
18
+ #include "lexbor/html/token.h"
19
+
20
+ #include "lexbor/tag/tag.h"
21
+ #include "lexbor/ns/ns.h"
22
+
23
+
24
+ /* State */
25
+ typedef const lxb_char_t *
26
+ (*lxb_html_tokenizer_state_f)(lxb_html_tokenizer_t *tkz,
27
+ const lxb_char_t *data, const lxb_char_t *end);
28
+
29
+ typedef lxb_html_token_t *
30
+ (*lxb_html_tokenizer_token_f)(lxb_html_tokenizer_t *tkz,
31
+ lxb_html_token_t *token, void *ctx);
32
+
33
+
34
+ struct lxb_html_tokenizer {
35
+ lxb_html_tokenizer_state_f state;
36
+ lxb_html_tokenizer_state_f state_return;
37
+
38
+ lxb_html_tokenizer_token_f callback_token_done;
39
+ void *callback_token_ctx;
40
+
41
+ lexbor_hash_t *tags;
42
+ lexbor_hash_t *attrs;
43
+ lexbor_mraw_t *attrs_mraw;
44
+
45
+ /* For a temp strings and other templary data */
46
+ lexbor_mraw_t *mraw;
47
+
48
+ /* Current process token */
49
+ lxb_html_token_t *token;
50
+
51
+ /* Memory for token and attr */
52
+ lexbor_dobject_t *dobj_token;
53
+ lexbor_dobject_t *dobj_token_attr;
54
+
55
+ /* Parse error */
56
+ lexbor_array_obj_t *parse_errors;
57
+
58
+ /*
59
+ * Leak abstractions.
60
+ * The only place where the specification causes mixing Tree Builder
61
+ * and Tokenizer. We kill all beauty.
62
+ * Current Tree parser. This is not ref (not ref count).
63
+ */
64
+ lxb_html_tree_t *tree;
65
+
66
+ /* Temp */
67
+ const lxb_char_t *markup;
68
+ const lxb_char_t *temp;
69
+ lxb_tag_id_t tmp_tag_id;
70
+
71
+ lxb_char_t *start;
72
+ lxb_char_t *pos;
73
+ const lxb_char_t *end;
74
+ const lxb_char_t *begin;
75
+ const lxb_char_t *last;
76
+
77
+ /* Entities */
78
+ const lexbor_sbst_entry_static_t *entity;
79
+ const lexbor_sbst_entry_static_t *entity_match;
80
+ uintptr_t entity_start;
81
+ uintptr_t entity_end;
82
+ uint32_t entity_length;
83
+ uint32_t entity_number;
84
+ bool is_attribute;
85
+
86
+ /* Process */
87
+ lxb_html_tokenizer_opt_t opt;
88
+ lxb_status_t status;
89
+ bool is_eof;
90
+
91
+ lxb_html_tokenizer_t *base;
92
+ size_t ref_count;
93
+ };
94
+
95
+
96
+ #include "lexbor/html/tokenizer/error.h"
97
+
98
+
99
+ extern const lxb_char_t *lxb_html_tokenizer_eof;
100
+
101
+ LXB_API lxb_html_tokenizer_t *
102
+ lxb_html_tokenizer_create(void);
103
+
104
+ LXB_API lxb_status_t
105
+ lxb_html_tokenizer_init(lxb_html_tokenizer_t *tkz);
106
+
107
+ LXB_API lxb_status_t
108
+ lxb_html_tokenizer_inherit(lxb_html_tokenizer_t *tkz_to,
109
+ lxb_html_tokenizer_t *tkz_from);
110
+
111
+ LXB_API lxb_html_tokenizer_t *
112
+ lxb_html_tokenizer_ref(lxb_html_tokenizer_t *tkz);
113
+
114
+ LXB_API lxb_html_tokenizer_t *
115
+ lxb_html_tokenizer_unref(lxb_html_tokenizer_t *tkz);
116
+
117
+ LXB_API void
118
+ lxb_html_tokenizer_clean(lxb_html_tokenizer_t *tkz);
119
+
120
+ LXB_API lxb_html_tokenizer_t *
121
+ lxb_html_tokenizer_destroy(lxb_html_tokenizer_t *tkz);
122
+
123
+ LXB_API lxb_status_t
124
+ lxb_html_tokenizer_tags_make(lxb_html_tokenizer_t *tkz, size_t table_size);
125
+
126
+ LXB_API void
127
+ lxb_html_tokenizer_tags_destroy(lxb_html_tokenizer_t *tkz);
128
+
129
+ LXB_API lxb_status_t
130
+ lxb_html_tokenizer_attrs_make(lxb_html_tokenizer_t *tkz, size_t table_size);
131
+
132
+ LXB_API void
133
+ lxb_html_tokenizer_attrs_destroy(lxb_html_tokenizer_t *tkz);
134
+
135
+ LXB_API lxb_status_t
136
+ lxb_html_tokenizer_begin(lxb_html_tokenizer_t *tkz);
137
+
138
+ LXB_API lxb_status_t
139
+ lxb_html_tokenizer_chunk(lxb_html_tokenizer_t *tkz,
140
+ const lxb_char_t *data, size_t size);
141
+
142
+ LXB_API lxb_status_t
143
+ lxb_html_tokenizer_end(lxb_html_tokenizer_t *tkz);
144
+
145
+
146
+ LXB_API const lxb_char_t *
147
+ lxb_html_tokenizer_change_incoming(lxb_html_tokenizer_t *tkz,
148
+ const lxb_char_t *pos);
149
+
150
+ LXB_API lxb_ns_id_t
151
+ lxb_html_tokenizer_current_namespace(lxb_html_tokenizer_t *tkz);
152
+
153
+ LXB_API void
154
+ lxb_html_tokenizer_set_state_by_tag(lxb_html_tokenizer_t *tkz, bool scripting,
155
+ lxb_tag_id_t tag_id, lxb_ns_id_t ns);
156
+
157
+
158
+ /*
159
+ * Inline functions
160
+ */
161
+ lxb_inline void
162
+ lxb_html_tokenizer_status_set(lxb_html_tokenizer_t *tkz, lxb_status_t status)
163
+ {
164
+ tkz->status = status;
165
+ }
166
+
167
+ lxb_inline void
168
+ lxb_html_tokenizer_tags_set(lxb_html_tokenizer_t *tkz, lexbor_hash_t *tags)
169
+ {
170
+ tkz->tags = tags;
171
+ }
172
+
173
+ lxb_inline lexbor_hash_t *
174
+ lxb_html_tokenizer_tags(lxb_html_tokenizer_t *tkz)
175
+ {
176
+ return tkz->tags;
177
+ }
178
+
179
+ lxb_inline void
180
+ lxb_html_tokenizer_attrs_set(lxb_html_tokenizer_t *tkz, lexbor_hash_t *attrs)
181
+ {
182
+ tkz->attrs = attrs;
183
+ }
184
+
185
+ lxb_inline lexbor_hash_t *
186
+ lxb_html_tokenizer_attrs(lxb_html_tokenizer_t *tkz)
187
+ {
188
+ return tkz->attrs;
189
+ }
190
+
191
+ lxb_inline void
192
+ lxb_html_tokenizer_attrs_mraw_set(lxb_html_tokenizer_t *tkz,
193
+ lexbor_mraw_t *mraw)
194
+ {
195
+ tkz->attrs_mraw = mraw;
196
+ }
197
+
198
+ lxb_inline lexbor_mraw_t *
199
+ lxb_html_tokenizer_attrs_mraw(lxb_html_tokenizer_t *tkz)
200
+ {
201
+ return tkz->attrs_mraw;
202
+ }
203
+
204
+ lxb_inline void
205
+ lxb_html_tokenizer_callback_token_done_set(lxb_html_tokenizer_t *tkz,
206
+ lxb_html_tokenizer_token_f call_func,
207
+ void *ctx)
208
+ {
209
+ tkz->callback_token_done = call_func;
210
+ tkz->callback_token_ctx = ctx;
211
+ }
212
+
213
+ lxb_inline void *
214
+ lxb_html_tokenizer_callback_token_done_ctx(lxb_html_tokenizer_t *tkz)
215
+ {
216
+ return tkz->callback_token_ctx;
217
+ }
218
+
219
+ lxb_inline void
220
+ lxb_html_tokenizer_state_set(lxb_html_tokenizer_t *tkz,
221
+ lxb_html_tokenizer_state_f state)
222
+ {
223
+ tkz->state = state;
224
+ }
225
+
226
+ lxb_inline void
227
+ lxb_html_tokenizer_tmp_tag_id_set(lxb_html_tokenizer_t *tkz,
228
+ lxb_tag_id_t tag_id)
229
+ {
230
+ tkz->tmp_tag_id = tag_id;
231
+ }
232
+
233
+ lxb_inline lxb_html_tree_t *
234
+ lxb_html_tokenizer_tree(lxb_html_tokenizer_t *tkz)
235
+ {
236
+ return tkz->tree;
237
+ }
238
+
239
+ lxb_inline void
240
+ lxb_html_tokenizer_tree_set(lxb_html_tokenizer_t *tkz, lxb_html_tree_t *tree)
241
+ {
242
+ tkz->tree = tree;
243
+ }
244
+
245
+ lxb_inline lexbor_mraw_t *
246
+ lxb_html_tokenizer_mraw(lxb_html_tokenizer_t *tkz)
247
+ {
248
+ return tkz->mraw;
249
+ }
250
+
251
+ lxb_inline lxb_status_t
252
+ lxb_html_tokenizer_temp_realloc(lxb_html_tokenizer_t *tkz, size_t size)
253
+ {
254
+ size_t length = tkz->pos - tkz->start;
255
+ size_t new_size = (tkz->end - tkz->start) + size + 4096;
256
+
257
+ tkz->start = (lxb_char_t *)lexbor_realloc(tkz->start, new_size);
258
+ if (tkz->start == NULL) {
259
+ tkz->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
260
+ return tkz->status;
261
+ }
262
+
263
+ tkz->pos = tkz->start + length;
264
+ tkz->end = tkz->start + new_size;
265
+
266
+ return LXB_STATUS_OK;
267
+ }
268
+
269
+ lxb_inline lxb_status_t
270
+ lxb_html_tokenizer_temp_append_data(lxb_html_tokenizer_t *tkz,
271
+ const lxb_char_t *data)
272
+ {
273
+ size_t size = data - tkz->begin;
274
+
275
+ if ((tkz->pos + size) > tkz->end) {
276
+ if(lxb_html_tokenizer_temp_realloc(tkz, size)) {
277
+ return tkz->status;
278
+ }
279
+ }
280
+
281
+ tkz->pos = (lxb_char_t *) memcpy(tkz->pos, tkz->begin, size) + size;
282
+
283
+ return LXB_STATUS_OK;
284
+ }
285
+
286
+ lxb_inline lxb_status_t
287
+ lxb_html_tokenizer_temp_append(lxb_html_tokenizer_t *tkz,
288
+ const lxb_char_t *data, size_t size)
289
+ {
290
+ if ((tkz->pos + size) > tkz->end) {
291
+ if(lxb_html_tokenizer_temp_realloc(tkz, size)) {
292
+ return tkz->status;
293
+ }
294
+ }
295
+
296
+ tkz->pos = (lxb_char_t *) memcpy(tkz->pos, data, size) + size;
297
+
298
+ return LXB_STATUS_OK;
299
+ }
300
+
301
+
302
+ /*
303
+ * No inline functions for ABI.
304
+ */
305
+ LXB_API void
306
+ lxb_html_tokenizer_status_set_noi(lxb_html_tokenizer_t *tkz,
307
+ lxb_status_t status);
308
+
309
+ LXB_API void
310
+ lxb_html_tokenizer_callback_token_done_set_noi(lxb_html_tokenizer_t *tkz,
311
+ lxb_html_tokenizer_token_f call_func,
312
+ void *ctx);
313
+
314
+ LXB_API void *
315
+ lxb_html_tokenizer_callback_token_done_ctx_noi(lxb_html_tokenizer_t *tkz);
316
+
317
+ LXB_API void
318
+ lxb_html_tokenizer_state_set_noi(lxb_html_tokenizer_t *tkz,
319
+ lxb_html_tokenizer_state_f state);
320
+
321
+ LXB_API void
322
+ lxb_html_tokenizer_tmp_tag_id_set_noi(lxb_html_tokenizer_t *tkz,
323
+ lxb_tag_id_t tag_id);
324
+
325
+ LXB_API lxb_html_tree_t *
326
+ lxb_html_tokenizer_tree_noi(lxb_html_tokenizer_t *tkz);
327
+
328
+ LXB_API void
329
+ lxb_html_tokenizer_tree_set_noi(lxb_html_tokenizer_t *tkz,
330
+ lxb_html_tree_t *tree);
331
+
332
+ LXB_API lexbor_mraw_t *
333
+ lxb_html_tokenizer_mraw_noi(lxb_html_tokenizer_t *tkz);
334
+
335
+ LXB_API lexbor_hash_t *
336
+ lxb_html_tokenizer_tags_noi(lxb_html_tokenizer_t *tkz);
337
+
338
+
339
+ #ifdef __cplusplus
340
+ } /* extern "C" */
341
+ #endif
342
+
343
+ #endif /* LEXBOR_HTML_TOKENIZER_H */
@@ -0,0 +1,241 @@
1
+ /*
2
+ * Copyright (C) 2018 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #include "lexbor/dom/interfaces/node.h"
8
+
9
+ #include "lexbor/html/tree/active_formatting.h"
10
+ #include "lexbor/html/tree/open_elements.h"
11
+ #include "lexbor/html/interfaces/element.h"
12
+
13
+
14
+ static lxb_html_element_t lxb_html_tree_active_formatting_marker_static;
15
+
16
+ static lxb_dom_node_t *lxb_html_tree_active_formatting_marker_node_static =
17
+ (lxb_dom_node_t *) &lxb_html_tree_active_formatting_marker_static;
18
+
19
+
20
+ lxb_html_element_t *
21
+ lxb_html_tree_active_formatting_marker(void)
22
+ {
23
+ return &lxb_html_tree_active_formatting_marker_static;
24
+ }
25
+
26
+ void
27
+ lxb_html_tree_active_formatting_up_to_last_marker(lxb_html_tree_t *tree)
28
+ {
29
+ void **list = tree->active_formatting->list;
30
+
31
+ while (tree->active_formatting->length != 0) {
32
+ tree->active_formatting->length--;
33
+
34
+ if (list[tree->active_formatting->length]
35
+ == &lxb_html_tree_active_formatting_marker_static)
36
+ {
37
+ break;
38
+ }
39
+ }
40
+ }
41
+
42
+ void
43
+ lxb_html_tree_active_formatting_remove_by_node(lxb_html_tree_t *tree,
44
+ lxb_dom_node_t *node)
45
+ {
46
+ size_t delta;
47
+ void **list = tree->active_formatting->list;
48
+ size_t idx = tree->active_formatting->length;
49
+
50
+ while (idx != 0) {
51
+ idx--;
52
+
53
+ if (list[idx] == node) {
54
+ delta = tree->active_formatting->length - idx - 1;
55
+
56
+ memmove(list + idx, list + idx + 1, sizeof(void *) * delta);
57
+
58
+ tree->active_formatting->length--;
59
+
60
+ break;
61
+ }
62
+ }
63
+ }
64
+
65
+ bool
66
+ lxb_html_tree_active_formatting_find_by_node(lxb_html_tree_t *tree,
67
+ lxb_dom_node_t *node,
68
+ size_t *return_pos)
69
+ {
70
+ void **list = tree->active_formatting->list;
71
+
72
+ for (size_t i = 0; i < tree->active_formatting->length; i++) {
73
+ if (list[i] == node) {
74
+ if (return_pos) {
75
+ *return_pos = i;
76
+ }
77
+
78
+ return true;
79
+ }
80
+ }
81
+
82
+ if (return_pos) {
83
+ *return_pos = 0;
84
+ }
85
+
86
+ return false;
87
+ }
88
+
89
+ bool
90
+ lxb_html_tree_active_formatting_find_by_node_reverse(lxb_html_tree_t *tree,
91
+ lxb_dom_node_t *node,
92
+ size_t *return_pos)
93
+ {
94
+ void **list = tree->active_formatting->list;
95
+ size_t len = tree->active_formatting->length;
96
+
97
+ while (len != 0) {
98
+ len--;
99
+
100
+ if (list[len] == node) {
101
+ if (return_pos) {
102
+ *return_pos = len;
103
+ }
104
+
105
+ return true;
106
+ }
107
+ }
108
+
109
+ if (return_pos) {
110
+ *return_pos = 0;
111
+ }
112
+
113
+ return false;
114
+ }
115
+
116
+ lxb_status_t
117
+ lxb_html_tree_active_formatting_reconstruct_elements(lxb_html_tree_t *tree)
118
+ {
119
+ /* Step 1 */
120
+ if (tree->active_formatting->length == 0) {
121
+ return LXB_STATUS_OK;
122
+ }
123
+
124
+ lexbor_array_t *af = tree->active_formatting;
125
+ void **list = af->list;
126
+
127
+ /* Step 2-3 */
128
+ size_t af_idx = af->length - 1;
129
+
130
+ if(list[af_idx] == &lxb_html_tree_active_formatting_marker_static
131
+ || lxb_html_tree_open_elements_find_by_node_reverse(tree, list[af_idx],
132
+ NULL))
133
+ {
134
+ return LXB_STATUS_OK;
135
+ }
136
+
137
+ /*
138
+ * Step 4-6
139
+ * Rewind
140
+ */
141
+ while (af_idx != 0) {
142
+ af_idx--;
143
+
144
+ if(list[af_idx] == &lxb_html_tree_active_formatting_marker_static ||
145
+ lxb_html_tree_open_elements_find_by_node_reverse(tree, list[af_idx],
146
+ NULL))
147
+ {
148
+ /* Step 7 */
149
+ af_idx++;
150
+
151
+ break;
152
+ }
153
+ }
154
+
155
+ /*
156
+ * Step 8-10
157
+ * Create
158
+ */
159
+ lxb_dom_node_t *node;
160
+ lxb_html_element_t *element;
161
+ lxb_html_token_t fake_token = {0};
162
+
163
+ while (af_idx < af->length) {
164
+ node = list[af_idx];
165
+
166
+ fake_token.tag_id = node->local_name;
167
+ fake_token.base_element = node;
168
+
169
+ element = lxb_html_tree_insert_html_element(tree, &fake_token);
170
+ if (element == NULL) {
171
+ return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
172
+ }
173
+
174
+ /* Step 9 */
175
+ list[af_idx] = lxb_dom_interface_node(element);
176
+
177
+ /* Step 10 */
178
+ af_idx++;
179
+ }
180
+
181
+ return LXB_STATUS_OK;
182
+ }
183
+
184
+ lxb_dom_node_t *
185
+ lxb_html_tree_active_formatting_between_last_marker(lxb_html_tree_t *tree,
186
+ lxb_tag_id_t tag_idx,
187
+ size_t *return_idx)
188
+ {
189
+ lxb_dom_node_t **list = (lxb_dom_node_t **) tree->active_formatting->list;
190
+ size_t idx = tree->active_formatting->length;
191
+
192
+ while (idx) {
193
+ idx--;
194
+
195
+ if (list[idx] == lxb_html_tree_active_formatting_marker_node_static) {
196
+ return NULL;
197
+ }
198
+
199
+ if (list[idx]->local_name == tag_idx && list[idx]->ns == LXB_NS_HTML) {
200
+ if (return_idx) {
201
+ *return_idx = idx;
202
+ }
203
+
204
+ return list[idx];
205
+ }
206
+ }
207
+
208
+ return NULL;
209
+ }
210
+
211
+ void
212
+ lxb_html_tree_active_formatting_push_with_check_dupl(lxb_html_tree_t *tree,
213
+ lxb_dom_node_t *node)
214
+ {
215
+ lxb_dom_node_t **list = (lxb_dom_node_t **) tree->active_formatting->list;
216
+ size_t idx = tree->active_formatting->length;
217
+ size_t earliest_idx = (idx ? (idx - 1) : 0);
218
+ size_t count = 0;
219
+
220
+ while (idx) {
221
+ idx--;
222
+
223
+ if (list[idx] == lxb_html_tree_active_formatting_marker_node_static) {
224
+ break;
225
+ }
226
+
227
+ if(list[idx]->local_name == node->local_name && list[idx]->ns == node->ns
228
+ && lxb_dom_element_compare(lxb_dom_interface_element(list[idx]),
229
+ lxb_dom_interface_element(node)))
230
+ {
231
+ count++;
232
+ earliest_idx = idx;
233
+ }
234
+ }
235
+
236
+ if(count >= 3) {
237
+ lxb_html_tree_active_formatting_remove(tree, earliest_idx);
238
+ }
239
+
240
+ lxb_html_tree_active_formatting_push(tree, node);
241
+ }
@@ -0,0 +1,117 @@
1
+ /*
2
+ * Copyright (C) 2018 Alexander Borisov
3
+ *
4
+ * Author: Alexander Borisov <borisov@lexbor.com>
5
+ */
6
+
7
+ #ifndef LEXBOR_HTML_ACTIVE_FORMATTING_H
8
+ #define LEXBOR_HTML_ACTIVE_FORMATTING_H
9
+
10
+ #ifdef __cplusplus
11
+ extern "C" {
12
+ #endif
13
+
14
+ #include "lexbor/core/array.h"
15
+
16
+ #include "lexbor/html/tree.h"
17
+
18
+
19
+ lxb_html_element_t *
20
+ lxb_html_tree_active_formatting_marker(void);
21
+
22
+ void
23
+ lxb_html_tree_active_formatting_up_to_last_marker(lxb_html_tree_t *tree);
24
+
25
+ void
26
+ lxb_html_tree_active_formatting_remove_by_node(lxb_html_tree_t *tree,
27
+ lxb_dom_node_t *node);
28
+
29
+ bool
30
+ lxb_html_tree_active_formatting_find_by_node(lxb_html_tree_t *tree,
31
+ lxb_dom_node_t *node,
32
+ size_t *return_pos);
33
+
34
+ bool
35
+ lxb_html_tree_active_formatting_find_by_node_reverse(lxb_html_tree_t *tree,
36
+ lxb_dom_node_t *node,
37
+ size_t *return_pos);
38
+
39
+ lxb_status_t
40
+ lxb_html_tree_active_formatting_reconstruct_elements(lxb_html_tree_t *tree);
41
+
42
+ lxb_dom_node_t *
43
+ lxb_html_tree_active_formatting_between_last_marker(lxb_html_tree_t *tree,
44
+ lxb_tag_id_t tag_idx,
45
+ size_t *return_idx);
46
+
47
+ void
48
+ lxb_html_tree_active_formatting_push_with_check_dupl(lxb_html_tree_t *tree,
49
+ lxb_dom_node_t *node);
50
+
51
+
52
+ /*
53
+ * Inline functions
54
+ */
55
+ lxb_inline lxb_dom_node_t *
56
+ lxb_html_tree_active_formatting_current_node(lxb_html_tree_t *tree)
57
+ {
58
+ if (tree->active_formatting->length == 0) {
59
+ return NULL;
60
+ }
61
+
62
+ return (lxb_dom_node_t *) tree->active_formatting->list
63
+ [ (tree->active_formatting->length - 1) ];
64
+ }
65
+
66
+ lxb_inline lxb_dom_node_t *
67
+ lxb_html_tree_active_formatting_first(lxb_html_tree_t *tree)
68
+ {
69
+ return (lxb_dom_node_t *) lexbor_array_get(tree->active_formatting, 0);
70
+ }
71
+
72
+ lxb_inline lxb_dom_node_t *
73
+ lxb_html_tree_active_formatting_get(lxb_html_tree_t *tree, size_t idx)
74
+ {
75
+ return (lxb_dom_node_t *) lexbor_array_get(tree->active_formatting, idx);
76
+ }
77
+
78
+ lxb_inline lxb_status_t
79
+ lxb_html_tree_active_formatting_push(lxb_html_tree_t *tree,
80
+ lxb_dom_node_t *node)
81
+ {
82
+ return lexbor_array_push(tree->active_formatting, node);
83
+ }
84
+
85
+ lxb_inline lxb_dom_node_t *
86
+ lxb_html_tree_active_formatting_pop(lxb_html_tree_t *tree)
87
+ {
88
+ return (lxb_dom_node_t *) lexbor_array_pop(tree->active_formatting);
89
+ }
90
+
91
+ lxb_inline lxb_status_t
92
+ lxb_html_tree_active_formatting_push_marker(lxb_html_tree_t *tree)
93
+ {
94
+ return lexbor_array_push(tree->active_formatting,
95
+ lxb_html_tree_active_formatting_marker());
96
+ }
97
+
98
+ lxb_inline lxb_status_t
99
+ lxb_html_tree_active_formatting_insert(lxb_html_tree_t *tree,
100
+ lxb_dom_node_t *node, size_t idx)
101
+ {
102
+ return lexbor_array_insert(tree->active_formatting, idx, node);
103
+ }
104
+
105
+ lxb_inline void
106
+ lxb_html_tree_active_formatting_remove(lxb_html_tree_t *tree, size_t idx)
107
+ {
108
+ lexbor_array_delete(tree->active_formatting, idx, 1);
109
+ }
110
+
111
+
112
+ #ifdef __cplusplus
113
+ } /* extern "C" */
114
+ #endif
115
+
116
+ #endif /* LEXBOR_HTML_ACTIVE_FORMATTING_H */
117
+