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 +1,368 @@
1
+ #include "nokolexbor.h"
2
+
3
+ extern VALUE mNokolexbor;
4
+ extern VALUE cNokolexborNode;
5
+ VALUE cNokolexborNodeSet;
6
+ extern rb_data_type_t nl_document_type;
7
+
8
+ VALUE nl_node_at_css(VALUE self, VALUE selector);
9
+ VALUE nl_node_css(VALUE self, VALUE selector);
10
+
11
+ typedef VALUE (*nl_node_find_f)(VALUE self, VALUE selector);
12
+
13
+ lxb_status_t
14
+ lexbor_array_push_unique(lexbor_array_t *array, void *value)
15
+ {
16
+ for (int i = 0; i < array->length; i++)
17
+ if (array->list[i] == value)
18
+ return LXB_STATUS_STOPPED;
19
+
20
+ return lexbor_array_push(array, value);
21
+ }
22
+
23
+ static void
24
+ free_nl_node_set(lexbor_array_t *array)
25
+ {
26
+ lexbor_array_destroy(array, true);
27
+ }
28
+
29
+ const rb_data_type_t nl_node_set_type = {
30
+ "Nokolexbor::NodeSet",
31
+ {
32
+ 0,
33
+ free_nl_node_set,
34
+ },
35
+ 0,
36
+ 0,
37
+ RUBY_TYPED_FREE_IMMEDIATELY,
38
+ };
39
+
40
+ lexbor_array_t *
41
+ nl_rb_node_set_unwrap(VALUE rb_node_set)
42
+ {
43
+ lexbor_array_t *array;
44
+ TypedData_Get_Struct(rb_node_set, lexbor_array_t, &nl_node_set_type, array);
45
+ return array;
46
+ }
47
+
48
+ static VALUE
49
+ nl_node_set_allocate(VALUE klass)
50
+ {
51
+ lexbor_array_t *array = lexbor_array_create();
52
+ return TypedData_Wrap_Struct(cNokolexborNodeSet, &nl_node_set_type, array);
53
+ }
54
+
55
+ VALUE
56
+ nl_rb_node_set_create_with_data(lexbor_array_t *array, VALUE rb_document)
57
+ {
58
+ if (array == NULL)
59
+ {
60
+ array = lexbor_array_create();
61
+ }
62
+ VALUE ret = TypedData_Wrap_Struct(cNokolexborNodeSet, &nl_node_set_type, array);
63
+ rb_iv_set(ret, "@document", rb_document);
64
+ return ret;
65
+ }
66
+
67
+ static VALUE
68
+ nl_node_set_length(VALUE self)
69
+ {
70
+ return INT2NUM(nl_rb_node_set_unwrap(self)->length);
71
+ }
72
+
73
+ static VALUE
74
+ nl_node_set_push(VALUE self, VALUE rb_node)
75
+ {
76
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
77
+ lxb_dom_node_t *node = nl_rb_node_unwrap(rb_node);
78
+
79
+ lxb_status_t status = lexbor_array_push_unique(array, node);
80
+ if (status != LXB_STATUS_OK && status != LXB_STATUS_STOPPED)
81
+ {
82
+ nl_raise_lexbor_error(status);
83
+ }
84
+
85
+ return self;
86
+ }
87
+
88
+ static VALUE
89
+ nl_node_set_delete(VALUE self, VALUE rb_node)
90
+ {
91
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
92
+ lxb_dom_node_t *node = nl_rb_node_unwrap(rb_node);
93
+
94
+ int i;
95
+ for (i = 0; i < array->length; i++)
96
+ if (array->list[i] == node)
97
+ {
98
+ break;
99
+ }
100
+
101
+ if (i >= array->length)
102
+ {
103
+ // not found
104
+ return Qnil;
105
+ }
106
+ lexbor_array_delete(array, i, 1);
107
+ return rb_node;
108
+ }
109
+
110
+ static VALUE
111
+ nl_node_set_is_include(VALUE self, VALUE rb_node)
112
+ {
113
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
114
+ lxb_dom_node_t *node = nl_rb_node_unwrap(rb_node);
115
+
116
+ for (int i = 0; i < array->length; i++)
117
+ if (array->list[i] == node)
118
+ {
119
+ return Qtrue;
120
+ }
121
+
122
+ return Qfalse;
123
+ }
124
+
125
+ static VALUE
126
+ nl_node_set_index_at(VALUE self, long offset)
127
+ {
128
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
129
+ if (offset >= (long)array->length || abs((int)offset) > (long)array->length)
130
+ {
131
+ return Qnil;
132
+ }
133
+
134
+ if (offset < 0)
135
+ {
136
+ offset += array->length;
137
+ }
138
+
139
+ lxb_dom_node_t *node = lexbor_array_get(array, offset);
140
+ return nl_rb_node_create(node, nl_rb_document_get(self));
141
+ }
142
+
143
+ static VALUE
144
+ nl_node_set_subseq(VALUE self, long beg, long len)
145
+ {
146
+ lexbor_array_t *old_array = nl_rb_node_set_unwrap(self);
147
+
148
+ if (beg > (long)old_array->length)
149
+ {
150
+ return Qnil;
151
+ }
152
+ if (beg < 0 || len < 0)
153
+ {
154
+ return Qnil;
155
+ }
156
+
157
+ if ((beg + len) > (long)old_array->length)
158
+ {
159
+ len = old_array->length - beg;
160
+ }
161
+
162
+ lexbor_array_t *new_array = lexbor_array_create();
163
+ if (len > 0)
164
+ {
165
+ lxb_status_t status = lexbor_array_init(new_array, len);
166
+ if (status != LXB_STATUS_OK)
167
+ {
168
+ nl_raise_lexbor_error(status);
169
+ }
170
+ }
171
+
172
+ for (int j = beg; j < beg + len; ++j)
173
+ {
174
+ lxb_status_t status = lexbor_array_push(new_array, old_array->list[j]);
175
+ if (status != LXB_STATUS_OK)
176
+ {
177
+ nl_raise_lexbor_error(status);
178
+ }
179
+ }
180
+ return TypedData_Wrap_Struct(cNokolexborNodeSet, &nl_node_set_type, new_array);
181
+ }
182
+
183
+ static VALUE
184
+ nl_node_set_slice(int argc, VALUE *argv, VALUE self)
185
+ {
186
+ VALUE arg;
187
+ long beg, len;
188
+
189
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
190
+
191
+ if (argc == 2)
192
+ {
193
+ beg = NUM2LONG(argv[0]);
194
+ len = NUM2LONG(argv[1]);
195
+ if (beg < 0)
196
+ {
197
+ beg += array->length;
198
+ }
199
+ return nl_node_set_subseq(self, beg, len);
200
+ }
201
+
202
+ if (argc != 1)
203
+ {
204
+ rb_scan_args(argc, argv, "11", NULL, NULL);
205
+ }
206
+ arg = argv[0];
207
+
208
+ if (FIXNUM_P(arg))
209
+ {
210
+ return nl_node_set_index_at(self, FIX2LONG(arg));
211
+ }
212
+
213
+ /* if arg is Range */
214
+ switch (rb_range_beg_len(arg, &beg, &len, array->length, 0))
215
+ {
216
+ case Qfalse:
217
+ break;
218
+ case Qnil:
219
+ return Qnil;
220
+ default:
221
+ return nl_node_set_subseq(self, beg, len);
222
+ }
223
+
224
+ return nl_node_set_index_at(self, NUM2LONG(arg));
225
+ }
226
+
227
+ static VALUE
228
+ nl_node_set_to_array(VALUE self)
229
+ {
230
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
231
+
232
+ VALUE list = rb_ary_new2(array->length);
233
+ VALUE doc = nl_rb_document_get(self);
234
+ for (int i = 0; i < array->length; i++)
235
+ {
236
+ lxb_dom_node_t *node = (lxb_dom_node_t *)array->list[i];
237
+ VALUE rb_node = nl_rb_node_create(node, doc);
238
+ rb_ary_push(list, rb_node);
239
+ }
240
+
241
+ return list;
242
+ }
243
+
244
+ static VALUE
245
+ nl_node_set_union(VALUE self, VALUE other)
246
+ {
247
+ if (!rb_obj_is_kind_of(other, cNokolexborNodeSet))
248
+ {
249
+ rb_raise(rb_eArgError, "Parameter must be a Nokolexbor::NodeSet");
250
+ }
251
+
252
+ lexbor_array_t *self_array = nl_rb_node_set_unwrap(self);
253
+ lexbor_array_t *other_array = nl_rb_node_set_unwrap(other);
254
+
255
+ if (self_array->length + other_array->length == 0)
256
+ {
257
+ return nl_rb_node_set_create_with_data(NULL, nl_rb_document_get(self));
258
+ }
259
+
260
+ lexbor_array_t *new_array = lexbor_array_create();
261
+ lxb_status_t status = lexbor_array_init(new_array, self_array->length + other_array->length);
262
+ if (status != LXB_STATUS_OK)
263
+ {
264
+ nl_raise_lexbor_error(status);
265
+ }
266
+
267
+ memcpy(new_array->list, self_array->list, sizeof(lxb_dom_node_t *) * self_array->length);
268
+ new_array->length = self_array->length;
269
+
270
+ for (int i = 0; i < other_array->length; i++)
271
+ {
272
+ lexbor_array_push_unique(new_array, other_array->list[i]);
273
+ }
274
+
275
+ return nl_rb_node_set_create_with_data(new_array, nl_rb_document_get(self));
276
+ }
277
+
278
+ static VALUE
279
+ nl_node_set_find(VALUE self, VALUE selector, nl_node_find_f finder)
280
+ {
281
+ VALUE rb_doc = nl_rb_document_get(self);
282
+ lxb_dom_document_t *doc;
283
+ TypedData_Get_Struct(rb_doc, lxb_dom_document_t, &nl_document_type, doc);
284
+ if (doc == NULL)
285
+ {
286
+ rb_raise(rb_eRuntimeError, "Error getting document");
287
+ }
288
+ lxb_dom_document_fragment_t *frag = lxb_dom_document_fragment_interface_create(doc);
289
+ if (frag == NULL)
290
+ {
291
+ rb_raise(rb_eRuntimeError, "Error creating document fragment");
292
+ }
293
+ lexbor_array_t *array = nl_rb_node_set_unwrap(self);
294
+
295
+ lexbor_array_t *backup_array = lexbor_array_create();
296
+ if (array->length > 0)
297
+ {
298
+ lxb_status_t status = lexbor_array_init(backup_array, array->length);
299
+ if (status != LXB_STATUS_OK)
300
+ {
301
+ nl_raise_lexbor_error(status);
302
+ }
303
+ }
304
+ // Backup original node data and re-group them into a fragment
305
+ for (int i = 0; i < array->length; i++)
306
+ {
307
+ lxb_dom_node_t *node = (lxb_dom_node_t *)array->list[i];
308
+ lxb_dom_node_t *backup_node = malloc(sizeof(lxb_dom_node_t));
309
+ if (backup_node == NULL)
310
+ {
311
+ nl_raise_lexbor_error(LXB_STATUS_ERROR_MEMORY_ALLOCATION);
312
+ }
313
+ memcpy(backup_node, node, sizeof(lxb_dom_node_t));
314
+ lxb_status_t status = lexbor_array_push(backup_array, backup_node);
315
+ if (status != LXB_STATUS_OK)
316
+ {
317
+ nl_raise_lexbor_error(LXB_STATUS_ERROR_MEMORY_ALLOCATION);
318
+ }
319
+ lxb_dom_node_insert_child(&frag->node, node);
320
+ }
321
+ VALUE rb_frag = nl_rb_node_create(&frag->node, nl_rb_document_get(self));
322
+
323
+ VALUE ret = finder(rb_frag, selector);
324
+
325
+ lxb_dom_document_fragment_interface_destroy(frag);
326
+ // Restore original node data
327
+ for (int i = 0; i < array->length; i++)
328
+ {
329
+ memcpy(array->list[i], backup_array->list[i], sizeof(lxb_dom_node_t));
330
+ free(backup_array->list[i]);
331
+ }
332
+ lexbor_array_destroy(backup_array, true);
333
+ return ret;
334
+ }
335
+
336
+ static VALUE
337
+ nl_node_set_at_css(VALUE self, VALUE selector)
338
+ {
339
+ return nl_node_set_find(self, selector, nl_node_at_css);
340
+ }
341
+
342
+ static VALUE
343
+ nl_node_set_css(VALUE self, VALUE selector)
344
+ {
345
+ return nl_node_set_find(self, selector, nl_node_css);
346
+ }
347
+
348
+ void Init_nl_node_set(void)
349
+ {
350
+ cNokolexborNodeSet = rb_define_class_under(mNokolexbor, "NodeSet", cNokolexborNode);
351
+
352
+ rb_define_alloc_func(cNokolexborNodeSet, nl_node_set_allocate);
353
+
354
+ rb_define_method(cNokolexborNodeSet, "length", nl_node_set_length, 0);
355
+ rb_define_method(cNokolexborNodeSet, "[]", nl_node_set_slice, -1);
356
+ rb_define_method(cNokolexborNodeSet, "slice", nl_node_set_slice, -1);
357
+ rb_define_method(cNokolexborNodeSet, "push", nl_node_set_push, 1);
358
+ rb_define_method(cNokolexborNodeSet, "|", nl_node_set_union, 1);
359
+ rb_define_method(cNokolexborNodeSet, "to_a", nl_node_set_to_array, 0);
360
+ rb_define_method(cNokolexborNodeSet, "delete", nl_node_set_delete, 1);
361
+ rb_define_method(cNokolexborNodeSet, "include?", nl_node_set_is_include, 1);
362
+ rb_define_method(cNokolexborNodeSet, "at_css", nl_node_set_at_css, 1);
363
+ rb_define_method(cNokolexborNodeSet, "css", nl_node_set_css, 1);
364
+
365
+ rb_define_alias(cNokolexborNodeSet, "<<", "push");
366
+ rb_define_alias(cNokolexborNodeSet, "size", "length");
367
+ rb_define_alias(cNokolexborNodeSet, "+", "|");
368
+ }
@@ -0,0 +1,200 @@
1
+ #include <ruby.h>
2
+ #include <ruby/util.h>
3
+ #include "nokolexbor.h"
4
+ #include "libxml.h"
5
+ #include "libxml/globals.h"
6
+ #include "libxml/xpath.h"
7
+ #include "libxml/xpathInternals.h"
8
+ #include "libxml/parserInternals.h"
9
+
10
+ extern VALUE mNokolexbor;
11
+ extern VALUE cNokolexborNodeSet;
12
+ VALUE cNokoLexborXpathContext;
13
+
14
+ static void
15
+ free_xml_xpath_context(xmlXPathContextPtr ctx)
16
+ {
17
+ xmlXPathFreeContext(ctx);
18
+ }
19
+
20
+ /*
21
+ * call-seq:
22
+ * register_ns(prefix, uri)
23
+ *
24
+ * Register the namespace with +prefix+ and +uri+.
25
+ */
26
+ static VALUE
27
+ rb_xml_xpath_context_register_ns(VALUE self, VALUE prefix, VALUE uri)
28
+ {
29
+ xmlXPathContextPtr ctx;
30
+ Data_Get_Struct(self, xmlXPathContext, ctx);
31
+
32
+ xmlXPathRegisterNs(ctx,
33
+ (const xmlChar *)StringValueCStr(prefix),
34
+ (const xmlChar *)StringValueCStr(uri));
35
+ return self;
36
+ }
37
+
38
+ /*
39
+ * call-seq:
40
+ * register_variable(name, value)
41
+ *
42
+ * Register the variable +name+ with +value+.
43
+ */
44
+ static VALUE
45
+ rb_xml_xpath_context_register_variable(VALUE self, VALUE name, VALUE value)
46
+ {
47
+ xmlXPathContextPtr ctx;
48
+ xmlXPathObjectPtr xmlValue;
49
+ Data_Get_Struct(self, xmlXPathContext, ctx);
50
+
51
+ xmlValue = xmlXPathNewCString(StringValueCStr(value));
52
+
53
+ xmlXPathRegisterVariable(ctx,
54
+ (const xmlChar *)StringValueCStr(name),
55
+ xmlValue);
56
+
57
+ return self;
58
+ }
59
+
60
+ /*
61
+ * convert an XPath object into a Ruby object of the appropriate type.
62
+ * returns Qundef if no conversion was possible.
63
+ */
64
+ static VALUE
65
+ xpath2ruby(xmlXPathObjectPtr c_xpath_object, xmlXPathContextPtr ctx, VALUE rb_document)
66
+ {
67
+ VALUE rb_retval;
68
+
69
+ switch (c_xpath_object->type)
70
+ {
71
+ case XPATH_STRING:
72
+ rb_retval = rb_utf8_str_new_cstr(c_xpath_object->stringval);
73
+ xmlFree(c_xpath_object->stringval);
74
+ return rb_retval;
75
+
76
+ case XPATH_NODESET:
77
+ {
78
+ if (c_xpath_object->nodesetval == NULL)
79
+ {
80
+ return nl_rb_node_set_create_with_data(NULL, rb_document);
81
+ }
82
+ if (c_xpath_object->nodesetval->nodeNr == 0)
83
+ {
84
+ return nl_rb_node_set_create_with_data(NULL, rb_document);
85
+ }
86
+
87
+ lexbor_array_t *array = lexbor_array_create();
88
+ lxb_status_t status = lexbor_array_init(array, c_xpath_object->nodesetval->nodeNr);
89
+ if (status != LXB_STATUS_OK)
90
+ {
91
+ nl_raise_lexbor_error(status);
92
+ }
93
+ memcpy(array->list, c_xpath_object->nodesetval->nodeTab, sizeof(lxb_dom_node_t *) * c_xpath_object->nodesetval->nodeNr);
94
+ array->length = c_xpath_object->nodesetval->nodeNr;
95
+ return nl_rb_node_set_create_with_data(array, rb_document);
96
+ }
97
+
98
+ case XPATH_NUMBER:
99
+ return rb_float_new(c_xpath_object->floatval);
100
+
101
+ case XPATH_BOOLEAN:
102
+ return (c_xpath_object->boolval == 1) ? Qtrue : Qfalse;
103
+
104
+ default:
105
+ return Qundef;
106
+ }
107
+ }
108
+
109
+ /*
110
+ * call-seq:
111
+ * evaluate(search_path, handler = nil)
112
+ *
113
+ * Evaluate the +search_path+ returning an XML::XPath object.
114
+ */
115
+ static VALUE
116
+ rb_xml_xpath_context_evaluate(int argc, VALUE *argv, VALUE self)
117
+ {
118
+ VALUE search_path, xpath_handler;
119
+ VALUE retval = Qnil;
120
+ xmlXPathContextPtr ctx;
121
+ xmlXPathObjectPtr xpath;
122
+ xmlChar *query;
123
+ VALUE errors = rb_ary_new();
124
+
125
+ Data_Get_Struct(self, xmlXPathContext, ctx);
126
+
127
+ if (rb_scan_args(argc, argv, "11", &search_path, &xpath_handler) == 1)
128
+ {
129
+ xpath_handler = Qnil;
130
+ }
131
+
132
+ query = (xmlChar *)StringValueCStr(search_path);
133
+
134
+ // if (Qnil != xpath_handler) {
135
+ // /* FIXME: not sure if this is the correct place to shove private data. */
136
+ // ctx->userData = (void *)xpath_handler;
137
+ // xmlXPathRegisterFuncLookup(ctx, handler_lookup, (void *)xpath_handler);
138
+ // }
139
+
140
+ // xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
141
+ // xmlSetGenericErrorFunc((void *)errors, generic_exception_pusher);
142
+
143
+ xpath = xmlXPathEvalExpression(query, ctx);
144
+
145
+ // xmlSetStructuredErrorFunc(NULL, NULL);
146
+ // xmlSetGenericErrorFunc(NULL, NULL);
147
+
148
+ if (xpath == NULL)
149
+ {
150
+ rb_exc_raise(rb_ary_entry(errors, 0));
151
+ }
152
+
153
+ retval = xpath2ruby(xpath, ctx, nl_rb_document_get(self));
154
+ if (retval == Qundef)
155
+ {
156
+ retval = rb_funcall(cNokolexborNodeSet, rb_intern("new"), 1, rb_ary_new());
157
+ }
158
+
159
+ // xmlXPathFreeNodeSetList(xpath);
160
+
161
+ return retval;
162
+ }
163
+
164
+ /*
165
+ * call-seq:
166
+ * new(node)
167
+ *
168
+ * Create a new XPathContext with +node+ as the reference point.
169
+ */
170
+ static VALUE
171
+ rb_xml_xpath_context_new(VALUE klass, VALUE rb_node)
172
+ {
173
+ xmlXPathContextPtr ctx;
174
+ VALUE self;
175
+
176
+ lxb_dom_node_t *node = nl_rb_node_unwrap(rb_node);
177
+
178
+ ctx = xmlXPathNewContext(node->owner_document);
179
+ ctx->node = node;
180
+
181
+ self = Data_Wrap_Struct(klass, 0, free_xml_xpath_context, ctx);
182
+ rb_iv_set(self, "@document", rb_iv_get(rb_node, "@document"));
183
+
184
+ return self;
185
+ }
186
+
187
+ void Init_nl_xpath_context(void)
188
+ {
189
+ xmlMemSetup((xmlFreeFunc)ruby_xfree, (xmlMallocFunc)ruby_xmalloc, (xmlReallocFunc)ruby_xrealloc, ruby_strdup);
190
+
191
+ cNokoLexborXpathContext = rb_define_class_under(mNokolexbor, "XPathContext", rb_cObject);
192
+
193
+ rb_undef_alloc_func(cNokoLexborXpathContext);
194
+
195
+ rb_define_singleton_method(cNokoLexborXpathContext, "new", rb_xml_xpath_context_new, 1);
196
+
197
+ rb_define_method(cNokoLexborXpathContext, "evaluate", rb_xml_xpath_context_evaluate, -1);
198
+ rb_define_method(cNokoLexborXpathContext, "register_variable", rb_xml_xpath_context_register_variable, 2);
199
+ rb_define_method(cNokoLexborXpathContext, "register_ns", rb_xml_xpath_context_register_ns, 2);
200
+ }
@@ -0,0 +1,63 @@
1
+ #include "nokolexbor.h"
2
+
3
+ VALUE mNokolexbor;
4
+ VALUE eLexborError;
5
+
6
+ void nl_raise_lexbor_error(lxb_status_t error)
7
+ {
8
+ switch (error)
9
+ {
10
+ case LXB_STATUS_ERROR:
11
+ rb_raise(eLexborError, "LXB_STATUS_ERROR");
12
+ case LXB_STATUS_ERROR_MEMORY_ALLOCATION:
13
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_MEMORY_ALLOCATION");
14
+ case LXB_STATUS_ERROR_OBJECT_IS_NULL:
15
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_OBJECT_IS_NULL");
16
+ case LXB_STATUS_ERROR_SMALL_BUFFER:
17
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_SMALL_BUFFER");
18
+ case LXB_STATUS_ERROR_INCOMPLETE_OBJECT:
19
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_INCOMPLETE_OBJECT");
20
+ case LXB_STATUS_ERROR_NO_FREE_SLOT:
21
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_NO_FREE_SLOT");
22
+ case LXB_STATUS_ERROR_TOO_SMALL_SIZE:
23
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_TOO_SMALL_SIZE");
24
+ case LXB_STATUS_ERROR_NOT_EXISTS:
25
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_NOT_EXISTS");
26
+ case LXB_STATUS_ERROR_WRONG_ARGS:
27
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_WRONG_ARGS");
28
+ case LXB_STATUS_ERROR_WRONG_STAGE:
29
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_WRONG_STAGE");
30
+ case LXB_STATUS_ERROR_UNEXPECTED_RESULT:
31
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_UNEXPECTED_RESULT");
32
+ case LXB_STATUS_ERROR_UNEXPECTED_DATA:
33
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_UNEXPECTED_DATA");
34
+ case LXB_STATUS_ERROR_OVERFLOW:
35
+ rb_raise(eLexborError, "LXB_STATUS_ERROR_OVERFLOW");
36
+ case LXB_STATUS_CONTINUE:
37
+ rb_raise(eLexborError, "LXB_STATUS_CONTINUE");
38
+ case LXB_STATUS_SMALL_BUFFER:
39
+ rb_raise(eLexborError, "LXB_STATUS_SMALL_BUFFER");
40
+ case LXB_STATUS_ABORTED:
41
+ rb_raise(eLexborError, "LXB_STATUS_ABORTED");
42
+ case LXB_STATUS_STOPPED:
43
+ rb_raise(eLexborError, "LXB_STATUS_STOPPED");
44
+ case LXB_STATUS_NEXT:
45
+ rb_raise(eLexborError, "LXB_STATUS_NEXT");
46
+ case LXB_STATUS_STOP:
47
+ rb_raise(eLexborError, "LXB_STATUS_STOP");
48
+ case LXB_STATUS_OK:
49
+ return;
50
+ default:
51
+ rb_raise(eLexborError, "LXB_ERROR_UKNOWN(%d)", error);
52
+ }
53
+ }
54
+
55
+ void Init_nokolexbor(void)
56
+ {
57
+ mNokolexbor = rb_define_module("Nokolexbor");
58
+ eLexborError = rb_define_class_under(mNokolexbor, "LexborError", rb_eStandardError);
59
+ Init_nl_node();
60
+ Init_nl_document();
61
+ Init_nl_node_set();
62
+ Init_nl_xpath_context();
63
+ }
@@ -0,0 +1,37 @@
1
+ #ifndef __NOKOLEXBOR_RUBY_H__
2
+ #define __NOKOLEXBOR_RUBY_H__
3
+
4
+ #include <ruby.h>
5
+
6
+ #include <lexbor/html/html.h>
7
+ #include <lexbor/css/css.h>
8
+ #include <lexbor/selectors/selectors.h>
9
+
10
+ extern VALUE cNokolexborDocument;
11
+
12
+ void Init_nl_document(void);
13
+ void Init_nl_node(void);
14
+ void Init_nl_node_set(void);
15
+ void Init_nl_xpath_context(void);
16
+
17
+ void nl_raise_lexbor_error(lxb_status_t error);
18
+ lxb_dom_node_t *nl_rb_node_unwrap(VALUE rb_node);
19
+ VALUE nl_rb_node_create(lxb_dom_node_t *node, VALUE rb_document);
20
+ VALUE nl_rb_node_set_create_with_data(lexbor_array_t *array, VALUE rb_document);
21
+
22
+ lxb_inline VALUE nl_rb_document_get(VALUE rb_node_or_doc)
23
+ {
24
+ if (rb_obj_class(rb_node_or_doc) == cNokolexborDocument)
25
+ {
26
+ return rb_node_or_doc;
27
+ }
28
+ return rb_iv_get(rb_node_or_doc, "@document");
29
+ }
30
+
31
+ const lxb_char_t *
32
+ lxb_dom_node_name_qualified(lxb_dom_node_t *node, size_t *len);
33
+
34
+ lxb_status_t
35
+ lexbor_array_push_unique(lexbor_array_t *array, void *value);
36
+
37
+ #endif