nokolexbor 0.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
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 +1,641 @@
1
+ /*
2
+ * Summary: internals routines and limits exported by the parser.
3
+ * Description: this module exports a number of internal parsing routines
4
+ * they are not really all intended for applications but
5
+ * can prove useful doing low level processing.
6
+ *
7
+ * Copy: See Copyright for the status of this software.
8
+ *
9
+ * Author: Daniel Veillard
10
+ */
11
+
12
+ #ifndef __XML_PARSER_INTERNALS_H__
13
+ #define __XML_PARSER_INTERNALS_H__
14
+
15
+ #include "xmlversion.h"
16
+ #include "parser.h"
17
+ #include "HTMLparser.h"
18
+ #include "chvalid.h"
19
+
20
+ #ifdef __cplusplus
21
+ extern "C" {
22
+ #endif
23
+
24
+ /**
25
+ * xmlParserMaxDepth:
26
+ *
27
+ * arbitrary depth limit for the XML documents that we allow to
28
+ * process. This is not a limitation of the parser but a safety
29
+ * boundary feature, use XML_PARSE_HUGE option to override it.
30
+ */
31
+ XMLPUBVAR unsigned int xmlParserMaxDepth;
32
+
33
+ /**
34
+ * XML_MAX_TEXT_LENGTH:
35
+ *
36
+ * Maximum size allowed for a single text node when building a tree.
37
+ * This is not a limitation of the parser but a safety boundary feature,
38
+ * use XML_PARSE_HUGE option to override it.
39
+ * Introduced in 2.9.0
40
+ */
41
+ #define XML_MAX_TEXT_LENGTH 10000000
42
+
43
+ /**
44
+ * XML_MAX_NAME_LENGTH:
45
+ *
46
+ * Maximum size allowed for a markup identifier.
47
+ * This is not a limitation of the parser but a safety boundary feature,
48
+ * use XML_PARSE_HUGE option to override it.
49
+ * Note that with the use of parsing dictionaries overriding the limit
50
+ * may result in more runtime memory usage in face of "unfriendly' content
51
+ * Introduced in 2.9.0
52
+ */
53
+ #define XML_MAX_NAME_LENGTH 50000
54
+
55
+ /**
56
+ * XML_MAX_DICTIONARY_LIMIT:
57
+ *
58
+ * Maximum size allowed by the parser for a dictionary by default
59
+ * This is not a limitation of the parser but a safety boundary feature,
60
+ * use XML_PARSE_HUGE option to override it.
61
+ * Introduced in 2.9.0
62
+ */
63
+ #define XML_MAX_DICTIONARY_LIMIT 10000000
64
+
65
+ /**
66
+ * XML_MAX_LOOKUP_LIMIT:
67
+ *
68
+ * Maximum size allowed by the parser for ahead lookup
69
+ * This is an upper boundary enforced by the parser to avoid bad
70
+ * behaviour on "unfriendly' content
71
+ * Introduced in 2.9.0
72
+ */
73
+ #define XML_MAX_LOOKUP_LIMIT 10000000
74
+
75
+ /**
76
+ * XML_MAX_NAMELEN:
77
+ *
78
+ * Identifiers can be longer, but this will be more costly
79
+ * at runtime.
80
+ */
81
+ #define XML_MAX_NAMELEN 100
82
+
83
+ /**
84
+ * INPUT_CHUNK:
85
+ *
86
+ * The parser tries to always have that amount of input ready.
87
+ * One of the point is providing context when reporting errors.
88
+ */
89
+ #define INPUT_CHUNK 250
90
+
91
+ /************************************************************************
92
+ * *
93
+ * UNICODE version of the macros. *
94
+ * *
95
+ ************************************************************************/
96
+ /**
97
+ * IS_BYTE_CHAR:
98
+ * @c: an byte value (int)
99
+ *
100
+ * Macro to check the following production in the XML spec:
101
+ *
102
+ * [2] Char ::= #x9 | #xA | #xD | [#x20...]
103
+ * any byte character in the accepted range
104
+ */
105
+ #define IS_BYTE_CHAR(c) xmlIsChar_ch(c)
106
+
107
+ /**
108
+ * IS_CHAR:
109
+ * @c: an UNICODE value (int)
110
+ *
111
+ * Macro to check the following production in the XML spec:
112
+ *
113
+ * [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
114
+ * | [#x10000-#x10FFFF]
115
+ * any Unicode character, excluding the surrogate blocks, FFFE, and FFFF.
116
+ */
117
+ #define IS_CHAR(c) xmlIsCharQ(c)
118
+
119
+ /**
120
+ * IS_CHAR_CH:
121
+ * @c: an xmlChar (usually an unsigned char)
122
+ *
123
+ * Behaves like IS_CHAR on single-byte value
124
+ */
125
+ #define IS_CHAR_CH(c) xmlIsChar_ch(c)
126
+
127
+ /**
128
+ * IS_BLANK:
129
+ * @c: an UNICODE value (int)
130
+ *
131
+ * Macro to check the following production in the XML spec:
132
+ *
133
+ * [3] S ::= (#x20 | #x9 | #xD | #xA)+
134
+ */
135
+ #define IS_BLANK(c) xmlIsBlankQ(c)
136
+
137
+ /**
138
+ * IS_BLANK_CH:
139
+ * @c: an xmlChar value (normally unsigned char)
140
+ *
141
+ * Behaviour same as IS_BLANK
142
+ */
143
+ #define IS_BLANK_CH(c) xmlIsBlank_ch(c)
144
+
145
+ /**
146
+ * IS_BASECHAR:
147
+ * @c: an UNICODE value (int)
148
+ *
149
+ * Macro to check the following production in the XML spec:
150
+ *
151
+ * [85] BaseChar ::= ... long list see REC ...
152
+ */
153
+ #define IS_BASECHAR(c) xmlIsBaseCharQ(c)
154
+
155
+ /**
156
+ * IS_DIGIT:
157
+ * @c: an UNICODE value (int)
158
+ *
159
+ * Macro to check the following production in the XML spec:
160
+ *
161
+ * [88] Digit ::= ... long list see REC ...
162
+ */
163
+ #define IS_DIGIT(c) xmlIsDigitQ(c)
164
+
165
+ /**
166
+ * IS_DIGIT_CH:
167
+ * @c: an xmlChar value (usually an unsigned char)
168
+ *
169
+ * Behaves like IS_DIGIT but with a single byte argument
170
+ */
171
+ #define IS_DIGIT_CH(c) xmlIsDigit_ch(c)
172
+
173
+ /**
174
+ * IS_COMBINING:
175
+ * @c: an UNICODE value (int)
176
+ *
177
+ * Macro to check the following production in the XML spec:
178
+ *
179
+ * [87] CombiningChar ::= ... long list see REC ...
180
+ */
181
+ #define IS_COMBINING(c) xmlIsCombiningQ(c)
182
+
183
+ /**
184
+ * IS_COMBINING_CH:
185
+ * @c: an xmlChar (usually an unsigned char)
186
+ *
187
+ * Always false (all combining chars > 0xff)
188
+ */
189
+ #define IS_COMBINING_CH(c) 0
190
+
191
+ /**
192
+ * IS_EXTENDER:
193
+ * @c: an UNICODE value (int)
194
+ *
195
+ * Macro to check the following production in the XML spec:
196
+ *
197
+ *
198
+ * [89] Extender ::= #x00B7 | #x02D0 | #x02D1 | #x0387 | #x0640 |
199
+ * #x0E46 | #x0EC6 | #x3005 | [#x3031-#x3035] |
200
+ * [#x309D-#x309E] | [#x30FC-#x30FE]
201
+ */
202
+ #define IS_EXTENDER(c) xmlIsExtenderQ(c)
203
+
204
+ /**
205
+ * IS_EXTENDER_CH:
206
+ * @c: an xmlChar value (usually an unsigned char)
207
+ *
208
+ * Behaves like IS_EXTENDER but with a single-byte argument
209
+ */
210
+ #define IS_EXTENDER_CH(c) xmlIsExtender_ch(c)
211
+
212
+ /**
213
+ * IS_IDEOGRAPHIC:
214
+ * @c: an UNICODE value (int)
215
+ *
216
+ * Macro to check the following production in the XML spec:
217
+ *
218
+ *
219
+ * [86] Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
220
+ */
221
+ #define IS_IDEOGRAPHIC(c) xmlIsIdeographicQ(c)
222
+
223
+ /**
224
+ * IS_LETTER:
225
+ * @c: an UNICODE value (int)
226
+ *
227
+ * Macro to check the following production in the XML spec:
228
+ *
229
+ *
230
+ * [84] Letter ::= BaseChar | Ideographic
231
+ */
232
+ #define IS_LETTER(c) (IS_BASECHAR(c) || IS_IDEOGRAPHIC(c))
233
+
234
+ /**
235
+ * IS_LETTER_CH:
236
+ * @c: an xmlChar value (normally unsigned char)
237
+ *
238
+ * Macro behaves like IS_LETTER, but only check base chars
239
+ *
240
+ */
241
+ #define IS_LETTER_CH(c) xmlIsBaseChar_ch(c)
242
+
243
+ /**
244
+ * IS_ASCII_LETTER:
245
+ * @c: an xmlChar value
246
+ *
247
+ * Macro to check [a-zA-Z]
248
+ *
249
+ */
250
+ #define IS_ASCII_LETTER(c) (((0x41 <= (c)) && ((c) <= 0x5a)) || \
251
+ ((0x61 <= (c)) && ((c) <= 0x7a)))
252
+
253
+ /**
254
+ * IS_ASCII_DIGIT:
255
+ * @c: an xmlChar value
256
+ *
257
+ * Macro to check [0-9]
258
+ *
259
+ */
260
+ #define IS_ASCII_DIGIT(c) ((0x30 <= (c)) && ((c) <= 0x39))
261
+
262
+ /**
263
+ * IS_PUBIDCHAR:
264
+ * @c: an UNICODE value (int)
265
+ *
266
+ * Macro to check the following production in the XML spec:
267
+ *
268
+ *
269
+ * [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%]
270
+ */
271
+ #define IS_PUBIDCHAR(c) xmlIsPubidCharQ(c)
272
+
273
+ /**
274
+ * IS_PUBIDCHAR_CH:
275
+ * @c: an xmlChar value (normally unsigned char)
276
+ *
277
+ * Same as IS_PUBIDCHAR but for single-byte value
278
+ */
279
+ #define IS_PUBIDCHAR_CH(c) xmlIsPubidChar_ch(c)
280
+
281
+ /**
282
+ * Global variables used for predefined strings.
283
+ */
284
+ XMLPUBVAR const xmlChar xmlStringText[];
285
+ XMLPUBVAR const xmlChar xmlStringTextNoenc[];
286
+ XMLPUBVAR const xmlChar xmlStringComment[];
287
+
288
+ /*
289
+ * Function to finish the work of the macros where needed.
290
+ */
291
+ XMLPUBFUN int XMLCALL xmlIsLetter (int c);
292
+
293
+ /**
294
+ * Parser context.
295
+ */
296
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
297
+ xmlCreateFileParserCtxt (const char *filename);
298
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
299
+ xmlCreateURLParserCtxt (const char *filename,
300
+ int options);
301
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
302
+ xmlCreateMemoryParserCtxt(const char *buffer,
303
+ int size);
304
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
305
+ xmlCreateEntityParserCtxt(const xmlChar *URL,
306
+ const xmlChar *ID,
307
+ const xmlChar *base);
308
+ XMLPUBFUN int XMLCALL
309
+ xmlSwitchEncoding (xmlParserCtxtPtr ctxt,
310
+ xmlCharEncoding enc);
311
+ XMLPUBFUN int XMLCALL
312
+ xmlSwitchToEncoding (xmlParserCtxtPtr ctxt,
313
+ xmlCharEncodingHandlerPtr handler);
314
+ XML_DEPRECATED
315
+ XMLPUBFUN int XMLCALL
316
+ xmlSwitchInputEncoding (xmlParserCtxtPtr ctxt,
317
+ xmlParserInputPtr input,
318
+ xmlCharEncodingHandlerPtr handler);
319
+
320
+ /**
321
+ * Input Streams.
322
+ */
323
+ XMLPUBFUN xmlParserInputPtr XMLCALL
324
+ xmlNewStringInputStream (xmlParserCtxtPtr ctxt,
325
+ const xmlChar *buffer);
326
+ XMLPUBFUN xmlParserInputPtr XMLCALL
327
+ xmlNewEntityInputStream (xmlParserCtxtPtr ctxt,
328
+ xmlEntityPtr entity);
329
+ XMLPUBFUN int XMLCALL
330
+ xmlPushInput (xmlParserCtxtPtr ctxt,
331
+ xmlParserInputPtr input);
332
+ XMLPUBFUN xmlChar XMLCALL
333
+ xmlPopInput (xmlParserCtxtPtr ctxt);
334
+ XMLPUBFUN void XMLCALL
335
+ xmlFreeInputStream (xmlParserInputPtr input);
336
+ XMLPUBFUN xmlParserInputPtr XMLCALL
337
+ xmlNewInputFromFile (xmlParserCtxtPtr ctxt,
338
+ const char *filename);
339
+ XMLPUBFUN xmlParserInputPtr XMLCALL
340
+ xmlNewInputStream (xmlParserCtxtPtr ctxt);
341
+
342
+ /**
343
+ * Namespaces.
344
+ */
345
+ XMLPUBFUN xmlChar * XMLCALL
346
+ xmlSplitQName (xmlParserCtxtPtr ctxt,
347
+ const xmlChar *name,
348
+ xmlChar **prefix);
349
+
350
+ /**
351
+ * Generic production rules.
352
+ */
353
+ XML_DEPRECATED
354
+ XMLPUBFUN const xmlChar * XMLCALL
355
+ xmlParseName (xmlParserCtxtPtr ctxt);
356
+ XML_DEPRECATED
357
+ XMLPUBFUN xmlChar * XMLCALL
358
+ xmlParseNmtoken (xmlParserCtxtPtr ctxt);
359
+ XML_DEPRECATED
360
+ XMLPUBFUN xmlChar * XMLCALL
361
+ xmlParseEntityValue (xmlParserCtxtPtr ctxt,
362
+ xmlChar **orig);
363
+ XML_DEPRECATED
364
+ XMLPUBFUN xmlChar * XMLCALL
365
+ xmlParseAttValue (xmlParserCtxtPtr ctxt);
366
+ XML_DEPRECATED
367
+ XMLPUBFUN xmlChar * XMLCALL
368
+ xmlParseSystemLiteral (xmlParserCtxtPtr ctxt);
369
+ XML_DEPRECATED
370
+ XMLPUBFUN xmlChar * XMLCALL
371
+ xmlParsePubidLiteral (xmlParserCtxtPtr ctxt);
372
+ XML_DEPRECATED
373
+ XMLPUBFUN void XMLCALL
374
+ xmlParseCharData (xmlParserCtxtPtr ctxt,
375
+ int cdata);
376
+ XML_DEPRECATED
377
+ XMLPUBFUN xmlChar * XMLCALL
378
+ xmlParseExternalID (xmlParserCtxtPtr ctxt,
379
+ xmlChar **publicID,
380
+ int strict);
381
+ XML_DEPRECATED
382
+ XMLPUBFUN void XMLCALL
383
+ xmlParseComment (xmlParserCtxtPtr ctxt);
384
+ XML_DEPRECATED
385
+ XMLPUBFUN const xmlChar * XMLCALL
386
+ xmlParsePITarget (xmlParserCtxtPtr ctxt);
387
+ XML_DEPRECATED
388
+ XMLPUBFUN void XMLCALL
389
+ xmlParsePI (xmlParserCtxtPtr ctxt);
390
+ XML_DEPRECATED
391
+ XMLPUBFUN void XMLCALL
392
+ xmlParseNotationDecl (xmlParserCtxtPtr ctxt);
393
+ XML_DEPRECATED
394
+ XMLPUBFUN void XMLCALL
395
+ xmlParseEntityDecl (xmlParserCtxtPtr ctxt);
396
+ XML_DEPRECATED
397
+ XMLPUBFUN int XMLCALL
398
+ xmlParseDefaultDecl (xmlParserCtxtPtr ctxt,
399
+ xmlChar **value);
400
+ XML_DEPRECATED
401
+ XMLPUBFUN xmlEnumerationPtr XMLCALL
402
+ xmlParseNotationType (xmlParserCtxtPtr ctxt);
403
+ XML_DEPRECATED
404
+ XMLPUBFUN xmlEnumerationPtr XMLCALL
405
+ xmlParseEnumerationType (xmlParserCtxtPtr ctxt);
406
+ XML_DEPRECATED
407
+ XMLPUBFUN int XMLCALL
408
+ xmlParseEnumeratedType (xmlParserCtxtPtr ctxt,
409
+ xmlEnumerationPtr *tree);
410
+ XML_DEPRECATED
411
+ XMLPUBFUN int XMLCALL
412
+ xmlParseAttributeType (xmlParserCtxtPtr ctxt,
413
+ xmlEnumerationPtr *tree);
414
+ XML_DEPRECATED
415
+ XMLPUBFUN void XMLCALL
416
+ xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt);
417
+ XML_DEPRECATED
418
+ XMLPUBFUN xmlElementContentPtr XMLCALL
419
+ xmlParseElementMixedContentDecl
420
+ (xmlParserCtxtPtr ctxt,
421
+ int inputchk);
422
+ XML_DEPRECATED
423
+ XMLPUBFUN xmlElementContentPtr XMLCALL
424
+ xmlParseElementChildrenContentDecl
425
+ (xmlParserCtxtPtr ctxt,
426
+ int inputchk);
427
+ XML_DEPRECATED
428
+ XMLPUBFUN int XMLCALL
429
+ xmlParseElementContentDecl(xmlParserCtxtPtr ctxt,
430
+ const xmlChar *name,
431
+ xmlElementContentPtr *result);
432
+ XML_DEPRECATED
433
+ XMLPUBFUN int XMLCALL
434
+ xmlParseElementDecl (xmlParserCtxtPtr ctxt);
435
+ XML_DEPRECATED
436
+ XMLPUBFUN void XMLCALL
437
+ xmlParseMarkupDecl (xmlParserCtxtPtr ctxt);
438
+ XML_DEPRECATED
439
+ XMLPUBFUN int XMLCALL
440
+ xmlParseCharRef (xmlParserCtxtPtr ctxt);
441
+ XML_DEPRECATED
442
+ XMLPUBFUN xmlEntityPtr XMLCALL
443
+ xmlParseEntityRef (xmlParserCtxtPtr ctxt);
444
+ XML_DEPRECATED
445
+ XMLPUBFUN void XMLCALL
446
+ xmlParseReference (xmlParserCtxtPtr ctxt);
447
+ XML_DEPRECATED
448
+ XMLPUBFUN void XMLCALL
449
+ xmlParsePEReference (xmlParserCtxtPtr ctxt);
450
+ XML_DEPRECATED
451
+ XMLPUBFUN void XMLCALL
452
+ xmlParseDocTypeDecl (xmlParserCtxtPtr ctxt);
453
+ #ifdef LIBXML_SAX1_ENABLED
454
+ XML_DEPRECATED
455
+ XMLPUBFUN const xmlChar * XMLCALL
456
+ xmlParseAttribute (xmlParserCtxtPtr ctxt,
457
+ xmlChar **value);
458
+ XML_DEPRECATED
459
+ XMLPUBFUN const xmlChar * XMLCALL
460
+ xmlParseStartTag (xmlParserCtxtPtr ctxt);
461
+ XML_DEPRECATED
462
+ XMLPUBFUN void XMLCALL
463
+ xmlParseEndTag (xmlParserCtxtPtr ctxt);
464
+ #endif /* LIBXML_SAX1_ENABLED */
465
+ XML_DEPRECATED
466
+ XMLPUBFUN void XMLCALL
467
+ xmlParseCDSect (xmlParserCtxtPtr ctxt);
468
+ XMLPUBFUN void XMLCALL
469
+ xmlParseContent (xmlParserCtxtPtr ctxt);
470
+ XML_DEPRECATED
471
+ XMLPUBFUN void XMLCALL
472
+ xmlParseElement (xmlParserCtxtPtr ctxt);
473
+ XML_DEPRECATED
474
+ XMLPUBFUN xmlChar * XMLCALL
475
+ xmlParseVersionNum (xmlParserCtxtPtr ctxt);
476
+ XML_DEPRECATED
477
+ XMLPUBFUN xmlChar * XMLCALL
478
+ xmlParseVersionInfo (xmlParserCtxtPtr ctxt);
479
+ XML_DEPRECATED
480
+ XMLPUBFUN xmlChar * XMLCALL
481
+ xmlParseEncName (xmlParserCtxtPtr ctxt);
482
+ XML_DEPRECATED
483
+ XMLPUBFUN const xmlChar * XMLCALL
484
+ xmlParseEncodingDecl (xmlParserCtxtPtr ctxt);
485
+ XML_DEPRECATED
486
+ XMLPUBFUN int XMLCALL
487
+ xmlParseSDDecl (xmlParserCtxtPtr ctxt);
488
+ XML_DEPRECATED
489
+ XMLPUBFUN void XMLCALL
490
+ xmlParseXMLDecl (xmlParserCtxtPtr ctxt);
491
+ XML_DEPRECATED
492
+ XMLPUBFUN void XMLCALL
493
+ xmlParseTextDecl (xmlParserCtxtPtr ctxt);
494
+ XML_DEPRECATED
495
+ XMLPUBFUN void XMLCALL
496
+ xmlParseMisc (xmlParserCtxtPtr ctxt);
497
+ XMLPUBFUN void XMLCALL
498
+ xmlParseExternalSubset (xmlParserCtxtPtr ctxt,
499
+ const xmlChar *ExternalID,
500
+ const xmlChar *SystemID);
501
+ /**
502
+ * XML_SUBSTITUTE_NONE:
503
+ *
504
+ * If no entities need to be substituted.
505
+ */
506
+ #define XML_SUBSTITUTE_NONE 0
507
+ /**
508
+ * XML_SUBSTITUTE_REF:
509
+ *
510
+ * Whether general entities need to be substituted.
511
+ */
512
+ #define XML_SUBSTITUTE_REF 1
513
+ /**
514
+ * XML_SUBSTITUTE_PEREF:
515
+ *
516
+ * Whether parameter entities need to be substituted.
517
+ */
518
+ #define XML_SUBSTITUTE_PEREF 2
519
+ /**
520
+ * XML_SUBSTITUTE_BOTH:
521
+ *
522
+ * Both general and parameter entities need to be substituted.
523
+ */
524
+ #define XML_SUBSTITUTE_BOTH 3
525
+
526
+ XMLPUBFUN xmlChar * XMLCALL
527
+ xmlStringDecodeEntities (xmlParserCtxtPtr ctxt,
528
+ const xmlChar *str,
529
+ int what,
530
+ xmlChar end,
531
+ xmlChar end2,
532
+ xmlChar end3);
533
+ XMLPUBFUN xmlChar * XMLCALL
534
+ xmlStringLenDecodeEntities (xmlParserCtxtPtr ctxt,
535
+ const xmlChar *str,
536
+ int len,
537
+ int what,
538
+ xmlChar end,
539
+ xmlChar end2,
540
+ xmlChar end3);
541
+
542
+ /*
543
+ * Generated by MACROS on top of parser.c c.f. PUSH_AND_POP.
544
+ */
545
+ XMLPUBFUN int XMLCALL nodePush (xmlParserCtxtPtr ctxt,
546
+ lxb_dom_node_t_ptr value);
547
+ XMLPUBFUN lxb_dom_node_t_ptr XMLCALL nodePop (xmlParserCtxtPtr ctxt);
548
+ XMLPUBFUN int XMLCALL inputPush (xmlParserCtxtPtr ctxt,
549
+ xmlParserInputPtr value);
550
+ XMLPUBFUN xmlParserInputPtr XMLCALL inputPop (xmlParserCtxtPtr ctxt);
551
+ XMLPUBFUN const xmlChar * XMLCALL namePop (xmlParserCtxtPtr ctxt);
552
+ XMLPUBFUN int XMLCALL namePush (xmlParserCtxtPtr ctxt,
553
+ const xmlChar *value);
554
+
555
+ /*
556
+ * other commodities shared between parser.c and parserInternals.
557
+ */
558
+ XMLPUBFUN int XMLCALL xmlSkipBlankChars (xmlParserCtxtPtr ctxt);
559
+ XMLPUBFUN int XMLCALL xmlStringCurrentChar (xmlParserCtxtPtr ctxt,
560
+ const xmlChar *cur,
561
+ int *len);
562
+ XMLPUBFUN void XMLCALL xmlParserHandlePEReference(xmlParserCtxtPtr ctxt);
563
+ XMLPUBFUN int XMLCALL xmlCheckLanguageID (const xmlChar *lang);
564
+
565
+ /*
566
+ * Really core function shared with HTML parser.
567
+ */
568
+ XMLPUBFUN int XMLCALL xmlCurrentChar (xmlParserCtxtPtr ctxt,
569
+ int *len);
570
+ XMLPUBFUN int XMLCALL xmlCopyCharMultiByte (xmlChar *out,
571
+ int val);
572
+ XMLPUBFUN int XMLCALL xmlCopyChar (int len,
573
+ xmlChar *out,
574
+ int val);
575
+ XMLPUBFUN void XMLCALL xmlNextChar (xmlParserCtxtPtr ctxt);
576
+ XMLPUBFUN void XMLCALL xmlParserInputShrink (xmlParserInputPtr in);
577
+
578
+ /*
579
+ * Specific function to keep track of entities references
580
+ * and used by the XSLT debugger.
581
+ */
582
+ #ifdef LIBXML_LEGACY_ENABLED
583
+ /**
584
+ * xmlEntityReferenceFunc:
585
+ * @ent: the entity
586
+ * @firstNode: the fist node in the chunk
587
+ * @lastNode: the last nod in the chunk
588
+ *
589
+ * Callback function used when one needs to be able to track back the
590
+ * provenance of a chunk of nodes inherited from an entity replacement.
591
+ */
592
+ typedef void (*xmlEntityReferenceFunc) (xmlEntityPtr ent,
593
+ lxb_dom_node_t_ptr firstNode,
594
+ lxb_dom_node_t_ptr lastNode);
595
+
596
+ XML_DEPRECATED
597
+ XMLPUBFUN void XMLCALL xmlSetEntityReferenceFunc (xmlEntityReferenceFunc func);
598
+
599
+ XML_DEPRECATED
600
+ XMLPUBFUN xmlChar * XMLCALL
601
+ xmlParseQuotedString (xmlParserCtxtPtr ctxt);
602
+ XML_DEPRECATED
603
+ XMLPUBFUN void XMLCALL
604
+ xmlParseNamespace (xmlParserCtxtPtr ctxt);
605
+ XML_DEPRECATED
606
+ XMLPUBFUN xmlChar * XMLCALL
607
+ xmlNamespaceParseNSDef (xmlParserCtxtPtr ctxt);
608
+ XML_DEPRECATED
609
+ XMLPUBFUN xmlChar * XMLCALL
610
+ xmlScanName (xmlParserCtxtPtr ctxt);
611
+ XML_DEPRECATED
612
+ XMLPUBFUN xmlChar * XMLCALL
613
+ xmlNamespaceParseNCName (xmlParserCtxtPtr ctxt);
614
+ XML_DEPRECATED
615
+ XMLPUBFUN void XMLCALL xmlParserHandleReference(xmlParserCtxtPtr ctxt);
616
+ XML_DEPRECATED
617
+ XMLPUBFUN xmlChar * XMLCALL
618
+ xmlNamespaceParseQName (xmlParserCtxtPtr ctxt,
619
+ xmlChar **prefix);
620
+ /**
621
+ * Entities
622
+ */
623
+ XML_DEPRECATED
624
+ XMLPUBFUN xmlChar * XMLCALL
625
+ xmlDecodeEntities (xmlParserCtxtPtr ctxt,
626
+ int len,
627
+ int what,
628
+ xmlChar end,
629
+ xmlChar end2,
630
+ xmlChar end3);
631
+ XML_DEPRECATED
632
+ XMLPUBFUN void XMLCALL
633
+ xmlHandleEntity (xmlParserCtxtPtr ctxt,
634
+ xmlEntityPtr entity);
635
+
636
+ #endif /* LIBXML_LEGACY_ENABLED */
637
+
638
+ #ifdef __cplusplus
639
+ }
640
+ #endif
641
+ #endif /* __XML_PARSER_INTERNALS_H__ */