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 +1,1264 @@
1
+ /*
2
+ * Summary: the core parser module
3
+ * Description: Interfaces, constants and types related to the XML parser
4
+ *
5
+ * Copy: See Copyright for the status of this software.
6
+ *
7
+ * Author: Daniel Veillard
8
+ */
9
+
10
+ #ifndef __XML_PARSER_H__
11
+ #define __XML_PARSER_H__
12
+
13
+ #include "xmlversion.h"
14
+ #include "tree.h"
15
+ #include "dict.h"
16
+ #include "hash.h"
17
+ #include "valid.h"
18
+ #include "entities.h"
19
+ #include "xmlerror.h"
20
+ #include "xmlstring.h"
21
+
22
+ #ifdef __cplusplus
23
+ extern "C" {
24
+ #endif
25
+
26
+ /**
27
+ * XML_DEFAULT_VERSION:
28
+ *
29
+ * The default version of XML used: 1.0
30
+ */
31
+ #define XML_DEFAULT_VERSION "1.0"
32
+
33
+ /**
34
+ * xmlParserInput:
35
+ *
36
+ * An xmlParserInput is an input flow for the XML processor.
37
+ * Each entity parsed is associated an xmlParserInput (except the
38
+ * few predefined ones). This is the case both for internal entities
39
+ * - in which case the flow is already completely in memory - or
40
+ * external entities - in which case we use the buf structure for
41
+ * progressive reading and I18N conversions to the internal UTF-8 format.
42
+ */
43
+
44
+ /**
45
+ * xmlParserInputDeallocate:
46
+ * @str: the string to deallocate
47
+ *
48
+ * Callback for freeing some parser input allocations.
49
+ */
50
+ typedef void (* xmlParserInputDeallocate)(xmlChar *str);
51
+
52
+ struct _xmlParserInput {
53
+ /* Input buffer */
54
+ xmlParserInputBufferPtr buf; /* UTF-8 encoded buffer */
55
+
56
+ const char *filename; /* The file analyzed, if any */
57
+ const char *directory; /* the directory/base of the file */
58
+ const xmlChar *base; /* Base of the array to parse */
59
+ const xmlChar *cur; /* Current char being parsed */
60
+ const xmlChar *end; /* end of the array to parse */
61
+ int length; /* length if known */
62
+ int line; /* Current line */
63
+ int col; /* Current column */
64
+ /*
65
+ * NOTE: consumed is only tested for equality in the parser code,
66
+ * so even if there is an overflow this should not give troubles
67
+ * for parsing very large instances.
68
+ */
69
+ unsigned long consumed; /* How many xmlChars already consumed */
70
+ xmlParserInputDeallocate free; /* function to deallocate the base */
71
+ const xmlChar *encoding; /* the encoding string for entity */
72
+ const xmlChar *version; /* the version string for entity */
73
+ int standalone; /* Was that entity marked standalone */
74
+ int id; /* an unique identifier for the entity */
75
+ };
76
+
77
+ /**
78
+ * xmlParserNodeInfo:
79
+ *
80
+ * The parser can be asked to collect Node information, i.e. at what
81
+ * place in the file they were detected.
82
+ * NOTE: This is off by default and not very well tested.
83
+ */
84
+ typedef struct _xmlParserNodeInfo xmlParserNodeInfo;
85
+ typedef xmlParserNodeInfo *xmlParserNodeInfoPtr;
86
+
87
+ struct _xmlParserNodeInfo {
88
+ const struct _xmlNode* node;
89
+ /* Position & line # that text that created the node begins & ends on */
90
+ unsigned long begin_pos;
91
+ unsigned long begin_line;
92
+ unsigned long end_pos;
93
+ unsigned long end_line;
94
+ };
95
+
96
+ typedef struct _xmlParserNodeInfoSeq xmlParserNodeInfoSeq;
97
+ typedef xmlParserNodeInfoSeq *xmlParserNodeInfoSeqPtr;
98
+ struct _xmlParserNodeInfoSeq {
99
+ unsigned long maximum;
100
+ unsigned long length;
101
+ xmlParserNodeInfo* buffer;
102
+ };
103
+
104
+ /**
105
+ * xmlParserInputState:
106
+ *
107
+ * The parser is now working also as a state based parser.
108
+ * The recursive one use the state info for entities processing.
109
+ */
110
+ typedef enum {
111
+ XML_PARSER_EOF = -1, /* nothing is to be parsed */
112
+ XML_PARSER_START = 0, /* nothing has been parsed */
113
+ XML_PARSER_MISC, /* Misc* before int subset */
114
+ XML_PARSER_PI, /* Within a processing instruction */
115
+ XML_PARSER_DTD, /* within some DTD content */
116
+ XML_PARSER_PROLOG, /* Misc* after internal subset */
117
+ XML_PARSER_COMMENT, /* within a comment */
118
+ XML_PARSER_START_TAG, /* within a start tag */
119
+ XML_PARSER_CONTENT, /* within the content */
120
+ XML_PARSER_CDATA_SECTION, /* within a CDATA section */
121
+ XML_PARSER_END_TAG, /* within a closing tag */
122
+ XML_PARSER_ENTITY_DECL, /* within an entity declaration */
123
+ XML_PARSER_ENTITY_VALUE, /* within an entity value in a decl */
124
+ XML_PARSER_ATTRIBUTE_VALUE, /* within an attribute value */
125
+ XML_PARSER_SYSTEM_LITERAL, /* within a SYSTEM value */
126
+ XML_PARSER_EPILOG, /* the Misc* after the last end tag */
127
+ XML_PARSER_IGNORE, /* within an IGNORED section */
128
+ XML_PARSER_PUBLIC_LITERAL /* within a PUBLIC value */
129
+ } xmlParserInputState;
130
+
131
+ /**
132
+ * XML_DETECT_IDS:
133
+ *
134
+ * Bit in the loadsubset context field to tell to do ID/REFs lookups.
135
+ * Use it to initialize xmlLoadExtDtdDefaultValue.
136
+ */
137
+ #define XML_DETECT_IDS 2
138
+
139
+ /**
140
+ * XML_COMPLETE_ATTRS:
141
+ *
142
+ * Bit in the loadsubset context field to tell to do complete the
143
+ * elements attributes lists with the ones defaulted from the DTDs.
144
+ * Use it to initialize xmlLoadExtDtdDefaultValue.
145
+ */
146
+ #define XML_COMPLETE_ATTRS 4
147
+
148
+ /**
149
+ * XML_SKIP_IDS:
150
+ *
151
+ * Bit in the loadsubset context field to tell to not do ID/REFs registration.
152
+ * Used to initialize xmlLoadExtDtdDefaultValue in some special cases.
153
+ */
154
+ #define XML_SKIP_IDS 8
155
+
156
+ /**
157
+ * xmlParserMode:
158
+ *
159
+ * A parser can operate in various modes
160
+ */
161
+ typedef enum {
162
+ XML_PARSE_UNKNOWN = 0,
163
+ XML_PARSE_DOM = 1,
164
+ XML_PARSE_SAX = 2,
165
+ XML_PARSE_PUSH_DOM = 3,
166
+ XML_PARSE_PUSH_SAX = 4,
167
+ XML_PARSE_READER = 5
168
+ } xmlParserMode;
169
+
170
+ typedef struct _xmlStartTag xmlStartTag;
171
+
172
+ /**
173
+ * xmlParserCtxt:
174
+ *
175
+ * The parser context.
176
+ * NOTE This doesn't completely define the parser state, the (current ?)
177
+ * design of the parser uses recursive function calls since this allow
178
+ * and easy mapping from the production rules of the specification
179
+ * to the actual code. The drawback is that the actual function call
180
+ * also reflect the parser state. However most of the parsing routines
181
+ * takes as the only argument the parser context pointer, so migrating
182
+ * to a state based parser for progressive parsing shouldn't be too hard.
183
+ */
184
+ struct _xmlParserCtxt {
185
+ struct _xmlSAXHandler *sax; /* The SAX handler */
186
+ void *userData; /* For SAX interface only, used by DOM build */
187
+ lxb_dom_document_t_ptr myDoc; /* the document being built */
188
+ int wellFormed; /* is the document well formed */
189
+ int replaceEntities; /* shall we replace entities ? */
190
+ const xmlChar *version; /* the XML version string */
191
+ const xmlChar *encoding; /* the declared encoding, if any */
192
+ int standalone; /* standalone document */
193
+ int html; /* an HTML(1) document
194
+ * 3 is HTML after <head>
195
+ * 10 is HTML after <body>
196
+ */
197
+
198
+ /* Input stream stack */
199
+ xmlParserInputPtr input; /* Current input stream */
200
+ int inputNr; /* Number of current input streams */
201
+ int inputMax; /* Max number of input streams */
202
+ xmlParserInputPtr *inputTab; /* stack of inputs */
203
+
204
+ /* Node analysis stack only used for DOM building */
205
+ lxb_dom_node_t_ptr node; /* Current parsed Node */
206
+ int nodeNr; /* Depth of the parsing stack */
207
+ int nodeMax; /* Max depth of the parsing stack */
208
+ lxb_dom_node_t_ptr *nodeTab; /* array of nodes */
209
+
210
+ int record_info; /* Whether node info should be kept */
211
+ xmlParserNodeInfoSeq node_seq; /* info about each node parsed */
212
+
213
+ int errNo; /* error code */
214
+
215
+ int hasExternalSubset; /* reference and external subset */
216
+ int hasPErefs; /* the internal subset has PE refs */
217
+ int external; /* are we parsing an external entity */
218
+
219
+ int valid; /* is the document valid */
220
+ int validate; /* shall we try to validate ? */
221
+ xmlValidCtxt vctxt; /* The validity context */
222
+
223
+ xmlParserInputState instate; /* current type of input */
224
+ int token; /* next char look-ahead */
225
+
226
+ char *directory; /* the data directory */
227
+
228
+ /* Node name stack */
229
+ const xmlChar *name; /* Current parsed Node */
230
+ int nameNr; /* Depth of the parsing stack */
231
+ int nameMax; /* Max depth of the parsing stack */
232
+ const xmlChar * *nameTab; /* array of nodes */
233
+
234
+ long nbChars; /* unused */
235
+ long checkIndex; /* used by progressive parsing lookup */
236
+ int keepBlanks; /* ugly but ... */
237
+ int disableSAX; /* SAX callbacks are disabled */
238
+ int inSubset; /* Parsing is in int 1/ext 2 subset */
239
+ const xmlChar * intSubName; /* name of subset */
240
+ xmlChar * extSubURI; /* URI of external subset */
241
+ xmlChar * extSubSystem; /* SYSTEM ID of external subset */
242
+
243
+ /* xml:space values */
244
+ int * space; /* Should the parser preserve spaces */
245
+ int spaceNr; /* Depth of the parsing stack */
246
+ int spaceMax; /* Max depth of the parsing stack */
247
+ int * spaceTab; /* array of space infos */
248
+
249
+ int depth; /* to prevent entity substitution loops */
250
+ xmlParserInputPtr entity; /* used to check entities boundaries */
251
+ int charset; /* encoding of the in-memory content
252
+ actually an xmlCharEncoding */
253
+ int nodelen; /* Those two fields are there to */
254
+ int nodemem; /* Speed up large node parsing */
255
+ int pedantic; /* signal pedantic warnings */
256
+ void *_private; /* For user data, libxml won't touch it */
257
+
258
+ int loadsubset; /* should the external subset be loaded */
259
+ int linenumbers; /* set line number in element content */
260
+ void *catalogs; /* document's own catalog */
261
+ int recovery; /* run in recovery mode */
262
+ int progressive; /* is this a progressive parsing */
263
+ xmlDictPtr dict; /* dictionary for the parser */
264
+ const xmlChar * *atts; /* array for the attributes callbacks */
265
+ int maxatts; /* the size of the array */
266
+ int docdict; /* use strings from dict to build tree */
267
+
268
+ /*
269
+ * pre-interned strings
270
+ */
271
+ const xmlChar *str_xml;
272
+ const xmlChar *str_xmlns;
273
+ const xmlChar *str_xml_ns;
274
+
275
+ /*
276
+ * Everything below is used only by the new SAX mode
277
+ */
278
+ int sax2; /* operating in the new SAX mode */
279
+ int nsNr; /* the number of inherited namespaces */
280
+ int nsMax; /* the size of the arrays */
281
+ const xmlChar * *nsTab; /* the array of prefix/namespace name */
282
+ int *attallocs; /* which attribute were allocated */
283
+ xmlStartTag *pushTab; /* array of data for push */
284
+ xmlHashTablePtr attsDefault; /* defaulted attributes if any */
285
+ xmlHashTablePtr attsSpecial; /* non-CDATA attributes if any */
286
+ int nsWellFormed; /* is the document XML Namespace okay */
287
+ int options; /* Extra options */
288
+
289
+ /*
290
+ * Those fields are needed only for streaming parsing so far
291
+ */
292
+ int dictNames; /* Use dictionary names for the tree */
293
+ int freeElemsNr; /* number of freed element nodes */
294
+ lxb_dom_node_t_ptr freeElems; /* List of freed element nodes */
295
+ int freeAttrsNr; /* number of freed attributes nodes */
296
+ lxb_dom_attr_t_ptr freeAttrs; /* List of freed attributes nodes */
297
+
298
+ /*
299
+ * the complete error information for the last error.
300
+ */
301
+ xmlError lastError;
302
+ xmlParserMode parseMode; /* the parser mode */
303
+ unsigned long nbentities; /* number of entities references */
304
+ unsigned long sizeentities; /* size of parsed entities */
305
+
306
+ /* for use by HTML non-recursive parser */
307
+ xmlParserNodeInfo *nodeInfo; /* Current NodeInfo */
308
+ int nodeInfoNr; /* Depth of the parsing stack */
309
+ int nodeInfoMax; /* Max depth of the parsing stack */
310
+ xmlParserNodeInfo *nodeInfoTab; /* array of nodeInfos */
311
+
312
+ int input_id; /* we need to label inputs */
313
+ unsigned long sizeentcopy; /* volume of entity copy */
314
+
315
+ int endCheckState; /* quote state for push parser */
316
+ };
317
+
318
+ /**
319
+ * xmlSAXLocator:
320
+ *
321
+ * A SAX Locator.
322
+ */
323
+ struct _xmlSAXLocator {
324
+ const xmlChar *(*getPublicId)(void *ctx);
325
+ const xmlChar *(*getSystemId)(void *ctx);
326
+ int (*getLineNumber)(void *ctx);
327
+ int (*getColumnNumber)(void *ctx);
328
+ };
329
+
330
+ /**
331
+ * xmlSAXHandler:
332
+ *
333
+ * A SAX handler is bunch of callbacks called by the parser when processing
334
+ * of the input generate data or structure information.
335
+ */
336
+
337
+ /**
338
+ * resolveEntitySAXFunc:
339
+ * @ctx: the user data (XML parser context)
340
+ * @publicId: The public ID of the entity
341
+ * @systemId: The system ID of the entity
342
+ *
343
+ * Callback:
344
+ * The entity loader, to control the loading of external entities,
345
+ * the application can either:
346
+ * - override this resolveEntity() callback in the SAX block
347
+ * - or better use the xmlSetExternalEntityLoader() function to
348
+ * set up it's own entity resolution routine
349
+ *
350
+ * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour.
351
+ */
352
+ typedef xmlParserInputPtr (*resolveEntitySAXFunc) (void *ctx,
353
+ const xmlChar *publicId,
354
+ const xmlChar *systemId);
355
+ /**
356
+ * internalSubsetSAXFunc:
357
+ * @ctx: the user data (XML parser context)
358
+ * @name: the root element name
359
+ * @ExternalID: the external ID
360
+ * @SystemID: the SYSTEM ID (e.g. filename or URL)
361
+ *
362
+ * Callback on internal subset declaration.
363
+ */
364
+ typedef void (*internalSubsetSAXFunc) (void *ctx,
365
+ const xmlChar *name,
366
+ const xmlChar *ExternalID,
367
+ const xmlChar *SystemID);
368
+ /**
369
+ * externalSubsetSAXFunc:
370
+ * @ctx: the user data (XML parser context)
371
+ * @name: the root element name
372
+ * @ExternalID: the external ID
373
+ * @SystemID: the SYSTEM ID (e.g. filename or URL)
374
+ *
375
+ * Callback on external subset declaration.
376
+ */
377
+ typedef void (*externalSubsetSAXFunc) (void *ctx,
378
+ const xmlChar *name,
379
+ const xmlChar *ExternalID,
380
+ const xmlChar *SystemID);
381
+ /**
382
+ * getEntitySAXFunc:
383
+ * @ctx: the user data (XML parser context)
384
+ * @name: The entity name
385
+ *
386
+ * Get an entity by name.
387
+ *
388
+ * Returns the xmlEntityPtr if found.
389
+ */
390
+ typedef xmlEntityPtr (*getEntitySAXFunc) (void *ctx,
391
+ const xmlChar *name);
392
+ /**
393
+ * getParameterEntitySAXFunc:
394
+ * @ctx: the user data (XML parser context)
395
+ * @name: The entity name
396
+ *
397
+ * Get a parameter entity by name.
398
+ *
399
+ * Returns the xmlEntityPtr if found.
400
+ */
401
+ typedef xmlEntityPtr (*getParameterEntitySAXFunc) (void *ctx,
402
+ const xmlChar *name);
403
+ /**
404
+ * entityDeclSAXFunc:
405
+ * @ctx: the user data (XML parser context)
406
+ * @name: the entity name
407
+ * @type: the entity type
408
+ * @publicId: The public ID of the entity
409
+ * @systemId: The system ID of the entity
410
+ * @content: the entity value (without processing).
411
+ *
412
+ * An entity definition has been parsed.
413
+ */
414
+ typedef void (*entityDeclSAXFunc) (void *ctx,
415
+ const xmlChar *name,
416
+ int type,
417
+ const xmlChar *publicId,
418
+ const xmlChar *systemId,
419
+ xmlChar *content);
420
+ /**
421
+ * notationDeclSAXFunc:
422
+ * @ctx: the user data (XML parser context)
423
+ * @name: The name of the notation
424
+ * @publicId: The public ID of the entity
425
+ * @systemId: The system ID of the entity
426
+ *
427
+ * What to do when a notation declaration has been parsed.
428
+ */
429
+ typedef void (*notationDeclSAXFunc)(void *ctx,
430
+ const xmlChar *name,
431
+ const xmlChar *publicId,
432
+ const xmlChar *systemId);
433
+ /**
434
+ * attributeDeclSAXFunc:
435
+ * @ctx: the user data (XML parser context)
436
+ * @elem: the name of the element
437
+ * @fullname: the attribute name
438
+ * @type: the attribute type
439
+ * @def: the type of default value
440
+ * @defaultValue: the attribute default value
441
+ * @tree: the tree of enumerated value set
442
+ *
443
+ * An attribute definition has been parsed.
444
+ */
445
+ typedef void (*attributeDeclSAXFunc)(void *ctx,
446
+ const xmlChar *elem,
447
+ const xmlChar *fullname,
448
+ int type,
449
+ int def,
450
+ const xmlChar *defaultValue,
451
+ xmlEnumerationPtr tree);
452
+ /**
453
+ * elementDeclSAXFunc:
454
+ * @ctx: the user data (XML parser context)
455
+ * @name: the element name
456
+ * @type: the element type
457
+ * @content: the element value tree
458
+ *
459
+ * An element definition has been parsed.
460
+ */
461
+ typedef void (*elementDeclSAXFunc)(void *ctx,
462
+ const xmlChar *name,
463
+ int type,
464
+ xmlElementContentPtr content);
465
+ /**
466
+ * unparsedEntityDeclSAXFunc:
467
+ * @ctx: the user data (XML parser context)
468
+ * @name: The name of the entity
469
+ * @publicId: The public ID of the entity
470
+ * @systemId: The system ID of the entity
471
+ * @notationName: the name of the notation
472
+ *
473
+ * What to do when an unparsed entity declaration is parsed.
474
+ */
475
+ typedef void (*unparsedEntityDeclSAXFunc)(void *ctx,
476
+ const xmlChar *name,
477
+ const xmlChar *publicId,
478
+ const xmlChar *systemId,
479
+ const xmlChar *notationName);
480
+ /**
481
+ * setDocumentLocatorSAXFunc:
482
+ * @ctx: the user data (XML parser context)
483
+ * @loc: A SAX Locator
484
+ *
485
+ * Receive the document locator at startup, actually xmlDefaultSAXLocator.
486
+ * Everything is available on the context, so this is useless in our case.
487
+ */
488
+ typedef void (*setDocumentLocatorSAXFunc) (void *ctx,
489
+ xmlSAXLocatorPtr loc);
490
+ /**
491
+ * startDocumentSAXFunc:
492
+ * @ctx: the user data (XML parser context)
493
+ *
494
+ * Called when the document start being processed.
495
+ */
496
+ typedef void (*startDocumentSAXFunc) (void *ctx);
497
+ /**
498
+ * endDocumentSAXFunc:
499
+ * @ctx: the user data (XML parser context)
500
+ *
501
+ * Called when the document end has been detected.
502
+ */
503
+ typedef void (*endDocumentSAXFunc) (void *ctx);
504
+ /**
505
+ * startElementSAXFunc:
506
+ * @ctx: the user data (XML parser context)
507
+ * @name: The element name, including namespace prefix
508
+ * @atts: An array of name/value attributes pairs, NULL terminated
509
+ *
510
+ * Called when an opening tag has been processed.
511
+ */
512
+ typedef void (*startElementSAXFunc) (void *ctx,
513
+ const xmlChar *name,
514
+ const xmlChar **atts);
515
+ /**
516
+ * endElementSAXFunc:
517
+ * @ctx: the user data (XML parser context)
518
+ * @name: The element name
519
+ *
520
+ * Called when the end of an element has been detected.
521
+ */
522
+ typedef void (*endElementSAXFunc) (void *ctx,
523
+ const xmlChar *name);
524
+ /**
525
+ * attributeSAXFunc:
526
+ * @ctx: the user data (XML parser context)
527
+ * @name: The attribute name, including namespace prefix
528
+ * @value: The attribute value
529
+ *
530
+ * Handle an attribute that has been read by the parser.
531
+ * The default handling is to convert the attribute into an
532
+ * DOM subtree and past it in a new xmlAttr element added to
533
+ * the element.
534
+ */
535
+ typedef void (*attributeSAXFunc) (void *ctx,
536
+ const xmlChar *name,
537
+ const xmlChar *value);
538
+ /**
539
+ * referenceSAXFunc:
540
+ * @ctx: the user data (XML parser context)
541
+ * @name: The entity name
542
+ *
543
+ * Called when an entity reference is detected.
544
+ */
545
+ typedef void (*referenceSAXFunc) (void *ctx,
546
+ const xmlChar *name);
547
+ /**
548
+ * charactersSAXFunc:
549
+ * @ctx: the user data (XML parser context)
550
+ * @ch: a xmlChar string
551
+ * @len: the number of xmlChar
552
+ *
553
+ * Receiving some chars from the parser.
554
+ */
555
+ typedef void (*charactersSAXFunc) (void *ctx,
556
+ const xmlChar *ch,
557
+ int len);
558
+ /**
559
+ * ignorableWhitespaceSAXFunc:
560
+ * @ctx: the user data (XML parser context)
561
+ * @ch: a xmlChar string
562
+ * @len: the number of xmlChar
563
+ *
564
+ * Receiving some ignorable whitespaces from the parser.
565
+ * UNUSED: by default the DOM building will use characters.
566
+ */
567
+ typedef void (*ignorableWhitespaceSAXFunc) (void *ctx,
568
+ const xmlChar *ch,
569
+ int len);
570
+ /**
571
+ * processingInstructionSAXFunc:
572
+ * @ctx: the user data (XML parser context)
573
+ * @target: the target name
574
+ * @data: the PI data's
575
+ *
576
+ * A processing instruction has been parsed.
577
+ */
578
+ typedef void (*processingInstructionSAXFunc) (void *ctx,
579
+ const xmlChar *target,
580
+ const xmlChar *data);
581
+ /**
582
+ * commentSAXFunc:
583
+ * @ctx: the user data (XML parser context)
584
+ * @value: the comment content
585
+ *
586
+ * A comment has been parsed.
587
+ */
588
+ typedef void (*commentSAXFunc) (void *ctx,
589
+ const xmlChar *value);
590
+ /**
591
+ * cdataBlockSAXFunc:
592
+ * @ctx: the user data (XML parser context)
593
+ * @value: The pcdata content
594
+ * @len: the block length
595
+ *
596
+ * Called when a pcdata block has been parsed.
597
+ */
598
+ typedef void (*cdataBlockSAXFunc) (
599
+ void *ctx,
600
+ const xmlChar *value,
601
+ int len);
602
+ /**
603
+ * warningSAXFunc:
604
+ * @ctx: an XML parser context
605
+ * @msg: the message to display/transmit
606
+ * @...: extra parameters for the message display
607
+ *
608
+ * Display and format a warning messages, callback.
609
+ */
610
+ typedef void (XMLCDECL *warningSAXFunc) (void *ctx,
611
+ const char *msg, ...) LIBXML_ATTR_FORMAT(2,3);
612
+ /**
613
+ * errorSAXFunc:
614
+ * @ctx: an XML parser context
615
+ * @msg: the message to display/transmit
616
+ * @...: extra parameters for the message display
617
+ *
618
+ * Display and format an error messages, callback.
619
+ */
620
+ typedef void (XMLCDECL *errorSAXFunc) (void *ctx,
621
+ const char *msg, ...) LIBXML_ATTR_FORMAT(2,3);
622
+ /**
623
+ * fatalErrorSAXFunc:
624
+ * @ctx: an XML parser context
625
+ * @msg: the message to display/transmit
626
+ * @...: extra parameters for the message display
627
+ *
628
+ * Display and format fatal error messages, callback.
629
+ * Note: so far fatalError() SAX callbacks are not used, error()
630
+ * get all the callbacks for errors.
631
+ */
632
+ typedef void (XMLCDECL *fatalErrorSAXFunc) (void *ctx,
633
+ const char *msg, ...) LIBXML_ATTR_FORMAT(2,3);
634
+ /**
635
+ * isStandaloneSAXFunc:
636
+ * @ctx: the user data (XML parser context)
637
+ *
638
+ * Is this document tagged standalone?
639
+ *
640
+ * Returns 1 if true
641
+ */
642
+ typedef int (*isStandaloneSAXFunc) (void *ctx);
643
+ /**
644
+ * hasInternalSubsetSAXFunc:
645
+ * @ctx: the user data (XML parser context)
646
+ *
647
+ * Does this document has an internal subset.
648
+ *
649
+ * Returns 1 if true
650
+ */
651
+ typedef int (*hasInternalSubsetSAXFunc) (void *ctx);
652
+
653
+ /**
654
+ * hasExternalSubsetSAXFunc:
655
+ * @ctx: the user data (XML parser context)
656
+ *
657
+ * Does this document has an external subset?
658
+ *
659
+ * Returns 1 if true
660
+ */
661
+ typedef int (*hasExternalSubsetSAXFunc) (void *ctx);
662
+
663
+ /************************************************************************
664
+ * *
665
+ * The SAX version 2 API extensions *
666
+ * *
667
+ ************************************************************************/
668
+ /**
669
+ * XML_SAX2_MAGIC:
670
+ *
671
+ * Special constant found in SAX2 blocks initialized fields
672
+ */
673
+ #define XML_SAX2_MAGIC 0xDEEDBEAF
674
+
675
+ /**
676
+ * startElementNsSAX2Func:
677
+ * @ctx: the user data (XML parser context)
678
+ * @localname: the local name of the element
679
+ * @prefix: the element namespace prefix if available
680
+ * @URI: the element namespace name if available
681
+ * @nb_namespaces: number of namespace definitions on that node
682
+ * @namespaces: pointer to the array of prefix/URI pairs namespace definitions
683
+ * @nb_attributes: the number of attributes on that node
684
+ * @nb_defaulted: the number of defaulted attributes. The defaulted
685
+ * ones are at the end of the array
686
+ * @attributes: pointer to the array of (localname/prefix/URI/value/end)
687
+ * attribute values.
688
+ *
689
+ * SAX2 callback when an element start has been detected by the parser.
690
+ * It provides the namespace information for the element, as well as
691
+ * the new namespace declarations on the element.
692
+ */
693
+
694
+ typedef void (*startElementNsSAX2Func) (void *ctx,
695
+ const xmlChar *localname,
696
+ const xmlChar *prefix,
697
+ const xmlChar *URI,
698
+ int nb_namespaces,
699
+ const xmlChar **namespaces,
700
+ int nb_attributes,
701
+ int nb_defaulted,
702
+ const xmlChar **attributes);
703
+
704
+ /**
705
+ * endElementNsSAX2Func:
706
+ * @ctx: the user data (XML parser context)
707
+ * @localname: the local name of the element
708
+ * @prefix: the element namespace prefix if available
709
+ * @URI: the element namespace name if available
710
+ *
711
+ * SAX2 callback when an element end has been detected by the parser.
712
+ * It provides the namespace information for the element.
713
+ */
714
+
715
+ typedef void (*endElementNsSAX2Func) (void *ctx,
716
+ const xmlChar *localname,
717
+ const xmlChar *prefix,
718
+ const xmlChar *URI);
719
+
720
+
721
+ struct _xmlSAXHandler {
722
+ internalSubsetSAXFunc internalSubset;
723
+ isStandaloneSAXFunc isStandalone;
724
+ hasInternalSubsetSAXFunc hasInternalSubset;
725
+ hasExternalSubsetSAXFunc hasExternalSubset;
726
+ resolveEntitySAXFunc resolveEntity;
727
+ getEntitySAXFunc getEntity;
728
+ entityDeclSAXFunc entityDecl;
729
+ notationDeclSAXFunc notationDecl;
730
+ attributeDeclSAXFunc attributeDecl;
731
+ elementDeclSAXFunc elementDecl;
732
+ unparsedEntityDeclSAXFunc unparsedEntityDecl;
733
+ setDocumentLocatorSAXFunc setDocumentLocator;
734
+ startDocumentSAXFunc startDocument;
735
+ endDocumentSAXFunc endDocument;
736
+ startElementSAXFunc startElement;
737
+ endElementSAXFunc endElement;
738
+ referenceSAXFunc reference;
739
+ charactersSAXFunc characters;
740
+ ignorableWhitespaceSAXFunc ignorableWhitespace;
741
+ processingInstructionSAXFunc processingInstruction;
742
+ commentSAXFunc comment;
743
+ warningSAXFunc warning;
744
+ errorSAXFunc error;
745
+ fatalErrorSAXFunc fatalError; /* unused error() get all the errors */
746
+ getParameterEntitySAXFunc getParameterEntity;
747
+ cdataBlockSAXFunc cdataBlock;
748
+ externalSubsetSAXFunc externalSubset;
749
+ unsigned int initialized;
750
+ /* The following fields are extensions available only on version 2 */
751
+ void *_private;
752
+ startElementNsSAX2Func startElementNs;
753
+ endElementNsSAX2Func endElementNs;
754
+ xmlStructuredErrorFunc serror;
755
+ };
756
+
757
+ /*
758
+ * SAX Version 1
759
+ */
760
+ typedef struct _xmlSAXHandlerV1 xmlSAXHandlerV1;
761
+ typedef xmlSAXHandlerV1 *xmlSAXHandlerV1Ptr;
762
+ struct _xmlSAXHandlerV1 {
763
+ internalSubsetSAXFunc internalSubset;
764
+ isStandaloneSAXFunc isStandalone;
765
+ hasInternalSubsetSAXFunc hasInternalSubset;
766
+ hasExternalSubsetSAXFunc hasExternalSubset;
767
+ resolveEntitySAXFunc resolveEntity;
768
+ getEntitySAXFunc getEntity;
769
+ entityDeclSAXFunc entityDecl;
770
+ notationDeclSAXFunc notationDecl;
771
+ attributeDeclSAXFunc attributeDecl;
772
+ elementDeclSAXFunc elementDecl;
773
+ unparsedEntityDeclSAXFunc unparsedEntityDecl;
774
+ setDocumentLocatorSAXFunc setDocumentLocator;
775
+ startDocumentSAXFunc startDocument;
776
+ endDocumentSAXFunc endDocument;
777
+ startElementSAXFunc startElement;
778
+ endElementSAXFunc endElement;
779
+ referenceSAXFunc reference;
780
+ charactersSAXFunc characters;
781
+ ignorableWhitespaceSAXFunc ignorableWhitespace;
782
+ processingInstructionSAXFunc processingInstruction;
783
+ commentSAXFunc comment;
784
+ warningSAXFunc warning;
785
+ errorSAXFunc error;
786
+ fatalErrorSAXFunc fatalError; /* unused error() get all the errors */
787
+ getParameterEntitySAXFunc getParameterEntity;
788
+ cdataBlockSAXFunc cdataBlock;
789
+ externalSubsetSAXFunc externalSubset;
790
+ unsigned int initialized;
791
+ };
792
+
793
+
794
+ /**
795
+ * xmlExternalEntityLoader:
796
+ * @URL: The System ID of the resource requested
797
+ * @ID: The Public ID of the resource requested
798
+ * @context: the XML parser context
799
+ *
800
+ * External entity loaders types.
801
+ *
802
+ * Returns the entity input parser.
803
+ */
804
+ typedef xmlParserInputPtr (*xmlExternalEntityLoader) (const char *URL,
805
+ const char *ID,
806
+ xmlParserCtxtPtr context);
807
+
808
+ #ifdef __cplusplus
809
+ }
810
+ #endif
811
+
812
+ #include "encoding.h"
813
+ #include "xmlIO.h"
814
+ #include "globals.h"
815
+
816
+ #ifdef __cplusplus
817
+ extern "C" {
818
+ #endif
819
+
820
+
821
+ /*
822
+ * Init/Cleanup
823
+ */
824
+ XMLPUBFUN void XMLCALL
825
+ xmlInitParser (void);
826
+ XMLPUBFUN void XMLCALL
827
+ xmlCleanupParser (void);
828
+
829
+ /*
830
+ * Input functions
831
+ */
832
+ XMLPUBFUN int XMLCALL
833
+ xmlParserInputRead (xmlParserInputPtr in,
834
+ int len);
835
+ XMLPUBFUN int XMLCALL
836
+ xmlParserInputGrow (xmlParserInputPtr in,
837
+ int len);
838
+
839
+ /*
840
+ * Basic parsing Interfaces
841
+ */
842
+ #ifdef LIBXML_SAX1_ENABLED
843
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
844
+ xmlParseDoc (const xmlChar *cur);
845
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
846
+ xmlParseFile (const char *filename);
847
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
848
+ xmlParseMemory (const char *buffer,
849
+ int size);
850
+ #endif /* LIBXML_SAX1_ENABLED */
851
+ XMLPUBFUN int XMLCALL
852
+ xmlSubstituteEntitiesDefault(int val);
853
+ XMLPUBFUN int XMLCALL
854
+ xmlKeepBlanksDefault (int val);
855
+ XMLPUBFUN void XMLCALL
856
+ xmlStopParser (xmlParserCtxtPtr ctxt);
857
+ XMLPUBFUN int XMLCALL
858
+ xmlPedanticParserDefault(int val);
859
+ XMLPUBFUN int XMLCALL
860
+ xmlLineNumbersDefault (int val);
861
+
862
+ #ifdef LIBXML_SAX1_ENABLED
863
+ /*
864
+ * Recovery mode
865
+ */
866
+ XML_DEPRECATED
867
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
868
+ xmlRecoverDoc (const xmlChar *cur);
869
+ XML_DEPRECATED
870
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
871
+ xmlRecoverMemory (const char *buffer,
872
+ int size);
873
+ XML_DEPRECATED
874
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
875
+ xmlRecoverFile (const char *filename);
876
+ #endif /* LIBXML_SAX1_ENABLED */
877
+
878
+ /*
879
+ * Less common routines and SAX interfaces
880
+ */
881
+ XMLPUBFUN int XMLCALL
882
+ xmlParseDocument (xmlParserCtxtPtr ctxt);
883
+ XMLPUBFUN int XMLCALL
884
+ xmlParseExtParsedEnt (xmlParserCtxtPtr ctxt);
885
+ #ifdef LIBXML_SAX1_ENABLED
886
+ XML_DEPRECATED
887
+ XMLPUBFUN int XMLCALL
888
+ xmlSAXUserParseFile (xmlSAXHandlerPtr sax,
889
+ void *user_data,
890
+ const char *filename);
891
+ XML_DEPRECATED
892
+ XMLPUBFUN int XMLCALL
893
+ xmlSAXUserParseMemory (xmlSAXHandlerPtr sax,
894
+ void *user_data,
895
+ const char *buffer,
896
+ int size);
897
+ XML_DEPRECATED
898
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
899
+ xmlSAXParseDoc (xmlSAXHandlerPtr sax,
900
+ const xmlChar *cur,
901
+ int recovery);
902
+ XML_DEPRECATED
903
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
904
+ xmlSAXParseMemory (xmlSAXHandlerPtr sax,
905
+ const char *buffer,
906
+ int size,
907
+ int recovery);
908
+ XML_DEPRECATED
909
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
910
+ xmlSAXParseMemoryWithData (xmlSAXHandlerPtr sax,
911
+ const char *buffer,
912
+ int size,
913
+ int recovery,
914
+ void *data);
915
+ XML_DEPRECATED
916
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
917
+ xmlSAXParseFile (xmlSAXHandlerPtr sax,
918
+ const char *filename,
919
+ int recovery);
920
+ XML_DEPRECATED
921
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
922
+ xmlSAXParseFileWithData (xmlSAXHandlerPtr sax,
923
+ const char *filename,
924
+ int recovery,
925
+ void *data);
926
+ XML_DEPRECATED
927
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
928
+ xmlSAXParseEntity (xmlSAXHandlerPtr sax,
929
+ const char *filename);
930
+ XML_DEPRECATED
931
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
932
+ xmlParseEntity (const char *filename);
933
+ #endif /* LIBXML_SAX1_ENABLED */
934
+
935
+ #ifdef LIBXML_VALID_ENABLED
936
+ XML_DEPRECATED
937
+ XMLPUBFUN xmlDtdPtr XMLCALL
938
+ xmlSAXParseDTD (xmlSAXHandlerPtr sax,
939
+ const xmlChar *ExternalID,
940
+ const xmlChar *SystemID);
941
+ XMLPUBFUN xmlDtdPtr XMLCALL
942
+ xmlParseDTD (const xmlChar *ExternalID,
943
+ const xmlChar *SystemID);
944
+ XMLPUBFUN xmlDtdPtr XMLCALL
945
+ xmlIOParseDTD (xmlSAXHandlerPtr sax,
946
+ xmlParserInputBufferPtr input,
947
+ xmlCharEncoding enc);
948
+ #endif /* LIBXML_VALID_ENABLE */
949
+ #ifdef LIBXML_SAX1_ENABLED
950
+ XMLPUBFUN int XMLCALL
951
+ xmlParseBalancedChunkMemory(lxb_dom_document_t_ptr doc,
952
+ xmlSAXHandlerPtr sax,
953
+ void *user_data,
954
+ int depth,
955
+ const xmlChar *string,
956
+ lxb_dom_node_t_ptr *lst);
957
+ #endif /* LIBXML_SAX1_ENABLED */
958
+ XMLPUBFUN xmlParserErrors XMLCALL
959
+ xmlParseInNodeContext (lxb_dom_node_t_ptr node,
960
+ const char *data,
961
+ int datalen,
962
+ int options,
963
+ lxb_dom_node_t_ptr *lst);
964
+ #ifdef LIBXML_SAX1_ENABLED
965
+ XMLPUBFUN int XMLCALL
966
+ xmlParseBalancedChunkMemoryRecover(lxb_dom_document_t_ptr doc,
967
+ xmlSAXHandlerPtr sax,
968
+ void *user_data,
969
+ int depth,
970
+ const xmlChar *string,
971
+ lxb_dom_node_t_ptr *lst,
972
+ int recover);
973
+ XML_DEPRECATED
974
+ XMLPUBFUN int XMLCALL
975
+ xmlParseExternalEntity (lxb_dom_document_t_ptr doc,
976
+ xmlSAXHandlerPtr sax,
977
+ void *user_data,
978
+ int depth,
979
+ const xmlChar *URL,
980
+ const xmlChar *ID,
981
+ lxb_dom_node_t_ptr *lst);
982
+ #endif /* LIBXML_SAX1_ENABLED */
983
+ XMLPUBFUN int XMLCALL
984
+ xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx,
985
+ const xmlChar *URL,
986
+ const xmlChar *ID,
987
+ lxb_dom_node_t_ptr *lst);
988
+
989
+ /*
990
+ * Parser contexts handling.
991
+ */
992
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
993
+ xmlNewParserCtxt (void);
994
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
995
+ xmlNewSAXParserCtxt (const xmlSAXHandler *sax,
996
+ void *userData);
997
+ XMLPUBFUN int XMLCALL
998
+ xmlInitParserCtxt (xmlParserCtxtPtr ctxt);
999
+ XMLPUBFUN void XMLCALL
1000
+ xmlClearParserCtxt (xmlParserCtxtPtr ctxt);
1001
+ XMLPUBFUN void XMLCALL
1002
+ xmlFreeParserCtxt (xmlParserCtxtPtr ctxt);
1003
+ #ifdef LIBXML_SAX1_ENABLED
1004
+ XML_DEPRECATED
1005
+ XMLPUBFUN void XMLCALL
1006
+ xmlSetupParserForBuffer (xmlParserCtxtPtr ctxt,
1007
+ const xmlChar* buffer,
1008
+ const char *filename);
1009
+ #endif /* LIBXML_SAX1_ENABLED */
1010
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
1011
+ xmlCreateDocParserCtxt (const xmlChar *cur);
1012
+
1013
+ #ifdef LIBXML_LEGACY_ENABLED
1014
+ /*
1015
+ * Reading/setting optional parsing features.
1016
+ */
1017
+ XML_DEPRECATED
1018
+ XMLPUBFUN int XMLCALL
1019
+ xmlGetFeaturesList (int *len,
1020
+ const char **result);
1021
+ XML_DEPRECATED
1022
+ XMLPUBFUN int XMLCALL
1023
+ xmlGetFeature (xmlParserCtxtPtr ctxt,
1024
+ const char *name,
1025
+ void *result);
1026
+ XML_DEPRECATED
1027
+ XMLPUBFUN int XMLCALL
1028
+ xmlSetFeature (xmlParserCtxtPtr ctxt,
1029
+ const char *name,
1030
+ void *value);
1031
+ #endif /* LIBXML_LEGACY_ENABLED */
1032
+
1033
+ #ifdef LIBXML_PUSH_ENABLED
1034
+ /*
1035
+ * Interfaces for the Push mode.
1036
+ */
1037
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
1038
+ xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax,
1039
+ void *user_data,
1040
+ const char *chunk,
1041
+ int size,
1042
+ const char *filename);
1043
+ XMLPUBFUN int XMLCALL
1044
+ xmlParseChunk (xmlParserCtxtPtr ctxt,
1045
+ const char *chunk,
1046
+ int size,
1047
+ int terminate);
1048
+ #endif /* LIBXML_PUSH_ENABLED */
1049
+
1050
+ /*
1051
+ * Special I/O mode.
1052
+ */
1053
+
1054
+ XMLPUBFUN xmlParserCtxtPtr XMLCALL
1055
+ xmlCreateIOParserCtxt (xmlSAXHandlerPtr sax,
1056
+ void *user_data,
1057
+ xmlInputReadCallback ioread,
1058
+ xmlInputCloseCallback ioclose,
1059
+ void *ioctx,
1060
+ xmlCharEncoding enc);
1061
+
1062
+ XMLPUBFUN xmlParserInputPtr XMLCALL
1063
+ xmlNewIOInputStream (xmlParserCtxtPtr ctxt,
1064
+ xmlParserInputBufferPtr input,
1065
+ xmlCharEncoding enc);
1066
+
1067
+ /*
1068
+ * Node infos.
1069
+ */
1070
+ XMLPUBFUN const xmlParserNodeInfo* XMLCALL
1071
+ xmlParserFindNodeInfo (const xmlParserCtxtPtr ctxt,
1072
+ const lxb_dom_node_t_ptr node);
1073
+ XMLPUBFUN void XMLCALL
1074
+ xmlInitNodeInfoSeq (xmlParserNodeInfoSeqPtr seq);
1075
+ XMLPUBFUN void XMLCALL
1076
+ xmlClearNodeInfoSeq (xmlParserNodeInfoSeqPtr seq);
1077
+ XMLPUBFUN unsigned long XMLCALL
1078
+ xmlParserFindNodeInfoIndex(const xmlParserNodeInfoSeqPtr seq,
1079
+ const lxb_dom_node_t_ptr node);
1080
+ XMLPUBFUN void XMLCALL
1081
+ xmlParserAddNodeInfo (xmlParserCtxtPtr ctxt,
1082
+ const xmlParserNodeInfoPtr info);
1083
+
1084
+ /*
1085
+ * External entities handling actually implemented in xmlIO.
1086
+ */
1087
+
1088
+ XMLPUBFUN void XMLCALL
1089
+ xmlSetExternalEntityLoader(xmlExternalEntityLoader f);
1090
+ XMLPUBFUN xmlExternalEntityLoader XMLCALL
1091
+ xmlGetExternalEntityLoader(void);
1092
+ XMLPUBFUN xmlParserInputPtr XMLCALL
1093
+ xmlLoadExternalEntity (const char *URL,
1094
+ const char *ID,
1095
+ xmlParserCtxtPtr ctxt);
1096
+
1097
+ /*
1098
+ * Index lookup, actually implemented in the encoding module
1099
+ */
1100
+ XMLPUBFUN long XMLCALL
1101
+ xmlByteConsumed (xmlParserCtxtPtr ctxt);
1102
+
1103
+ /*
1104
+ * New set of simpler/more flexible APIs
1105
+ */
1106
+ /**
1107
+ * xmlParserOption:
1108
+ *
1109
+ * This is the set of XML parser options that can be passed down
1110
+ * to the xmlReadDoc() and similar calls.
1111
+ */
1112
+ typedef enum {
1113
+ XML_PARSE_RECOVER = 1<<0, /* recover on errors */
1114
+ XML_PARSE_NOENT = 1<<1, /* substitute entities */
1115
+ XML_PARSE_DTDLOAD = 1<<2, /* load the external subset */
1116
+ XML_PARSE_DTDATTR = 1<<3, /* default DTD attributes */
1117
+ XML_PARSE_DTDVALID = 1<<4, /* validate with the DTD */
1118
+ XML_PARSE_NOERROR = 1<<5, /* suppress error reports */
1119
+ XML_PARSE_NOWARNING = 1<<6, /* suppress warning reports */
1120
+ XML_PARSE_PEDANTIC = 1<<7, /* pedantic error reporting */
1121
+ XML_PARSE_NOBLANKS = 1<<8, /* remove blank nodes */
1122
+ XML_PARSE_SAX1 = 1<<9, /* use the SAX1 interface internally */
1123
+ XML_PARSE_XINCLUDE = 1<<10,/* Implement XInclude substitution */
1124
+ XML_PARSE_NONET = 1<<11,/* Forbid network access */
1125
+ XML_PARSE_NODICT = 1<<12,/* Do not reuse the context dictionary */
1126
+ XML_PARSE_NSCLEAN = 1<<13,/* remove redundant namespaces declarations */
1127
+ XML_PARSE_NOCDATA = 1<<14,/* merge CDATA as text nodes */
1128
+ XML_PARSE_NOXINCNODE= 1<<15,/* do not generate XINCLUDE START/END nodes */
1129
+ XML_PARSE_COMPACT = 1<<16,/* compact small text nodes; no modification of
1130
+ the tree allowed afterwards (will possibly
1131
+ crash if you try to modify the tree) */
1132
+ XML_PARSE_OLD10 = 1<<17,/* parse using XML-1.0 before update 5 */
1133
+ XML_PARSE_NOBASEFIX = 1<<18,/* do not fixup XINCLUDE xml:base uris */
1134
+ XML_PARSE_HUGE = 1<<19,/* relax any hardcoded limit from the parser */
1135
+ XML_PARSE_OLDSAX = 1<<20,/* parse using SAX2 interface before 2.7.0 */
1136
+ XML_PARSE_IGNORE_ENC= 1<<21,/* ignore internal document encoding hint */
1137
+ XML_PARSE_BIG_LINES = 1<<22 /* Store big lines numbers in text PSVI field */
1138
+ } xmlParserOption;
1139
+
1140
+ XMLPUBFUN void XMLCALL
1141
+ xmlCtxtReset (xmlParserCtxtPtr ctxt);
1142
+ XMLPUBFUN int XMLCALL
1143
+ xmlCtxtResetPush (xmlParserCtxtPtr ctxt,
1144
+ const char *chunk,
1145
+ int size,
1146
+ const char *filename,
1147
+ const char *encoding);
1148
+ XMLPUBFUN int XMLCALL
1149
+ xmlCtxtUseOptions (xmlParserCtxtPtr ctxt,
1150
+ int options);
1151
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1152
+ xmlReadDoc (const xmlChar *cur,
1153
+ const char *URL,
1154
+ const char *encoding,
1155
+ int options);
1156
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1157
+ xmlReadFile (const char *URL,
1158
+ const char *encoding,
1159
+ int options);
1160
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1161
+ xmlReadMemory (const char *buffer,
1162
+ int size,
1163
+ const char *URL,
1164
+ const char *encoding,
1165
+ int options);
1166
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1167
+ xmlReadFd (int fd,
1168
+ const char *URL,
1169
+ const char *encoding,
1170
+ int options);
1171
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1172
+ xmlReadIO (xmlInputReadCallback ioread,
1173
+ xmlInputCloseCallback ioclose,
1174
+ void *ioctx,
1175
+ const char *URL,
1176
+ const char *encoding,
1177
+ int options);
1178
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1179
+ xmlCtxtReadDoc (xmlParserCtxtPtr ctxt,
1180
+ const xmlChar *cur,
1181
+ const char *URL,
1182
+ const char *encoding,
1183
+ int options);
1184
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1185
+ xmlCtxtReadFile (xmlParserCtxtPtr ctxt,
1186
+ const char *filename,
1187
+ const char *encoding,
1188
+ int options);
1189
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1190
+ xmlCtxtReadMemory (xmlParserCtxtPtr ctxt,
1191
+ const char *buffer,
1192
+ int size,
1193
+ const char *URL,
1194
+ const char *encoding,
1195
+ int options);
1196
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1197
+ xmlCtxtReadFd (xmlParserCtxtPtr ctxt,
1198
+ int fd,
1199
+ const char *URL,
1200
+ const char *encoding,
1201
+ int options);
1202
+ XMLPUBFUN lxb_dom_document_t_ptr XMLCALL
1203
+ xmlCtxtReadIO (xmlParserCtxtPtr ctxt,
1204
+ xmlInputReadCallback ioread,
1205
+ xmlInputCloseCallback ioclose,
1206
+ void *ioctx,
1207
+ const char *URL,
1208
+ const char *encoding,
1209
+ int options);
1210
+
1211
+ /*
1212
+ * Library wide options
1213
+ */
1214
+ /**
1215
+ * xmlFeature:
1216
+ *
1217
+ * Used to examine the existence of features that can be enabled
1218
+ * or disabled at compile-time.
1219
+ * They used to be called XML_FEATURE_xxx but this clashed with Expat
1220
+ */
1221
+ typedef enum {
1222
+ XML_WITH_THREAD = 1,
1223
+ XML_WITH_TREE = 2,
1224
+ XML_WITH_OUTPUT = 3,
1225
+ XML_WITH_PUSH = 4,
1226
+ XML_WITH_READER = 5,
1227
+ XML_WITH_PATTERN = 6,
1228
+ XML_WITH_WRITER = 7,
1229
+ XML_WITH_SAX1 = 8,
1230
+ XML_WITH_FTP = 9,
1231
+ XML_WITH_HTTP = 10,
1232
+ XML_WITH_VALID = 11,
1233
+ XML_WITH_HTML = 12,
1234
+ XML_WITH_LEGACY = 13,
1235
+ XML_WITH_C14N = 14,
1236
+ XML_WITH_CATALOG = 15,
1237
+ XML_WITH_XPATH = 16,
1238
+ XML_WITH_XPTR = 17,
1239
+ XML_WITH_XINCLUDE = 18,
1240
+ XML_WITH_ICONV = 19,
1241
+ XML_WITH_ISO8859X = 20,
1242
+ XML_WITH_UNICODE = 21,
1243
+ XML_WITH_REGEXP = 22,
1244
+ XML_WITH_AUTOMATA = 23,
1245
+ XML_WITH_EXPR = 24,
1246
+ XML_WITH_SCHEMAS = 25,
1247
+ XML_WITH_SCHEMATRON = 26,
1248
+ XML_WITH_MODULES = 27,
1249
+ XML_WITH_DEBUG = 28,
1250
+ XML_WITH_DEBUG_MEM = 29,
1251
+ XML_WITH_DEBUG_RUN = 30,
1252
+ XML_WITH_ZLIB = 31,
1253
+ XML_WITH_ICU = 32,
1254
+ XML_WITH_LZMA = 33,
1255
+ XML_WITH_NONE = 99999 /* just to be sure of allocation size */
1256
+ } xmlFeature;
1257
+
1258
+ XMLPUBFUN int XMLCALL
1259
+ xmlHasFeature (xmlFeature feature);
1260
+
1261
+ #ifdef __cplusplus
1262
+ }
1263
+ #endif
1264
+ #endif /* __XML_PARSER_H__ */