nokogiri 1.6.8.1 → 1.13.4

Sign up to get free protection for your applications and to get access to all the features.

Potentially problematic release.


This version of nokogiri might be problematic. Click here for more details.

Files changed (354) hide show
  1. checksums.yaml +5 -5
  2. data/Gemfile +3 -20
  3. data/LICENSE-DEPENDENCIES.md +1903 -0
  4. data/LICENSE.md +9 -0
  5. data/README.md +197 -83
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +16 -22
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +751 -432
  10. data/ext/nokogiri/gumbo.c +584 -0
  11. data/ext/nokogiri/html4_document.c +166 -0
  12. data/ext/nokogiri/html4_element_description.c +294 -0
  13. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  14. data/ext/nokogiri/html4_sax_parser_context.c +120 -0
  15. data/ext/nokogiri/html4_sax_push_parser.c +95 -0
  16. data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
  17. data/ext/nokogiri/nokogiri.c +228 -95
  18. data/ext/nokogiri/nokogiri.h +190 -98
  19. data/ext/nokogiri/test_global_handlers.c +40 -0
  20. data/ext/nokogiri/xml_attr.c +49 -40
  21. data/ext/nokogiri/xml_attribute_decl.c +18 -18
  22. data/ext/nokogiri/xml_cdata.c +24 -23
  23. data/ext/nokogiri/xml_comment.c +19 -26
  24. data/ext/nokogiri/xml_document.c +291 -217
  25. data/ext/nokogiri/xml_document_fragment.c +12 -16
  26. data/ext/nokogiri/xml_dtd.c +56 -50
  27. data/ext/nokogiri/xml_element_content.c +31 -26
  28. data/ext/nokogiri/xml_element_decl.c +22 -22
  29. data/ext/nokogiri/xml_encoding_handler.c +43 -18
  30. data/ext/nokogiri/xml_entity_decl.c +32 -30
  31. data/ext/nokogiri/xml_entity_reference.c +16 -18
  32. data/ext/nokogiri/xml_namespace.c +61 -58
  33. data/ext/nokogiri/xml_node.c +1194 -729
  34. data/ext/nokogiri/xml_node_set.c +178 -165
  35. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  36. data/ext/nokogiri/xml_reader.c +226 -175
  37. data/ext/nokogiri/xml_relax_ng.c +52 -28
  38. data/ext/nokogiri/xml_sax_parser.c +115 -114
  39. data/ext/nokogiri/xml_sax_parser_context.c +105 -86
  40. data/ext/nokogiri/xml_sax_push_parser.c +88 -35
  41. data/ext/nokogiri/xml_schema.c +112 -33
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +13 -17
  44. data/ext/nokogiri/xml_xpath_context.c +223 -115
  45. data/ext/nokogiri/xslt_stylesheet.c +265 -173
  46. data/gumbo-parser/CHANGES.md +63 -0
  47. data/gumbo-parser/Makefile +101 -0
  48. data/gumbo-parser/THANKS +27 -0
  49. data/gumbo-parser/src/Makefile +34 -0
  50. data/gumbo-parser/src/README.md +41 -0
  51. data/gumbo-parser/src/ascii.c +75 -0
  52. data/gumbo-parser/src/ascii.h +115 -0
  53. data/gumbo-parser/src/attribute.c +42 -0
  54. data/gumbo-parser/src/attribute.h +17 -0
  55. data/gumbo-parser/src/char_ref.c +22225 -0
  56. data/gumbo-parser/src/char_ref.h +29 -0
  57. data/gumbo-parser/src/char_ref.rl +2154 -0
  58. data/gumbo-parser/src/error.c +626 -0
  59. data/gumbo-parser/src/error.h +148 -0
  60. data/gumbo-parser/src/foreign_attrs.c +104 -0
  61. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  62. data/gumbo-parser/src/gumbo.h +943 -0
  63. data/gumbo-parser/src/insertion_mode.h +33 -0
  64. data/gumbo-parser/src/macros.h +91 -0
  65. data/gumbo-parser/src/parser.c +4875 -0
  66. data/gumbo-parser/src/parser.h +41 -0
  67. data/gumbo-parser/src/replacement.h +33 -0
  68. data/gumbo-parser/src/string_buffer.c +103 -0
  69. data/gumbo-parser/src/string_buffer.h +68 -0
  70. data/gumbo-parser/src/string_piece.c +48 -0
  71. data/gumbo-parser/src/svg_attrs.c +174 -0
  72. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  73. data/gumbo-parser/src/svg_tags.c +137 -0
  74. data/gumbo-parser/src/svg_tags.gperf +55 -0
  75. data/gumbo-parser/src/tag.c +222 -0
  76. data/gumbo-parser/src/tag_lookup.c +382 -0
  77. data/gumbo-parser/src/tag_lookup.gperf +169 -0
  78. data/gumbo-parser/src/tag_lookup.h +13 -0
  79. data/gumbo-parser/src/token_buffer.c +79 -0
  80. data/gumbo-parser/src/token_buffer.h +71 -0
  81. data/gumbo-parser/src/token_type.h +17 -0
  82. data/gumbo-parser/src/tokenizer.c +3463 -0
  83. data/gumbo-parser/src/tokenizer.h +112 -0
  84. data/gumbo-parser/src/tokenizer_states.h +339 -0
  85. data/gumbo-parser/src/utf8.c +245 -0
  86. data/gumbo-parser/src/utf8.h +164 -0
  87. data/gumbo-parser/src/util.c +68 -0
  88. data/gumbo-parser/src/util.h +30 -0
  89. data/gumbo-parser/src/vector.c +111 -0
  90. data/gumbo-parser/src/vector.h +45 -0
  91. data/lib/nokogiri/class_resolver.rb +67 -0
  92. data/lib/nokogiri/css/node.rb +10 -8
  93. data/lib/nokogiri/css/parser.rb +410 -372
  94. data/lib/nokogiri/css/parser.y +260 -244
  95. data/lib/nokogiri/css/parser_extras.rb +54 -49
  96. data/lib/nokogiri/css/syntax_error.rb +3 -1
  97. data/lib/nokogiri/css/tokenizer.rb +107 -104
  98. data/lib/nokogiri/css/tokenizer.rex +7 -6
  99. data/lib/nokogiri/css/xpath_visitor.rb +226 -92
  100. data/lib/nokogiri/css.rb +50 -17
  101. data/lib/nokogiri/decorators/slop.rb +9 -7
  102. data/lib/nokogiri/extension.rb +31 -0
  103. data/lib/nokogiri/gumbo.rb +15 -0
  104. data/lib/nokogiri/html.rb +38 -27
  105. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  106. data/lib/nokogiri/{html → html4}/document.rb +104 -106
  107. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  108. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  109. data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
  110. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  111. data/lib/nokogiri/{html → html4}/sax/parser.rb +24 -13
  112. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  113. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  114. data/lib/nokogiri/html4.rb +46 -0
  115. data/lib/nokogiri/html5/document.rb +91 -0
  116. data/lib/nokogiri/html5/document_fragment.rb +83 -0
  117. data/lib/nokogiri/html5/node.rb +100 -0
  118. data/lib/nokogiri/html5.rb +478 -0
  119. data/lib/nokogiri/jruby/dependencies.rb +21 -0
  120. data/lib/nokogiri/syntax_error.rb +2 -0
  121. data/lib/nokogiri/version/constant.rb +6 -0
  122. data/lib/nokogiri/version/info.rb +222 -0
  123. data/lib/nokogiri/version.rb +3 -107
  124. data/lib/nokogiri/xml/attr.rb +6 -3
  125. data/lib/nokogiri/xml/attribute_decl.rb +3 -1
  126. data/lib/nokogiri/xml/builder.rb +97 -53
  127. data/lib/nokogiri/xml/cdata.rb +3 -1
  128. data/lib/nokogiri/xml/character_data.rb +2 -0
  129. data/lib/nokogiri/xml/document.rb +231 -93
  130. data/lib/nokogiri/xml/document_fragment.rb +57 -44
  131. data/lib/nokogiri/xml/dtd.rb +4 -2
  132. data/lib/nokogiri/xml/element_content.rb +2 -0
  133. data/lib/nokogiri/xml/element_decl.rb +3 -1
  134. data/lib/nokogiri/xml/entity_decl.rb +4 -2
  135. data/lib/nokogiri/xml/entity_reference.rb +20 -0
  136. data/lib/nokogiri/xml/namespace.rb +3 -0
  137. data/lib/nokogiri/xml/node/save_options.rb +10 -5
  138. data/lib/nokogiri/xml/node.rb +911 -337
  139. data/lib/nokogiri/xml/node_set.rb +141 -84
  140. data/lib/nokogiri/xml/notation.rb +13 -0
  141. data/lib/nokogiri/xml/parse_options.rb +23 -9
  142. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  143. data/lib/nokogiri/xml/pp/node.rb +25 -26
  144. data/lib/nokogiri/xml/pp.rb +4 -2
  145. data/lib/nokogiri/xml/processing_instruction.rb +3 -1
  146. data/lib/nokogiri/xml/reader.rb +23 -28
  147. data/lib/nokogiri/xml/relax_ng.rb +8 -2
  148. data/lib/nokogiri/xml/sax/document.rb +45 -49
  149. data/lib/nokogiri/xml/sax/parser.rb +43 -40
  150. data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
  151. data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
  152. data/lib/nokogiri/xml/sax.rb +6 -4
  153. data/lib/nokogiri/xml/schema.rb +19 -9
  154. data/lib/nokogiri/xml/searchable.rb +138 -89
  155. data/lib/nokogiri/xml/syntax_error.rb +26 -1
  156. data/lib/nokogiri/xml/text.rb +2 -0
  157. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  158. data/lib/nokogiri/xml/xpath.rb +15 -4
  159. data/lib/nokogiri/xml/xpath_context.rb +3 -3
  160. data/lib/nokogiri/xml.rb +39 -37
  161. data/lib/nokogiri/xslt/stylesheet.rb +3 -1
  162. data/lib/nokogiri/xslt.rb +29 -20
  163. data/lib/nokogiri.rb +49 -65
  164. data/lib/xsd/xmlparser/nokogiri.rb +26 -24
  165. data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
  166. data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
  167. data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
  168. data/patches/libxml2/0004-use-glibc-strlen.patch +53 -0
  169. data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  170. data/patches/libxml2/0006-update-automake-files-for-arm64.patch +3040 -0
  171. data/patches/libxml2/0008-htmlParseComment-handle-abruptly-closed-comments.patch +61 -0
  172. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  173. data/patches/libxml2/0010-Revert-Different-approach-to-fix-quadratic-behavior.patch +45 -0
  174. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
  175. data/ports/archives/libxml2-2.9.13.tar.xz +0 -0
  176. data/ports/archives/libxslt-1.1.35.tar.xz +0 -0
  177. metadata +233 -258
  178. data/.autotest +0 -26
  179. data/.cross_rubies +0 -9
  180. data/.editorconfig +0 -17
  181. data/.gemtest +0 -0
  182. data/.travis.yml +0 -51
  183. data/CHANGELOG.rdoc +0 -1160
  184. data/CONTRIBUTING.md +0 -42
  185. data/C_CODING_STYLE.rdoc +0 -33
  186. data/LICENSE.txt +0 -31
  187. data/Manifest.txt +0 -364
  188. data/ROADMAP.md +0 -111
  189. data/Rakefile +0 -375
  190. data/STANDARD_RESPONSES.md +0 -47
  191. data/Y_U_NO_GEMSPEC.md +0 -155
  192. data/appveyor.yml +0 -22
  193. data/build_all +0 -45
  194. data/ext/nokogiri/html_document.c +0 -170
  195. data/ext/nokogiri/html_document.h +0 -10
  196. data/ext/nokogiri/html_element_description.c +0 -279
  197. data/ext/nokogiri/html_element_description.h +0 -10
  198. data/ext/nokogiri/html_entity_lookup.c +0 -32
  199. data/ext/nokogiri/html_entity_lookup.h +0 -8
  200. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  201. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  202. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  203. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  204. data/ext/nokogiri/xml_attr.h +0 -9
  205. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  206. data/ext/nokogiri/xml_cdata.h +0 -9
  207. data/ext/nokogiri/xml_comment.h +0 -9
  208. data/ext/nokogiri/xml_document.h +0 -23
  209. data/ext/nokogiri/xml_document_fragment.h +0 -10
  210. data/ext/nokogiri/xml_dtd.h +0 -10
  211. data/ext/nokogiri/xml_element_content.h +0 -10
  212. data/ext/nokogiri/xml_element_decl.h +0 -9
  213. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  214. data/ext/nokogiri/xml_entity_decl.h +0 -10
  215. data/ext/nokogiri/xml_entity_reference.h +0 -9
  216. data/ext/nokogiri/xml_io.c +0 -60
  217. data/ext/nokogiri/xml_io.h +0 -11
  218. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  219. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  220. data/ext/nokogiri/xml_namespace.h +0 -13
  221. data/ext/nokogiri/xml_node.h +0 -13
  222. data/ext/nokogiri/xml_node_set.h +0 -13
  223. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  224. data/ext/nokogiri/xml_reader.h +0 -10
  225. data/ext/nokogiri/xml_relax_ng.h +0 -9
  226. data/ext/nokogiri/xml_sax_parser.h +0 -39
  227. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  228. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  229. data/ext/nokogiri/xml_schema.h +0 -9
  230. data/ext/nokogiri/xml_syntax_error.h +0 -13
  231. data/ext/nokogiri/xml_text.h +0 -9
  232. data/ext/nokogiri/xml_xpath_context.h +0 -10
  233. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  234. data/lib/nokogiri/html/document_fragment.rb +0 -39
  235. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  236. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  237. data/patches/sort-patches-by-date +0 -25
  238. data/ports/archives/libxml2-2.9.4.tar.gz +0 -0
  239. data/ports/archives/libxslt-1.1.29.tar.gz +0 -0
  240. data/suppressions/README.txt +0 -1
  241. data/suppressions/nokogiri_ree-1.8.7.358.supp +0 -61
  242. data/suppressions/nokogiri_ruby-1.8.7.370.supp +0 -0
  243. data/suppressions/nokogiri_ruby-1.9.2.320.supp +0 -28
  244. data/suppressions/nokogiri_ruby-1.9.3.327.supp +0 -28
  245. data/tasks/test.rb +0 -100
  246. data/test/css/test_nthiness.rb +0 -226
  247. data/test/css/test_parser.rb +0 -369
  248. data/test/css/test_tokenizer.rb +0 -215
  249. data/test/css/test_xpath_visitor.rb +0 -96
  250. data/test/decorators/test_slop.rb +0 -20
  251. data/test/files/2ch.html +0 -108
  252. data/test/files/GH_1042.html +0 -18
  253. data/test/files/address_book.rlx +0 -12
  254. data/test/files/address_book.xml +0 -10
  255. data/test/files/atom.xml +0 -344
  256. data/test/files/bar/bar.xsd +0 -4
  257. data/test/files/bogus.xml +0 -0
  258. data/test/files/dont_hurt_em_why.xml +0 -422
  259. data/test/files/encoding.html +0 -82
  260. data/test/files/encoding.xhtml +0 -84
  261. data/test/files/exslt.xml +0 -8
  262. data/test/files/exslt.xslt +0 -35
  263. data/test/files/foo/foo.xsd +0 -4
  264. data/test/files/metacharset.html +0 -10
  265. data/test/files/namespace_pressure_test.xml +0 -1684
  266. data/test/files/noencoding.html +0 -47
  267. data/test/files/po.xml +0 -32
  268. data/test/files/po.xsd +0 -66
  269. data/test/files/saml/saml20assertion_schema.xsd +0 -283
  270. data/test/files/saml/saml20protocol_schema.xsd +0 -302
  271. data/test/files/saml/xenc_schema.xsd +0 -146
  272. data/test/files/saml/xmldsig_schema.xsd +0 -318
  273. data/test/files/shift_jis.html +0 -10
  274. data/test/files/shift_jis.xml +0 -5
  275. data/test/files/shift_jis_no_charset.html +0 -9
  276. data/test/files/slow-xpath.xml +0 -25509
  277. data/test/files/snuggles.xml +0 -3
  278. data/test/files/staff.dtd +0 -10
  279. data/test/files/staff.xml +0 -59
  280. data/test/files/staff.xslt +0 -32
  281. data/test/files/test_document_url/bar.xml +0 -2
  282. data/test/files/test_document_url/document.dtd +0 -4
  283. data/test/files/test_document_url/document.xml +0 -6
  284. data/test/files/tlm.html +0 -851
  285. data/test/files/to_be_xincluded.xml +0 -2
  286. data/test/files/valid_bar.xml +0 -2
  287. data/test/files/xinclude.xml +0 -4
  288. data/test/helper.rb +0 -181
  289. data/test/html/sax/test_parser.rb +0 -141
  290. data/test/html/sax/test_parser_context.rb +0 -46
  291. data/test/html/sax/test_push_parser.rb +0 -87
  292. data/test/html/test_builder.rb +0 -164
  293. data/test/html/test_document.rb +0 -701
  294. data/test/html/test_document_encoding.rb +0 -145
  295. data/test/html/test_document_fragment.rb +0 -301
  296. data/test/html/test_element_description.rb +0 -105
  297. data/test/html/test_named_characters.rb +0 -14
  298. data/test/html/test_node.rb +0 -212
  299. data/test/html/test_node_encoding.rb +0 -85
  300. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
  301. data/test/namespaces/test_namespaces_aliased_default.rb +0 -24
  302. data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
  303. data/test/namespaces/test_namespaces_in_cloned_doc.rb +0 -31
  304. data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
  305. data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -80
  306. data/test/namespaces/test_namespaces_preservation.rb +0 -31
  307. data/test/test_convert_xpath.rb +0 -135
  308. data/test/test_css_cache.rb +0 -45
  309. data/test/test_encoding_handler.rb +0 -48
  310. data/test/test_memory_leak.rb +0 -156
  311. data/test/test_nokogiri.rb +0 -138
  312. data/test/test_soap4r_sax.rb +0 -52
  313. data/test/test_xslt_transforms.rb +0 -314
  314. data/test/xml/node/test_save_options.rb +0 -28
  315. data/test/xml/node/test_subclass.rb +0 -44
  316. data/test/xml/sax/test_parser.rb +0 -394
  317. data/test/xml/sax/test_parser_context.rb +0 -115
  318. data/test/xml/sax/test_push_parser.rb +0 -157
  319. data/test/xml/test_attr.rb +0 -67
  320. data/test/xml/test_attribute_decl.rb +0 -86
  321. data/test/xml/test_builder.rb +0 -341
  322. data/test/xml/test_c14n.rb +0 -180
  323. data/test/xml/test_cdata.rb +0 -48
  324. data/test/xml/test_comment.rb +0 -40
  325. data/test/xml/test_document.rb +0 -982
  326. data/test/xml/test_document_encoding.rb +0 -31
  327. data/test/xml/test_document_fragment.rb +0 -271
  328. data/test/xml/test_dtd.rb +0 -187
  329. data/test/xml/test_dtd_encoding.rb +0 -31
  330. data/test/xml/test_element_content.rb +0 -56
  331. data/test/xml/test_element_decl.rb +0 -73
  332. data/test/xml/test_entity_decl.rb +0 -122
  333. data/test/xml/test_entity_reference.rb +0 -251
  334. data/test/xml/test_namespace.rb +0 -96
  335. data/test/xml/test_node.rb +0 -1244
  336. data/test/xml/test_node_attributes.rb +0 -115
  337. data/test/xml/test_node_encoding.rb +0 -69
  338. data/test/xml/test_node_inheritance.rb +0 -32
  339. data/test/xml/test_node_reparenting.rb +0 -549
  340. data/test/xml/test_node_set.rb +0 -775
  341. data/test/xml/test_parse_options.rb +0 -64
  342. data/test/xml/test_processing_instruction.rb +0 -30
  343. data/test/xml/test_reader.rb +0 -589
  344. data/test/xml/test_reader_encoding.rb +0 -134
  345. data/test/xml/test_relax_ng.rb +0 -60
  346. data/test/xml/test_schema.rb +0 -142
  347. data/test/xml/test_syntax_error.rb +0 -30
  348. data/test/xml/test_text.rb +0 -60
  349. data/test/xml/test_unparented_node.rb +0 -440
  350. data/test/xml/test_xinclude.rb +0 -83
  351. data/test/xml/test_xpath.rb +0 -445
  352. data/test/xslt/test_custom_functions.rb +0 -133
  353. data/test/xslt/test_exception_handling.rb +0 -37
  354. data/test_all +0 -107
@@ -1,87 +0,0 @@
1
- #include <html_sax_push_parser.h>
2
-
3
- /*
4
- * call-seq:
5
- * native_write(chunk, last_chunk)
6
- *
7
- * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
8
- */
9
- static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
10
- {
11
- xmlParserCtxtPtr ctx;
12
- const char * chunk = NULL;
13
- int size = 0;
14
-
15
-
16
- Data_Get_Struct(self, xmlParserCtxt, ctx);
17
-
18
- if(Qnil != _chunk) {
19
- chunk = StringValuePtr(_chunk);
20
- size = (int)RSTRING_LEN(_chunk);
21
- }
22
-
23
- if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
24
- if (!(ctx->options & XML_PARSE_RECOVER)) {
25
- xmlErrorPtr e = xmlCtxtGetLastError(ctx);
26
- Nokogiri_error_raise(NULL, e);
27
- }
28
- }
29
-
30
- return self;
31
- }
32
-
33
- /*
34
- * call-seq:
35
- * initialize_native(xml_sax, filename)
36
- *
37
- * Initialize the push parser with +xml_sax+ using +filename+
38
- */
39
- static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
40
- VALUE encoding)
41
- {
42
- htmlSAXHandlerPtr sax;
43
- const char * filename = NULL;
44
- htmlParserCtxtPtr ctx;
45
- xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
46
-
47
- Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
48
-
49
- if(_filename != Qnil) filename = StringValueCStr(_filename);
50
-
51
- if (!NIL_P(encoding)) {
52
- enc = xmlParseCharEncoding(StringValueCStr(encoding));
53
- if (enc == XML_CHAR_ENCODING_ERROR)
54
- rb_raise(rb_eArgError, "Unsupported Encoding");
55
- }
56
-
57
- ctx = htmlCreatePushParserCtxt(
58
- sax,
59
- NULL,
60
- NULL,
61
- 0,
62
- filename,
63
- enc
64
- );
65
- if(ctx == NULL)
66
- rb_raise(rb_eRuntimeError, "Could not create a parser context");
67
-
68
- ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
69
-
70
- ctx->sax2 = 1;
71
- DATA_PTR(self) = ctx;
72
- return self;
73
- }
74
-
75
- VALUE cNokogiriHtmlSaxPushParser;
76
- void init_html_sax_push_parser()
77
- {
78
- VALUE nokogiri = rb_define_module("Nokogiri");
79
- VALUE html = rb_define_module_under(nokogiri, "HTML");
80
- VALUE sax = rb_define_module_under(html, "SAX");
81
- VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
82
-
83
- cNokogiriHtmlSaxPushParser = klass;
84
-
85
- rb_define_private_method(klass, "initialize_native", initialize_native, 3);
86
- rb_define_private_method(klass, "native_write", native_write, 2);
87
- }
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_HTML_SAX_PUSH_PARSER
2
- #define NOKOGIRI_HTML_SAX_PUSH_PARSER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_html_sax_push_parser();
7
-
8
- extern VALUE cNokogiriHtmlSaxPushParser ;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ATTR
2
- #define NOKOGIRI_XML_ATTR
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_attr();
7
-
8
- extern VALUE cNokogiriXmlAttr;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ATTRIBUTE_DECL
2
- #define NOKOGIRI_XML_ATTRIBUTE_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_attribute_decl();
7
-
8
- extern VALUE cNokogiriXmlAttributeDecl;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_CDATA
2
- #define NOKOGIRI_XML_CDATA
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_cdata();
7
-
8
- extern VALUE cNokogiriXmlCData;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_COMMENT
2
- #define NOKOGIRI_XML_COMMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_comment();
7
-
8
- extern VALUE cNokogiriXmlComment;
9
- #endif
@@ -1,23 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DOCUMENT
2
- #define NOKOGIRI_XML_DOCUMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- struct _nokogiriTuple {
7
- VALUE doc;
8
- st_table *unlinkedNodes;
9
- VALUE node_cache;
10
- };
11
- typedef struct _nokogiriTuple nokogiriTuple;
12
- typedef nokogiriTuple * nokogiriTuplePtr;
13
-
14
- void init_xml_document();
15
- VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc);
16
-
17
- #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
18
- #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)
19
- #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
20
- #define DOC_NODE_CACHE(x) (((nokogiriTuplePtr)(x->_private))->node_cache)
21
-
22
- extern VALUE cNokogiriXmlDocument ;
23
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DOCUMENT_FRAGMENT
2
- #define NOKOGIRI_XML_DOCUMENT_FRAGMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_document_fragment();
7
-
8
- extern VALUE cNokogiriXmlDocumentFragment;
9
- #endif
10
-
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DTD
2
- #define NOKOGIRI_XML_DTD
3
-
4
- #include <nokogiri.h>
5
-
6
- extern VALUE cNokogiriXmlDtd;
7
-
8
- void init_xml_dtd();
9
-
10
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ELEMENT_CONTENT
2
- #define NOKOGIRI_XML_ELEMENT_CONTENT
3
-
4
- #include <nokogiri.h>
5
-
6
-
7
- VALUE Nokogiri_wrap_element_content(VALUE doc, xmlElementContentPtr element);
8
- void init_xml_element_content();
9
-
10
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ELEMENT_DECL
2
- #define NOKOGIRI_XML_ELEMENT_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_element_decl();
7
-
8
- extern VALUE cNokogiriXmlElementDecl;
9
- #endif
@@ -1,8 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENCODING_HANDLER
2
- #define NOKOGIRI_XML_ENCODING_HANDLER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_encoding_handler();
7
-
8
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENTITY_DECL
2
- #define NOKOGIRI_XML_ENTITY_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_entity_decl();
7
-
8
- extern VALUE cNokogiriXmlEntityDecl;
9
- #endif
10
-
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENTITY_REFERENCE
2
- #define NOKOGIRI_XML_ENTITY_REFERENCE
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_entity_reference();
7
-
8
- extern VALUE cNokogiriXmlEntityReference;
9
- #endif
@@ -1,60 +0,0 @@
1
- #include <xml_io.h>
2
-
3
- static ID id_read, id_write;
4
-
5
- VALUE read_check(VALUE *args) {
6
- return rb_funcall(args[0], id_read, 1, args[1]);
7
- }
8
-
9
- VALUE read_failed(void) {
10
- return Qundef;
11
- }
12
-
13
- int io_read_callback(void * ctx, char * buffer, int len) {
14
- VALUE string, args[2];
15
- size_t str_len, safe_len;
16
-
17
- args[0] = (VALUE)ctx;
18
- args[1] = INT2NUM(len);
19
-
20
- string = rb_rescue(read_check, (VALUE)args, read_failed, 0);
21
-
22
- if (NIL_P(string)) return 0;
23
- if (string == Qundef) return -1;
24
-
25
- str_len = (size_t)RSTRING_LEN(string);
26
- safe_len = str_len > (size_t)len ? (size_t)len : str_len;
27
- memcpy(buffer, StringValuePtr(string), safe_len);
28
-
29
- return (int)safe_len;
30
- }
31
-
32
- VALUE write_check(VALUE *args) {
33
- return rb_funcall(args[0], id_write, 1, args[1]);
34
- }
35
-
36
- VALUE write_failed(void) {
37
- return Qundef;
38
- }
39
-
40
- int io_write_callback(void * ctx, char * buffer, int len) {
41
- VALUE args[2], size;
42
-
43
- args[0] = (VALUE)ctx;
44
- args[1] = rb_str_new(buffer, (long)len);
45
-
46
- size = rb_rescue(write_check, (VALUE)args, write_failed, 0);
47
-
48
- if (size == Qundef) return -1;
49
-
50
- return NUM2INT(size);
51
- }
52
-
53
- int io_close_callback(void * ctx) {
54
- return 0;
55
- }
56
-
57
- void init_nokogiri_io() {
58
- id_read = rb_intern("read");
59
- id_write = rb_intern("write");
60
- }
@@ -1,11 +0,0 @@
1
- #ifndef NOKOGIRI_XML_IO
2
- #define NOKOGIRI_XML_IO
3
-
4
- #include <nokogiri.h>
5
-
6
- int io_read_callback(void * ctx, char * buffer, int len);
7
- int io_write_callback(void * ctx, char * buffer, int len);
8
- int io_close_callback(void * ctx);
9
- void init_nokogiri_io();
10
-
11
- #endif
@@ -1,112 +0,0 @@
1
- #ifndef HAVE_XMLFIRSTELEMENTCHILD
2
- #include <libxml/tree.h>
3
- /**
4
- * xmlFirstElementChild:
5
- * @parent: the parent node
6
- *
7
- * Finds the first child node of that element which is a Element node
8
- * Note the handling of entities references is different than in
9
- * the W3C DOM element traversal spec since we don't have back reference
10
- * from entities content to entities references.
11
- *
12
- * Returns the first element child or NULL if not available
13
- */
14
- xmlNodePtr
15
- xmlFirstElementChild(xmlNodePtr parent) {
16
- xmlNodePtr cur = NULL;
17
-
18
- if (parent == NULL)
19
- return(NULL);
20
- switch (parent->type) {
21
- case XML_ELEMENT_NODE:
22
- case XML_ENTITY_NODE:
23
- case XML_DOCUMENT_NODE:
24
- case XML_HTML_DOCUMENT_NODE:
25
- cur = parent->children;
26
- break;
27
- default:
28
- return(NULL);
29
- }
30
- while (cur != NULL) {
31
- if (cur->type == XML_ELEMENT_NODE)
32
- return(cur);
33
- cur = cur->next;
34
- }
35
- return(NULL);
36
- }
37
-
38
- /**
39
- * xmlNextElementSibling:
40
- * @node: the current node
41
- *
42
- * Finds the first closest next sibling of the node which is an
43
- * element node.
44
- * Note the handling of entities references is different than in
45
- * the W3C DOM element traversal spec since we don't have back reference
46
- * from entities content to entities references.
47
- *
48
- * Returns the next element sibling or NULL if not available
49
- */
50
- xmlNodePtr
51
- xmlNextElementSibling(xmlNodePtr node) {
52
- if (node == NULL)
53
- return(NULL);
54
- switch (node->type) {
55
- case XML_ELEMENT_NODE:
56
- case XML_TEXT_NODE:
57
- case XML_CDATA_SECTION_NODE:
58
- case XML_ENTITY_REF_NODE:
59
- case XML_ENTITY_NODE:
60
- case XML_PI_NODE:
61
- case XML_COMMENT_NODE:
62
- case XML_DTD_NODE:
63
- case XML_XINCLUDE_START:
64
- case XML_XINCLUDE_END:
65
- node = node->next;
66
- break;
67
- default:
68
- return(NULL);
69
- }
70
- while (node != NULL) {
71
- if (node->type == XML_ELEMENT_NODE)
72
- return(node);
73
- node = node->next;
74
- }
75
- return(NULL);
76
- }
77
-
78
- /**
79
- * xmlLastElementChild:
80
- * @parent: the parent node
81
- *
82
- * Finds the last child node of that element which is a Element node
83
- * Note the handling of entities references is different than in
84
- * the W3C DOM element traversal spec since we don't have back reference
85
- * from entities content to entities references.
86
- *
87
- * Returns the last element child or NULL if not available
88
- */
89
- xmlNodePtr
90
- xmlLastElementChild(xmlNodePtr parent) {
91
- xmlNodePtr cur = NULL;
92
-
93
- if (parent == NULL)
94
- return(NULL);
95
- switch (parent->type) {
96
- case XML_ELEMENT_NODE:
97
- case XML_ENTITY_NODE:
98
- case XML_DOCUMENT_NODE:
99
- case XML_HTML_DOCUMENT_NODE:
100
- cur = parent->last;
101
- break;
102
- default:
103
- return(NULL);
104
- }
105
- while (cur != NULL) {
106
- if (cur->type == XML_ELEMENT_NODE)
107
- return(cur);
108
- cur = cur->prev;
109
- }
110
- return(NULL);
111
- }
112
- #endif
@@ -1,12 +0,0 @@
1
- #ifndef HAVE_XMLFIRSTELEMENTCHILD
2
-
3
- #ifndef XML_LIBXML2_HACKS
4
- #define XML_LIBXML2_HACKS
5
-
6
- xmlNodePtr xmlFirstElementChild(xmlNodePtr parent);
7
- xmlNodePtr xmlNextElementSibling(xmlNodePtr node);
8
- xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
9
-
10
- #endif
11
-
12
- #endif
@@ -1,13 +0,0 @@
1
- #ifndef NOKOGIRI_XML_NAMESPACE
2
- #define NOKOGIRI_XML_NAMESPACE
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_namespace();
7
-
8
- extern VALUE cNokogiriXmlNamespace ;
9
-
10
- VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node) ;
11
- VALUE Nokogiri_wrap_xml_namespace2(VALUE document, xmlNsPtr node) ;
12
-
13
- #endif
@@ -1,13 +0,0 @@
1
- #ifndef NOKOGIRI_XML_NODE
2
- #define NOKOGIRI_XML_NODE
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_node();
7
-
8
- extern VALUE cNokogiriXmlNode ;
9
- extern VALUE cNokogiriXmlElement ;
10
-
11
- VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node) ;
12
- void Nokogiri_xml_node_properties(xmlNodePtr node, VALUE attr_hash) ;
13
- #endif
@@ -1,13 +0,0 @@
1
- #ifndef NOKOGIRI_XML_NODE_SET
2
- #define NOKOGIRI_XML_NODE_SET
3
-
4
- #include <nokogiri.h>
5
- void init_xml_node_set();
6
-
7
- extern VALUE cNokogiriXmlNodeSet ;
8
- VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set, VALUE document) ;
9
- VALUE Nokogiri_wrap_xml_node_set_node(xmlNodePtr node, VALUE node_set) ;
10
- VALUE Nokogiri_wrap_xml_node_set_namespace(xmlNsPtr node, VALUE node_set) ;
11
- int Nokogiri_namespace_eh(xmlNodePtr node) ;
12
-
13
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_PROCESSING_INSTRUCTION
2
- #define NOKOGIRI_XML_PROCESSING_INSTRUCTION
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_processing_instruction();
7
-
8
- extern VALUE cNokogiriXmlProcessingInstruction;
9
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_READER
2
- #define NOKOGIRI_XML_READER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_reader();
7
-
8
- extern VALUE cNokogiriXmlReader;
9
-
10
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_RELAX_NG
2
- #define NOKOGIRI_XML_RELAX_NG
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_relax_ng();
7
-
8
- extern VALUE cNokogiriXmlRelaxNG;
9
- #endif
@@ -1,39 +0,0 @@
1
- #ifndef NOKOGIRI_XML_SAX_PARSER
2
- #define NOKOGIRI_XML_SAX_PARSER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_sax_parser();
7
-
8
- extern VALUE cNokogiriXmlSaxParser ;
9
-
10
- typedef struct _nokogiriSAXTuple {
11
- xmlParserCtxtPtr ctxt;
12
- VALUE self;
13
- } nokogiriSAXTuple;
14
-
15
- typedef nokogiriSAXTuple * nokogiriSAXTuplePtr;
16
-
17
- #define NOKOGIRI_SAX_SELF(_ctxt) \
18
- ((nokogiriSAXTuplePtr)(_ctxt))->self
19
-
20
- #define NOKOGIRI_SAX_CTXT(_ctxt) \
21
- ((nokogiriSAXTuplePtr)(_ctxt))->ctxt
22
-
23
- #define NOKOGIRI_SAX_TUPLE_NEW(_ctxt, _self) \
24
- nokogiri_sax_tuple_new(_ctxt, _self)
25
-
26
- static inline nokogiriSAXTuplePtr
27
- nokogiri_sax_tuple_new(xmlParserCtxtPtr ctxt, VALUE self)
28
- {
29
- nokogiriSAXTuplePtr tuple = malloc(sizeof(nokogiriSAXTuple));
30
- tuple->self = self;
31
- tuple->ctxt = ctxt;
32
- return tuple;
33
- }
34
-
35
- #define NOKOGIRI_SAX_TUPLE_DESTROY(_tuple) \
36
- free(_tuple) \
37
-
38
- #endif
39
-
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_SAX_PARSER_CONTEXT
2
- #define NOKOGIRI_XML_SAX_PARSER_CONTEXT
3
-
4
- #include <nokogiri.h>
5
-
6
- extern VALUE cNokogiriXmlSaxParserContext;
7
-
8
- void init_xml_sax_parser_context();
9
-
10
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_SAX_PUSH_PARSER
2
- #define NOKOGIRI_XML_SAX_PUSH_PARSER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_sax_push_parser();
7
-
8
- extern VALUE cNokogiriXmlSaxPushParser ;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_SCHEMA
2
- #define NOKOGIRI_XML_SCHEMA
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_schema();
7
-
8
- extern VALUE cNokogiriXmlSchema;
9
- #endif
@@ -1,13 +0,0 @@
1
- #ifndef NOKOGIRI_XML_SYNTAX_ERROR
2
- #define NOKOGIRI_XML_SYNTAX_ERROR
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_syntax_error();
7
- VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error);
8
- void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error);
9
- NORETURN(void Nokogiri_error_raise(void * ctx, xmlErrorPtr error));
10
-
11
- extern VALUE cNokogiriXmlSyntaxError;
12
- #endif
13
-
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_TEXT
2
- #define NOKOGIRI_XML_TEXT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_text();
7
-
8
- extern VALUE cNokogiriXmlText ;
9
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_XPATH_CONTEXT
2
- #define NOKOGIRI_XML_XPATH_CONTEXT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_xpath_context();
7
- void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler, const char* function_name) ;
8
-
9
- extern VALUE cNokogiriXmlXpathContext;
10
- #endif
@@ -1,14 +0,0 @@
1
- #ifndef NOKOGIRI_XSLT_STYLESHEET
2
- #define NOKOGIRI_XSLT_STYLESHEET
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xslt_stylesheet();
7
-
8
- extern VALUE cNokogiriXsltStylesheet ;
9
-
10
- typedef struct _nokogiriXsltStylesheetTuple {
11
- xsltStylesheetPtr ss;
12
- VALUE func_instances;
13
- } nokogiriXsltStylesheetTuple;
14
- #endif
@@ -1,39 +0,0 @@
1
- module Nokogiri
2
- module HTML
3
- class DocumentFragment < Nokogiri::XML::DocumentFragment
4
- ####
5
- # Create a Nokogiri::XML::DocumentFragment from +tags+, using +encoding+
6
- def self.parse tags, encoding = nil
7
- doc = HTML::Document.new
8
-
9
- encoding ||= tags.respond_to?(:encoding) ? tags.encoding.name : 'UTF-8'
10
- doc.encoding = encoding
11
-
12
- new(doc, tags)
13
- end
14
-
15
- def initialize document, tags = nil, ctx = nil
16
- return self unless tags
17
-
18
- if ctx
19
- preexisting_errors = document.errors.dup
20
- node_set = ctx.parse("<div>#{tags}</div>")
21
- node_set.first.children.each { |child| child.parent = self } unless node_set.empty?
22
- self.errors = document.errors - preexisting_errors
23
- else
24
- # This is a horrible hack, but I don't care
25
- if tags.strip =~ /^<body/i
26
- path = "/html/body"
27
- else
28
- path = "/html/body/node()"
29
- end
30
-
31
- temp_doc = HTML::Document.parse "<html><body>#{tags}", nil, document.encoding
32
- temp_doc.xpath(path).each { |child| child.parent = self }
33
- self.errors = temp_doc.errors
34
- end
35
- children
36
- end
37
- end
38
- end
39
- end