nokogiri 1.8.5 → 1.13.9

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 (353) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +3 -21
  3. data/LICENSE-DEPENDENCIES.md +1159 -868
  4. data/LICENSE.md +5 -28
  5. data/README.md +196 -90
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +13 -59
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +765 -420
  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 +119 -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 -91
  18. data/ext/nokogiri/nokogiri.h +199 -88
  19. data/ext/nokogiri/test_global_handlers.c +40 -0
  20. data/ext/nokogiri/xml_attr.c +42 -37
  21. data/ext/nokogiri/xml_attribute_decl.c +21 -21
  22. data/ext/nokogiri/xml_cdata.c +14 -19
  23. data/ext/nokogiri/xml_comment.c +19 -26
  24. data/ext/nokogiri/xml_document.c +296 -217
  25. data/ext/nokogiri/xml_document_fragment.c +12 -16
  26. data/ext/nokogiri/xml_dtd.c +64 -58
  27. data/ext/nokogiri/xml_element_content.c +31 -26
  28. data/ext/nokogiri/xml_element_decl.c +25 -25
  29. data/ext/nokogiri/xml_encoding_handler.c +43 -18
  30. data/ext/nokogiri/xml_entity_decl.c +37 -35
  31. data/ext/nokogiri/xml_entity_reference.c +16 -18
  32. data/ext/nokogiri/xml_namespace.c +99 -54
  33. data/ext/nokogiri/xml_node.c +1107 -658
  34. data/ext/nokogiri/xml_node_set.c +178 -166
  35. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  36. data/ext/nokogiri/xml_reader.c +277 -175
  37. data/ext/nokogiri/xml_relax_ng.c +52 -28
  38. data/ext/nokogiri/xml_sax_parser.c +112 -112
  39. data/ext/nokogiri/xml_sax_parser_context.c +112 -86
  40. data/ext/nokogiri/xml_sax_push_parser.c +36 -27
  41. data/ext/nokogiri/xml_schema.c +114 -35
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +14 -18
  44. data/ext/nokogiri/xml_xpath_context.c +226 -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 +397 -377
  94. data/lib/nokogiri/css/parser.y +250 -245
  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 +3 -2
  99. data/lib/nokogiri/css/xpath_visitor.rb +218 -91
  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 +103 -105
  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 +17 -16
  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 -108
  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 +224 -86
  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 +2 -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 +895 -377
  139. data/lib/nokogiri/xml/node_set.rb +92 -65
  140. data/lib/nokogiri/xml/notation.rb +13 -0
  141. data/lib/nokogiri/xml/parse_options.rb +22 -8
  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 +21 -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 +38 -34
  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 +112 -72
  155. data/lib/nokogiri/xml/syntax_error.rb +6 -4
  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 +38 -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/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  169. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  170. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
  171. data/ports/archives/libxml2-2.10.3.tar.xz +0 -0
  172. data/ports/archives/libxslt-1.1.37.tar.xz +0 -0
  173. metadata +211 -266
  174. data/.autotest +0 -22
  175. data/.cross_rubies +0 -8
  176. data/.editorconfig +0 -17
  177. data/.gemtest +0 -0
  178. data/.travis.yml +0 -63
  179. data/CHANGELOG.md +0 -1368
  180. data/CONTRIBUTING.md +0 -42
  181. data/C_CODING_STYLE.rdoc +0 -33
  182. data/Gemfile-libxml-ruby +0 -3
  183. data/Manifest.txt +0 -370
  184. data/ROADMAP.md +0 -111
  185. data/Rakefile +0 -348
  186. data/SECURITY.md +0 -19
  187. data/STANDARD_RESPONSES.md +0 -47
  188. data/Y_U_NO_GEMSPEC.md +0 -155
  189. data/appveyor.yml +0 -29
  190. data/build_all +0 -44
  191. data/ext/nokogiri/html_document.c +0 -170
  192. data/ext/nokogiri/html_document.h +0 -10
  193. data/ext/nokogiri/html_element_description.c +0 -279
  194. data/ext/nokogiri/html_element_description.h +0 -10
  195. data/ext/nokogiri/html_entity_lookup.c +0 -32
  196. data/ext/nokogiri/html_entity_lookup.h +0 -8
  197. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  198. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  199. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  200. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  201. data/ext/nokogiri/xml_attr.h +0 -9
  202. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  203. data/ext/nokogiri/xml_cdata.h +0 -9
  204. data/ext/nokogiri/xml_comment.h +0 -9
  205. data/ext/nokogiri/xml_document.h +0 -23
  206. data/ext/nokogiri/xml_document_fragment.h +0 -10
  207. data/ext/nokogiri/xml_dtd.h +0 -10
  208. data/ext/nokogiri/xml_element_content.h +0 -10
  209. data/ext/nokogiri/xml_element_decl.h +0 -9
  210. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  211. data/ext/nokogiri/xml_entity_decl.h +0 -10
  212. data/ext/nokogiri/xml_entity_reference.h +0 -9
  213. data/ext/nokogiri/xml_io.c +0 -61
  214. data/ext/nokogiri/xml_io.h +0 -11
  215. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  216. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  217. data/ext/nokogiri/xml_namespace.h +0 -15
  218. data/ext/nokogiri/xml_node.h +0 -13
  219. data/ext/nokogiri/xml_node_set.h +0 -12
  220. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  221. data/ext/nokogiri/xml_reader.h +0 -10
  222. data/ext/nokogiri/xml_relax_ng.h +0 -9
  223. data/ext/nokogiri/xml_sax_parser.h +0 -39
  224. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  225. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  226. data/ext/nokogiri/xml_schema.h +0 -9
  227. data/ext/nokogiri/xml_syntax_error.h +0 -13
  228. data/ext/nokogiri/xml_text.h +0 -9
  229. data/ext/nokogiri/xml_xpath_context.h +0 -10
  230. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  231. data/lib/nokogiri/html/document_fragment.rb +0 -49
  232. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  233. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  234. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  235. data/patches/libxml2/0002-Fix-nullptr-deref-with-XPath-logic-ops.patch +0 -54
  236. data/patches/libxml2/0003-Fix-infinite-loop-in-LZMA-decompression.patch +0 -50
  237. data/patches/sort-patches-by-date +0 -25
  238. data/ports/archives/libxml2-2.9.8.tar.gz +0 -0
  239. data/ports/archives/libxslt-1.1.32.tar.gz +0 -0
  240. data/suppressions/README.txt +0 -1
  241. data/suppressions/nokogiri_ruby-2.supp +0 -10
  242. data/tasks/test.rb +0 -100
  243. data/test/css/test_nthiness.rb +0 -226
  244. data/test/css/test_parser.rb +0 -386
  245. data/test/css/test_tokenizer.rb +0 -215
  246. data/test/css/test_xpath_visitor.rb +0 -96
  247. data/test/decorators/test_slop.rb +0 -23
  248. data/test/files/2ch.html +0 -108
  249. data/test/files/GH_1042.html +0 -18
  250. data/test/files/address_book.rlx +0 -12
  251. data/test/files/address_book.xml +0 -10
  252. data/test/files/atom.xml +0 -344
  253. data/test/files/bar/bar.xsd +0 -4
  254. data/test/files/bogus.xml +0 -0
  255. data/test/files/dont_hurt_em_why.xml +0 -422
  256. data/test/files/encoding.html +0 -82
  257. data/test/files/encoding.xhtml +0 -84
  258. data/test/files/exslt.xml +0 -8
  259. data/test/files/exslt.xslt +0 -35
  260. data/test/files/foo/foo.xsd +0 -4
  261. data/test/files/metacharset.html +0 -10
  262. data/test/files/namespace_pressure_test.xml +0 -1684
  263. data/test/files/noencoding.html +0 -47
  264. data/test/files/po.xml +0 -32
  265. data/test/files/po.xsd +0 -66
  266. data/test/files/saml/saml20assertion_schema.xsd +0 -283
  267. data/test/files/saml/saml20protocol_schema.xsd +0 -302
  268. data/test/files/saml/xenc_schema.xsd +0 -146
  269. data/test/files/saml/xmldsig_schema.xsd +0 -318
  270. data/test/files/shift_jis.html +0 -10
  271. data/test/files/shift_jis.xml +0 -5
  272. data/test/files/shift_jis_no_charset.html +0 -9
  273. data/test/files/slow-xpath.xml +0 -25509
  274. data/test/files/snuggles.xml +0 -3
  275. data/test/files/staff.dtd +0 -10
  276. data/test/files/staff.xml +0 -59
  277. data/test/files/staff.xslt +0 -32
  278. data/test/files/test_document_url/bar.xml +0 -2
  279. data/test/files/test_document_url/document.dtd +0 -4
  280. data/test/files/test_document_url/document.xml +0 -6
  281. data/test/files/tlm.html +0 -851
  282. data/test/files/to_be_xincluded.xml +0 -2
  283. data/test/files/valid_bar.xml +0 -2
  284. data/test/files/xinclude.xml +0 -4
  285. data/test/helper.rb +0 -271
  286. data/test/html/sax/test_parser.rb +0 -168
  287. data/test/html/sax/test_parser_context.rb +0 -46
  288. data/test/html/sax/test_parser_text.rb +0 -163
  289. data/test/html/sax/test_push_parser.rb +0 -87
  290. data/test/html/test_attributes.rb +0 -85
  291. data/test/html/test_builder.rb +0 -164
  292. data/test/html/test_document.rb +0 -712
  293. data/test/html/test_document_encoding.rb +0 -143
  294. data/test/html/test_document_fragment.rb +0 -310
  295. data/test/html/test_element_description.rb +0 -105
  296. data/test/html/test_named_characters.rb +0 -14
  297. data/test/html/test_node.rb +0 -212
  298. data/test/html/test_node_encoding.rb +0 -91
  299. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
  300. data/test/namespaces/test_namespaces_aliased_default.rb +0 -24
  301. data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
  302. data/test/namespaces/test_namespaces_in_cloned_doc.rb +0 -31
  303. data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
  304. data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -80
  305. data/test/namespaces/test_namespaces_preservation.rb +0 -31
  306. data/test/test_convert_xpath.rb +0 -135
  307. data/test/test_css_cache.rb +0 -47
  308. data/test/test_encoding_handler.rb +0 -48
  309. data/test/test_memory_leak.rb +0 -156
  310. data/test/test_nokogiri.rb +0 -138
  311. data/test/test_soap4r_sax.rb +0 -52
  312. data/test/test_xslt_transforms.rb +0 -314
  313. data/test/xml/node/test_save_options.rb +0 -28
  314. data/test/xml/node/test_subclass.rb +0 -44
  315. data/test/xml/sax/test_parser.rb +0 -402
  316. data/test/xml/sax/test_parser_context.rb +0 -115
  317. data/test/xml/sax/test_parser_text.rb +0 -202
  318. data/test/xml/sax/test_push_parser.rb +0 -265
  319. data/test/xml/test_attr.rb +0 -74
  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 -54
  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 -298
  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 -262
  334. data/test/xml/test_namespace.rb +0 -96
  335. data/test/xml/test_node.rb +0 -1325
  336. data/test/xml/test_node_attributes.rb +0 -115
  337. data/test/xml/test_node_encoding.rb +0 -75
  338. data/test/xml/test_node_inheritance.rb +0 -32
  339. data/test/xml/test_node_reparenting.rb +0 -592
  340. data/test/xml/test_node_set.rb +0 -809
  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 -620
  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 -36
  348. data/test/xml/test_text.rb +0 -60
  349. data/test/xml/test_unparented_node.rb +0 -483
  350. data/test/xml/test_xinclude.rb +0 -83
  351. data/test/xml/test_xpath.rb +0 -470
  352. data/test/xslt/test_custom_functions.rb +0 -133
  353. data/test/xslt/test_exception_handling.rb +0 -37
@@ -0,0 +1,166 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriHtml4Document ;
4
+
5
+ static ID id_encoding_found;
6
+ static ID id_to_s;
7
+
8
+ /*
9
+ * call-seq:
10
+ * new
11
+ *
12
+ * Create a new document
13
+ */
14
+ static VALUE
15
+ rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
16
+ {
17
+ VALUE uri, external_id, rest, rb_doc;
18
+ htmlDocPtr doc;
19
+
20
+ rb_scan_args(argc, argv, "0*", &rest);
21
+ uri = rb_ary_entry(rest, (long)0);
22
+ external_id = rb_ary_entry(rest, (long)1);
23
+
24
+ doc = htmlNewDoc(
25
+ RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
26
+ RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
27
+ );
28
+ rb_doc = noko_xml_document_wrap_with_init_args(klass, doc, argc, argv);
29
+ return rb_doc ;
30
+ }
31
+
32
+ /*
33
+ * call-seq:
34
+ * read_io(io, url, encoding, options)
35
+ *
36
+ * Read the HTML document from +io+ with given +url+, +encoding+,
37
+ * and +options+. See Nokogiri::HTML4.parse
38
+ */
39
+ static VALUE
40
+ rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
41
+ {
42
+ VALUE rb_doc;
43
+ VALUE rb_error_list = rb_ary_new();
44
+ htmlDocPtr c_doc;
45
+ const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
46
+ const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
47
+ int options = NUM2INT(rb_options);
48
+
49
+ xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
50
+
51
+ c_doc = htmlReadIO(noko_io_read, noko_io_close, (void *)rb_io, c_url, c_encoding, options);
52
+
53
+ xmlSetStructuredErrorFunc(NULL, NULL);
54
+
55
+ /*
56
+ * If EncodingFound has occurred in EncodingReader, make sure to do
57
+ * a cleanup and propagate the error.
58
+ */
59
+ if (rb_respond_to(rb_io, id_encoding_found)) {
60
+ VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
61
+ if (!NIL_P(encoding_found)) {
62
+ xmlFreeDoc(c_doc);
63
+ rb_exc_raise(encoding_found);
64
+ }
65
+ }
66
+
67
+ if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
68
+ VALUE rb_error ;
69
+
70
+ xmlFreeDoc(c_doc);
71
+
72
+ rb_error = rb_ary_entry(rb_error_list, 0);
73
+ if (rb_error == Qnil) {
74
+ rb_raise(rb_eRuntimeError, "Could not parse document");
75
+ } else {
76
+ VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
77
+ exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
78
+ exception_message);
79
+ rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
80
+ }
81
+
82
+ return Qnil;
83
+ }
84
+
85
+ rb_doc = noko_xml_document_wrap(klass, c_doc);
86
+ rb_iv_set(rb_doc, "@errors", rb_error_list);
87
+ return rb_doc;
88
+ }
89
+
90
+ /*
91
+ * call-seq:
92
+ * read_memory(string, url, encoding, options)
93
+ *
94
+ * Read the HTML document contained in +string+ with given +url+, +encoding+,
95
+ * and +options+. See Nokogiri::HTML4.parse
96
+ */
97
+ static VALUE
98
+ rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
99
+ {
100
+ VALUE rb_doc;
101
+ VALUE rb_error_list = rb_ary_new();
102
+ htmlDocPtr c_doc;
103
+ const char *c_buffer = StringValuePtr(rb_html);
104
+ const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
105
+ const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
106
+ int html_len = (int)RSTRING_LEN(rb_html);
107
+ int options = NUM2INT(rb_options);
108
+
109
+ xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
110
+
111
+ c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
112
+
113
+ xmlSetStructuredErrorFunc(NULL, NULL);
114
+
115
+ if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
116
+ VALUE rb_error ;
117
+
118
+ xmlFreeDoc(c_doc);
119
+
120
+ rb_error = rb_ary_entry(rb_error_list, 0);
121
+ if (rb_error == Qnil) {
122
+ rb_raise(rb_eRuntimeError, "Could not parse document");
123
+ } else {
124
+ VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
125
+ exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
126
+ exception_message);
127
+ rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
128
+ }
129
+
130
+ return Qnil;
131
+ }
132
+
133
+ rb_doc = noko_xml_document_wrap(klass, c_doc);
134
+ rb_iv_set(rb_doc, "@errors", rb_error_list);
135
+ return rb_doc;
136
+ }
137
+
138
+ /*
139
+ * call-seq:
140
+ * type
141
+ *
142
+ * The type for this document
143
+ */
144
+ static VALUE
145
+ rb_html_document_type(VALUE self)
146
+ {
147
+ htmlDocPtr doc;
148
+ Data_Get_Struct(self, xmlDoc, doc);
149
+ return INT2NUM((long)doc->type);
150
+ }
151
+
152
+ void
153
+ noko_init_html_document()
154
+ {
155
+ assert(cNokogiriXmlDocument);
156
+ cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);
157
+
158
+ rb_define_singleton_method(cNokogiriHtml4Document, "read_memory", rb_html_document_s_read_memory, 4);
159
+ rb_define_singleton_method(cNokogiriHtml4Document, "read_io", rb_html_document_s_read_io, 4);
160
+ rb_define_singleton_method(cNokogiriHtml4Document, "new", rb_html_document_s_new, -1);
161
+
162
+ rb_define_method(cNokogiriHtml4Document, "type", rb_html_document_type, 0);
163
+
164
+ id_encoding_found = rb_intern("encoding_found");
165
+ id_to_s = rb_intern("to_s");
166
+ }
@@ -0,0 +1,294 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriHtml4ElementDescription ;
4
+
5
+ /*
6
+ * call-seq:
7
+ * required_attributes
8
+ *
9
+ * A list of required attributes for this element
10
+ */
11
+ static VALUE
12
+ required_attributes(VALUE self)
13
+ {
14
+ const htmlElemDesc *description;
15
+ VALUE list;
16
+ int i;
17
+
18
+ Data_Get_Struct(self, htmlElemDesc, description);
19
+
20
+ list = rb_ary_new();
21
+
22
+ if (NULL == description->attrs_req) { return list; }
23
+
24
+ for (i = 0; description->attrs_depr[i]; i++) {
25
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
26
+ }
27
+
28
+ return list;
29
+ }
30
+
31
+ /*
32
+ * call-seq:
33
+ * deprecated_attributes
34
+ *
35
+ * A list of deprecated attributes for this element
36
+ */
37
+ static VALUE
38
+ deprecated_attributes(VALUE self)
39
+ {
40
+ const htmlElemDesc *description;
41
+ VALUE list;
42
+ int i;
43
+
44
+ Data_Get_Struct(self, htmlElemDesc, description);
45
+
46
+ list = rb_ary_new();
47
+
48
+ if (NULL == description->attrs_depr) { return list; }
49
+
50
+ for (i = 0; description->attrs_depr[i]; i++) {
51
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
52
+ }
53
+
54
+ return list;
55
+ }
56
+
57
+ /*
58
+ * call-seq:
59
+ * optional_attributes
60
+ *
61
+ * A list of optional attributes for this element
62
+ */
63
+ static VALUE
64
+ optional_attributes(VALUE self)
65
+ {
66
+ const htmlElemDesc *description;
67
+ VALUE list;
68
+ int i;
69
+
70
+ Data_Get_Struct(self, htmlElemDesc, description);
71
+
72
+ list = rb_ary_new();
73
+
74
+ if (NULL == description->attrs_opt) { return list; }
75
+
76
+ for (i = 0; description->attrs_opt[i]; i++) {
77
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
78
+ }
79
+
80
+ return list;
81
+ }
82
+
83
+ /*
84
+ * call-seq:
85
+ * default_sub_element
86
+ *
87
+ * The default sub element for this element
88
+ */
89
+ static VALUE
90
+ default_sub_element(VALUE self)
91
+ {
92
+ const htmlElemDesc *description;
93
+ Data_Get_Struct(self, htmlElemDesc, description);
94
+
95
+ if (description->defaultsubelt) {
96
+ return NOKOGIRI_STR_NEW2(description->defaultsubelt);
97
+ }
98
+
99
+ return Qnil;
100
+ }
101
+
102
+ /*
103
+ * call-seq:
104
+ * sub_elements
105
+ *
106
+ * A list of allowed sub elements for this element.
107
+ */
108
+ static VALUE
109
+ sub_elements(VALUE self)
110
+ {
111
+ const htmlElemDesc *description;
112
+ VALUE list;
113
+ int i;
114
+
115
+ Data_Get_Struct(self, htmlElemDesc, description);
116
+
117
+ list = rb_ary_new();
118
+
119
+ if (NULL == description->subelts) { return list; }
120
+
121
+ for (i = 0; description->subelts[i]; i++) {
122
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
123
+ }
124
+
125
+ return list;
126
+ }
127
+
128
+ /*
129
+ * call-seq:
130
+ * description
131
+ *
132
+ * The description for this element
133
+ */
134
+ static VALUE
135
+ description(VALUE self)
136
+ {
137
+ const htmlElemDesc *description;
138
+ Data_Get_Struct(self, htmlElemDesc, description);
139
+
140
+ return NOKOGIRI_STR_NEW2(description->desc);
141
+ }
142
+
143
+ /*
144
+ * call-seq:
145
+ * inline?
146
+ *
147
+ * Is this element an inline element?
148
+ */
149
+ static VALUE
150
+ inline_eh(VALUE self)
151
+ {
152
+ const htmlElemDesc *description;
153
+ Data_Get_Struct(self, htmlElemDesc, description);
154
+
155
+ if (description->isinline) { return Qtrue; }
156
+ return Qfalse;
157
+ }
158
+
159
+ /*
160
+ * call-seq:
161
+ * deprecated?
162
+ *
163
+ * Is this element deprecated?
164
+ */
165
+ static VALUE
166
+ deprecated_eh(VALUE self)
167
+ {
168
+ const htmlElemDesc *description;
169
+ Data_Get_Struct(self, htmlElemDesc, description);
170
+
171
+ if (description->depr) { return Qtrue; }
172
+ return Qfalse;
173
+ }
174
+
175
+ /*
176
+ * call-seq:
177
+ * empty?
178
+ *
179
+ * Is this an empty element?
180
+ */
181
+ static VALUE
182
+ empty_eh(VALUE self)
183
+ {
184
+ const htmlElemDesc *description;
185
+ Data_Get_Struct(self, htmlElemDesc, description);
186
+
187
+ if (description->empty) { return Qtrue; }
188
+ return Qfalse;
189
+ }
190
+
191
+ /*
192
+ * call-seq:
193
+ * save_end_tag?
194
+ *
195
+ * Should the end tag be saved?
196
+ */
197
+ static VALUE
198
+ save_end_tag_eh(VALUE self)
199
+ {
200
+ const htmlElemDesc *description;
201
+ Data_Get_Struct(self, htmlElemDesc, description);
202
+
203
+ if (description->saveEndTag) { return Qtrue; }
204
+ return Qfalse;
205
+ }
206
+
207
+ /*
208
+ * call-seq:
209
+ * implied_end_tag?
210
+ *
211
+ * Can the end tag be implied for this tag?
212
+ */
213
+ static VALUE
214
+ implied_end_tag_eh(VALUE self)
215
+ {
216
+ const htmlElemDesc *description;
217
+ Data_Get_Struct(self, htmlElemDesc, description);
218
+
219
+ if (description->endTag) { return Qtrue; }
220
+ return Qfalse;
221
+ }
222
+
223
+ /*
224
+ * call-seq:
225
+ * implied_start_tag?
226
+ *
227
+ * Can the start tag be implied for this tag?
228
+ */
229
+ static VALUE
230
+ implied_start_tag_eh(VALUE self)
231
+ {
232
+ const htmlElemDesc *description;
233
+ Data_Get_Struct(self, htmlElemDesc, description);
234
+
235
+ if (description->startTag) { return Qtrue; }
236
+ return Qfalse;
237
+ }
238
+
239
+ /*
240
+ * call-seq:
241
+ * name
242
+ *
243
+ * Get the tag name for this ElemementDescription
244
+ */
245
+ static VALUE
246
+ name(VALUE self)
247
+ {
248
+ const htmlElemDesc *description;
249
+ Data_Get_Struct(self, htmlElemDesc, description);
250
+
251
+ if (NULL == description->name) { return Qnil; }
252
+ return NOKOGIRI_STR_NEW2(description->name);
253
+ }
254
+
255
+ /*
256
+ * call-seq:
257
+ * [](tag_name)
258
+ *
259
+ * Get ElemementDescription for +tag_name+
260
+ */
261
+ static VALUE
262
+ get_description(VALUE klass, VALUE tag_name)
263
+ {
264
+ const htmlElemDesc *description = htmlTagLookup(
265
+ (const xmlChar *)StringValueCStr(tag_name)
266
+ );
267
+
268
+ if (NULL == description) { return Qnil; }
269
+ return Data_Wrap_Struct(klass, 0, 0, DISCARD_CONST_QUAL(void *, description));
270
+ }
271
+
272
+ void
273
+ noko_init_html_element_description()
274
+ {
275
+ cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
276
+
277
+ rb_undef_alloc_func(cNokogiriHtml4ElementDescription);
278
+
279
+ rb_define_singleton_method(cNokogiriHtml4ElementDescription, "[]", get_description, 1);
280
+
281
+ rb_define_method(cNokogiriHtml4ElementDescription, "name", name, 0);
282
+ rb_define_method(cNokogiriHtml4ElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
283
+ rb_define_method(cNokogiriHtml4ElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
284
+ rb_define_method(cNokogiriHtml4ElementDescription, "save_end_tag?", save_end_tag_eh, 0);
285
+ rb_define_method(cNokogiriHtml4ElementDescription, "empty?", empty_eh, 0);
286
+ rb_define_method(cNokogiriHtml4ElementDescription, "deprecated?", deprecated_eh, 0);
287
+ rb_define_method(cNokogiriHtml4ElementDescription, "inline?", inline_eh, 0);
288
+ rb_define_method(cNokogiriHtml4ElementDescription, "description", description, 0);
289
+ rb_define_method(cNokogiriHtml4ElementDescription, "sub_elements", sub_elements, 0);
290
+ rb_define_method(cNokogiriHtml4ElementDescription, "default_sub_element", default_sub_element, 0);
291
+ rb_define_method(cNokogiriHtml4ElementDescription, "optional_attributes", optional_attributes, 0);
292
+ rb_define_method(cNokogiriHtml4ElementDescription, "deprecated_attributes", deprecated_attributes, 0);
293
+ rb_define_method(cNokogiriHtml4ElementDescription, "required_attributes", required_attributes, 0);
294
+ }
@@ -0,0 +1,37 @@
1
+ #include <nokogiri.h>
2
+
3
+ static VALUE cNokogiriHtml4EntityLookup;
4
+
5
+ /*
6
+ * call-seq:
7
+ * get(key)
8
+ *
9
+ * Get the HTML4::EntityDescription for +key+
10
+ */
11
+ static VALUE
12
+ get(VALUE _, VALUE rb_entity_name)
13
+ {
14
+ VALUE cNokogiriHtml4EntityDescription;
15
+ const htmlEntityDesc *c_entity_desc;
16
+ VALUE rb_constructor_args[3];
17
+
18
+ c_entity_desc = htmlEntityLookup((const xmlChar *)StringValueCStr(rb_entity_name));
19
+ if (NULL == c_entity_desc) {
20
+ return Qnil;
21
+ }
22
+
23
+ rb_constructor_args[0] = INT2NUM((long)c_entity_desc->value);
24
+ rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
25
+ rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
26
+
27
+ cNokogiriHtml4EntityDescription = rb_const_get_at(mNokogiriHtml4, rb_intern("EntityDescription"));
28
+ return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtml4EntityDescription);
29
+ }
30
+
31
+ void
32
+ noko_init_html_entity_lookup()
33
+ {
34
+ cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
35
+
36
+ rb_define_method(cNokogiriHtml4EntityLookup, "get", get, 1);
37
+ }
@@ -0,0 +1,119 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriHtml4SaxParserContext ;
4
+
5
+ static void
6
+ deallocate(xmlParserCtxtPtr ctxt)
7
+ {
8
+ NOKOGIRI_DEBUG_START(ctxt);
9
+
10
+ ctxt->sax = NULL;
11
+
12
+ htmlFreeParserCtxt(ctxt);
13
+
14
+ NOKOGIRI_DEBUG_END(ctxt);
15
+ }
16
+
17
+ static VALUE
18
+ parse_memory(VALUE klass, VALUE data, VALUE encoding)
19
+ {
20
+ htmlParserCtxtPtr ctxt;
21
+
22
+ Check_Type(data, T_STRING);
23
+
24
+ if (!(int)RSTRING_LEN(data)) {
25
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
26
+ }
27
+
28
+ ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
29
+ (int)RSTRING_LEN(data));
30
+ if (ctxt->sax) {
31
+ xmlFree(ctxt->sax);
32
+ ctxt->sax = NULL;
33
+ }
34
+
35
+ if (RTEST(encoding)) {
36
+ xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
37
+ if (enc != NULL) {
38
+ xmlSwitchToEncoding(ctxt, enc);
39
+ if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
40
+ rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
41
+ StringValueCStr(encoding));
42
+ }
43
+ }
44
+ }
45
+
46
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
47
+ }
48
+
49
+ static VALUE
50
+ parse_file(VALUE klass, VALUE filename, VALUE encoding)
51
+ {
52
+ htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
53
+ StringValueCStr(filename),
54
+ StringValueCStr(encoding)
55
+ );
56
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
57
+ }
58
+
59
+ static VALUE
60
+ parse_doc(VALUE ctxt_val)
61
+ {
62
+ htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
63
+ htmlParseDocument(ctxt);
64
+ return Qnil;
65
+ }
66
+
67
+ static VALUE
68
+ parse_doc_finalize(VALUE ctxt_val)
69
+ {
70
+ htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
71
+
72
+ if (ctxt->myDoc) {
73
+ xmlFreeDoc(ctxt->myDoc);
74
+ }
75
+
76
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
77
+ return Qnil;
78
+ }
79
+
80
+ static VALUE
81
+ parse_with(VALUE self, VALUE sax_handler)
82
+ {
83
+ htmlParserCtxtPtr ctxt;
84
+ htmlSAXHandlerPtr sax;
85
+
86
+ if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
87
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
88
+ }
89
+
90
+ Data_Get_Struct(self, htmlParserCtxt, ctxt);
91
+ Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
92
+
93
+ /* Free the sax handler since we'll assign our own */
94
+ if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
95
+ xmlFree(ctxt->sax);
96
+ }
97
+
98
+ ctxt->sax = sax;
99
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
100
+
101
+ xmlSetStructuredErrorFunc(NULL, NULL);
102
+
103
+ rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
104
+
105
+ return self;
106
+ }
107
+
108
+ void
109
+ noko_init_html_sax_parser_context()
110
+ {
111
+ assert(cNokogiriXmlSaxParserContext);
112
+ cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext",
113
+ cNokogiriXmlSaxParserContext);
114
+
115
+ rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "memory", parse_memory, 2);
116
+ rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "file", parse_file, 2);
117
+
118
+ rb_define_method(cNokogiriHtml4SaxParserContext, "parse_with", parse_with, 1);
119
+ }
@@ -0,0 +1,95 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriHtml4SaxPushParser;
4
+
5
+ /*
6
+ * call-seq:
7
+ * native_write(chunk, last_chunk)
8
+ *
9
+ * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
10
+ */
11
+ static VALUE
12
+ native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
13
+ {
14
+ xmlParserCtxtPtr ctx;
15
+ const char *chunk = NULL;
16
+ int size = 0;
17
+ int status = 0;
18
+ libxmlStructuredErrorHandlerState handler_state;
19
+
20
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
21
+
22
+ if (Qnil != _chunk) {
23
+ chunk = StringValuePtr(_chunk);
24
+ size = (int)RSTRING_LEN(_chunk);
25
+ }
26
+
27
+ Nokogiri_structured_error_func_save_and_set(&handler_state, NULL, NULL);
28
+
29
+ status = htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0);
30
+
31
+ Nokogiri_structured_error_func_restore(&handler_state);
32
+
33
+ if ((status != 0) && !(ctx->options & XML_PARSE_RECOVER)) {
34
+ // TODO: there appear to be no tests for this block
35
+ xmlErrorPtr e = xmlCtxtGetLastError(ctx);
36
+ Nokogiri_error_raise(NULL, e);
37
+ }
38
+
39
+ return self;
40
+ }
41
+
42
+ /*
43
+ * call-seq:
44
+ * initialize_native(xml_sax, filename)
45
+ *
46
+ * Initialize the push parser with +xml_sax+ using +filename+
47
+ */
48
+ static VALUE
49
+ initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
50
+ VALUE encoding)
51
+ {
52
+ htmlSAXHandlerPtr sax;
53
+ const char *filename = NULL;
54
+ htmlParserCtxtPtr ctx;
55
+ xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
56
+
57
+ Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
58
+
59
+ if (_filename != Qnil) { filename = StringValueCStr(_filename); }
60
+
61
+ if (!NIL_P(encoding)) {
62
+ enc = xmlParseCharEncoding(StringValueCStr(encoding));
63
+ if (enc == XML_CHAR_ENCODING_ERROR) {
64
+ rb_raise(rb_eArgError, "Unsupported Encoding");
65
+ }
66
+ }
67
+
68
+ ctx = htmlCreatePushParserCtxt(
69
+ sax,
70
+ NULL,
71
+ NULL,
72
+ 0,
73
+ filename,
74
+ enc
75
+ );
76
+ if (ctx == NULL) {
77
+ rb_raise(rb_eRuntimeError, "Could not create a parser context");
78
+ }
79
+
80
+ ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
81
+
82
+ ctx->sax2 = 1;
83
+ DATA_PTR(self) = ctx;
84
+ return self;
85
+ }
86
+
87
+ void
88
+ noko_init_html_sax_push_parser()
89
+ {
90
+ assert(cNokogiriXmlSaxPushParser);
91
+ cNokogiriHtml4SaxPushParser = rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);
92
+
93
+ rb_define_private_method(cNokogiriHtml4SaxPushParser, "initialize_native", initialize_native, 3);
94
+ rb_define_private_method(cNokogiriHtml4SaxPushParser, "native_write", native_write, 2);
95
+ }