nokogiri 1.5.10 → 1.12.5

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 (328) hide show
  1. checksums.yaml +7 -0
  2. data/Gemfile +3 -0
  3. data/LICENSE-DEPENDENCIES.md +1903 -0
  4. data/LICENSE.md +9 -0
  5. data/README.md +278 -0
  6. data/bin/nokogiri +50 -10
  7. data/dependencies.yml +74 -0
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +944 -100
  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 +232 -87
  18. data/ext/nokogiri/nokogiri.h +188 -129
  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 +29 -21
  24. data/ext/nokogiri/xml_document.c +305 -201
  25. data/ext/nokogiri/xml_document_fragment.c +13 -15
  26. data/ext/nokogiri/xml_dtd.c +54 -48
  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 +30 -19
  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 +74 -32
  33. data/ext/nokogiri/xml_node.c +808 -503
  34. data/ext/nokogiri/xml_node_set.c +239 -208
  35. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  36. data/ext/nokogiri/xml_reader.c +198 -186
  37. data/ext/nokogiri/xml_relax_ng.c +52 -28
  38. data/ext/nokogiri/xml_sax_parser.c +123 -125
  39. data/ext/nokogiri/xml_sax_parser_context.c +138 -79
  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 +50 -23
  43. data/ext/nokogiri/xml_text.c +14 -18
  44. data/ext/nokogiri/xml_xpath_context.c +162 -98
  45. data/ext/nokogiri/xslt_stylesheet.c +162 -168
  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 +4886 -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/css/node.rb +1 -50
  92. data/lib/nokogiri/css/parser.rb +317 -286
  93. data/lib/nokogiri/css/parser.y +57 -43
  94. data/lib/nokogiri/css/parser_extras.rb +39 -36
  95. data/lib/nokogiri/css/syntax_error.rb +2 -1
  96. data/lib/nokogiri/css/tokenizer.rb +105 -103
  97. data/lib/nokogiri/css/tokenizer.rex +5 -5
  98. data/lib/nokogiri/css/xpath_visitor.rb +137 -48
  99. data/lib/nokogiri/css.rb +15 -14
  100. data/lib/nokogiri/decorators/slop.rb +13 -5
  101. data/lib/nokogiri/extension.rb +31 -0
  102. data/lib/nokogiri/gumbo.rb +14 -0
  103. data/lib/nokogiri/html.rb +32 -27
  104. data/lib/nokogiri/{html → html4}/builder.rb +3 -2
  105. data/lib/nokogiri/{html → html4}/document.rb +118 -50
  106. data/lib/nokogiri/{html → html4}/document_fragment.rb +20 -11
  107. data/lib/nokogiri/{html → html4}/element_description.rb +2 -1
  108. data/lib/nokogiri/{html → html4}/element_description_defaults.rb +2 -1
  109. data/lib/nokogiri/{html → html4}/entity_lookup.rb +2 -1
  110. data/lib/nokogiri/{html → html4}/sax/parser.rb +22 -14
  111. data/lib/nokogiri/html4/sax/parser_context.rb +19 -0
  112. data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
  113. data/lib/nokogiri/html4.rb +40 -0
  114. data/lib/nokogiri/html5/document.rb +74 -0
  115. data/lib/nokogiri/html5/document_fragment.rb +80 -0
  116. data/lib/nokogiri/html5/node.rb +93 -0
  117. data/lib/nokogiri/html5.rb +473 -0
  118. data/lib/nokogiri/jruby/dependencies.rb +20 -0
  119. data/lib/nokogiri/syntax_error.rb +1 -0
  120. data/lib/nokogiri/version/constant.rb +5 -0
  121. data/lib/nokogiri/version/info.rb +215 -0
  122. data/lib/nokogiri/version.rb +3 -91
  123. data/lib/nokogiri/xml/attr.rb +1 -0
  124. data/lib/nokogiri/xml/attribute_decl.rb +1 -0
  125. data/lib/nokogiri/xml/builder.rb +75 -33
  126. data/lib/nokogiri/xml/cdata.rb +1 -0
  127. data/lib/nokogiri/xml/character_data.rb +1 -0
  128. data/lib/nokogiri/xml/document.rb +157 -54
  129. data/lib/nokogiri/xml/document_fragment.rb +55 -8
  130. data/lib/nokogiri/xml/dtd.rb +15 -4
  131. data/lib/nokogiri/xml/element_content.rb +1 -0
  132. data/lib/nokogiri/xml/element_decl.rb +1 -0
  133. data/lib/nokogiri/xml/entity_decl.rb +1 -0
  134. data/lib/nokogiri/xml/entity_reference.rb +19 -0
  135. data/lib/nokogiri/xml/namespace.rb +1 -0
  136. data/lib/nokogiri/xml/node/save_options.rb +2 -1
  137. data/lib/nokogiri/xml/node.rb +712 -431
  138. data/lib/nokogiri/xml/node_set.rb +140 -123
  139. data/lib/nokogiri/xml/notation.rb +1 -0
  140. data/lib/nokogiri/xml/parse_options.rb +31 -0
  141. data/lib/nokogiri/xml/pp/character_data.rb +1 -0
  142. data/lib/nokogiri/xml/pp/node.rb +1 -0
  143. data/lib/nokogiri/xml/pp.rb +3 -2
  144. data/lib/nokogiri/xml/processing_instruction.rb +1 -0
  145. data/lib/nokogiri/xml/reader.rb +9 -12
  146. data/lib/nokogiri/xml/relax_ng.rb +7 -2
  147. data/lib/nokogiri/xml/sax/document.rb +25 -30
  148. data/lib/nokogiri/xml/sax/parser.rb +8 -8
  149. data/lib/nokogiri/xml/sax/parser_context.rb +1 -0
  150. data/lib/nokogiri/xml/sax/push_parser.rb +1 -0
  151. data/lib/nokogiri/xml/sax.rb +5 -4
  152. data/lib/nokogiri/xml/schema.rb +13 -4
  153. data/lib/nokogiri/xml/searchable.rb +239 -0
  154. data/lib/nokogiri/xml/syntax_error.rb +25 -1
  155. data/lib/nokogiri/xml/text.rb +1 -0
  156. data/lib/nokogiri/xml/xpath/syntax_error.rb +2 -1
  157. data/lib/nokogiri/xml/xpath.rb +4 -5
  158. data/lib/nokogiri/xml/xpath_context.rb +1 -0
  159. data/lib/nokogiri/xml.rb +37 -35
  160. data/lib/nokogiri/xslt/stylesheet.rb +2 -1
  161. data/lib/nokogiri/xslt.rb +17 -16
  162. data/lib/nokogiri.rb +55 -58
  163. data/lib/xsd/xmlparser/nokogiri.rb +1 -0
  164. data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
  165. data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
  166. data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
  167. data/patches/libxml2/0004-use-glibc-strlen.patch +53 -0
  168. data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  169. data/patches/libxml2/0006-update-automake-files-for-arm64.patch +2511 -0
  170. data/patches/libxml2/0007-Fix-XPath-recursion-limit.patch +31 -0
  171. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +2511 -0
  172. data/patches/libxslt/0002-Fix-xml2-config-check-in-configure-script.patch +19 -0
  173. data/ports/archives/libxml2-2.9.12.tar.gz +0 -0
  174. data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
  175. metadata +307 -459
  176. data/.autotest +0 -26
  177. data/.gemtest +0 -0
  178. data/CHANGELOG.ja.rdoc +0 -785
  179. data/CHANGELOG.rdoc +0 -783
  180. data/C_CODING_STYLE.rdoc +0 -33
  181. data/Manifest.txt +0 -303
  182. data/README.ja.rdoc +0 -106
  183. data/README.rdoc +0 -175
  184. data/ROADMAP.md +0 -90
  185. data/Rakefile +0 -228
  186. data/STANDARD_RESPONSES.md +0 -47
  187. data/Y_U_NO_GEMSPEC.md +0 -155
  188. data/build_all +0 -105
  189. data/ext/nokogiri/html_document.c +0 -170
  190. data/ext/nokogiri/html_document.h +0 -10
  191. data/ext/nokogiri/html_element_description.c +0 -279
  192. data/ext/nokogiri/html_element_description.h +0 -10
  193. data/ext/nokogiri/html_entity_lookup.c +0 -32
  194. data/ext/nokogiri/html_entity_lookup.h +0 -8
  195. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  196. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  197. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  198. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  199. data/ext/nokogiri/xml_attr.h +0 -9
  200. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  201. data/ext/nokogiri/xml_cdata.h +0 -9
  202. data/ext/nokogiri/xml_comment.h +0 -9
  203. data/ext/nokogiri/xml_document.h +0 -23
  204. data/ext/nokogiri/xml_document_fragment.h +0 -10
  205. data/ext/nokogiri/xml_dtd.h +0 -10
  206. data/ext/nokogiri/xml_element_content.h +0 -10
  207. data/ext/nokogiri/xml_element_decl.h +0 -9
  208. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  209. data/ext/nokogiri/xml_entity_decl.h +0 -10
  210. data/ext/nokogiri/xml_entity_reference.h +0 -9
  211. data/ext/nokogiri/xml_io.c +0 -56
  212. data/ext/nokogiri/xml_io.h +0 -11
  213. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  214. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  215. data/ext/nokogiri/xml_namespace.h +0 -13
  216. data/ext/nokogiri/xml_node.h +0 -13
  217. data/ext/nokogiri/xml_node_set.h +0 -14
  218. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  219. data/ext/nokogiri/xml_reader.h +0 -10
  220. data/ext/nokogiri/xml_relax_ng.h +0 -9
  221. data/ext/nokogiri/xml_sax_parser.h +0 -39
  222. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  223. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  224. data/ext/nokogiri/xml_schema.h +0 -9
  225. data/ext/nokogiri/xml_syntax_error.h +0 -13
  226. data/ext/nokogiri/xml_text.h +0 -9
  227. data/ext/nokogiri/xml_xpath_context.h +0 -10
  228. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  229. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  230. data/lib/nokogiri/html/sax/push_parser.rb +0 -16
  231. data/tasks/cross_compile.rb +0 -150
  232. data/tasks/nokogiri.org.rb +0 -24
  233. data/tasks/test.rb +0 -95
  234. data/test/css/test_nthiness.rb +0 -159
  235. data/test/css/test_parser.rb +0 -341
  236. data/test/css/test_tokenizer.rb +0 -198
  237. data/test/css/test_xpath_visitor.rb +0 -91
  238. data/test/decorators/test_slop.rb +0 -16
  239. data/test/files/2ch.html +0 -108
  240. data/test/files/address_book.rlx +0 -12
  241. data/test/files/address_book.xml +0 -10
  242. data/test/files/bar/bar.xsd +0 -4
  243. data/test/files/dont_hurt_em_why.xml +0 -422
  244. data/test/files/encoding.html +0 -82
  245. data/test/files/encoding.xhtml +0 -84
  246. data/test/files/exslt.xml +0 -8
  247. data/test/files/exslt.xslt +0 -35
  248. data/test/files/foo/foo.xsd +0 -4
  249. data/test/files/metacharset.html +0 -10
  250. data/test/files/noencoding.html +0 -47
  251. data/test/files/po.xml +0 -32
  252. data/test/files/po.xsd +0 -66
  253. data/test/files/shift_jis.html +0 -10
  254. data/test/files/shift_jis.xml +0 -5
  255. data/test/files/snuggles.xml +0 -3
  256. data/test/files/staff.dtd +0 -10
  257. data/test/files/staff.xml +0 -59
  258. data/test/files/staff.xslt +0 -32
  259. data/test/files/test_document_url/bar.xml +0 -2
  260. data/test/files/test_document_url/document.dtd +0 -4
  261. data/test/files/test_document_url/document.xml +0 -6
  262. data/test/files/tlm.html +0 -850
  263. data/test/files/to_be_xincluded.xml +0 -2
  264. data/test/files/valid_bar.xml +0 -2
  265. data/test/files/xinclude.xml +0 -4
  266. data/test/helper.rb +0 -154
  267. data/test/html/sax/test_parser.rb +0 -141
  268. data/test/html/sax/test_parser_context.rb +0 -46
  269. data/test/html/test_builder.rb +0 -164
  270. data/test/html/test_document.rb +0 -552
  271. data/test/html/test_document_encoding.rb +0 -138
  272. data/test/html/test_document_fragment.rb +0 -261
  273. data/test/html/test_element_description.rb +0 -105
  274. data/test/html/test_named_characters.rb +0 -14
  275. data/test/html/test_node.rb +0 -196
  276. data/test/html/test_node_encoding.rb +0 -27
  277. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
  278. data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
  279. data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
  280. data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -66
  281. data/test/test_convert_xpath.rb +0 -135
  282. data/test/test_css_cache.rb +0 -45
  283. data/test/test_encoding_handler.rb +0 -46
  284. data/test/test_memory_leak.rb +0 -156
  285. data/test/test_nokogiri.rb +0 -132
  286. data/test/test_reader.rb +0 -555
  287. data/test/test_soap4r_sax.rb +0 -52
  288. data/test/test_xslt_transforms.rb +0 -254
  289. data/test/xml/node/test_save_options.rb +0 -28
  290. data/test/xml/node/test_subclass.rb +0 -44
  291. data/test/xml/sax/test_parser.rb +0 -366
  292. data/test/xml/sax/test_parser_context.rb +0 -106
  293. data/test/xml/sax/test_push_parser.rb +0 -157
  294. data/test/xml/test_attr.rb +0 -64
  295. data/test/xml/test_attribute_decl.rb +0 -86
  296. data/test/xml/test_builder.rb +0 -306
  297. data/test/xml/test_c14n.rb +0 -151
  298. data/test/xml/test_cdata.rb +0 -48
  299. data/test/xml/test_comment.rb +0 -29
  300. data/test/xml/test_document.rb +0 -828
  301. data/test/xml/test_document_encoding.rb +0 -28
  302. data/test/xml/test_document_fragment.rb +0 -223
  303. data/test/xml/test_dtd.rb +0 -103
  304. data/test/xml/test_dtd_encoding.rb +0 -33
  305. data/test/xml/test_element_content.rb +0 -56
  306. data/test/xml/test_element_decl.rb +0 -73
  307. data/test/xml/test_entity_decl.rb +0 -122
  308. data/test/xml/test_entity_reference.rb +0 -245
  309. data/test/xml/test_namespace.rb +0 -95
  310. data/test/xml/test_node.rb +0 -1137
  311. data/test/xml/test_node_attributes.rb +0 -96
  312. data/test/xml/test_node_encoding.rb +0 -107
  313. data/test/xml/test_node_inheritance.rb +0 -32
  314. data/test/xml/test_node_reparenting.rb +0 -374
  315. data/test/xml/test_node_set.rb +0 -755
  316. data/test/xml/test_parse_options.rb +0 -64
  317. data/test/xml/test_processing_instruction.rb +0 -30
  318. data/test/xml/test_reader_encoding.rb +0 -142
  319. data/test/xml/test_relax_ng.rb +0 -60
  320. data/test/xml/test_schema.rb +0 -103
  321. data/test/xml/test_syntax_error.rb +0 -12
  322. data/test/xml/test_text.rb +0 -45
  323. data/test/xml/test_unparented_node.rb +0 -422
  324. data/test/xml/test_xinclude.rb +0 -83
  325. data/test/xml/test_xpath.rb +0 -295
  326. data/test/xslt/test_custom_functions.rb +0 -133
  327. data/test/xslt/test_exception_handling.rb +0 -37
  328. data/test_all +0 -81
@@ -1,16 +1,17 @@
1
- #include <xml_sax_parser_context.h>
1
+ #include <nokogiri.h>
2
2
 
3
3
  VALUE cNokogiriXmlSaxParserContext ;
4
4
 
5
- static void deallocate(xmlParserCtxtPtr ctxt)
5
+ static void
6
+ deallocate(xmlParserCtxtPtr ctxt)
6
7
  {
7
- NOKOGIRI_DEBUG_START(handler);
8
+ NOKOGIRI_DEBUG_START(ctxt);
8
9
 
9
10
  ctxt->sax = NULL;
10
11
 
11
12
  xmlFreeParserCtxt(ctxt);
12
13
 
13
- NOKOGIRI_DEBUG_END(handler);
14
+ NOKOGIRI_DEBUG_END(ctxt);
14
15
  }
15
16
 
16
17
  /*
@@ -22,19 +23,19 @@ static void deallocate(xmlParserCtxtPtr ctxt)
22
23
  static VALUE
23
24
  parse_io(VALUE klass, VALUE io, VALUE encoding)
24
25
  {
25
- xmlParserCtxtPtr ctxt;
26
- xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding);
27
-
28
- ctxt = xmlCreateIOParserCtxt(NULL, NULL,
29
- (xmlInputReadCallback)io_read_callback,
30
- (xmlInputCloseCallback)io_close_callback,
31
- (void *)io, enc);
32
- if (ctxt->sax) {
33
- xmlFree(ctxt->sax);
34
- ctxt->sax = NULL;
35
- }
36
-
37
- return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
26
+ xmlParserCtxtPtr ctxt;
27
+ xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding);
28
+
29
+ ctxt = xmlCreateIOParserCtxt(NULL, NULL,
30
+ (xmlInputReadCallback)noko_io_read,
31
+ (xmlInputCloseCallback)noko_io_close,
32
+ (void *)io, enc);
33
+ if (ctxt->sax) {
34
+ xmlFree(ctxt->sax);
35
+ ctxt->sax = NULL;
36
+ }
37
+
38
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
38
39
  }
39
40
 
40
41
  /*
@@ -43,9 +44,10 @@ parse_io(VALUE klass, VALUE io, VALUE encoding)
43
44
  *
44
45
  * Parse file given +filename+
45
46
  */
46
- static VALUE parse_file(VALUE klass, VALUE filename)
47
+ static VALUE
48
+ parse_file(VALUE klass, VALUE filename)
47
49
  {
48
- xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValuePtr(filename));
50
+ xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValueCStr(filename));
49
51
  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
50
52
  }
51
53
 
@@ -58,41 +60,44 @@ static VALUE parse_file(VALUE klass, VALUE filename)
58
60
  static VALUE
59
61
  parse_memory(VALUE klass, VALUE data)
60
62
  {
61
- xmlParserCtxtPtr ctxt;
63
+ xmlParserCtxtPtr ctxt;
62
64
 
63
- if (NIL_P(data))
64
- rb_raise(rb_eArgError, "data cannot be nil");
65
- if (!(int)RSTRING_LEN(data))
66
- rb_raise(rb_eRuntimeError, "data cannot be empty");
65
+ if (NIL_P(data)) {
66
+ rb_raise(rb_eArgError, "data cannot be nil");
67
+ }
68
+ if (!(int)RSTRING_LEN(data)) {
69
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
70
+ }
67
71
 
68
- ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(data),
69
- (int)RSTRING_LEN(data));
70
- if (ctxt->sax) {
71
- xmlFree(ctxt->sax);
72
- ctxt->sax = NULL;
73
- }
72
+ ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(data),
73
+ (int)RSTRING_LEN(data));
74
+ if (ctxt->sax) {
75
+ xmlFree(ctxt->sax);
76
+ ctxt->sax = NULL;
77
+ }
74
78
 
75
- return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
79
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
76
80
  }
77
81
 
78
82
  static VALUE
79
83
  parse_doc(VALUE ctxt_val)
80
84
  {
81
- xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
82
- xmlParseDocument(ctxt);
83
- return Qnil;
85
+ xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
86
+ xmlParseDocument(ctxt);
87
+ return Qnil;
84
88
  }
85
89
 
86
90
  static VALUE
87
91
  parse_doc_finalize(VALUE ctxt_val)
88
92
  {
89
- xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
93
+ xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
90
94
 
91
- if (NULL != ctxt->myDoc)
92
- xmlFreeDoc(ctxt->myDoc);
95
+ if (NULL != ctxt->myDoc) {
96
+ xmlFreeDoc(ctxt->myDoc);
97
+ }
93
98
 
94
- NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
95
- return Qnil;
99
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
100
+ return Qnil;
96
101
  }
97
102
 
98
103
  /*
@@ -104,25 +109,29 @@ parse_doc_finalize(VALUE ctxt_val)
104
109
  static VALUE
105
110
  parse_with(VALUE self, VALUE sax_handler)
106
111
  {
107
- xmlParserCtxtPtr ctxt;
108
- xmlSAXHandlerPtr sax;
112
+ xmlParserCtxtPtr ctxt;
113
+ xmlSAXHandlerPtr sax;
114
+
115
+ if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
116
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
117
+ }
109
118
 
110
- if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
111
- rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
119
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
120
+ Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
112
121
 
113
- Data_Get_Struct(self, xmlParserCtxt, ctxt);
114
- Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
122
+ /* Free the sax handler since we'll assign our own */
123
+ if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
124
+ xmlFree(ctxt->sax);
125
+ }
115
126
 
116
- /* Free the sax handler since we'll assign our own */
117
- if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
118
- xmlFree(ctxt->sax);
127
+ ctxt->sax = sax;
128
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
119
129
 
120
- ctxt->sax = sax;
121
- ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
130
+ xmlSetStructuredErrorFunc(NULL, NULL);
122
131
 
123
- rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
132
+ rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
124
133
 
125
- return Qnil;
134
+ return Qnil;
126
135
  }
127
136
 
128
137
  /*
@@ -132,15 +141,17 @@ parse_with(VALUE self, VALUE sax_handler)
132
141
  * Should this parser replace entities? &amp; will get converted to '&' if
133
142
  * set to true
134
143
  */
135
- static VALUE set_replace_entities(VALUE self, VALUE value)
144
+ static VALUE
145
+ set_replace_entities(VALUE self, VALUE value)
136
146
  {
137
147
  xmlParserCtxtPtr ctxt;
138
148
  Data_Get_Struct(self, xmlParserCtxt, ctxt);
139
149
 
140
- if(Qfalse == value)
150
+ if (Qfalse == value) {
141
151
  ctxt->replaceEntities = 0;
142
- else
152
+ } else {
143
153
  ctxt->replaceEntities = 1;
154
+ }
144
155
 
145
156
  return value;
146
157
  }
@@ -152,15 +163,17 @@ static VALUE set_replace_entities(VALUE self, VALUE value)
152
163
  * Should this parser replace entities? &amp; will get converted to '&' if
153
164
  * set to true
154
165
  */
155
- static VALUE get_replace_entities(VALUE self)
166
+ static VALUE
167
+ get_replace_entities(VALUE self)
156
168
  {
157
169
  xmlParserCtxtPtr ctxt;
158
170
  Data_Get_Struct(self, xmlParserCtxt, ctxt);
159
171
 
160
- if(0 == ctxt->replaceEntities)
172
+ if (0 == ctxt->replaceEntities) {
161
173
  return Qfalse;
162
- else
174
+ } else {
163
175
  return Qtrue;
176
+ }
164
177
  }
165
178
 
166
179
  /*
@@ -168,7 +181,8 @@ static VALUE get_replace_entities(VALUE self)
168
181
  *
169
182
  * Get the current line the parser context is processing.
170
183
  */
171
- static VALUE line(VALUE self)
184
+ static VALUE
185
+ line(VALUE self)
172
186
  {
173
187
  xmlParserCtxtPtr ctxt;
174
188
  xmlParserInputPtr io;
@@ -176,8 +190,9 @@ static VALUE line(VALUE self)
176
190
  Data_Get_Struct(self, xmlParserCtxt, ctxt);
177
191
 
178
192
  io = ctxt->input;
179
- if(io)
193
+ if (io) {
180
194
  return INT2NUM(io->line);
195
+ }
181
196
 
182
197
  return Qnil;
183
198
  }
@@ -187,7 +202,8 @@ static VALUE line(VALUE self)
187
202
  *
188
203
  * Get the current column the parser context is processing.
189
204
  */
190
- static VALUE column(VALUE self)
205
+ static VALUE
206
+ column(VALUE self)
191
207
  {
192
208
  xmlParserCtxtPtr ctxt;
193
209
  xmlParserInputPtr io;
@@ -195,28 +211,71 @@ static VALUE column(VALUE self)
195
211
  Data_Get_Struct(self, xmlParserCtxt, ctxt);
196
212
 
197
213
  io = ctxt->input;
198
- if(io)
214
+ if (io) {
199
215
  return INT2NUM(io->col);
216
+ }
200
217
 
201
218
  return Qnil;
202
219
  }
203
220
 
204
- void init_xml_sax_parser_context()
221
+ /*
222
+ * call-seq:
223
+ * recovery=(boolean)
224
+ *
225
+ * Should this parser recover from structural errors? It will not stop processing
226
+ * file on structural errors if set to true
227
+ */
228
+ static VALUE
229
+ set_recovery(VALUE self, VALUE value)
230
+ {
231
+ xmlParserCtxtPtr ctxt;
232
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
233
+
234
+ if (value == Qfalse) {
235
+ ctxt->recovery = 0;
236
+ } else {
237
+ ctxt->recovery = 1;
238
+ }
239
+
240
+ return value;
241
+ }
242
+
243
+ /*
244
+ * call-seq:
245
+ * recovery
246
+ *
247
+ * Should this parser recover from structural errors? It will not stop processing
248
+ * file on structural errors if set to true
249
+ */
250
+ static VALUE
251
+ get_recovery(VALUE self)
205
252
  {
206
- VALUE nokogiri = rb_define_module("Nokogiri");
207
- VALUE xml = rb_define_module_under(nokogiri, "XML");
208
- VALUE sax = rb_define_module_under(xml, "SAX");
209
- VALUE klass = rb_define_class_under(sax, "ParserContext", rb_cObject);
210
-
211
- cNokogiriXmlSaxParserContext = klass;
212
-
213
- rb_define_singleton_method(klass, "io", parse_io, 2);
214
- rb_define_singleton_method(klass, "memory", parse_memory, 1);
215
- rb_define_singleton_method(klass, "file", parse_file, 1);
216
-
217
- rb_define_method(klass, "parse_with", parse_with, 1);
218
- rb_define_method(klass, "replace_entities=", set_replace_entities, 1);
219
- rb_define_method(klass, "replace_entities", get_replace_entities, 0);
220
- rb_define_method(klass, "line", line, 0);
221
- rb_define_method(klass, "column", column, 0);
253
+ xmlParserCtxtPtr ctxt;
254
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
255
+
256
+ if (ctxt->recovery == 0) {
257
+ return Qfalse;
258
+ } else {
259
+ return Qtrue;
260
+ }
261
+ }
262
+
263
+ void
264
+ noko_init_xml_sax_parser_context()
265
+ {
266
+ cNokogiriXmlSaxParserContext = rb_define_class_under(mNokogiriXmlSax, "ParserContext", rb_cObject);
267
+
268
+ rb_undef_alloc_func(cNokogiriXmlSaxParserContext);
269
+
270
+ rb_define_singleton_method(cNokogiriXmlSaxParserContext, "io", parse_io, 2);
271
+ rb_define_singleton_method(cNokogiriXmlSaxParserContext, "memory", parse_memory, 1);
272
+ rb_define_singleton_method(cNokogiriXmlSaxParserContext, "file", parse_file, 1);
273
+
274
+ rb_define_method(cNokogiriXmlSaxParserContext, "parse_with", parse_with, 1);
275
+ rb_define_method(cNokogiriXmlSaxParserContext, "replace_entities=", set_replace_entities, 1);
276
+ rb_define_method(cNokogiriXmlSaxParserContext, "replace_entities", get_replace_entities, 0);
277
+ rb_define_method(cNokogiriXmlSaxParserContext, "recovery=", set_recovery, 1);
278
+ rb_define_method(cNokogiriXmlSaxParserContext, "recovery", get_recovery, 0);
279
+ rb_define_method(cNokogiriXmlSaxParserContext, "line", line, 0);
280
+ rb_define_method(cNokogiriXmlSaxParserContext, "column", column, 0);
222
281
  }
@@ -1,16 +1,20 @@
1
- #include <xml_sax_push_parser.h>
1
+ #include <nokogiri.h>
2
2
 
3
- static void deallocate(xmlParserCtxtPtr ctx)
3
+ VALUE cNokogiriXmlSaxPushParser ;
4
+
5
+ static void
6
+ deallocate(xmlParserCtxtPtr ctx)
4
7
  {
5
8
  NOKOGIRI_DEBUG_START(ctx);
6
- if(ctx != NULL) {
9
+ if (ctx != NULL) {
7
10
  NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
8
11
  xmlFreeParserCtxt(ctx);
9
12
  }
10
13
  NOKOGIRI_DEBUG_END(ctx);
11
14
  }
12
15
 
13
- static VALUE allocate(VALUE klass)
16
+ static VALUE
17
+ allocate(VALUE klass)
14
18
  {
15
19
  return Data_Wrap_Struct(klass, NULL, deallocate, NULL);
16
20
  }
@@ -21,21 +25,24 @@ static VALUE allocate(VALUE klass)
21
25
  *
22
26
  * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
23
27
  */
24
- static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
28
+ static VALUE
29
+ native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
25
30
  {
26
31
  xmlParserCtxtPtr ctx;
27
- const char * chunk = NULL;
32
+ const char *chunk = NULL;
28
33
  int size = 0;
29
34
 
30
35
 
31
36
  Data_Get_Struct(self, xmlParserCtxt, ctx);
32
37
 
33
- if(Qnil != _chunk) {
38
+ if (Qnil != _chunk) {
34
39
  chunk = StringValuePtr(_chunk);
35
40
  size = (int)RSTRING_LEN(_chunk);
36
41
  }
37
42
 
38
- if(xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
43
+ xmlSetStructuredErrorFunc(NULL, NULL);
44
+
45
+ if (xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
39
46
  if (!(ctx->options & XML_PARSE_RECOVER)) {
40
47
  xmlErrorPtr e = xmlCtxtGetLastError(ctx);
41
48
  Nokogiri_error_raise(NULL, e);
@@ -51,25 +58,27 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
51
58
  *
52
59
  * Initialize the push parser with +xml_sax+ using +filename+
53
60
  */
54
- static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
61
+ static VALUE
62
+ initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
55
63
  {
56
64
  xmlSAXHandlerPtr sax;
57
- const char * filename = NULL;
65
+ const char *filename = NULL;
58
66
  xmlParserCtxtPtr ctx;
59
67
 
60
68
  Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
61
69
 
62
- if(_filename != Qnil) filename = StringValuePtr(_filename);
70
+ if (_filename != Qnil) { filename = StringValueCStr(_filename); }
63
71
 
64
72
  ctx = xmlCreatePushParserCtxt(
65
- sax,
66
- NULL,
67
- NULL,
68
- 0,
69
- filename
70
- );
71
- if(ctx == NULL)
73
+ sax,
74
+ NULL,
75
+ NULL,
76
+ 0,
77
+ filename
78
+ );
79
+ if (ctx == NULL) {
72
80
  rb_raise(rb_eRuntimeError, "Could not create a parser context");
81
+ }
73
82
 
74
83
  ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
75
84
 
@@ -78,7 +87,8 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
78
87
  return self;
79
88
  }
80
89
 
81
- static VALUE get_options(VALUE self)
90
+ static VALUE
91
+ get_options(VALUE self)
82
92
  {
83
93
  xmlParserCtxtPtr ctx;
84
94
  Data_Get_Struct(self, xmlParserCtxt, ctx);
@@ -86,30 +96,73 @@ static VALUE get_options(VALUE self)
86
96
  return INT2NUM(ctx->options);
87
97
  }
88
98
 
89
- static VALUE set_options(VALUE self, VALUE options)
99
+ static VALUE
100
+ set_options(VALUE self, VALUE options)
90
101
  {
91
102
  xmlParserCtxtPtr ctx;
92
103
  Data_Get_Struct(self, xmlParserCtxt, ctx);
93
104
 
94
- if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0)
105
+ if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0) {
95
106
  rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
107
+ }
96
108
 
97
109
  return Qnil;
98
110
  }
99
111
 
100
- VALUE cNokogiriXmlSaxPushParser ;
101
- void init_xml_sax_push_parser()
112
+ /*
113
+ * call-seq:
114
+ * replace_entities
115
+ *
116
+ * Should this parser replace entities? &amp; will get converted to '&' if
117
+ * set to true
118
+ */
119
+ static VALUE
120
+ get_replace_entities(VALUE self)
121
+ {
122
+ xmlParserCtxtPtr ctx;
123
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
124
+
125
+ if (0 == ctx->replaceEntities) {
126
+ return Qfalse;
127
+ } else {
128
+ return Qtrue;
129
+ }
130
+ }
131
+
132
+ /*
133
+ * call-seq:
134
+ * replace_entities=(boolean)
135
+ *
136
+ * Should this parser replace entities? &amp; will get converted to '&' if
137
+ * set to true
138
+ */
139
+ static VALUE
140
+ set_replace_entities(VALUE self, VALUE value)
141
+ {
142
+ xmlParserCtxtPtr ctx;
143
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
144
+
145
+ if (Qfalse == value) {
146
+ ctx->replaceEntities = 0;
147
+ } else {
148
+ ctx->replaceEntities = 1;
149
+ }
150
+
151
+ return value;
152
+ }
153
+
154
+ void
155
+ noko_init_xml_sax_push_parser()
102
156
  {
103
- VALUE nokogiri = rb_define_module("Nokogiri");
104
- VALUE xml = rb_define_module_under(nokogiri, "XML");
105
- VALUE sax = rb_define_module_under(xml, "SAX");
106
- VALUE klass = rb_define_class_under(sax, "PushParser", rb_cObject);
107
-
108
- cNokogiriXmlSaxPushParser = klass;
109
-
110
- rb_define_alloc_func(klass, allocate);
111
- rb_define_private_method(klass, "initialize_native", initialize_native, 2);
112
- rb_define_private_method(klass, "native_write", native_write, 2);
113
- rb_define_method(klass, "options", get_options, 0);
114
- rb_define_method(klass, "options=", set_options, 1);
157
+ cNokogiriXmlSaxPushParser = rb_define_class_under(mNokogiriXmlSax, "PushParser", rb_cObject);
158
+
159
+ rb_define_alloc_func(cNokogiriXmlSaxPushParser, allocate);
160
+
161
+ rb_define_method(cNokogiriXmlSaxPushParser, "options", get_options, 0);
162
+ rb_define_method(cNokogiriXmlSaxPushParser, "options=", set_options, 1);
163
+ rb_define_method(cNokogiriXmlSaxPushParser, "replace_entities", get_replace_entities, 0);
164
+ rb_define_method(cNokogiriXmlSaxPushParser, "replace_entities=", set_replace_entities, 1);
165
+
166
+ rb_define_private_method(cNokogiriXmlSaxPushParser, "initialize_native", initialize_native, 2);
167
+ rb_define_private_method(cNokogiriXmlSaxPushParser, "native_write", native_write, 2);
115
168
  }