nokogiri-backupify 1.5.0.beta.4

Sign up to get free protection for your applications and to get access to all the features.
Files changed (275) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +509 -0
  3. data/CHANGELOG.rdoc +490 -0
  4. data/Manifest.txt +274 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +150 -0
  7. data/Rakefile +217 -0
  8. data/bin/nokogiri +54 -0
  9. data/deps.rip +5 -0
  10. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  11. data/ext/java/nokogiri/HtmlDocument.java +146 -0
  12. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  13. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  14. data/ext/java/nokogiri/HtmlSaxParserContext.java +256 -0
  15. data/ext/java/nokogiri/NokogiriService.java +466 -0
  16. data/ext/java/nokogiri/XmlAttr.java +183 -0
  17. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  18. data/ext/java/nokogiri/XmlCdata.java +89 -0
  19. data/ext/java/nokogiri/XmlComment.java +84 -0
  20. data/ext/java/nokogiri/XmlDocument.java +514 -0
  21. data/ext/java/nokogiri/XmlDocumentFragment.java +216 -0
  22. data/ext/java/nokogiri/XmlDtd.java +464 -0
  23. data/ext/java/nokogiri/XmlElement.java +221 -0
  24. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  25. data/ext/java/nokogiri/XmlElementDecl.java +147 -0
  26. data/ext/java/nokogiri/XmlEntityDecl.java +161 -0
  27. data/ext/java/nokogiri/XmlEntityReference.java +75 -0
  28. data/ext/java/nokogiri/XmlNamespace.java +127 -0
  29. data/ext/java/nokogiri/XmlNode.java +1392 -0
  30. data/ext/java/nokogiri/XmlNodeSet.java +284 -0
  31. data/ext/java/nokogiri/XmlProcessingInstruction.java +103 -0
  32. data/ext/java/nokogiri/XmlReader.java +409 -0
  33. data/ext/java/nokogiri/XmlRelaxng.java +199 -0
  34. data/ext/java/nokogiri/XmlSaxParserContext.java +353 -0
  35. data/ext/java/nokogiri/XmlSaxPushParser.java +182 -0
  36. data/ext/java/nokogiri/XmlSchema.java +175 -0
  37. data/ext/java/nokogiri/XmlSyntaxError.java +114 -0
  38. data/ext/java/nokogiri/XmlText.java +135 -0
  39. data/ext/java/nokogiri/XmlXpathContext.java +175 -0
  40. data/ext/java/nokogiri/XsltStylesheet.java +181 -0
  41. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +205 -0
  42. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  43. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +80 -0
  44. data/ext/java/nokogiri/internals/NokogiriHandler.java +326 -0
  45. data/ext/java/nokogiri/internals/NokogiriHelpers.java +583 -0
  46. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +170 -0
  47. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +118 -0
  48. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +73 -0
  49. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  50. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
  51. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +120 -0
  52. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +56 -0
  53. data/ext/java/nokogiri/internals/ParserContext.java +278 -0
  54. data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
  55. data/ext/java/nokogiri/internals/ReaderNode.java +473 -0
  56. data/ext/java/nokogiri/internals/SaveContext.java +282 -0
  57. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +68 -0
  58. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  59. data/ext/java/nokogiri/internals/XmlDomParser.java +77 -0
  60. data/ext/java/nokogiri/internals/XmlDomParserContext.java +233 -0
  61. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  62. data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
  63. data/ext/nokogiri/depend +358 -0
  64. data/ext/nokogiri/extconf.rb +124 -0
  65. data/ext/nokogiri/html_document.c +154 -0
  66. data/ext/nokogiri/html_document.h +10 -0
  67. data/ext/nokogiri/html_element_description.c +276 -0
  68. data/ext/nokogiri/html_element_description.h +10 -0
  69. data/ext/nokogiri/html_entity_lookup.c +32 -0
  70. data/ext/nokogiri/html_entity_lookup.h +8 -0
  71. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  72. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  73. data/ext/nokogiri/nokogiri.c +92 -0
  74. data/ext/nokogiri/nokogiri.h +160 -0
  75. data/ext/nokogiri/xml_attr.c +94 -0
  76. data/ext/nokogiri/xml_attr.h +9 -0
  77. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  78. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  79. data/ext/nokogiri/xml_cdata.c +56 -0
  80. data/ext/nokogiri/xml_cdata.h +9 -0
  81. data/ext/nokogiri/xml_comment.c +54 -0
  82. data/ext/nokogiri/xml_comment.h +9 -0
  83. data/ext/nokogiri/xml_document.c +478 -0
  84. data/ext/nokogiri/xml_document.h +23 -0
  85. data/ext/nokogiri/xml_document_fragment.c +48 -0
  86. data/ext/nokogiri/xml_document_fragment.h +10 -0
  87. data/ext/nokogiri/xml_dtd.c +202 -0
  88. data/ext/nokogiri/xml_dtd.h +10 -0
  89. data/ext/nokogiri/xml_element_content.c +123 -0
  90. data/ext/nokogiri/xml_element_content.h +10 -0
  91. data/ext/nokogiri/xml_element_decl.c +69 -0
  92. data/ext/nokogiri/xml_element_decl.h +9 -0
  93. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  94. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  95. data/ext/nokogiri/xml_entity_decl.c +110 -0
  96. data/ext/nokogiri/xml_entity_decl.h +10 -0
  97. data/ext/nokogiri/xml_entity_reference.c +52 -0
  98. data/ext/nokogiri/xml_entity_reference.h +9 -0
  99. data/ext/nokogiri/xml_io.c +31 -0
  100. data/ext/nokogiri/xml_io.h +11 -0
  101. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  102. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  103. data/ext/nokogiri/xml_namespace.c +84 -0
  104. data/ext/nokogiri/xml_namespace.h +13 -0
  105. data/ext/nokogiri/xml_node.c +1384 -0
  106. data/ext/nokogiri/xml_node.h +13 -0
  107. data/ext/nokogiri/xml_node_set.c +418 -0
  108. data/ext/nokogiri/xml_node_set.h +9 -0
  109. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  110. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  111. data/ext/nokogiri/xml_reader.c +684 -0
  112. data/ext/nokogiri/xml_reader.h +10 -0
  113. data/ext/nokogiri/xml_relax_ng.c +161 -0
  114. data/ext/nokogiri/xml_relax_ng.h +9 -0
  115. data/ext/nokogiri/xml_sax_parser.c +288 -0
  116. data/ext/nokogiri/xml_sax_parser.h +39 -0
  117. data/ext/nokogiri/xml_sax_parser_context.c +199 -0
  118. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  119. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  120. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  121. data/ext/nokogiri/xml_schema.c +205 -0
  122. data/ext/nokogiri/xml_schema.h +9 -0
  123. data/ext/nokogiri/xml_syntax_error.c +58 -0
  124. data/ext/nokogiri/xml_syntax_error.h +13 -0
  125. data/ext/nokogiri/xml_text.c +50 -0
  126. data/ext/nokogiri/xml_text.h +9 -0
  127. data/ext/nokogiri/xml_xpath_context.c +309 -0
  128. data/ext/nokogiri/xml_xpath_context.h +9 -0
  129. data/ext/nokogiri/xslt_stylesheet.c +258 -0
  130. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  131. data/lib/isorelax.jar +0 -0
  132. data/lib/jing.jar +0 -0
  133. data/lib/nekodtd.jar +0 -0
  134. data/lib/nekohtml.jar +0 -0
  135. data/lib/nokogiri.rb +143 -0
  136. data/lib/nokogiri/css.rb +23 -0
  137. data/lib/nokogiri/css/node.rb +99 -0
  138. data/lib/nokogiri/css/parser.rb +677 -0
  139. data/lib/nokogiri/css/parser.y +237 -0
  140. data/lib/nokogiri/css/parser_extras.rb +91 -0
  141. data/lib/nokogiri/css/syntax_error.rb +7 -0
  142. data/lib/nokogiri/css/tokenizer.rb +152 -0
  143. data/lib/nokogiri/css/tokenizer.rex +55 -0
  144. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  145. data/lib/nokogiri/decorators/slop.rb +35 -0
  146. data/lib/nokogiri/html.rb +36 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +221 -0
  149. data/lib/nokogiri/html/document_fragment.rb +41 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  152. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  153. data/lib/nokogiri/html/sax/parser.rb +52 -0
  154. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  155. data/lib/nokogiri/syntax_error.rb +4 -0
  156. data/lib/nokogiri/version.rb +35 -0
  157. data/lib/nokogiri/xml.rb +67 -0
  158. data/lib/nokogiri/xml/attr.rb +14 -0
  159. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  160. data/lib/nokogiri/xml/builder.rb +418 -0
  161. data/lib/nokogiri/xml/cdata.rb +11 -0
  162. data/lib/nokogiri/xml/character_data.rb +7 -0
  163. data/lib/nokogiri/xml/document.rb +218 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +84 -0
  165. data/lib/nokogiri/xml/dtd.rb +22 -0
  166. data/lib/nokogiri/xml/element_content.rb +36 -0
  167. data/lib/nokogiri/xml/element_decl.rb +13 -0
  168. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  169. data/lib/nokogiri/xml/namespace.rb +13 -0
  170. data/lib/nokogiri/xml/node.rb +907 -0
  171. data/lib/nokogiri/xml/node/save_options.rb +45 -0
  172. data/lib/nokogiri/xml/node_set.rb +350 -0
  173. data/lib/nokogiri/xml/notation.rb +6 -0
  174. data/lib/nokogiri/xml/parse_options.rb +85 -0
  175. data/lib/nokogiri/xml/pp.rb +2 -0
  176. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  177. data/lib/nokogiri/xml/pp/node.rb +56 -0
  178. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  179. data/lib/nokogiri/xml/reader.rb +112 -0
  180. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  181. data/lib/nokogiri/xml/sax.rb +4 -0
  182. data/lib/nokogiri/xml/sax/document.rb +164 -0
  183. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  184. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  185. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  186. data/lib/nokogiri/xml/schema.rb +57 -0
  187. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  188. data/lib/nokogiri/xml/text.rb +9 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +52 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xercesImpl.jar +0 -0
  195. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  196. data/tasks/cross_compile.rb +177 -0
  197. data/tasks/test.rb +94 -0
  198. data/test/css/test_nthiness.rb +159 -0
  199. data/test/css/test_parser.rb +303 -0
  200. data/test/css/test_tokenizer.rb +198 -0
  201. data/test/css/test_xpath_visitor.rb +85 -0
  202. data/test/decorators/test_slop.rb +16 -0
  203. data/test/files/2ch.html +108 -0
  204. data/test/files/address_book.rlx +12 -0
  205. data/test/files/address_book.xml +10 -0
  206. data/test/files/bar/bar.xsd +4 -0
  207. data/test/files/dont_hurt_em_why.xml +422 -0
  208. data/test/files/exslt.xml +8 -0
  209. data/test/files/exslt.xslt +35 -0
  210. data/test/files/foo/foo.xsd +4 -0
  211. data/test/files/po.xml +32 -0
  212. data/test/files/po.xsd +66 -0
  213. data/test/files/shift_jis.html +10 -0
  214. data/test/files/shift_jis.xml +5 -0
  215. data/test/files/snuggles.xml +3 -0
  216. data/test/files/staff.dtd +10 -0
  217. data/test/files/staff.xml +59 -0
  218. data/test/files/staff.xslt +32 -0
  219. data/test/files/tlm.html +850 -0
  220. data/test/files/valid_bar.xml +2 -0
  221. data/test/helper.rb +171 -0
  222. data/test/html/sax/test_parser.rb +136 -0
  223. data/test/html/sax/test_parser_context.rb +48 -0
  224. data/test/html/test_builder.rb +164 -0
  225. data/test/html/test_document.rb +457 -0
  226. data/test/html/test_document_encoding.rb +123 -0
  227. data/test/html/test_document_fragment.rb +255 -0
  228. data/test/html/test_element_description.rb +100 -0
  229. data/test/html/test_named_characters.rb +14 -0
  230. data/test/html/test_node.rb +190 -0
  231. data/test/html/test_node_encoding.rb +27 -0
  232. data/test/test_convert_xpath.rb +135 -0
  233. data/test/test_css_cache.rb +45 -0
  234. data/test/test_encoding_handler.rb +46 -0
  235. data/test/test_memory_leak.rb +52 -0
  236. data/test/test_nokogiri.rb +132 -0
  237. data/test/test_reader.rb +403 -0
  238. data/test/test_soap4r_sax.rb +52 -0
  239. data/test/test_xslt_transforms.rb +189 -0
  240. data/test/xml/node/test_save_options.rb +20 -0
  241. data/test/xml/node/test_subclass.rb +44 -0
  242. data/test/xml/sax/test_parser.rb +338 -0
  243. data/test/xml/sax/test_parser_context.rb +113 -0
  244. data/test/xml/sax/test_push_parser.rb +156 -0
  245. data/test/xml/test_attr.rb +65 -0
  246. data/test/xml/test_attribute_decl.rb +86 -0
  247. data/test/xml/test_builder.rb +210 -0
  248. data/test/xml/test_cdata.rb +50 -0
  249. data/test/xml/test_comment.rb +29 -0
  250. data/test/xml/test_document.rb +675 -0
  251. data/test/xml/test_document_encoding.rb +26 -0
  252. data/test/xml/test_document_fragment.rb +192 -0
  253. data/test/xml/test_dtd.rb +107 -0
  254. data/test/xml/test_dtd_encoding.rb +33 -0
  255. data/test/xml/test_element_content.rb +56 -0
  256. data/test/xml/test_element_decl.rb +73 -0
  257. data/test/xml/test_entity_decl.rb +122 -0
  258. data/test/xml/test_entity_reference.rb +21 -0
  259. data/test/xml/test_namespace.rb +70 -0
  260. data/test/xml/test_node.rb +899 -0
  261. data/test/xml/test_node_attributes.rb +34 -0
  262. data/test/xml/test_node_encoding.rb +107 -0
  263. data/test/xml/test_node_reparenting.rb +321 -0
  264. data/test/xml/test_node_set.rb +708 -0
  265. data/test/xml/test_parse_options.rb +52 -0
  266. data/test/xml/test_processing_instruction.rb +30 -0
  267. data/test/xml/test_reader_encoding.rb +126 -0
  268. data/test/xml/test_relax_ng.rb +60 -0
  269. data/test/xml/test_schema.rb +89 -0
  270. data/test/xml/test_syntax_error.rb +12 -0
  271. data/test/xml/test_text.rb +47 -0
  272. data/test/xml/test_unparented_node.rb +381 -0
  273. data/test/xml/test_xpath.rb +237 -0
  274. data/test/xslt/test_custom_functions.rb +94 -0
  275. metadata +525 -0
@@ -0,0 +1,10 @@
1
+ #ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
2
+ #define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_element_description();
7
+
8
+ extern VALUE cNokogiriHtmlElementDescription ;
9
+
10
+ #endif
@@ -0,0 +1,32 @@
1
+ #include <html_entity_lookup.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * get(key)
6
+ *
7
+ * Get the HTML::EntityDescription for +key+
8
+ */
9
+ static VALUE get(VALUE self, VALUE key)
10
+ {
11
+ const htmlEntityDesc * desc =
12
+ htmlEntityLookup((const xmlChar *)StringValuePtr(key));
13
+ VALUE klass, args[3];
14
+
15
+ if(NULL == desc) return Qnil;
16
+ klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
17
+
18
+ args[0] = INT2NUM((long)desc->value);
19
+ args[1] = NOKOGIRI_STR_NEW2(desc->name);
20
+ args[2] = NOKOGIRI_STR_NEW2(desc->desc);
21
+
22
+ return rb_class_new_instance(3, args, klass);
23
+ }
24
+
25
+ void init_html_entity_lookup()
26
+ {
27
+ VALUE nokogiri = rb_define_module("Nokogiri");
28
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
29
+ VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
30
+
31
+ rb_define_method(klass, "get", get, 1);
32
+ }
@@ -0,0 +1,8 @@
1
+ #ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
2
+ #define NOKOGIRI_HTML_ENTITY_LOOKUP
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_entity_lookup();
7
+
8
+ #endif
@@ -0,0 +1,94 @@
1
+ #include <html_sax_parser_context.h>
2
+
3
+ VALUE cNokogiriHtmlSaxParserContext ;
4
+
5
+ static void deallocate(xmlParserCtxtPtr ctxt)
6
+ {
7
+ NOKOGIRI_DEBUG_START(handler);
8
+
9
+ ctxt->sax = NULL;
10
+
11
+ htmlFreeParserCtxt(ctxt);
12
+
13
+ NOKOGIRI_DEBUG_END(handler);
14
+ }
15
+
16
+ static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
17
+ {
18
+ htmlParserCtxtPtr ctxt;
19
+
20
+ if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
21
+ if(!(int)RSTRING_LEN(data))
22
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
23
+
24
+ ctxt = htmlCreateMemoryParserCtxt(
25
+ StringValuePtr(data),
26
+ (int)RSTRING_LEN(data)
27
+ );
28
+
29
+ if(RTEST(encoding)) {
30
+ xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValuePtr(encoding));
31
+ if(enc != NULL) {
32
+ xmlSwitchToEncoding(ctxt, enc);
33
+ if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
34
+ rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
35
+ StringValuePtr(encoding));
36
+ }
37
+ }
38
+ }
39
+
40
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
41
+ }
42
+
43
+ static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
44
+ {
45
+ htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
46
+ StringValuePtr(filename),
47
+ StringValuePtr(encoding)
48
+ );
49
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
50
+ }
51
+
52
+ static VALUE parse_with(VALUE self, VALUE sax_handler)
53
+ {
54
+ htmlParserCtxtPtr ctxt;
55
+ htmlSAXHandlerPtr sax;
56
+
57
+ if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
58
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
59
+
60
+ Data_Get_Struct(self, htmlParserCtxt, ctxt);
61
+ Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
62
+
63
+ /* Free the sax handler since we'll assign our own */
64
+ if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
65
+ xmlFree(ctxt->sax);
66
+
67
+ ctxt->sax = sax;
68
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
69
+
70
+ htmlParseDocument(ctxt);
71
+
72
+ if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
73
+
74
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
75
+ return self;
76
+ }
77
+
78
+ void init_html_sax_parser_context()
79
+ {
80
+ VALUE nokogiri = rb_define_module("Nokogiri");
81
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
82
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
83
+ VALUE sax = rb_define_module_under(xml, "SAX");
84
+ VALUE hsax = rb_define_module_under(html, "SAX");
85
+ VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
86
+ VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
87
+
88
+ cNokogiriHtmlSaxParserContext = klass;
89
+
90
+ rb_define_singleton_method(klass, "memory", parse_memory, 2);
91
+ rb_define_singleton_method(klass, "file", parse_file, 2);
92
+
93
+ rb_define_method(klass, "parse_with", parse_with, 1);
94
+ }
@@ -0,0 +1,11 @@
1
+ #ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
2
+ #define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ extern VALUE cNokogiriHtmlSaxParserContext;
7
+
8
+ void init_html_sax_parser_context();
9
+
10
+ #endif
11
+
@@ -0,0 +1,92 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE mNokogiri ;
4
+ VALUE mNokogiriXml ;
5
+ VALUE mNokogiriHtml ;
6
+ VALUE mNokogiriXslt ;
7
+ VALUE mNokogiriXmlSax ;
8
+ VALUE mNokogiriHtmlSax ;
9
+
10
+ #ifdef USE_INCLUDED_VASPRINTF
11
+ /*
12
+ * I srsly hate windows. it doesn't have vasprintf.
13
+ * Thank you Geoffroy Couprie for this implementation of vasprintf!
14
+ */
15
+ int vasprintf (char **strp, const char *fmt, va_list ap)
16
+ {
17
+ int len = vsnprintf (NULL, 0, fmt, ap) + 1;
18
+ char *res = (char *)malloc((unsigned int)len);
19
+ if (res == NULL)
20
+ return -1;
21
+ *strp = res;
22
+ return vsnprintf(res, (unsigned int)len, fmt, ap);
23
+ }
24
+ #endif
25
+
26
+ void Init_nokogiri()
27
+ {
28
+ #ifndef __MACRUBY__
29
+ xmlMemSetup(
30
+ (xmlFreeFunc)ruby_xfree,
31
+ (xmlMallocFunc)ruby_xmalloc,
32
+ (xmlReallocFunc)ruby_xrealloc,
33
+ strdup
34
+ );
35
+ #endif
36
+
37
+ mNokogiri = rb_define_module("Nokogiri");
38
+ mNokogiriXml = rb_define_module_under(mNokogiri, "XML");
39
+ mNokogiriHtml = rb_define_module_under(mNokogiri, "HTML");
40
+ mNokogiriXslt = rb_define_module_under(mNokogiri, "XSLT");
41
+ mNokogiriXmlSax = rb_define_module_under(mNokogiriXml, "SAX");
42
+ mNokogiriHtmlSax = rb_define_module_under(mNokogiriHtml, "SAX");
43
+
44
+ rb_const_set( mNokogiri,
45
+ rb_intern("LIBXML_VERSION"),
46
+ NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
47
+ );
48
+ rb_const_set( mNokogiri,
49
+ rb_intern("LIBXML_PARSER_VERSION"),
50
+ NOKOGIRI_STR_NEW2(xmlParserVersion)
51
+ );
52
+
53
+ #ifdef LIBXML_ICONV_ENABLED
54
+ rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
55
+ #else
56
+ rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
57
+ #endif
58
+
59
+ xmlInitParser();
60
+
61
+ init_xml_document();
62
+ init_html_document();
63
+ init_xml_node();
64
+ init_xml_document_fragment();
65
+ init_xml_text();
66
+ init_xml_cdata();
67
+ init_xml_processing_instruction();
68
+ init_xml_attr();
69
+ init_xml_entity_reference();
70
+ init_xml_comment();
71
+ init_xml_node_set();
72
+ init_xml_xpath_context();
73
+ init_xml_sax_parser_context();
74
+ init_xml_sax_parser();
75
+ init_xml_sax_push_parser();
76
+ init_xml_reader();
77
+ init_xml_dtd();
78
+ init_xml_element_content();
79
+ init_xml_attribute_decl();
80
+ init_xml_element_decl();
81
+ init_xml_entity_decl();
82
+ init_xml_namespace();
83
+ init_html_sax_parser_context();
84
+ init_xslt_stylesheet();
85
+ init_xml_syntax_error();
86
+ init_html_entity_lookup();
87
+ init_html_element_description();
88
+ init_xml_schema();
89
+ init_xml_relax_ng();
90
+ init_nokogiri_io();
91
+ init_xml_encoding_handler();
92
+ }
@@ -0,0 +1,160 @@
1
+ #ifndef NOKOGIRI_NATIVE
2
+ #define NOKOGIRI_NATIVE
3
+
4
+ #include <stdlib.h>
5
+ #include <string.h>
6
+ #include <assert.h>
7
+ #include <stdarg.h>
8
+
9
+ #ifdef USE_INCLUDED_VASPRINTF
10
+ int vasprintf (char **strp, const char *fmt, va_list ap);
11
+ #else
12
+
13
+ #define _GNU_SOURCE
14
+ # include <stdio.h>
15
+ #undef _GNU_SOURCE
16
+
17
+ #endif
18
+
19
+ #include <libxml/parser.h>
20
+ #include <libxml/entities.h>
21
+ #include <libxml/parserInternals.h>
22
+ #include <libxml/xpath.h>
23
+ #include <libxml/xpathInternals.h>
24
+ #include <libxml/xmlreader.h>
25
+ #include <libxml/xmlsave.h>
26
+ #include <libxml/xmlschemas.h>
27
+ #include <libxml/HTMLparser.h>
28
+ #include <libxml/HTMLtree.h>
29
+ #include <libxml/relaxng.h>
30
+ #include <libxslt/extensions.h>
31
+ #include <ruby.h>
32
+
33
+ #ifdef HAVE_RUBY_ENCODING_H
34
+ #include <ruby/st.h>
35
+ #else
36
+ #include <st.h>
37
+ #endif
38
+
39
+ #ifndef UNUSED
40
+ # if defined(__GNUC__)
41
+ # define MAYBE_UNUSED(name) name __attribute__((unused))
42
+ # define UNUSED(name) MAYBE_UNUSED(UNUSED_ ## name)
43
+ # else
44
+ # define MAYBE_UNUSED(name) name
45
+ # define UNUSED(name) name
46
+ # endif
47
+ #endif
48
+
49
+ #ifndef NORETURN
50
+ # if defined(__GNUC__)
51
+ # define NORETURN(name) __attribute__((noreturn)) name
52
+ # else
53
+ # define NORETURN(name) name
54
+ # endif
55
+ #endif
56
+
57
+ #ifdef HAVE_RUBY_ENCODING_H
58
+
59
+ #include <ruby/encoding.h>
60
+
61
+ #define NOKOGIRI_STR_NEW2(str) \
62
+ NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
63
+
64
+ #define NOKOGIRI_STR_NEW(str, len) \
65
+ rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
66
+
67
+ #else
68
+
69
+ #define NOKOGIRI_STR_NEW2(str) \
70
+ rb_str_new2((const char *)(str))
71
+
72
+ #define NOKOGIRI_STR_NEW(str, len) \
73
+ rb_str_new((const char *)(str), (long)(len))
74
+ #endif
75
+
76
+ #define RBSTR_OR_QNIL(_str) \
77
+ (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
78
+
79
+ #include <xml_libxml2_hacks.h>
80
+
81
+ #include <xml_io.h>
82
+ #include <xml_document.h>
83
+ #include <html_entity_lookup.h>
84
+ #include <html_document.h>
85
+ #include <xml_node.h>
86
+ #include <xml_text.h>
87
+ #include <xml_cdata.h>
88
+ #include <xml_attr.h>
89
+ #include <xml_processing_instruction.h>
90
+ #include <xml_entity_reference.h>
91
+ #include <xml_document_fragment.h>
92
+ #include <xml_comment.h>
93
+ #include <xml_node_set.h>
94
+ #include <xml_dtd.h>
95
+ #include <xml_attribute_decl.h>
96
+ #include <xml_element_decl.h>
97
+ #include <xml_entity_decl.h>
98
+ #include <xml_xpath_context.h>
99
+ #include <xml_element_content.h>
100
+ #include <xml_sax_parser_context.h>
101
+ #include <xml_sax_parser.h>
102
+ #include <xml_sax_push_parser.h>
103
+ #include <xml_reader.h>
104
+ #include <html_sax_parser_context.h>
105
+ #include <xslt_stylesheet.h>
106
+ #include <xml_syntax_error.h>
107
+ #include <xml_schema.h>
108
+ #include <xml_relax_ng.h>
109
+ #include <html_element_description.h>
110
+ #include <xml_namespace.h>
111
+ #include <xml_encoding_handler.h>
112
+
113
+ extern VALUE mNokogiri ;
114
+ extern VALUE mNokogiriXml ;
115
+ extern VALUE mNokogiriXmlSax ;
116
+ extern VALUE mNokogiriHtml ;
117
+ extern VALUE mNokogiriHtmlSax ;
118
+ extern VALUE mNokogiriXslt ;
119
+
120
+ #define NOKOGIRI_ROOT_NODE(_node) \
121
+ st_insert(((nokogiriTuplePtr)(_node)->doc->_private)->unlinkedNodes, (st_data_t)(_node), (st_data_t)(_node))
122
+
123
+ #define NOKOGIRI_ROOT_NSDEF(_nsDef, _doc) \
124
+ st_insert(((nokogiriTuplePtr)(_doc)->_private)->unlinkedNodes, (st_data_t)(_nsDef), (st_data_t)(_nsDef))
125
+
126
+ #ifdef DEBUG
127
+
128
+ #define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
129
+ #define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
130
+
131
+ #else
132
+
133
+ #define NOKOGIRI_DEBUG_START(p)
134
+ #define NOKOGIRI_DEBUG_END(p)
135
+
136
+ #endif
137
+
138
+ #ifndef RSTRING_PTR
139
+ #define RSTRING_PTR(s) (RSTRING(s)->ptr)
140
+ #endif
141
+
142
+ #ifndef RSTRING_LEN
143
+ #define RSTRING_LEN(s) (RSTRING(s)->len)
144
+ #endif
145
+
146
+ #ifndef RARRAY_PTR
147
+ #define RARRAY_PTR(a) RARRAY(a)->ptr
148
+ #endif
149
+
150
+ #ifndef RARRAY_LEN
151
+ #define RARRAY_LEN(a) RARRAY(a)->len
152
+ #endif
153
+
154
+ #ifndef __builtin_expect
155
+ # if defined(__GNUC__)
156
+ # define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
157
+ # endif
158
+ #endif
159
+
160
+ #endif
@@ -0,0 +1,94 @@
1
+ #include <xml_attr.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * value=(content)
6
+ *
7
+ * Set the value for this Attr to +content+
8
+ */
9
+ static VALUE set_value(VALUE self, VALUE content)
10
+ {
11
+ xmlAttrPtr attr;
12
+ Data_Get_Struct(self, xmlAttr, attr);
13
+
14
+ if(attr->children) xmlFreeNodeList(attr->children);
15
+
16
+ attr->children = attr->last = NULL;
17
+
18
+ if(content) {
19
+ xmlChar *buffer;
20
+ xmlNode *tmp;
21
+
22
+ /* Encode our content */
23
+ buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
24
+
25
+ attr->children = xmlStringGetNodeList(attr->doc, buffer);
26
+ attr->last = NULL;
27
+ tmp = attr->children;
28
+
29
+ /* Loop through the children */
30
+ for(tmp = attr->children; tmp; tmp = tmp->next) {
31
+ tmp->parent = (xmlNode *)attr;
32
+ tmp->doc = attr->doc;
33
+ if(tmp->next == NULL) attr->last = tmp;
34
+ }
35
+
36
+ /* Free up memory */
37
+ xmlFree(buffer);
38
+ }
39
+
40
+ return content;
41
+ }
42
+
43
+ /*
44
+ * call-seq:
45
+ * new(document, name)
46
+ *
47
+ * Create a new Attr element on the +document+ with +name+
48
+ */
49
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
50
+ {
51
+ xmlDocPtr xml_doc;
52
+ VALUE document;
53
+ VALUE name;
54
+ VALUE rest;
55
+ xmlAttrPtr node;
56
+ VALUE rb_node;
57
+
58
+ rb_scan_args(argc, argv, "2*", &document, &name, &rest);
59
+
60
+ Data_Get_Struct(document, xmlDoc, xml_doc);
61
+
62
+ node = xmlNewDocProp(
63
+ xml_doc,
64
+ (const xmlChar *)StringValuePtr(name),
65
+ NULL
66
+ );
67
+
68
+ NOKOGIRI_ROOT_NODE((xmlNodePtr)node);
69
+
70
+ rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
71
+ rb_obj_call_init(rb_node, argc, argv);
72
+
73
+ if(rb_block_given_p()) rb_yield(rb_node);
74
+
75
+ return rb_node;
76
+ }
77
+
78
+ VALUE cNokogiriXmlAttr;
79
+ void init_xml_attr()
80
+ {
81
+ VALUE nokogiri = rb_define_module("Nokogiri");
82
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
83
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
84
+
85
+ /*
86
+ * Attr represents a Attr node in an xml document.
87
+ */
88
+ VALUE klass = rb_define_class_under(xml, "Attr", node);
89
+
90
+ cNokogiriXmlAttr = klass;
91
+
92
+ rb_define_singleton_method(klass, "new", new, -1);
93
+ rb_define_method(klass, "value=", set_value, 1);
94
+ }