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 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_ATTR
2
+ #define NOKOGIRI_XML_ATTR
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_attr();
7
+
8
+ extern VALUE cNokogiriXmlAttr;
9
+ #endif
@@ -0,0 +1,70 @@
1
+ #include <xml_attribute_decl.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * attribute_type
6
+ *
7
+ * The attribute_type for this AttributeDecl
8
+ */
9
+ static VALUE attribute_type(VALUE self)
10
+ {
11
+ xmlAttributePtr node;
12
+ Data_Get_Struct(self, xmlAttribute, node);
13
+ return INT2NUM((long)node->atype);
14
+ }
15
+
16
+ /*
17
+ * call-seq:
18
+ * default
19
+ *
20
+ * The default value
21
+ */
22
+ static VALUE default_value(VALUE self)
23
+ {
24
+ xmlAttributePtr node;
25
+ Data_Get_Struct(self, xmlAttribute, node);
26
+
27
+ if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
28
+ return Qnil;
29
+ }
30
+
31
+ /*
32
+ * call-seq:
33
+ * enumeration
34
+ *
35
+ * An enumeration of possible values
36
+ */
37
+ static VALUE enumeration(VALUE self)
38
+ {
39
+ xmlAttributePtr node;
40
+ xmlEnumerationPtr enm;
41
+ VALUE list;
42
+
43
+ Data_Get_Struct(self, xmlAttribute, node);
44
+
45
+ list = rb_ary_new();
46
+ enm = node->tree;
47
+
48
+ while(enm) {
49
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
50
+ enm = enm->next;
51
+ }
52
+
53
+ return list;
54
+ }
55
+
56
+ VALUE cNokogiriXmlAttributeDecl;
57
+
58
+ void init_xml_attribute_decl()
59
+ {
60
+ VALUE nokogiri = rb_define_module("Nokogiri");
61
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
62
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
63
+ VALUE klass = rb_define_class_under(xml, "AttributeDecl", node);
64
+
65
+ cNokogiriXmlAttributeDecl = klass;
66
+
67
+ rb_define_method(klass, "attribute_type", attribute_type, 0);
68
+ rb_define_method(klass, "default", default_value, 0);
69
+ rb_define_method(klass, "enumeration", enumeration, 0);
70
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_ATTRIBUTE_DECL
2
+ #define NOKOGIRI_XML_ATTRIBUTE_DECL
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_attribute_decl();
7
+
8
+ extern VALUE cNokogiriXmlAttributeDecl;
9
+ #endif
@@ -0,0 +1,56 @@
1
+ #include <xml_cdata.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * new(document, content)
6
+ *
7
+ * Create a new CDATA element on the +document+ with +content+
8
+ */
9
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
10
+ {
11
+ xmlDocPtr xml_doc;
12
+ xmlNodePtr node;
13
+ VALUE doc;
14
+ VALUE content;
15
+ VALUE rest;
16
+ VALUE rb_node;
17
+
18
+ rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
19
+
20
+ Data_Get_Struct(doc, xmlDoc, xml_doc);
21
+
22
+ node = xmlNewCDataBlock(
23
+ xml_doc->doc,
24
+ NIL_P(content) ? NULL : (const xmlChar *)StringValuePtr(content),
25
+ NIL_P(content) ? 0 : (int)RSTRING_LEN(content)
26
+ );
27
+
28
+ NOKOGIRI_ROOT_NODE(node);
29
+
30
+ rb_node = Nokogiri_wrap_xml_node(klass, node);
31
+ rb_obj_call_init(rb_node, argc, argv);
32
+
33
+ if(rb_block_given_p()) rb_yield(rb_node);
34
+
35
+ return rb_node;
36
+ }
37
+
38
+ VALUE cNokogiriXmlCData;
39
+ void init_xml_cdata()
40
+ {
41
+ VALUE nokogiri = rb_define_module("Nokogiri");
42
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
43
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
44
+ VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
45
+ VALUE text = rb_define_class_under(xml, "Text", char_data);
46
+
47
+ /*
48
+ * CData represents a CData node in an xml document.
49
+ */
50
+ VALUE klass = rb_define_class_under(xml, "CDATA", text);
51
+
52
+
53
+ cNokogiriXmlCData = klass;
54
+
55
+ rb_define_singleton_method(klass, "new", new, -1);
56
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_CDATA
2
+ #define NOKOGIRI_XML_CDATA
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_cdata();
7
+
8
+ extern VALUE cNokogiriXmlCData;
9
+ #endif
@@ -0,0 +1,54 @@
1
+ #include <xml_comment.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * new(document, content)
6
+ *
7
+ * Create a new Comment element on the +document+ with +content+
8
+ */
9
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
10
+ {
11
+ xmlDocPtr xml_doc;
12
+ xmlNodePtr node;
13
+ VALUE document;
14
+ VALUE content;
15
+ VALUE rest;
16
+ VALUE rb_node;
17
+
18
+ rb_scan_args(argc, argv, "2*", &document, &content, &rest);
19
+
20
+ Data_Get_Struct(document, xmlDoc, xml_doc);
21
+
22
+ node = xmlNewDocComment(
23
+ xml_doc,
24
+ (const xmlChar *)StringValuePtr(content)
25
+ );
26
+
27
+ rb_node = Nokogiri_wrap_xml_node(klass, node);
28
+ rb_obj_call_init(rb_node, argc, argv);
29
+
30
+ NOKOGIRI_ROOT_NODE(node);
31
+
32
+ if(rb_block_given_p()) rb_yield(rb_node);
33
+
34
+ return rb_node;
35
+ }
36
+
37
+ VALUE cNokogiriXmlComment;
38
+ void init_xml_comment()
39
+ {
40
+ VALUE nokogiri = rb_define_module("Nokogiri");
41
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
42
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
43
+ VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
44
+
45
+ /*
46
+ * Comment represents a comment node in an xml document.
47
+ */
48
+ VALUE klass = rb_define_class_under(xml, "Comment", char_data);
49
+
50
+
51
+ cNokogiriXmlComment = klass;
52
+
53
+ rb_define_singleton_method(klass, "new", new, -1);
54
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_COMMENT
2
+ #define NOKOGIRI_XML_COMMENT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_comment();
7
+
8
+ extern VALUE cNokogiriXmlComment;
9
+ #endif
@@ -0,0 +1,478 @@
1
+ #include <xml_document.h>
2
+
3
+ static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
4
+ {
5
+ switch(node->type) {
6
+ case XML_ATTRIBUTE_NODE:
7
+ xmlFreePropList((xmlAttrPtr)node);
8
+ break;
9
+ case XML_NAMESPACE_DECL:
10
+ xmlFree(node);
11
+ break;
12
+ default:
13
+ if(node->parent == NULL) {
14
+ xmlAddChild((xmlNodePtr)doc, node);
15
+ }
16
+ }
17
+ return ST_CONTINUE;
18
+ }
19
+
20
+ static void dealloc(xmlDocPtr doc)
21
+ {
22
+ xmlDeregisterNodeFunc func;
23
+ st_table *node_hash;
24
+
25
+ NOKOGIRI_DEBUG_START(doc);
26
+ func = xmlDeregisterNodeDefault(NULL);
27
+
28
+ node_hash = DOC_UNLINKED_NODE_HASH(doc);
29
+
30
+ st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
31
+ st_free_table(node_hash);
32
+
33
+ free(doc->_private);
34
+ doc->_private = NULL;
35
+ xmlFreeDoc(doc);
36
+
37
+ xmlDeregisterNodeDefault(func);
38
+ NOKOGIRI_DEBUG_END(doc);
39
+ }
40
+
41
+ static void recursively_remove_namespaces_from_node(xmlNodePtr node)
42
+ {
43
+ xmlNodePtr child ;
44
+ xmlAttrPtr property ;
45
+
46
+ xmlSetNs(node, NULL);
47
+
48
+ for (child = node->children ; child ; child = child->next)
49
+ recursively_remove_namespaces_from_node(child);
50
+
51
+ if (node->nsDef) {
52
+ xmlFreeNsList(node->nsDef);
53
+ node->nsDef = NULL;
54
+ }
55
+
56
+ if (node->properties != NULL) {
57
+ property = node->properties ;
58
+ while (property != NULL) {
59
+ if (property->ns) property->ns = NULL ;
60
+ property = property->next ;
61
+ }
62
+ }
63
+ }
64
+
65
+ /*
66
+ * call-seq:
67
+ * url
68
+ *
69
+ * Get the url name for this document.
70
+ */
71
+ static VALUE url(VALUE self)
72
+ {
73
+ xmlDocPtr doc;
74
+ Data_Get_Struct(self, xmlDoc, doc);
75
+
76
+ if(doc->URL) return NOKOGIRI_STR_NEW2(doc->URL);
77
+
78
+ return Qnil;
79
+ }
80
+
81
+ /*
82
+ * call-seq:
83
+ * root=
84
+ *
85
+ * Set the root element on this document
86
+ */
87
+ static VALUE set_root(VALUE self, VALUE root)
88
+ {
89
+ xmlDocPtr doc;
90
+ xmlNodePtr new_root;
91
+ xmlNodePtr old_root;
92
+
93
+ Data_Get_Struct(self, xmlDoc, doc);
94
+
95
+ old_root = NULL;
96
+
97
+ if(NIL_P(root)) {
98
+ old_root = xmlDocGetRootElement(doc);
99
+
100
+ if(old_root) {
101
+ xmlUnlinkNode(old_root);
102
+ NOKOGIRI_ROOT_NODE(old_root);
103
+ }
104
+
105
+ return root;
106
+ }
107
+
108
+ Data_Get_Struct(root, xmlNode, new_root);
109
+
110
+
111
+ /* If the new root's document is not the same as the current document,
112
+ * then we need to dup the node in to this document. */
113
+ if(new_root->doc != doc) {
114
+ old_root = xmlDocGetRootElement(doc);
115
+ if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
116
+ rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
117
+ }
118
+ }
119
+
120
+ xmlDocSetRootElement(doc, new_root);
121
+ if(old_root) NOKOGIRI_ROOT_NODE(old_root);
122
+ return root;
123
+ }
124
+
125
+ /*
126
+ * call-seq:
127
+ * root
128
+ *
129
+ * Get the root node for this document.
130
+ */
131
+ static VALUE root(VALUE self)
132
+ {
133
+ xmlDocPtr doc;
134
+ xmlNodePtr root;
135
+
136
+ Data_Get_Struct(self, xmlDoc, doc);
137
+
138
+ root = xmlDocGetRootElement(doc);
139
+
140
+ if(!root) return Qnil;
141
+ return Nokogiri_wrap_xml_node(Qnil, root) ;
142
+ }
143
+
144
+ /*
145
+ * call-seq:
146
+ * encoding= encoding
147
+ *
148
+ * Set the encoding string for this Document
149
+ */
150
+ static VALUE set_encoding(VALUE self, VALUE encoding)
151
+ {
152
+ xmlDocPtr doc;
153
+ Data_Get_Struct(self, xmlDoc, doc);
154
+
155
+ doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
156
+
157
+ return encoding;
158
+ }
159
+
160
+ /*
161
+ * call-seq:
162
+ * encoding
163
+ *
164
+ * Get the encoding for this Document
165
+ */
166
+ static VALUE encoding(VALUE self)
167
+ {
168
+ xmlDocPtr doc;
169
+ Data_Get_Struct(self, xmlDoc, doc);
170
+
171
+ if(!doc->encoding) return Qnil;
172
+ return NOKOGIRI_STR_NEW2(doc->encoding);
173
+ }
174
+
175
+ /*
176
+ * call-seq:
177
+ * version
178
+ *
179
+ * Get the XML version for this Document
180
+ */
181
+ static VALUE version(VALUE self)
182
+ {
183
+ xmlDocPtr doc;
184
+ Data_Get_Struct(self, xmlDoc, doc);
185
+
186
+ if(!doc->version) return Qnil;
187
+ return NOKOGIRI_STR_NEW2(doc->version);
188
+ }
189
+
190
+ /*
191
+ * call-seq:
192
+ * read_io(io, url, encoding, options)
193
+ *
194
+ * Create a new document from an IO object
195
+ */
196
+ static VALUE read_io( VALUE klass,
197
+ VALUE io,
198
+ VALUE url,
199
+ VALUE encoding,
200
+ VALUE options )
201
+ {
202
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
203
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
204
+ VALUE error_list = rb_ary_new();
205
+ VALUE document;
206
+ xmlDocPtr doc;
207
+
208
+ xmlResetLastError();
209
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
210
+
211
+ doc = xmlReadIO(
212
+ (xmlInputReadCallback)io_read_callback,
213
+ (xmlInputCloseCallback)io_close_callback,
214
+ (void *)io,
215
+ c_url,
216
+ c_enc,
217
+ (int)NUM2INT(options)
218
+ );
219
+ xmlSetStructuredErrorFunc(NULL, NULL);
220
+
221
+ if(doc == NULL) {
222
+ xmlErrorPtr error;
223
+
224
+ xmlFreeDoc(doc);
225
+
226
+ error = xmlGetLastError();
227
+ if(error)
228
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
229
+ else
230
+ rb_raise(rb_eRuntimeError, "Could not parse document");
231
+
232
+ return Qnil;
233
+ }
234
+
235
+ document = Nokogiri_wrap_xml_document(klass, doc);
236
+ rb_iv_set(document, "@errors", error_list);
237
+ return document;
238
+ }
239
+
240
+ /*
241
+ * call-seq:
242
+ * read_memory(string, url, encoding, options)
243
+ *
244
+ * Create a new document from a String
245
+ */
246
+ static VALUE read_memory( VALUE klass,
247
+ VALUE string,
248
+ VALUE url,
249
+ VALUE encoding,
250
+ VALUE options )
251
+ {
252
+ const char * c_buffer = StringValuePtr(string);
253
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
254
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
255
+ int len = (int)RSTRING_LEN(string);
256
+ VALUE error_list = rb_ary_new();
257
+ VALUE document;
258
+ xmlDocPtr doc;
259
+
260
+ xmlResetLastError();
261
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
262
+ doc = xmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
263
+ xmlSetStructuredErrorFunc(NULL, NULL);
264
+
265
+ if(doc == NULL) {
266
+ xmlErrorPtr error;
267
+
268
+ xmlFreeDoc(doc);
269
+
270
+ error = xmlGetLastError();
271
+ if(error)
272
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
273
+ else
274
+ rb_raise(rb_eRuntimeError, "Could not parse document");
275
+
276
+ return Qnil;
277
+ }
278
+
279
+ document = Nokogiri_wrap_xml_document(klass, doc);
280
+ rb_iv_set(document, "@errors", error_list);
281
+ return document;
282
+ }
283
+
284
+ /*
285
+ * call-seq:
286
+ * dup
287
+ *
288
+ * Copy this Document. An optional depth may be passed in, but it defaults
289
+ * to a deep copy. 0 is a shallow copy, 1 is a deep copy.
290
+ */
291
+ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
292
+ {
293
+ xmlDocPtr doc, dup;
294
+ VALUE level;
295
+
296
+ if(rb_scan_args(argc, argv, "01", &level) == 0)
297
+ level = INT2NUM((long)1);
298
+
299
+ Data_Get_Struct(self, xmlDoc, doc);
300
+
301
+ dup = xmlCopyDoc(doc, (int)NUM2INT(level));
302
+ if(dup == NULL) return Qnil;
303
+
304
+ dup->type = doc->type;
305
+ return Nokogiri_wrap_xml_document(rb_obj_class(self), dup);
306
+ }
307
+
308
+ /*
309
+ * call-seq:
310
+ * new(version = default)
311
+ *
312
+ * Create a new document with +version+ (defaults to "1.0")
313
+ */
314
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
315
+ {
316
+ xmlDocPtr doc;
317
+ VALUE version, rest, rb_doc ;
318
+
319
+ rb_scan_args(argc, argv, "0*", &rest);
320
+ version = rb_ary_entry(rest, (long)0);
321
+ if (NIL_P(version)) version = rb_str_new2("1.0");
322
+
323
+ doc = xmlNewDoc((xmlChar *)StringValuePtr(version));
324
+ rb_doc = Nokogiri_wrap_xml_document(klass, doc);
325
+ rb_obj_call_init(rb_doc, argc, argv);
326
+ return rb_doc ;
327
+ }
328
+
329
+ /*
330
+ * call-seq:
331
+ * remove_namespaces!
332
+ *
333
+ * Remove all namespaces from all nodes in the document.
334
+ *
335
+ * This could be useful for developers who either don't understand namespaces
336
+ * or don't care about them.
337
+ *
338
+ * The following example shows a use case, and you can decide for yourself
339
+ * whether this is a good thing or not:
340
+ *
341
+ * doc = Nokogiri::XML <<-EOXML
342
+ * <root>
343
+ * <car xmlns:part="http://general-motors.com/">
344
+ * <part:tire>Michelin Model XGV</part:tire>
345
+ * </car>
346
+ * <bicycle xmlns:part="http://schwinn.com/">
347
+ * <part:tire>I'm a bicycle tire!</part:tire>
348
+ * </bicycle>
349
+ * </root>
350
+ * EOXML
351
+ *
352
+ * doc.xpath("//tire").to_s # => ""
353
+ * doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => "<part:tire>Michelin Model XGV</part:tire>"
354
+ * doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => "<part:tire>I'm a bicycle tire!</part:tire>"
355
+ *
356
+ * doc.remove_namespaces!
357
+ *
358
+ * doc.xpath("//tire").to_s # => "<tire>Michelin Model XGV</tire><tire>I'm a bicycle tire!</tire>"
359
+ * doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => ""
360
+ * doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => ""
361
+ *
362
+ * For more information on why this probably is *not* a good thing in general,
363
+ * please direct your browser to
364
+ * http://tenderlovemaking.com/2009/04/23/namespaces-in-xml/
365
+ */
366
+ VALUE remove_namespaces_bang(VALUE self)
367
+ {
368
+ xmlDocPtr doc ;
369
+ Data_Get_Struct(self, xmlDoc, doc);
370
+
371
+ recursively_remove_namespaces_from_node((xmlNodePtr)doc);
372
+ return self;
373
+ }
374
+
375
+ /* call-seq: doc.create_entity(name, type, external_id, system_id, content)
376
+ *
377
+ * Create a new entity named +name+.
378
+ *
379
+ * +type+ is an integer representing the type of entity to be created, and it
380
+ * defaults to Nokogiri::XML::EntityDecl::INTERNAL_GENERAL. See
381
+ * the constants on Nokogiri::XML::EntityDecl for more information.
382
+ *
383
+ * +external_id+, +system_id+, and +content+ set the External ID, System ID,
384
+ * and content respectively. All of these parameters are optional.
385
+ */
386
+ static VALUE create_entity(int argc, VALUE *argv, VALUE self)
387
+ {
388
+ VALUE name;
389
+ VALUE type;
390
+ VALUE external_id;
391
+ VALUE system_id;
392
+ VALUE content;
393
+ xmlEntityPtr ptr;
394
+ xmlDocPtr doc ;
395
+
396
+ Data_Get_Struct(self, xmlDoc, doc);
397
+
398
+ rb_scan_args(argc, argv, "14", &name, &type, &external_id, &system_id,
399
+ &content);
400
+
401
+ xmlResetLastError();
402
+ ptr = xmlAddDocEntity(
403
+ doc,
404
+ (xmlChar *)(NIL_P(name) ? NULL : StringValuePtr(name)),
405
+ (int) (NIL_P(type) ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(type)),
406
+ (xmlChar *)(NIL_P(external_id) ? NULL : StringValuePtr(external_id)),
407
+ (xmlChar *)(NIL_P(system_id) ? NULL : StringValuePtr(system_id)),
408
+ (xmlChar *)(NIL_P(content) ? NULL : StringValuePtr(content))
409
+ );
410
+
411
+ if(NULL == ptr) {
412
+ xmlErrorPtr error = xmlGetLastError();
413
+ if(error)
414
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
415
+ else
416
+ rb_raise(rb_eRuntimeError, "Could not create entity");
417
+
418
+ return Qnil;
419
+ }
420
+
421
+ return Nokogiri_wrap_xml_node(cNokogiriXmlEntityDecl, (xmlNodePtr)ptr);
422
+ }
423
+
424
+ VALUE cNokogiriXmlDocument ;
425
+ void init_xml_document()
426
+ {
427
+ VALUE nokogiri = rb_define_module("Nokogiri");
428
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
429
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
430
+
431
+ /*
432
+ * Nokogiri::XML::Document wraps an xml document.
433
+ */
434
+ VALUE klass = rb_define_class_under(xml, "Document", node);
435
+
436
+ cNokogiriXmlDocument = klass;
437
+
438
+ rb_define_singleton_method(klass, "read_memory", read_memory, 4);
439
+ rb_define_singleton_method(klass, "read_io", read_io, 4);
440
+ rb_define_singleton_method(klass, "new", new, -1);
441
+
442
+ rb_define_method(klass, "root", root, 0);
443
+ rb_define_method(klass, "root=", set_root, 1);
444
+ rb_define_method(klass, "encoding", encoding, 0);
445
+ rb_define_method(klass, "encoding=", set_encoding, 1);
446
+ rb_define_method(klass, "version", version, 0);
447
+ rb_define_method(klass, "dup", duplicate_node, -1);
448
+ rb_define_method(klass, "url", url, 0);
449
+ rb_define_method(klass, "create_entity", create_entity, -1);
450
+ rb_define_method(klass, "remove_namespaces!", remove_namespaces_bang, 0);
451
+ }
452
+
453
+
454
+ /* this takes klass as a param because it's used for HtmlDocument, too. */
455
+ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
456
+ {
457
+ nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
458
+
459
+ VALUE rb_doc = Data_Wrap_Struct(
460
+ klass ? klass : cNokogiriXmlDocument,
461
+ 0,
462
+ dealloc,
463
+ doc
464
+ );
465
+
466
+ VALUE cache = rb_ary_new();
467
+ rb_iv_set(rb_doc, "@decorators", Qnil);
468
+ rb_iv_set(rb_doc, "@node_cache", cache);
469
+
470
+ tuple->doc = (void *)rb_doc;
471
+ tuple->unlinkedNodes = st_init_numtable_with_size(128);
472
+ tuple->node_cache = cache;
473
+ doc->_private = tuple ;
474
+
475
+ rb_obj_call_init(rb_doc, 0, NULL);
476
+
477
+ return rb_doc ;
478
+ }