superfeedr-nokogiri 1.4.0.20091116183308

Sign up to get free protection for your applications and to get access to all the features.
Files changed (270) hide show
  1. data/.autotest +27 -0
  2. data/CHANGELOG.ja.rdoc +330 -0
  3. data/CHANGELOG.rdoc +314 -0
  4. data/Manifest.txt +269 -0
  5. data/README.ja.rdoc +105 -0
  6. data/README.rdoc +118 -0
  7. data/Rakefile +244 -0
  8. data/bin/nokogiri +49 -0
  9. data/ext/nokogiri/extconf.rb +145 -0
  10. data/ext/nokogiri/html_document.c +145 -0
  11. data/ext/nokogiri/html_document.h +10 -0
  12. data/ext/nokogiri/html_element_description.c +272 -0
  13. data/ext/nokogiri/html_element_description.h +10 -0
  14. data/ext/nokogiri/html_entity_lookup.c +32 -0
  15. data/ext/nokogiri/html_entity_lookup.h +8 -0
  16. data/ext/nokogiri/html_sax_parser_context.c +92 -0
  17. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  18. data/ext/nokogiri/nokogiri.c +89 -0
  19. data/ext/nokogiri/nokogiri.h +145 -0
  20. data/ext/nokogiri/xml_attr.c +92 -0
  21. data/ext/nokogiri/xml_attr.h +9 -0
  22. data/ext/nokogiri/xml_attribute_decl.c +67 -0
  23. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  24. data/ext/nokogiri/xml_cdata.c +54 -0
  25. data/ext/nokogiri/xml_cdata.h +9 -0
  26. data/ext/nokogiri/xml_comment.c +52 -0
  27. data/ext/nokogiri/xml_comment.h +9 -0
  28. data/ext/nokogiri/xml_document.c +388 -0
  29. data/ext/nokogiri/xml_document.h +24 -0
  30. data/ext/nokogiri/xml_document_fragment.c +46 -0
  31. data/ext/nokogiri/xml_document_fragment.h +10 -0
  32. data/ext/nokogiri/xml_dtd.c +192 -0
  33. data/ext/nokogiri/xml_dtd.h +10 -0
  34. data/ext/nokogiri/xml_element_content.c +123 -0
  35. data/ext/nokogiri/xml_element_content.h +10 -0
  36. data/ext/nokogiri/xml_element_decl.c +69 -0
  37. data/ext/nokogiri/xml_element_decl.h +9 -0
  38. data/ext/nokogiri/xml_entity_decl.c +97 -0
  39. data/ext/nokogiri/xml_entity_decl.h +10 -0
  40. data/ext/nokogiri/xml_entity_reference.c +50 -0
  41. data/ext/nokogiri/xml_entity_reference.h +9 -0
  42. data/ext/nokogiri/xml_io.c +31 -0
  43. data/ext/nokogiri/xml_io.h +11 -0
  44. data/ext/nokogiri/xml_namespace.c +74 -0
  45. data/ext/nokogiri/xml_namespace.h +12 -0
  46. data/ext/nokogiri/xml_node.c +1060 -0
  47. data/ext/nokogiri/xml_node.h +13 -0
  48. data/ext/nokogiri/xml_node_set.c +397 -0
  49. data/ext/nokogiri/xml_node_set.h +9 -0
  50. data/ext/nokogiri/xml_processing_instruction.c +54 -0
  51. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  52. data/ext/nokogiri/xml_reader.c +593 -0
  53. data/ext/nokogiri/xml_reader.h +10 -0
  54. data/ext/nokogiri/xml_relax_ng.c +159 -0
  55. data/ext/nokogiri/xml_relax_ng.h +9 -0
  56. data/ext/nokogiri/xml_sax_parser.c +286 -0
  57. data/ext/nokogiri/xml_sax_parser.h +43 -0
  58. data/ext/nokogiri/xml_sax_parser_context.c +155 -0
  59. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  60. data/ext/nokogiri/xml_sax_push_parser.c +114 -0
  61. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  62. data/ext/nokogiri/xml_schema.c +156 -0
  63. data/ext/nokogiri/xml_schema.h +9 -0
  64. data/ext/nokogiri/xml_syntax_error.c +261 -0
  65. data/ext/nokogiri/xml_syntax_error.h +13 -0
  66. data/ext/nokogiri/xml_text.c +48 -0
  67. data/ext/nokogiri/xml_text.h +9 -0
  68. data/ext/nokogiri/xml_xpath.c +53 -0
  69. data/ext/nokogiri/xml_xpath.h +11 -0
  70. data/ext/nokogiri/xml_xpath_context.c +239 -0
  71. data/ext/nokogiri/xml_xpath_context.h +9 -0
  72. data/ext/nokogiri/xslt_stylesheet.c +131 -0
  73. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  74. data/lib/nokogiri.rb +116 -0
  75. data/lib/nokogiri/css.rb +25 -0
  76. data/lib/nokogiri/css/generated_parser.rb +646 -0
  77. data/lib/nokogiri/css/generated_tokenizer.rb +142 -0
  78. data/lib/nokogiri/css/node.rb +99 -0
  79. data/lib/nokogiri/css/parser.rb +82 -0
  80. data/lib/nokogiri/css/parser.y +227 -0
  81. data/lib/nokogiri/css/syntax_error.rb +7 -0
  82. data/lib/nokogiri/css/tokenizer.rb +7 -0
  83. data/lib/nokogiri/css/tokenizer.rex +54 -0
  84. data/lib/nokogiri/css/xpath_visitor.rb +162 -0
  85. data/lib/nokogiri/decorators/slop.rb +33 -0
  86. data/lib/nokogiri/ffi/html/document.rb +28 -0
  87. data/lib/nokogiri/ffi/html/element_description.rb +85 -0
  88. data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
  89. data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
  90. data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
  91. data/lib/nokogiri/ffi/libxml.rb +356 -0
  92. data/lib/nokogiri/ffi/structs/common_node.rb +26 -0
  93. data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
  94. data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
  95. data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
  96. data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
  97. data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
  98. data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
  99. data/lib/nokogiri/ffi/structs/xml_document.rb +108 -0
  100. data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
  101. data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
  102. data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
  103. data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
  104. data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
  105. data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
  106. data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
  107. data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
  108. data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
  109. data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
  110. data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
  111. data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
  112. data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +15 -0
  113. data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
  114. data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
  115. data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
  116. data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +37 -0
  117. data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
  118. data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
  119. data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
  120. data/lib/nokogiri/ffi/xml/attr.rb +41 -0
  121. data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
  122. data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
  123. data/lib/nokogiri/ffi/xml/comment.rb +18 -0
  124. data/lib/nokogiri/ffi/xml/document.rb +135 -0
  125. data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
  126. data/lib/nokogiri/ffi/xml/dtd.rb +69 -0
  127. data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
  128. data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
  129. data/lib/nokogiri/ffi/xml/entity_decl.rb +27 -0
  130. data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
  131. data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
  132. data/lib/nokogiri/ffi/xml/node.rb +444 -0
  133. data/lib/nokogiri/ffi/xml/node_set.rb +133 -0
  134. data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
  135. data/lib/nokogiri/ffi/xml/reader.rb +227 -0
  136. data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
  137. data/lib/nokogiri/ffi/xml/sax/parser.rb +142 -0
  138. data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
  139. data/lib/nokogiri/ffi/xml/sax/push_parser.rb +39 -0
  140. data/lib/nokogiri/ffi/xml/schema.rb +92 -0
  141. data/lib/nokogiri/ffi/xml/syntax_error.rb +91 -0
  142. data/lib/nokogiri/ffi/xml/text.rb +18 -0
  143. data/lib/nokogiri/ffi/xml/xpath.rb +19 -0
  144. data/lib/nokogiri/ffi/xml/xpath_context.rb +135 -0
  145. data/lib/nokogiri/ffi/xslt/stylesheet.rb +47 -0
  146. data/lib/nokogiri/html.rb +35 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +88 -0
  149. data/lib/nokogiri/html/document_fragment.rb +15 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  152. data/lib/nokogiri/html/sax/parser.rb +48 -0
  153. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  154. data/lib/nokogiri/syntax_error.rb +4 -0
  155. data/lib/nokogiri/version.rb +33 -0
  156. data/lib/nokogiri/version_warning.rb +11 -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 +405 -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 +131 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +69 -0
  165. data/lib/nokogiri/xml/dtd.rb +11 -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 +15 -0
  169. data/lib/nokogiri/xml/fragment_handler.rb +71 -0
  170. data/lib/nokogiri/xml/namespace.rb +13 -0
  171. data/lib/nokogiri/xml/node.rb +665 -0
  172. data/lib/nokogiri/xml/node/save_options.rb +42 -0
  173. data/lib/nokogiri/xml/node_set.rb +307 -0
  174. data/lib/nokogiri/xml/notation.rb +6 -0
  175. data/lib/nokogiri/xml/parse_options.rb +85 -0
  176. data/lib/nokogiri/xml/pp.rb +2 -0
  177. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  178. data/lib/nokogiri/xml/pp/node.rb +56 -0
  179. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  180. data/lib/nokogiri/xml/reader.rb +74 -0
  181. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  182. data/lib/nokogiri/xml/sax.rb +4 -0
  183. data/lib/nokogiri/xml/sax/document.rb +160 -0
  184. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  185. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  186. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  187. data/lib/nokogiri/xml/schema.rb +61 -0
  188. data/lib/nokogiri/xml/syntax_error.rb +38 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +8 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +48 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xsd/xmlparser/nokogiri.rb +71 -0
  195. data/tasks/test.rb +100 -0
  196. data/test/css/test_nthiness.rb +159 -0
  197. data/test/css/test_parser.rb +277 -0
  198. data/test/css/test_tokenizer.rb +183 -0
  199. data/test/css/test_xpath_visitor.rb +76 -0
  200. data/test/ffi/test_document.rb +35 -0
  201. data/test/files/2ch.html +108 -0
  202. data/test/files/address_book.rlx +12 -0
  203. data/test/files/address_book.xml +10 -0
  204. data/test/files/bar/bar.xsd +4 -0
  205. data/test/files/dont_hurt_em_why.xml +422 -0
  206. data/test/files/exslt.xml +8 -0
  207. data/test/files/exslt.xslt +35 -0
  208. data/test/files/foo/foo.xsd +4 -0
  209. data/test/files/po.xml +32 -0
  210. data/test/files/po.xsd +66 -0
  211. data/test/files/shift_jis.html +10 -0
  212. data/test/files/shift_jis.xml +5 -0
  213. data/test/files/snuggles.xml +3 -0
  214. data/test/files/staff.dtd +10 -0
  215. data/test/files/staff.xml +59 -0
  216. data/test/files/staff.xslt +32 -0
  217. data/test/files/tlm.html +850 -0
  218. data/test/files/valid_bar.xml +2 -0
  219. data/test/helper.rb +136 -0
  220. data/test/html/sax/test_parser.rb +64 -0
  221. data/test/html/sax/test_parser_context.rb +48 -0
  222. data/test/html/test_builder.rb +164 -0
  223. data/test/html/test_document.rb +390 -0
  224. data/test/html/test_document_encoding.rb +77 -0
  225. data/test/html/test_document_fragment.rb +132 -0
  226. data/test/html/test_element_description.rb +94 -0
  227. data/test/html/test_named_characters.rb +14 -0
  228. data/test/html/test_node.rb +228 -0
  229. data/test/html/test_node_encoding.rb +27 -0
  230. data/test/test_convert_xpath.rb +135 -0
  231. data/test/test_css_cache.rb +45 -0
  232. data/test/test_gc.rb +15 -0
  233. data/test/test_memory_leak.rb +77 -0
  234. data/test/test_nokogiri.rb +134 -0
  235. data/test/test_reader.rb +358 -0
  236. data/test/test_xslt_transforms.rb +131 -0
  237. data/test/xml/node/test_save_options.rb +20 -0
  238. data/test/xml/node/test_subclass.rb +44 -0
  239. data/test/xml/sax/test_parser.rb +307 -0
  240. data/test/xml/sax/test_parser_context.rb +56 -0
  241. data/test/xml/sax/test_push_parser.rb +131 -0
  242. data/test/xml/test_attr.rb +38 -0
  243. data/test/xml/test_attribute_decl.rb +82 -0
  244. data/test/xml/test_builder.rb +167 -0
  245. data/test/xml/test_cdata.rb +38 -0
  246. data/test/xml/test_comment.rb +29 -0
  247. data/test/xml/test_document.rb +607 -0
  248. data/test/xml/test_document_encoding.rb +26 -0
  249. data/test/xml/test_document_fragment.rb +138 -0
  250. data/test/xml/test_dtd.rb +82 -0
  251. data/test/xml/test_dtd_encoding.rb +33 -0
  252. data/test/xml/test_element_content.rb +56 -0
  253. data/test/xml/test_element_decl.rb +73 -0
  254. data/test/xml/test_entity_decl.rb +83 -0
  255. data/test/xml/test_entity_reference.rb +21 -0
  256. data/test/xml/test_namespace.rb +68 -0
  257. data/test/xml/test_node.rb +889 -0
  258. data/test/xml/test_node_attributes.rb +34 -0
  259. data/test/xml/test_node_encoding.rb +107 -0
  260. data/test/xml/test_node_set.rb +531 -0
  261. data/test/xml/test_parse_options.rb +52 -0
  262. data/test/xml/test_processing_instruction.rb +30 -0
  263. data/test/xml/test_reader_encoding.rb +126 -0
  264. data/test/xml/test_relax_ng.rb +60 -0
  265. data/test/xml/test_schema.rb +89 -0
  266. data/test/xml/test_syntax_error.rb +27 -0
  267. data/test/xml/test_text.rb +30 -0
  268. data/test/xml/test_unparented_node.rb +381 -0
  269. data/test/xml/test_xpath.rb +106 -0
  270. metadata +430 -0
@@ -0,0 +1,388 @@
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
+ default:
10
+ if(node->parent == NULL) {
11
+ xmlAddChild((xmlNodePtr)doc, node);
12
+ }
13
+ }
14
+ return ST_CONTINUE;
15
+ }
16
+
17
+ static void dealloc(xmlDocPtr doc)
18
+ {
19
+ NOKOGIRI_DEBUG_START(doc);
20
+
21
+ st_table *node_hash = DOC_UNLINKED_NODE_HASH(doc);
22
+
23
+ xmlDeregisterNodeFunc func = xmlDeregisterNodeDefault(NULL);
24
+
25
+ st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
26
+ st_free_table(node_hash);
27
+
28
+ free(doc->_private);
29
+ doc->_private = NULL;
30
+ xmlFreeDoc(doc);
31
+
32
+ xmlDeregisterNodeDefault(func);
33
+
34
+ NOKOGIRI_DEBUG_END(doc);
35
+ }
36
+
37
+ static void recursively_remove_namespaces_from_node(xmlNodePtr node)
38
+ {
39
+ xmlNodePtr child ;
40
+
41
+ xmlSetNs(node, NULL);
42
+
43
+ for (child = node->children ; child ; child = child->next)
44
+ recursively_remove_namespaces_from_node(child);
45
+ }
46
+
47
+ /*
48
+ * call-seq:
49
+ * url
50
+ *
51
+ * Get the url name for this document.
52
+ */
53
+ static VALUE url(VALUE self)
54
+ {
55
+ xmlDocPtr doc;
56
+ Data_Get_Struct(self, xmlDoc, doc);
57
+
58
+ if(doc->URL) return NOKOGIRI_STR_NEW2(doc->URL);
59
+
60
+ return Qnil;
61
+ }
62
+
63
+ /*
64
+ * call-seq:
65
+ * root=
66
+ *
67
+ * Set the root element on this document
68
+ */
69
+ static VALUE set_root(VALUE self, VALUE root)
70
+ {
71
+ xmlDocPtr doc;
72
+ xmlNodePtr new_root;
73
+
74
+ Data_Get_Struct(self, xmlDoc, doc);
75
+ Data_Get_Struct(root, xmlNode, new_root);
76
+
77
+ xmlNodePtr old_root = NULL;
78
+
79
+ /* If the new root's document is not the same as the current document,
80
+ * then we need to dup the node in to this document. */
81
+ if(new_root->doc != doc) {
82
+ old_root = xmlDocGetRootElement(doc);
83
+ if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
84
+ rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
85
+ }
86
+ }
87
+
88
+ xmlDocSetRootElement(doc, new_root);
89
+ if(old_root) NOKOGIRI_ROOT_NODE(old_root);
90
+ return root;
91
+ }
92
+
93
+ /*
94
+ * call-seq:
95
+ * root
96
+ *
97
+ * Get the root node for this document.
98
+ */
99
+ static VALUE root(VALUE self)
100
+ {
101
+ xmlDocPtr doc;
102
+ Data_Get_Struct(self, xmlDoc, doc);
103
+
104
+ xmlNodePtr root = xmlDocGetRootElement(doc);
105
+
106
+ if(!root) return Qnil;
107
+ return Nokogiri_wrap_xml_node(Qnil, root) ;
108
+ }
109
+
110
+ /*
111
+ * call-seq:
112
+ * encoding= encoding
113
+ *
114
+ * Set the encoding string for this Document
115
+ */
116
+ static VALUE set_encoding(VALUE self, VALUE encoding)
117
+ {
118
+ xmlDocPtr doc;
119
+ Data_Get_Struct(self, xmlDoc, doc);
120
+
121
+ doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
122
+
123
+ return encoding;
124
+ }
125
+
126
+ /*
127
+ * call-seq:
128
+ * encoding
129
+ *
130
+ * Get the encoding for this Document
131
+ */
132
+ static VALUE encoding(VALUE self)
133
+ {
134
+ xmlDocPtr doc;
135
+ Data_Get_Struct(self, xmlDoc, doc);
136
+
137
+ if(!doc->encoding) return Qnil;
138
+ return NOKOGIRI_STR_NEW2(doc->encoding);
139
+ }
140
+
141
+ /*
142
+ * call-seq:
143
+ * version
144
+ *
145
+ * Get the XML version for this Document
146
+ */
147
+ static VALUE version(VALUE self)
148
+ {
149
+ xmlDocPtr doc;
150
+ Data_Get_Struct(self, xmlDoc, doc);
151
+
152
+ if(!doc->version) return Qnil;
153
+ return NOKOGIRI_STR_NEW2(doc->version);
154
+ }
155
+
156
+ /*
157
+ * call-seq:
158
+ * read_io(io, url, encoding, options)
159
+ *
160
+ * Create a new document from an IO object
161
+ */
162
+ static VALUE read_io( VALUE klass,
163
+ VALUE io,
164
+ VALUE url,
165
+ VALUE encoding,
166
+ VALUE options )
167
+ {
168
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
169
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
170
+ VALUE error_list = rb_ary_new();
171
+
172
+ xmlResetLastError();
173
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
174
+
175
+ xmlDocPtr doc = xmlReadIO(
176
+ (xmlInputReadCallback)io_read_callback,
177
+ (xmlInputCloseCallback)io_close_callback,
178
+ (void *)io,
179
+ c_url,
180
+ c_enc,
181
+ (int)NUM2INT(options)
182
+ );
183
+ xmlSetStructuredErrorFunc(NULL, NULL);
184
+
185
+ if(doc == NULL) {
186
+ xmlFreeDoc(doc);
187
+
188
+ xmlErrorPtr error = xmlGetLastError();
189
+ if(error)
190
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
191
+ else
192
+ rb_raise(rb_eRuntimeError, "Could not parse document");
193
+
194
+ return Qnil;
195
+ }
196
+
197
+ VALUE document = Nokogiri_wrap_xml_document(klass, doc);
198
+ rb_iv_set(document, "@errors", error_list);
199
+ return document;
200
+ }
201
+
202
+ /*
203
+ * call-seq:
204
+ * read_memory(string, url, encoding, options)
205
+ *
206
+ * Create a new document from a String
207
+ */
208
+ static VALUE read_memory( VALUE klass,
209
+ VALUE string,
210
+ VALUE url,
211
+ VALUE encoding,
212
+ VALUE options )
213
+ {
214
+ const char * c_buffer = StringValuePtr(string);
215
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
216
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
217
+ int len = RSTRING_LEN(string);
218
+ VALUE error_list = rb_ary_new();
219
+
220
+ xmlResetLastError();
221
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
222
+ xmlDocPtr doc = xmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
223
+ xmlSetStructuredErrorFunc(NULL, NULL);
224
+
225
+ if(doc == NULL) {
226
+ xmlFreeDoc(doc);
227
+
228
+ xmlErrorPtr error = xmlGetLastError();
229
+ if(error)
230
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
231
+ else
232
+ rb_raise(rb_eRuntimeError, "Could not parse document");
233
+
234
+ return Qnil;
235
+ }
236
+
237
+ VALUE document = Nokogiri_wrap_xml_document(klass, doc);
238
+ rb_iv_set(document, "@errors", error_list);
239
+ return document;
240
+ }
241
+
242
+ /*
243
+ * call-seq:
244
+ * dup
245
+ *
246
+ * Copy this Document. An optional depth may be passed in, but it defaults
247
+ * to a deep copy. 0 is a shallow copy, 1 is a deep copy.
248
+ */
249
+ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
250
+ {
251
+ VALUE level;
252
+
253
+ if(rb_scan_args(argc, argv, "01", &level) == 0)
254
+ level = INT2NUM((long)1);
255
+
256
+ xmlDocPtr doc, dup;
257
+ Data_Get_Struct(self, xmlDoc, doc);
258
+
259
+ dup = xmlCopyDoc(doc, (int)NUM2INT(level));
260
+ if(dup == NULL) return Qnil;
261
+
262
+ dup->type = doc->type;
263
+ return Nokogiri_wrap_xml_document(RBASIC(self)->klass, dup);
264
+ }
265
+
266
+ /*
267
+ * call-seq:
268
+ * new(version = default)
269
+ *
270
+ * Create a new document with +version+ (defaults to "1.0")
271
+ */
272
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
273
+ {
274
+ VALUE version, rest, rb_doc ;
275
+
276
+ rb_scan_args(argc, argv, "0*", &rest);
277
+ version = rb_ary_entry(rest, (long)0);
278
+ if (NIL_P(Qnil)) {
279
+ version = rb_str_new2("1.0");
280
+ }
281
+
282
+ xmlDocPtr doc = xmlNewDoc((xmlChar *)StringValuePtr(version));
283
+ rb_doc = Nokogiri_wrap_xml_document(klass, doc);
284
+ rb_obj_call_init(rb_doc, argc, argv);
285
+ return rb_doc ;
286
+ }
287
+
288
+ /*
289
+ * call-seq:
290
+ * remove_namespaces!
291
+ *
292
+ * Remove all namespaces from all nodes in the document.
293
+ *
294
+ * This could be useful for developers who either don't understand namespaces
295
+ * or don't care about them.
296
+ *
297
+ * The following example shows a use case, and you can decide for yourself
298
+ * whether this is a good thing or not:
299
+ *
300
+ * doc = Nokogiri::XML <<-EOXML
301
+ * <root>
302
+ * <car xmlns:part="http://general-motors.com/">
303
+ * <part:tire>Michelin Model XGV</part:tire>
304
+ * </car>
305
+ * <bicycle xmlns:part="http://schwinn.com/">
306
+ * <part:tire>I'm a bicycle tire!</part:tire>
307
+ * </bicycle>
308
+ * </root>
309
+ * EOXML
310
+ *
311
+ * doc.xpath("//tire").to_s # => ""
312
+ * doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => "<part:tire>Michelin Model XGV</part:tire>"
313
+ * doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => "<part:tire>I'm a bicycle tire!</part:tire>"
314
+ *
315
+ * doc.remove_namespaces!
316
+ *
317
+ * doc.xpath("//tire").to_s # => "<tire>Michelin Model XGV</tire><tire>I'm a bicycle tire!</tire>"
318
+ * doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => ""
319
+ * doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => ""
320
+ *
321
+ * For more information on why this probably is *not* a good thing in general,
322
+ * please direct your browser to
323
+ * http://tenderlovemaking.com/2009/04/23/namespaces-in-xml/
324
+ */
325
+ VALUE remove_namespaces_bang(VALUE self)
326
+ {
327
+ xmlDocPtr doc ;
328
+ Data_Get_Struct(self, xmlDoc, doc);
329
+
330
+ recursively_remove_namespaces_from_node(doc);
331
+ return self;
332
+ }
333
+
334
+
335
+ VALUE cNokogiriXmlDocument ;
336
+ void init_xml_document()
337
+ {
338
+ VALUE nokogiri = rb_define_module("Nokogiri");
339
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
340
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
341
+
342
+ /*
343
+ * Nokogiri::XML::Document wraps an xml document.
344
+ */
345
+ VALUE klass = rb_define_class_under(xml, "Document", node);
346
+
347
+ cNokogiriXmlDocument = klass;
348
+
349
+ rb_define_singleton_method(klass, "read_memory", read_memory, 4);
350
+ rb_define_singleton_method(klass, "read_io", read_io, 4);
351
+ rb_define_singleton_method(klass, "new", new, -1);
352
+
353
+ rb_define_method(klass, "root", root, 0);
354
+ rb_define_method(klass, "root=", set_root, 1);
355
+ rb_define_method(klass, "encoding", encoding, 0);
356
+ rb_define_method(klass, "encoding=", set_encoding, 1);
357
+ rb_define_method(klass, "version", version, 0);
358
+ rb_define_method(klass, "dup", duplicate_node, -1);
359
+ rb_define_method(klass, "url", url, 0);
360
+ rb_define_method(klass, "remove_namespaces!", remove_namespaces_bang, 0);
361
+ }
362
+
363
+
364
+ /* this takes klass as a param because it's used for HtmlDocument, too. */
365
+ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
366
+ {
367
+ nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
368
+
369
+ VALUE rb_doc = Data_Wrap_Struct(
370
+ klass ? klass : cNokogiriXmlDocument,
371
+ 0,
372
+ dealloc,
373
+ doc
374
+ );
375
+
376
+ VALUE cache = rb_ary_new();
377
+ rb_iv_set(rb_doc, "@decorators", Qnil);
378
+ rb_iv_set(rb_doc, "@node_cache", cache);
379
+
380
+ tuple->doc = (void *)rb_doc;
381
+ tuple->unlinkedNodes = st_init_numtable_with_size(128);
382
+ tuple->node_cache = cache;
383
+ doc->_private = tuple ;
384
+
385
+ rb_obj_call_init(rb_doc, 0, NULL);
386
+
387
+ return rb_doc ;
388
+ }
@@ -0,0 +1,24 @@
1
+ #ifndef NOKOGIRI_XML_DOCUMENT
2
+ #define NOKOGIRI_XML_DOCUMENT
3
+
4
+ #include <nokogiri.h>
5
+ #include "st.h"
6
+
7
+ struct _nokogiriTuple {
8
+ xmlDocPtr doc;
9
+ st_table *unlinkedNodes;
10
+ VALUE node_cache;
11
+ };
12
+ typedef struct _nokogiriTuple nokogiriTuple;
13
+ typedef nokogiriTuple * nokogiriTuplePtr;
14
+
15
+ void init_xml_document();
16
+ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc);
17
+
18
+ #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
19
+ #define DOC_RUBY_OBJECT(x) ((VALUE)((nokogiriTuplePtr)(x->_private))->doc)
20
+ #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
21
+ #define DOC_NODE_CACHE(x) ((VALUE)((nokogiriTuplePtr)(x->_private))->node_cache)
22
+
23
+ extern VALUE cNokogiriXmlDocument ;
24
+ #endif
@@ -0,0 +1,46 @@
1
+ #include <xml_document_fragment.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * new(document)
6
+ *
7
+ * Create a new DocumentFragment element on the +document+
8
+ */
9
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
10
+ {
11
+ xmlDocPtr xml_doc;
12
+ VALUE document;
13
+ VALUE rest;
14
+
15
+ rb_scan_args(argc, argv, "1*", &document, &rest);
16
+
17
+ Data_Get_Struct(document, xmlDoc, xml_doc);
18
+
19
+ xmlNodePtr node = xmlNewDocFragment(xml_doc->doc);
20
+
21
+ NOKOGIRI_ROOT_NODE(node);
22
+
23
+ VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
24
+ rb_obj_call_init(rb_node, argc, argv);
25
+
26
+ if(rb_block_given_p()) rb_yield(rb_node);
27
+
28
+ return rb_node;
29
+ }
30
+
31
+ VALUE cNokogiriXmlDocumentFragment;
32
+ void init_xml_document_fragment()
33
+ {
34
+ VALUE nokogiri = rb_define_module("Nokogiri");
35
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
36
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
37
+
38
+ /*
39
+ * DocumentFragment represents a DocumentFragment node in an xml document.
40
+ */
41
+ VALUE klass = rb_define_class_under(xml, "DocumentFragment", node);
42
+
43
+ cNokogiriXmlDocumentFragment = klass;
44
+
45
+ rb_define_singleton_method(klass, "new", new, -1);
46
+ }