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 +1,10 @@
1
+ #ifndef NOKOGIRI_XML_READER
2
+ #define NOKOGIRI_XML_READER
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_reader();
7
+
8
+ extern VALUE cNokogiriXmlReader;
9
+
10
+ #endif
@@ -0,0 +1,159 @@
1
+ #include <xml_relax_ng.h>
2
+
3
+ static void dealloc(xmlRelaxNGPtr schema)
4
+ {
5
+ NOKOGIRI_DEBUG_START(schema);
6
+ xmlRelaxNGFree(schema);
7
+ NOKOGIRI_DEBUG_END(schema);
8
+ }
9
+
10
+ /*
11
+ * call-seq:
12
+ * validate_document(document)
13
+ *
14
+ * Validate a Nokogiri::XML::Document against this RelaxNG schema.
15
+ */
16
+ static VALUE validate_document(VALUE self, VALUE document)
17
+ {
18
+ xmlDocPtr doc;
19
+ xmlRelaxNGPtr schema;
20
+
21
+ Data_Get_Struct(self, xmlRelaxNG, schema);
22
+ Data_Get_Struct(document, xmlDoc, doc);
23
+
24
+ VALUE errors = rb_ary_new();
25
+
26
+ xmlRelaxNGValidCtxtPtr valid_ctxt = xmlRelaxNGNewValidCtxt(schema);
27
+
28
+ if(NULL == valid_ctxt) {
29
+ // we have a problem
30
+ rb_raise(rb_eRuntimeError, "Could not create a validation context");
31
+ }
32
+
33
+ #ifdef HAVE_XMLRELAXNGSETVALIDSTRUCTUREDERRORS
34
+ xmlRelaxNGSetValidStructuredErrors(
35
+ valid_ctxt,
36
+ Nokogiri_error_array_pusher,
37
+ (void *)errors
38
+ );
39
+ #endif
40
+
41
+ xmlRelaxNGValidateDoc(valid_ctxt, doc);
42
+
43
+ xmlRelaxNGFreeValidCtxt(valid_ctxt);
44
+
45
+ return errors;
46
+ }
47
+
48
+ /*
49
+ * call-seq:
50
+ * read_memory(string)
51
+ *
52
+ * Create a new RelaxNG from the contents of +string+
53
+ */
54
+ static VALUE read_memory(VALUE klass, VALUE content)
55
+ {
56
+ xmlRelaxNGParserCtxtPtr ctx = xmlRelaxNGNewMemParserCtxt(
57
+ (const char *)StringValuePtr(content),
58
+ RSTRING_LEN(content)
59
+ );
60
+
61
+ VALUE errors = rb_ary_new();
62
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
63
+
64
+ #ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
65
+ xmlRelaxNGSetParserStructuredErrors(
66
+ ctx,
67
+ Nokogiri_error_array_pusher,
68
+ (void *)errors
69
+ );
70
+ #endif
71
+
72
+ xmlRelaxNGPtr schema = xmlRelaxNGParse(ctx);
73
+
74
+ xmlSetStructuredErrorFunc(NULL, NULL);
75
+ xmlRelaxNGFreeParserCtxt(ctx);
76
+
77
+ if(NULL == schema) {
78
+ xmlErrorPtr error = xmlGetLastError();
79
+ if(error)
80
+ Nokogiri_error_raise(NULL, error);
81
+ else
82
+ rb_raise(rb_eRuntimeError, "Could not parse document");
83
+
84
+ return Qnil;
85
+ }
86
+
87
+ VALUE rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
88
+ rb_iv_set(rb_schema, "@errors", errors);
89
+
90
+ return rb_schema;
91
+ }
92
+
93
+ /*
94
+ * call-seq:
95
+ * from_document(doc)
96
+ *
97
+ * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
98
+ */
99
+ static VALUE from_document(VALUE klass, VALUE document)
100
+ {
101
+ xmlDocPtr doc;
102
+ Data_Get_Struct(document, xmlDoc, doc);
103
+
104
+ // In case someone passes us a node. ugh.
105
+ doc = doc->doc;
106
+
107
+ xmlRelaxNGParserCtxtPtr ctx = xmlRelaxNGNewDocParserCtxt(doc);
108
+
109
+ VALUE errors = rb_ary_new();
110
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
111
+
112
+ #ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
113
+ xmlRelaxNGSetParserStructuredErrors(
114
+ ctx,
115
+ Nokogiri_error_array_pusher,
116
+ (void *)errors
117
+ );
118
+ #endif
119
+
120
+ xmlRelaxNGPtr schema = xmlRelaxNGParse(ctx);
121
+
122
+ xmlSetStructuredErrorFunc(NULL, NULL);
123
+ if (! is_2_6_16()) {
124
+ xmlRelaxNGFreeParserCtxt(ctx);
125
+ }
126
+
127
+ if(NULL == schema) {
128
+ xmlErrorPtr error = xmlGetLastError();
129
+ if(error)
130
+ Nokogiri_error_raise(NULL, error);
131
+ else
132
+ rb_raise(rb_eRuntimeError, "Could not parse document");
133
+
134
+ return Qnil;
135
+ }
136
+
137
+ if (is_2_6_16()) {
138
+ xmlRelaxNGFreeParserCtxt(ctx);
139
+ }
140
+
141
+ VALUE rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
142
+ rb_iv_set(rb_schema, "@errors", errors);
143
+
144
+ return rb_schema;
145
+ }
146
+
147
+ VALUE cNokogiriXmlRelaxNG;
148
+ void init_xml_relax_ng()
149
+ {
150
+ VALUE nokogiri = rb_define_module("Nokogiri");
151
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
152
+ VALUE klass = rb_define_class_under(xml, "RelaxNG", cNokogiriXmlSchema);
153
+
154
+ cNokogiriXmlRelaxNG = klass;
155
+
156
+ rb_define_singleton_method(klass, "read_memory", read_memory, 1);
157
+ rb_define_singleton_method(klass, "from_document", from_document, 1);
158
+ rb_define_private_method(klass, "validate_document", validate_document, 1);
159
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_RELAX_NG
2
+ #define NOKOGIRI_XML_RELAX_NG
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_relax_ng();
7
+
8
+ extern VALUE cNokogiriXmlRelaxNG;
9
+ #endif
@@ -0,0 +1,286 @@
1
+ #include <xml_sax_parser.h>
2
+
3
+ static ID id_start_document, id_end_document, id_start_element, id_end_element;
4
+ static ID id_start_element_namespace, id_end_element_namespace;
5
+ static ID id_comment, id_characters, id_xmldecl, id_error, id_warning;
6
+ static ID id_cdata_block, id_cAttribute;
7
+
8
+ #define STRING_OR_NULL(str) \
9
+ (RTEST(str) ? StringValuePtr(str) : NULL)
10
+
11
+ #define RBSTR_OR_QNIL(_str) \
12
+ (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
13
+
14
+ static void start_document(void * ctx)
15
+ {
16
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
17
+ VALUE doc = rb_iv_get(self, "@document");
18
+
19
+ xmlParserCtxtPtr ctxt = NOKOGIRI_SAX_CTXT(ctx);
20
+
21
+ if(NULL != ctxt && ctxt->html != 1) {
22
+ if(ctxt->standalone != -1) { // -1 means there was no declaration
23
+ VALUE encoding = ctxt->encoding ?
24
+ NOKOGIRI_STR_NEW2(ctxt->encoding) :
25
+ Qnil;
26
+
27
+ VALUE version = ctxt->version ?
28
+ NOKOGIRI_STR_NEW2(ctxt->version) :
29
+ Qnil;
30
+
31
+ VALUE standalone = Qnil;
32
+
33
+ switch(ctxt->standalone)
34
+ {
35
+ case 0:
36
+ standalone = NOKOGIRI_STR_NEW2("no");
37
+ break;
38
+ case 1:
39
+ standalone = NOKOGIRI_STR_NEW2("yes");
40
+ break;
41
+ }
42
+
43
+ rb_funcall(doc, id_xmldecl, 3, version, encoding, standalone);
44
+ }
45
+ }
46
+
47
+ rb_funcall(doc, id_start_document, 0);
48
+ }
49
+
50
+ static void end_document(void * ctx)
51
+ {
52
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
53
+ VALUE doc = rb_iv_get(self, "@document");
54
+ rb_funcall(doc, id_end_document, 0);
55
+ }
56
+
57
+ static void start_element(void * ctx, const xmlChar *name, const xmlChar **atts)
58
+ {
59
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
60
+ VALUE doc = rb_iv_get(self, "@document");
61
+ VALUE attributes = rb_ary_new();
62
+ const xmlChar * attr;
63
+ int i = 0;
64
+ if(atts) {
65
+ while((attr = atts[i]) != NULL) {
66
+ rb_ary_push(attributes, NOKOGIRI_STR_NEW2(attr));
67
+ i++;
68
+ }
69
+ }
70
+
71
+ rb_funcall( doc,
72
+ id_start_element,
73
+ 2,
74
+ NOKOGIRI_STR_NEW2(name),
75
+ attributes
76
+ );
77
+ }
78
+
79
+ static void end_element(void * ctx, const xmlChar *name)
80
+ {
81
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
82
+ VALUE doc = rb_iv_get(self, "@document");
83
+ rb_funcall(doc, id_end_element, 1, NOKOGIRI_STR_NEW2(name));
84
+ }
85
+
86
+ static VALUE attributes_as_list(
87
+ VALUE self,
88
+ int nb_attributes,
89
+ const xmlChar ** attributes)
90
+ {
91
+ VALUE list = rb_ary_new2((long)nb_attributes);
92
+
93
+ VALUE attr_klass = rb_const_get(cNokogiriXmlSaxParser, id_cAttribute);
94
+ if (attributes) {
95
+ /* Each attribute is an array of [localname, prefix, URI, value, end] */
96
+ int i;
97
+ for (i = 0; i < nb_attributes * 5; i += 5) {
98
+ VALUE argv[4];
99
+ argv[0] = RBSTR_OR_QNIL(attributes[i + 0]); /* localname */
100
+ argv[1] = RBSTR_OR_QNIL(attributes[i + 1]); /* prefix */
101
+ argv[2] = RBSTR_OR_QNIL(attributes[i + 2]); /* URI */
102
+
103
+ /* value */
104
+ argv[3] = NOKOGIRI_STR_NEW((const char*)attributes[i+3],
105
+ (attributes[i+4] - attributes[i+3]));
106
+
107
+ VALUE attribute = rb_class_new_instance(4, argv, attr_klass);
108
+ rb_ary_push(list, attribute);
109
+ }
110
+ }
111
+
112
+ return list;
113
+ }
114
+
115
+ static void
116
+ start_element_ns (
117
+ void * ctx,
118
+ const xmlChar * localname,
119
+ const xmlChar * prefix,
120
+ const xmlChar * uri,
121
+ int nb_namespaces,
122
+ const xmlChar ** namespaces,
123
+ int nb_attributes,
124
+ int nb_defaulted,
125
+ const xmlChar ** attributes)
126
+ {
127
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
128
+ VALUE doc = rb_iv_get(self, "@document");
129
+
130
+ VALUE attribute_list = attributes_as_list(self, nb_attributes, attributes);
131
+
132
+ VALUE ns_list = rb_ary_new2((long)nb_namespaces);
133
+
134
+ if (namespaces) {
135
+ int i;
136
+ for (i = 0; i < nb_namespaces * 2; i += 2)
137
+ {
138
+ rb_ary_push(ns_list,
139
+ rb_ary_new3((long)2,
140
+ RBSTR_OR_QNIL(namespaces[i + 0]),
141
+ RBSTR_OR_QNIL(namespaces[i + 1])
142
+ )
143
+ );
144
+ }
145
+ }
146
+
147
+ rb_funcall( doc,
148
+ id_start_element_namespace,
149
+ 5,
150
+ NOKOGIRI_STR_NEW2(localname),
151
+ attribute_list,
152
+ RBSTR_OR_QNIL(prefix),
153
+ RBSTR_OR_QNIL(uri),
154
+ ns_list
155
+ );
156
+ }
157
+
158
+ /**
159
+ * end_element_ns was borrowed heavily from libxml-ruby.
160
+ */
161
+ static void
162
+ end_element_ns (
163
+ void * ctx,
164
+ const xmlChar * localname,
165
+ const xmlChar * prefix,
166
+ const xmlChar * uri)
167
+ {
168
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
169
+ VALUE doc = rb_iv_get(self, "@document");
170
+
171
+ rb_funcall(doc, id_end_element_namespace, 3,
172
+ NOKOGIRI_STR_NEW2(localname),
173
+ RBSTR_OR_QNIL(prefix),
174
+ RBSTR_OR_QNIL(uri)
175
+ );
176
+ }
177
+
178
+ static void characters_func(void * ctx, const xmlChar * ch, int len)
179
+ {
180
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
181
+ VALUE doc = rb_iv_get(self, "@document");
182
+ VALUE str = NOKOGIRI_STR_NEW(ch, len);
183
+ rb_funcall(doc, id_characters, 1, str);
184
+ }
185
+
186
+ static void comment_func(void * ctx, const xmlChar * value)
187
+ {
188
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
189
+ VALUE doc = rb_iv_get(self, "@document");
190
+ VALUE str = NOKOGIRI_STR_NEW2(value);
191
+ rb_funcall(doc, id_comment, 1, str);
192
+ }
193
+
194
+ static void warning_func(void * ctx, const char *msg, ...)
195
+ {
196
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
197
+ VALUE doc = rb_iv_get(self, "@document");
198
+ char * message;
199
+
200
+ va_list args;
201
+ va_start(args, msg);
202
+ vasprintf(&message, msg, args);
203
+ va_end(args);
204
+
205
+ rb_funcall(doc, id_warning, 1, NOKOGIRI_STR_NEW2(message));
206
+ free(message);
207
+ }
208
+
209
+ static void error_func(void * ctx, const char *msg, ...)
210
+ {
211
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
212
+ VALUE doc = rb_iv_get(self, "@document");
213
+ char * message;
214
+
215
+ va_list args;
216
+ va_start(args, msg);
217
+ vasprintf(&message, msg, args);
218
+ va_end(args);
219
+
220
+ rb_funcall(doc, id_error, 1, NOKOGIRI_STR_NEW2(message));
221
+ free(message);
222
+ }
223
+
224
+ static void cdata_block(void * ctx, const xmlChar * value, int len)
225
+ {
226
+ VALUE self = NOKOGIRI_SAX_SELF(ctx);
227
+ VALUE doc = rb_iv_get(self, "@document");
228
+ VALUE string = NOKOGIRI_STR_NEW(value, len);
229
+ rb_funcall(doc, id_cdata_block, 1, string);
230
+ }
231
+
232
+ static void deallocate(xmlSAXHandlerPtr handler)
233
+ {
234
+ NOKOGIRI_DEBUG_START(handler);
235
+ free(handler);
236
+ NOKOGIRI_DEBUG_END(handler);
237
+ }
238
+
239
+ static VALUE allocate(VALUE klass)
240
+ {
241
+ xmlSAXHandlerPtr handler = calloc(1, sizeof(xmlSAXHandler));
242
+
243
+ xmlSetStructuredErrorFunc(NULL, NULL);
244
+
245
+ handler->startDocument = start_document;
246
+ handler->endDocument = end_document;
247
+ handler->startElement = start_element;
248
+ handler->endElement = end_element;
249
+ handler->startElementNs = start_element_ns;
250
+ handler->endElementNs = end_element_ns;
251
+ handler->characters = characters_func;
252
+ handler->comment = comment_func;
253
+ handler->warning = warning_func;
254
+ handler->error = error_func;
255
+ handler->cdataBlock = cdata_block;
256
+ handler->initialized = XML_SAX2_MAGIC;
257
+
258
+ return Data_Wrap_Struct(klass, NULL, deallocate, handler);
259
+ }
260
+
261
+ VALUE cNokogiriXmlSaxParser ;
262
+ void init_xml_sax_parser()
263
+ {
264
+ VALUE nokogiri = rb_define_module("Nokogiri");
265
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
266
+ VALUE sax = rb_define_module_under(xml, "SAX");
267
+ VALUE klass = rb_define_class_under(sax, "Parser", rb_cObject);
268
+
269
+ cNokogiriXmlSaxParser = klass;
270
+
271
+ rb_define_alloc_func(klass, allocate);
272
+
273
+ id_start_document = rb_intern("start_document");
274
+ id_end_document = rb_intern("end_document");
275
+ id_start_element = rb_intern("start_element");
276
+ id_end_element = rb_intern("end_element");
277
+ id_comment = rb_intern("comment");
278
+ id_characters = rb_intern("characters");
279
+ id_xmldecl = rb_intern("xmldecl");
280
+ id_error = rb_intern("error");
281
+ id_warning = rb_intern("warning");
282
+ id_cdata_block = rb_intern("cdata_block");
283
+ id_cAttribute = rb_intern("Attribute");
284
+ id_start_element_namespace = rb_intern("start_element_namespace");
285
+ id_end_element_namespace = rb_intern("end_element_namespace");
286
+ }