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_HTML_DOCUMENT
2
+ #define NOKOGIRI_HTML_DOCUMENT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_document();
7
+
8
+ extern VALUE cNokogiriHtmlDocument ;
9
+
10
+ #endif
@@ -0,0 +1,272 @@
1
+ #include <html_element_description.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * required_attributes
6
+ *
7
+ * A list of required attributes for this element
8
+ */
9
+ static VALUE required_attributes(VALUE self)
10
+ {
11
+ htmlElemDesc * description;
12
+ Data_Get_Struct(self, htmlElemDesc, description);
13
+
14
+ VALUE list = rb_ary_new();
15
+
16
+ if(NULL == description->attrs_req) return list;
17
+
18
+ int i = 0;
19
+ while(description->attrs_req[i]) {
20
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
21
+ i++;
22
+ }
23
+
24
+ return list;
25
+ }
26
+
27
+ /*
28
+ * call-seq:
29
+ * deprecated_attributes
30
+ *
31
+ * A list of deprecated attributes for this element
32
+ */
33
+ static VALUE deprecated_attributes(VALUE self)
34
+ {
35
+ htmlElemDesc * description;
36
+ Data_Get_Struct(self, htmlElemDesc, description);
37
+
38
+ VALUE list = rb_ary_new();
39
+
40
+ if(NULL == description->attrs_depr) return list;
41
+
42
+ int i = 0;
43
+ while(description->attrs_depr[i]) {
44
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
45
+ i++;
46
+ }
47
+
48
+ return list;
49
+ }
50
+
51
+ /*
52
+ * call-seq:
53
+ * optional_attributes
54
+ *
55
+ * A list of optional attributes for this element
56
+ */
57
+ static VALUE optional_attributes(VALUE self)
58
+ {
59
+ htmlElemDesc * description;
60
+ Data_Get_Struct(self, htmlElemDesc, description);
61
+
62
+ VALUE list = rb_ary_new();
63
+
64
+ if(NULL == description->attrs_opt) return list;
65
+
66
+ int i = 0;
67
+ while(description->attrs_opt[i]) {
68
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
69
+ i++;
70
+ }
71
+
72
+ return list;
73
+ }
74
+
75
+ /*
76
+ * call-seq:
77
+ * default_sub_element
78
+ *
79
+ * The default sub element for this element
80
+ */
81
+ static VALUE default_sub_element(VALUE self)
82
+ {
83
+ htmlElemDesc * description;
84
+ Data_Get_Struct(self, htmlElemDesc, description);
85
+
86
+ return NOKOGIRI_STR_NEW2(description->defaultsubelt);
87
+ }
88
+
89
+ /*
90
+ * call-seq:
91
+ * sub_elements
92
+ *
93
+ * A list of allowed sub elements for this element.
94
+ */
95
+ static VALUE sub_elements(VALUE self)
96
+ {
97
+ htmlElemDesc * description;
98
+ Data_Get_Struct(self, htmlElemDesc, description);
99
+
100
+ VALUE list = rb_ary_new();
101
+
102
+ if(NULL == description->subelts) return list;
103
+
104
+ int i = 0;
105
+ while(description->subelts[i]) {
106
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
107
+ i++;
108
+ }
109
+
110
+ return list;
111
+ }
112
+
113
+ /*
114
+ * call-seq:
115
+ * description
116
+ *
117
+ * The description for this element
118
+ */
119
+ static VALUE description(VALUE self)
120
+ {
121
+ htmlElemDesc * description;
122
+ Data_Get_Struct(self, htmlElemDesc, description);
123
+
124
+ return NOKOGIRI_STR_NEW2(description->desc);
125
+ }
126
+
127
+ /*
128
+ * call-seq:
129
+ * inline?
130
+ *
131
+ * Is this element an inline element?
132
+ */
133
+ static VALUE inline_eh(VALUE self)
134
+ {
135
+ htmlElemDesc * description;
136
+ Data_Get_Struct(self, htmlElemDesc, description);
137
+
138
+ if(description->isinline) return Qtrue;
139
+ return Qfalse;
140
+ }
141
+
142
+ /*
143
+ * call-seq:
144
+ * deprecated?
145
+ *
146
+ * Is this element deprecated?
147
+ */
148
+ static VALUE deprecated_eh(VALUE self)
149
+ {
150
+ htmlElemDesc * description;
151
+ Data_Get_Struct(self, htmlElemDesc, description);
152
+
153
+ if(description->depr) return Qtrue;
154
+ return Qfalse;
155
+ }
156
+
157
+ /*
158
+ * call-seq:
159
+ * empty?
160
+ *
161
+ * Is this an empty element?
162
+ */
163
+ static VALUE empty_eh(VALUE self)
164
+ {
165
+ htmlElemDesc * description;
166
+ Data_Get_Struct(self, htmlElemDesc, description);
167
+
168
+ if(description->empty) return Qtrue;
169
+ return Qfalse;
170
+ }
171
+
172
+ /*
173
+ * call-seq:
174
+ * save_end_tag?
175
+ *
176
+ * Should the end tag be saved?
177
+ */
178
+ static VALUE save_end_tag_eh(VALUE self)
179
+ {
180
+ htmlElemDesc * description;
181
+ Data_Get_Struct(self, htmlElemDesc, description);
182
+
183
+ if(description->saveEndTag) return Qtrue;
184
+ return Qfalse;
185
+ }
186
+
187
+ /*
188
+ * call-seq:
189
+ * implied_end_tag?
190
+ *
191
+ * Can the end tag be implied for this tag?
192
+ */
193
+ static VALUE implied_end_tag_eh(VALUE self)
194
+ {
195
+ htmlElemDesc * description;
196
+ Data_Get_Struct(self, htmlElemDesc, description);
197
+
198
+ if(description->endTag) return Qtrue;
199
+ return Qfalse;
200
+ }
201
+
202
+ /*
203
+ * call-seq:
204
+ * implied_start_tag?
205
+ *
206
+ * Can the start tag be implied for this tag?
207
+ */
208
+ static VALUE implied_start_tag_eh(VALUE self)
209
+ {
210
+ htmlElemDesc * description;
211
+ Data_Get_Struct(self, htmlElemDesc, description);
212
+
213
+ if(description->startTag) return Qtrue;
214
+ return Qfalse;
215
+ }
216
+
217
+ /*
218
+ * call-seq:
219
+ * name
220
+ *
221
+ * Get the tag name for this ElemementDescription
222
+ */
223
+ static VALUE name(VALUE self)
224
+ {
225
+ htmlElemDesc * description;
226
+ Data_Get_Struct(self, htmlElemDesc, description);
227
+
228
+ if(NULL == description->name) return Qnil;
229
+ return NOKOGIRI_STR_NEW2(description->name);
230
+ }
231
+
232
+ /*
233
+ * call-seq:
234
+ * [](tag_name)
235
+ *
236
+ * Get ElemementDescription for +tag_name+
237
+ */
238
+ static VALUE get_description(VALUE klass, VALUE tag_name)
239
+ {
240
+ const htmlElemDesc * description = htmlTagLookup(
241
+ (const xmlChar *)StringValuePtr(tag_name)
242
+ );
243
+
244
+ if(NULL == description) return Qnil;
245
+ return Data_Wrap_Struct(klass, 0, 0, description);
246
+ }
247
+
248
+ VALUE cNokogiriHtmlElementDescription ;
249
+ void init_html_element_description()
250
+ {
251
+ VALUE nokogiri = rb_define_module("Nokogiri");
252
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
253
+ VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
254
+
255
+ cNokogiriHtmlElementDescription = klass;
256
+
257
+ rb_define_singleton_method(klass, "[]", get_description, 1);
258
+
259
+ rb_define_method(klass, "name", name, 0);
260
+ rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
261
+ rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
262
+ rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
263
+ rb_define_method(klass, "empty?", empty_eh, 0);
264
+ rb_define_method(klass, "deprecated?", deprecated_eh, 0);
265
+ rb_define_method(klass, "inline?", inline_eh, 0);
266
+ rb_define_method(klass, "description", description, 0);
267
+ rb_define_method(klass, "sub_elements", sub_elements, 0);
268
+ rb_define_method(klass, "default_sub_element", default_sub_element, 0);
269
+ rb_define_method(klass, "optional_attributes", optional_attributes, 0);
270
+ rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
271
+ rb_define_method(klass, "required_attributes", required_attributes, 0);
272
+ }
@@ -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
+
14
+ if(NULL == desc) return Qnil;
15
+ VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
16
+
17
+ VALUE args[3];
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 +1,92 @@
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
+ if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
19
+ if(!(int)RSTRING_LEN(data))
20
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
21
+
22
+ htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(
23
+ StringValuePtr(data),
24
+ (int)RSTRING_LEN(data)
25
+ );
26
+
27
+ if(RTEST(encoding)) {
28
+ xmlCharEncoding enc = xmlParseCharEncoding(StringValuePtr(encoding));
29
+ if(enc != XML_CHAR_ENCODING_ERROR) {
30
+ xmlSwitchEncoding(ctxt, enc);
31
+ if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
32
+ rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
33
+ StringValuePtr(encoding));
34
+ }
35
+ }
36
+ }
37
+
38
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
39
+ }
40
+
41
+ static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
42
+ {
43
+ htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
44
+ StringValuePtr(filename),
45
+ StringValuePtr(encoding)
46
+ );
47
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
48
+ }
49
+
50
+ static VALUE parse_with(VALUE self, VALUE sax_handler)
51
+ {
52
+ if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
53
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
54
+
55
+ htmlParserCtxtPtr ctxt;
56
+ Data_Get_Struct(self, htmlParserCtxt, ctxt);
57
+
58
+ htmlSAXHandlerPtr sax;
59
+ Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
60
+
61
+ // Free the sax handler since we'll assign our own
62
+ if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
63
+ xmlFree(ctxt->sax);
64
+
65
+ ctxt->sax = sax;
66
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
67
+
68
+ htmlParseDocument(ctxt);
69
+
70
+ if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
71
+
72
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
73
+ return self;
74
+ }
75
+
76
+ void init_html_sax_parser_context()
77
+ {
78
+ VALUE nokogiri = rb_define_module("Nokogiri");
79
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
80
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
81
+ VALUE sax = rb_define_module_under(xml, "SAX");
82
+ VALUE hsax = rb_define_module_under(html, "SAX");
83
+ VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
84
+ VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
85
+
86
+ cNokogiriHtmlSaxParserContext = klass;
87
+
88
+ rb_define_singleton_method(klass, "memory", parse_memory, 2);
89
+ rb_define_singleton_method(klass, "file", parse_file, 2);
90
+
91
+ rb_define_method(klass, "parse_with", parse_with, 1);
92
+ }
@@ -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 +1,89 @@
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
+ int is_2_6_16(void)
27
+ {
28
+ return (strcmp(xmlParserVersion, "20616") <= 0) ? 1 : 0 ;
29
+ }
30
+
31
+ void Init_nokogiri()
32
+ {
33
+ xmlMemSetup(
34
+ (xmlFreeFunc)ruby_xfree,
35
+ (xmlMallocFunc)ruby_xmalloc,
36
+ (xmlReallocFunc)ruby_xrealloc,
37
+ strdup
38
+ );
39
+
40
+ mNokogiri = rb_define_module("Nokogiri");
41
+ mNokogiriXml = rb_define_module_under(mNokogiri, "XML");
42
+ mNokogiriHtml = rb_define_module_under(mNokogiri, "HTML");
43
+ mNokogiriXslt = rb_define_module_under(mNokogiri, "XSLT");
44
+ mNokogiriXmlSax = rb_define_module_under(mNokogiriXml, "SAX");
45
+ mNokogiriHtmlSax = rb_define_module_under(mNokogiriHtml, "SAX");
46
+
47
+ rb_const_set( mNokogiri,
48
+ rb_intern("LIBXML_VERSION"),
49
+ NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
50
+ );
51
+ rb_const_set( mNokogiri,
52
+ rb_intern("LIBXML_PARSER_VERSION"),
53
+ NOKOGIRI_STR_NEW2(xmlParserVersion)
54
+ );
55
+
56
+ xmlInitParser();
57
+
58
+ init_xml_document();
59
+ init_html_document();
60
+ init_xml_node();
61
+ init_xml_document_fragment();
62
+ init_xml_text();
63
+ init_xml_cdata();
64
+ init_xml_processing_instruction();
65
+ init_xml_attr();
66
+ init_xml_entity_reference();
67
+ init_xml_comment();
68
+ init_xml_node_set();
69
+ init_xml_xpath_context();
70
+ init_xml_xpath();
71
+ init_xml_sax_parser_context();
72
+ init_xml_sax_parser();
73
+ init_xml_sax_push_parser();
74
+ init_xml_reader();
75
+ init_xml_dtd();
76
+ init_xml_element_content();
77
+ init_xml_attribute_decl();
78
+ init_xml_element_decl();
79
+ init_xml_entity_decl();
80
+ init_xml_namespace();
81
+ init_html_sax_parser_context();
82
+ init_xslt_stylesheet();
83
+ init_xml_syntax_error();
84
+ init_html_entity_lookup();
85
+ init_html_element_description();
86
+ init_xml_schema();
87
+ init_xml_relax_ng();
88
+ init_nokogiri_io();
89
+ }