glebm-nokogiri 1.4.2.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (277) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +411 -0
  3. data/CHANGELOG.rdoc +397 -0
  4. data/Manifest.txt +276 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +132 -0
  7. data/Rakefile +183 -0
  8. data/bin/nokogiri +49 -0
  9. data/deps.rip +5 -0
  10. data/ext/nokogiri/extconf.rb +97 -0
  11. data/ext/nokogiri/html_document.c +154 -0
  12. data/ext/nokogiri/html_document.h +10 -0
  13. data/ext/nokogiri/html_element_description.c +276 -0
  14. data/ext/nokogiri/html_element_description.h +10 -0
  15. data/ext/nokogiri/html_entity_lookup.c +32 -0
  16. data/ext/nokogiri/html_entity_lookup.h +8 -0
  17. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  18. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  19. data/ext/nokogiri/nokogiri.c +95 -0
  20. data/ext/nokogiri/nokogiri.h +153 -0
  21. data/ext/nokogiri/xml_attr.c +94 -0
  22. data/ext/nokogiri/xml_attr.h +9 -0
  23. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  24. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  25. data/ext/nokogiri/xml_cdata.c +56 -0
  26. data/ext/nokogiri/xml_cdata.h +9 -0
  27. data/ext/nokogiri/xml_comment.c +54 -0
  28. data/ext/nokogiri/xml_comment.h +9 -0
  29. data/ext/nokogiri/xml_document.c +464 -0
  30. data/ext/nokogiri/xml_document.h +23 -0
  31. data/ext/nokogiri/xml_document_fragment.c +48 -0
  32. data/ext/nokogiri/xml_document_fragment.h +10 -0
  33. data/ext/nokogiri/xml_dtd.c +202 -0
  34. data/ext/nokogiri/xml_dtd.h +10 -0
  35. data/ext/nokogiri/xml_element_content.c +123 -0
  36. data/ext/nokogiri/xml_element_content.h +10 -0
  37. data/ext/nokogiri/xml_element_decl.c +69 -0
  38. data/ext/nokogiri/xml_element_decl.h +9 -0
  39. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  40. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  41. data/ext/nokogiri/xml_entity_decl.c +110 -0
  42. data/ext/nokogiri/xml_entity_decl.h +10 -0
  43. data/ext/nokogiri/xml_entity_reference.c +52 -0
  44. data/ext/nokogiri/xml_entity_reference.h +9 -0
  45. data/ext/nokogiri/xml_io.c +31 -0
  46. data/ext/nokogiri/xml_io.h +11 -0
  47. data/ext/nokogiri/xml_namespace.c +84 -0
  48. data/ext/nokogiri/xml_namespace.h +13 -0
  49. data/ext/nokogiri/xml_node.c +1347 -0
  50. data/ext/nokogiri/xml_node.h +13 -0
  51. data/ext/nokogiri/xml_node_set.c +418 -0
  52. data/ext/nokogiri/xml_node_set.h +9 -0
  53. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  54. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  55. data/ext/nokogiri/xml_reader.c +665 -0
  56. data/ext/nokogiri/xml_reader.h +10 -0
  57. data/ext/nokogiri/xml_relax_ng.c +168 -0
  58. data/ext/nokogiri/xml_relax_ng.h +9 -0
  59. data/ext/nokogiri/xml_sax_parser.c +286 -0
  60. data/ext/nokogiri/xml_sax_parser.h +39 -0
  61. data/ext/nokogiri/xml_sax_parser_context.c +159 -0
  62. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  63. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  64. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  65. data/ext/nokogiri/xml_schema.c +205 -0
  66. data/ext/nokogiri/xml_schema.h +9 -0
  67. data/ext/nokogiri/xml_syntax_error.c +58 -0
  68. data/ext/nokogiri/xml_syntax_error.h +13 -0
  69. data/ext/nokogiri/xml_text.c +50 -0
  70. data/ext/nokogiri/xml_text.h +9 -0
  71. data/ext/nokogiri/xml_xpath_context.c +276 -0
  72. data/ext/nokogiri/xml_xpath_context.h +9 -0
  73. data/ext/nokogiri/xslt_stylesheet.c +142 -0
  74. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  75. data/lib/nokogiri.rb +133 -0
  76. data/lib/nokogiri/css.rb +25 -0
  77. data/lib/nokogiri/css/generated_parser.rb +669 -0
  78. data/lib/nokogiri/css/generated_tokenizer.rb +145 -0
  79. data/lib/nokogiri/css/node.rb +99 -0
  80. data/lib/nokogiri/css/parser.rb +82 -0
  81. data/lib/nokogiri/css/parser.y +232 -0
  82. data/lib/nokogiri/css/syntax_error.rb +7 -0
  83. data/lib/nokogiri/css/tokenizer.rb +7 -0
  84. data/lib/nokogiri/css/tokenizer.rex +55 -0
  85. data/lib/nokogiri/css/xpath_visitor.rb +169 -0
  86. data/lib/nokogiri/decorators/slop.rb +33 -0
  87. data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
  88. data/lib/nokogiri/ffi/html/document.rb +28 -0
  89. data/lib/nokogiri/ffi/html/element_description.rb +81 -0
  90. data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
  91. data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
  92. data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
  93. data/lib/nokogiri/ffi/libxml.rb +386 -0
  94. data/lib/nokogiri/ffi/structs/common_node.rb +38 -0
  95. data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
  96. data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
  97. data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
  98. data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
  99. data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
  100. data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
  101. data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
  102. data/lib/nokogiri/ffi/structs/xml_document.rb +117 -0
  103. data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
  104. data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
  105. data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
  106. data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
  107. data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
  108. data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
  109. data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
  110. data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
  111. data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
  112. data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
  113. data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
  114. data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
  115. data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
  116. data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
  117. data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
  118. data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
  119. data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +38 -0
  120. data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
  121. data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
  122. data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
  123. data/lib/nokogiri/ffi/weak_bucket.rb +40 -0
  124. data/lib/nokogiri/ffi/xml/attr.rb +41 -0
  125. data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
  126. data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
  127. data/lib/nokogiri/ffi/xml/comment.rb +18 -0
  128. data/lib/nokogiri/ffi/xml/document.rb +162 -0
  129. data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
  130. data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
  131. data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
  132. data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
  133. data/lib/nokogiri/ffi/xml/entity_decl.rb +36 -0
  134. data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
  135. data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
  136. data/lib/nokogiri/ffi/xml/node.rb +556 -0
  137. data/lib/nokogiri/ffi/xml/node_set.rb +149 -0
  138. data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
  139. data/lib/nokogiri/ffi/xml/reader.rb +232 -0
  140. data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
  141. data/lib/nokogiri/ffi/xml/sax/parser.rb +135 -0
  142. data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
  143. data/lib/nokogiri/ffi/xml/sax/push_parser.rb +51 -0
  144. data/lib/nokogiri/ffi/xml/schema.rb +109 -0
  145. data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
  146. data/lib/nokogiri/ffi/xml/text.rb +18 -0
  147. data/lib/nokogiri/ffi/xml/xpath.rb +9 -0
  148. data/lib/nokogiri/ffi/xml/xpath_context.rb +148 -0
  149. data/lib/nokogiri/ffi/xslt/stylesheet.rb +53 -0
  150. data/lib/nokogiri/html.rb +35 -0
  151. data/lib/nokogiri/html/builder.rb +35 -0
  152. data/lib/nokogiri/html/document.rb +90 -0
  153. data/lib/nokogiri/html/document_fragment.rb +36 -0
  154. data/lib/nokogiri/html/element_description.rb +23 -0
  155. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  156. data/lib/nokogiri/html/sax/parser.rb +48 -0
  157. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  158. data/lib/nokogiri/syntax_error.rb +4 -0
  159. data/lib/nokogiri/version.rb +37 -0
  160. data/lib/nokogiri/version_warning.rb +14 -0
  161. data/lib/nokogiri/xml.rb +67 -0
  162. data/lib/nokogiri/xml/attr.rb +14 -0
  163. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  164. data/lib/nokogiri/xml/builder.rb +418 -0
  165. data/lib/nokogiri/xml/cdata.rb +11 -0
  166. data/lib/nokogiri/xml/character_data.rb +7 -0
  167. data/lib/nokogiri/xml/document.rb +194 -0
  168. data/lib/nokogiri/xml/document_fragment.rb +77 -0
  169. data/lib/nokogiri/xml/dtd.rb +11 -0
  170. data/lib/nokogiri/xml/element_content.rb +36 -0
  171. data/lib/nokogiri/xml/element_decl.rb +13 -0
  172. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  173. data/lib/nokogiri/xml/namespace.rb +13 -0
  174. data/lib/nokogiri/xml/node.rb +793 -0
  175. data/lib/nokogiri/xml/node/save_options.rb +42 -0
  176. data/lib/nokogiri/xml/node_set.rb +325 -0
  177. data/lib/nokogiri/xml/notation.rb +6 -0
  178. data/lib/nokogiri/xml/parse_options.rb +85 -0
  179. data/lib/nokogiri/xml/pp.rb +2 -0
  180. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  181. data/lib/nokogiri/xml/pp/node.rb +56 -0
  182. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  183. data/lib/nokogiri/xml/reader.rb +74 -0
  184. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  185. data/lib/nokogiri/xml/sax.rb +4 -0
  186. data/lib/nokogiri/xml/sax/document.rb +160 -0
  187. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  188. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  189. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  190. data/lib/nokogiri/xml/schema.rb +57 -0
  191. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  192. data/lib/nokogiri/xml/text.rb +9 -0
  193. data/lib/nokogiri/xml/xpath.rb +10 -0
  194. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  195. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  196. data/lib/nokogiri/xslt.rb +48 -0
  197. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  198. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  199. data/tasks/cross_compile.rb +158 -0
  200. data/tasks/test.rb +94 -0
  201. data/test/css/test_nthiness.rb +159 -0
  202. data/test/css/test_parser.rb +282 -0
  203. data/test/css/test_tokenizer.rb +190 -0
  204. data/test/css/test_xpath_visitor.rb +85 -0
  205. data/test/ffi/test_document.rb +35 -0
  206. data/test/files/2ch.html +108 -0
  207. data/test/files/address_book.rlx +12 -0
  208. data/test/files/address_book.xml +10 -0
  209. data/test/files/bar/bar.xsd +4 -0
  210. data/test/files/dont_hurt_em_why.xml +422 -0
  211. data/test/files/exslt.xml +8 -0
  212. data/test/files/exslt.xslt +35 -0
  213. data/test/files/foo/foo.xsd +4 -0
  214. data/test/files/po.xml +32 -0
  215. data/test/files/po.xsd +66 -0
  216. data/test/files/shift_jis.html +10 -0
  217. data/test/files/shift_jis.xml +5 -0
  218. data/test/files/snuggles.xml +3 -0
  219. data/test/files/staff.dtd +10 -0
  220. data/test/files/staff.xml +59 -0
  221. data/test/files/staff.xslt +32 -0
  222. data/test/files/tlm.html +850 -0
  223. data/test/files/valid_bar.xml +2 -0
  224. data/test/helper.rb +169 -0
  225. data/test/html/sax/test_parser.rb +74 -0
  226. data/test/html/sax/test_parser_context.rb +48 -0
  227. data/test/html/test_builder.rb +164 -0
  228. data/test/html/test_document.rb +398 -0
  229. data/test/html/test_document_encoding.rb +77 -0
  230. data/test/html/test_document_fragment.rb +182 -0
  231. data/test/html/test_element_description.rb +98 -0
  232. data/test/html/test_named_characters.rb +14 -0
  233. data/test/html/test_node.rb +181 -0
  234. data/test/html/test_node_encoding.rb +27 -0
  235. data/test/test_convert_xpath.rb +135 -0
  236. data/test/test_css_cache.rb +45 -0
  237. data/test/test_encoding_handler.rb +46 -0
  238. data/test/test_memory_leak.rb +87 -0
  239. data/test/test_nokogiri.rb +138 -0
  240. data/test/test_reader.rb +386 -0
  241. data/test/test_soap4r_sax.rb +52 -0
  242. data/test/test_xslt_transforms.rb +188 -0
  243. data/test/xml/node/test_save_options.rb +20 -0
  244. data/test/xml/node/test_subclass.rb +44 -0
  245. data/test/xml/sax/test_parser.rb +307 -0
  246. data/test/xml/sax/test_parser_context.rb +63 -0
  247. data/test/xml/sax/test_push_parser.rb +139 -0
  248. data/test/xml/test_attr.rb +38 -0
  249. data/test/xml/test_attribute_decl.rb +82 -0
  250. data/test/xml/test_builder.rb +210 -0
  251. data/test/xml/test_cdata.rb +50 -0
  252. data/test/xml/test_comment.rb +29 -0
  253. data/test/xml/test_document.rb +668 -0
  254. data/test/xml/test_document_encoding.rb +26 -0
  255. data/test/xml/test_document_fragment.rb +180 -0
  256. data/test/xml/test_dtd.rb +82 -0
  257. data/test/xml/test_dtd_encoding.rb +33 -0
  258. data/test/xml/test_element_content.rb +56 -0
  259. data/test/xml/test_element_decl.rb +73 -0
  260. data/test/xml/test_entity_decl.rb +120 -0
  261. data/test/xml/test_entity_reference.rb +21 -0
  262. data/test/xml/test_namespace.rb +68 -0
  263. data/test/xml/test_node.rb +865 -0
  264. data/test/xml/test_node_attributes.rb +34 -0
  265. data/test/xml/test_node_encoding.rb +107 -0
  266. data/test/xml/test_node_reparenting.rb +293 -0
  267. data/test/xml/test_node_set.rb +649 -0
  268. data/test/xml/test_parse_options.rb +52 -0
  269. data/test/xml/test_processing_instruction.rb +30 -0
  270. data/test/xml/test_reader_encoding.rb +126 -0
  271. data/test/xml/test_relax_ng.rb +60 -0
  272. data/test/xml/test_schema.rb +89 -0
  273. data/test/xml/test_syntax_error.rb +12 -0
  274. data/test/xml/test_text.rb +38 -0
  275. data/test/xml/test_unparented_node.rb +381 -0
  276. data/test/xml/test_xpath.rb +138 -0
  277. metadata +533 -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 +1,276 @@
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
+ VALUE list;
13
+ int i;
14
+
15
+ Data_Get_Struct(self, htmlElemDesc, description);
16
+
17
+ list = rb_ary_new();
18
+
19
+ if(NULL == description->attrs_req) return list;
20
+
21
+ for(i = 0; description->attrs_depr[i]; i++) {
22
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
23
+ }
24
+
25
+ return list;
26
+ }
27
+
28
+ /*
29
+ * call-seq:
30
+ * deprecated_attributes
31
+ *
32
+ * A list of deprecated attributes for this element
33
+ */
34
+ static VALUE deprecated_attributes(VALUE self)
35
+ {
36
+ htmlElemDesc * description;
37
+ VALUE list;
38
+ int i;
39
+
40
+ Data_Get_Struct(self, htmlElemDesc, description);
41
+
42
+ list = rb_ary_new();
43
+
44
+ if(NULL == description->attrs_depr) return list;
45
+
46
+ for(i = 0; description->attrs_depr[i]; i++) {
47
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
48
+ }
49
+
50
+ return list;
51
+ }
52
+
53
+ /*
54
+ * call-seq:
55
+ * optional_attributes
56
+ *
57
+ * A list of optional attributes for this element
58
+ */
59
+ static VALUE optional_attributes(VALUE self)
60
+ {
61
+ htmlElemDesc * description;
62
+ VALUE list;
63
+ int i;
64
+
65
+ Data_Get_Struct(self, htmlElemDesc, description);
66
+
67
+ list = rb_ary_new();
68
+
69
+ if(NULL == description->attrs_opt) return list;
70
+
71
+ for(i = 0; description->attrs_opt[i]; i++) {
72
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
73
+ }
74
+
75
+ return list;
76
+ }
77
+
78
+ /*
79
+ * call-seq:
80
+ * default_sub_element
81
+ *
82
+ * The default sub element for this element
83
+ */
84
+ static VALUE default_sub_element(VALUE self)
85
+ {
86
+ htmlElemDesc * description;
87
+ Data_Get_Struct(self, htmlElemDesc, description);
88
+
89
+ return NOKOGIRI_STR_NEW2(description->defaultsubelt);
90
+ }
91
+
92
+ /*
93
+ * call-seq:
94
+ * sub_elements
95
+ *
96
+ * A list of allowed sub elements for this element.
97
+ */
98
+ static VALUE sub_elements(VALUE self)
99
+ {
100
+ htmlElemDesc * description;
101
+ VALUE list;
102
+ int i;
103
+
104
+ Data_Get_Struct(self, htmlElemDesc, description);
105
+
106
+ list = rb_ary_new();
107
+
108
+ if(NULL == description->subelts) return list;
109
+
110
+ for(i = 0; description->subelts[i]; i++) {
111
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
112
+ }
113
+
114
+ return list;
115
+ }
116
+
117
+ /*
118
+ * call-seq:
119
+ * description
120
+ *
121
+ * The description for this element
122
+ */
123
+ static VALUE description(VALUE self)
124
+ {
125
+ htmlElemDesc * description;
126
+ Data_Get_Struct(self, htmlElemDesc, description);
127
+
128
+ return NOKOGIRI_STR_NEW2(description->desc);
129
+ }
130
+
131
+ /*
132
+ * call-seq:
133
+ * inline?
134
+ *
135
+ * Is this element an inline element?
136
+ */
137
+ static VALUE inline_eh(VALUE self)
138
+ {
139
+ htmlElemDesc * description;
140
+ Data_Get_Struct(self, htmlElemDesc, description);
141
+
142
+ if(description->isinline) return Qtrue;
143
+ return Qfalse;
144
+ }
145
+
146
+ /*
147
+ * call-seq:
148
+ * deprecated?
149
+ *
150
+ * Is this element deprecated?
151
+ */
152
+ static VALUE deprecated_eh(VALUE self)
153
+ {
154
+ htmlElemDesc * description;
155
+ Data_Get_Struct(self, htmlElemDesc, description);
156
+
157
+ if(description->depr) return Qtrue;
158
+ return Qfalse;
159
+ }
160
+
161
+ /*
162
+ * call-seq:
163
+ * empty?
164
+ *
165
+ * Is this an empty element?
166
+ */
167
+ static VALUE empty_eh(VALUE self)
168
+ {
169
+ htmlElemDesc * description;
170
+ Data_Get_Struct(self, htmlElemDesc, description);
171
+
172
+ if(description->empty) return Qtrue;
173
+ return Qfalse;
174
+ }
175
+
176
+ /*
177
+ * call-seq:
178
+ * save_end_tag?
179
+ *
180
+ * Should the end tag be saved?
181
+ */
182
+ static VALUE save_end_tag_eh(VALUE self)
183
+ {
184
+ htmlElemDesc * description;
185
+ Data_Get_Struct(self, htmlElemDesc, description);
186
+
187
+ if(description->saveEndTag) return Qtrue;
188
+ return Qfalse;
189
+ }
190
+
191
+ /*
192
+ * call-seq:
193
+ * implied_end_tag?
194
+ *
195
+ * Can the end tag be implied for this tag?
196
+ */
197
+ static VALUE implied_end_tag_eh(VALUE self)
198
+ {
199
+ htmlElemDesc * description;
200
+ Data_Get_Struct(self, htmlElemDesc, description);
201
+
202
+ if(description->endTag) return Qtrue;
203
+ return Qfalse;
204
+ }
205
+
206
+ /*
207
+ * call-seq:
208
+ * implied_start_tag?
209
+ *
210
+ * Can the start tag be implied for this tag?
211
+ */
212
+ static VALUE implied_start_tag_eh(VALUE self)
213
+ {
214
+ htmlElemDesc * description;
215
+ Data_Get_Struct(self, htmlElemDesc, description);
216
+
217
+ if(description->startTag) return Qtrue;
218
+ return Qfalse;
219
+ }
220
+
221
+ /*
222
+ * call-seq:
223
+ * name
224
+ *
225
+ * Get the tag name for this ElemementDescription
226
+ */
227
+ static VALUE name(VALUE self)
228
+ {
229
+ htmlElemDesc * description;
230
+ Data_Get_Struct(self, htmlElemDesc, description);
231
+
232
+ if(NULL == description->name) return Qnil;
233
+ return NOKOGIRI_STR_NEW2(description->name);
234
+ }
235
+
236
+ /*
237
+ * call-seq:
238
+ * [](tag_name)
239
+ *
240
+ * Get ElemementDescription for +tag_name+
241
+ */
242
+ static VALUE get_description(VALUE klass, VALUE tag_name)
243
+ {
244
+ const htmlElemDesc * description = htmlTagLookup(
245
+ (const xmlChar *)StringValuePtr(tag_name)
246
+ );
247
+
248
+ if(NULL == description) return Qnil;
249
+ return Data_Wrap_Struct(klass, 0, 0, (void *)description);
250
+ }
251
+
252
+ VALUE cNokogiriHtmlElementDescription ;
253
+ void init_html_element_description()
254
+ {
255
+ VALUE nokogiri = rb_define_module("Nokogiri");
256
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
257
+ VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
258
+
259
+ cNokogiriHtmlElementDescription = klass;
260
+
261
+ rb_define_singleton_method(klass, "[]", get_description, 1);
262
+
263
+ rb_define_method(klass, "name", name, 0);
264
+ rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
265
+ rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
266
+ rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
267
+ rb_define_method(klass, "empty?", empty_eh, 0);
268
+ rb_define_method(klass, "deprecated?", deprecated_eh, 0);
269
+ rb_define_method(klass, "inline?", inline_eh, 0);
270
+ rb_define_method(klass, "description", description, 0);
271
+ rb_define_method(klass, "sub_elements", sub_elements, 0);
272
+ rb_define_method(klass, "default_sub_element", default_sub_element, 0);
273
+ rb_define_method(klass, "optional_attributes", optional_attributes, 0);
274
+ rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
275
+ rb_define_method(klass, "required_attributes", required_attributes, 0);
276
+ }
@@ -0,0 +1,10 @@
1
+ #ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
2
+ #define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_element_description();
7
+
8
+ extern VALUE cNokogiriHtmlElementDescription ;
9
+
10
+ #endif
@@ -0,0 +1,32 @@
1
+ #include <html_entity_lookup.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * get(key)
6
+ *
7
+ * Get the HTML::EntityDescription for +key+
8
+ */
9
+ static VALUE get(VALUE self, VALUE key)
10
+ {
11
+ const htmlEntityDesc * desc =
12
+ htmlEntityLookup((const xmlChar *)StringValuePtr(key));
13
+ VALUE klass, args[3];
14
+
15
+ if(NULL == desc) return Qnil;
16
+ klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
17
+
18
+ args[0] = INT2NUM((long)desc->value);
19
+ args[1] = NOKOGIRI_STR_NEW2(desc->name);
20
+ args[2] = NOKOGIRI_STR_NEW2(desc->desc);
21
+
22
+ return rb_class_new_instance(3, args, klass);
23
+ }
24
+
25
+ void init_html_entity_lookup()
26
+ {
27
+ VALUE nokogiri = rb_define_module("Nokogiri");
28
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
29
+ VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
30
+
31
+ rb_define_method(klass, "get", get, 1);
32
+ }
@@ -0,0 +1,8 @@
1
+ #ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
2
+ #define NOKOGIRI_HTML_ENTITY_LOOKUP
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_entity_lookup();
7
+
8
+ #endif
@@ -0,0 +1,94 @@
1
+ #include <html_sax_parser_context.h>
2
+
3
+ VALUE cNokogiriHtmlSaxParserContext ;
4
+
5
+ static void deallocate(xmlParserCtxtPtr ctxt)
6
+ {
7
+ NOKOGIRI_DEBUG_START(handler);
8
+
9
+ ctxt->sax = NULL;
10
+
11
+ htmlFreeParserCtxt(ctxt);
12
+
13
+ NOKOGIRI_DEBUG_END(handler);
14
+ }
15
+
16
+ static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
17
+ {
18
+ htmlParserCtxtPtr ctxt;
19
+
20
+ if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
21
+ if(!(int)RSTRING_LEN(data))
22
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
23
+
24
+ ctxt = htmlCreateMemoryParserCtxt(
25
+ StringValuePtr(data),
26
+ (int)RSTRING_LEN(data)
27
+ );
28
+
29
+ if(RTEST(encoding)) {
30
+ xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValuePtr(encoding));
31
+ if(enc != NULL) {
32
+ xmlSwitchToEncoding(ctxt, enc);
33
+ if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
34
+ rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
35
+ StringValuePtr(encoding));
36
+ }
37
+ }
38
+ }
39
+
40
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
41
+ }
42
+
43
+ static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
44
+ {
45
+ htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
46
+ StringValuePtr(filename),
47
+ StringValuePtr(encoding)
48
+ );
49
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
50
+ }
51
+
52
+ static VALUE parse_with(VALUE self, VALUE sax_handler)
53
+ {
54
+ htmlParserCtxtPtr ctxt;
55
+ htmlSAXHandlerPtr sax;
56
+
57
+ if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
58
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
59
+
60
+ Data_Get_Struct(self, htmlParserCtxt, ctxt);
61
+ Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
62
+
63
+ /* Free the sax handler since we'll assign our own */
64
+ if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
65
+ xmlFree(ctxt->sax);
66
+
67
+ ctxt->sax = sax;
68
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
69
+
70
+ htmlParseDocument(ctxt);
71
+
72
+ if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
73
+
74
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
75
+ return self;
76
+ }
77
+
78
+ void init_html_sax_parser_context()
79
+ {
80
+ VALUE nokogiri = rb_define_module("Nokogiri");
81
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
82
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
83
+ VALUE sax = rb_define_module_under(xml, "SAX");
84
+ VALUE hsax = rb_define_module_under(html, "SAX");
85
+ VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
86
+ VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
87
+
88
+ cNokogiriHtmlSaxParserContext = klass;
89
+
90
+ rb_define_singleton_method(klass, "memory", parse_memory, 2);
91
+ rb_define_singleton_method(klass, "file", parse_file, 2);
92
+
93
+ rb_define_method(klass, "parse_with", parse_with, 1);
94
+ }
@@ -0,0 +1,11 @@
1
+ #ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
2
+ #define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ extern VALUE cNokogiriHtmlSaxParserContext;
7
+
8
+ void init_html_sax_parser_context();
9
+
10
+ #endif
11
+