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 +1,19 @@
1
+ module Nokogiri
2
+ module XML
3
+ class XPath
4
+
5
+ attr_accessor :cstruct # :nodoc:
6
+
7
+ def node_set # :nodoc:
8
+ ptr = cstruct[:nodesetval] if cstruct[:nodesetval]
9
+ ptr = LibXML.xmlXPathNodeSetCreate(nil) if ptr.null?
10
+
11
+ set = XML::NodeSet.new(@document)
12
+ set.cstruct = LibXML::XmlNodeSet.new(ptr)
13
+ set.document = @document
14
+ set
15
+ end
16
+
17
+ end
18
+ end
19
+ end
@@ -0,0 +1,135 @@
1
+ module Nokogiri
2
+ module XML
3
+ class XPathContext
4
+
5
+ attr_accessor :cstruct # :nodoc:
6
+
7
+ def register_ns(prefix, uri) # :nodoc:
8
+ LibXML.xmlXPathRegisterNs(cstruct, prefix, uri)
9
+ end
10
+
11
+ def evaluate(search_path, xpath_handler=nil) # :nodoc:
12
+ lookup = nil # to keep lambda in scope long enough to avoid a possible GC tragedy
13
+ query = search_path.to_s
14
+
15
+ if xpath_handler
16
+ lookup = lambda do |ctx, name, uri|
17
+ return nil unless xpath_handler.respond_to?(name)
18
+ ruby_funcall name, xpath_handler
19
+ end
20
+ LibXML.xmlXPathRegisterFuncLookup(cstruct, lookup, nil);
21
+ end
22
+
23
+ exception_handler = lambda do |ctx, error|
24
+ raise XPath::SyntaxError.wrap(error)
25
+ end
26
+ LibXML.xmlResetLastError()
27
+ LibXML.xmlSetStructuredErrorFunc(nil, exception_handler)
28
+
29
+ generic_exception_handler = lambda do |ctx, msg|
30
+ raise RuntimeError.new(msg) # TODO: varargs
31
+ end
32
+ LibXML.xmlSetGenericErrorFunc(nil, generic_exception_handler)
33
+
34
+ xpath_ptr = LibXML.xmlXPathEvalExpression(query, cstruct)
35
+
36
+ LibXML.xmlSetStructuredErrorFunc(nil, nil)
37
+ LibXML.xmlSetGenericErrorFunc(nil, nil)
38
+
39
+ if xpath_ptr.null?
40
+ error = LibXML.xmlGetLastError()
41
+ raise XPath::SyntaxError.wrap(error)
42
+ end
43
+
44
+ xpath = XML::XPath.new
45
+ xpath.cstruct = LibXML::XmlXpathObject.new(xpath_ptr)
46
+ xpath.document = cstruct[:doc]
47
+ xpath
48
+ end
49
+
50
+ def self.new(node) # :nodoc:
51
+ LibXML.xmlXPathInit()
52
+
53
+ ptr = LibXML.xmlXPathNewContext(node.cstruct[:doc])
54
+
55
+ ctx = allocate
56
+ ctx.cstruct = LibXML::XmlXpathContext.new(ptr)
57
+ ctx.cstruct[:node] = node.cstruct
58
+ ctx
59
+ end
60
+
61
+ private
62
+
63
+ #
64
+ # returns a lambda that will call the handler function with marshalled parameters
65
+ #
66
+ def ruby_funcall(name, xpath_handler) # :nodoc:
67
+ lambda do |ctx, nargs|
68
+ parser_context = LibXML::XmlXpathParserContext.new(ctx)
69
+ context = parser_context.context
70
+ doc = context.doc.ruby_doc
71
+
72
+ params = []
73
+
74
+ nargs.times do |j|
75
+ obj = LibXML::XmlXpathObject.new(LibXML.valuePop(ctx))
76
+ case obj[:type]
77
+ when LibXML::XmlXpathObject::XPATH_STRING
78
+ params.unshift obj[:stringval]
79
+ when LibXML::XmlXpathObject::XPATH_BOOLEAN
80
+ params.unshift obj[:boolval] == 1
81
+ when LibXML::XmlXpathObject::XPATH_NUMBER
82
+ params.unshift obj[:floatval]
83
+ when LibXML::XmlXpathObject::XPATH_NODESET
84
+ ns_ptr = LibXML::XmlNodeSet.new(obj[:nodesetval])
85
+ set = NodeSet.allocate
86
+ set.cstruct = ns_ptr
87
+ params.unshift set
88
+ else
89
+ char_ptr = params.unshift LibXML.xmlXPathCastToString(obj)
90
+ string = char_ptr.read_string
91
+ LibXML.xmlFree(char_ptr)
92
+ string
93
+ end
94
+ end
95
+
96
+ result = xpath_handler.send(name, *params)
97
+
98
+ case result.class.to_s
99
+ when Fixnum.to_s, Float.to_s, Bignum.to_s
100
+ LibXML.xmlXPathReturnNumber(ctx, result)
101
+ when String.to_s
102
+ LibXML.xmlXPathReturnString(
103
+ ctx,
104
+ LibXML.xmlXPathWrapCString(result)
105
+ )
106
+ when TrueClass.to_s
107
+ LibXML.xmlXPathReturnTrue(ctx)
108
+ when FalseClass.to_s
109
+ LibXML.xmlXPathReturnFalse(ctx)
110
+ when NilClass.to_s
111
+ ;
112
+ when Array.to_s
113
+ node_set = XML::NodeSet.new(doc, result)
114
+ LibXML.xmlXPathReturnNodeSet(
115
+ ctx,
116
+ LibXML.xmlXPathNodeSetMerge(nil, node_set.cstruct)
117
+ )
118
+ else
119
+ if result.is_a?(XML::NodeSet)
120
+ LibXML.xmlXPathReturnNodeSet(
121
+ ctx,
122
+ LibXML.xmlXPathNodeSetMerge(nil, result.cstruct)
123
+ )
124
+ else
125
+ raise RuntimeError.new("Invalid return type #{result.class.inspect}")
126
+ end
127
+ end
128
+
129
+ nil
130
+ end # lambda
131
+ end # ruby_funcall
132
+
133
+ end
134
+ end
135
+ end
@@ -0,0 +1,47 @@
1
+ module Nokogiri
2
+ module XSLT
3
+ class Stylesheet
4
+
5
+ attr_accessor :cstruct # :nodoc:
6
+
7
+ def self.parse_stylesheet_doc(document) # :nodoc:
8
+ LibXML.exsltRegisterAll
9
+
10
+ generic_exception_handler = lambda do |ctx, msg|
11
+ raise RuntimeError.new(msg) # TODO: varargs
12
+ end
13
+ LibXML.xsltSetGenericErrorFunc(nil, generic_exception_handler)
14
+
15
+ ss = LibXML.xsltParseStylesheetDoc(LibXML.xmlCopyDoc(document.cstruct, 1)) # 1 => recursive
16
+
17
+ LibXML.xsltSetGenericErrorFunc(nil, nil)
18
+
19
+ obj = allocate
20
+ obj.cstruct = LibXML::XsltStylesheet.new(ss)
21
+ obj
22
+ end
23
+
24
+ def serialize(document) # :nodoc:
25
+ buf_ptr = FFI::MemoryPointer.new :pointer
26
+ buf_len = FFI::MemoryPointer.new :int
27
+ LibXML.xsltSaveResultToString(buf_ptr, buf_len, document.cstruct, cstruct)
28
+ buf = Nokogiri::LibXML::XmlAlloc.new(buf_ptr.read_pointer)
29
+ buf.pointer.read_string(buf_len.read_int)
30
+ end
31
+
32
+ def transform(document, params=[]) # :nodoc:
33
+ param_arr = FFI::MemoryPointer.new(:pointer, params.length + 1)
34
+ params.each_with_index do |param, j|
35
+ param_arr[j].put_pointer(0, FFI::MemoryPointer.from_string(param.to_s))
36
+ end
37
+ param_arr[params.length].put_pointer(0,nil)
38
+
39
+ ptr = LibXML.xsltApplyStylesheet(cstruct, document.cstruct, param_arr)
40
+ raise(RuntimeError, "could not perform xslt transform on document") if ptr.null?
41
+
42
+ XML::Document.wrap(ptr)
43
+ end
44
+
45
+ end
46
+ end
47
+ end
@@ -0,0 +1,35 @@
1
+ require 'nokogiri/html/entity_lookup'
2
+ require 'nokogiri/html/document'
3
+ require 'nokogiri/html/document_fragment'
4
+ require 'nokogiri/html/sax/parser_context'
5
+ require 'nokogiri/html/sax/parser'
6
+ require 'nokogiri/html/element_description'
7
+
8
+ module Nokogiri
9
+ class << self
10
+ ###
11
+ # Parse HTML. Convenience method for Nokogiri::HTML::Document.parse
12
+ def HTML thing, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block
13
+ Nokogiri::HTML::Document.parse(thing, url, encoding, options, &block)
14
+ end
15
+ end
16
+
17
+ module HTML
18
+ class << self
19
+ ###
20
+ # Parse HTML. Convenience method for Nokogiri::HTML::Document.parse
21
+ def parse thing, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block
22
+ Document.parse(thing, url, encoding, options, &block)
23
+ end
24
+
25
+ ####
26
+ # Parse a fragment from +string+ in to a NodeSet.
27
+ def fragment string
28
+ HTML::DocumentFragment.parse(string)
29
+ end
30
+ end
31
+
32
+ # Instance of Nokogiri::HTML::EntityLookup
33
+ NamedCharacters = EntityLookup.new
34
+ end
35
+ end
@@ -0,0 +1,35 @@
1
+ module Nokogiri
2
+ module HTML
3
+ ###
4
+ # Nokogiri HTML builder is used for building HTML documents. It is very
5
+ # similar to the Nokogiri::XML::Builder. In fact, you should go read the
6
+ # documentation for Nokogiri::XML::Builder before reading this
7
+ # documentation.
8
+ #
9
+ # == Synopsis:
10
+ #
11
+ # Create an HTML document with a body that has an onload attribute, and a
12
+ # span tag with a class of "bold" that has content of "Hello world".
13
+ #
14
+ # builder = Nokogiri::HTML::Builder.new do |doc|
15
+ # doc.html {
16
+ # doc.body(:onload => 'some_func();') {
17
+ # doc.span.bold {
18
+ # doc.text "Hello world"
19
+ # }
20
+ # }
21
+ # }
22
+ # end
23
+ # puts builder.to_html
24
+ #
25
+ # The HTML builder inherits from the XML builder, so make sure to read the
26
+ # Nokogiri::XML::Builder documentation.
27
+ class Builder < Nokogiri::XML::Builder
28
+ ###
29
+ # Convert the builder to HTML
30
+ def to_html
31
+ @doc.to_html
32
+ end
33
+ end
34
+ end
35
+ end
@@ -0,0 +1,88 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class Document < Nokogiri::XML::Document
4
+ ###
5
+ # Get the meta tag encoding for this document. If there is no meta tag,
6
+ # then nil is returned
7
+ def meta_encoding
8
+ return nil unless meta = css('meta').find { |node|
9
+ node['http-equiv'] =~ /Content-Type/i
10
+ }
11
+
12
+ /charset\s*=\s*([\w\d-]+)/i.match(meta['content'])[1]
13
+ end
14
+
15
+ ###
16
+ # Set the meta tag encoding for this document. If there is no meta
17
+ # content tag, nil is returned and the encoding is not set.
18
+ def meta_encoding= encoding
19
+ return nil unless meta = css('meta').find { |node|
20
+ node['http-equiv'] =~ /Content-Type/i
21
+ }
22
+
23
+ meta['content'] = "text/html; charset=%s" % encoding
24
+ encoding
25
+ end
26
+
27
+ ####
28
+ # Serialize Node using +options+. Save options can also be set using a
29
+ # block. See SaveOptions.
30
+ #
31
+ # These two statements are equivalent:
32
+ #
33
+ # node.serialize(:encoding => 'UTF-8', :save_with => FORMAT | AS_XML)
34
+ #
35
+ # or
36
+ #
37
+ # node.serialize(:encoding => 'UTF-8') do |config|
38
+ # config.format.as_xml
39
+ # end
40
+ #
41
+ def serialize options = {}, &block
42
+ options[:save_with] ||= XML::Node::SaveOptions::FORMAT |
43
+ XML::Node::SaveOptions::AS_HTML |
44
+ XML::Node::SaveOptions::NO_DECLARATION |
45
+ XML::Node::SaveOptions::NO_EMPTY_TAGS
46
+ super
47
+ end
48
+
49
+ ####
50
+ # Create a Nokogiri::XML::DocumentFragment from +tags+
51
+ def fragment tags = nil
52
+ DocumentFragment.new(self, tags)
53
+ end
54
+
55
+ class << self
56
+ ###
57
+ # Parse HTML. +thing+ may be a String, or any object that
58
+ # responds to _read_ and _close_ such as an IO, or StringIO.
59
+ # +url+ is resource where this document is located. +encoding+ is the
60
+ # encoding that should be used when processing the document. +options+
61
+ # is a number that sets options in the parser, such as
62
+ # Nokogiri::XML::ParseOptions::RECOVER. See the constants in
63
+ # Nokogiri::XML::ParseOptions.
64
+ def parse string_or_io, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block
65
+
66
+ options = Nokogiri::XML::ParseOptions.new(options) if Fixnum === options
67
+ # Give the options to the user
68
+ yield options if block_given?
69
+
70
+ if string_or_io.respond_to?(:encoding)
71
+ encoding ||= string_or_io.encoding.name
72
+ end
73
+
74
+ if string_or_io.respond_to?(:read)
75
+ url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
76
+ return read_io(string_or_io, url, encoding, options.to_i)
77
+ end
78
+
79
+ # read_memory pukes on empty docs
80
+ return new if string_or_io.nil? or string_or_io.empty?
81
+
82
+ read_memory(string_or_io, url, encoding, options.to_i)
83
+ end
84
+ end
85
+
86
+ end
87
+ end
88
+ end
@@ -0,0 +1,15 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class DocumentFragment < Nokogiri::XML::DocumentFragment
4
+
5
+ class << self
6
+ ####
7
+ # Create a Nokogiri::XML::DocumentFragment from +tags+
8
+ def parse tags
9
+ self.new(HTML::Document.new, tags)
10
+ end
11
+ end
12
+
13
+ end
14
+ end
15
+ end
@@ -0,0 +1,23 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class ElementDescription
4
+ ###
5
+ # Is this element a block element?
6
+ def block?
7
+ !inline?
8
+ end
9
+
10
+ ###
11
+ # Convert this description to a string
12
+ def to_s
13
+ "#{name}: #{description}"
14
+ end
15
+
16
+ ###
17
+ # Inspection information
18
+ def inspect
19
+ "#<#{self.class.name}: #{name} #{description}>"
20
+ end
21
+ end
22
+ end
23
+ end
@@ -0,0 +1,13 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class EntityDescription < Struct.new(:value, :name, :description); end
4
+
5
+ class EntityLookup
6
+ ###
7
+ # Look up entity with +name+
8
+ def [] name
9
+ (val = get(name)) && val.value
10
+ end
11
+ end
12
+ end
13
+ end
@@ -0,0 +1,48 @@
1
+ module Nokogiri
2
+ module HTML
3
+ ###
4
+ # Nokogiri lets you write a SAX parser to process HTML but get HTML
5
+ # correction features.
6
+ #
7
+ # See Nokogiri::HTML::SAX::Parser for a basic example of using a
8
+ # SAX parser with HTML.
9
+ #
10
+ # For more information on SAX parsers, see Nokogiri::XML::SAX
11
+ module SAX
12
+ ###
13
+ # This class lets you perform SAX style parsing on HTML with HTML
14
+ # error correction.
15
+ #
16
+ # Here is a basic usage example:
17
+ #
18
+ # class MyDoc < Nokogiri::XML::SAX::Document
19
+ # def start_element name, attributes = []
20
+ # puts "found a #{name}"
21
+ # end
22
+ # end
23
+ #
24
+ # parser = Nokogiri::HTML::SAX::Parser.new(MyDoc.new)
25
+ # parser.parse(File.read(ARGV[0], 'rb'))
26
+ #
27
+ # For more information on SAX parsers, see Nokogiri::XML::SAX
28
+ class Parser < Nokogiri::XML::SAX::Parser
29
+ ###
30
+ # Parse html stored in +data+ using +encoding+
31
+ def parse_memory data, encoding = 'UTF-8'
32
+ raise ArgumentError unless data
33
+ return unless data.length > 0
34
+ ParserContext.memory(data, encoding).parse_with self
35
+ end
36
+
37
+ ###
38
+ # Parse a file with +filename+
39
+ def parse_file filename, encoding = 'UTF-8'
40
+ raise ArgumentError unless filename
41
+ raise Errno::ENOENT unless File.exists?(filename)
42
+ raise Errno::EISDIR if File.directory?(filename)
43
+ ParserContext.file(filename, encoding).parse_with self
44
+ end
45
+ end
46
+ end
47
+ end
48
+ end