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 +1,142 @@
1
+ module Nokogiri
2
+ module XML
3
+ module SAX
4
+ class Parser
5
+ # :stopdoc:
6
+
7
+ attr_accessor :cstruct
8
+
9
+ def self.new(doc = XML::SAX::Document.new, encoding = 'UTF-8')
10
+ parser = allocate
11
+ parser.document = doc
12
+ parser.encoding = encoding
13
+ parser.cstruct = LibXML::XmlSaxHandler.allocate
14
+ parser.send(:setup_lambdas)
15
+ parser.instance_variable_set(:@ctxt, nil)
16
+ parser
17
+ end
18
+
19
+ private
20
+
21
+ def setup_lambdas
22
+ @closures = {} # we need to keep references to the closures to avoid GC
23
+
24
+ [ :startDocument, :endDocument, :startElement, :endElement, :characters,
25
+ :comment, :warning, :error, :cdataBlock, :startElementNs, :endElementNs ].each do |sym|
26
+ @closures[sym] = lambda { |*args| send("__internal__#{sym}", *args) } # "i'm your private dancer", etc.
27
+ end
28
+
29
+ @closures.each { |k,v| cstruct[k] = v }
30
+
31
+ cstruct[:initialized] = Nokogiri::LibXML::XmlSaxHandler::XML_SAX2_MAGIC
32
+ end
33
+
34
+ def __internal__startDocument(_)
35
+ if @ctxt && @ctxt[:html] == 0 && @ctxt[:standalone] != -1
36
+ standalone = {
37
+ 0 => 'no',
38
+ 1 => 'yes',
39
+ }[@ctxt[:standalone]]
40
+
41
+ @document.xmldecl @ctxt[:version], @ctxt[:encoding], standalone
42
+ end
43
+ @document.start_document
44
+ end
45
+
46
+ def __internal__endDocument(_)
47
+ @document.end_document
48
+ end
49
+
50
+ def __internal__startElement(_, name, attributes)
51
+ attrs = []
52
+ unless attributes.null?
53
+ j = 0
54
+ while ! (value = attributes.get_pointer(LibXML.pointer_offset(j))).null?
55
+ attrs << value.read_string
56
+ j += 1
57
+ end
58
+ end
59
+ @document.start_element name, attrs
60
+ end
61
+
62
+ def __internal__endElement(_, name)
63
+ @document.end_element name
64
+ end
65
+
66
+ def __internal__characters(_, data, data_length)
67
+ @document.characters data.slice(0, data_length)
68
+ end
69
+
70
+ def __internal__comment(_, data)
71
+ @document.comment data
72
+ end
73
+
74
+ def __internal__warning(_, msg)
75
+ # TODO: vasprintf here
76
+ @document.warning(msg)
77
+ end
78
+
79
+ def __internal__error(_, msg)
80
+ # TODO: vasprintf here
81
+ @document.error(msg)
82
+ end
83
+
84
+ def __internal__cdataBlock(_, data, data_length)
85
+ @document.cdata_block data.slice(0, data_length)
86
+ end
87
+
88
+ def __internal__startElementNs(_, localname, prefix, uri, nb_namespaces, namespaces, nb_attributes, nb_defaulted, attributes)
89
+ localname = localname.null? ? nil : localname.read_string
90
+ prefix = prefix .null? ? nil : prefix .read_string
91
+ uri = uri .null? ? nil : uri .read_string
92
+
93
+ attr_list = []
94
+ ns_list = []
95
+
96
+ if ! attributes.null?
97
+ # Each attribute is an array of [localname, prefix, URI, value, end]
98
+ (0..(nb_attributes-1)*5).step(5) do |j|
99
+ key = attributes.get_pointer(LibXML.pointer_offset(j)).read_string
100
+ attr_prefix = attributes.get_pointer(LibXML.pointer_offset(j + 1))
101
+ attr_prefix = attr_prefix.null? ? nil : attr_prefix.read_string
102
+ attr_uri = attributes.get_pointer(LibXML.pointer_offset(j + 2))
103
+ attr_uri = attr_uri.null? ? nil : attr_uri.read_string
104
+ value_length = attributes.get_pointer(LibXML.pointer_offset(j+4)).address \
105
+ - attributes.get_pointer(LibXML.pointer_offset(j+3)).address
106
+ value = attributes.get_pointer(LibXML.pointer_offset(j+3)).get_string(0, value_length)
107
+ attr_list << Attribute.new(key, attr_prefix, attr_uri, value)
108
+ end
109
+ end
110
+
111
+ if ! namespaces.null?
112
+ (0..(nb_namespaces-1)*2).step(2) do |j|
113
+ key = namespaces.get_pointer(LibXML.pointer_offset(j))
114
+ key = key.null? ? nil : key.read_string
115
+ value = namespaces.get_pointer(LibXML.pointer_offset(j+1))
116
+ value = value.null? ? nil : value.read_string
117
+ ns_list << [key, value]
118
+ end
119
+ end
120
+
121
+ @document.start_element_namespace(
122
+ localname,
123
+ attr_list,
124
+ prefix,
125
+ uri,
126
+ ns_list
127
+ )
128
+ end
129
+
130
+ def __internal__endElementNs(_, localname, prefix, uri)
131
+ localname = localname.null? ? nil : localname.read_string
132
+ prefix = prefix .null? ? nil : prefix .read_string
133
+ uri = uri .null? ? nil : uri .read_string
134
+
135
+ @document.end_element_namespace(localname, prefix, uri)
136
+ end
137
+
138
+ # :startdoc:
139
+ end
140
+ end
141
+ end
142
+ end
@@ -0,0 +1,67 @@
1
+ # :stopdoc:
2
+ module Nokogiri
3
+ module XML
4
+ module SAX
5
+ class ParserContext
6
+ attr_accessor :cstruct
7
+ attr_accessor :reader_callback
8
+
9
+ def self.memory data
10
+ raise(ArgumentError, 'data cannot be nil') if data.nil?
11
+ raise('data cannot be empty') if data.length == 0
12
+ ctx = LibXML::XmlParserContext.new(
13
+ LibXML.xmlCreateMemoryParserCtxt data, data.length
14
+ )
15
+ pc = allocate
16
+ pc.cstruct = ctx
17
+ pc
18
+ end
19
+
20
+ def self.io io, encoding
21
+ reader_callback = IoCallbacks.reader(io) # keep a reference to prevent it from being GC'd
22
+ sax_ctx = LibXML.xmlCreateIOParserCtxt(
23
+ nil,
24
+ nil,
25
+ reader_callback,
26
+ nil,
27
+ nil,
28
+ encoding
29
+ )
30
+ pc = allocate
31
+ pc.cstruct = LibXML::XmlParserContext.new sax_ctx
32
+ pc.reader_callback = reader_callback
33
+ pc
34
+ end
35
+
36
+ def self.file filename
37
+ ctx = LibXML.xmlCreateFileParserCtxt filename
38
+ pc = allocate
39
+ pc.cstruct = LibXML::XmlParserContext.new ctx
40
+ pc
41
+ end
42
+
43
+ def parse_with sax_handler, type = :xml
44
+ raise ArgumentError unless XML::SAX::Parser === sax_handler
45
+ sax = sax_handler.cstruct
46
+ cstruct[:sax] = sax
47
+
48
+ sax_handler.instance_variable_set(:@ctxt, cstruct)
49
+
50
+ LibXML.send(:"#{type}ParseDocument", cstruct)
51
+
52
+ cstruct[:sax] = nil
53
+ LibXML.xmlFreeDoc cstruct[:myDoc] unless cstruct[:myDoc].null?
54
+ end
55
+
56
+ def replace_entities=(value)
57
+ self.cstruct[:replaceEntities] = value ? 1 : 0
58
+ end
59
+
60
+ def replace_entities
61
+ self.cstruct[:replaceEntities] == 0 ? false : true
62
+ end
63
+ end
64
+ end
65
+ end
66
+ end
67
+ # :startdoc:
@@ -0,0 +1,39 @@
1
+ module Nokogiri
2
+ module XML
3
+ module SAX
4
+ class PushParser
5
+
6
+ attr_accessor :cstruct # :nodoc:
7
+
8
+ private
9
+
10
+ def native_write(chunk, last_chunk) # :nodoc:
11
+ size = 0
12
+ unless chunk.nil?
13
+ chunk = chunk.to_s
14
+ size = chunk.length
15
+ end
16
+
17
+ rcode = LibXML.xmlParseChunk(cstruct, chunk, size, last_chunk ? 1 : 0)
18
+ if rcode != 0
19
+ error = LibXML.xmlCtxtGetLastError(cstruct)
20
+ raise Nokogiri::XML::SyntaxError.wrap(error)
21
+ end
22
+
23
+ self
24
+ end
25
+
26
+ def initialize_native(sax, filename) # :nodoc:
27
+ filename = filename.to_s unless filename.nil?
28
+ ctx_ptr = LibXML.xmlCreatePushParserCtxt(
29
+ sax.cstruct, nil, nil, 0, filename
30
+ )
31
+ raise(RuntimeError, "Could not create a parser context") if ctx_ptr.null?
32
+ self.cstruct = LibXML::XmlSaxPushParserContext.new(ctx_ptr) ;
33
+ self
34
+ end
35
+
36
+ end
37
+ end
38
+ end
39
+ end
@@ -0,0 +1,92 @@
1
+ module Nokogiri
2
+ module XML
3
+ class Schema
4
+ # :stopdoc:
5
+
6
+ attr_accessor :cstruct
7
+
8
+ def validate_document document
9
+ errors = []
10
+
11
+ ctx = LibXML.xmlSchemaNewValidCtxt(cstruct)
12
+ raise RuntimeError.new("Could not create a validation context") if ctx.null?
13
+
14
+ LibXML.xmlSchemaSetValidStructuredErrors(ctx,
15
+ SyntaxError.error_array_pusher(errors), nil) unless Nokogiri.is_2_6_16?
16
+
17
+ LibXML.xmlSchemaValidateDoc(ctx, document.cstruct)
18
+
19
+ LibXML.xmlSchemaFreeValidCtxt(ctx)
20
+
21
+ errors
22
+ end
23
+ private :validate_document
24
+
25
+ def self.read_memory content
26
+ content_copy = FFI::MemoryPointer.from_string(content)
27
+ ctx = LibXML.xmlSchemaNewMemParserCtxt(content_copy, content.length)
28
+
29
+ errors = []
30
+
31
+ LibXML.xmlSetStructuredErrorFunc(nil, SyntaxError.error_array_pusher(errors))
32
+ LibXML.xmlSchemaSetParserStructuredErrors(ctx, SyntaxError.error_array_pusher(errors), nil) unless Nokogiri.is_2_6_16?
33
+
34
+ schema_ptr = LibXML.xmlSchemaParse(ctx)
35
+
36
+ LibXML.xmlSetStructuredErrorFunc(nil, nil)
37
+ LibXML.xmlSchemaFreeParserCtxt(ctx)
38
+
39
+ if schema_ptr.null?
40
+ error = LibXML.xmlGetLastError
41
+ if error
42
+ raise SyntaxError.wrap(error)
43
+ else
44
+ raise RuntimeError, "Could not parse document"
45
+ end
46
+ end
47
+
48
+ schema = allocate
49
+ schema.cstruct = LibXML::XmlSchema.new schema_ptr
50
+ schema.errors = errors
51
+ schema
52
+ end
53
+
54
+ def self.from_document document
55
+ ctx = LibXML.xmlSchemaNewDocParserCtxt(document.document.cstruct)
56
+
57
+ errors = []
58
+
59
+ LibXML.xmlSetStructuredErrorFunc(nil, SyntaxError.error_array_pusher(errors))
60
+ unless Nokogiri.is_2_6_16?
61
+ LibXML.xmlSchemaSetParserStructuredErrors(
62
+ ctx,
63
+ SyntaxError.error_array_pusher(errors),
64
+ nil
65
+ )
66
+ end
67
+
68
+ schema_ptr = LibXML.xmlSchemaParse(ctx)
69
+
70
+ LibXML.xmlSetStructuredErrorFunc(nil, nil)
71
+ LibXML.xmlSchemaFreeParserCtxt(ctx)
72
+
73
+ if schema_ptr.null?
74
+ error = LibXML.xmlGetLastError
75
+ if error
76
+ raise SyntaxError.wrap(error)
77
+ else
78
+ raise RuntimeError, "Could not parse document"
79
+ end
80
+ end
81
+
82
+ schema = allocate
83
+ schema.cstruct = LibXML::XmlSchema.new schema_ptr
84
+ schema.errors = errors
85
+ schema
86
+ end
87
+
88
+ # :startdoc:
89
+ end
90
+ end
91
+ end
92
+
@@ -0,0 +1,91 @@
1
+ # :stopdoc:
2
+ module Nokogiri
3
+ module XML
4
+ class SyntaxError < ::Nokogiri::SyntaxError
5
+
6
+ attr_accessor :cstruct
7
+
8
+ def initialize(message)
9
+ self.cstruct = LibXML::XmlSyntaxError.new(LibXML::XmlSyntaxError.allocate())
10
+ self.cstruct[:message] = FFI::MemoryPointer.from_string(message)
11
+ end
12
+
13
+ def domain
14
+ cstruct[:domain]
15
+ end
16
+
17
+ def code
18
+ cstruct[:code]
19
+ end
20
+
21
+ def message
22
+ val = cstruct[:message]
23
+ val.null? ? nil : val.read_string
24
+ end
25
+ undef_method :inspect
26
+ alias_method :inspect, :message
27
+ undef_method :to_s
28
+ alias_method :to_s, :message
29
+
30
+ def message=(string)
31
+ unless cstruct[:message].null?
32
+ LibXML.xmlFree(cstruct[:message])
33
+ end
34
+ cstruct[:message] = string
35
+ end
36
+
37
+ def level
38
+ cstruct[:level]
39
+ end
40
+
41
+ def file
42
+ cstruct[:file].null? ? nil : cstruct[:file]
43
+ end
44
+
45
+ def line
46
+ cstruct[:line]
47
+ end
48
+
49
+ def str1
50
+ cstruct[:str1].null? ? nil : cstruct[:str1]
51
+ end
52
+
53
+ def str2
54
+ cstruct[:str].null? ? nil : cstruct[:str]
55
+ end
56
+
57
+ def str3
58
+ cstruct[:str3].null? ? nil : cstruct[:str3]
59
+ end
60
+
61
+ def int1
62
+ cstruct[:int1]
63
+ end
64
+
65
+ def column
66
+ cstruct[:int2]
67
+ end
68
+ alias_method :int2, :column
69
+
70
+ class << self
71
+ def error_array_pusher(array)
72
+ Proc.new do |_ignored_, error|
73
+ array << wrap(error) if array
74
+ end
75
+ end
76
+
77
+ def wrap(error_ptr)
78
+ error_struct = LibXML::XmlSyntaxError.allocate
79
+ LibXML.xmlCopyError(error_ptr, error_struct)
80
+ error_cstruct = LibXML::XmlSyntaxError.new(error_struct)
81
+ error = self.allocate # will generate XML::XPath::SyntaxError or XML::SyntaxError
82
+ error.cstruct = error_cstruct
83
+ error
84
+ end
85
+ end
86
+
87
+ end
88
+ end
89
+
90
+ end
91
+ # :startdoc:
@@ -0,0 +1,18 @@
1
+ module Nokogiri
2
+ module XML
3
+ class Text < CharacterData
4
+
5
+ def self.new(string, document, *rest) # :nodoc:
6
+ node_ptr = LibXML.xmlNewText(string)
7
+ node_cstruct = LibXML::XmlNode.new(node_ptr)
8
+ node_cstruct[:doc] = document.cstruct[:doc]
9
+
10
+ node = Node.wrap(node_cstruct, self)
11
+ node.send :initialize, string, document, *rest
12
+ yield node if block_given?
13
+ node
14
+ end
15
+
16
+ end
17
+ end
18
+ end