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 +1,7 @@
1
+ require 'nokogiri/syntax_error'
2
+ module Nokogiri
3
+ module CSS
4
+ class SyntaxError < ::Nokogiri::SyntaxError
5
+ end
6
+ end
7
+ end
@@ -0,0 +1,7 @@
1
+ module Nokogiri
2
+ module CSS
3
+ class Tokenizer < GeneratedTokenizer
4
+ alias :scan :scan_setup
5
+ end
6
+ end
7
+ end
@@ -0,0 +1,54 @@
1
+ module Nokogiri
2
+ module CSS
3
+ class GeneratedTokenizer < GeneratedParser
4
+
5
+ macro
6
+ nl \n|\r\n|\r|\f
7
+ w [\s\r\n\f]*
8
+ nonascii [^\0-\177]
9
+ num -?([0-9]+|[0-9]*\.[0-9]+)
10
+ unicode \\[0-9A-Fa-f]{1,6}(\r\n|[\s\n\r\t\f])?
11
+
12
+ escape {unicode}|\\[^\n\r\f0-9A-Fa-f]
13
+ nmchar [_A-Za-z0-9-]|{nonascii}|{escape}
14
+ nmstart [_A-Za-z]|{nonascii}|{escape}
15
+ ident [-@]?({nmstart})({nmchar})*
16
+ name ({nmchar})+
17
+ string1 "([^\n\r\f"]|{nl}|{nonascii}|{escape})*"
18
+ string2 '([^\n\r\f']|{nl}|{nonascii}|{escape})*'
19
+ string {string1}|{string2}
20
+
21
+ rule
22
+
23
+ # [:state] pattern [actions]
24
+
25
+ {ident}\(\s* { [:FUNCTION, text] }
26
+ {ident} { [:IDENT, text] }
27
+ \#{name} { [:HASH, text] }
28
+ {w}~={w} { [:INCLUDES, text] }
29
+ {w}\|={w} { [:DASHMATCH, text] }
30
+ {w}\^={w} { [:PREFIXMATCH, text] }
31
+ {w}\$={w} { [:SUFFIXMATCH, text] }
32
+ {w}\*={w} { [:SUBSTRINGMATCH, text] }
33
+ {w}!={w} { [:NOT_EQUAL, text] }
34
+ {w}={w} { [:EQUAL, text] }
35
+ {w}\) { [:RPAREN, text] }
36
+ {w}\[{w} { [:LSQUARE, text] }
37
+ {w}\] { [:RSQUARE, text] }
38
+ {w}\+{w} { [:PLUS, text] }
39
+ {w}>{w} { [:GREATER, text] }
40
+ {w},{w} { [:COMMA, text] }
41
+ {w}~{w} { [:TILDE, text] }
42
+ \:not\({w} { [:NOT, text] }
43
+ {num} { [:NUMBER, text] }
44
+ {w}\/\/{w} { [:DOUBLESLASH, text] }
45
+ {w}\/{w} { [:SLASH, text] }
46
+
47
+ U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})? {[:UNICODE_RANGE, text] }
48
+
49
+ [\s\t\r\n\f]+ { [:S, text] }
50
+ {string} { [:STRING, text] }
51
+ . { [text, text] }
52
+ end
53
+ end
54
+ end
@@ -0,0 +1,162 @@
1
+ module Nokogiri
2
+ module CSS
3
+ class XPathVisitor # :nodoc:
4
+ def visit_function node
5
+ # note that nth-child and nth-last-child are preprocessed in css/node.rb.
6
+ msg = :"visit_function_#{node.value.first.gsub(/[(]/, '')}"
7
+ return self.send(msg, node) if self.respond_to?(msg)
8
+
9
+ case node.value.first
10
+ when /^text\(/
11
+ 'child::text()'
12
+ when /^self\(/
13
+ "self::#{node.value[1]}"
14
+ when /^(eq|nth|nth-of-type|nth-child)\(/
15
+ if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :AN_PLUS_B
16
+ an_plus_b(node.value[1])
17
+ else
18
+ "position() = " + node.value[1]
19
+ end
20
+ when /^(first|first-of-type)\(/
21
+ "position() = 1"
22
+ when /^(last|last-of-type)\(/
23
+ "position() = last()"
24
+ when /^(nth-last-child|nth-last-of-type)\(/
25
+ "position() = last() - #{node.value[1]}"
26
+ when /^contains\(/
27
+ "contains(., #{node.value[1]})"
28
+ when /^gt\(/
29
+ "position() > #{node.value[1]}"
30
+ when /^only-child\(/
31
+ "last() = 1"
32
+ when /^comment\(/
33
+ "comment()"
34
+ else
35
+ args = ['.'] + node.value[1..-1]
36
+ "#{node.value.first}#{args.join(', ')})"
37
+ end
38
+ end
39
+
40
+ def visit_not node
41
+ 'not(' + node.value.first.accept(self) + ')'
42
+ end
43
+
44
+ def visit_preceding_selector node
45
+ node.value.last.accept(self) +
46
+ '[preceding-sibling::' +
47
+ node.value.first.accept(self) +
48
+ ']'
49
+ end
50
+
51
+ def visit_id node
52
+ node.value.first =~ /^#(.*)$/
53
+ "@id = '#{$1}'"
54
+ end
55
+
56
+ def visit_attribute_condition node
57
+ attribute = if (node.value.first.type == :FUNCTION) or (node.value.first.value.first =~ /::/)
58
+ ''
59
+ else
60
+ '@'
61
+ end
62
+ attribute += node.value.first.accept(self)
63
+
64
+ # Support non-standard css
65
+ attribute.gsub!(/^@@/, '@')
66
+
67
+ return attribute unless node.value.length == 3
68
+
69
+ value = node.value.last
70
+ value = "'#{value}'" if value !~ /^['"]/
71
+
72
+ case node.value[1]
73
+ when :equal
74
+ attribute + " = " + "#{value}"
75
+ when :not_equal
76
+ attribute + " != " + "#{value}"
77
+ when :substring_match
78
+ "contains(#{attribute}, #{value})"
79
+ when :prefix_match
80
+ "starts-with(#{attribute}, #{value})"
81
+ when :dash_match
82
+ "#{attribute} = #{value} or starts-with(#{attribute}, concat(#{value}, '-'))"
83
+ when :includes
84
+ "contains(concat(\" \", #{attribute}, \" \"),concat(\" \", #{value}, \" \"))"
85
+ when :suffix_match
86
+ "substring(#{attribute}, string-length(#{attribute}) - " +
87
+ "string-length(#{value}) + 1, string-length(#{value})) = #{value}"
88
+ else
89
+ attribute + " #{node.value[1]} " + "#{value}"
90
+ end
91
+ end
92
+
93
+ def visit_pseudo_class node
94
+ if node.value.first.is_a?(Nokogiri::CSS::Node) and node.value.first.type == :FUNCTION
95
+ node.value.first.accept(self)
96
+ else
97
+ msg = :"visit_pseudo_class_#{node.value.first.gsub(/[(]/, '')}"
98
+ return self.send(msg, node) if self.respond_to?(msg)
99
+
100
+ case node.value.first
101
+ when "first" then "position() = 1"
102
+ when "last" then "position() = last()"
103
+ when "first-of-type" then "position() = 1"
104
+ when "last-of-type" then "position() = last()"
105
+ when "only-of-type" then "last() = 1"
106
+ when "empty" then "not(node())"
107
+ when "parent" then "node()"
108
+ when "root" then "not(parent::*)"
109
+ else
110
+ node.value.first + "(.)"
111
+ end
112
+ end
113
+ end
114
+
115
+ def visit_class_condition node
116
+ "contains(concat(' ', @class, ' '), ' #{node.value.first} ')"
117
+ end
118
+
119
+ {
120
+ 'combinator' => ' and ',
121
+ 'direct_adjacent_selector' => "/following-sibling::*[1]/self::",
122
+ 'descendant_selector' => '//',
123
+ 'child_selector' => '/',
124
+ }.each do |k,v|
125
+ class_eval %{
126
+ def visit_#{k} node
127
+ "\#{node.value.first.accept(self)}#{v}\#{node.value.last.accept(self)}"
128
+ end
129
+ }
130
+ end
131
+
132
+ def visit_conditional_selector node
133
+ node.value.first.accept(self) + '[' +
134
+ node.value.last.accept(self) + ']'
135
+ end
136
+
137
+ def visit_element_name node
138
+ node.value.first
139
+ end
140
+
141
+ def accept node
142
+ node.accept(self)
143
+ end
144
+
145
+ private
146
+ def an_plus_b node
147
+ raise ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}" unless node.value.size == 4
148
+
149
+ a = node.value[0].to_i
150
+ b = node.value[3].to_i
151
+
152
+ if (b == 0)
153
+ return "(position() mod #{a}) = 0"
154
+ else
155
+ compare = (a < 0) ? "<=" : ">="
156
+ return "(position() #{compare} #{b}) and (((position()-#{b}) mod #{a.abs}) = 0)"
157
+ end
158
+ end
159
+
160
+ end
161
+ end
162
+ end
@@ -0,0 +1,33 @@
1
+ module Nokogiri
2
+ module Decorators
3
+ ###
4
+ # The Slop decorator implements method missing such that a methods may be
5
+ # used instead of XPath or CSS. See Nokogiri.Slop
6
+ module Slop
7
+ ###
8
+ # look for node with +name+. See Nokogiri.Slop
9
+ def method_missing name, *args, &block
10
+ if args.empty?
11
+ list = xpath("./#{name}")
12
+ elsif args.first.is_a? Hash
13
+ hash = args.first
14
+ if hash[:css]
15
+ list = css("#{name}#{hash[:css]}")
16
+ elsif hash[:xpath]
17
+ conds = Array(hash[:xpath]).join(' and ')
18
+ list = xpath("./#{name}[#{conds}]")
19
+ end
20
+ else
21
+ CSS::Parser.without_cache do
22
+ list = xpath(
23
+ *CSS.xpath_for("#{name}#{args.first}", :prefix => "./")
24
+ )
25
+ end
26
+ end
27
+
28
+ super if list.empty?
29
+ list.length == 1 ? list.first : list
30
+ end
31
+ end
32
+ end
33
+ end
@@ -0,0 +1,28 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class Document < XML::Document
4
+
5
+ attr_accessor :cstruct # :nodoc:
6
+
7
+ def self.new(*args) # :nodoc:
8
+ uri = args[0]
9
+ external_id = args[1]
10
+ doc = wrap(LibXML.htmlNewDoc(uri, external_id))
11
+ doc.send :initialize, *args
12
+ doc
13
+ end
14
+
15
+ def self.read_io(io, url, encoding, options) # :nodoc:
16
+ wrap_with_error_handling do
17
+ LibXML.htmlReadIO(IoCallbacks.reader(io), nil, nil, url, encoding, options)
18
+ end
19
+ end
20
+
21
+ def self.read_memory(string, url, encoding, options) # :nodoc:
22
+ wrap_with_error_handling do
23
+ LibXML.htmlReadMemory(string, string.length, url, encoding, options)
24
+ end
25
+ end
26
+ end
27
+ end
28
+ end
@@ -0,0 +1,85 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class ElementDescription
4
+
5
+ attr_accessor :cstruct # :nodoc:
6
+
7
+ def required_attributes # :nodoc:
8
+ get_string_array_from :attrs_req
9
+ end
10
+
11
+ def deprecated_attributes # :nodoc:
12
+ get_string_array_from :attrs_depr
13
+ end
14
+
15
+ def optional_attributes # :nodoc:
16
+ get_string_array_from :attrs_opt
17
+ end
18
+
19
+ def default_sub_element # :nodoc:
20
+ cstruct[:defaultsubelt]
21
+ end
22
+
23
+ def sub_elements # :nodoc:
24
+ get_string_array_from :subelts
25
+ end
26
+
27
+ def description # :nodoc:
28
+ cstruct[:desc]
29
+ end
30
+
31
+ def inline? # :nodoc:
32
+ cstruct[:isinline] != 0
33
+ end
34
+
35
+ def deprecated? # :nodoc:
36
+ cstruct[:depr] != 0
37
+ end
38
+
39
+ def empty? # :nodoc:
40
+ cstruct[:empty] != 0
41
+ end
42
+
43
+ def save_end_tag? # :nodoc:
44
+ cstruct[:saveEndTag] != 0
45
+ end
46
+
47
+ def implied_end_tag? # :nodoc:
48
+ cstruct[:endTag] != 0
49
+ end
50
+
51
+ def implied_start_tag? # :nodoc:
52
+ cstruct[:startTag] != 0
53
+ end
54
+
55
+ def name # :nodoc:
56
+ cstruct[:name]
57
+ end
58
+
59
+ def self.[](tag_name) # :nodoc:
60
+ ptr = LibXML.htmlTagLookup(tag_name)
61
+ return nil if ptr.null?
62
+
63
+ desc = allocate
64
+ desc.cstruct = LibXML::HtmlElemDesc.new(ptr)
65
+ desc
66
+ end
67
+
68
+ private
69
+
70
+ def get_string_array_from(sym) # :nodoc:
71
+ list = []
72
+ return list if cstruct[sym].null?
73
+
74
+ j = 0
75
+ while (ptr = cstruct[sym].get_pointer(j * FFI.type_size(:pointer))) && ! ptr.null?
76
+ list << ptr.read_string
77
+ j += 1
78
+ end
79
+
80
+ list
81
+ end
82
+
83
+ end
84
+ end
85
+ end
@@ -0,0 +1,16 @@
1
+ module Nokogiri
2
+ module HTML
3
+ class EntityLookup
4
+
5
+ def get(key) # :nodoc:
6
+ ptr = LibXML.htmlEntityLookup(key.to_s)
7
+ return nil if ptr.null?
8
+
9
+ cstruct = LibXML::HtmlEntityDesc.new(ptr)
10
+ EntityDescription.new cstruct[:value], cstruct[:name], cstruct[:desc]
11
+ end
12
+
13
+ end
14
+ end
15
+ end
16
+
@@ -0,0 +1,38 @@
1
+ # :stopdoc:
2
+ module Nokogiri
3
+ module HTML
4
+ module SAX
5
+ class ParserContext < Nokogiri::XML::SAX::ParserContext
6
+ attr_accessor :cstruct
7
+
8
+ def self.file filename, encoding
9
+ ctx = LibXML.htmlCreateFileParserCtxt filename, encoding
10
+ pc = allocate
11
+ pc.cstruct = LibXML::XmlParserContext.new ctx
12
+ pc
13
+ end
14
+
15
+ def self.memory data, encoding
16
+ raise ArgumentError unless data
17
+ raise "data cannot be empty" unless data.length > 0
18
+
19
+ ctx = LibXML.htmlCreateMemoryParserCtxt data, data.length
20
+ pc = allocate
21
+ pc.cstruct = LibXML::XmlParserContext.new ctx
22
+ if encoding
23
+ enc = LibXML.xmlParseCharEncoding(encoding)
24
+ if enc != LibXML::XML_CHAR_ENCODING_ERROR
25
+ LibXML.xmlSwitchEncoding(ctx, enc)
26
+ end
27
+ end
28
+ pc
29
+ end
30
+
31
+ def parse_with sax_handler, type = :html
32
+ super
33
+ end
34
+ end
35
+ end
36
+ end
37
+ end
38
+ # :startdoc:
@@ -0,0 +1,42 @@
1
+ # :stopdoc:
2
+ module Nokogiri
3
+ module IoCallbacks
4
+
5
+ class << self
6
+
7
+ def plain_old_reader(io)
8
+ lambda do |ctx, buffer, len|
9
+ string = io.read(len)
10
+ return 0 if string.nil?
11
+ buffer.put_bytes(0, string, 0, string.length)
12
+ string.length
13
+ end
14
+ end
15
+
16
+ if defined?(FFI::IO.native_read)
17
+ def ffi_io_native_reader(io)
18
+ if io.is_a?(StringIO)
19
+ plain_old_reader(io)
20
+ else
21
+ lambda do |ctx, buffer, len|
22
+ rcode = FFI::IO.native_read(io, buffer, len)
23
+ (rcode < 0) ? 0 : rcode
24
+ end
25
+ end
26
+ end
27
+ alias :reader :ffi_io_native_reader
28
+ else
29
+ alias :reader :plain_old_reader
30
+ end
31
+
32
+ def writer(io)
33
+ lambda do |context, buffer, len|
34
+ io.write buffer
35
+ len
36
+ end
37
+ end
38
+ end
39
+
40
+ end
41
+ end
42
+ # :startdoc: