nokogiri-backupify 1.5.0.beta.4

Sign up to get free protection for your applications and to get access to all the features.
Files changed (275) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +509 -0
  3. data/CHANGELOG.rdoc +490 -0
  4. data/Manifest.txt +274 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +150 -0
  7. data/Rakefile +217 -0
  8. data/bin/nokogiri +54 -0
  9. data/deps.rip +5 -0
  10. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  11. data/ext/java/nokogiri/HtmlDocument.java +146 -0
  12. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  13. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  14. data/ext/java/nokogiri/HtmlSaxParserContext.java +256 -0
  15. data/ext/java/nokogiri/NokogiriService.java +466 -0
  16. data/ext/java/nokogiri/XmlAttr.java +183 -0
  17. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  18. data/ext/java/nokogiri/XmlCdata.java +89 -0
  19. data/ext/java/nokogiri/XmlComment.java +84 -0
  20. data/ext/java/nokogiri/XmlDocument.java +514 -0
  21. data/ext/java/nokogiri/XmlDocumentFragment.java +216 -0
  22. data/ext/java/nokogiri/XmlDtd.java +464 -0
  23. data/ext/java/nokogiri/XmlElement.java +221 -0
  24. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  25. data/ext/java/nokogiri/XmlElementDecl.java +147 -0
  26. data/ext/java/nokogiri/XmlEntityDecl.java +161 -0
  27. data/ext/java/nokogiri/XmlEntityReference.java +75 -0
  28. data/ext/java/nokogiri/XmlNamespace.java +127 -0
  29. data/ext/java/nokogiri/XmlNode.java +1392 -0
  30. data/ext/java/nokogiri/XmlNodeSet.java +284 -0
  31. data/ext/java/nokogiri/XmlProcessingInstruction.java +103 -0
  32. data/ext/java/nokogiri/XmlReader.java +409 -0
  33. data/ext/java/nokogiri/XmlRelaxng.java +199 -0
  34. data/ext/java/nokogiri/XmlSaxParserContext.java +353 -0
  35. data/ext/java/nokogiri/XmlSaxPushParser.java +182 -0
  36. data/ext/java/nokogiri/XmlSchema.java +175 -0
  37. data/ext/java/nokogiri/XmlSyntaxError.java +114 -0
  38. data/ext/java/nokogiri/XmlText.java +135 -0
  39. data/ext/java/nokogiri/XmlXpathContext.java +175 -0
  40. data/ext/java/nokogiri/XsltStylesheet.java +181 -0
  41. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +205 -0
  42. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  43. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +80 -0
  44. data/ext/java/nokogiri/internals/NokogiriHandler.java +326 -0
  45. data/ext/java/nokogiri/internals/NokogiriHelpers.java +583 -0
  46. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +170 -0
  47. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +118 -0
  48. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +73 -0
  49. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  50. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
  51. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +120 -0
  52. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +56 -0
  53. data/ext/java/nokogiri/internals/ParserContext.java +278 -0
  54. data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
  55. data/ext/java/nokogiri/internals/ReaderNode.java +473 -0
  56. data/ext/java/nokogiri/internals/SaveContext.java +282 -0
  57. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +68 -0
  58. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  59. data/ext/java/nokogiri/internals/XmlDomParser.java +77 -0
  60. data/ext/java/nokogiri/internals/XmlDomParserContext.java +233 -0
  61. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  62. data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
  63. data/ext/nokogiri/depend +358 -0
  64. data/ext/nokogiri/extconf.rb +124 -0
  65. data/ext/nokogiri/html_document.c +154 -0
  66. data/ext/nokogiri/html_document.h +10 -0
  67. data/ext/nokogiri/html_element_description.c +276 -0
  68. data/ext/nokogiri/html_element_description.h +10 -0
  69. data/ext/nokogiri/html_entity_lookup.c +32 -0
  70. data/ext/nokogiri/html_entity_lookup.h +8 -0
  71. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  72. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  73. data/ext/nokogiri/nokogiri.c +92 -0
  74. data/ext/nokogiri/nokogiri.h +160 -0
  75. data/ext/nokogiri/xml_attr.c +94 -0
  76. data/ext/nokogiri/xml_attr.h +9 -0
  77. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  78. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  79. data/ext/nokogiri/xml_cdata.c +56 -0
  80. data/ext/nokogiri/xml_cdata.h +9 -0
  81. data/ext/nokogiri/xml_comment.c +54 -0
  82. data/ext/nokogiri/xml_comment.h +9 -0
  83. data/ext/nokogiri/xml_document.c +478 -0
  84. data/ext/nokogiri/xml_document.h +23 -0
  85. data/ext/nokogiri/xml_document_fragment.c +48 -0
  86. data/ext/nokogiri/xml_document_fragment.h +10 -0
  87. data/ext/nokogiri/xml_dtd.c +202 -0
  88. data/ext/nokogiri/xml_dtd.h +10 -0
  89. data/ext/nokogiri/xml_element_content.c +123 -0
  90. data/ext/nokogiri/xml_element_content.h +10 -0
  91. data/ext/nokogiri/xml_element_decl.c +69 -0
  92. data/ext/nokogiri/xml_element_decl.h +9 -0
  93. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  94. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  95. data/ext/nokogiri/xml_entity_decl.c +110 -0
  96. data/ext/nokogiri/xml_entity_decl.h +10 -0
  97. data/ext/nokogiri/xml_entity_reference.c +52 -0
  98. data/ext/nokogiri/xml_entity_reference.h +9 -0
  99. data/ext/nokogiri/xml_io.c +31 -0
  100. data/ext/nokogiri/xml_io.h +11 -0
  101. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  102. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  103. data/ext/nokogiri/xml_namespace.c +84 -0
  104. data/ext/nokogiri/xml_namespace.h +13 -0
  105. data/ext/nokogiri/xml_node.c +1384 -0
  106. data/ext/nokogiri/xml_node.h +13 -0
  107. data/ext/nokogiri/xml_node_set.c +418 -0
  108. data/ext/nokogiri/xml_node_set.h +9 -0
  109. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  110. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  111. data/ext/nokogiri/xml_reader.c +684 -0
  112. data/ext/nokogiri/xml_reader.h +10 -0
  113. data/ext/nokogiri/xml_relax_ng.c +161 -0
  114. data/ext/nokogiri/xml_relax_ng.h +9 -0
  115. data/ext/nokogiri/xml_sax_parser.c +288 -0
  116. data/ext/nokogiri/xml_sax_parser.h +39 -0
  117. data/ext/nokogiri/xml_sax_parser_context.c +199 -0
  118. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  119. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  120. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  121. data/ext/nokogiri/xml_schema.c +205 -0
  122. data/ext/nokogiri/xml_schema.h +9 -0
  123. data/ext/nokogiri/xml_syntax_error.c +58 -0
  124. data/ext/nokogiri/xml_syntax_error.h +13 -0
  125. data/ext/nokogiri/xml_text.c +50 -0
  126. data/ext/nokogiri/xml_text.h +9 -0
  127. data/ext/nokogiri/xml_xpath_context.c +309 -0
  128. data/ext/nokogiri/xml_xpath_context.h +9 -0
  129. data/ext/nokogiri/xslt_stylesheet.c +258 -0
  130. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  131. data/lib/isorelax.jar +0 -0
  132. data/lib/jing.jar +0 -0
  133. data/lib/nekodtd.jar +0 -0
  134. data/lib/nekohtml.jar +0 -0
  135. data/lib/nokogiri.rb +143 -0
  136. data/lib/nokogiri/css.rb +23 -0
  137. data/lib/nokogiri/css/node.rb +99 -0
  138. data/lib/nokogiri/css/parser.rb +677 -0
  139. data/lib/nokogiri/css/parser.y +237 -0
  140. data/lib/nokogiri/css/parser_extras.rb +91 -0
  141. data/lib/nokogiri/css/syntax_error.rb +7 -0
  142. data/lib/nokogiri/css/tokenizer.rb +152 -0
  143. data/lib/nokogiri/css/tokenizer.rex +55 -0
  144. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  145. data/lib/nokogiri/decorators/slop.rb +35 -0
  146. data/lib/nokogiri/html.rb +36 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +221 -0
  149. data/lib/nokogiri/html/document_fragment.rb +41 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  152. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  153. data/lib/nokogiri/html/sax/parser.rb +52 -0
  154. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  155. data/lib/nokogiri/syntax_error.rb +4 -0
  156. data/lib/nokogiri/version.rb +35 -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 +418 -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 +218 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +84 -0
  165. data/lib/nokogiri/xml/dtd.rb +22 -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 +19 -0
  169. data/lib/nokogiri/xml/namespace.rb +13 -0
  170. data/lib/nokogiri/xml/node.rb +907 -0
  171. data/lib/nokogiri/xml/node/save_options.rb +45 -0
  172. data/lib/nokogiri/xml/node_set.rb +350 -0
  173. data/lib/nokogiri/xml/notation.rb +6 -0
  174. data/lib/nokogiri/xml/parse_options.rb +85 -0
  175. data/lib/nokogiri/xml/pp.rb +2 -0
  176. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  177. data/lib/nokogiri/xml/pp/node.rb +56 -0
  178. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  179. data/lib/nokogiri/xml/reader.rb +112 -0
  180. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  181. data/lib/nokogiri/xml/sax.rb +4 -0
  182. data/lib/nokogiri/xml/sax/document.rb +164 -0
  183. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  184. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  185. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  186. data/lib/nokogiri/xml/schema.rb +57 -0
  187. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  188. data/lib/nokogiri/xml/text.rb +9 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +52 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xercesImpl.jar +0 -0
  195. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  196. data/tasks/cross_compile.rb +177 -0
  197. data/tasks/test.rb +94 -0
  198. data/test/css/test_nthiness.rb +159 -0
  199. data/test/css/test_parser.rb +303 -0
  200. data/test/css/test_tokenizer.rb +198 -0
  201. data/test/css/test_xpath_visitor.rb +85 -0
  202. data/test/decorators/test_slop.rb +16 -0
  203. data/test/files/2ch.html +108 -0
  204. data/test/files/address_book.rlx +12 -0
  205. data/test/files/address_book.xml +10 -0
  206. data/test/files/bar/bar.xsd +4 -0
  207. data/test/files/dont_hurt_em_why.xml +422 -0
  208. data/test/files/exslt.xml +8 -0
  209. data/test/files/exslt.xslt +35 -0
  210. data/test/files/foo/foo.xsd +4 -0
  211. data/test/files/po.xml +32 -0
  212. data/test/files/po.xsd +66 -0
  213. data/test/files/shift_jis.html +10 -0
  214. data/test/files/shift_jis.xml +5 -0
  215. data/test/files/snuggles.xml +3 -0
  216. data/test/files/staff.dtd +10 -0
  217. data/test/files/staff.xml +59 -0
  218. data/test/files/staff.xslt +32 -0
  219. data/test/files/tlm.html +850 -0
  220. data/test/files/valid_bar.xml +2 -0
  221. data/test/helper.rb +171 -0
  222. data/test/html/sax/test_parser.rb +136 -0
  223. data/test/html/sax/test_parser_context.rb +48 -0
  224. data/test/html/test_builder.rb +164 -0
  225. data/test/html/test_document.rb +457 -0
  226. data/test/html/test_document_encoding.rb +123 -0
  227. data/test/html/test_document_fragment.rb +255 -0
  228. data/test/html/test_element_description.rb +100 -0
  229. data/test/html/test_named_characters.rb +14 -0
  230. data/test/html/test_node.rb +190 -0
  231. data/test/html/test_node_encoding.rb +27 -0
  232. data/test/test_convert_xpath.rb +135 -0
  233. data/test/test_css_cache.rb +45 -0
  234. data/test/test_encoding_handler.rb +46 -0
  235. data/test/test_memory_leak.rb +52 -0
  236. data/test/test_nokogiri.rb +132 -0
  237. data/test/test_reader.rb +403 -0
  238. data/test/test_soap4r_sax.rb +52 -0
  239. data/test/test_xslt_transforms.rb +189 -0
  240. data/test/xml/node/test_save_options.rb +20 -0
  241. data/test/xml/node/test_subclass.rb +44 -0
  242. data/test/xml/sax/test_parser.rb +338 -0
  243. data/test/xml/sax/test_parser_context.rb +113 -0
  244. data/test/xml/sax/test_push_parser.rb +156 -0
  245. data/test/xml/test_attr.rb +65 -0
  246. data/test/xml/test_attribute_decl.rb +86 -0
  247. data/test/xml/test_builder.rb +210 -0
  248. data/test/xml/test_cdata.rb +50 -0
  249. data/test/xml/test_comment.rb +29 -0
  250. data/test/xml/test_document.rb +675 -0
  251. data/test/xml/test_document_encoding.rb +26 -0
  252. data/test/xml/test_document_fragment.rb +192 -0
  253. data/test/xml/test_dtd.rb +107 -0
  254. data/test/xml/test_dtd_encoding.rb +33 -0
  255. data/test/xml/test_element_content.rb +56 -0
  256. data/test/xml/test_element_decl.rb +73 -0
  257. data/test/xml/test_entity_decl.rb +122 -0
  258. data/test/xml/test_entity_reference.rb +21 -0
  259. data/test/xml/test_namespace.rb +70 -0
  260. data/test/xml/test_node.rb +899 -0
  261. data/test/xml/test_node_attributes.rb +34 -0
  262. data/test/xml/test_node_encoding.rb +107 -0
  263. data/test/xml/test_node_reparenting.rb +321 -0
  264. data/test/xml/test_node_set.rb +708 -0
  265. data/test/xml/test_parse_options.rb +52 -0
  266. data/test/xml/test_processing_instruction.rb +30 -0
  267. data/test/xml/test_reader_encoding.rb +126 -0
  268. data/test/xml/test_relax_ng.rb +60 -0
  269. data/test/xml/test_schema.rb +89 -0
  270. data/test/xml/test_syntax_error.rb +12 -0
  271. data/test/xml/test_text.rb +47 -0
  272. data/test/xml/test_unparented_node.rb +381 -0
  273. data/test/xml/test_xpath.rb +237 -0
  274. data/test/xslt/test_custom_functions.rb +94 -0
  275. metadata +525 -0
@@ -0,0 +1,55 @@
1
+ module Nokogiri
2
+ module CSS
3
+ class Tokenizer
4
+
5
+ macro
6
+ nl \n|\r\n|\r|\f
7
+ w [\s]*
8
+ nonascii [^\0-\177]
9
+ num -?([0-9]+|[0-9]*\.[0-9]+)
10
+ unicode \\[0-9A-Fa-f]{1,6}(\r\n|[\s])?
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
+ has\({w} { [:HAS, text] }
26
+ {ident}\({w} { [:FUNCTION, text] }
27
+ {ident} { [:IDENT, text] }
28
+ \#{name} { [:HASH, text] }
29
+ {w}~={w} { [:INCLUDES, text] }
30
+ {w}\|={w} { [:DASHMATCH, text] }
31
+ {w}\^={w} { [:PREFIXMATCH, text] }
32
+ {w}\$={w} { [:SUFFIXMATCH, text] }
33
+ {w}\*={w} { [:SUBSTRINGMATCH, text] }
34
+ {w}!={w} { [:NOT_EQUAL, text] }
35
+ {w}={w} { [:EQUAL, text] }
36
+ {w}\) { [:RPAREN, text] }
37
+ {w}\[{w} { [:LSQUARE, text] }
38
+ {w}\] { [:RSQUARE, text] }
39
+ {w}\+{w} { [:PLUS, text] }
40
+ {w}>{w} { [:GREATER, text] }
41
+ {w},{w} { [:COMMA, text] }
42
+ {w}~{w} { [:TILDE, text] }
43
+ \:not\({w} { [:NOT, text] }
44
+ {num} { [:NUMBER, text] }
45
+ {w}\/\/{w} { [:DOUBLESLASH, text] }
46
+ {w}\/{w} { [:SLASH, text] }
47
+
48
+ U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})? {[:UNICODE_RANGE, text] }
49
+
50
+ [\s]+ { [:S, text] }
51
+ {string} { [:STRING, text] }
52
+ . { [text, text] }
53
+ end
54
+ end
55
+ end
@@ -0,0 +1,171 @@
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\(/
15
+ "position() = #{node.value[1]}"
16
+ when /^(nth|nth-of-type|nth-child)\(/
17
+ if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :AN_PLUS_B
18
+ an_plus_b(node.value[1])
19
+ else
20
+ "position() = #{node.value[1]}"
21
+ end
22
+ when /^(nth-last-child|nth-last-of-type)\(/
23
+ if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :AN_PLUS_B
24
+ an_plus_b(node.value[1], :last => true)
25
+ else
26
+ index = node.value[1].to_i - 1
27
+ index == 0 ? "position() = last()" : "position() = last() - #{index}"
28
+ end
29
+ when /^(first|first-of-type)\(/
30
+ "position() = 1"
31
+ when /^(last|last-of-type)\(/
32
+ "position() = last()"
33
+ when /^contains\(/
34
+ "contains(., #{node.value[1]})"
35
+ when /^gt\(/
36
+ "position() > #{node.value[1]}"
37
+ when /^only-child\(/
38
+ "last() = 1"
39
+ when /^comment\(/
40
+ "comment()"
41
+ when /^has\(/
42
+ node.value[1].accept(self)
43
+ else
44
+ args = ['.'] + node.value[1..-1]
45
+ "#{node.value.first}#{args.join(', ')})"
46
+ end
47
+ end
48
+
49
+ def visit_not node
50
+ child = node.value.first
51
+ if :ELEMENT_NAME == child.type
52
+ "not(self::#{child.accept(self)})"
53
+ else
54
+ "not(#{child.accept(self)})"
55
+ end
56
+ end
57
+
58
+ def visit_id node
59
+ node.value.first =~ /^#(.*)$/
60
+ "@id = '#{$1}'"
61
+ end
62
+
63
+ def visit_attribute_condition node
64
+ attribute = if (node.value.first.type == :FUNCTION) or (node.value.first.value.first =~ /::/)
65
+ ''
66
+ else
67
+ '@'
68
+ end
69
+ attribute += node.value.first.accept(self)
70
+
71
+ # Support non-standard css
72
+ attribute.gsub!(/^@@/, '@')
73
+
74
+ return attribute unless node.value.length == 3
75
+
76
+ value = node.value.last
77
+ value = "'#{value}'" if value !~ /^['"]/
78
+
79
+ case node.value[1]
80
+ when :equal
81
+ attribute + " = " + "#{value}"
82
+ when :not_equal
83
+ attribute + " != " + "#{value}"
84
+ when :substring_match
85
+ "contains(#{attribute}, #{value})"
86
+ when :prefix_match
87
+ "starts-with(#{attribute}, #{value})"
88
+ when :dash_match
89
+ "#{attribute} = #{value} or starts-with(#{attribute}, concat(#{value}, '-'))"
90
+ when :includes
91
+ "contains(concat(\" \", #{attribute}, \" \"),concat(\" \", #{value}, \" \"))"
92
+ when :suffix_match
93
+ "substring(#{attribute}, string-length(#{attribute}) - " +
94
+ "string-length(#{value}) + 1, string-length(#{value})) = #{value}"
95
+ else
96
+ attribute + " #{node.value[1]} " + "#{value}"
97
+ end
98
+ end
99
+
100
+ def visit_pseudo_class node
101
+ if node.value.first.is_a?(Nokogiri::CSS::Node) and node.value.first.type == :FUNCTION
102
+ node.value.first.accept(self)
103
+ else
104
+ msg = :"visit_pseudo_class_#{node.value.first.gsub(/[(]/, '')}"
105
+ return self.send(msg, node) if self.respond_to?(msg)
106
+
107
+ case node.value.first
108
+ when "first", "first-child" then "position() = 1"
109
+ when "last", "last-child" then "position() = last()"
110
+ when "first-of-type" then "position() = 1"
111
+ when "last-of-type" then "position() = last()"
112
+ when "only-of-type" then "last() = 1"
113
+ when "empty" then "not(node())"
114
+ when "parent" then "node()"
115
+ when "root" then "not(parent::*)"
116
+ else
117
+ node.value.first + "(.)"
118
+ end
119
+ end
120
+ end
121
+
122
+ def visit_class_condition node
123
+ "contains(concat(' ', @class, ' '), ' #{node.value.first} ')"
124
+ end
125
+
126
+ {
127
+ 'combinator' => ' and ',
128
+ 'direct_adjacent_selector' => "/following-sibling::*[1]/self::",
129
+ 'preceding_selector' => "/following-sibling::",
130
+ 'descendant_selector' => '//',
131
+ 'child_selector' => '/',
132
+ }.each do |k,v|
133
+ class_eval %{
134
+ def visit_#{k} node
135
+ "\#{node.value.first.accept(self)}#{v}\#{node.value.last.accept(self)}"
136
+ end
137
+ }
138
+ end
139
+
140
+ def visit_conditional_selector node
141
+ node.value.first.accept(self) + '[' +
142
+ node.value.last.accept(self) + ']'
143
+ end
144
+
145
+ def visit_element_name node
146
+ node.value.first
147
+ end
148
+
149
+ def accept node
150
+ node.accept(self)
151
+ end
152
+
153
+ private
154
+ def an_plus_b node, options={}
155
+ raise ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}" unless node.value.size == 4
156
+
157
+ a = node.value[0].to_i
158
+ b = node.value[3].to_i
159
+ position = options[:last] ? "(last()-position()+1)" : "position()"
160
+
161
+ if (b == 0)
162
+ return "(#{position} mod #{a}) = 0"
163
+ else
164
+ compare = (a < 0) ? "<=" : ">="
165
+ return "(#{position} #{compare} #{b}) and (((#{position}-#{b}) mod #{a.abs}) = 0)"
166
+ end
167
+ end
168
+
169
+ end
170
+ end
171
+ end
@@ -0,0 +1,35 @@
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
+ prefix = implied_xpath_context
11
+
12
+ if args.empty?
13
+ list = xpath("#{prefix}#{name.to_s.sub(/^_/, '')}")
14
+ elsif args.first.is_a? Hash
15
+ hash = args.first
16
+ if hash[:css]
17
+ list = css("#{name}#{hash[:css]}")
18
+ elsif hash[:xpath]
19
+ conds = Array(hash[:xpath]).join(' and ')
20
+ list = xpath("#{prefix}#{name}[#{conds}]")
21
+ end
22
+ else
23
+ CSS::Parser.without_cache do
24
+ list = xpath(
25
+ *CSS.xpath_for("#{name}#{args.first}", :prefix => prefix)
26
+ )
27
+ end
28
+ end
29
+
30
+ super if list.empty?
31
+ list.length == 1 ? list.first : list
32
+ end
33
+ end
34
+ end
35
+ end
@@ -0,0 +1,36 @@
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
+ require 'nokogiri/html/element_description_defaults'
8
+
9
+ module Nokogiri
10
+ class << self
11
+ ###
12
+ # Parse HTML. Convenience method for Nokogiri::HTML::Document.parse
13
+ def HTML thing, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block
14
+ Nokogiri::HTML::Document.parse(thing, url, encoding, options, &block)
15
+ end
16
+ end
17
+
18
+ module HTML
19
+ class << self
20
+ ###
21
+ # Parse HTML. Convenience method for Nokogiri::HTML::Document.parse
22
+ def parse thing, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block
23
+ Document.parse(thing, url, encoding, options, &block)
24
+ end
25
+
26
+ ####
27
+ # Parse a fragment from +string+ in to a NodeSet.
28
+ def fragment string, encoding = nil
29
+ HTML::DocumentFragment.parse string, encoding
30
+ end
31
+ end
32
+
33
+ # Instance of Nokogiri::HTML::EntityLookup
34
+ NamedCharacters = EntityLookup.new
35
+ end
36
+ 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 +1,221 @@
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
+ meta = meta_content_type and
9
+ /charset\s*=\s*([\w-]+)/i.match(meta['content'])[1]
10
+ end
11
+
12
+ ###
13
+ # Set the meta tag encoding for this document. If there is no meta
14
+ # content tag, the encoding is not set.
15
+ def meta_encoding= encoding
16
+ meta = meta_content_type and
17
+ meta['content'] = "text/html; charset=%s" % encoding
18
+ end
19
+
20
+ def meta_content_type
21
+ css('meta').find { |node|
22
+ node['http-equiv'] =~ /\AContent-Type\z/i
23
+ }
24
+ end
25
+ private :meta_content_type
26
+
27
+ ###
28
+ # Get the title string of this document. Return nil if there is
29
+ # no title tag.
30
+ def title
31
+ title = at('head title') and title.inner_text
32
+ end
33
+
34
+ ###
35
+ # Set the title string of this document. If there is no head
36
+ # element, the title is not set.
37
+ def title=(text)
38
+ unless title = at('head title')
39
+ head = at('head') or return nil
40
+ title = Nokogiri::XML::Node.new('title', self)
41
+ head << title
42
+ end
43
+ title.children = XML::Text.new(text, self)
44
+ end
45
+
46
+ ####
47
+ # Serialize Node using +options+. Save options can also be set using a
48
+ # block. See SaveOptions.
49
+ #
50
+ # These two statements are equivalent:
51
+ #
52
+ # node.serialize(:encoding => 'UTF-8', :save_with => FORMAT | AS_XML)
53
+ #
54
+ # or
55
+ #
56
+ # node.serialize(:encoding => 'UTF-8') do |config|
57
+ # config.format.as_xml
58
+ # end
59
+ #
60
+ def serialize options = {}
61
+ options[:save_with] ||= XML::Node::SaveOptions::FORMAT |
62
+ XML::Node::SaveOptions::AS_HTML |
63
+ XML::Node::SaveOptions::NO_DECLARATION |
64
+ XML::Node::SaveOptions::NO_EMPTY_TAGS
65
+ super
66
+ end
67
+
68
+ ####
69
+ # Create a Nokogiri::XML::DocumentFragment from +tags+
70
+ def fragment tags = nil
71
+ DocumentFragment.new(self, tags, self.root)
72
+ end
73
+
74
+ class << self
75
+ ###
76
+ # Parse HTML. +thing+ may be a String, or any object that
77
+ # responds to _read_ and _close_ such as an IO, or StringIO.
78
+ # +url+ is resource where this document is located. +encoding+ is the
79
+ # encoding that should be used when processing the document. +options+
80
+ # is a number that sets options in the parser, such as
81
+ # Nokogiri::XML::ParseOptions::RECOVER. See the constants in
82
+ # Nokogiri::XML::ParseOptions.
83
+ def parse string_or_io, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML
84
+
85
+ options = Nokogiri::XML::ParseOptions.new(options) if Fixnum === options
86
+ # Give the options to the user
87
+ yield options if block_given?
88
+
89
+ if string_or_io.respond_to?(:encoding)
90
+ unless string_or_io.encoding.name == "ASCII-8BIT"
91
+ encoding ||= string_or_io.encoding.name
92
+ end
93
+ end
94
+
95
+ if string_or_io.respond_to?(:read)
96
+ url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
97
+ if !encoding
98
+ # Perform further encoding detection that libxml2 does
99
+ # not do.
100
+ string_or_io = EncodingReader.new(string_or_io)
101
+ begin
102
+ return read_io(string_or_io, url, encoding, options.to_i)
103
+ rescue EncodingFoundException => e
104
+ # A retry is required because libxml2 has a problem in
105
+ # that it cannot switch encoding well in the middle of
106
+ # parsing, especially if it has already seen a
107
+ # non-ASCII character when it finds an encoding hint.
108
+ encoding = e.encoding
109
+ end
110
+ end
111
+ return read_io(string_or_io, url, encoding, options.to_i)
112
+ end
113
+
114
+ # read_memory pukes on empty docs
115
+ return new if string_or_io.nil? or string_or_io.empty?
116
+
117
+ if !encoding
118
+ encoding = EncodingReader.detect_encoding(string_or_io)
119
+ end
120
+
121
+ read_memory(string_or_io, url, encoding, options.to_i)
122
+ end
123
+ end
124
+
125
+ class EncodingFoundException < Exception # :nodoc:
126
+ attr_reader :encoding
127
+
128
+ def initialize(encoding)
129
+ @encoding = encoding
130
+ super("encoding found: %s" % encoding)
131
+ end
132
+ end
133
+
134
+ class EncodingReader # :nodoc:
135
+ class SAXHandler < Nokogiri::XML::SAX::Document # :nodoc:
136
+ attr_reader :encoding
137
+
138
+ def found(encoding)
139
+ @encoding = encoding
140
+ throw :found
141
+ end
142
+
143
+ def not_found(encoding)
144
+ found nil
145
+ end
146
+
147
+ def start_element(name, attrs = [])
148
+ case name
149
+ when 'head'
150
+ @head = true
151
+ when 'body'
152
+ not_found
153
+ when 'meta'
154
+ @head or return
155
+ attr = Hash[attrs]
156
+ http_equiv = attr['http-equiv'] and
157
+ http_equiv.match(/\AContent-Type\z/i) and
158
+ content = attr['content'] and
159
+ m = content.match(/;\s*charset\s*=\s*([\w-]+)/) and
160
+ found m[1]
161
+ end
162
+ end
163
+
164
+ def end_element(name)
165
+ if name == 'head'
166
+ not_found
167
+ end
168
+ end
169
+ end
170
+
171
+ def self.detect_encoding(chunk)
172
+ m = chunk.match(/\A(<\?xml[ \t\r\n]+[^>]*>)/) and
173
+ return Nokogiri.XML(m[1]).encoding
174
+
175
+ handler = SAXHandler.new
176
+ parser = Nokogiri::HTML::SAX::Parser.new(handler)
177
+ catch(:found) {
178
+ parser.parse(chunk)
179
+ }
180
+ handler.encoding
181
+ rescue => e
182
+ nil
183
+ end
184
+
185
+ def initialize(io)
186
+ @io = io
187
+ @firstchunk = nil
188
+ end
189
+
190
+ def read(len)
191
+ # no support for a call without len
192
+
193
+ if !@firstchunk
194
+ @firstchunk = @io.read(len) or return nil
195
+
196
+ # This implementation expects and assumes that the first
197
+ # call from htmlReadIO() is made with a length long enough
198
+ # (~1KB) to achieve further encoding detection that
199
+ # libxml2 does not do.
200
+ if encoding = EncodingReader.detect_encoding(@firstchunk)
201
+ raise EncodingFoundException, encoding
202
+ end
203
+
204
+ # This chunk is stored for the next read in retry.
205
+ return @firstchunk
206
+ end
207
+
208
+ ret = @firstchunk.slice!(0, len)
209
+ if (len -= ret.length) > 0
210
+ rest = @io.read(len) and ret << rest
211
+ end
212
+ if ret.empty?
213
+ nil
214
+ else
215
+ ret
216
+ end
217
+ end
218
+ end
219
+ end
220
+ end
221
+ end