nokogiri-fitzsimmons 1.5.5.3-java

Sign up to get free protection for your applications and to get access to all the features.
Files changed (297) hide show
  1. data/.autotest +26 -0
  2. data/.gemtest +0 -0
  3. data/CHANGELOG.ja.rdoc +666 -0
  4. data/CHANGELOG.rdoc +659 -0
  5. data/C_CODING_STYLE.rdoc +33 -0
  6. data/Manifest.txt +295 -0
  7. data/README.ja.rdoc +106 -0
  8. data/README.rdoc +178 -0
  9. data/ROADMAP.md +86 -0
  10. data/Rakefile +194 -0
  11. data/STANDARD_RESPONSES.md +47 -0
  12. data/Y_U_NO_GEMSPEC.md +155 -0
  13. data/bin/nokogiri +63 -0
  14. data/build_all +58 -0
  15. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  16. data/ext/java/nokogiri/HtmlDocument.java +163 -0
  17. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  18. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  19. data/ext/java/nokogiri/HtmlSaxParserContext.java +259 -0
  20. data/ext/java/nokogiri/NokogiriService.java +598 -0
  21. data/ext/java/nokogiri/XmlAttr.java +190 -0
  22. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  23. data/ext/java/nokogiri/XmlCdata.java +84 -0
  24. data/ext/java/nokogiri/XmlComment.java +95 -0
  25. data/ext/java/nokogiri/XmlDocument.java +580 -0
  26. data/ext/java/nokogiri/XmlDocumentFragment.java +234 -0
  27. data/ext/java/nokogiri/XmlDtd.java +469 -0
  28. data/ext/java/nokogiri/XmlElement.java +97 -0
  29. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  30. data/ext/java/nokogiri/XmlElementDecl.java +152 -0
  31. data/ext/java/nokogiri/XmlEntityDecl.java +162 -0
  32. data/ext/java/nokogiri/XmlEntityReference.java +97 -0
  33. data/ext/java/nokogiri/XmlNamespace.java +215 -0
  34. data/ext/java/nokogiri/XmlNode.java +1534 -0
  35. data/ext/java/nokogiri/XmlNodeSet.java +270 -0
  36. data/ext/java/nokogiri/XmlProcessingInstruction.java +99 -0
  37. data/ext/java/nokogiri/XmlReader.java +456 -0
  38. data/ext/java/nokogiri/XmlRelaxng.java +144 -0
  39. data/ext/java/nokogiri/XmlSaxParserContext.java +356 -0
  40. data/ext/java/nokogiri/XmlSaxPushParser.java +215 -0
  41. data/ext/java/nokogiri/XmlSchema.java +324 -0
  42. data/ext/java/nokogiri/XmlSyntaxError.java +136 -0
  43. data/ext/java/nokogiri/XmlText.java +119 -0
  44. data/ext/java/nokogiri/XmlXpathContext.java +203 -0
  45. data/ext/java/nokogiri/XsltStylesheet.java +360 -0
  46. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +243 -0
  47. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  48. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +86 -0
  49. data/ext/java/nokogiri/internals/NokogiriHandler.java +333 -0
  50. data/ext/java/nokogiri/internals/NokogiriHelpers.java +800 -0
  51. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +163 -0
  52. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +130 -0
  53. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +74 -0
  54. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  55. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +79 -0
  56. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +141 -0
  57. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +73 -0
  58. data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +67 -0
  59. data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
  60. data/ext/java/nokogiri/internals/ParserContext.java +288 -0
  61. data/ext/java/nokogiri/internals/ReaderNode.java +531 -0
  62. data/ext/java/nokogiri/internals/SaveContextVisitor.java +775 -0
  63. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +76 -0
  64. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  65. data/ext/java/nokogiri/internals/XmlDomParser.java +105 -0
  66. data/ext/java/nokogiri/internals/XmlDomParserContext.java +266 -0
  67. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  68. data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
  69. data/ext/nokogiri/depend +358 -0
  70. data/ext/nokogiri/extconf.rb +142 -0
  71. data/ext/nokogiri/html_document.c +170 -0
  72. data/ext/nokogiri/html_document.h +10 -0
  73. data/ext/nokogiri/html_element_description.c +276 -0
  74. data/ext/nokogiri/html_element_description.h +10 -0
  75. data/ext/nokogiri/html_entity_lookup.c +32 -0
  76. data/ext/nokogiri/html_entity_lookup.h +8 -0
  77. data/ext/nokogiri/html_sax_parser_context.c +116 -0
  78. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  79. data/ext/nokogiri/html_sax_push_parser.c +87 -0
  80. data/ext/nokogiri/html_sax_push_parser.h +9 -0
  81. data/ext/nokogiri/nokogiri.c +133 -0
  82. data/ext/nokogiri/nokogiri.h +160 -0
  83. data/ext/nokogiri/xml_attr.c +94 -0
  84. data/ext/nokogiri/xml_attr.h +9 -0
  85. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  86. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  87. data/ext/nokogiri/xml_cdata.c +56 -0
  88. data/ext/nokogiri/xml_cdata.h +9 -0
  89. data/ext/nokogiri/xml_comment.c +54 -0
  90. data/ext/nokogiri/xml_comment.h +9 -0
  91. data/ext/nokogiri/xml_document.c +576 -0
  92. data/ext/nokogiri/xml_document.h +23 -0
  93. data/ext/nokogiri/xml_document_fragment.c +48 -0
  94. data/ext/nokogiri/xml_document_fragment.h +10 -0
  95. data/ext/nokogiri/xml_dtd.c +202 -0
  96. data/ext/nokogiri/xml_dtd.h +10 -0
  97. data/ext/nokogiri/xml_element_content.c +123 -0
  98. data/ext/nokogiri/xml_element_content.h +10 -0
  99. data/ext/nokogiri/xml_element_decl.c +69 -0
  100. data/ext/nokogiri/xml_element_decl.h +9 -0
  101. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  102. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  103. data/ext/nokogiri/xml_entity_decl.c +110 -0
  104. data/ext/nokogiri/xml_entity_decl.h +10 -0
  105. data/ext/nokogiri/xml_entity_reference.c +52 -0
  106. data/ext/nokogiri/xml_entity_reference.h +9 -0
  107. data/ext/nokogiri/xml_io.c +56 -0
  108. data/ext/nokogiri/xml_io.h +11 -0
  109. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  110. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  111. data/ext/nokogiri/xml_namespace.c +78 -0
  112. data/ext/nokogiri/xml_namespace.h +13 -0
  113. data/ext/nokogiri/xml_node.c +1480 -0
  114. data/ext/nokogiri/xml_node.h +13 -0
  115. data/ext/nokogiri/xml_node_set.c +467 -0
  116. data/ext/nokogiri/xml_node_set.h +14 -0
  117. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  118. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  119. data/ext/nokogiri/xml_reader.c +684 -0
  120. data/ext/nokogiri/xml_reader.h +10 -0
  121. data/ext/nokogiri/xml_relax_ng.c +161 -0
  122. data/ext/nokogiri/xml_relax_ng.h +9 -0
  123. data/ext/nokogiri/xml_sax_parser.c +293 -0
  124. data/ext/nokogiri/xml_sax_parser.h +39 -0
  125. data/ext/nokogiri/xml_sax_parser_context.c +222 -0
  126. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  127. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  128. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  129. data/ext/nokogiri/xml_schema.c +205 -0
  130. data/ext/nokogiri/xml_schema.h +9 -0
  131. data/ext/nokogiri/xml_syntax_error.c +58 -0
  132. data/ext/nokogiri/xml_syntax_error.h +13 -0
  133. data/ext/nokogiri/xml_text.c +52 -0
  134. data/ext/nokogiri/xml_text.h +9 -0
  135. data/ext/nokogiri/xml_xpath_context.c +319 -0
  136. data/ext/nokogiri/xml_xpath_context.h +10 -0
  137. data/ext/nokogiri/xslt_stylesheet.c +270 -0
  138. data/ext/nokogiri/xslt_stylesheet.h +14 -0
  139. data/lib/isorelax.jar +0 -0
  140. data/lib/jing.jar +0 -0
  141. data/lib/nekodtd.jar +0 -0
  142. data/lib/nekohtml.jar +0 -0
  143. data/lib/nokogiri.rb +127 -0
  144. data/lib/nokogiri/css.rb +27 -0
  145. data/lib/nokogiri/css/node.rb +102 -0
  146. data/lib/nokogiri/css/parser.rb +720 -0
  147. data/lib/nokogiri/css/parser.y +258 -0
  148. data/lib/nokogiri/css/parser_extras.rb +91 -0
  149. data/lib/nokogiri/css/syntax_error.rb +7 -0
  150. data/lib/nokogiri/css/tokenizer.rb +152 -0
  151. data/lib/nokogiri/css/tokenizer.rex +55 -0
  152. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  153. data/lib/nokogiri/decorators/slop.rb +35 -0
  154. data/lib/nokogiri/html.rb +37 -0
  155. data/lib/nokogiri/html/builder.rb +35 -0
  156. data/lib/nokogiri/html/document.rb +254 -0
  157. data/lib/nokogiri/html/document_fragment.rb +41 -0
  158. data/lib/nokogiri/html/element_description.rb +23 -0
  159. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  160. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  161. data/lib/nokogiri/html/sax/parser.rb +52 -0
  162. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  163. data/lib/nokogiri/html/sax/push_parser.rb +16 -0
  164. data/lib/nokogiri/nokogiri.jar +0 -0
  165. data/lib/nokogiri/syntax_error.rb +4 -0
  166. data/lib/nokogiri/version.rb +88 -0
  167. data/lib/nokogiri/xml.rb +73 -0
  168. data/lib/nokogiri/xml/attr.rb +14 -0
  169. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  170. data/lib/nokogiri/xml/builder.rb +431 -0
  171. data/lib/nokogiri/xml/cdata.rb +11 -0
  172. data/lib/nokogiri/xml/character_data.rb +7 -0
  173. data/lib/nokogiri/xml/document.rb +267 -0
  174. data/lib/nokogiri/xml/document_fragment.rb +103 -0
  175. data/lib/nokogiri/xml/dtd.rb +22 -0
  176. data/lib/nokogiri/xml/element_content.rb +36 -0
  177. data/lib/nokogiri/xml/element_decl.rb +13 -0
  178. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  179. data/lib/nokogiri/xml/namespace.rb +13 -0
  180. data/lib/nokogiri/xml/node.rb +946 -0
  181. data/lib/nokogiri/xml/node/save_options.rb +61 -0
  182. data/lib/nokogiri/xml/node_set.rb +357 -0
  183. data/lib/nokogiri/xml/notation.rb +6 -0
  184. data/lib/nokogiri/xml/parse_options.rb +98 -0
  185. data/lib/nokogiri/xml/pp.rb +2 -0
  186. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  187. data/lib/nokogiri/xml/pp/node.rb +56 -0
  188. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  189. data/lib/nokogiri/xml/reader.rb +112 -0
  190. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  191. data/lib/nokogiri/xml/sax.rb +4 -0
  192. data/lib/nokogiri/xml/sax/document.rb +164 -0
  193. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  194. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  195. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  196. data/lib/nokogiri/xml/schema.rb +63 -0
  197. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  198. data/lib/nokogiri/xml/text.rb +9 -0
  199. data/lib/nokogiri/xml/xpath.rb +10 -0
  200. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  201. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  202. data/lib/nokogiri/xslt.rb +56 -0
  203. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  204. data/lib/xercesImpl.jar +0 -0
  205. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  206. data/tasks/cross_compile.rb +153 -0
  207. data/tasks/nokogiri.org.rb +24 -0
  208. data/tasks/test.rb +95 -0
  209. data/test/css/test_nthiness.rb +159 -0
  210. data/test/css/test_parser.rb +341 -0
  211. data/test/css/test_tokenizer.rb +198 -0
  212. data/test/css/test_xpath_visitor.rb +91 -0
  213. data/test/decorators/test_slop.rb +16 -0
  214. data/test/files/2ch.html +108 -0
  215. data/test/files/address_book.rlx +12 -0
  216. data/test/files/address_book.xml +10 -0
  217. data/test/files/bar/bar.xsd +4 -0
  218. data/test/files/dont_hurt_em_why.xml +422 -0
  219. data/test/files/encoding.html +82 -0
  220. data/test/files/encoding.xhtml +84 -0
  221. data/test/files/exslt.xml +8 -0
  222. data/test/files/exslt.xslt +35 -0
  223. data/test/files/foo/foo.xsd +4 -0
  224. data/test/files/metacharset.html +10 -0
  225. data/test/files/noencoding.html +47 -0
  226. data/test/files/po.xml +32 -0
  227. data/test/files/po.xsd +66 -0
  228. data/test/files/shift_jis.html +10 -0
  229. data/test/files/shift_jis.xml +5 -0
  230. data/test/files/snuggles.xml +3 -0
  231. data/test/files/staff.dtd +10 -0
  232. data/test/files/staff.xml +59 -0
  233. data/test/files/staff.xslt +32 -0
  234. data/test/files/tlm.html +850 -0
  235. data/test/files/to_be_xincluded.xml +2 -0
  236. data/test/files/valid_bar.xml +2 -0
  237. data/test/files/xinclude.xml +4 -0
  238. data/test/helper.rb +147 -0
  239. data/test/html/sax/test_parser.rb +138 -0
  240. data/test/html/sax/test_parser_context.rb +46 -0
  241. data/test/html/test_builder.rb +164 -0
  242. data/test/html/test_document.rb +529 -0
  243. data/test/html/test_document_encoding.rb +138 -0
  244. data/test/html/test_document_fragment.rb +254 -0
  245. data/test/html/test_element_description.rb +100 -0
  246. data/test/html/test_named_characters.rb +14 -0
  247. data/test/html/test_node.rb +188 -0
  248. data/test/html/test_node_encoding.rb +27 -0
  249. data/test/test_convert_xpath.rb +135 -0
  250. data/test/test_css_cache.rb +45 -0
  251. data/test/test_encoding_handler.rb +46 -0
  252. data/test/test_memory_leak.rb +152 -0
  253. data/test/test_nokogiri.rb +132 -0
  254. data/test/test_reader.rb +488 -0
  255. data/test/test_soap4r_sax.rb +52 -0
  256. data/test/test_xslt_transforms.rb +254 -0
  257. data/test/xml/node/test_save_options.rb +28 -0
  258. data/test/xml/node/test_subclass.rb +44 -0
  259. data/test/xml/sax/test_parser.rb +338 -0
  260. data/test/xml/sax/test_parser_context.rb +106 -0
  261. data/test/xml/sax/test_push_parser.rb +157 -0
  262. data/test/xml/test_attr.rb +64 -0
  263. data/test/xml/test_attribute_decl.rb +86 -0
  264. data/test/xml/test_builder.rb +248 -0
  265. data/test/xml/test_c14n.rb +151 -0
  266. data/test/xml/test_cdata.rb +48 -0
  267. data/test/xml/test_comment.rb +29 -0
  268. data/test/xml/test_document.rb +742 -0
  269. data/test/xml/test_document_encoding.rb +28 -0
  270. data/test/xml/test_document_fragment.rb +216 -0
  271. data/test/xml/test_dtd.rb +103 -0
  272. data/test/xml/test_dtd_encoding.rb +33 -0
  273. data/test/xml/test_element_content.rb +56 -0
  274. data/test/xml/test_element_decl.rb +73 -0
  275. data/test/xml/test_entity_decl.rb +122 -0
  276. data/test/xml/test_entity_reference.rb +235 -0
  277. data/test/xml/test_namespace.rb +75 -0
  278. data/test/xml/test_node.rb +1029 -0
  279. data/test/xml/test_node_attributes.rb +53 -0
  280. data/test/xml/test_node_encoding.rb +107 -0
  281. data/test/xml/test_node_inheritance.rb +32 -0
  282. data/test/xml/test_node_reparenting.rb +374 -0
  283. data/test/xml/test_node_set.rb +755 -0
  284. data/test/xml/test_parse_options.rb +64 -0
  285. data/test/xml/test_processing_instruction.rb +30 -0
  286. data/test/xml/test_reader_encoding.rb +142 -0
  287. data/test/xml/test_relax_ng.rb +60 -0
  288. data/test/xml/test_schema.rb +94 -0
  289. data/test/xml/test_syntax_error.rb +12 -0
  290. data/test/xml/test_text.rb +45 -0
  291. data/test/xml/test_unparented_node.rb +413 -0
  292. data/test/xml/test_xinclude.rb +83 -0
  293. data/test/xml/test_xpath.rb +295 -0
  294. data/test/xslt/test_custom_functions.rb +129 -0
  295. data/test/xslt/test_exception_handling.rb +37 -0
  296. data/test_all +84 -0
  297. metadata +571 -0
@@ -0,0 +1,14 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module HTML
5
+ class TestNamedCharacters < Nokogiri::TestCase
6
+ def test_named_character
7
+ copy = NamedCharacters.get('copy')
8
+ assert_equal 169, NamedCharacters['copy']
9
+ assert_equal copy.value, NamedCharacters['copy']
10
+ assert copy.description
11
+ end
12
+ end
13
+ end
14
+ end
@@ -0,0 +1,188 @@
1
+ require "helper"
2
+
3
+ require 'nkf'
4
+
5
+ module Nokogiri
6
+ module HTML
7
+ class TestNode < Nokogiri::TestCase
8
+ def setup
9
+ super
10
+ @html = Nokogiri::HTML(<<-eohtml)
11
+ <html>
12
+ <head></head>
13
+ <body>
14
+ <div class='baz'><a href="foo" class="bar">first</a></div>
15
+ </body>
16
+ </html>
17
+ eohtml
18
+ end
19
+
20
+ def test_to_a
21
+ assert_equal [['class', 'bar'], ['href', 'foo']],@html.at('a').to_a.sort
22
+ end
23
+
24
+ def test_attr
25
+ node = @html.at('div.baz')
26
+ assert_equal node['class'], node.attr('class')
27
+ end
28
+
29
+ def test_get_attribute
30
+ element = @html.at('div')
31
+ assert_equal 'baz', element.get_attribute('class')
32
+ assert_equal 'baz', element['class']
33
+ element['href'] = "javascript:alert(\"AGGA-KA-BOO!\")"
34
+ assert_match(/%22AGGA-KA-BOO!%22/, element.to_html)
35
+ end
36
+
37
+ def test_css_path_round_trip
38
+ doc = Nokogiri::HTML(File.read(HTML_FILE))
39
+ %w{ #header small div[2] div.post body }.each do |css_sel|
40
+ ele = doc.at css_sel
41
+ assert_equal ele, doc.at(ele.css_path), ele.css_path
42
+ end
43
+ end
44
+
45
+ def test_path_round_trip
46
+ doc = Nokogiri::HTML(File.read(HTML_FILE))
47
+ %w{ #header small div[2] div.post body }.each do |css_sel|
48
+ ele = doc.at css_sel
49
+ assert_equal ele, doc.at(ele.path), ele.path
50
+ end
51
+ end
52
+
53
+ def test_append_with_document
54
+ assert_raises(ArgumentError) do
55
+ @html.root << Nokogiri::HTML::Document.new
56
+ end
57
+ end
58
+
59
+ ###
60
+ # Make sure a document that doesn't declare a meta encoding returns
61
+ # nil.
62
+ def test_meta_encoding
63
+ assert_nil @html.meta_encoding
64
+ end
65
+
66
+ def test_description
67
+ assert desc = @html.at('a.bar').description
68
+ assert_equal 'a', desc.name
69
+ end
70
+
71
+ def test_ancestors_with_selector
72
+ assert node = @html.at('a.bar').child
73
+ assert list = node.ancestors('.baz')
74
+ assert_equal 1, list.length
75
+ assert_equal 'div', list.first.name
76
+ end
77
+
78
+ def test_matches_inside_fragment
79
+ fragment = DocumentFragment.new @html
80
+ fragment << XML::Node.new('a', @html)
81
+
82
+ a = fragment.children.last
83
+ assert a.matches?('a'), 'a should match'
84
+ end
85
+
86
+ def test_css_matches?
87
+ assert node = @html.at('a.bar')
88
+ assert node.matches?('a.bar')
89
+ end
90
+
91
+ def test_xpath_matches?
92
+ assert node = @html.at('//a')
93
+ assert node.matches?('//a')
94
+ end
95
+
96
+ def test_unlink_then_swap
97
+ node = @html.at('a')
98
+ node.unlink
99
+
100
+ another_node = @html.at('div')
101
+ assert another_node, 'should have a node'
102
+
103
+ # This used to segv
104
+ assert node.add_previous_sibling another_node
105
+ end
106
+
107
+ def test_swap
108
+ @html.at('div').swap('<a href="foo">bar</a>')
109
+ a_tag = @html.css('a').first
110
+ assert_equal 'body', a_tag.parent.name
111
+ assert_equal 0, @html.css('div').length
112
+ end
113
+
114
+ def test_swap_with_regex_characters
115
+ @html.at('div').swap('<a href="foo">ba)r</a>')
116
+ a_tag = @html.css('a').first
117
+ assert_equal 'ba)r', a_tag.text
118
+ end
119
+
120
+ def test_attribute_decodes_entities
121
+ node = @html.at('div')
122
+ node['href'] = 'foo&bar'
123
+ assert_equal 'foo&bar', node['href']
124
+ node['href'] += '&baz'
125
+ assert_equal 'foo&bar&baz', node['href']
126
+ end
127
+
128
+ def test_parse_config_option
129
+ node = @html.at('div')
130
+ options = nil
131
+ node.parse("<div></div>") do |config|
132
+ options = config
133
+ end
134
+ assert_equal Nokogiri::XML::ParseOptions::DEFAULT_HTML, options.to_i
135
+ end
136
+
137
+ def test_fragment_handler_does_not_regurge_on_invalid_attributes
138
+ iframe = %Q{<iframe style="width: 0%; height: 0px" src="http://someurl" allowtransparency></iframe>}
139
+ assert @html.at('div').fragment(iframe)
140
+ end
141
+
142
+ def test_fragment
143
+ fragment = @html.fragment(<<-eohtml)
144
+ hello
145
+ <div class="foo">
146
+ <p>bar</p>
147
+ </div>
148
+ world
149
+ eohtml
150
+ assert_match(/^hello/, fragment.inner_html.strip)
151
+ assert_equal 3, fragment.children.length
152
+ assert p_tag = fragment.css('p').first
153
+ assert_equal 'div', p_tag.parent.name
154
+ assert_equal 'foo', p_tag.parent['class']
155
+ end
156
+
157
+ def test_fragment_serialization
158
+ fragment = Nokogiri::HTML.fragment("<div>foo</div>")
159
+ assert_equal "<div>foo</div>", fragment.serialize.chomp
160
+ assert_equal "<div>foo</div>", fragment.to_xml.chomp
161
+ assert_equal "<div>foo</div>", fragment.inner_html
162
+ assert_equal "<div>foo</div>", fragment.to_html
163
+ assert_equal "<div>foo</div>", fragment.to_s
164
+ end
165
+
166
+ def test_to_html_does_not_contain_entities
167
+ return unless defined?(NKF) # NKF is not implemented on Rubinius as of 2009-11-23
168
+ html = NKF.nkf("-e --msdos", <<-EOH)
169
+ <html><body>
170
+ <p> test paragraph
171
+ foo bar </p>
172
+ </body></html>
173
+ EOH
174
+ nokogiri = Nokogiri::HTML.parse(html)
175
+
176
+ if RUBY_PLATFORM =~ /java/
177
+ # NKF linebreak modes are not supported as of jruby 1.2
178
+ # see http://jira.codehaus.org/browse/JRUBY-3602 for status
179
+ assert_equal "<p>testparagraph\nfoobar</p>",
180
+ nokogiri.at("p").to_html.gsub(/ /, '')
181
+ else
182
+ assert_equal "<p>testparagraph\r\nfoobar</p>",
183
+ nokogiri.at("p").to_html.gsub(/ /, '')
184
+ end
185
+ end
186
+ end
187
+ end
188
+ end
@@ -0,0 +1,27 @@
1
+ # -*- coding: utf-8 -*-
2
+ require "helper"
3
+
4
+ module Nokogiri
5
+ module HTML
6
+ if RUBY_VERSION =~ /^1\.9/
7
+ class TestNodeEncoding < Nokogiri::TestCase
8
+ def test_inner_html
9
+ doc = Nokogiri::HTML File.open(SHIFT_JIS_HTML, 'rb')
10
+
11
+ hello = "こんにちは"
12
+
13
+ contents = doc.at('h2').inner_html
14
+ assert_equal doc.encoding, contents.encoding.name
15
+ assert_match hello.encode('Shift_JIS'), contents
16
+
17
+ contents = doc.at('h2').inner_html(:encoding => 'UTF-8')
18
+ assert_match hello, contents
19
+
20
+ doc.encoding = 'UTF-8'
21
+ contents = doc.at('h2').inner_html
22
+ assert_match hello, contents
23
+ end
24
+ end
25
+ end
26
+ end
27
+ end
@@ -0,0 +1,135 @@
1
+ require "helper"
2
+
3
+ class TestConvertXPath < Nokogiri::TestCase
4
+
5
+ def setup
6
+ super
7
+ @N = Nokogiri(File.read(HTML_FILE))
8
+ end
9
+
10
+ def assert_syntactical_equivalence(hpath, xpath, match, &blk)
11
+ blk ||= lambda {|j| j.first}
12
+ assert_equal match, blk.call(@N.search(xpath)), "xpath result did not match"
13
+ end
14
+
15
+ def test_child_tag
16
+ assert_syntactical_equivalence("h1[a]", ".//h1[child::a]", "Tender Lovemaking") do |j|
17
+ j.inner_text
18
+ end
19
+ end
20
+
21
+ def test_child_tag_equals
22
+ assert_syntactical_equivalence("h1[a='Tender Lovemaking']", ".//h1[child::a = 'Tender Lovemaking']", "Tender Lovemaking") do |j|
23
+ j.inner_text
24
+ end
25
+ end
26
+
27
+ def test_filter_contains
28
+ assert_syntactical_equivalence("title:contains('Tender')", ".//title[contains(., 'Tender')]",
29
+ "Tender Lovemaking ") do |j|
30
+ j.inner_text
31
+ end
32
+ end
33
+
34
+ def test_filter_comment
35
+ assert_syntactical_equivalence("div comment()[2]", ".//div//comment()[position() = 2]", "<!-- end of header -->") do |j|
36
+ j.first.to_s
37
+ end
38
+ end
39
+
40
+ def test_filter_text
41
+ assert_syntactical_equivalence("a[text()]", ".//a[normalize-space(child::text())]", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
42
+ j.first.to_s
43
+ end
44
+ assert_syntactical_equivalence("a[text()='Tender Lovemaking']", ".//a[normalize-space(child::text()) = 'Tender Lovemaking']", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
45
+ j.first.to_s
46
+ end
47
+ assert_syntactical_equivalence("a/text()", ".//a/child::text()", "Tender Lovemaking") do |j|
48
+ j.first.to_s
49
+ end
50
+ assert_syntactical_equivalence("h2//a[text()!='Back Home!']", ".//h2//a[normalize-space(child::text()) != 'Back Home!']", "Meow meow meow meow meow") do |j|
51
+ j.first.inner_text
52
+ end
53
+ end
54
+
55
+ def test_filter_by_attr
56
+ assert_syntactical_equivalence("a[@href='http://blog.geminigeek.com/wordpress-theme']",
57
+ ".//a[@href = 'http://blog.geminigeek.com/wordpress-theme']",
58
+ "http://blog.geminigeek.com/wordpress-theme") do |j|
59
+ j.first["href"]
60
+ end
61
+ end
62
+
63
+ def test_css_id
64
+ assert_syntactical_equivalence("#linkcat-7", ".//*[@id = 'linkcat-7']", "linkcat-7") do |j|
65
+ j.first["id"]
66
+ end
67
+ assert_syntactical_equivalence("li#linkcat-7", ".//li[@id = 'linkcat-7']", "linkcat-7") do |j|
68
+ j.first["id"]
69
+ end
70
+ end
71
+
72
+ def test_css_class
73
+ assert_syntactical_equivalence(".cat-item-15", ".//*[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
74
+ "cat-item cat-item-15") do |j|
75
+ j.first["class"]
76
+ end
77
+ assert_syntactical_equivalence("li.cat-item-15", ".//li[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
78
+ "cat-item cat-item-15") do |j|
79
+ j.first["class"]
80
+ end
81
+ end
82
+
83
+ def test_css_tags
84
+ assert_syntactical_equivalence("div li a", ".//div//li//a", "http://brobinius.org/") do |j|
85
+ j.first.inner_text
86
+ end
87
+ assert_syntactical_equivalence("div li > a", ".//div//li/a", "http://brobinius.org/") do |j|
88
+ j.first.inner_text
89
+ end
90
+ assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
91
+ j.first.inner_text
92
+ end
93
+ assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
94
+ j.first.inner_text
95
+ end
96
+ end
97
+
98
+ def test_positional
99
+ assert_syntactical_equivalence("div/div:first()", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
100
+ j.first.inner_text.gsub(/[\r\n]/, '')
101
+ end
102
+ assert_syntactical_equivalence("div/div:first", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
103
+ j.first.inner_text.gsub(/[\r\n]/, '')
104
+ end
105
+ assert_syntactical_equivalence("div//a:last()", ".//div//a[position() = last()]", "Wordpress") do |j|
106
+ j.last.inner_text
107
+ end
108
+ assert_syntactical_equivalence("div//a:last", ".//div//a[position() = last()]", "Wordpress") do |j|
109
+ j.last.inner_text
110
+ end
111
+ end
112
+
113
+ def test_multiple_filters
114
+ assert_syntactical_equivalence("a[@rel='bookmark'][1]", ".//a[@rel = 'bookmark' and position() = 1]", "Back Home!") do |j|
115
+ j.first.inner_text
116
+ end
117
+ end
118
+
119
+ # TODO:
120
+ # doc/'title ~ link' -> links that are siblings of title
121
+ # doc/'p[@class~="final"]' -> class includes string (whitespacy)
122
+ # doc/'p[text()*="final"]' -> class includes string (index) (broken: always returns true?)
123
+ # doc/'p[text()$="final"]' -> /final$/
124
+ # doc/'p[text()|="final"]' -> /^final$/
125
+ # doc/'p[text()^="final"]' -> string starts with 'final
126
+ # nth_first
127
+ # nth_last
128
+ # even
129
+ # odd
130
+ # first-child, nth-child, last-child, nth-last-child, nth-last-of-type
131
+ # only-of-type, only-child
132
+ # parent
133
+ # empty
134
+ # root
135
+ end
@@ -0,0 +1,45 @@
1
+ require "helper"
2
+
3
+ class TestCssCache < Nokogiri::TestCase
4
+
5
+ def setup
6
+ super
7
+ @css = "a1 > b2 > c3"
8
+ @parse_result = Nokogiri::CSS.parse(@css)
9
+ @to_xpath_result = @parse_result.map {|ast| ast.to_xpath}
10
+ Nokogiri::CSS::Parser.class_eval do
11
+ class << @cache
12
+ alias :old_bracket :[]
13
+ attr_reader :count
14
+ def [](key)
15
+ @count ||= 0
16
+ @count += 1
17
+ old_bracket(key)
18
+ end
19
+ end
20
+ end
21
+ assert Nokogiri::CSS::Parser.cache_on?
22
+ end
23
+
24
+ def teardown
25
+ Nokogiri::CSS::Parser.clear_cache
26
+ Nokogiri::CSS::Parser.set_cache true
27
+ end
28
+
29
+ [ false, true ].each do |cache_setting|
30
+ define_method "test_css_cache_#{cache_setting ? "true" : "false"}" do
31
+ times = cache_setting ? 4 : nil
32
+
33
+ Nokogiri::CSS::Parser.set_cache cache_setting
34
+
35
+ Nokogiri::CSS.xpath_for(@css)
36
+ Nokogiri::CSS.xpath_for(@css)
37
+ Nokogiri::CSS::Parser.new.xpath_for(@css)
38
+ Nokogiri::CSS::Parser.new.xpath_for(@css)
39
+
40
+ assert_equal(times, Nokogiri::CSS::Parser.class_eval { @cache.count })
41
+ end
42
+ end
43
+
44
+
45
+ end
@@ -0,0 +1,46 @@
1
+ # -*- coding: utf-8 -*-
2
+
3
+ require "helper"
4
+
5
+ class TestEncodingHandler < Nokogiri::TestCase
6
+ def teardown
7
+ Nokogiri::EncodingHandler.clear_aliases!
8
+ end
9
+
10
+ def test_get
11
+ assert_not_nil Nokogiri::EncodingHandler['UTF-8']
12
+ assert_nil Nokogiri::EncodingHandler['alsdkjfhaldskjfh']
13
+ end
14
+
15
+ def test_name
16
+ eh = Nokogiri::EncodingHandler['UTF-8']
17
+ assert_equal "UTF-8", eh.name
18
+ end
19
+
20
+ def test_alias
21
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-18')
22
+ assert_equal 'UTF-8', Nokogiri::EncodingHandler['UTF-18'].name
23
+ end
24
+
25
+ def test_cleanup_aliases
26
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
27
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
28
+ assert_not_nil Nokogiri::EncodingHandler['UTF-9']
29
+
30
+ Nokogiri::EncodingHandler.clear_aliases!
31
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
32
+ end
33
+
34
+ def test_delete
35
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
36
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
37
+ assert_not_nil Nokogiri::EncodingHandler['UTF-9']
38
+
39
+ Nokogiri::EncodingHandler.delete 'UTF-9'
40
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
41
+ end
42
+
43
+ def test_delete_non_existent
44
+ assert_nil Nokogiri::EncodingHandler.delete('UTF-9')
45
+ end
46
+ end