nokogiri-maven 1.5.0-java

Sign up to get free protection for your applications and to get access to all the features.
Files changed (278) hide show
  1. data/CHANGELOG.ja.rdoc +544 -0
  2. data/CHANGELOG.rdoc +532 -0
  3. data/Manifest.txt +283 -0
  4. data/README.ja.rdoc +106 -0
  5. data/README.rdoc +174 -0
  6. data/Rakefile +164 -0
  7. data/bin/nokogiri +53 -0
  8. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  9. data/ext/java/nokogiri/HtmlDocument.java +119 -0
  10. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  11. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  12. data/ext/java/nokogiri/HtmlSaxParserContext.java +259 -0
  13. data/ext/java/nokogiri/NokogiriService.java +590 -0
  14. data/ext/java/nokogiri/XmlAttr.java +180 -0
  15. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  16. data/ext/java/nokogiri/XmlCdata.java +84 -0
  17. data/ext/java/nokogiri/XmlComment.java +86 -0
  18. data/ext/java/nokogiri/XmlDocument.java +519 -0
  19. data/ext/java/nokogiri/XmlDocumentFragment.java +223 -0
  20. data/ext/java/nokogiri/XmlDtd.java +469 -0
  21. data/ext/java/nokogiri/XmlElement.java +195 -0
  22. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  23. data/ext/java/nokogiri/XmlElementDecl.java +152 -0
  24. data/ext/java/nokogiri/XmlEntityDecl.java +162 -0
  25. data/ext/java/nokogiri/XmlEntityReference.java +97 -0
  26. data/ext/java/nokogiri/XmlNamespace.java +183 -0
  27. data/ext/java/nokogiri/XmlNode.java +1378 -0
  28. data/ext/java/nokogiri/XmlNodeSet.java +267 -0
  29. data/ext/java/nokogiri/XmlProcessingInstruction.java +99 -0
  30. data/ext/java/nokogiri/XmlReader.java +408 -0
  31. data/ext/java/nokogiri/XmlRelaxng.java +144 -0
  32. data/ext/java/nokogiri/XmlSaxParserContext.java +367 -0
  33. data/ext/java/nokogiri/XmlSaxPushParser.java +184 -0
  34. data/ext/java/nokogiri/XmlSchema.java +324 -0
  35. data/ext/java/nokogiri/XmlSyntaxError.java +119 -0
  36. data/ext/java/nokogiri/XmlText.java +119 -0
  37. data/ext/java/nokogiri/XmlXpathContext.java +199 -0
  38. data/ext/java/nokogiri/XsltStylesheet.java +197 -0
  39. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +204 -0
  40. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  41. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +86 -0
  42. data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
  43. data/ext/java/nokogiri/internals/NokogiriHelpers.java +639 -0
  44. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +167 -0
  45. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +130 -0
  46. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +74 -0
  47. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  48. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +79 -0
  49. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +141 -0
  50. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +73 -0
  51. data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +67 -0
  52. data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +86 -0
  53. data/ext/java/nokogiri/internals/ParserContext.java +276 -0
  54. data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
  55. data/ext/java/nokogiri/internals/ReaderNode.java +531 -0
  56. data/ext/java/nokogiri/internals/SaveContextVisitor.java +567 -0
  57. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +76 -0
  58. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  59. data/ext/java/nokogiri/internals/XmlDomParser.java +76 -0
  60. data/ext/java/nokogiri/internals/XmlDomParserContext.java +244 -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 +115 -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 +56 -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 +1385 -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 +293 -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 +264 -0
  130. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  131. data/lib/nokogiri.rb +127 -0
  132. data/lib/nokogiri/css.rb +27 -0
  133. data/lib/nokogiri/css/node.rb +99 -0
  134. data/lib/nokogiri/css/parser.rb +677 -0
  135. data/lib/nokogiri/css/parser.y +237 -0
  136. data/lib/nokogiri/css/parser_extras.rb +91 -0
  137. data/lib/nokogiri/css/syntax_error.rb +7 -0
  138. data/lib/nokogiri/css/tokenizer.rb +152 -0
  139. data/lib/nokogiri/css/tokenizer.rex +55 -0
  140. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  141. data/lib/nokogiri/decorators/slop.rb +35 -0
  142. data/lib/nokogiri/html.rb +36 -0
  143. data/lib/nokogiri/html/builder.rb +35 -0
  144. data/lib/nokogiri/html/document.rb +213 -0
  145. data/lib/nokogiri/html/document_fragment.rb +41 -0
  146. data/lib/nokogiri/html/element_description.rb +23 -0
  147. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  148. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  149. data/lib/nokogiri/html/sax/parser.rb +52 -0
  150. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  151. data/lib/nokogiri/nokogiri.jar +0 -0
  152. data/lib/nokogiri/syntax_error.rb +4 -0
  153. data/lib/nokogiri/version.rb +88 -0
  154. data/lib/nokogiri/xml.rb +67 -0
  155. data/lib/nokogiri/xml/attr.rb +14 -0
  156. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  157. data/lib/nokogiri/xml/builder.rb +425 -0
  158. data/lib/nokogiri/xml/cdata.rb +11 -0
  159. data/lib/nokogiri/xml/character_data.rb +7 -0
  160. data/lib/nokogiri/xml/document.rb +234 -0
  161. data/lib/nokogiri/xml/document_fragment.rb +98 -0
  162. data/lib/nokogiri/xml/dtd.rb +22 -0
  163. data/lib/nokogiri/xml/element_content.rb +36 -0
  164. data/lib/nokogiri/xml/element_decl.rb +13 -0
  165. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  166. data/lib/nokogiri/xml/namespace.rb +13 -0
  167. data/lib/nokogiri/xml/node.rb +915 -0
  168. data/lib/nokogiri/xml/node/save_options.rb +61 -0
  169. data/lib/nokogiri/xml/node_set.rb +357 -0
  170. data/lib/nokogiri/xml/notation.rb +6 -0
  171. data/lib/nokogiri/xml/parse_options.rb +93 -0
  172. data/lib/nokogiri/xml/pp.rb +2 -0
  173. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  174. data/lib/nokogiri/xml/pp/node.rb +56 -0
  175. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  176. data/lib/nokogiri/xml/reader.rb +112 -0
  177. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  178. data/lib/nokogiri/xml/sax.rb +4 -0
  179. data/lib/nokogiri/xml/sax/document.rb +164 -0
  180. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  181. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  182. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  183. data/lib/nokogiri/xml/schema.rb +63 -0
  184. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  185. data/lib/nokogiri/xml/text.rb +9 -0
  186. data/lib/nokogiri/xml/xpath.rb +10 -0
  187. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  188. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  189. data/lib/nokogiri/xslt.rb +52 -0
  190. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  191. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  192. data/nokogiri_help_responses.md +40 -0
  193. data/tasks/cross_compile.rb +152 -0
  194. data/tasks/nokogiri.org.rb +18 -0
  195. data/tasks/test.rb +94 -0
  196. data/test/css/test_nthiness.rb +159 -0
  197. data/test/css/test_parser.rb +303 -0
  198. data/test/css/test_tokenizer.rb +198 -0
  199. data/test/css/test_xpath_visitor.rb +85 -0
  200. data/test/decorators/test_slop.rb +16 -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/encoding.html +82 -0
  207. data/test/files/encoding.xhtml +84 -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/metacharset.html +10 -0
  212. data/test/files/noencoding.html +47 -0
  213. data/test/files/po.xml +32 -0
  214. data/test/files/po.xsd +66 -0
  215. data/test/files/shift_jis.html +10 -0
  216. data/test/files/shift_jis.xml +5 -0
  217. data/test/files/snuggles.xml +3 -0
  218. data/test/files/staff.dtd +10 -0
  219. data/test/files/staff.xml +59 -0
  220. data/test/files/staff.xslt +32 -0
  221. data/test/files/tlm.html +850 -0
  222. data/test/files/valid_bar.xml +2 -0
  223. data/test/helper.rb +173 -0
  224. data/test/html/sax/test_parser.rb +136 -0
  225. data/test/html/sax/test_parser_context.rb +48 -0
  226. data/test/html/test_builder.rb +164 -0
  227. data/test/html/test_document.rb +472 -0
  228. data/test/html/test_document_encoding.rb +138 -0
  229. data/test/html/test_document_fragment.rb +255 -0
  230. data/test/html/test_element_description.rb +100 -0
  231. data/test/html/test_named_characters.rb +14 -0
  232. data/test/html/test_node.rb +190 -0
  233. data/test/html/test_node_encoding.rb +27 -0
  234. data/test/test_convert_xpath.rb +135 -0
  235. data/test/test_css_cache.rb +45 -0
  236. data/test/test_encoding_handler.rb +46 -0
  237. data/test/test_memory_leak.rb +72 -0
  238. data/test/test_nokogiri.rb +132 -0
  239. data/test/test_reader.rb +425 -0
  240. data/test/test_soap4r_sax.rb +52 -0
  241. data/test/test_xslt_transforms.rb +193 -0
  242. data/test/xml/node/test_save_options.rb +28 -0
  243. data/test/xml/node/test_subclass.rb +44 -0
  244. data/test/xml/sax/test_parser.rb +338 -0
  245. data/test/xml/sax/test_parser_context.rb +113 -0
  246. data/test/xml/sax/test_push_parser.rb +156 -0
  247. data/test/xml/test_attr.rb +65 -0
  248. data/test/xml/test_attribute_decl.rb +86 -0
  249. data/test/xml/test_builder.rb +227 -0
  250. data/test/xml/test_cdata.rb +50 -0
  251. data/test/xml/test_comment.rb +29 -0
  252. data/test/xml/test_document.rb +697 -0
  253. data/test/xml/test_document_encoding.rb +26 -0
  254. data/test/xml/test_document_fragment.rb +192 -0
  255. data/test/xml/test_dtd.rb +107 -0
  256. data/test/xml/test_dtd_encoding.rb +33 -0
  257. data/test/xml/test_element_content.rb +56 -0
  258. data/test/xml/test_element_decl.rb +73 -0
  259. data/test/xml/test_entity_decl.rb +122 -0
  260. data/test/xml/test_entity_reference.rb +21 -0
  261. data/test/xml/test_namespace.rb +70 -0
  262. data/test/xml/test_node.rb +917 -0
  263. data/test/xml/test_node_attributes.rb +34 -0
  264. data/test/xml/test_node_encoding.rb +107 -0
  265. data/test/xml/test_node_reparenting.rb +334 -0
  266. data/test/xml/test_node_set.rb +742 -0
  267. data/test/xml/test_parse_options.rb +52 -0
  268. data/test/xml/test_processing_instruction.rb +30 -0
  269. data/test/xml/test_reader_encoding.rb +126 -0
  270. data/test/xml/test_relax_ng.rb +60 -0
  271. data/test/xml/test_schema.rb +94 -0
  272. data/test/xml/test_syntax_error.rb +12 -0
  273. data/test/xml/test_text.rb +47 -0
  274. data/test/xml/test_unparented_node.rb +381 -0
  275. data/test/xml/test_xpath.rb +237 -0
  276. data/test/xslt/test_custom_functions.rb +94 -0
  277. data/test/xslt/test_exception_handling.rb +37 -0
  278. metadata +552 -0
@@ -0,0 +1,472 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module HTML
5
+ class TestDocument < Nokogiri::TestCase
6
+ def setup
7
+ super
8
+ @html = Nokogiri::HTML.parse(File.read(HTML_FILE))
9
+ end
10
+
11
+ def test_nil_css
12
+ # Behavior is undefined but shouldn't break
13
+ assert @html.css(nil)
14
+ assert @html.xpath(nil)
15
+ end
16
+
17
+ def test_exceptions_remove_newlines
18
+ errors = @html.errors
19
+ assert errors.length > 0, 'has errors'
20
+ errors.each do |error|
21
+ assert_equal(error.to_s.chomp, error.to_s)
22
+ end
23
+ end
24
+
25
+ def test_fragment
26
+ fragment = @html.fragment
27
+ assert_equal 0, fragment.children.length
28
+ end
29
+
30
+ def test_document_takes_config_block
31
+ options = nil
32
+ Nokogiri::HTML(File.read(HTML_FILE), HTML_FILE) do |cfg|
33
+ options = cfg
34
+ options.nonet.nowarning.dtdattr
35
+ end
36
+ assert options.nonet?
37
+ assert options.nowarning?
38
+ assert options.dtdattr?
39
+ end
40
+
41
+ def test_parse_takes_config_block
42
+ options = nil
43
+ Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE) do |cfg|
44
+ options = cfg
45
+ options.nonet.nowarning.dtdattr
46
+ end
47
+ assert options.nonet?
48
+ assert options.nowarning?
49
+ assert options.dtdattr?
50
+ end
51
+
52
+ def test_subclass
53
+ klass = Class.new(Nokogiri::HTML::Document)
54
+ doc = klass.new
55
+ assert_instance_of klass, doc
56
+ end
57
+
58
+ def test_subclass_initialize
59
+ klass = Class.new(Nokogiri::HTML::Document) do
60
+ attr_accessor :initialized_with
61
+
62
+ def initialize(*args)
63
+ @initialized_with = args
64
+ end
65
+ end
66
+ doc = klass.new("uri", "external_id", 1)
67
+ assert_equal ["uri", "external_id", 1], doc.initialized_with
68
+ end
69
+
70
+ def test_subclass_dup
71
+ klass = Class.new(Nokogiri::HTML::Document)
72
+ doc = klass.new.dup
73
+ assert_instance_of klass, doc
74
+ end
75
+
76
+ def test_subclass_parse
77
+ klass = Class.new(Nokogiri::HTML::Document)
78
+ doc = klass.parse(File.read(HTML_FILE))
79
+ assert_equal @html.to_s, doc.to_s
80
+ assert_instance_of klass, doc
81
+ end
82
+
83
+ def test_document_parse_method
84
+ html = Nokogiri::HTML::Document.parse(File.read(HTML_FILE))
85
+ assert_equal @html.to_s, html.to_s
86
+ end
87
+
88
+ ###
89
+ # Nokogiri::HTML returns an empty Document when given a blank string GH#11
90
+ def test_empty_string_returns_empty_doc
91
+ doc = Nokogiri::HTML('')
92
+ assert_instance_of Nokogiri::HTML::Document, doc
93
+ assert_nil doc.root
94
+ end
95
+
96
+ unless Nokogiri.uses_libxml? && %w[2 6] === LIBXML_VERSION.split('.')[0..1]
97
+ # FIXME: this is a hack around broken libxml versions
98
+ def test_to_xhtml_with_indent
99
+ doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
100
+ doc = Nokogiri::HTML(doc.to_xhtml(:indent => 2))
101
+ assert_indent 2, doc
102
+ end
103
+
104
+ def test_write_to_xhtml_with_indent
105
+ io = StringIO.new
106
+ doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
107
+ doc.write_xhtml_to io, :indent => 5
108
+ io.rewind
109
+ doc = Nokogiri::HTML(io.read)
110
+ assert_indent 5, doc
111
+ end
112
+ end
113
+
114
+ def test_swap_should_not_exist
115
+ assert_raises(NoMethodError) {
116
+ @html.swap
117
+ }
118
+ end
119
+
120
+ def test_namespace_should_not_exist
121
+ assert_raises(NoMethodError) {
122
+ @html.namespace
123
+ }
124
+ end
125
+
126
+ def test_meta_encoding
127
+ assert_equal 'UTF-8', @html.meta_encoding
128
+
129
+ html = Nokogiri::HTML(<<-eohtml)
130
+ <html>
131
+ <head>
132
+ <meta http-equiv="X-Content-Type" content="text/html; charset=Shift_JIS">
133
+ </head>
134
+ <body>
135
+ foo
136
+ </body>
137
+ </html>
138
+ eohtml
139
+ assert_nil html.meta_encoding
140
+ end
141
+
142
+ def test_meta_encoding=
143
+ @html.meta_encoding = 'EUC-JP'
144
+ assert_equal 'EUC-JP', @html.meta_encoding
145
+ end
146
+
147
+ def test_title
148
+ assert_equal 'Tender Lovemaking ', @html.title
149
+ doc = Nokogiri::HTML('<html><body>foo</body></html>')
150
+ assert_nil doc.title
151
+ end
152
+
153
+ def test_title=()
154
+ doc = Nokogiri::HTML(<<eohtml)
155
+ <html>
156
+ <head>
157
+ <title>old</title>
158
+ </head>
159
+ <body>
160
+ foo
161
+ </body>
162
+ </html>
163
+ eohtml
164
+ doc.title = 'new'
165
+ assert_equal 'new', doc.title
166
+
167
+ doc = Nokogiri::HTML(<<eohtml)
168
+ <html>
169
+ <head>
170
+ </head>
171
+ <body>
172
+ foo
173
+ </body>
174
+ </html>
175
+ eohtml
176
+ doc.title = 'new'
177
+ assert_equal 'new', doc.title
178
+
179
+ doc = Nokogiri::HTML(<<eohtml)
180
+ <html>
181
+ <body>
182
+ foo
183
+ </body>
184
+ </html>
185
+ eohtml
186
+ doc.title = 'new'
187
+ if Nokogiri.uses_libxml?
188
+ assert_nil doc.title
189
+ else
190
+ assert_equal 'new', doc.title
191
+ end
192
+ end
193
+
194
+ def test_meta_encoding_without_head
195
+ html = Nokogiri::HTML('<html><body>foo</body></html>')
196
+ assert_nil html.meta_encoding
197
+
198
+ html.meta_encoding = 'EUC-JP'
199
+ assert_nil html.meta_encoding
200
+ end
201
+
202
+ def test_root_node_parent_is_document
203
+ parent = @html.root.parent
204
+ assert_equal @html, parent
205
+ assert_instance_of Nokogiri::HTML::Document, parent
206
+ end
207
+
208
+ def test_parse_handles_nil_gracefully
209
+ assert_nothing_raised do
210
+ @doc = Nokogiri::HTML::Document.parse(nil)
211
+ end
212
+ assert_instance_of Nokogiri::HTML::Document, @doc
213
+ end
214
+
215
+ def test_parse_empty_document
216
+ doc = Nokogiri::HTML("\n")
217
+ assert_equal 0, doc.css('a').length
218
+ assert_equal 0, doc.xpath('//a').length
219
+ assert_equal 0, doc.search('//a').length
220
+ end
221
+
222
+ def test_HTML_function
223
+ html = Nokogiri::HTML(File.read(HTML_FILE))
224
+ assert html.html?
225
+ end
226
+
227
+ def test_parse_io
228
+ assert File.open(HTML_FILE, 'rb') { |f|
229
+ Document.read_io(f, nil, 'UTF-8',
230
+ XML::ParseOptions::NOERROR | XML::ParseOptions::NOWARNING
231
+ )
232
+ }
233
+ end
234
+
235
+ def test_parse_temp_file
236
+ temp_html_file = Tempfile.new("TEMP_HTML_FILE")
237
+ File.open(HTML_FILE, 'rb') { |f| temp_html_file.write f.read }
238
+ temp_html_file.close
239
+ temp_html_file.open
240
+ assert_equal Nokogiri::HTML.parse(File.read(HTML_FILE)).xpath('//div/a').length,
241
+ Nokogiri::HTML.parse(temp_html_file).xpath('//div/a').length
242
+ end
243
+
244
+ def test_to_xhtml
245
+ assert_match 'XHTML', @html.to_xhtml
246
+ assert_match 'XHTML', @html.to_xhtml(:encoding => 'UTF-8')
247
+ assert_match 'UTF-8', @html.to_xhtml(:encoding => 'UTF-8')
248
+ end
249
+
250
+ def test_no_xml_header
251
+ html = Nokogiri::HTML(<<-eohtml)
252
+ <html>
253
+ </html>
254
+ eohtml
255
+ assert html.to_html.length > 0, 'html length is too short'
256
+ assert_no_match(/^<\?xml/, html.to_html)
257
+ end
258
+
259
+ def test_document_has_error
260
+ html = Nokogiri::HTML(<<-eohtml)
261
+ <html>
262
+ <body>
263
+ <div awesome="asdf>
264
+ <p>inside div tag</p>
265
+ </div>
266
+ <p>outside div tag</p>
267
+ </body>
268
+ </html>
269
+ eohtml
270
+ assert html.errors.length > 0
271
+ end
272
+
273
+ def test_relative_css
274
+ html = Nokogiri::HTML(<<-eohtml)
275
+ <html>
276
+ <body>
277
+ <div>
278
+ <p>inside div tag</p>
279
+ </div>
280
+ <p>outside div tag</p>
281
+ </body>
282
+ </html>
283
+ eohtml
284
+ set = html.search('div').search('p')
285
+ assert_equal(1, set.length)
286
+ assert_equal('inside div tag', set.first.inner_text)
287
+ end
288
+
289
+ def test_multi_css
290
+ html = Nokogiri::HTML(<<-eohtml)
291
+ <html>
292
+ <body>
293
+ <div>
294
+ <p>p tag</p>
295
+ <a>a tag</a>
296
+ </div>
297
+ </body>
298
+ </html>
299
+ eohtml
300
+ set = html.css('p, a')
301
+ assert_equal(2, set.length)
302
+ assert_equal ['a tag', 'p tag'].sort, set.map { |x| x.content }.sort
303
+ end
304
+
305
+ def test_inner_text
306
+ html = Nokogiri::HTML(<<-eohtml)
307
+ <html>
308
+ <body>
309
+ <div>
310
+ <p>
311
+ Hello world!
312
+ </p>
313
+ </div>
314
+ </body>
315
+ </html>
316
+ eohtml
317
+ node = html.xpath('//div').first
318
+ assert_equal('Hello world!', node.inner_text.strip)
319
+ end
320
+
321
+ def test_find_by_xpath
322
+ found = @html.xpath('//div/a')
323
+ assert_equal 3, found.length
324
+ end
325
+
326
+ def test_find_by_css
327
+ found = @html.css('div > a')
328
+ assert_equal 3, found.length
329
+ end
330
+
331
+ def test_find_by_css_with_square_brackets
332
+ found = @html.css("div[@id='header'] > h1")
333
+ found = @html.css("div[@id='header'] h1") # this blows up on commit 6fa0f6d329d9dbf1cc21c0ac72f7e627bb4c05fc
334
+ assert_equal 1, found.length
335
+ end
336
+
337
+ def test_find_with_function
338
+ assert @html.css("div:awesome() h1", Class.new {
339
+ def awesome divs
340
+ [divs.first]
341
+ end
342
+ }.new)
343
+ end
344
+
345
+ def test_dup_shallow
346
+ found = @html.search('//div/a').first
347
+ dup = found.dup(0)
348
+ assert dup
349
+ assert_equal '', dup.content
350
+ end
351
+
352
+ def test_search_can_handle_xpath_and_css
353
+ found = @html.search('//div/a', 'div > p')
354
+ length = @html.xpath('//div/a').length +
355
+ @html.css('div > p').length
356
+ assert_equal length, found.length
357
+ end
358
+
359
+ def test_dup_document
360
+ assert dup = @html.dup
361
+ assert_not_equal dup, @html
362
+ assert @html.html?
363
+ assert_instance_of Nokogiri::HTML::Document, dup
364
+ assert dup.html?, 'duplicate should be html'
365
+ assert_equal @html.to_s, dup.to_s
366
+ end
367
+
368
+ def test_dup_document_shallow
369
+ assert dup = @html.dup(0)
370
+ assert_not_equal dup, @html
371
+ end
372
+
373
+ def test_dup
374
+ found = @html.search('//div/a').first
375
+ dup = found.dup
376
+ assert dup
377
+ assert_equal found.content, dup.content
378
+ assert_equal found.document, dup.document
379
+ end
380
+
381
+ def test_inner_html
382
+ html = Nokogiri::HTML(<<-eohtml)
383
+ <html>
384
+ <body>
385
+ <div>
386
+ <p>
387
+ Hello world!
388
+ </p>
389
+ </div>
390
+ </body>
391
+ </html>
392
+ eohtml
393
+ node = html.xpath('//div').first
394
+ assert_equal('<p>Helloworld!</p>', node.inner_html.gsub(/\s/, ''))
395
+ end
396
+
397
+ def test_round_trip
398
+ doc = Nokogiri::HTML(@html.inner_html)
399
+ assert_equal @html.root.to_html, doc.root.to_html
400
+ end
401
+
402
+ def test_fragment_contains_text_node
403
+ fragment = Nokogiri::HTML.fragment('fooo')
404
+ assert_equal 1, fragment.children.length
405
+ assert_equal 'fooo', fragment.inner_text
406
+ end
407
+
408
+ def test_fragment_includes_two_tags
409
+ assert_equal 2, Nokogiri::HTML.fragment("<br/><hr/>").children.length
410
+ end
411
+
412
+ def test_relative_css_finder
413
+ doc = Nokogiri::HTML(<<-eohtml)
414
+ <html>
415
+ <body>
416
+ <div class="red">
417
+ <p>
418
+ inside red
419
+ </p>
420
+ </div>
421
+ <div class="green">
422
+ <p>
423
+ inside green
424
+ </p>
425
+ </div>
426
+ </body>
427
+ </html>
428
+ eohtml
429
+ red_divs = doc.css('div.red')
430
+ assert_equal 1, red_divs.length
431
+ p_tags = red_divs.first.css('p')
432
+ assert_equal 1, p_tags.length
433
+ assert_equal 'inside red', p_tags.first.text.strip
434
+ end
435
+
436
+ def test_find_classes
437
+ doc = Nokogiri::HTML(<<-eohtml)
438
+ <html>
439
+ <body>
440
+ <p class="red">RED</p>
441
+ <p class="awesome red">RED</p>
442
+ <p class="notred">GREEN</p>
443
+ <p class="green notred">GREEN</p>
444
+ </body>
445
+ </html>
446
+ eohtml
447
+ list = doc.css('.red')
448
+ assert_equal 2, list.length
449
+ assert_equal %w{ RED RED }, list.map { |x| x.text }
450
+ end
451
+
452
+ def test_parse_can_take_io
453
+ html = nil
454
+ File.open(HTML_FILE, 'rb') { |f|
455
+ html = Nokogiri::HTML(f)
456
+ }
457
+ assert html.html?
458
+ end
459
+
460
+ def test_html?
461
+ assert !@html.xml?
462
+ assert @html.html?
463
+ end
464
+
465
+ def test_serialize
466
+ assert @html.serialize
467
+ assert @html.to_html
468
+ end
469
+ end
470
+ end
471
+ end
472
+