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 +1,2 @@
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <bar />
@@ -0,0 +1,136 @@
1
+ #Process.setrlimit(Process::RLIMIT_CORE, Process::RLIM_INFINITY) unless RUBY_PLATFORM =~ /(java|mswin|mingw)/i
2
+ $VERBOSE = true
3
+ require 'rubygems'
4
+ require 'test/unit'
5
+ require 'fileutils'
6
+ require 'tempfile'
7
+ require 'pp'
8
+
9
+ %w(../lib ../ext).each do |path|
10
+ $LOAD_PATH.unshift(File.expand_path(File.join(File.dirname(__FILE__), path)))
11
+ end
12
+
13
+ require 'nokogiri'
14
+
15
+ warn "#{__FILE__}:#{__LINE__}: libxml version info: #{Nokogiri::VERSION_INFO.inspect}"
16
+
17
+ module Nokogiri
18
+ class TestCase < Test::Unit::TestCase
19
+ ASSETS_DIR = File.expand_path File.join(File.dirname(__FILE__), 'files')
20
+ XML_FILE = File.join(ASSETS_DIR, 'staff.xml')
21
+ XSLT_FILE = File.join(ASSETS_DIR, 'staff.xslt')
22
+ EXSLT_FILE = File.join(ASSETS_DIR, 'exslt.xslt')
23
+ EXML_FILE = File.join(ASSETS_DIR, 'exslt.xml')
24
+ HTML_FILE = File.join(ASSETS_DIR, 'tlm.html')
25
+ NICH_FILE = File.join(ASSETS_DIR, '2ch.html')
26
+ SHIFT_JIS_XML = File.join(ASSETS_DIR, 'shift_jis.xml')
27
+ SHIFT_JIS_HTML = File.join(ASSETS_DIR, 'shift_jis.html')
28
+ PO_XML_FILE = File.join(ASSETS_DIR, 'po.xml')
29
+ PO_SCHEMA_FILE = File.join(ASSETS_DIR, 'po.xsd')
30
+ ADDRESS_SCHEMA_FILE = File.join(ASSETS_DIR, 'address_book.rlx')
31
+ ADDRESS_XML_FILE = File.join(ASSETS_DIR, 'address_book.xml')
32
+ SNUGGLES_FILE = File.join(ASSETS_DIR, 'snuggles.xml')
33
+
34
+ unless RUBY_VERSION >= '1.9'
35
+ undef :default_test
36
+ end
37
+
38
+ def setup
39
+ warn "#{name}" if ENV['TESTOPTS'] == '-v'
40
+ end
41
+
42
+ def teardown
43
+ if ENV['NOKOGIRI_GC']
44
+ STDOUT.putc '!'
45
+ GC.start
46
+ end
47
+ end
48
+
49
+ def assert_indent amount, doc, message = nil
50
+ nodes = []
51
+ doc.traverse do |node|
52
+ nodes << node if node.text? && node.blank?
53
+ end
54
+ assert nodes.length > 0
55
+ nodes.each do |node|
56
+ len = node.content.gsub(/[\r\n]/, '').length
57
+ assert_equal(0, len % amount, message)
58
+ end
59
+ end
60
+ end
61
+
62
+ module SAX
63
+ class TestCase < Nokogiri::TestCase
64
+ class Doc < XML::SAX::Document
65
+ attr_reader :start_elements, :start_document_called
66
+ attr_reader :end_elements, :end_document_called
67
+ attr_reader :data, :comments, :cdata_blocks, :start_elements_namespace
68
+ attr_reader :errors, :warnings, :end_elements_namespace
69
+ attr_reader :xmldecls
70
+
71
+ def xmldecl version, encoding, standalone
72
+ @xmldecls = [version, encoding, standalone].compact
73
+ super
74
+ end
75
+
76
+ def start_document
77
+ @start_document_called = true
78
+ super
79
+ end
80
+
81
+ def end_document
82
+ @end_document_called = true
83
+ super
84
+ end
85
+
86
+ def error error
87
+ (@errors ||= []) << error
88
+ super
89
+ end
90
+
91
+ def warning warning
92
+ (@warning ||= []) << warning
93
+ super
94
+ end
95
+
96
+ def start_element *args
97
+ (@start_elements ||= []) << args
98
+ super
99
+ end
100
+
101
+ def start_element_namespace *args
102
+ (@start_elements_namespace ||= []) << args
103
+ super
104
+ end
105
+
106
+ def end_element *args
107
+ (@end_elements ||= []) << args
108
+ super
109
+ end
110
+
111
+ def end_element_namespace *args
112
+ (@end_elements_namespace ||= []) << args
113
+ super
114
+ end
115
+
116
+ def characters string
117
+ @data ||= []
118
+ @data += [string]
119
+ super
120
+ end
121
+
122
+ def comment string
123
+ @comments ||= []
124
+ @comments += [string]
125
+ super
126
+ end
127
+
128
+ def cdata_block string
129
+ @cdata_blocks ||= []
130
+ @cdata_blocks += [string]
131
+ super
132
+ end
133
+ end
134
+ end
135
+ end
136
+ end
@@ -0,0 +1,64 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module HTML
5
+ module SAX
6
+ class TestParser < Nokogiri::SAX::TestCase
7
+ def setup
8
+ super
9
+ @parser = HTML::SAX::Parser.new(Doc.new)
10
+ end
11
+
12
+ def test_parse_empty_document
13
+ # This caused a segfault in libxml 2.6.x
14
+ assert_nothing_raised { @parser.parse '' }
15
+ end
16
+
17
+ def test_parse_empty_file
18
+ # Make sure empty files don't break stuff
19
+ empty_file_name = File.join(Dir.tmpdir, 'bogus.xml')
20
+ FileUtils.touch empty_file_name
21
+ assert_nothing_raised { @parser.parse_file empty_file_name }
22
+ end
23
+
24
+ def test_parse_file
25
+ @parser.parse_file(HTML_FILE)
26
+ assert_equal 1110, @parser.document.end_elements.length
27
+ end
28
+
29
+ def test_parse_file_nil_argument
30
+ assert_raises(ArgumentError) {
31
+ @parser.parse_file(nil)
32
+ }
33
+ end
34
+
35
+ def test_parse_file_non_existant
36
+ assert_raise Errno::ENOENT do
37
+ @parser.parse_file('foo')
38
+ end
39
+ end
40
+
41
+ def test_parse_file_with_dir
42
+ assert_raise Errno::EISDIR do
43
+ @parser.parse_file(File.dirname(__FILE__))
44
+ end
45
+ end
46
+
47
+ def test_parse_memory_nil
48
+ assert_raise ArgumentError do
49
+ @parser.parse_memory(nil)
50
+ end
51
+ end
52
+
53
+ def test_parse_document
54
+ @parser.parse_memory(<<-eoxml)
55
+ <p>Paragraph 1</p>
56
+ <p>Paragraph 2</p>
57
+ eoxml
58
+ assert_equal([["html", []], ["body", []], ["p", []], ["p", []]],
59
+ @parser.document.start_elements)
60
+ end
61
+ end
62
+ end
63
+ end
64
+ end
@@ -0,0 +1,48 @@
1
+ # -*- coding: utf-8 -*-
2
+
3
+ require "helper"
4
+
5
+ module Nokogiri
6
+ module HTML
7
+ module SAX
8
+ class TestParserContext < Nokogiri::SAX::TestCase
9
+ def test_from_io
10
+ assert_nothing_raised do
11
+ ctx = ParserContext.new StringIO.new('fo'), 'UTF-8'
12
+ end
13
+ end
14
+
15
+ def test_from_string
16
+ assert_nothing_raised do
17
+ ctx = ParserContext.new 'blah blah'
18
+ end
19
+ end
20
+
21
+ def test_parse_with
22
+ ctx = ParserContext.new 'blah'
23
+ assert_raises ArgumentError do
24
+ ctx.parse_with nil
25
+ end
26
+ end
27
+
28
+ def test_parse_with_sax_parser
29
+ assert_nothing_raised do
30
+ xml = "<root />"
31
+ ctx = ParserContext.new xml
32
+ parser = Parser.new Doc.new
33
+ ctx.parse_with parser
34
+ end
35
+ end
36
+
37
+ def test_from_file
38
+ assert_nothing_raised do
39
+ ctx = ParserContext.file HTML_FILE, 'UTF-8'
40
+ parser = Parser.new Doc.new
41
+ ctx.parse_with parser
42
+ end
43
+ end
44
+ end
45
+ end
46
+ end
47
+ end
48
+
@@ -0,0 +1,164 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module HTML
5
+ class TestBuilder < Nokogiri::TestCase
6
+ def test_top_level_function_builds
7
+ foo = nil
8
+ Nokogiri() { |xml| foo = xml }
9
+ assert_instance_of Nokogiri::HTML::Builder, foo
10
+ end
11
+
12
+ def test_builder_with_explicit_tags
13
+ html_doc = Nokogiri::HTML::Builder.new {
14
+ div.slide(:class => 'another_class') {
15
+ node = Nokogiri::XML::Node.new("id", doc)
16
+ node.content = "hello"
17
+ insert(node)
18
+ }
19
+ }.doc
20
+ assert_equal 1, html_doc.css('div.slide > id').length
21
+ assert_equal 'hello', html_doc.at('div.slide > id').content
22
+ end
23
+
24
+ def test_hash_as_attributes_for_attribute_method
25
+ html = Nokogiri::HTML::Builder.new { ||
26
+ div.slide(:class => 'another_class') {
27
+ span 'Slide 1'
28
+ }
29
+ }.to_html
30
+ assert_match 'class="slide another_class"', html
31
+ end
32
+
33
+ def test_hash_as_attributes
34
+ builder = Nokogiri::HTML::Builder.new do
35
+ div(:id => 'awesome') {
36
+ h1 "america"
37
+ }
38
+ end
39
+ assert_equal('<div id="awesome"><h1>america</h1></div>',
40
+ builder.doc.root.to_html.gsub(/\n/, '').gsub(/>\s*</, '><'))
41
+ end
42
+
43
+ def test_href_with_attributes
44
+ uri = 'http://tenderlovemaking.com/'
45
+ built = Nokogiri::XML::Builder.new {
46
+ div {
47
+ a('King Khan & The Shrines', :href => uri)
48
+ }
49
+ }
50
+ assert_equal 'http://tenderlovemaking.com/',
51
+ built.doc.at('a')[:href]
52
+ end
53
+
54
+ def test_tag_nesting
55
+ builder = Nokogiri::HTML::Builder.new do
56
+ body {
57
+ span.left ''
58
+ span.middle {
59
+ div.icon ''
60
+ }
61
+ span.right ''
62
+ }
63
+ end
64
+ assert node = builder.doc.css('span.right').first
65
+ assert_equal 'middle', node.previous_sibling['class']
66
+ end
67
+
68
+ def test_has_ampersand
69
+ builder = Nokogiri::HTML::Builder.new do
70
+ div.rad.thing! {
71
+ text "<awe&some>"
72
+ b "hello & world"
73
+ }
74
+ end
75
+ assert_equal(
76
+ '<div class="rad" id="thing">&lt;awe&amp;some&gt;<b>hello &amp; world</b></div>',
77
+ builder.doc.root.to_html.gsub(/\n/, ''))
78
+ end
79
+
80
+ def test_multi_tags
81
+ builder = Nokogiri::HTML::Builder.new do
82
+ div.rad.thing! {
83
+ text "<awesome>"
84
+ b "hello"
85
+ }
86
+ end
87
+ assert_equal(
88
+ '<div class="rad" id="thing">&lt;awesome&gt;<b>hello</b></div>',
89
+ builder.doc.root.to_html.gsub(/\n/, ''))
90
+ end
91
+
92
+ def test_attributes_plus_block
93
+ builder = Nokogiri::HTML::Builder.new do
94
+ div.rad.thing! {
95
+ text "<awesome>"
96
+ }
97
+ end
98
+ assert_equal('<div class="rad" id="thing">&lt;awesome&gt;</div>',
99
+ builder.doc.root.to_html.chomp)
100
+ end
101
+
102
+ def test_builder_adds_attributes
103
+ builder = Nokogiri::HTML::Builder.new do
104
+ div.rad.thing! "tender div"
105
+ end
106
+ assert_equal('<div class="rad" id="thing">tender div</div>',
107
+ builder.doc.root.to_html.chomp)
108
+ end
109
+
110
+ def test_bold_tag
111
+ builder = Nokogiri::HTML::Builder.new do
112
+ b "bold tag"
113
+ end
114
+ assert_equal('<b>bold tag</b>', builder.doc.root.to_html.chomp)
115
+ end
116
+
117
+ def test_html_then_body_tag
118
+ builder = Nokogiri::HTML::Builder.new do
119
+ html {
120
+ body {
121
+ b "bold tag"
122
+ }
123
+ }
124
+ end
125
+ assert_equal('<html><body><b>bold tag</b></body></html>',
126
+ builder.doc.root.to_html.chomp.gsub(/>\s*</, '><'))
127
+ end
128
+
129
+ def test_instance_eval_with_delegation_to_block_context
130
+ class << self
131
+ def foo
132
+ "foo!"
133
+ end
134
+ end
135
+
136
+ builder = Nokogiri::HTML::Builder.new { text foo }
137
+ assert builder.to_html.include?("foo!")
138
+ end
139
+
140
+ def test_builder_with_param
141
+ doc = Nokogiri::HTML::Builder.new { |html|
142
+ html.body {
143
+ html.p "hello world"
144
+ }
145
+ }.doc
146
+
147
+ assert node = doc.xpath('//body/p').first
148
+ assert_equal 'hello world', node.content
149
+ end
150
+
151
+ def test_builder_with_id
152
+ text = "hello world"
153
+ doc = Nokogiri::HTML::Builder.new { |html|
154
+ html.body {
155
+ html.id_ text
156
+ }
157
+ }.doc
158
+
159
+ assert node = doc.xpath('//body/id').first
160
+ assert_equal text, node.content
161
+ end
162
+ end
163
+ end
164
+ end
@@ -0,0 +1,390 @@
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_fragment
12
+ fragment = @html.fragment
13
+ assert_equal 0, fragment.children.length
14
+ end
15
+
16
+ def test_document_takes_config_block
17
+ options = nil
18
+ Nokogiri::HTML(File.read(HTML_FILE), HTML_FILE) do |cfg|
19
+ options = cfg
20
+ options.nonet.nowarning.dtdattr
21
+ end
22
+ assert options.nonet?
23
+ assert options.nowarning?
24
+ assert options.dtdattr?
25
+ end
26
+
27
+ def test_parse_takes_config_block
28
+ options = nil
29
+ Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE) do |cfg|
30
+ options = cfg
31
+ options.nonet.nowarning.dtdattr
32
+ end
33
+ assert options.nonet?
34
+ assert options.nowarning?
35
+ assert options.dtdattr?
36
+ end
37
+
38
+ def test_subclass
39
+ klass = Class.new(Nokogiri::HTML::Document)
40
+ doc = klass.new
41
+ assert_instance_of klass, doc
42
+ end
43
+
44
+ def test_subclass_initialize
45
+ klass = Class.new(Nokogiri::HTML::Document) do
46
+ attr_accessor :initialized_with
47
+
48
+ def initialize(*args)
49
+ @initialized_with = args
50
+ end
51
+ end
52
+ doc = klass.new("uri", "external_id", 1)
53
+ assert_equal ["uri", "external_id", 1], doc.initialized_with
54
+ end
55
+
56
+ def test_subclass_dup
57
+ klass = Class.new(Nokogiri::HTML::Document)
58
+ doc = klass.new.dup
59
+ assert_instance_of klass, doc
60
+ end
61
+
62
+ def test_subclass_parse
63
+ klass = Class.new(Nokogiri::HTML::Document)
64
+ doc = klass.parse(File.read(HTML_FILE))
65
+ assert_equal @html.to_s, doc.to_s
66
+ assert_instance_of klass, doc
67
+ end
68
+
69
+ def test_document_parse_method
70
+ html = Nokogiri::HTML::Document.parse(File.read(HTML_FILE))
71
+ assert_equal @html.to_s, html.to_s
72
+ end
73
+
74
+ ###
75
+ # Nokogiri::HTML returns an empty Document when given a blank string GH#11
76
+ def test_empty_string_returns_empty_doc
77
+ doc = Nokogiri::HTML('')
78
+ assert_instance_of Nokogiri::HTML::Document, doc
79
+ assert_nil doc.root
80
+ end
81
+
82
+ unless %w[2 6] === LIBXML_VERSION.split('.')[0..1]
83
+ # FIXME: this is a hack around broken libxml versions
84
+ def test_to_xhtml_with_indent
85
+ doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
86
+ doc = Nokogiri::HTML(doc.to_xhtml(:indent => 2))
87
+ assert_indent 2, doc
88
+ end
89
+
90
+ def test_write_to_xhtml_with_indent
91
+ io = StringIO.new
92
+ doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
93
+ doc.write_xhtml_to io, :indent => 5
94
+ io.rewind
95
+ doc = Nokogiri::HTML(io.read)
96
+ assert_indent 5, doc
97
+ end
98
+ end
99
+
100
+ def test_swap_should_not_exist
101
+ assert_raises(NoMethodError) {
102
+ @html.swap
103
+ }
104
+ end
105
+
106
+ def test_namespace_should_not_exist
107
+ assert_raises(NoMethodError) {
108
+ @html.namespace
109
+ }
110
+ end
111
+
112
+ def test_meta_encoding
113
+ assert_equal 'UTF-8', @html.meta_encoding
114
+ end
115
+
116
+ def test_meta_encoding=
117
+ @html.meta_encoding = 'EUC-JP'
118
+ assert_equal 'EUC-JP', @html.meta_encoding
119
+ end
120
+
121
+ def test_meta_encoding_without_head
122
+ html = Nokogiri::HTML('<html><body>foo</body></html>')
123
+ assert_nil html.meta_encoding
124
+
125
+ html.meta_encoding = 'EUC-JP'
126
+ assert_nil html.meta_encoding
127
+ end
128
+
129
+ def test_root_node_parent_is_document
130
+ parent = @html.root.parent
131
+ assert_equal @html, parent
132
+ assert_instance_of Nokogiri::HTML::Document, parent
133
+ end
134
+
135
+ def test_parse_handles_nil_gracefully
136
+ assert_nothing_raised do
137
+ @doc = Nokogiri::HTML::Document.parse(nil)
138
+ end
139
+ assert_instance_of Nokogiri::HTML::Document, @doc
140
+ end
141
+
142
+ def test_parse_empty_document
143
+ doc = Nokogiri::HTML("\n")
144
+ assert_equal 0, doc.css('a').length
145
+ assert_equal 0, doc.xpath('//a').length
146
+ assert_equal 0, doc.search('//a').length
147
+ end
148
+
149
+ def test_HTML_function
150
+ html = Nokogiri::HTML(File.read(HTML_FILE))
151
+ assert html.html?
152
+ end
153
+
154
+ def test_parse_io
155
+ assert doc = File.open(HTML_FILE, 'rb') { |f|
156
+ Document.read_io(f, nil, 'UTF-8',
157
+ XML::ParseOptions::NOERROR | XML::ParseOptions::NOWARNING
158
+ )
159
+ }
160
+ end
161
+
162
+ def test_to_xhtml
163
+ assert_match 'XHTML', @html.to_xhtml
164
+ assert_match 'XHTML', @html.to_xhtml(:encoding => 'UTF-8')
165
+ assert_match 'UTF-8', @html.to_xhtml(:encoding => 'UTF-8')
166
+ end
167
+
168
+ def test_no_xml_header
169
+ html = Nokogiri::HTML(<<-eohtml)
170
+ <html>
171
+ </html>
172
+ eohtml
173
+ assert html.to_html.length > 0, 'html length is too short'
174
+ assert_no_match(/^<\?xml/, html.to_html)
175
+ end
176
+
177
+ def test_document_has_error
178
+ html = Nokogiri::HTML(<<-eohtml)
179
+ <html>
180
+ <body>
181
+ <div awesome="asdf>
182
+ <p>inside div tag</p>
183
+ </div>
184
+ <p>outside div tag</p>
185
+ </body>
186
+ </html>
187
+ eohtml
188
+ assert html.errors.length > 0
189
+ end
190
+
191
+ def test_relative_css
192
+ html = Nokogiri::HTML(<<-eohtml)
193
+ <html>
194
+ <body>
195
+ <div>
196
+ <p>inside div tag</p>
197
+ </div>
198
+ <p>outside div tag</p>
199
+ </body>
200
+ </html>
201
+ eohtml
202
+ set = html.search('div').search('p')
203
+ assert_equal(1, set.length)
204
+ assert_equal('inside div tag', set.first.inner_text)
205
+ end
206
+
207
+ def test_multi_css
208
+ html = Nokogiri::HTML(<<-eohtml)
209
+ <html>
210
+ <body>
211
+ <div>
212
+ <p>p tag</p>
213
+ <a>a tag</a>
214
+ </div>
215
+ </body>
216
+ </html>
217
+ eohtml
218
+ set = html.css('p, a')
219
+ assert_equal(2, set.length)
220
+ assert_equal ['a tag', 'p tag'].sort, set.map { |x| x.content }.sort
221
+ end
222
+
223
+ def test_inner_text
224
+ html = Nokogiri::HTML(<<-eohtml)
225
+ <html>
226
+ <body>
227
+ <div>
228
+ <p>
229
+ Hello world!
230
+ </p>
231
+ </div>
232
+ </body>
233
+ </html>
234
+ eohtml
235
+ node = html.xpath('//div').first
236
+ assert_equal('Hello world!', node.inner_text.strip)
237
+ end
238
+
239
+ def test_find_by_xpath
240
+ found = @html.xpath('//div/a')
241
+ assert_equal 3, found.length
242
+ end
243
+
244
+ def test_find_by_css
245
+ found = @html.css('div > a')
246
+ assert_equal 3, found.length
247
+ end
248
+
249
+ def test_find_by_css_with_square_brackets
250
+ found = @html.css("div[@id='header'] > h1")
251
+ found = @html.css("div[@id='header'] h1") # this blows up on commit 6fa0f6d329d9dbf1cc21c0ac72f7e627bb4c05fc
252
+ assert_equal 1, found.length
253
+ end
254
+
255
+ def test_find_with_function
256
+ found = @html.css("div:awesome() h1", Class.new {
257
+ def awesome divs
258
+ [divs.first]
259
+ end
260
+ }.new)
261
+ end
262
+
263
+ def test_dup_shallow
264
+ found = @html.search('//div/a').first
265
+ dup = found.dup(0)
266
+ assert dup
267
+ assert_equal '', dup.content
268
+ end
269
+
270
+ def test_search_can_handle_xpath_and_css
271
+ found = @html.search('//div/a', 'div > p')
272
+ length = @html.xpath('//div/a').length +
273
+ @html.css('div > p').length
274
+ assert_equal length, found.length
275
+ end
276
+
277
+ def test_dup_document
278
+ assert dup = @html.dup
279
+ assert_not_equal dup, @html
280
+ assert @html.html?
281
+ assert_instance_of Nokogiri::HTML::Document, dup
282
+ assert dup.html?, 'duplicate should be html'
283
+ assert_equal @html.to_s, dup.to_s
284
+ end
285
+
286
+ def test_dup_document_shallow
287
+ assert dup = @html.dup(0)
288
+ assert_not_equal dup, @html
289
+ end
290
+
291
+ def test_dup
292
+ found = @html.search('//div/a').first
293
+ dup = found.dup
294
+ assert dup
295
+ assert_equal found.content, dup.content
296
+ assert_equal found.document, dup.document
297
+ end
298
+
299
+ def test_inner_html
300
+ html = Nokogiri::HTML(<<-eohtml)
301
+ <html>
302
+ <body>
303
+ <div>
304
+ <p>
305
+ Hello world!
306
+ </p>
307
+ </div>
308
+ </body>
309
+ </html>
310
+ eohtml
311
+ node = html.xpath('//div').first
312
+ assert_equal('<p>Helloworld!</p>', node.inner_html.gsub(/\s/, ''))
313
+ end
314
+
315
+ def test_round_trip
316
+ doc = Nokogiri::HTML(@html.inner_html)
317
+ assert_equal @html.root.to_html, doc.root.to_html
318
+ end
319
+
320
+ def test_fragment_contains_text_node
321
+ fragment = Nokogiri::HTML.fragment('fooo')
322
+ assert_equal 1, fragment.children.length
323
+ assert_equal 'fooo', fragment.inner_text
324
+ end
325
+
326
+ def test_fragment_includes_two_tags
327
+ assert_equal 2, Nokogiri::HTML.fragment("<br/><hr/>").children.length
328
+ end
329
+
330
+ def test_relative_css_finder
331
+ doc = Nokogiri::HTML(<<-eohtml)
332
+ <html>
333
+ <body>
334
+ <div class="red">
335
+ <p>
336
+ inside red
337
+ </p>
338
+ </div>
339
+ <div class="green">
340
+ <p>
341
+ inside green
342
+ </p>
343
+ </div>
344
+ </body>
345
+ </html>
346
+ eohtml
347
+ red_divs = doc.css('div.red')
348
+ assert_equal 1, red_divs.length
349
+ p_tags = red_divs.first.css('p')
350
+ assert_equal 1, p_tags.length
351
+ assert_equal 'inside red', p_tags.first.text.strip
352
+ end
353
+
354
+ def test_find_classes
355
+ doc = Nokogiri::HTML(<<-eohtml)
356
+ <html>
357
+ <body>
358
+ <p class="red">RED</p>
359
+ <p class="awesome red">RED</p>
360
+ <p class="notred">GREEN</p>
361
+ <p class="green notred">GREEN</p>
362
+ </body>
363
+ </html>
364
+ eohtml
365
+ list = doc.css('.red')
366
+ assert_equal 2, list.length
367
+ assert_equal %w{ RED RED }, list.map { |x| x.text }
368
+ end
369
+
370
+ def test_parse_can_take_io
371
+ html = nil
372
+ File.open(HTML_FILE, 'rb') { |f|
373
+ html = Nokogiri::HTML(f)
374
+ }
375
+ assert html.html?
376
+ end
377
+
378
+ def test_html?
379
+ assert !@html.xml?
380
+ assert @html.html?
381
+ end
382
+
383
+ def test_serialize
384
+ assert @html.serialize
385
+ assert @html.to_html
386
+ end
387
+ end
388
+ end
389
+ end
390
+