glebm-nokogiri 1.4.2.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (277) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +411 -0
  3. data/CHANGELOG.rdoc +397 -0
  4. data/Manifest.txt +276 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +132 -0
  7. data/Rakefile +183 -0
  8. data/bin/nokogiri +49 -0
  9. data/deps.rip +5 -0
  10. data/ext/nokogiri/extconf.rb +97 -0
  11. data/ext/nokogiri/html_document.c +154 -0
  12. data/ext/nokogiri/html_document.h +10 -0
  13. data/ext/nokogiri/html_element_description.c +276 -0
  14. data/ext/nokogiri/html_element_description.h +10 -0
  15. data/ext/nokogiri/html_entity_lookup.c +32 -0
  16. data/ext/nokogiri/html_entity_lookup.h +8 -0
  17. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  18. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  19. data/ext/nokogiri/nokogiri.c +95 -0
  20. data/ext/nokogiri/nokogiri.h +153 -0
  21. data/ext/nokogiri/xml_attr.c +94 -0
  22. data/ext/nokogiri/xml_attr.h +9 -0
  23. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  24. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  25. data/ext/nokogiri/xml_cdata.c +56 -0
  26. data/ext/nokogiri/xml_cdata.h +9 -0
  27. data/ext/nokogiri/xml_comment.c +54 -0
  28. data/ext/nokogiri/xml_comment.h +9 -0
  29. data/ext/nokogiri/xml_document.c +464 -0
  30. data/ext/nokogiri/xml_document.h +23 -0
  31. data/ext/nokogiri/xml_document_fragment.c +48 -0
  32. data/ext/nokogiri/xml_document_fragment.h +10 -0
  33. data/ext/nokogiri/xml_dtd.c +202 -0
  34. data/ext/nokogiri/xml_dtd.h +10 -0
  35. data/ext/nokogiri/xml_element_content.c +123 -0
  36. data/ext/nokogiri/xml_element_content.h +10 -0
  37. data/ext/nokogiri/xml_element_decl.c +69 -0
  38. data/ext/nokogiri/xml_element_decl.h +9 -0
  39. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  40. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  41. data/ext/nokogiri/xml_entity_decl.c +110 -0
  42. data/ext/nokogiri/xml_entity_decl.h +10 -0
  43. data/ext/nokogiri/xml_entity_reference.c +52 -0
  44. data/ext/nokogiri/xml_entity_reference.h +9 -0
  45. data/ext/nokogiri/xml_io.c +31 -0
  46. data/ext/nokogiri/xml_io.h +11 -0
  47. data/ext/nokogiri/xml_namespace.c +84 -0
  48. data/ext/nokogiri/xml_namespace.h +13 -0
  49. data/ext/nokogiri/xml_node.c +1347 -0
  50. data/ext/nokogiri/xml_node.h +13 -0
  51. data/ext/nokogiri/xml_node_set.c +418 -0
  52. data/ext/nokogiri/xml_node_set.h +9 -0
  53. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  54. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  55. data/ext/nokogiri/xml_reader.c +665 -0
  56. data/ext/nokogiri/xml_reader.h +10 -0
  57. data/ext/nokogiri/xml_relax_ng.c +168 -0
  58. data/ext/nokogiri/xml_relax_ng.h +9 -0
  59. data/ext/nokogiri/xml_sax_parser.c +286 -0
  60. data/ext/nokogiri/xml_sax_parser.h +39 -0
  61. data/ext/nokogiri/xml_sax_parser_context.c +159 -0
  62. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  63. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  64. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  65. data/ext/nokogiri/xml_schema.c +205 -0
  66. data/ext/nokogiri/xml_schema.h +9 -0
  67. data/ext/nokogiri/xml_syntax_error.c +58 -0
  68. data/ext/nokogiri/xml_syntax_error.h +13 -0
  69. data/ext/nokogiri/xml_text.c +50 -0
  70. data/ext/nokogiri/xml_text.h +9 -0
  71. data/ext/nokogiri/xml_xpath_context.c +276 -0
  72. data/ext/nokogiri/xml_xpath_context.h +9 -0
  73. data/ext/nokogiri/xslt_stylesheet.c +142 -0
  74. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  75. data/lib/nokogiri.rb +133 -0
  76. data/lib/nokogiri/css.rb +25 -0
  77. data/lib/nokogiri/css/generated_parser.rb +669 -0
  78. data/lib/nokogiri/css/generated_tokenizer.rb +145 -0
  79. data/lib/nokogiri/css/node.rb +99 -0
  80. data/lib/nokogiri/css/parser.rb +82 -0
  81. data/lib/nokogiri/css/parser.y +232 -0
  82. data/lib/nokogiri/css/syntax_error.rb +7 -0
  83. data/lib/nokogiri/css/tokenizer.rb +7 -0
  84. data/lib/nokogiri/css/tokenizer.rex +55 -0
  85. data/lib/nokogiri/css/xpath_visitor.rb +169 -0
  86. data/lib/nokogiri/decorators/slop.rb +33 -0
  87. data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
  88. data/lib/nokogiri/ffi/html/document.rb +28 -0
  89. data/lib/nokogiri/ffi/html/element_description.rb +81 -0
  90. data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
  91. data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
  92. data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
  93. data/lib/nokogiri/ffi/libxml.rb +386 -0
  94. data/lib/nokogiri/ffi/structs/common_node.rb +38 -0
  95. data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
  96. data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
  97. data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
  98. data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
  99. data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
  100. data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
  101. data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
  102. data/lib/nokogiri/ffi/structs/xml_document.rb +117 -0
  103. data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
  104. data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
  105. data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
  106. data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
  107. data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
  108. data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
  109. data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
  110. data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
  111. data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
  112. data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
  113. data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
  114. data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
  115. data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
  116. data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
  117. data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
  118. data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
  119. data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +38 -0
  120. data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
  121. data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
  122. data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
  123. data/lib/nokogiri/ffi/weak_bucket.rb +40 -0
  124. data/lib/nokogiri/ffi/xml/attr.rb +41 -0
  125. data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
  126. data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
  127. data/lib/nokogiri/ffi/xml/comment.rb +18 -0
  128. data/lib/nokogiri/ffi/xml/document.rb +162 -0
  129. data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
  130. data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
  131. data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
  132. data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
  133. data/lib/nokogiri/ffi/xml/entity_decl.rb +36 -0
  134. data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
  135. data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
  136. data/lib/nokogiri/ffi/xml/node.rb +556 -0
  137. data/lib/nokogiri/ffi/xml/node_set.rb +149 -0
  138. data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
  139. data/lib/nokogiri/ffi/xml/reader.rb +232 -0
  140. data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
  141. data/lib/nokogiri/ffi/xml/sax/parser.rb +135 -0
  142. data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
  143. data/lib/nokogiri/ffi/xml/sax/push_parser.rb +51 -0
  144. data/lib/nokogiri/ffi/xml/schema.rb +109 -0
  145. data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
  146. data/lib/nokogiri/ffi/xml/text.rb +18 -0
  147. data/lib/nokogiri/ffi/xml/xpath.rb +9 -0
  148. data/lib/nokogiri/ffi/xml/xpath_context.rb +148 -0
  149. data/lib/nokogiri/ffi/xslt/stylesheet.rb +53 -0
  150. data/lib/nokogiri/html.rb +35 -0
  151. data/lib/nokogiri/html/builder.rb +35 -0
  152. data/lib/nokogiri/html/document.rb +90 -0
  153. data/lib/nokogiri/html/document_fragment.rb +36 -0
  154. data/lib/nokogiri/html/element_description.rb +23 -0
  155. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  156. data/lib/nokogiri/html/sax/parser.rb +48 -0
  157. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  158. data/lib/nokogiri/syntax_error.rb +4 -0
  159. data/lib/nokogiri/version.rb +37 -0
  160. data/lib/nokogiri/version_warning.rb +14 -0
  161. data/lib/nokogiri/xml.rb +67 -0
  162. data/lib/nokogiri/xml/attr.rb +14 -0
  163. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  164. data/lib/nokogiri/xml/builder.rb +418 -0
  165. data/lib/nokogiri/xml/cdata.rb +11 -0
  166. data/lib/nokogiri/xml/character_data.rb +7 -0
  167. data/lib/nokogiri/xml/document.rb +194 -0
  168. data/lib/nokogiri/xml/document_fragment.rb +77 -0
  169. data/lib/nokogiri/xml/dtd.rb +11 -0
  170. data/lib/nokogiri/xml/element_content.rb +36 -0
  171. data/lib/nokogiri/xml/element_decl.rb +13 -0
  172. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  173. data/lib/nokogiri/xml/namespace.rb +13 -0
  174. data/lib/nokogiri/xml/node.rb +793 -0
  175. data/lib/nokogiri/xml/node/save_options.rb +42 -0
  176. data/lib/nokogiri/xml/node_set.rb +325 -0
  177. data/lib/nokogiri/xml/notation.rb +6 -0
  178. data/lib/nokogiri/xml/parse_options.rb +85 -0
  179. data/lib/nokogiri/xml/pp.rb +2 -0
  180. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  181. data/lib/nokogiri/xml/pp/node.rb +56 -0
  182. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  183. data/lib/nokogiri/xml/reader.rb +74 -0
  184. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  185. data/lib/nokogiri/xml/sax.rb +4 -0
  186. data/lib/nokogiri/xml/sax/document.rb +160 -0
  187. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  188. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  189. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  190. data/lib/nokogiri/xml/schema.rb +57 -0
  191. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  192. data/lib/nokogiri/xml/text.rb +9 -0
  193. data/lib/nokogiri/xml/xpath.rb +10 -0
  194. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  195. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  196. data/lib/nokogiri/xslt.rb +48 -0
  197. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  198. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  199. data/tasks/cross_compile.rb +158 -0
  200. data/tasks/test.rb +94 -0
  201. data/test/css/test_nthiness.rb +159 -0
  202. data/test/css/test_parser.rb +282 -0
  203. data/test/css/test_tokenizer.rb +190 -0
  204. data/test/css/test_xpath_visitor.rb +85 -0
  205. data/test/ffi/test_document.rb +35 -0
  206. data/test/files/2ch.html +108 -0
  207. data/test/files/address_book.rlx +12 -0
  208. data/test/files/address_book.xml +10 -0
  209. data/test/files/bar/bar.xsd +4 -0
  210. data/test/files/dont_hurt_em_why.xml +422 -0
  211. data/test/files/exslt.xml +8 -0
  212. data/test/files/exslt.xslt +35 -0
  213. data/test/files/foo/foo.xsd +4 -0
  214. data/test/files/po.xml +32 -0
  215. data/test/files/po.xsd +66 -0
  216. data/test/files/shift_jis.html +10 -0
  217. data/test/files/shift_jis.xml +5 -0
  218. data/test/files/snuggles.xml +3 -0
  219. data/test/files/staff.dtd +10 -0
  220. data/test/files/staff.xml +59 -0
  221. data/test/files/staff.xslt +32 -0
  222. data/test/files/tlm.html +850 -0
  223. data/test/files/valid_bar.xml +2 -0
  224. data/test/helper.rb +169 -0
  225. data/test/html/sax/test_parser.rb +74 -0
  226. data/test/html/sax/test_parser_context.rb +48 -0
  227. data/test/html/test_builder.rb +164 -0
  228. data/test/html/test_document.rb +398 -0
  229. data/test/html/test_document_encoding.rb +77 -0
  230. data/test/html/test_document_fragment.rb +182 -0
  231. data/test/html/test_element_description.rb +98 -0
  232. data/test/html/test_named_characters.rb +14 -0
  233. data/test/html/test_node.rb +181 -0
  234. data/test/html/test_node_encoding.rb +27 -0
  235. data/test/test_convert_xpath.rb +135 -0
  236. data/test/test_css_cache.rb +45 -0
  237. data/test/test_encoding_handler.rb +46 -0
  238. data/test/test_memory_leak.rb +87 -0
  239. data/test/test_nokogiri.rb +138 -0
  240. data/test/test_reader.rb +386 -0
  241. data/test/test_soap4r_sax.rb +52 -0
  242. data/test/test_xslt_transforms.rb +188 -0
  243. data/test/xml/node/test_save_options.rb +20 -0
  244. data/test/xml/node/test_subclass.rb +44 -0
  245. data/test/xml/sax/test_parser.rb +307 -0
  246. data/test/xml/sax/test_parser_context.rb +63 -0
  247. data/test/xml/sax/test_push_parser.rb +139 -0
  248. data/test/xml/test_attr.rb +38 -0
  249. data/test/xml/test_attribute_decl.rb +82 -0
  250. data/test/xml/test_builder.rb +210 -0
  251. data/test/xml/test_cdata.rb +50 -0
  252. data/test/xml/test_comment.rb +29 -0
  253. data/test/xml/test_document.rb +668 -0
  254. data/test/xml/test_document_encoding.rb +26 -0
  255. data/test/xml/test_document_fragment.rb +180 -0
  256. data/test/xml/test_dtd.rb +82 -0
  257. data/test/xml/test_dtd_encoding.rb +33 -0
  258. data/test/xml/test_element_content.rb +56 -0
  259. data/test/xml/test_element_decl.rb +73 -0
  260. data/test/xml/test_entity_decl.rb +120 -0
  261. data/test/xml/test_entity_reference.rb +21 -0
  262. data/test/xml/test_namespace.rb +68 -0
  263. data/test/xml/test_node.rb +865 -0
  264. data/test/xml/test_node_attributes.rb +34 -0
  265. data/test/xml/test_node_encoding.rb +107 -0
  266. data/test/xml/test_node_reparenting.rb +293 -0
  267. data/test/xml/test_node_set.rb +649 -0
  268. data/test/xml/test_parse_options.rb +52 -0
  269. data/test/xml/test_processing_instruction.rb +30 -0
  270. data/test/xml/test_reader_encoding.rb +126 -0
  271. data/test/xml/test_relax_ng.rb +60 -0
  272. data/test/xml/test_schema.rb +89 -0
  273. data/test/xml/test_syntax_error.rb +12 -0
  274. data/test/xml/test_text.rb +38 -0
  275. data/test/xml/test_unparented_node.rb +381 -0
  276. data/test/xml/test_xpath.rb +138 -0
  277. metadata +533 -0
@@ -0,0 +1,94 @@
1
+ namespace :test do
2
+ desc "run test suite with aggressive GC"
3
+ task :gc => :build do
4
+ ENV['NOKOGIRI_GC'] = "true"
5
+ Rake::Task["test"].invoke
6
+ end
7
+
8
+ desc "find call-seq in the rdoc"
9
+ task :rdoc_call_seq => 'docs' do
10
+ Dir['doc/**/*.html'].each { |docfile|
11
+ next if docfile =~ /\.src/
12
+ puts "FAIL: #{docfile}" if File.read(docfile) =~ /call-seq/
13
+ }
14
+ end
15
+
16
+ desc "find all undocumented things"
17
+ task :rdoc => 'docs' do
18
+ base = File.expand_path(File.join(File.dirname(__FILE__), '..', 'doc'))
19
+ require 'test/unit'
20
+ test = Class.new(Test::Unit::TestCase)
21
+ Dir["#{base}/**/*.html"].each { |docfile|
22
+ test.class_eval(<<-eotest)
23
+ def test_#{docfile.sub("#{base}/", '').gsub(/[\/\.-]/, '_')}
24
+ assert_no_match(
25
+ /Not documented/,
26
+ File.read('#{docfile}'),
27
+ '#{docfile} has undocumented things'
28
+ )
29
+ end
30
+ eotest
31
+ }
32
+ end
33
+
34
+ desc "Test against multiple versions of libxml2 (MULTIXML2_DIR=directory)"
35
+ task :multixml2 do
36
+ MULTI_XML = File.join(ENV['HOME'], '.multixml2')
37
+ unless File.exists?(MULTI_XML)
38
+ %w{ versions install build }.each { |x|
39
+ FileUtils.mkdir_p(File.join(MULTI_XML, x))
40
+ }
41
+ Dir.chdir File.join(MULTI_XML, 'versions') do
42
+ require 'net/ftp'
43
+ ftp = Net::FTP.new('xmlsoft.org')
44
+ ftp.login('anonymous', 'anonymous')
45
+ ftp.chdir('libxml2')
46
+ ftp.list('libxml2-2.*.tar.gz').each do |x|
47
+ file = x[/[^\s]*$/]
48
+ puts "Downloading #{file}"
49
+ ftp.getbinaryfile(file)
50
+ end
51
+ end
52
+ end
53
+
54
+ # Build any libxml2 versions in $HOME/.multixml2/versions that
55
+ # haven't been built yet
56
+ Dir[File.join(MULTI_XML, 'versions','*.tar.gz')].each do |f|
57
+ filename = File.basename(f, '.tar.gz')
58
+
59
+ install_dir = File.join(MULTI_XML, 'install', filename)
60
+ next if File.exists?(install_dir)
61
+
62
+ Dir.chdir File.join(MULTI_XML, 'versions') do
63
+ system "tar zxvf #{f} -C #{File.join(MULTI_XML, 'build')}"
64
+ end
65
+
66
+ Dir.chdir File.join(MULTI_XML, 'build', filename) do
67
+ system "./configure --prefix=#{install_dir}"
68
+ system "make && make install"
69
+ end
70
+ end
71
+
72
+ test_results = {}
73
+ libxslt = Dir[File.join(MULTI_XML, 'install', 'libxslt*')].first
74
+
75
+ directories = ENV['MULTIXML2_DIR'] ? [ENV['MULTIXML2_DIR']] : Dir[File.join(MULTI_XML, 'install', '*')]
76
+ directories.sort.reverse.each do |xml2_version|
77
+ next unless xml2_version =~ /libxml2/
78
+ extopts = "--with-xml2-include=#{xml2_version}/include/libxml2 --with-xml2-lib=#{xml2_version}/lib --with-xslt-dir=#{libxslt} --with-iconv-dir=/usr"
79
+ cmd = "#{$0} clean test EXTOPTS='#{extopts}' LD_LIBRARY_PATH='#{xml2_version}/lib'"
80
+
81
+ version = File.basename(xml2_version)
82
+ result = system(cmd)
83
+ test_results[version] = {
84
+ :result => result,
85
+ :cmd => cmd
86
+ }
87
+ end
88
+ test_results.sort_by { |k,v| k }.each do |k,v|
89
+ passed = v[:result]
90
+ puts "#{k}: #{passed ? 'PASS' : 'FAIL'}"
91
+ puts "repro: #{v[:cmd]}" unless passed
92
+ end
93
+ end
94
+ end
@@ -0,0 +1,159 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module CSS
5
+ class TestNthiness < Nokogiri::TestCase
6
+ def setup
7
+ super
8
+ doc = <<EOF
9
+ <html>
10
+ <table>
11
+ <tr><td>row1 </td></tr>
12
+ <tr><td>row2 </td></tr>
13
+ <tr><td>row3 </td></tr>
14
+ <tr><td>row4 </td></tr>
15
+ <tr><td>row5 </td></tr>
16
+ <tr><td>row6 </td></tr>
17
+ <tr><td>row7 </td></tr>
18
+ <tr><td>row8 </td></tr>
19
+ <tr><td>row9 </td></tr>
20
+ <tr><td>row10 </td></tr>
21
+ <tr><td>row11 </td></tr>
22
+ <tr><td>row12 </td></tr>
23
+ <tr><td>row13 </td></tr>
24
+ <tr><td>row14 </td></tr>
25
+ </table>
26
+ <div>
27
+ <b>bold1 </b>
28
+ <i>italic1 </i>
29
+ <b>bold2 </b>
30
+ <i>italic2 </i>
31
+ <p>para1 </p>
32
+ <b>bold3 </b>
33
+ </div>
34
+ <div>
35
+ <p>para2 </p>
36
+ <p>para3 </p>
37
+ </div>
38
+ <div>
39
+ <p>para4 </p>
40
+ </div>
41
+ <p class='empty'></p>
42
+ <p class='not-empty'><b></b></p>
43
+ </html>
44
+ EOF
45
+ @parser = Nokogiri.HTML doc
46
+ end
47
+
48
+
49
+ def test_even
50
+ assert_result_rows [2,4,6,8,10,12,14], @parser.search("table/tr:nth(even)")
51
+ end
52
+
53
+ def test_odd
54
+ assert_result_rows [1,3,5,7,9,11,13], @parser.search("table/tr:nth(odd)")
55
+ end
56
+
57
+ def test_2n
58
+ assert_equal @parser.search("table/tr:nth(even)").inner_text, @parser.search("table/tr:nth(2n)").inner_text
59
+ end
60
+
61
+ def test_2np1
62
+ assert_equal @parser.search("table/tr:nth(odd)").inner_text, @parser.search("table/tr:nth(2n+1)").inner_text
63
+ end
64
+
65
+ def test_4np3
66
+ assert_result_rows [3,7,11], @parser.search("table/tr:nth(4n+3)")
67
+ end
68
+
69
+ def test_3np4
70
+ assert_result_rows [4,7,10,13], @parser.search("table/tr:nth(3n+4)")
71
+ end
72
+
73
+ def test_mnp3
74
+ assert_result_rows [1,2,3], @parser.search("table/tr:nth(-n+3)")
75
+ end
76
+
77
+ def test_np3
78
+ assert_result_rows [3,4,5,6,7,8,9,10,11,12,13,14], @parser.search("table/tr:nth(n+3)")
79
+ end
80
+
81
+ def test_first
82
+ assert_result_rows [1], @parser.search("table/tr:first")
83
+ assert_result_rows [1], @parser.search("table/tr:first()")
84
+ end
85
+
86
+ def test_last
87
+ assert_result_rows [14], @parser.search("table/tr:last")
88
+ assert_result_rows [14], @parser.search("table/tr:last()")
89
+ end
90
+
91
+ def test_first_child
92
+ assert_result_rows [1], @parser.search("div/b:first-child"), "bold"
93
+ assert_result_rows [1], @parser.search("table/tr:first-child")
94
+ end
95
+
96
+ def test_last_child
97
+ assert_result_rows [3], @parser.search("div/b:last-child"), "bold"
98
+ assert_result_rows [14], @parser.search("table/tr:last-child")
99
+ end
100
+
101
+ def test_first_of_type
102
+ assert_result_rows [1], @parser.search("table/tr:first-of-type")
103
+ assert_result_rows [1], @parser.search("div/b:first-of-type"), "bold"
104
+ end
105
+
106
+ def test_last_of_type
107
+ assert_result_rows [14], @parser.search("table/tr:last-of-type")
108
+ assert_result_rows [3], @parser.search("div/b:last-of-type"), "bold"
109
+ end
110
+
111
+ def test_only_of_type
112
+ assert_result_rows [1,4], @parser.search("div/p:only-of-type"), "para"
113
+ end
114
+
115
+ def test_only_child
116
+ assert_result_rows [4], @parser.search("div/p:only-child"), "para"
117
+ end
118
+
119
+ def test_empty
120
+ result = @parser.search("p:empty")
121
+ assert_equal 1, result.size, "unexpected number of rows returned: '#{result.inner_text}'"
122
+ assert_equal 'empty', result.first['class']
123
+ end
124
+
125
+ def test_parent
126
+ result = @parser.search("p:parent")
127
+ assert_equal 5, result.size
128
+ 0.upto(3) do |j|
129
+ assert_equal "para#{j+1} ", result[j].inner_text
130
+ end
131
+ assert_equal "not-empty", result[4]['class']
132
+ end
133
+
134
+ def test_siblings
135
+ doc = <<-EOF
136
+ <html><body><div>
137
+ <p id="1">p1 </p>
138
+ <p id="2">p2 </p>
139
+ <p id="3">p3 </p>
140
+ <p id="4">p4 </p>
141
+ <p id="5">p5 </p>
142
+ EOF
143
+ parser = Nokogiri.HTML doc
144
+ assert_equal 2, parser.search("#3 ~ p").size
145
+ assert_equal "p4 p5 ", parser.search("#3 ~ p").inner_text
146
+ assert_equal 0, parser.search("#5 ~ p").size
147
+
148
+ assert_equal 1, parser.search("#3 + p").size
149
+ assert_equal "p4 ", parser.search("#3 + p").inner_text
150
+ assert_equal 0, parser.search("#5 + p").size
151
+ end
152
+
153
+ def assert_result_rows intarray, result, word="row"
154
+ assert_equal intarray.size, result.size, "unexpected number of rows returned: '#{result.inner_text}'"
155
+ assert_equal intarray.map{|j| "#{word}#{j}"}.join(' '), result.inner_text.strip, result.inner_text
156
+ end
157
+ end
158
+ end
159
+ end
@@ -0,0 +1,282 @@
1
+ require "helper"
2
+
3
+ module Nokogiri
4
+ module CSS
5
+ class TestParser < Nokogiri::TestCase
6
+ def setup
7
+ super
8
+ @parser = Nokogiri::CSS::Parser.new
9
+ end
10
+
11
+ def test_extra_single_quote
12
+ assert_raises(CSS::SyntaxError) { @parser.parse("'") }
13
+ end
14
+
15
+ def test_syntax_error_raised
16
+ assert_raises(CSS::SyntaxError) { @parser.parse("a[x=]") }
17
+ end
18
+
19
+ def test_find_by_type
20
+ ast = @parser.parse("a:nth-child(2)").first
21
+ matches = ast.find_by_type(
22
+ [:CONDITIONAL_SELECTOR,
23
+ [:ELEMENT_NAME],
24
+ [:PSEUDO_CLASS,
25
+ [:FUNCTION]
26
+ ]
27
+ ]
28
+ )
29
+ assert_equal(1, matches.length)
30
+ assert_equal(ast, matches.first)
31
+ end
32
+
33
+ def test_to_type
34
+ ast = @parser.parse("a:nth-child(2)").first
35
+ assert_equal(
36
+ [:CONDITIONAL_SELECTOR,
37
+ [:ELEMENT_NAME],
38
+ [:PSEUDO_CLASS,
39
+ [:FUNCTION]
40
+ ]
41
+ ], ast.to_type
42
+ )
43
+ end
44
+
45
+ def test_to_a
46
+ asts = @parser.parse("a:nth-child(2)")
47
+ assert_equal(
48
+ [:CONDITIONAL_SELECTOR,
49
+ [:ELEMENT_NAME, ["a"]],
50
+ [:PSEUDO_CLASS,
51
+ [:FUNCTION, ["nth-child("], ["2"]]
52
+ ]
53
+ ], asts.first.to_a
54
+ )
55
+ end
56
+
57
+ def test_has
58
+ assert_xpath "//a[b]", @parser.parse("a:has(b)")
59
+ assert_xpath "//a[b/c]", @parser.parse("a:has(b > c)")
60
+ end
61
+
62
+ def test_dashmatch
63
+ assert_xpath "//a[@class = 'bar' or starts-with(@class, concat('bar', '-'))]",
64
+ @parser.parse("a[@class|='bar']")
65
+ assert_xpath "//a[@class = 'bar' or starts-with(@class, concat('bar', '-'))]",
66
+ @parser.parse("a[@class |= 'bar']")
67
+ end
68
+
69
+ def test_includes
70
+ assert_xpath "//a[contains(concat(\" \", @class, \" \"),concat(\" \", 'bar', \" \"))]",
71
+ @parser.parse("a[@class~='bar']")
72
+ assert_xpath "//a[contains(concat(\" \", @class, \" \"),concat(\" \", 'bar', \" \"))]",
73
+ @parser.parse("a[@class ~= 'bar']")
74
+ end
75
+
76
+ def test_function_with_arguments
77
+ assert_xpath "//*[position() = 2 and self::a]",
78
+ @parser.parse("a[2]")
79
+ assert_xpath "//*[position() = 2 and self::a]",
80
+ @parser.parse("a:nth-child(2)")
81
+ end
82
+
83
+ def test_carrot
84
+ assert_xpath "//a[starts-with(@id, 'Boing')]",
85
+ @parser.parse("a[id^='Boing']")
86
+ assert_xpath "//a[starts-with(@id, 'Boing')]",
87
+ @parser.parse("a[id ^= 'Boing']")
88
+ end
89
+
90
+ def test_suffix_match
91
+ assert_xpath "//a[substring(@id, string-length(@id) - string-length('Boing') + 1, string-length('Boing')) = 'Boing']",
92
+ @parser.parse("a[id$='Boing']")
93
+ assert_xpath "//a[substring(@id, string-length(@id) - string-length('Boing') + 1, string-length('Boing')) = 'Boing']",
94
+ @parser.parse("a[id $= 'Boing']")
95
+ end
96
+
97
+ def test_attributes_with_at
98
+ ## This is non standard CSS
99
+ assert_xpath "//a[@id = 'Boing']",
100
+ @parser.parse("a[@id='Boing']")
101
+ assert_xpath "//a[@id = 'Boing']",
102
+ @parser.parse("a[@id = 'Boing']")
103
+ end
104
+
105
+ def test_attributes_with_at_and_stuff
106
+ ## This is non standard CSS
107
+ assert_xpath "//a[@id = 'Boing']//div",
108
+ @parser.parse("a[@id='Boing'] div")
109
+ end
110
+
111
+ def test_not_equal
112
+ ## This is non standard CSS
113
+ assert_xpath "//a[child::text() != 'Boing']",
114
+ @parser.parse("a[text()!='Boing']")
115
+ assert_xpath "//a[child::text() != 'Boing']",
116
+ @parser.parse("a[text() != 'Boing']")
117
+ end
118
+
119
+ def test_function
120
+ ## This is non standard CSS
121
+ assert_xpath "//a[child::text()]",
122
+ @parser.parse("a[text()]")
123
+
124
+ ## This is non standard CSS
125
+ assert_xpath "//child::text()",
126
+ @parser.parse("text()")
127
+
128
+ ## This is non standard CSS
129
+ assert_xpath "//a[contains(child::text(), 'Boing')]",
130
+ @parser.parse("a[text()*='Boing']")
131
+ assert_xpath "//a[contains(child::text(), 'Boing')]",
132
+ @parser.parse("a[text() *= 'Boing']")
133
+
134
+ ## This is non standard CSS
135
+ assert_xpath "//script//comment()",
136
+ @parser.parse("script comment()")
137
+ end
138
+
139
+ def test_nonstandard_nth_selectors
140
+ ## These are non standard CSS
141
+ assert_xpath '//a[position() = 99]', @parser.parse('a:eq(99)')
142
+ assert_xpath '//a[position() = 1]', @parser.parse('a:first') # no parens
143
+ assert_xpath '//a[position() = last()]', @parser.parse('a:last') # no parens
144
+ assert_xpath '//a[position() = 99]', @parser.parse('a:nth(99)')
145
+ assert_xpath '//a[position() = 1]', @parser.parse('a:first()')
146
+ assert_xpath '//a[position() = last()]', @parser.parse('a:last()')
147
+ assert_xpath '//a[node()]', @parser.parse('a:parent')
148
+ end
149
+
150
+ def test_standard_nth_selectors
151
+ assert_xpath '//a[position() = 99]', @parser.parse('a:nth-of-type(99)')
152
+ assert_xpath '//a[position() = 1]', @parser.parse('a:first-of-type()')
153
+ assert_xpath '//a[position() = last()]', @parser.parse('a:last-of-type()')
154
+ assert_xpath '//a[position() = 1]', @parser.parse('a:first-of-type') # no parens
155
+ assert_xpath '//a[position() = last()]', @parser.parse('a:last-of-type') # no parens
156
+ assert_xpath '//a[position() = last() - 99]', @parser.parse('a:nth-last-of-type(99)')
157
+ assert_xpath '//a[position() = last() - 99]', @parser.parse('a:nth-last-of-type(99)')
158
+ end
159
+
160
+ def test_nth_child_selectors
161
+ assert_xpath '//*[position() = 1 and self::a]', @parser.parse('a:first-child')
162
+ assert_xpath '//*[position() = last() and self::a]', @parser.parse('a:last-child')
163
+ assert_xpath '//*[position() = 99 and self::a]', @parser.parse('a:nth-child(99)')
164
+ assert_xpath '//*[position() = last() - 99 and self::a]', @parser.parse('a:nth-last-child(99)')
165
+ end
166
+
167
+ def test_miscellaneous_selectors
168
+ assert_xpath '//*[last() = 1 and self::a]',
169
+ @parser.parse('a:only-child')
170
+ assert_xpath '//a[last() = 1]', @parser.parse('a:only-of-type')
171
+ assert_xpath '//a[not(node())]', @parser.parse('a:empty')
172
+ end
173
+
174
+ def test_nth_a_n_plus_b
175
+ assert_xpath '//a[(position() mod 2) = 0]', @parser.parse('a:nth-of-type(2n)')
176
+ assert_xpath '//a[(position() >= 1) and (((position()-1) mod 2) = 0)]', @parser.parse('a:nth-of-type(2n+1)')
177
+ assert_xpath '//a[(position() mod 2) = 0]', @parser.parse('a:nth-of-type(even)')
178
+ assert_xpath '//a[(position() >= 1) and (((position()-1) mod 2) = 0)]', @parser.parse('a:nth-of-type(odd)')
179
+ assert_xpath '//a[(position() >= 3) and (((position()-3) mod 4) = 0)]', @parser.parse('a:nth-of-type(4n+3)')
180
+ assert_xpath '//a[(position() <= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(-1n+3)')
181
+ assert_xpath '//a[(position() <= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(-n+3)')
182
+ assert_xpath '//a[(position() >= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(1n+3)')
183
+ assert_xpath '//a[(position() >= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(n+3)')
184
+ end
185
+
186
+ def test_preceding_selector
187
+ assert_xpath "//F[preceding-sibling::E]",
188
+ @parser.parse("E ~ F")
189
+ end
190
+
191
+ def test_direct_preceding_selector
192
+ assert_xpath "//E/following-sibling::*[1]/self::F",
193
+ @parser.parse("E + F")
194
+ end
195
+
196
+ def test_attribute
197
+ assert_xpath "//h1[@a = 'Tender Lovemaking']",
198
+ @parser.parse("h1[a='Tender Lovemaking']")
199
+ end
200
+
201
+ def test_id
202
+ assert_xpath "//*[@id = 'foo']", @parser.parse('#foo')
203
+ end
204
+
205
+ def test_pseudo_class_no_ident
206
+ assert_xpath "//*[link(.)]", @parser.parse(':link')
207
+ end
208
+
209
+ def test_pseudo_class
210
+ assert_xpath "//a[link(.)]", @parser.parse('a:link')
211
+ assert_xpath "//a[visited(.)]", @parser.parse('a:visited')
212
+ assert_xpath "//a[hover(.)]", @parser.parse('a:hover')
213
+ assert_xpath "//a[active(.)]", @parser.parse('a:active')
214
+ assert_xpath "//a[active(.) and contains(concat(' ', @class, ' '), ' foo ')]",
215
+ @parser.parse('a:active.foo')
216
+ end
217
+
218
+ def test_star
219
+ assert_xpath "//*", @parser.parse('*')
220
+ assert_xpath "//*[contains(concat(' ', @class, ' '), ' pastoral ')]",
221
+ @parser.parse('*.pastoral')
222
+ end
223
+
224
+ def test_class
225
+ assert_xpath "//*[contains(concat(' ', @class, ' '), ' a ') and contains(concat(' ', @class, ' '), ' b ')]",
226
+ @parser.parse('.a.b')
227
+ assert_xpath "//*[contains(concat(' ', @class, ' '), ' awesome ')]",
228
+ @parser.parse('.awesome')
229
+ assert_xpath "//foo[contains(concat(' ', @class, ' '), ' awesome ')]",
230
+ @parser.parse('foo.awesome')
231
+ assert_xpath "//foo//*[contains(concat(' ', @class, ' '), ' awesome ')]",
232
+ @parser.parse('foo .awesome')
233
+ end
234
+
235
+ def test_not_so_simple_not
236
+ assert_xpath "//*[@id = 'p' and not(contains(concat(' ', @class, ' '), ' a '))]",
237
+ @parser.parse('#p:not(.a)')
238
+ assert_xpath "//p[contains(concat(' ', @class, ' '), ' a ') and not(contains(concat(' ', @class, ' '), ' b '))]",
239
+ @parser.parse('p.a:not(.b)')
240
+ assert_xpath "//p[@a = 'foo' and not(contains(concat(' ', @class, ' '), ' b '))]",
241
+ @parser.parse("p[a='foo']:not(.b)")
242
+ end
243
+
244
+ def test_ident
245
+ assert_xpath '//x', @parser.parse('x')
246
+ end
247
+
248
+ def test_parse_space
249
+ assert_xpath '//x//y', @parser.parse('x y')
250
+ end
251
+
252
+ def test_parse_descendant
253
+ assert_xpath '//x/y', @parser.parse('x > y')
254
+ end
255
+
256
+ def test_parse_slash
257
+ ## This is non standard CSS
258
+ assert_xpath '//x/y', @parser.parse('x/y')
259
+ end
260
+
261
+ def test_parse_doubleslash
262
+ ## This is non standard CSS
263
+ assert_xpath '//x//y', @parser.parse('x//y')
264
+ end
265
+
266
+ def test_multi_path
267
+ assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y, y > z')
268
+ assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y,y > z')
269
+ ###
270
+ # TODO: should we make this work?
271
+ # assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y | y > z')
272
+ end
273
+
274
+ def assert_xpath expecteds, asts
275
+ expecteds = [expecteds].flatten
276
+ expecteds.zip(asts).each do |expected, actual|
277
+ assert_equal expected, actual.to_xpath
278
+ end
279
+ end
280
+ end
281
+ end
282
+ end