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 +1,142 @@
1
+ ENV['RC_ARCHS'] = '' if RUBY_PLATFORM =~ /darwin/
2
+
3
+ # :stopdoc:
4
+
5
+ require 'mkmf'
6
+
7
+ RbConfig::MAKEFILE_CONFIG['CC'] = ENV['CC'] if ENV['CC']
8
+
9
+ ROOT = File.expand_path(File.join(File.dirname(__FILE__), '..', '..'))
10
+ LIBDIR = RbConfig::CONFIG['libdir']
11
+ INCLUDEDIR = RbConfig::CONFIG['includedir']
12
+
13
+ if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'macruby'
14
+ $LIBRUBYARG_STATIC.gsub!(/-static/, '')
15
+ end
16
+
17
+ $CFLAGS << " #{ENV["CFLAGS"]}"
18
+ $LIBS << " #{ENV["LIBS"]}"
19
+
20
+ if RbConfig::CONFIG['target_os'] == 'mingw32' || RbConfig::CONFIG['target_os'] =~ /mswin/
21
+ $CFLAGS << " -DXP_WIN -DXP_WIN32 -DUSE_INCLUDED_VASPRINTF"
22
+ elsif RbConfig::CONFIG['target_os'] =~ /solaris/
23
+ $CFLAGS << " -DUSE_INCLUDED_VASPRINTF"
24
+ else
25
+ $CFLAGS << " -g -DXP_UNIX"
26
+ end
27
+
28
+ if RbConfig::MAKEFILE_CONFIG['CC'] =~ /mingw/
29
+ $CFLAGS << " -DIN_LIBXML"
30
+ end
31
+
32
+ if RbConfig::MAKEFILE_CONFIG['CC'] =~ /gcc/
33
+ $CFLAGS << " -O3" unless $CFLAGS[/-O\d/]
34
+ $CFLAGS << " -Wall -Wcast-qual -Wwrite-strings -Wconversion -Wmissing-noreturn -Winline"
35
+ end
36
+
37
+ if RbConfig::CONFIG['target_os'] =~ /mswin/
38
+ lib_prefix = 'lib'
39
+
40
+ # There's no default include/lib dir on Windows. Let's just add the Ruby ones
41
+ # and resort on the search path specified by INCLUDE and LIB environment
42
+ # variables
43
+ HEADER_DIRS = [INCLUDEDIR]
44
+ LIB_DIRS = [LIBDIR]
45
+ XML2_HEADER_DIRS = [File.join(INCLUDEDIR, "libxml2"), INCLUDEDIR]
46
+
47
+ else
48
+ lib_prefix = ''
49
+
50
+ HEADER_DIRS = [
51
+ # First search /opt/local for macports
52
+ '/opt/local/include',
53
+
54
+ # Then search /usr/local for people that installed from source
55
+ '/usr/local/include',
56
+
57
+ # Check the ruby install locations
58
+ INCLUDEDIR,
59
+
60
+ # Finally fall back to /usr
61
+ '/usr/include',
62
+ '/usr/include/libxml2',
63
+ ]
64
+
65
+ LIB_DIRS = [
66
+ # First search /opt/local for macports
67
+ '/opt/local/lib',
68
+
69
+ # Then search /usr/local for people that installed from source
70
+ '/usr/local/lib',
71
+
72
+ # Check the ruby install locations
73
+ LIBDIR,
74
+
75
+ # Finally fall back to /usr
76
+ '/usr/lib',
77
+ ]
78
+
79
+ XML2_HEADER_DIRS = [
80
+ '/opt/local/include/libxml2',
81
+ '/usr/local/include/libxml2',
82
+ File.join(INCLUDEDIR, "libxml2")
83
+ ] + HEADER_DIRS
84
+
85
+ # If the user has homebrew installed, use the libxml2 inside homebrew
86
+ brew_prefix = `brew --prefix libxml2 2> /dev/null`.chomp
87
+ unless brew_prefix.empty?
88
+ LIB_DIRS.unshift File.join(brew_prefix, 'lib')
89
+ XML2_HEADER_DIRS.unshift File.join(brew_prefix, 'include/libxml2')
90
+ end
91
+ end
92
+
93
+ dir_config('zlib', HEADER_DIRS, LIB_DIRS)
94
+ dir_config('iconv', HEADER_DIRS, LIB_DIRS)
95
+ dir_config('xml2', XML2_HEADER_DIRS, LIB_DIRS)
96
+ dir_config('xslt', HEADER_DIRS, LIB_DIRS)
97
+
98
+ def asplode(lib)
99
+ abort "-----\n#{lib} is missing. please visit http://nokogiri.org/tutorials/installing_nokogiri.html for help with installing dependencies.\n-----"
100
+ end
101
+
102
+ pkg_config('libxslt')
103
+ pkg_config('libxml-2.0')
104
+ pkg_config('libiconv')
105
+
106
+ def have_iconv?
107
+ %w{ iconv_open libiconv_open }.any? do |method|
108
+ have_func(method, 'iconv.h') or
109
+ have_library('iconv', method, 'iconv.h') or
110
+ find_library('iconv', method, 'iconv.h')
111
+ end
112
+ end
113
+
114
+ asplode "libxml2" unless find_header('libxml/parser.h')
115
+ asplode "libxslt" unless find_header('libxslt/xslt.h')
116
+ asplode "libexslt" unless find_header('libexslt/exslt.h')
117
+ asplode "libiconv" unless have_iconv?
118
+ asplode "libxml2" unless find_library("#{lib_prefix}xml2", 'xmlParseDoc')
119
+ asplode "libxslt" unless find_library("#{lib_prefix}xslt", 'xsltParseStylesheetDoc')
120
+ asplode "libexslt" unless find_library("#{lib_prefix}exslt", 'exsltFuncRegister')
121
+
122
+ unless have_func('xmlHasFeature')
123
+ abort "-----\nThe function 'xmlHasFeature' is missing from your installation of libxml2. Likely this means that your installed version of libxml2 is old enough that nokogiri will not work well. To get around this problem, please upgrade your installation of libxml2.
124
+
125
+ Please visit http://nokogiri.org/tutorials/installing_nokogiri.html for more help!"
126
+ end
127
+
128
+ have_func 'xmlFirstElementChild'
129
+ have_func('xmlRelaxNGSetParserStructuredErrors')
130
+ have_func('xmlRelaxNGSetParserStructuredErrors')
131
+ have_func('xmlRelaxNGSetValidStructuredErrors')
132
+ have_func('xmlSchemaSetValidStructuredErrors')
133
+ have_func('xmlSchemaSetParserStructuredErrors')
134
+
135
+ if ENV['CPUPROFILE']
136
+ unless find_library('profiler', 'ProfilerEnable', *LIB_DIRS)
137
+ abort "google performance tools are not installed"
138
+ end
139
+ end
140
+
141
+ create_makefile('nokogiri/nokogiri')
142
+ # :startdoc:
@@ -0,0 +1,170 @@
1
+ #include <html_document.h>
2
+
3
+ static ID id_encoding_found;
4
+
5
+ /*
6
+ * call-seq:
7
+ * new
8
+ *
9
+ * Create a new document
10
+ */
11
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
12
+ {
13
+ VALUE uri, external_id, rest, rb_doc;
14
+ htmlDocPtr doc;
15
+
16
+ rb_scan_args(argc, argv, "0*", &rest);
17
+ uri = rb_ary_entry(rest, (long)0);
18
+ external_id = rb_ary_entry(rest, (long)1);
19
+
20
+ doc = htmlNewDoc(
21
+ RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
22
+ RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
23
+ );
24
+ rb_doc = Nokogiri_wrap_xml_document(klass, doc);
25
+ rb_obj_call_init(rb_doc, argc, argv);
26
+ return rb_doc ;
27
+ }
28
+
29
+ /*
30
+ * call-seq:
31
+ * read_io(io, url, encoding, options)
32
+ *
33
+ * Read the HTML document from +io+ with given +url+, +encoding+,
34
+ * and +options+. See Nokogiri::HTML.parse
35
+ */
36
+ static VALUE read_io( VALUE klass,
37
+ VALUE io,
38
+ VALUE url,
39
+ VALUE encoding,
40
+ VALUE options )
41
+ {
42
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
43
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
44
+ VALUE error_list = rb_ary_new();
45
+ VALUE document;
46
+ htmlDocPtr doc;
47
+
48
+ xmlResetLastError();
49
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
50
+
51
+ doc = htmlReadIO(
52
+ io_read_callback,
53
+ io_close_callback,
54
+ (void *)io,
55
+ c_url,
56
+ c_enc,
57
+ (int)NUM2INT(options)
58
+ );
59
+ xmlSetStructuredErrorFunc(NULL, NULL);
60
+
61
+ /*
62
+ * If EncodingFound has occurred in EncodingReader, make sure to do
63
+ * a cleanup and propagate the error.
64
+ */
65
+ if (rb_respond_to(io, id_encoding_found)) {
66
+ VALUE encoding_found = rb_funcall(io, id_encoding_found, 0);
67
+ if (!NIL_P(encoding_found)) {
68
+ xmlFreeDoc(doc);
69
+ rb_exc_raise(encoding_found);
70
+ }
71
+ }
72
+
73
+ if(doc == NULL) {
74
+ xmlErrorPtr error;
75
+
76
+ xmlFreeDoc(doc);
77
+
78
+ error = xmlGetLastError();
79
+ if(error)
80
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
81
+ else
82
+ rb_raise(rb_eRuntimeError, "Could not parse document");
83
+
84
+ return Qnil;
85
+ }
86
+
87
+ document = Nokogiri_wrap_xml_document(klass, doc);
88
+ rb_iv_set(document, "@errors", error_list);
89
+ return document;
90
+ }
91
+
92
+ /*
93
+ * call-seq:
94
+ * read_memory(string, url, encoding, options)
95
+ *
96
+ * Read the HTML document contained in +string+ with given +url+, +encoding+,
97
+ * and +options+. See Nokogiri::HTML.parse
98
+ */
99
+ static VALUE read_memory( VALUE klass,
100
+ VALUE string,
101
+ VALUE url,
102
+ VALUE encoding,
103
+ VALUE options )
104
+ {
105
+ const char * c_buffer = StringValuePtr(string);
106
+ const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
107
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
108
+ int len = (int)RSTRING_LEN(string);
109
+ VALUE error_list = rb_ary_new();
110
+ VALUE document;
111
+ htmlDocPtr doc;
112
+
113
+ xmlResetLastError();
114
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
115
+
116
+ doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
117
+ xmlSetStructuredErrorFunc(NULL, NULL);
118
+
119
+ if(doc == NULL) {
120
+ xmlErrorPtr error;
121
+
122
+ xmlFreeDoc(doc);
123
+
124
+ error = xmlGetLastError();
125
+ if(error)
126
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
127
+ else
128
+ rb_raise(rb_eRuntimeError, "Could not parse document");
129
+
130
+ return Qnil;
131
+ }
132
+
133
+ document = Nokogiri_wrap_xml_document(klass, doc);
134
+ rb_iv_set(document, "@errors", error_list);
135
+ return document;
136
+ }
137
+
138
+ /*
139
+ * call-seq:
140
+ * type
141
+ *
142
+ * The type for this document
143
+ */
144
+ static VALUE type(VALUE self)
145
+ {
146
+ htmlDocPtr doc;
147
+ Data_Get_Struct(self, xmlDoc, doc);
148
+ return INT2NUM((long)doc->type);
149
+ }
150
+
151
+ VALUE cNokogiriHtmlDocument ;
152
+ void init_html_document()
153
+ {
154
+ VALUE nokogiri = rb_define_module("Nokogiri");
155
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
156
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
157
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
158
+ VALUE xml_doc = rb_define_class_under(xml, "Document", node);
159
+ VALUE klass = rb_define_class_under(html, "Document", xml_doc);
160
+
161
+ cNokogiriHtmlDocument = klass;
162
+
163
+ rb_define_singleton_method(klass, "read_memory", read_memory, 4);
164
+ rb_define_singleton_method(klass, "read_io", read_io, 4);
165
+ rb_define_singleton_method(klass, "new", new, -1);
166
+
167
+ rb_define_method(klass, "type", type, 0);
168
+
169
+ id_encoding_found = rb_intern("encoding_found");
170
+ }
@@ -0,0 +1,10 @@
1
+ #ifndef NOKOGIRI_HTML_DOCUMENT
2
+ #define NOKOGIRI_HTML_DOCUMENT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_document();
7
+
8
+ extern VALUE cNokogiriHtmlDocument ;
9
+
10
+ #endif
@@ -0,0 +1,276 @@
1
+ #include <html_element_description.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * required_attributes
6
+ *
7
+ * A list of required attributes for this element
8
+ */
9
+ static VALUE required_attributes(VALUE self)
10
+ {
11
+ htmlElemDesc * description;
12
+ VALUE list;
13
+ int i;
14
+
15
+ Data_Get_Struct(self, htmlElemDesc, description);
16
+
17
+ list = rb_ary_new();
18
+
19
+ if(NULL == description->attrs_req) return list;
20
+
21
+ for(i = 0; description->attrs_depr[i]; i++) {
22
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
23
+ }
24
+
25
+ return list;
26
+ }
27
+
28
+ /*
29
+ * call-seq:
30
+ * deprecated_attributes
31
+ *
32
+ * A list of deprecated attributes for this element
33
+ */
34
+ static VALUE deprecated_attributes(VALUE self)
35
+ {
36
+ htmlElemDesc * description;
37
+ VALUE list;
38
+ int i;
39
+
40
+ Data_Get_Struct(self, htmlElemDesc, description);
41
+
42
+ list = rb_ary_new();
43
+
44
+ if(NULL == description->attrs_depr) return list;
45
+
46
+ for(i = 0; description->attrs_depr[i]; i++) {
47
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
48
+ }
49
+
50
+ return list;
51
+ }
52
+
53
+ /*
54
+ * call-seq:
55
+ * optional_attributes
56
+ *
57
+ * A list of optional attributes for this element
58
+ */
59
+ static VALUE optional_attributes(VALUE self)
60
+ {
61
+ htmlElemDesc * description;
62
+ VALUE list;
63
+ int i;
64
+
65
+ Data_Get_Struct(self, htmlElemDesc, description);
66
+
67
+ list = rb_ary_new();
68
+
69
+ if(NULL == description->attrs_opt) return list;
70
+
71
+ for(i = 0; description->attrs_opt[i]; i++) {
72
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
73
+ }
74
+
75
+ return list;
76
+ }
77
+
78
+ /*
79
+ * call-seq:
80
+ * default_sub_element
81
+ *
82
+ * The default sub element for this element
83
+ */
84
+ static VALUE default_sub_element(VALUE self)
85
+ {
86
+ htmlElemDesc * description;
87
+ Data_Get_Struct(self, htmlElemDesc, description);
88
+
89
+ return NOKOGIRI_STR_NEW2(description->defaultsubelt);
90
+ }
91
+
92
+ /*
93
+ * call-seq:
94
+ * sub_elements
95
+ *
96
+ * A list of allowed sub elements for this element.
97
+ */
98
+ static VALUE sub_elements(VALUE self)
99
+ {
100
+ htmlElemDesc * description;
101
+ VALUE list;
102
+ int i;
103
+
104
+ Data_Get_Struct(self, htmlElemDesc, description);
105
+
106
+ list = rb_ary_new();
107
+
108
+ if(NULL == description->subelts) return list;
109
+
110
+ for(i = 0; description->subelts[i]; i++) {
111
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
112
+ }
113
+
114
+ return list;
115
+ }
116
+
117
+ /*
118
+ * call-seq:
119
+ * description
120
+ *
121
+ * The description for this element
122
+ */
123
+ static VALUE description(VALUE self)
124
+ {
125
+ htmlElemDesc * description;
126
+ Data_Get_Struct(self, htmlElemDesc, description);
127
+
128
+ return NOKOGIRI_STR_NEW2(description->desc);
129
+ }
130
+
131
+ /*
132
+ * call-seq:
133
+ * inline?
134
+ *
135
+ * Is this element an inline element?
136
+ */
137
+ static VALUE inline_eh(VALUE self)
138
+ {
139
+ htmlElemDesc * description;
140
+ Data_Get_Struct(self, htmlElemDesc, description);
141
+
142
+ if(description->isinline) return Qtrue;
143
+ return Qfalse;
144
+ }
145
+
146
+ /*
147
+ * call-seq:
148
+ * deprecated?
149
+ *
150
+ * Is this element deprecated?
151
+ */
152
+ static VALUE deprecated_eh(VALUE self)
153
+ {
154
+ htmlElemDesc * description;
155
+ Data_Get_Struct(self, htmlElemDesc, description);
156
+
157
+ if(description->depr) return Qtrue;
158
+ return Qfalse;
159
+ }
160
+
161
+ /*
162
+ * call-seq:
163
+ * empty?
164
+ *
165
+ * Is this an empty element?
166
+ */
167
+ static VALUE empty_eh(VALUE self)
168
+ {
169
+ htmlElemDesc * description;
170
+ Data_Get_Struct(self, htmlElemDesc, description);
171
+
172
+ if(description->empty) return Qtrue;
173
+ return Qfalse;
174
+ }
175
+
176
+ /*
177
+ * call-seq:
178
+ * save_end_tag?
179
+ *
180
+ * Should the end tag be saved?
181
+ */
182
+ static VALUE save_end_tag_eh(VALUE self)
183
+ {
184
+ htmlElemDesc * description;
185
+ Data_Get_Struct(self, htmlElemDesc, description);
186
+
187
+ if(description->saveEndTag) return Qtrue;
188
+ return Qfalse;
189
+ }
190
+
191
+ /*
192
+ * call-seq:
193
+ * implied_end_tag?
194
+ *
195
+ * Can the end tag be implied for this tag?
196
+ */
197
+ static VALUE implied_end_tag_eh(VALUE self)
198
+ {
199
+ htmlElemDesc * description;
200
+ Data_Get_Struct(self, htmlElemDesc, description);
201
+
202
+ if(description->endTag) return Qtrue;
203
+ return Qfalse;
204
+ }
205
+
206
+ /*
207
+ * call-seq:
208
+ * implied_start_tag?
209
+ *
210
+ * Can the start tag be implied for this tag?
211
+ */
212
+ static VALUE implied_start_tag_eh(VALUE self)
213
+ {
214
+ htmlElemDesc * description;
215
+ Data_Get_Struct(self, htmlElemDesc, description);
216
+
217
+ if(description->startTag) return Qtrue;
218
+ return Qfalse;
219
+ }
220
+
221
+ /*
222
+ * call-seq:
223
+ * name
224
+ *
225
+ * Get the tag name for this ElemementDescription
226
+ */
227
+ static VALUE name(VALUE self)
228
+ {
229
+ htmlElemDesc * description;
230
+ Data_Get_Struct(self, htmlElemDesc, description);
231
+
232
+ if(NULL == description->name) return Qnil;
233
+ return NOKOGIRI_STR_NEW2(description->name);
234
+ }
235
+
236
+ /*
237
+ * call-seq:
238
+ * [](tag_name)
239
+ *
240
+ * Get ElemementDescription for +tag_name+
241
+ */
242
+ static VALUE get_description(VALUE klass, VALUE tag_name)
243
+ {
244
+ const htmlElemDesc * description = htmlTagLookup(
245
+ (const xmlChar *)StringValuePtr(tag_name)
246
+ );
247
+
248
+ if(NULL == description) return Qnil;
249
+ return Data_Wrap_Struct(klass, 0, 0, (void *)description);
250
+ }
251
+
252
+ VALUE cNokogiriHtmlElementDescription ;
253
+ void init_html_element_description()
254
+ {
255
+ VALUE nokogiri = rb_define_module("Nokogiri");
256
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
257
+ VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
258
+
259
+ cNokogiriHtmlElementDescription = klass;
260
+
261
+ rb_define_singleton_method(klass, "[]", get_description, 1);
262
+
263
+ rb_define_method(klass, "name", name, 0);
264
+ rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
265
+ rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
266
+ rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
267
+ rb_define_method(klass, "empty?", empty_eh, 0);
268
+ rb_define_method(klass, "deprecated?", deprecated_eh, 0);
269
+ rb_define_method(klass, "inline?", inline_eh, 0);
270
+ rb_define_method(klass, "description", description, 0);
271
+ rb_define_method(klass, "sub_elements", sub_elements, 0);
272
+ rb_define_method(klass, "default_sub_element", default_sub_element, 0);
273
+ rb_define_method(klass, "optional_attributes", optional_attributes, 0);
274
+ rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
275
+ rb_define_method(klass, "required_attributes", required_attributes, 0);
276
+ }