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 +1,144 @@
1
+ /**
2
+ * (The MIT License)
3
+ *
4
+ * Copyright (c) 2008 - 2011:
5
+ *
6
+ * * {Aaron Patterson}[http://tenderlovemaking.com]
7
+ * * {Mike Dalessio}[http://mike.daless.io]
8
+ * * {Charles Nutter}[http://blog.headius.com]
9
+ * * {Sergio Arbeo}[http://www.serabe.com]
10
+ * * {Patrick Mahoney}[http://polycrystal.org]
11
+ * * {Yoko Harada}[http://yokolet.blogspot.com]
12
+ *
13
+ * Permission is hereby granted, free of charge, to any person obtaining
14
+ * a copy of this software and associated documentation files (the
15
+ * 'Software'), to deal in the Software without restriction, including
16
+ * without limitation the rights to use, copy, modify, merge, publish,
17
+ * distribute, sublicense, and/or sell copies of the Software, and to
18
+ * permit persons to whom the Software is furnished to do so, subject to
19
+ * the following conditions:
20
+ *
21
+ * The above copyright notice and this permission notice shall be
22
+ * included in all copies or substantial portions of the Software.
23
+ *
24
+ * THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND,
25
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
26
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
27
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
28
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
29
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
30
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
31
+ */
32
+
33
+ package nokogiri;
34
+
35
+ import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
36
+
37
+ import java.io.ByteArrayInputStream;
38
+ import java.io.IOException;
39
+ import java.io.InputStream;
40
+ import java.io.StringWriter;
41
+ import java.io.UnsupportedEncodingException;
42
+
43
+ import javax.xml.transform.Source;
44
+ import javax.xml.transform.TransformerConfigurationException;
45
+ import javax.xml.transform.TransformerException;
46
+ import javax.xml.transform.TransformerFactory;
47
+ import javax.xml.transform.dom.DOMSource;
48
+ import javax.xml.transform.stream.StreamResult;
49
+ import javax.xml.transform.stream.StreamSource;
50
+
51
+ import org.iso_relax.verifier.Schema;
52
+ import org.iso_relax.verifier.Verifier;
53
+ import org.iso_relax.verifier.VerifierConfigurationException;
54
+ import org.iso_relax.verifier.VerifierFactory;
55
+ import org.jruby.Ruby;
56
+ import org.jruby.RubyClass;
57
+ import org.jruby.anno.JRubyClass;
58
+ import org.jruby.runtime.ThreadContext;
59
+ import org.w3c.dom.Document;
60
+ import org.xml.sax.ErrorHandler;
61
+ import org.xml.sax.SAXException;
62
+
63
+ /**
64
+ * Class for Nokogiri::XML::RelaxNG
65
+ *
66
+ * @author sergio
67
+ * @author Yoko Harada <yokolet@gmail.com>
68
+ */
69
+ @JRubyClass(name="Nokogiri::XML::RelaxNG", parent="Nokogiri::XML::Schema")
70
+ public class XmlRelaxng extends XmlSchema {
71
+ private Verifier verifier;
72
+
73
+ public XmlRelaxng(Ruby ruby, RubyClass klazz) {
74
+ super(ruby, klazz);
75
+ }
76
+
77
+ private void setVerifier(Verifier verifier) {
78
+ this.verifier = verifier;
79
+ }
80
+
81
+ static XmlSchema createSchemaInstance(ThreadContext context, RubyClass klazz, Source source) {
82
+ Ruby runtime = context.getRuntime();
83
+ XmlRelaxng xmlRelaxng = (XmlRelaxng) NokogiriService.XML_RELAXNG_ALLOCATOR.allocate(runtime, klazz);
84
+ xmlRelaxng.setInstanceVariable("@errors", runtime.newEmptyArray());
85
+
86
+ try {
87
+ Schema schema = xmlRelaxng.getSchema(source, context);
88
+ xmlRelaxng.setVerifier(schema.newVerifier());
89
+ return xmlRelaxng;
90
+ } catch (VerifierConfigurationException ex) {
91
+ throw context.getRuntime().newRuntimeError("Could not parse document: " + ex.getMessage());
92
+ }
93
+ }
94
+
95
+ private Schema getSchema(Source source, ThreadContext context) {
96
+ InputStream is = null;
97
+ VerifierFactory factory = new com.thaiopensource.relaxng.jarv.VerifierFactoryImpl();
98
+ if (source instanceof StreamSource) {
99
+ StreamSource ss = (StreamSource)source;
100
+ is = ss.getInputStream();
101
+ } else { //if (this.source instanceof DOMSource)
102
+ DOMSource ds = (DOMSource)source;
103
+ StringWriter xmlAsWriter = new StringWriter();
104
+ StreamResult result = new StreamResult(xmlAsWriter);
105
+ try {
106
+ TransformerFactory.newInstance().newTransformer().transform(ds, result);
107
+ } catch (TransformerConfigurationException ex) {
108
+ throw context.getRuntime()
109
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
110
+ } catch (TransformerException ex) {
111
+ throw context.getRuntime()
112
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
113
+ }
114
+ try {
115
+ is = new ByteArrayInputStream(xmlAsWriter.toString().getBytes("UTF-8"));
116
+ } catch (UnsupportedEncodingException ex) {
117
+ throw context.getRuntime()
118
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
119
+ }
120
+ }
121
+
122
+ try {
123
+ return factory.compileSchema(is);
124
+ } catch (VerifierConfigurationException ex) {
125
+ throw context.getRuntime()
126
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
127
+ } catch (SAXException ex) {
128
+ throw context.getRuntime()
129
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
130
+ } catch (IOException ex) {
131
+ throw context.getRuntime().newIOError(ex.getMessage());
132
+ }
133
+ }
134
+
135
+ @Override
136
+ protected void setErrorHandler(ErrorHandler errorHandler) {
137
+ verifier.setErrorHandler(errorHandler);
138
+ }
139
+
140
+ @Override
141
+ protected void validate(Document document) throws SAXException, IOException {
142
+ verifier.verify(document);
143
+ }
144
+ }
@@ -0,0 +1,356 @@
1
+ /**
2
+ * (The MIT License)
3
+ *
4
+ * Copyright (c) 2008 - 2011:
5
+ *
6
+ * * {Aaron Patterson}[http://tenderlovemaking.com]
7
+ * * {Mike Dalessio}[http://mike.daless.io]
8
+ * * {Charles Nutter}[http://blog.headius.com]
9
+ * * {Sergio Arbeo}[http://www.serabe.com]
10
+ * * {Patrick Mahoney}[http://polycrystal.org]
11
+ * * {Yoko Harada}[http://yokolet.blogspot.com]
12
+ *
13
+ * Permission is hereby granted, free of charge, to any person obtaining
14
+ * a copy of this software and associated documentation files (the
15
+ * 'Software'), to deal in the Software without restriction, including
16
+ * without limitation the rights to use, copy, modify, merge, publish,
17
+ * distribute, sublicense, and/or sell copies of the Software, and to
18
+ * permit persons to whom the Software is furnished to do so, subject to
19
+ * the following conditions:
20
+ *
21
+ * The above copyright notice and this permission notice shall be
22
+ * included in all copies or substantial portions of the Software.
23
+ *
24
+ * THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND,
25
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
26
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
27
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
28
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
29
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
30
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
31
+ */
32
+
33
+ package nokogiri;
34
+
35
+ import static nokogiri.internals.NokogiriHelpers.isWhitespaceText;
36
+ import static org.jruby.javasupport.util.RuntimeHelpers.invoke;
37
+
38
+ import java.io.IOException;
39
+ import java.io.InputStream;
40
+
41
+ import nokogiri.internals.NokogiriHandler;
42
+ import nokogiri.internals.ParserContext;
43
+ import nokogiri.internals.XmlSaxParser;
44
+
45
+ import org.apache.xerces.parsers.AbstractSAXParser;
46
+ import org.jruby.Ruby;
47
+ import org.jruby.RubyClass;
48
+ import org.jruby.RubyFixnum;
49
+ import org.jruby.RubyModule;
50
+ import org.jruby.RubyObjectAdapter;
51
+ import org.jruby.anno.JRubyClass;
52
+ import org.jruby.anno.JRubyMethod;
53
+ import org.jruby.exceptions.RaiseException;
54
+ import org.jruby.javasupport.JavaEmbedUtils;
55
+ import org.jruby.runtime.ThreadContext;
56
+ import org.jruby.runtime.builtin.IRubyObject;
57
+ import org.xml.sax.ContentHandler;
58
+ import org.xml.sax.ErrorHandler;
59
+ import org.xml.sax.SAXException;
60
+ import org.xml.sax.SAXNotRecognizedException;
61
+ import org.xml.sax.SAXNotSupportedException;
62
+ import org.xml.sax.SAXParseException;
63
+
64
+ /**
65
+ * Base class for the SAX parsers.
66
+ *
67
+ * @author Patrick Mahoney <pat@polycrystal.org>
68
+ * @author Yoko Harada <yokolet@gmail.com>
69
+ */
70
+ @JRubyClass(name="Nokogiri::XML::SAX::ParserContext")
71
+ public class XmlSaxParserContext extends ParserContext {
72
+ protected static final String FEATURE_NAMESPACES =
73
+ "http://xml.org/sax/features/namespaces";
74
+ protected static final String FEATURE_NAMESPACE_PREFIXES =
75
+ "http://xml.org/sax/features/namespace-prefixes";
76
+ protected static final String FEATURE_LOAD_EXTERNAL_DTD =
77
+ "http://apache.org/xml/features/nonvalidating/load-external-dtd";
78
+
79
+ protected AbstractSAXParser parser;
80
+
81
+ protected NokogiriHandler handler = null;
82
+ private IRubyObject replaceEntities;
83
+
84
+ public XmlSaxParserContext(final Ruby ruby, RubyClass rubyClass) {
85
+ super(ruby, rubyClass);
86
+ }
87
+
88
+ protected void initialize(Ruby runtime) {
89
+ replaceEntities = runtime.getTrue();
90
+ try {
91
+ parser = createParser();
92
+ } catch (SAXException se) {
93
+ throw RaiseException.createNativeRaiseException(runtime, se);
94
+ }
95
+ }
96
+
97
+ /**
98
+ * Create and return a copy of this object.
99
+ *
100
+ * @return a clone of this object
101
+ */
102
+ @Override
103
+ public Object clone() throws CloneNotSupportedException {
104
+ return super.clone();
105
+ }
106
+
107
+ protected AbstractSAXParser createParser() throws SAXException {
108
+ XmlSaxParser parser = new XmlSaxParser();
109
+ parser.setFeature(FEATURE_NAMESPACE_PREFIXES, true);
110
+ parser.setFeature(FEATURE_LOAD_EXTERNAL_DTD, false);
111
+ return parser;
112
+ }
113
+
114
+ /**
115
+ * Create a new parser context that will parse the string
116
+ * <code>data</code>.
117
+ */
118
+ @JRubyMethod(name="memory", meta=true)
119
+ public static IRubyObject parse_memory(ThreadContext context,
120
+ IRubyObject klazz,
121
+ IRubyObject data) {
122
+ XmlSaxParserContext ctx = (XmlSaxParserContext) NokogiriService.XML_SAXPARSER_CONTEXT_ALLOCATOR.allocate(context.getRuntime(), (RubyClass) klazz);
123
+ ctx.initialize(context.getRuntime());
124
+ ctx.setInputSource(context, data, context.getRuntime().getNil());
125
+ return ctx;
126
+ }
127
+
128
+ /**
129
+ * Create a new parser context that will read from the file
130
+ * <code>data</code> and parse.
131
+ */
132
+ @JRubyMethod(name="file", meta=true)
133
+ public static IRubyObject parse_file(ThreadContext context,
134
+ IRubyObject klazz,
135
+ IRubyObject data) {
136
+ XmlSaxParserContext ctx = (XmlSaxParserContext) NokogiriService.XML_SAXPARSER_CONTEXT_ALLOCATOR.allocate(context.getRuntime(), (RubyClass) klazz);
137
+ ctx.initialize(context.getRuntime());
138
+ ctx.setInputSourceFile(context, data);
139
+ return ctx;
140
+ }
141
+
142
+ /**
143
+ * Create a new parser context that will read from the IO or
144
+ * StringIO <code>data</code> and parse.
145
+ *
146
+ * TODO: Currently ignores encoding <code>enc</code>.
147
+ */
148
+ @JRubyMethod(name="io", meta=true)
149
+ public static IRubyObject parse_io(ThreadContext context,
150
+ IRubyObject klazz,
151
+ IRubyObject data,
152
+ IRubyObject enc) {
153
+ //int encoding = (int)enc.convertToInteger().getLongValue();
154
+ XmlSaxParserContext ctx = (XmlSaxParserContext) NokogiriService.XML_SAXPARSER_CONTEXT_ALLOCATOR.allocate(context.getRuntime(), (RubyClass) klazz);
155
+ ctx.initialize(context.getRuntime());
156
+ ctx.setInputSource(context, data, context.getRuntime().getNil());
157
+ return ctx;
158
+ }
159
+
160
+ /**
161
+ * Create a new parser context that will read from a raw input
162
+ * stream. Not a JRuby method. Meant to be run in a separate
163
+ * thread by XmlSaxPushParser.
164
+ */
165
+ public static IRubyObject parse_stream(ThreadContext context,
166
+ IRubyObject klazz,
167
+ InputStream stream) {
168
+ XmlSaxParserContext ctx = (XmlSaxParserContext) NokogiriService.XML_SAXPARSER_CONTEXT_ALLOCATOR.allocate(context.getRuntime(), (RubyClass) klazz);
169
+ ctx.initialize(context.getRuntime());
170
+ ctx.setInputSource(stream);
171
+ return ctx;
172
+ }
173
+
174
+ /**
175
+ * Set a property of the underlying parser.
176
+ */
177
+ protected void setProperty(String key, Object val)
178
+ throws SAXNotRecognizedException, SAXNotSupportedException {
179
+ parser.setProperty(key, val);
180
+ }
181
+
182
+ protected void setContentHandler(ContentHandler handler) {
183
+ parser.setContentHandler(handler);
184
+ }
185
+
186
+ protected void setErrorHandler(ErrorHandler handler) {
187
+ parser.setErrorHandler(handler);
188
+ }
189
+
190
+ public NokogiriHandler getNokogiriHandler() {
191
+ return handler;
192
+ }
193
+
194
+ /**
195
+ * Perform any initialization prior to parsing with the handler
196
+ * <code>handlerRuby</code>. Convenience hook for subclasses.
197
+ */
198
+ protected void preParse(ThreadContext context,
199
+ IRubyObject handlerRuby,
200
+ NokogiriHandler handler) {
201
+ ((XmlSaxParser) parser).setXmlDeclHandler(handler);
202
+ }
203
+
204
+ protected void postParse(ThreadContext context,
205
+ IRubyObject handlerRuby,
206
+ NokogiriHandler handler) {
207
+ // noop
208
+ }
209
+
210
+ protected void do_parse() throws SAXException, IOException {
211
+ parser.parse(getInputSource());
212
+ }
213
+
214
+ @JRubyMethod
215
+ public IRubyObject parse_with(ThreadContext context,
216
+ IRubyObject handlerRuby) {
217
+ Ruby ruby = context.getRuntime();
218
+
219
+ if(!invoke(context, handlerRuby, "respond_to?",
220
+ ruby.newSymbol("document")).isTrue()) {
221
+ String msg = "argument must respond_to document";
222
+ throw ruby.newArgumentError(msg);
223
+ }
224
+
225
+ handler = new NokogiriHandler(ruby, handlerRuby);
226
+ preParse(context, handlerRuby, handler);
227
+
228
+ setContentHandler(handler);
229
+ setErrorHandler(handler);
230
+
231
+ try{
232
+ setProperty("http://xml.org/sax/properties/lexical-handler",
233
+ handler);
234
+ } catch(Exception ex) {
235
+ throw ruby.newRuntimeError(
236
+ "Problem while creating XML SAX Parser: " + ex.toString());
237
+ }
238
+
239
+ try{
240
+ try {
241
+ do_parse();
242
+ } catch(SAXParseException spe) {
243
+ // A bad document (<foo><bar></foo>) should call the
244
+ // error handler instead of raising a SAX exception.
245
+
246
+ // However, an EMPTY document should raise a
247
+ // RuntimeError. This is a bit kludgy, but AFAIK SAX
248
+ // doesn't distinguish between empty and bad whereas
249
+ // Nokogiri does.
250
+ String message = spe.getMessage();
251
+ if ("Premature end of file.".matches(message) && stringDataSize < 1) {
252
+ throw ruby.newRuntimeError(
253
+ "couldn't parse document: " + message);
254
+ } else {
255
+ handler.error(spe);
256
+ }
257
+
258
+ }
259
+ } catch(SAXException se) {
260
+ throw RaiseException.createNativeRaiseException(ruby, se);
261
+ } catch(IOException ioe) {
262
+ throw ruby.newIOErrorFromException(ioe);
263
+ }
264
+
265
+ postParse(context, handlerRuby, handler);
266
+
267
+ //maybeTrimLeadingAndTrailingWhitespace(context, handlerRuby);
268
+
269
+ return ruby.getNil();
270
+ }
271
+
272
+ /**
273
+ * Can take a boolean assignment.
274
+ *
275
+ * @param context
276
+ * @param value
277
+ * @return
278
+ */
279
+ @JRubyMethod(name = "replace_entities=")
280
+ public IRubyObject set_replace_entities(ThreadContext context,
281
+ IRubyObject value) {
282
+ if (!value.isTrue()) replaceEntities = context.getRuntime().getFalse();
283
+ else replaceEntities = context.getRuntime().getTrue();
284
+
285
+ return this;
286
+ }
287
+
288
+ @JRubyMethod(name="replace_entities")
289
+ public IRubyObject get_replace_entities(ThreadContext context) {
290
+ return replaceEntities;
291
+ }
292
+
293
+
294
+ /**
295
+ * If the handler's document is a FragmentHandler, attempt to trim
296
+ * leading and trailing whitespace.
297
+ *
298
+ * This is a bit hackish and depends heavily on the internals of
299
+ * FragmentHandler.
300
+ */
301
+ protected void maybeTrimLeadingAndTrailingWhitespace(ThreadContext context,
302
+ IRubyObject parser) {
303
+ final String path = "Nokogiri::XML::FragmentHandler";
304
+ RubyObjectAdapter adapter = JavaEmbedUtils.newObjectAdapter();
305
+ RubyModule mod =
306
+ context.getRuntime().getClassFromPath(path);
307
+
308
+ IRubyObject handler = adapter.getInstanceVariable(parser, "@document");
309
+ if (handler == null || handler.isNil() || !adapter.isKindOf(handler, mod))
310
+ return;
311
+ IRubyObject stack = adapter.getInstanceVariable(handler, "@stack");
312
+ if (stack == null || stack.isNil())
313
+ return;
314
+ // doc is finally a DocumentFragment whose nodes we can check
315
+ IRubyObject doc = adapter.callMethod(stack, "first");
316
+ if (doc == null || doc.isNil())
317
+ return;
318
+
319
+ IRubyObject children;
320
+
321
+ for (;;) {
322
+ children = adapter.callMethod(doc, "children");
323
+ IRubyObject first = adapter.callMethod(children, "first");
324
+ if (isWhitespaceText(context, first))
325
+ adapter.callMethod(first, "unlink");
326
+ else
327
+ break;
328
+ }
329
+
330
+ for (;;) {
331
+ children = adapter.callMethod(doc, "children");
332
+ IRubyObject last = adapter.callMethod(children, "last");
333
+ if (isWhitespaceText(context, last))
334
+ adapter.callMethod(last, "unlink");
335
+ else
336
+ break;
337
+ }
338
+
339
+ // While we have a document, normalize it.
340
+ ((XmlNode) doc).normalize();
341
+ }
342
+
343
+ @JRubyMethod(name="column")
344
+ public IRubyObject column(ThreadContext context) {
345
+ Integer number = handler.getColumn();
346
+ if (number == null) return context.getRuntime().getNil();
347
+ else return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
348
+ }
349
+
350
+ @JRubyMethod(name="line")
351
+ public IRubyObject line(ThreadContext context) {
352
+ Integer number = handler.getLine();
353
+ if (number == null) return context.getRuntime().getNil();
354
+ else return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
355
+ }
356
+ }