nokogiri-backupify 1.5.0.beta.4

Sign up to get free protection for your applications and to get access to all the features.
Files changed (275) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +509 -0
  3. data/CHANGELOG.rdoc +490 -0
  4. data/Manifest.txt +274 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +150 -0
  7. data/Rakefile +217 -0
  8. data/bin/nokogiri +54 -0
  9. data/deps.rip +5 -0
  10. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  11. data/ext/java/nokogiri/HtmlDocument.java +146 -0
  12. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  13. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  14. data/ext/java/nokogiri/HtmlSaxParserContext.java +256 -0
  15. data/ext/java/nokogiri/NokogiriService.java +466 -0
  16. data/ext/java/nokogiri/XmlAttr.java +183 -0
  17. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  18. data/ext/java/nokogiri/XmlCdata.java +89 -0
  19. data/ext/java/nokogiri/XmlComment.java +84 -0
  20. data/ext/java/nokogiri/XmlDocument.java +514 -0
  21. data/ext/java/nokogiri/XmlDocumentFragment.java +216 -0
  22. data/ext/java/nokogiri/XmlDtd.java +464 -0
  23. data/ext/java/nokogiri/XmlElement.java +221 -0
  24. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  25. data/ext/java/nokogiri/XmlElementDecl.java +147 -0
  26. data/ext/java/nokogiri/XmlEntityDecl.java +161 -0
  27. data/ext/java/nokogiri/XmlEntityReference.java +75 -0
  28. data/ext/java/nokogiri/XmlNamespace.java +127 -0
  29. data/ext/java/nokogiri/XmlNode.java +1392 -0
  30. data/ext/java/nokogiri/XmlNodeSet.java +284 -0
  31. data/ext/java/nokogiri/XmlProcessingInstruction.java +103 -0
  32. data/ext/java/nokogiri/XmlReader.java +409 -0
  33. data/ext/java/nokogiri/XmlRelaxng.java +199 -0
  34. data/ext/java/nokogiri/XmlSaxParserContext.java +353 -0
  35. data/ext/java/nokogiri/XmlSaxPushParser.java +182 -0
  36. data/ext/java/nokogiri/XmlSchema.java +175 -0
  37. data/ext/java/nokogiri/XmlSyntaxError.java +114 -0
  38. data/ext/java/nokogiri/XmlText.java +135 -0
  39. data/ext/java/nokogiri/XmlXpathContext.java +175 -0
  40. data/ext/java/nokogiri/XsltStylesheet.java +181 -0
  41. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +205 -0
  42. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  43. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +80 -0
  44. data/ext/java/nokogiri/internals/NokogiriHandler.java +326 -0
  45. data/ext/java/nokogiri/internals/NokogiriHelpers.java +583 -0
  46. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +170 -0
  47. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +118 -0
  48. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +73 -0
  49. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  50. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
  51. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +120 -0
  52. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +56 -0
  53. data/ext/java/nokogiri/internals/ParserContext.java +278 -0
  54. data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
  55. data/ext/java/nokogiri/internals/ReaderNode.java +473 -0
  56. data/ext/java/nokogiri/internals/SaveContext.java +282 -0
  57. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +68 -0
  58. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  59. data/ext/java/nokogiri/internals/XmlDomParser.java +77 -0
  60. data/ext/java/nokogiri/internals/XmlDomParserContext.java +233 -0
  61. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  62. data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
  63. data/ext/nokogiri/depend +358 -0
  64. data/ext/nokogiri/extconf.rb +124 -0
  65. data/ext/nokogiri/html_document.c +154 -0
  66. data/ext/nokogiri/html_document.h +10 -0
  67. data/ext/nokogiri/html_element_description.c +276 -0
  68. data/ext/nokogiri/html_element_description.h +10 -0
  69. data/ext/nokogiri/html_entity_lookup.c +32 -0
  70. data/ext/nokogiri/html_entity_lookup.h +8 -0
  71. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  72. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  73. data/ext/nokogiri/nokogiri.c +92 -0
  74. data/ext/nokogiri/nokogiri.h +160 -0
  75. data/ext/nokogiri/xml_attr.c +94 -0
  76. data/ext/nokogiri/xml_attr.h +9 -0
  77. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  78. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  79. data/ext/nokogiri/xml_cdata.c +56 -0
  80. data/ext/nokogiri/xml_cdata.h +9 -0
  81. data/ext/nokogiri/xml_comment.c +54 -0
  82. data/ext/nokogiri/xml_comment.h +9 -0
  83. data/ext/nokogiri/xml_document.c +478 -0
  84. data/ext/nokogiri/xml_document.h +23 -0
  85. data/ext/nokogiri/xml_document_fragment.c +48 -0
  86. data/ext/nokogiri/xml_document_fragment.h +10 -0
  87. data/ext/nokogiri/xml_dtd.c +202 -0
  88. data/ext/nokogiri/xml_dtd.h +10 -0
  89. data/ext/nokogiri/xml_element_content.c +123 -0
  90. data/ext/nokogiri/xml_element_content.h +10 -0
  91. data/ext/nokogiri/xml_element_decl.c +69 -0
  92. data/ext/nokogiri/xml_element_decl.h +9 -0
  93. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  94. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  95. data/ext/nokogiri/xml_entity_decl.c +110 -0
  96. data/ext/nokogiri/xml_entity_decl.h +10 -0
  97. data/ext/nokogiri/xml_entity_reference.c +52 -0
  98. data/ext/nokogiri/xml_entity_reference.h +9 -0
  99. data/ext/nokogiri/xml_io.c +31 -0
  100. data/ext/nokogiri/xml_io.h +11 -0
  101. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  102. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  103. data/ext/nokogiri/xml_namespace.c +84 -0
  104. data/ext/nokogiri/xml_namespace.h +13 -0
  105. data/ext/nokogiri/xml_node.c +1384 -0
  106. data/ext/nokogiri/xml_node.h +13 -0
  107. data/ext/nokogiri/xml_node_set.c +418 -0
  108. data/ext/nokogiri/xml_node_set.h +9 -0
  109. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  110. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  111. data/ext/nokogiri/xml_reader.c +684 -0
  112. data/ext/nokogiri/xml_reader.h +10 -0
  113. data/ext/nokogiri/xml_relax_ng.c +161 -0
  114. data/ext/nokogiri/xml_relax_ng.h +9 -0
  115. data/ext/nokogiri/xml_sax_parser.c +288 -0
  116. data/ext/nokogiri/xml_sax_parser.h +39 -0
  117. data/ext/nokogiri/xml_sax_parser_context.c +199 -0
  118. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  119. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  120. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  121. data/ext/nokogiri/xml_schema.c +205 -0
  122. data/ext/nokogiri/xml_schema.h +9 -0
  123. data/ext/nokogiri/xml_syntax_error.c +58 -0
  124. data/ext/nokogiri/xml_syntax_error.h +13 -0
  125. data/ext/nokogiri/xml_text.c +50 -0
  126. data/ext/nokogiri/xml_text.h +9 -0
  127. data/ext/nokogiri/xml_xpath_context.c +309 -0
  128. data/ext/nokogiri/xml_xpath_context.h +9 -0
  129. data/ext/nokogiri/xslt_stylesheet.c +258 -0
  130. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  131. data/lib/isorelax.jar +0 -0
  132. data/lib/jing.jar +0 -0
  133. data/lib/nekodtd.jar +0 -0
  134. data/lib/nekohtml.jar +0 -0
  135. data/lib/nokogiri.rb +143 -0
  136. data/lib/nokogiri/css.rb +23 -0
  137. data/lib/nokogiri/css/node.rb +99 -0
  138. data/lib/nokogiri/css/parser.rb +677 -0
  139. data/lib/nokogiri/css/parser.y +237 -0
  140. data/lib/nokogiri/css/parser_extras.rb +91 -0
  141. data/lib/nokogiri/css/syntax_error.rb +7 -0
  142. data/lib/nokogiri/css/tokenizer.rb +152 -0
  143. data/lib/nokogiri/css/tokenizer.rex +55 -0
  144. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  145. data/lib/nokogiri/decorators/slop.rb +35 -0
  146. data/lib/nokogiri/html.rb +36 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +221 -0
  149. data/lib/nokogiri/html/document_fragment.rb +41 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  152. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  153. data/lib/nokogiri/html/sax/parser.rb +52 -0
  154. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  155. data/lib/nokogiri/syntax_error.rb +4 -0
  156. data/lib/nokogiri/version.rb +35 -0
  157. data/lib/nokogiri/xml.rb +67 -0
  158. data/lib/nokogiri/xml/attr.rb +14 -0
  159. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  160. data/lib/nokogiri/xml/builder.rb +418 -0
  161. data/lib/nokogiri/xml/cdata.rb +11 -0
  162. data/lib/nokogiri/xml/character_data.rb +7 -0
  163. data/lib/nokogiri/xml/document.rb +218 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +84 -0
  165. data/lib/nokogiri/xml/dtd.rb +22 -0
  166. data/lib/nokogiri/xml/element_content.rb +36 -0
  167. data/lib/nokogiri/xml/element_decl.rb +13 -0
  168. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  169. data/lib/nokogiri/xml/namespace.rb +13 -0
  170. data/lib/nokogiri/xml/node.rb +907 -0
  171. data/lib/nokogiri/xml/node/save_options.rb +45 -0
  172. data/lib/nokogiri/xml/node_set.rb +350 -0
  173. data/lib/nokogiri/xml/notation.rb +6 -0
  174. data/lib/nokogiri/xml/parse_options.rb +85 -0
  175. data/lib/nokogiri/xml/pp.rb +2 -0
  176. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  177. data/lib/nokogiri/xml/pp/node.rb +56 -0
  178. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  179. data/lib/nokogiri/xml/reader.rb +112 -0
  180. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  181. data/lib/nokogiri/xml/sax.rb +4 -0
  182. data/lib/nokogiri/xml/sax/document.rb +164 -0
  183. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  184. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  185. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  186. data/lib/nokogiri/xml/schema.rb +57 -0
  187. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  188. data/lib/nokogiri/xml/text.rb +9 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +52 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xercesImpl.jar +0 -0
  195. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  196. data/tasks/cross_compile.rb +177 -0
  197. data/tasks/test.rb +94 -0
  198. data/test/css/test_nthiness.rb +159 -0
  199. data/test/css/test_parser.rb +303 -0
  200. data/test/css/test_tokenizer.rb +198 -0
  201. data/test/css/test_xpath_visitor.rb +85 -0
  202. data/test/decorators/test_slop.rb +16 -0
  203. data/test/files/2ch.html +108 -0
  204. data/test/files/address_book.rlx +12 -0
  205. data/test/files/address_book.xml +10 -0
  206. data/test/files/bar/bar.xsd +4 -0
  207. data/test/files/dont_hurt_em_why.xml +422 -0
  208. data/test/files/exslt.xml +8 -0
  209. data/test/files/exslt.xslt +35 -0
  210. data/test/files/foo/foo.xsd +4 -0
  211. data/test/files/po.xml +32 -0
  212. data/test/files/po.xsd +66 -0
  213. data/test/files/shift_jis.html +10 -0
  214. data/test/files/shift_jis.xml +5 -0
  215. data/test/files/snuggles.xml +3 -0
  216. data/test/files/staff.dtd +10 -0
  217. data/test/files/staff.xml +59 -0
  218. data/test/files/staff.xslt +32 -0
  219. data/test/files/tlm.html +850 -0
  220. data/test/files/valid_bar.xml +2 -0
  221. data/test/helper.rb +171 -0
  222. data/test/html/sax/test_parser.rb +136 -0
  223. data/test/html/sax/test_parser_context.rb +48 -0
  224. data/test/html/test_builder.rb +164 -0
  225. data/test/html/test_document.rb +457 -0
  226. data/test/html/test_document_encoding.rb +123 -0
  227. data/test/html/test_document_fragment.rb +255 -0
  228. data/test/html/test_element_description.rb +100 -0
  229. data/test/html/test_named_characters.rb +14 -0
  230. data/test/html/test_node.rb +190 -0
  231. data/test/html/test_node_encoding.rb +27 -0
  232. data/test/test_convert_xpath.rb +135 -0
  233. data/test/test_css_cache.rb +45 -0
  234. data/test/test_encoding_handler.rb +46 -0
  235. data/test/test_memory_leak.rb +52 -0
  236. data/test/test_nokogiri.rb +132 -0
  237. data/test/test_reader.rb +403 -0
  238. data/test/test_soap4r_sax.rb +52 -0
  239. data/test/test_xslt_transforms.rb +189 -0
  240. data/test/xml/node/test_save_options.rb +20 -0
  241. data/test/xml/node/test_subclass.rb +44 -0
  242. data/test/xml/sax/test_parser.rb +338 -0
  243. data/test/xml/sax/test_parser_context.rb +113 -0
  244. data/test/xml/sax/test_push_parser.rb +156 -0
  245. data/test/xml/test_attr.rb +65 -0
  246. data/test/xml/test_attribute_decl.rb +86 -0
  247. data/test/xml/test_builder.rb +210 -0
  248. data/test/xml/test_cdata.rb +50 -0
  249. data/test/xml/test_comment.rb +29 -0
  250. data/test/xml/test_document.rb +675 -0
  251. data/test/xml/test_document_encoding.rb +26 -0
  252. data/test/xml/test_document_fragment.rb +192 -0
  253. data/test/xml/test_dtd.rb +107 -0
  254. data/test/xml/test_dtd_encoding.rb +33 -0
  255. data/test/xml/test_element_content.rb +56 -0
  256. data/test/xml/test_element_decl.rb +73 -0
  257. data/test/xml/test_entity_decl.rb +122 -0
  258. data/test/xml/test_entity_reference.rb +21 -0
  259. data/test/xml/test_namespace.rb +70 -0
  260. data/test/xml/test_node.rb +899 -0
  261. data/test/xml/test_node_attributes.rb +34 -0
  262. data/test/xml/test_node_encoding.rb +107 -0
  263. data/test/xml/test_node_reparenting.rb +321 -0
  264. data/test/xml/test_node_set.rb +708 -0
  265. data/test/xml/test_parse_options.rb +52 -0
  266. data/test/xml/test_processing_instruction.rb +30 -0
  267. data/test/xml/test_reader_encoding.rb +126 -0
  268. data/test/xml/test_relax_ng.rb +60 -0
  269. data/test/xml/test_schema.rb +89 -0
  270. data/test/xml/test_syntax_error.rb +12 -0
  271. data/test/xml/test_text.rb +47 -0
  272. data/test/xml/test_unparented_node.rb +381 -0
  273. data/test/xml/test_xpath.rb +237 -0
  274. data/test/xslt/test_custom_functions.rb +94 -0
  275. metadata +525 -0
@@ -0,0 +1,199 @@
1
+ /**
2
+ * (The MIT License)
3
+ *
4
+ * Copyright (c) 2008 - 2010:
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 java.io.ByteArrayInputStream;
36
+ import java.io.IOException;
37
+ import java.io.InputStream;
38
+ import java.io.StringWriter;
39
+ import java.io.UnsupportedEncodingException;
40
+
41
+ import javax.xml.transform.TransformerConfigurationException;
42
+ import javax.xml.transform.TransformerException;
43
+ import javax.xml.transform.TransformerFactory;
44
+ import javax.xml.transform.dom.DOMSource;
45
+ import javax.xml.transform.stream.StreamResult;
46
+ import javax.xml.transform.stream.StreamSource;
47
+
48
+ import nokogiri.internals.SchemaErrorHandler;
49
+
50
+ import org.iso_relax.verifier.Schema;
51
+ import org.iso_relax.verifier.Verifier;
52
+ import org.iso_relax.verifier.VerifierConfigurationException;
53
+ import org.iso_relax.verifier.VerifierFactory;
54
+ import org.jruby.Ruby;
55
+ import org.jruby.RubyArray;
56
+ import org.jruby.RubyClass;
57
+ import org.jruby.anno.JRubyClass;
58
+ import org.jruby.anno.JRubyMethod;
59
+ import org.jruby.runtime.ThreadContext;
60
+ import org.jruby.runtime.Visibility;
61
+ import org.jruby.runtime.builtin.IRubyObject;
62
+ import org.w3c.dom.Document;
63
+ import org.xml.sax.ErrorHandler;
64
+ import org.xml.sax.SAXException;
65
+
66
+ /**
67
+ * Class for Nokogiri::XML::RelaxNG
68
+ *
69
+ * @author sergio
70
+ */
71
+ @JRubyClass(name="Nokogiri::XML::RelaxNG", parent="Nokogiri::XML::Schema")
72
+ public class XmlRelaxng extends XmlSchema{
73
+
74
+ public XmlRelaxng(Ruby ruby, RubyClass klazz) {
75
+ super(ruby, klazz);
76
+ }
77
+
78
+ private Schema getSchema(ThreadContext context) {
79
+ InputStream is = null;
80
+ VerifierFactory factory = new com.thaiopensource.relaxng.jarv.VerifierFactoryImpl();
81
+ if(this.source instanceof StreamSource) {
82
+ StreamSource ss = (StreamSource) this.source;
83
+ is = ss.getInputStream();
84
+ } else /*if (this.source instanceof DOMSource)*/{
85
+ DOMSource ds = (DOMSource) this.source;
86
+ StringWriter xmlAsWriter = new StringWriter();
87
+ StreamResult result = new StreamResult(xmlAsWriter);
88
+ try {
89
+ TransformerFactory.newInstance().newTransformer().transform(ds, result);
90
+ } catch (TransformerConfigurationException ex) {
91
+ throw context.getRuntime()
92
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
93
+ } catch (TransformerException ex) {
94
+ throw context.getRuntime()
95
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
96
+ }
97
+ try {
98
+ is = new ByteArrayInputStream(xmlAsWriter.toString().getBytes("UTF-8"));
99
+ } catch (UnsupportedEncodingException ex) {
100
+ throw context.getRuntime()
101
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
102
+ }
103
+ }
104
+
105
+ try {
106
+ return factory.compileSchema(is);
107
+ } catch (VerifierConfigurationException ex) {
108
+ throw context.getRuntime()
109
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
110
+ } catch (SAXException ex) {
111
+ throw context.getRuntime()
112
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
113
+ } catch (IOException ex) {
114
+ throw context.getRuntime().newIOError(ex.getMessage());
115
+ }
116
+ }
117
+
118
+ //
119
+ // protected static XmlSchema createSchemaFromSource(ThreadContext context,
120
+ // IRubyObject klazz, Source source) {
121
+ //
122
+ // Ruby ruby = context.getRuntime();
123
+ //
124
+ // XmlSchema schema = new XmlSchema(ruby, (RubyClass) klazz);
125
+ //
126
+ // try {
127
+ // schema.schema = SchemaFactory.newInstance(XMLConstants.W3C_XML_SCHEMA_NS_URI)
128
+ // .newSchema(source);
129
+ // } catch(SAXException ex) {
130
+ // throw ruby.newRuntimeError("Could not parse document: "+ex.getMessage());
131
+ // }
132
+ //
133
+ // schema.setInstanceVariable("@errors", ruby.newEmptyArray());
134
+ //
135
+ // return schema;
136
+ // }
137
+ //
138
+ // @JRubyMethod(meta=true)
139
+ // public static IRubyObject from_document(ThreadContext context,
140
+ // IRubyObject klazz, IRubyObject document) {
141
+ // XmlDocument doc = ((XmlDocument) ((XmlNode) document).document(context));
142
+ //
143
+ // RubyArray errors = (RubyArray) doc.getInstanceVariable("@errors");
144
+ //
145
+ // if(!errors.isEmpty()) {
146
+ // throw new RaiseException((XmlSyntaxError) errors.first());
147
+ // }
148
+ //
149
+ // DOMSource source = new DOMSource(doc.getDocument());
150
+ //
151
+ // IRubyObject uri = doc.url(context);
152
+ //
153
+ // if(!uri.isNil()) {
154
+ // source.setSystemId(uri.convertToString().asJavaString());
155
+ // }
156
+ //
157
+ // return createSchemaFromSource(context, klazz, source);
158
+ // }
159
+ //
160
+ // @JRubyMethod(meta=true)
161
+ // public static IRubyObject read_memory(ThreadContext context,
162
+ // IRubyObject klazz, IRubyObject content) {
163
+ //
164
+ // String data = content.convertToString().asJavaString();
165
+ //
166
+ // return createSchemaFromSource(context, klazz,
167
+ // new StreamSource(new StringReader(data)));
168
+ // }
169
+ //
170
+ @Override
171
+ @JRubyMethod(visibility=Visibility.PRIVATE)
172
+ public IRubyObject validate_document(ThreadContext context, IRubyObject document) {
173
+ Ruby ruby = context.getRuntime();
174
+
175
+ Document doc = ((XmlDocument) document).getDocument();
176
+
177
+ Schema schema = this.getSchema(context);
178
+
179
+ Verifier verifier;
180
+ try {
181
+ verifier = schema.newVerifier();
182
+ } catch (VerifierConfigurationException ex) {
183
+ throw context.getRuntime()
184
+ .newRuntimeError("Could not parse document: "+ex.getMessage());
185
+ }
186
+
187
+ RubyArray errors = (RubyArray) this.getInstanceVariable("@errors");
188
+ ErrorHandler errorHandler = new SchemaErrorHandler(ruby, errors);
189
+
190
+ verifier.setErrorHandler(errorHandler);
191
+ try {
192
+ verifier.verify(doc);
193
+ } catch (SAXException ex) {
194
+ errors.append(new XmlSyntaxError(ruby, ex));
195
+ }
196
+
197
+ return errors;
198
+ }
199
+ }
@@ -0,0 +1,353 @@
1
+ /**
2
+ * (The MIT License)
3
+ *
4
+ * Copyright (c) 2008 - 2010:
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.rubyStringToString;
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
+ */
69
+ @JRubyClass(name="Nokogiri::XML::SAX::ParserContext")
70
+ public class XmlSaxParserContext extends ParserContext {
71
+ protected static final String FEATURE_NAMESPACES =
72
+ "http://xml.org/sax/features/namespaces";
73
+ protected static final String FEATURE_NAMESPACE_PREFIXES =
74
+ "http://xml.org/sax/features/namespace-prefixes";
75
+ protected static final String FEATURE_LOAD_EXTERNAL_DTD =
76
+ "http://apache.org/xml/features/nonvalidating/load-external-dtd";
77
+
78
+ protected AbstractSAXParser parser;
79
+
80
+ protected NokogiriHandler handler = null;
81
+ private IRubyObject replaceEntities;
82
+
83
+ public XmlSaxParserContext(final Ruby ruby, RubyClass rubyClass) {
84
+ super(ruby, rubyClass);
85
+ replaceEntities = ruby.getTrue();
86
+ try {
87
+ parser = createParser();
88
+ } catch (SAXException se) {
89
+ throw RaiseException.createNativeRaiseException(ruby, se);
90
+ }
91
+ }
92
+
93
+ protected AbstractSAXParser createParser() throws SAXException {
94
+ XmlSaxParser parser = new XmlSaxParser();
95
+ parser.setFeature(FEATURE_NAMESPACE_PREFIXES, true);
96
+ parser.setFeature(FEATURE_LOAD_EXTERNAL_DTD, false);
97
+ return parser;
98
+ }
99
+
100
+ /**
101
+ * Create a new parser context that will parse the string
102
+ * <code>data</code>.
103
+ */
104
+ @JRubyMethod(name="memory", meta=true)
105
+ public static IRubyObject parse_memory(ThreadContext context,
106
+ IRubyObject klazz,
107
+ IRubyObject data) {
108
+ XmlSaxParserContext ctx = new XmlSaxParserContext(context.getRuntime(),
109
+ (RubyClass) klazz);
110
+ ctx.setInputSource(context, data);
111
+ return ctx;
112
+ }
113
+
114
+ /**
115
+ * Create a new parser context that will read from the file
116
+ * <code>data</code> and parse.
117
+ */
118
+ @JRubyMethod(name="file", meta=true)
119
+ public static IRubyObject parse_file(ThreadContext context,
120
+ IRubyObject klazz,
121
+ IRubyObject data) {
122
+ XmlSaxParserContext ctx = new XmlSaxParserContext(context.getRuntime(),
123
+ (RubyClass) klazz);
124
+ ctx.setInputSourceFile(context, data);
125
+ return ctx;
126
+ }
127
+
128
+ /**
129
+ * Create a new parser context that will read from the IO or
130
+ * StringIO <code>data</code> and parse.
131
+ *
132
+ * TODO: Currently ignores encoding <code>enc</code>.
133
+ */
134
+ @JRubyMethod(name="io", meta=true)
135
+ public static IRubyObject parse_io(ThreadContext context,
136
+ IRubyObject klazz,
137
+ IRubyObject data,
138
+ IRubyObject enc) {
139
+ //int encoding = (int)enc.convertToInteger().getLongValue();
140
+ XmlSaxParserContext ctx = new XmlSaxParserContext(context.getRuntime(),
141
+ (RubyClass) klazz);
142
+ ctx.setInputSource(context, data);
143
+ return ctx;
144
+ }
145
+
146
+ /**
147
+ * Create a new parser context that will read from a raw input
148
+ * stream. Not a JRuby method. Meant to be run in a separate
149
+ * thread by XmlSaxPushParser.
150
+ */
151
+ public static IRubyObject parse_stream(ThreadContext context,
152
+ IRubyObject klazz,
153
+ InputStream stream) {
154
+ XmlSaxParserContext ctx =
155
+ new XmlSaxParserContext(context.getRuntime(), (RubyClass)klazz);
156
+ ctx.setInputSource(stream);
157
+ return ctx;
158
+ }
159
+
160
+ /**
161
+ * Set a property of the underlying parser.
162
+ */
163
+ protected void setProperty(String key, Object val)
164
+ throws SAXNotRecognizedException, SAXNotSupportedException {
165
+ parser.setProperty(key, val);
166
+ }
167
+
168
+ protected void setContentHandler(ContentHandler handler) {
169
+ parser.setContentHandler(handler);
170
+ }
171
+
172
+ protected void setErrorHandler(ErrorHandler handler) {
173
+ parser.setErrorHandler(handler);
174
+ }
175
+
176
+ public NokogiriHandler getNokogiriHandler() {
177
+ return handler;
178
+ }
179
+
180
+ /**
181
+ * Perform any initialization prior to parsing with the handler
182
+ * <code>handlerRuby</code>. Convenience hook for subclasses.
183
+ */
184
+ protected void preParse(ThreadContext context,
185
+ IRubyObject handlerRuby,
186
+ NokogiriHandler handler) {
187
+ ((XmlSaxParser) parser).setXmlDeclHandler(handler);
188
+ }
189
+
190
+ protected void postParse(ThreadContext context,
191
+ IRubyObject handlerRuby,
192
+ NokogiriHandler handler) {
193
+ // noop
194
+ }
195
+
196
+ protected void do_parse() throws SAXException, IOException {
197
+ parser.parse(getInputSource());
198
+ }
199
+
200
+ @JRubyMethod
201
+ public IRubyObject parse_with(ThreadContext context,
202
+ IRubyObject handlerRuby) {
203
+ Ruby ruby = context.getRuntime();
204
+
205
+ if(!invoke(context, handlerRuby, "respond_to?",
206
+ ruby.newSymbol("document")).isTrue()) {
207
+ String msg = "argument must respond_to document";
208
+ throw ruby.newArgumentError(msg);
209
+ }
210
+
211
+ handler = new NokogiriHandler(ruby, handlerRuby);
212
+ preParse(context, handlerRuby, handler);
213
+
214
+ setContentHandler(handler);
215
+ setErrorHandler(handler);
216
+
217
+ try{
218
+ setProperty("http://xml.org/sax/properties/lexical-handler",
219
+ handler);
220
+ } catch(Exception ex) {
221
+ throw ruby.newRuntimeError(
222
+ "Problem while creating XML SAX Parser: " + ex.toString());
223
+ }
224
+
225
+ try{
226
+ try {
227
+ do_parse();
228
+ } catch(SAXParseException spe) {
229
+ // A bad document (<foo><bar></foo>) should call the
230
+ // error handler instead of raising a SAX exception.
231
+
232
+ // However, an EMPTY document should raise a
233
+ // RuntimeError. This is a bit kludgy, but AFAIK SAX
234
+ // doesn't distinguish between empty and bad whereas
235
+ // Nokogiri does.
236
+ String message = spe.getMessage();
237
+ if ("Premature end of file.".matches(message)) {
238
+ throw ruby.newRuntimeError(
239
+ "couldn't parse document: " + message);
240
+ } else {
241
+ handler.error(spe);
242
+ }
243
+
244
+ }
245
+ } catch(SAXException se) {
246
+ throw RaiseException.createNativeRaiseException(ruby, se);
247
+ } catch(IOException ioe) {
248
+ throw ruby.newIOErrorFromException(ioe);
249
+ }
250
+
251
+ postParse(context, handlerRuby, handler);
252
+
253
+ //maybeTrimLeadingAndTrailingWhitespace(context, handlerRuby);
254
+
255
+ return ruby.getNil();
256
+ }
257
+
258
+ /**
259
+ * Can take a boolean assignment.
260
+ *
261
+ * @param context
262
+ * @param value
263
+ * @return
264
+ */
265
+ @JRubyMethod(name = "replace_entities=")
266
+ public IRubyObject set_replace_entities(ThreadContext context,
267
+ IRubyObject value) {
268
+ if (!value.isTrue()) replaceEntities = context.getRuntime().getFalse();
269
+ else replaceEntities = context.getRuntime().getTrue();
270
+
271
+ return this;
272
+ }
273
+
274
+ @JRubyMethod(name="replace_entities")
275
+ public IRubyObject get_replace_entities(ThreadContext context) {
276
+ return replaceEntities;
277
+ }
278
+
279
+
280
+ /**
281
+ * If the handler's document is a FragmentHandler, attempt to trim
282
+ * leading and trailing whitespace.
283
+ *
284
+ * This is a bit hackish and depends heavily on the internals of
285
+ * FragmentHandler.
286
+ */
287
+ protected void maybeTrimLeadingAndTrailingWhitespace(ThreadContext context,
288
+ IRubyObject parser) {
289
+ final String path = "Nokogiri::XML::FragmentHandler";
290
+ RubyObjectAdapter adapter = JavaEmbedUtils.newObjectAdapter();
291
+ RubyModule mod =
292
+ context.getRuntime().getClassFromPath(path);
293
+
294
+ IRubyObject handler = adapter.getInstanceVariable(parser, "@document");
295
+ if (handler == null || handler.isNil() || !adapter.isKindOf(handler, mod))
296
+ return;
297
+ IRubyObject stack = adapter.getInstanceVariable(handler, "@stack");
298
+ if (stack == null || stack.isNil())
299
+ return;
300
+ // doc is finally a DocumentFragment whose nodes we can check
301
+ IRubyObject doc = adapter.callMethod(stack, "first");
302
+ if (doc == null || doc.isNil())
303
+ return;
304
+
305
+ IRubyObject children;
306
+
307
+ for (;;) {
308
+ children = adapter.callMethod(doc, "children");
309
+ IRubyObject first = adapter.callMethod(children, "first");
310
+ if (isWhitespaceText(context, first))
311
+ adapter.callMethod(first, "unlink");
312
+ else
313
+ break;
314
+ }
315
+
316
+ for (;;) {
317
+ children = adapter.callMethod(doc, "children");
318
+ IRubyObject last = adapter.callMethod(children, "last");
319
+ if (isWhitespaceText(context, last))
320
+ adapter.callMethod(last, "unlink");
321
+ else
322
+ break;
323
+ }
324
+
325
+ // While we have a document, normalize it.
326
+ ((XmlNode) doc).normalize();
327
+ }
328
+
329
+ protected boolean isWhitespaceText(ThreadContext context, IRubyObject obj) {
330
+ if (obj == null || obj.isNil()) return false;
331
+
332
+ XmlNode node = (XmlNode) obj;
333
+ if (!(node instanceof XmlText))
334
+ return false;
335
+
336
+ String content = rubyStringToString(node.content(context));
337
+ return content.trim().length() == 0;
338
+ }
339
+
340
+ @JRubyMethod(name="column")
341
+ public IRubyObject column(ThreadContext context) {
342
+ Integer number = handler.getColumn();
343
+ if (number == null) return context.getRuntime().getNil();
344
+ else return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
345
+ }
346
+
347
+ @JRubyMethod(name="line")
348
+ public IRubyObject line(ThreadContext context) {
349
+ Integer number = handler.getLine();
350
+ if (number == null) return context.getRuntime().getNil();
351
+ else return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
352
+ }
353
+ }