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 +1,205 @@
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.internals;
34
+
35
+ import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
36
+ import static nokogiri.internals.NokogiriHelpers.isNamespace;
37
+ import nokogiri.HtmlDocument;
38
+ import nokogiri.XmlDocument;
39
+
40
+ import org.apache.xerces.parsers.DOMParser;
41
+ import org.apache.xerces.xni.Augmentations;
42
+ import org.apache.xerces.xni.QName;
43
+ import org.apache.xerces.xni.XMLAttributes;
44
+ import org.apache.xerces.xni.XNIException;
45
+ import org.apache.xerces.xni.parser.XMLDocumentFilter;
46
+ import org.apache.xerces.xni.parser.XMLParserConfiguration;
47
+ import org.cyberneko.html.HTMLConfiguration;
48
+ import org.cyberneko.html.filters.DefaultFilter;
49
+ import org.jruby.Ruby;
50
+ import org.jruby.RubyClass;
51
+ import org.jruby.runtime.ThreadContext;
52
+ import org.jruby.runtime.builtin.IRubyObject;
53
+ import org.w3c.dom.Document;
54
+
55
+ /**
56
+ * Parser for HtmlDocument. This class actually parses HtmlDocument using NekoHtml.
57
+ *
58
+ * @author sergio
59
+ */
60
+ public class HtmlDomParserContext extends XmlDomParserContext {
61
+
62
+ public HtmlDomParserContext(Ruby runtime, IRubyObject options) {
63
+ super(runtime, options);
64
+ }
65
+
66
+ public HtmlDomParserContext(Ruby runtime, IRubyObject encoding, IRubyObject options) {
67
+ super(runtime, encoding, options);
68
+ }
69
+
70
+ @Override
71
+ protected void initErrorHandler() {
72
+ if (options.strict) {
73
+ errorHandler = new NokogiriStrictErrorHandler(options.noError, options.noWarning);
74
+ } else {
75
+ errorHandler = new NokogiriNonStrictErrorHandler4NekoHtml(options.noError, options.noWarning);
76
+ }
77
+ }
78
+
79
+ @Override
80
+ protected void initParser(Ruby runtime) {
81
+ XMLParserConfiguration config = new HTMLConfiguration();
82
+ XMLDocumentFilter removeNSAttrsFilter = new RemoveNSAttrsFilter();
83
+ XMLDocumentFilter elementValidityCheckFilter = new ElementValidityCheckFilter(errorHandler);
84
+ XMLDocumentFilter[] filters = { removeNSAttrsFilter, elementValidityCheckFilter};
85
+
86
+ config.setErrorHandler(this.errorHandler);
87
+ parser = new DOMParser(config);
88
+
89
+ // see http://nekohtml.sourceforge.net/settings.html for details
90
+ setProperty("http://cyberneko.org/html/properties/default-encoding", java_encoding);
91
+ setProperty("http://cyberneko.org/html/properties/names/elems", "lower");
92
+ setProperty("http://cyberneko.org/html/properties/names/attrs", "lower");
93
+ setFeature("http://cyberneko.org/html/features/report-errors", true);
94
+ setFeature("http://xml.org/sax/features/namespaces", false);
95
+ setProperty("http://cyberneko.org/html/properties/filters", filters);
96
+ setFeature("http://cyberneko.org/html/features/insert-doctype", true);
97
+ }
98
+
99
+ /**
100
+ * Enable NekoHTML feature for balancing tags in a document fragment.
101
+ *
102
+ * This method is used in XmlNode#in_context method.
103
+ */
104
+ public void enableDocumentFragment() {
105
+ setFeature("http://cyberneko.org/html/features/balance-tags/document-fragment", true);
106
+ }
107
+
108
+ @Override
109
+ protected XmlDocument getNewEmptyDocument(ThreadContext context) {
110
+ IRubyObject[] args = new IRubyObject[0];
111
+ return (XmlDocument) XmlDocument.rbNew(context,
112
+ getNokogiriClass(context.getRuntime(), "Nokogiri::HTML::Document"),
113
+ args);
114
+ }
115
+
116
+ @Override
117
+ protected XmlDocument wrapDocument(ThreadContext context,
118
+ RubyClass klass,
119
+ Document doc) {
120
+ HtmlDocument htmlDocument = new HtmlDocument(context.getRuntime(), klass, doc);
121
+ htmlDocument.setEncoding(ruby_encoding);
122
+ return htmlDocument;
123
+ }
124
+
125
+ /**
126
+ * Filter to strip out attributes that pertain to XML namespaces.
127
+ *
128
+ * @author sergio
129
+ * @author Patrick Mahoney <pat@polycrystal.org>
130
+ */
131
+ public static class RemoveNSAttrsFilter extends DefaultFilter {
132
+ @Override
133
+ public void startElement(QName element, XMLAttributes attrs,
134
+ Augmentations augs) throws XNIException {
135
+ int i;
136
+ for (i = 0; i < attrs.getLength(); ++i) {
137
+ if (isNamespace(attrs.getQName(i))) {
138
+ attrs.removeAttributeAt(i);
139
+ --i;
140
+ }
141
+ }
142
+
143
+ element.uri = null;
144
+ super.startElement(element, attrs, augs);
145
+ }
146
+ }
147
+
148
+ public static class ElementValidityCheckFilter extends DefaultFilter {
149
+ private NokogiriErrorHandler errorHandler;
150
+
151
+ private ElementValidityCheckFilter(NokogiriErrorHandler errorHandler) {
152
+ this.errorHandler = errorHandler;
153
+ }
154
+
155
+ // element names from xhtml1-strict.dtd
156
+ private static String[][] element_names = {
157
+ {"a", "abbr", "acronym", "address", "area"},
158
+ {"b", "base", "basefont", "bdo", "big", "blockquote", "body", "br", "button"},
159
+ {"caption", "cite", "code", "col", "colgroup"},
160
+ {"dd", "del", "dfn", "div", "dl", "dt"},
161
+ {"em"},
162
+ {"fieldset", "font", "form", "frame", "frameset"},
163
+ {}, // g
164
+ {"h1", "h2", "h3", "h4", "h5", "h6", "head", "hr", "html"},
165
+ {"i", "iframe", "img", "input", "ins"},
166
+ {}, // j
167
+ {"kbd"},
168
+ {"label", "legend", "li", "link"},
169
+ {"map", "meta"},
170
+ {"noframes", "noscript"},
171
+ {"object", "ol", "optgroup", "option"},
172
+ {"p", "param", "pre"},
173
+ {"q"},
174
+ {}, // r
175
+ {"s", "samp", "script", "select", "small", "span", "strike", "strong", "style", "sub", "sup"},
176
+ {"table", "tbody", "td", "textarea", "tfoot", "th", "thead", "title", "tr", "tt"},
177
+ {"u", "ul"},
178
+ {"var"},
179
+ {}, // w
180
+ {}, // x
181
+ {}, // y
182
+ {} // z
183
+ };
184
+
185
+ private boolean isValid(String testee) {
186
+ char[] c = testee.toCharArray();
187
+ int index = new Integer(c[0]) - 97;
188
+ if (index > 25) return false;
189
+ for (int i=0; i<element_names[index].length; i++) {
190
+ if (testee.equals(element_names[index][i])) {
191
+ return true;
192
+ }
193
+ }
194
+ return false;
195
+ }
196
+
197
+ @Override
198
+ public void startElement(QName name, XMLAttributes attrs, Augmentations augs) throws XNIException {
199
+ if (!isValid(name.rawname)) {
200
+ errorHandler.getErrors().add(new Exception("Tag " + name.rawname + " invalid"));
201
+ }
202
+ super.startElement(name, attrs, augs);
203
+ }
204
+ }
205
+ }
@@ -0,0 +1,73 @@
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.internals;
34
+
35
+ import java.util.Hashtable;
36
+ import nokogiri.XmlDocument;
37
+ import org.w3c.dom.Document;
38
+
39
+ /**
40
+ * Currently, this class is not used anywhere.
41
+ * I'm not sure what for this class was written.(Yoko)
42
+ *
43
+ * @author sergio
44
+ */
45
+ public class NokogiriDocumentCache {
46
+
47
+ private static NokogiriDocumentCache instance;
48
+ protected Hashtable<Document, XmlDocument> cache;
49
+
50
+ private NokogiriDocumentCache() {
51
+ this.cache = new Hashtable<Document, XmlDocument>();
52
+ }
53
+
54
+ public static NokogiriDocumentCache getInstance() {
55
+ if(instance == null) {
56
+ instance = new NokogiriDocumentCache();
57
+ }
58
+ return instance;
59
+ }
60
+
61
+ public XmlDocument getXmlDocument(Document doc) {
62
+ return this.cache.get(doc);
63
+ }
64
+
65
+ public void putDocument(Document doc, XmlDocument xmlDoc) {
66
+ this.cache.put(doc, xmlDoc);
67
+ }
68
+
69
+ public XmlDocument removeDocument(Document doc) {
70
+ return this.cache.remove(doc);
71
+ }
72
+
73
+ }
@@ -0,0 +1,80 @@
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.internals;
34
+
35
+ import java.util.ArrayList;
36
+ import java.util.List;
37
+
38
+ import nokogiri.XmlSyntaxError;
39
+
40
+ import org.apache.xerces.xni.parser.XMLErrorHandler;
41
+ import org.jruby.Ruby;
42
+ import org.jruby.runtime.ThreadContext;
43
+ import org.jruby.runtime.builtin.IRubyObject;
44
+ import org.xml.sax.ErrorHandler;
45
+
46
+ /**
47
+ * Super class of error handlers.
48
+ *
49
+ * XMLErrorHandler is used by nokogiri.internals.HtmlDomParserContext since NekoHtml
50
+ * uses this type of the error handler.
51
+ *
52
+ * @author sergio
53
+ */
54
+ public abstract class NokogiriErrorHandler implements ErrorHandler, XMLErrorHandler {
55
+ protected List<Exception> errors;
56
+ protected boolean noerror;
57
+ protected boolean nowarning;
58
+
59
+ public NokogiriErrorHandler(boolean noerror, boolean nowarning) {
60
+ errors = new ArrayList<Exception>();
61
+ this.noerror = noerror;
62
+ this.nowarning = nowarning;
63
+ }
64
+
65
+ public List<Exception> getErrors() { return errors; }
66
+
67
+ public List<IRubyObject> getErrorsReadyForRuby(ThreadContext context){
68
+ Ruby ruby = context.getRuntime();
69
+ List<IRubyObject> res = new ArrayList<IRubyObject>();
70
+ for(int i = 0; i < errors.size(); i++) {
71
+ res.add(new XmlSyntaxError(ruby, errors.get(i)));
72
+ }
73
+ return res;
74
+ }
75
+
76
+ protected boolean usesNekoHtml(String domain) {
77
+ if ("http://cyberneko.org/html".equals(domain)) return true;
78
+ else return false;
79
+ }
80
+ }
@@ -0,0 +1,326 @@
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.internals;
34
+
35
+ import static nokogiri.internals.NokogiriHelpers.getLocalPart;
36
+ import static nokogiri.internals.NokogiriHelpers.getPrefix;
37
+ import static nokogiri.internals.NokogiriHelpers.isNamespace;
38
+ import static nokogiri.internals.NokogiriHelpers.stringOrNil;
39
+
40
+ import java.util.ArrayDeque;
41
+ import java.util.LinkedList;
42
+
43
+ import nokogiri.XmlSyntaxError;
44
+
45
+ import org.jruby.Ruby;
46
+ import org.jruby.RubyArray;
47
+ import org.jruby.RubyClass;
48
+ import org.jruby.RubyObject;
49
+ import org.jruby.javasupport.util.RuntimeHelpers;
50
+ import org.jruby.runtime.ThreadContext;
51
+ import org.jruby.runtime.builtin.IRubyObject;
52
+ import org.xml.sax.Attributes;
53
+ import org.xml.sax.Locator;
54
+ import org.xml.sax.SAXException;
55
+ import org.xml.sax.SAXParseException;
56
+ import org.xml.sax.ext.DefaultHandler2;
57
+
58
+ /**
59
+ * A handler for SAX parsing.
60
+ *
61
+ * @author sergio
62
+ */
63
+ public class NokogiriHandler extends DefaultHandler2 implements XmlDeclHandler {
64
+
65
+ boolean inCDATA = false;
66
+
67
+ private Ruby ruby;
68
+ private RubyClass attrClass;
69
+ private IRubyObject object;
70
+
71
+ /**
72
+ * Stores parse errors with the most-recent error last.
73
+ *
74
+ * TODO: should these be stored in the document 'errors' array?
75
+ * Currently only string messages are stored there.
76
+ */
77
+ private LinkedList<XmlSyntaxError> errors = new LinkedList<XmlSyntaxError>();
78
+
79
+ private Locator locator;
80
+ private ArrayDeque<Integer> lines;
81
+ private ArrayDeque<Integer> columns;
82
+ private static String htmlParserName = "Nokogiri::HTML::SAX::Parser";
83
+ private boolean needEmptyAttrCheck = false;
84
+
85
+ public NokogiriHandler(Ruby ruby, IRubyObject object) {
86
+ this.ruby = ruby;
87
+ this.attrClass = (RubyClass) ruby.getClassFromPath("Nokogiri::XML::SAX::Parser::Attribute");
88
+ this.object = object;
89
+ lines = new ArrayDeque<Integer>();
90
+ columns = new ArrayDeque<Integer>();
91
+ String objectName = object.getMetaClass().getName();
92
+ if (htmlParserName.equals(objectName)) needEmptyAttrCheck = true;
93
+ }
94
+
95
+ public void setDocumentLocator(Locator locator) {
96
+ this.locator = locator;
97
+ }
98
+
99
+ @Override
100
+ public void startDocument() throws SAXException {
101
+ call("start_document");
102
+ }
103
+
104
+ public void xmlDecl(String version, String encoding, String standalone) {
105
+ call("xmldecl", stringOrNil(ruby, version),
106
+ stringOrNil(ruby, encoding),
107
+ stringOrNil(ruby, standalone));
108
+ }
109
+
110
+ @Override
111
+ public void endDocument() throws SAXException {
112
+ call("end_document");
113
+ }
114
+
115
+ /*
116
+ * This has to call either "start_element" or
117
+ * "start_element_namespace" depending on whether there are any
118
+ * namespace attributes.
119
+ *
120
+ * Attributes that define namespaces are passed in a separate
121
+ * array of of <code>[:prefix, :uri]</code> arrays and are not
122
+ * passed with the other attributes.
123
+ */
124
+ @Override
125
+ public void startElement(String uri, String localName, String qName, Attributes attrs) throws SAXException {
126
+ // for attributes other than namespace attrs
127
+ RubyArray rubyAttr = RubyArray.newArray(ruby);
128
+ // for namespace defining attributes
129
+ RubyArray rubyNSAttr = RubyArray.newArray(ruby);
130
+
131
+ ThreadContext context = ruby.getCurrentContext();
132
+ boolean fromFragmentHandler = false; // isFromFragmentHandler();
133
+
134
+ lines.add(locator.getLineNumber());
135
+ columns.add(locator.getColumnNumber() - 1); // libxml counts from 0 while java does from 1
136
+
137
+ for (int i = 0; i < attrs.getLength(); i++) {
138
+ String u = attrs.getURI(i);
139
+ String qn = attrs.getQName(i);
140
+ String ln = attrs.getLocalName(i);
141
+ String val = attrs.getValue(i);
142
+ String pre;
143
+
144
+ pre = getPrefix(qn);
145
+ if (ln == null || ln.equals("")) ln = getLocalPart(qn);
146
+
147
+ if (isNamespace(qn) && !fromFragmentHandler) {
148
+ // I haven't figured the reason out yet, but, in somewhere,
149
+ // namespace is converted to array in array in array and cause
150
+ // TypeError at line 45 in fragment_handler.rb
151
+ RubyArray ns = RubyArray.newArray(ruby, 2);
152
+ if (ln.equals("xmlns")) ln = null;
153
+ ns.add(stringOrNil(ruby, ln));
154
+ ns.add(ruby.newString(val));
155
+ rubyNSAttr.add(ns);
156
+ } else {
157
+ IRubyObject[] args = null;
158
+ if (needEmptyAttrCheck) {
159
+ if (isEmptyAttr(ln)) {
160
+ args = new IRubyObject[3];
161
+ args[0] = stringOrNil(ruby, ln);
162
+ args[1] = stringOrNil(ruby, pre);
163
+ args[2] = stringOrNil(ruby, u);
164
+ }
165
+ }
166
+ if (args == null) {
167
+ args = new IRubyObject[4];
168
+ args[0] = stringOrNil(ruby, ln);
169
+ args[1] = stringOrNil(ruby, pre);
170
+ args[2] = stringOrNil(ruby, u);
171
+ args[3] = stringOrNil(ruby, val);
172
+ }
173
+
174
+ IRubyObject attr = RuntimeHelpers.invoke(context, attrClass, "new", args);
175
+ rubyAttr.add(attr);
176
+ }
177
+ }
178
+
179
+ if (localName == null || localName.equals("")) localName = getLocalPart(qName);
180
+ call("start_element_namespace",
181
+ stringOrNil(ruby, localName),
182
+ rubyAttr,
183
+ stringOrNil(ruby, getPrefix(qName)),
184
+ stringOrNil(ruby, uri),
185
+ rubyNSAttr);
186
+ }
187
+
188
+ private static String[] emptyAttrs =
189
+ {"checked", "compact", "declare", "defer", "disabled", "ismap", "multiple",
190
+ "noresize", "nohref", "noshade", "nowrap", "readonly", "selected"};
191
+
192
+ private boolean isEmptyAttr(String name) {
193
+ for (String emptyAttr : emptyAttrs) {
194
+ if (emptyAttr.equals(name)) return true;
195
+ }
196
+ return false;
197
+ }
198
+
199
+ public Integer getLine() {
200
+ return lines.pop();
201
+ }
202
+
203
+ public Integer getColumn() {
204
+ return columns.pop();
205
+ }
206
+
207
+ private boolean isFromFragmentHandler() {
208
+ if (object != null && object instanceof RubyObject) {
209
+ RubyObject rubyObj = (RubyObject)object;
210
+ IRubyObject document = rubyObj.getInstanceVariable("@document");
211
+ if (document != null) {
212
+ String name = document.getMetaClass().getName();
213
+ if ("Nokogiri::XML::FragmentHandler".equals(name)) {
214
+ return true;
215
+ }
216
+ }
217
+ }
218
+ return false;
219
+ }
220
+
221
+ @Override
222
+ public void endElement(String uri, String localName, String qName) throws SAXException {
223
+ call("end_element_namespace",
224
+ stringOrNil(ruby, localName),
225
+ stringOrNil(ruby, getPrefix(qName)),
226
+ stringOrNil(ruby, uri));
227
+ }
228
+
229
+ @Override
230
+ public void characters(char[] ch, int start, int length) throws SAXException {
231
+ String target = inCDATA ? "cdata_block" : "characters";
232
+ call(target, ruby.newString(new String(ch, start, length)));
233
+ }
234
+
235
+ @Override
236
+ public void comment(char[] ch, int start, int length) throws SAXException {
237
+ call("comment", ruby.newString(new String(ch, start, length)));
238
+ }
239
+
240
+ @Override
241
+ public void startCDATA() throws SAXException {
242
+ inCDATA = true;
243
+ }
244
+
245
+ @Override
246
+ public void endCDATA() throws SAXException {
247
+ inCDATA = false;
248
+ }
249
+
250
+ @Override
251
+ public void error(SAXParseException saxpe) {
252
+ addError(XmlSyntaxError.createError(ruby, saxpe));
253
+ call("error", ruby.newString(saxpe.getMessage()));
254
+ }
255
+
256
+ @Override
257
+ public void fatalError(SAXParseException saxpe) throws SAXException
258
+ {
259
+ addError(XmlSyntaxError.createFatalError(ruby, saxpe));
260
+ call("error", ruby.newString(saxpe.getMessage()));
261
+ }
262
+
263
+ @Override
264
+ public void warning(SAXParseException saxpe) {
265
+ //System.out.println("warning: " + saxpe);
266
+ call("warning", ruby.newString(saxpe.getMessage()));
267
+ }
268
+
269
+ protected synchronized void addError(XmlSyntaxError e) {
270
+ errors.add(e);
271
+ }
272
+
273
+ public synchronized int getErrorCount() {
274
+ return errors.size();
275
+ }
276
+
277
+ public synchronized IRubyObject getLastError() {
278
+ return errors.getLast();
279
+ }
280
+
281
+ private void call(String methodName) {
282
+ ThreadContext context = ruby.getCurrentContext();
283
+ RuntimeHelpers.invoke(context, document(context), methodName);
284
+ }
285
+
286
+ private void call(String methodName, IRubyObject argument) {
287
+ ThreadContext context = ruby.getCurrentContext();
288
+ RuntimeHelpers.invoke(context, document(context), methodName, argument);
289
+ }
290
+
291
+ private void call(String methodName, IRubyObject arg1, IRubyObject arg2) {
292
+ ThreadContext context = ruby.getCurrentContext();
293
+ RuntimeHelpers.invoke(context, document(context), methodName, arg1, arg2);
294
+ }
295
+
296
+ private void call(String methodName, IRubyObject arg1, IRubyObject arg2,
297
+ IRubyObject arg3) {
298
+ ThreadContext context = ruby.getCurrentContext();
299
+ RuntimeHelpers.invoke(context, document(context), methodName,
300
+ arg1, arg2, arg3);
301
+ }
302
+
303
+ private void call(String methodName,
304
+ IRubyObject arg0,
305
+ IRubyObject arg1,
306
+ IRubyObject arg2,
307
+ IRubyObject arg3,
308
+ IRubyObject arg4) {
309
+ IRubyObject[] args = new IRubyObject[5];
310
+ args[0] = arg0;
311
+ args[1] = arg1;
312
+ args[2] = arg2;
313
+ args[3] = arg3;
314
+ args[4] = arg4;
315
+ ThreadContext context = ruby.getCurrentContext();
316
+ RuntimeHelpers.invoke(context, document(context), methodName, args);
317
+ }
318
+
319
+ private IRubyObject document(ThreadContext context) {
320
+ if (object instanceof RubyObject) {
321
+ return ((RubyObject)object).fastGetInstanceVariable("@document");
322
+ }
323
+ return context.getRuntime().getNil();
324
+ }
325
+
326
+ }