Nokogiri_precompiled_aarch64_dedshit 1.14.5

Sign up to get free protection for your applications and to get access to all the features.
Files changed (263) hide show
  1. checksums.yaml +7 -0
  2. data/Gemfile +44 -0
  3. data/LICENSE-DEPENDENCIES.md +2224 -0
  4. data/LICENSE.md +9 -0
  5. data/README.md +287 -0
  6. data/bin/nokogiri +131 -0
  7. data/dependencies.yml +41 -0
  8. data/ext/java/nokogiri/Html4Document.java +157 -0
  9. data/ext/java/nokogiri/Html4ElementDescription.java +133 -0
  10. data/ext/java/nokogiri/Html4EntityLookup.java +63 -0
  11. data/ext/java/nokogiri/Html4SaxParserContext.java +289 -0
  12. data/ext/java/nokogiri/Html4SaxPushParser.java +213 -0
  13. data/ext/java/nokogiri/NokogiriService.java +613 -0
  14. data/ext/java/nokogiri/XmlAttr.java +154 -0
  15. data/ext/java/nokogiri/XmlAttributeDecl.java +119 -0
  16. data/ext/java/nokogiri/XmlCdata.java +60 -0
  17. data/ext/java/nokogiri/XmlComment.java +77 -0
  18. data/ext/java/nokogiri/XmlDocument.java +705 -0
  19. data/ext/java/nokogiri/XmlDocumentFragment.java +163 -0
  20. data/ext/java/nokogiri/XmlDtd.java +516 -0
  21. data/ext/java/nokogiri/XmlElement.java +44 -0
  22. data/ext/java/nokogiri/XmlElementContent.java +412 -0
  23. data/ext/java/nokogiri/XmlElementDecl.java +148 -0
  24. data/ext/java/nokogiri/XmlEntityDecl.java +151 -0
  25. data/ext/java/nokogiri/XmlEntityReference.java +79 -0
  26. data/ext/java/nokogiri/XmlNamespace.java +193 -0
  27. data/ext/java/nokogiri/XmlNode.java +1938 -0
  28. data/ext/java/nokogiri/XmlNodeSet.java +463 -0
  29. data/ext/java/nokogiri/XmlProcessingInstruction.java +79 -0
  30. data/ext/java/nokogiri/XmlReader.java +615 -0
  31. data/ext/java/nokogiri/XmlRelaxng.java +133 -0
  32. data/ext/java/nokogiri/XmlSaxParserContext.java +329 -0
  33. data/ext/java/nokogiri/XmlSaxPushParser.java +288 -0
  34. data/ext/java/nokogiri/XmlSchema.java +423 -0
  35. data/ext/java/nokogiri/XmlSyntaxError.java +137 -0
  36. data/ext/java/nokogiri/XmlText.java +90 -0
  37. data/ext/java/nokogiri/XmlXpathContext.java +305 -0
  38. data/ext/java/nokogiri/XsltStylesheet.java +368 -0
  39. data/ext/java/nokogiri/internals/ClosedStreamException.java +13 -0
  40. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
  41. data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +27 -0
  42. data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +178 -0
  43. data/ext/java/nokogiri/internals/NokogiriDomParser.java +99 -0
  44. data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +140 -0
  45. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +65 -0
  46. data/ext/java/nokogiri/internals/NokogiriHandler.java +339 -0
  47. data/ext/java/nokogiri/internals/NokogiriHelpers.java +817 -0
  48. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +228 -0
  49. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +110 -0
  50. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +86 -0
  51. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +107 -0
  52. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +62 -0
  53. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +165 -0
  54. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +50 -0
  55. data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +37 -0
  56. data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +70 -0
  57. data/ext/java/nokogiri/internals/ParserContext.java +262 -0
  58. data/ext/java/nokogiri/internals/ReaderNode.java +564 -0
  59. data/ext/java/nokogiri/internals/SaveContextVisitor.java +865 -0
  60. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +50 -0
  61. data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +174 -0
  62. data/ext/java/nokogiri/internals/XmlDeclHandler.java +11 -0
  63. data/ext/java/nokogiri/internals/XmlDomParserContext.java +265 -0
  64. data/ext/java/nokogiri/internals/XmlSaxParser.java +40 -0
  65. data/ext/java/nokogiri/internals/c14n/AttrCompare.java +122 -0
  66. data/ext/java/nokogiri/internals/c14n/C14nHelper.java +178 -0
  67. data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +43 -0
  68. data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +106 -0
  69. data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +278 -0
  70. data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +664 -0
  71. data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +45 -0
  72. data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +45 -0
  73. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +388 -0
  74. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +308 -0
  75. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +47 -0
  76. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +51 -0
  77. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +51 -0
  78. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +50 -0
  79. data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +660 -0
  80. data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +194 -0
  81. data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +77 -0
  82. data/ext/java/nokogiri/internals/c14n/Constants.java +45 -0
  83. data/ext/java/nokogiri/internals/c14n/ElementProxy.java +325 -0
  84. data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +106 -0
  85. data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +86 -0
  86. data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +181 -0
  87. data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +87 -0
  88. data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +452 -0
  89. data/ext/java/nokogiri/internals/c14n/NodeFilter.java +52 -0
  90. data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +190 -0
  91. data/ext/java/nokogiri/internals/c14n/XMLUtils.java +540 -0
  92. data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1712 -0
  93. data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +737 -0
  94. data/ext/nokogiri/depend +38 -0
  95. data/ext/nokogiri/extconf.rb +1086 -0
  96. data/ext/nokogiri/gumbo.c +594 -0
  97. data/ext/nokogiri/html4_document.c +167 -0
  98. data/ext/nokogiri/html4_element_description.c +294 -0
  99. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  100. data/ext/nokogiri/html4_sax_parser_context.c +116 -0
  101. data/ext/nokogiri/html4_sax_push_parser.c +95 -0
  102. data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
  103. data/ext/nokogiri/nokogiri.c +265 -0
  104. data/ext/nokogiri/nokogiri.h +235 -0
  105. data/ext/nokogiri/test_global_handlers.c +42 -0
  106. data/ext/nokogiri/xml_attr.c +103 -0
  107. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  108. data/ext/nokogiri/xml_cdata.c +57 -0
  109. data/ext/nokogiri/xml_comment.c +62 -0
  110. data/ext/nokogiri/xml_document.c +689 -0
  111. data/ext/nokogiri/xml_document_fragment.c +44 -0
  112. data/ext/nokogiri/xml_dtd.c +210 -0
  113. data/ext/nokogiri/xml_element_content.c +128 -0
  114. data/ext/nokogiri/xml_element_decl.c +69 -0
  115. data/ext/nokogiri/xml_encoding_handler.c +104 -0
  116. data/ext/nokogiri/xml_entity_decl.c +112 -0
  117. data/ext/nokogiri/xml_entity_reference.c +50 -0
  118. data/ext/nokogiri/xml_namespace.c +186 -0
  119. data/ext/nokogiri/xml_node.c +2426 -0
  120. data/ext/nokogiri/xml_node_set.c +496 -0
  121. data/ext/nokogiri/xml_processing_instruction.c +54 -0
  122. data/ext/nokogiri/xml_reader.c +794 -0
  123. data/ext/nokogiri/xml_relax_ng.c +164 -0
  124. data/ext/nokogiri/xml_sax_parser.c +316 -0
  125. data/ext/nokogiri/xml_sax_parser_context.c +283 -0
  126. data/ext/nokogiri/xml_sax_push_parser.c +166 -0
  127. data/ext/nokogiri/xml_schema.c +260 -0
  128. data/ext/nokogiri/xml_syntax_error.c +85 -0
  129. data/ext/nokogiri/xml_text.c +48 -0
  130. data/ext/nokogiri/xml_xpath_context.c +415 -0
  131. data/ext/nokogiri/xslt_stylesheet.c +363 -0
  132. data/gumbo-parser/CHANGES.md +63 -0
  133. data/gumbo-parser/Makefile +111 -0
  134. data/gumbo-parser/THANKS +27 -0
  135. data/gumbo-parser/src/Makefile +34 -0
  136. data/gumbo-parser/src/README.md +41 -0
  137. data/gumbo-parser/src/ascii.c +75 -0
  138. data/gumbo-parser/src/ascii.h +115 -0
  139. data/gumbo-parser/src/attribute.c +42 -0
  140. data/gumbo-parser/src/attribute.h +17 -0
  141. data/gumbo-parser/src/char_ref.c +22225 -0
  142. data/gumbo-parser/src/char_ref.h +29 -0
  143. data/gumbo-parser/src/char_ref.rl +2154 -0
  144. data/gumbo-parser/src/error.c +626 -0
  145. data/gumbo-parser/src/error.h +148 -0
  146. data/gumbo-parser/src/foreign_attrs.c +104 -0
  147. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  148. data/gumbo-parser/src/insertion_mode.h +33 -0
  149. data/gumbo-parser/src/macros.h +91 -0
  150. data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
  151. data/gumbo-parser/src/parser.c +4878 -0
  152. data/gumbo-parser/src/parser.h +41 -0
  153. data/gumbo-parser/src/replacement.h +33 -0
  154. data/gumbo-parser/src/string_buffer.c +103 -0
  155. data/gumbo-parser/src/string_buffer.h +68 -0
  156. data/gumbo-parser/src/string_piece.c +48 -0
  157. data/gumbo-parser/src/svg_attrs.c +174 -0
  158. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  159. data/gumbo-parser/src/svg_tags.c +137 -0
  160. data/gumbo-parser/src/svg_tags.gperf +55 -0
  161. data/gumbo-parser/src/tag.c +223 -0
  162. data/gumbo-parser/src/tag_lookup.c +382 -0
  163. data/gumbo-parser/src/tag_lookup.gperf +170 -0
  164. data/gumbo-parser/src/tag_lookup.h +13 -0
  165. data/gumbo-parser/src/token_buffer.c +79 -0
  166. data/gumbo-parser/src/token_buffer.h +71 -0
  167. data/gumbo-parser/src/token_type.h +17 -0
  168. data/gumbo-parser/src/tokenizer.c +3463 -0
  169. data/gumbo-parser/src/tokenizer.h +112 -0
  170. data/gumbo-parser/src/tokenizer_states.h +339 -0
  171. data/gumbo-parser/src/utf8.c +245 -0
  172. data/gumbo-parser/src/utf8.h +164 -0
  173. data/gumbo-parser/src/util.c +66 -0
  174. data/gumbo-parser/src/util.h +34 -0
  175. data/gumbo-parser/src/vector.c +111 -0
  176. data/gumbo-parser/src/vector.h +45 -0
  177. data/lib/nokogiri/class_resolver.rb +67 -0
  178. data/lib/nokogiri/css/node.rb +54 -0
  179. data/lib/nokogiri/css/parser.rb +770 -0
  180. data/lib/nokogiri/css/parser.y +277 -0
  181. data/lib/nokogiri/css/parser_extras.rb +96 -0
  182. data/lib/nokogiri/css/syntax_error.rb +9 -0
  183. data/lib/nokogiri/css/tokenizer.rb +155 -0
  184. data/lib/nokogiri/css/tokenizer.rex +56 -0
  185. data/lib/nokogiri/css/xpath_visitor.rb +359 -0
  186. data/lib/nokogiri/css.rb +66 -0
  187. data/lib/nokogiri/decorators/slop.rb +44 -0
  188. data/lib/nokogiri/encoding_handler.rb +57 -0
  189. data/lib/nokogiri/extension.rb +32 -0
  190. data/lib/nokogiri/gumbo.rb +15 -0
  191. data/lib/nokogiri/html.rb +48 -0
  192. data/lib/nokogiri/html4/builder.rb +37 -0
  193. data/lib/nokogiri/html4/document.rb +214 -0
  194. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  195. data/lib/nokogiri/html4/element_description.rb +25 -0
  196. data/lib/nokogiri/html4/element_description_defaults.rb +572 -0
  197. data/lib/nokogiri/html4/encoding_reader.rb +121 -0
  198. data/lib/nokogiri/html4/entity_lookup.rb +15 -0
  199. data/lib/nokogiri/html4/sax/parser.rb +63 -0
  200. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  201. data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
  202. data/lib/nokogiri/html4.rb +47 -0
  203. data/lib/nokogiri/html5/document.rb +168 -0
  204. data/lib/nokogiri/html5/document_fragment.rb +90 -0
  205. data/lib/nokogiri/html5/node.rb +98 -0
  206. data/lib/nokogiri/html5.rb +389 -0
  207. data/lib/nokogiri/jruby/dependencies.rb +3 -0
  208. data/lib/nokogiri/jruby/isorelax/isorelax/20030108/isorelax-20030108.jar +0 -0
  209. data/lib/nokogiri/jruby/net/sf/saxon/Saxon-HE/9.6.0-4/Saxon-HE-9.6.0-4.jar +0 -0
  210. data/lib/nokogiri/jruby/net/sourceforge/htmlunit/neko-htmlunit/2.63.0/neko-htmlunit-2.63.0.jar +0 -0
  211. data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
  212. data/lib/nokogiri/jruby/nu/validator/jing/20200702VNU/jing-20200702VNU.jar +0 -0
  213. data/lib/nokogiri/jruby/org/nokogiri/nekodtd/0.1.11.noko2/nekodtd-0.1.11.noko2.jar +0 -0
  214. data/lib/nokogiri/jruby/xalan/serializer/2.7.3/serializer-2.7.3.jar +0 -0
  215. data/lib/nokogiri/jruby/xalan/xalan/2.7.3/xalan-2.7.3.jar +0 -0
  216. data/lib/nokogiri/jruby/xerces/xercesImpl/2.12.2/xercesImpl-2.12.2.jar +0 -0
  217. data/lib/nokogiri/jruby/xml-apis/xml-apis/1.4.01/xml-apis-1.4.01.jar +0 -0
  218. data/lib/nokogiri/syntax_error.rb +6 -0
  219. data/lib/nokogiri/version/constant.rb +6 -0
  220. data/lib/nokogiri/version/info.rb +223 -0
  221. data/lib/nokogiri/version.rb +4 -0
  222. data/lib/nokogiri/xml/attr.rb +66 -0
  223. data/lib/nokogiri/xml/attribute_decl.rb +20 -0
  224. data/lib/nokogiri/xml/builder.rb +487 -0
  225. data/lib/nokogiri/xml/cdata.rb +13 -0
  226. data/lib/nokogiri/xml/character_data.rb +9 -0
  227. data/lib/nokogiri/xml/document.rb +471 -0
  228. data/lib/nokogiri/xml/document_fragment.rb +205 -0
  229. data/lib/nokogiri/xml/dtd.rb +34 -0
  230. data/lib/nokogiri/xml/element_content.rb +38 -0
  231. data/lib/nokogiri/xml/element_decl.rb +15 -0
  232. data/lib/nokogiri/xml/entity_decl.rb +21 -0
  233. data/lib/nokogiri/xml/entity_reference.rb +20 -0
  234. data/lib/nokogiri/xml/namespace.rb +58 -0
  235. data/lib/nokogiri/xml/node/save_options.rb +68 -0
  236. data/lib/nokogiri/xml/node.rb +1563 -0
  237. data/lib/nokogiri/xml/node_set.rb +447 -0
  238. data/lib/nokogiri/xml/notation.rb +19 -0
  239. data/lib/nokogiri/xml/parse_options.rb +213 -0
  240. data/lib/nokogiri/xml/pp/character_data.rb +21 -0
  241. data/lib/nokogiri/xml/pp/node.rb +57 -0
  242. data/lib/nokogiri/xml/pp.rb +4 -0
  243. data/lib/nokogiri/xml/processing_instruction.rb +11 -0
  244. data/lib/nokogiri/xml/reader.rb +105 -0
  245. data/lib/nokogiri/xml/relax_ng.rb +38 -0
  246. data/lib/nokogiri/xml/sax/document.rb +167 -0
  247. data/lib/nokogiri/xml/sax/parser.rb +125 -0
  248. data/lib/nokogiri/xml/sax/parser_context.rb +21 -0
  249. data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
  250. data/lib/nokogiri/xml/sax.rb +6 -0
  251. data/lib/nokogiri/xml/schema.rb +73 -0
  252. data/lib/nokogiri/xml/searchable.rb +270 -0
  253. data/lib/nokogiri/xml/syntax_error.rb +72 -0
  254. data/lib/nokogiri/xml/text.rb +11 -0
  255. data/lib/nokogiri/xml/xpath/syntax_error.rb +13 -0
  256. data/lib/nokogiri/xml/xpath.rb +21 -0
  257. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  258. data/lib/nokogiri/xml.rb +76 -0
  259. data/lib/nokogiri/xslt/stylesheet.rb +27 -0
  260. data/lib/nokogiri/xslt.rb +65 -0
  261. data/lib/nokogiri.rb +120 -0
  262. data/lib/xsd/xmlparser/nokogiri.rb +106 -0
  263. metadata +391 -0
@@ -0,0 +1,178 @@
1
+ /**
2
+ *
3
+ */
4
+ package nokogiri.internals;
5
+
6
+ import java.io.ByteArrayInputStream;
7
+ import java.io.IOException;
8
+ import java.io.InputStream;
9
+ import java.util.LinkedList;
10
+ import java.util.List;
11
+ import java.util.concurrent.Callable;
12
+ import java.util.concurrent.Future;
13
+ import java.util.concurrent.FutureTask;
14
+ import java.util.concurrent.LinkedBlockingQueue;
15
+
16
+ import nokogiri.XmlSaxPushParser;
17
+
18
+ /**
19
+ * A smart input stream that signals the caller when a chunk of data is consumed
20
+ * from the stream. The main use of this stream is to synchronize the
21
+ * {@link XmlSaxPushParser} and the {@link XmlSaxParser} which runs in a
22
+ * different thread.
23
+ *
24
+ * @author John Shahid <jvshahid@gmail.com>
25
+ */
26
+ public class NokogiriBlockingQueueInputStream extends InputStream
27
+ {
28
+ private final LinkedBlockingQueue<Task> queue;
29
+ protected Task currentTask;
30
+ protected ByteArrayInputStream currentStream;
31
+ protected int position;
32
+ protected boolean closed = false;
33
+
34
+ public static final ByteArrayInputStream END = new ByteArrayInputStream(new byte[0]);
35
+
36
+ private static class Task extends FutureTask<Void>
37
+ {
38
+ private final ByteArrayInputStream stream;
39
+
40
+ public
41
+ Task(ByteArrayInputStream stream)
42
+ {
43
+ super(new Callable<Void>() {
44
+ @Override
45
+ public Void call() throws Exception {
46
+ // TODO Auto-generated method stub
47
+ return null;
48
+ }
49
+ });
50
+ this.stream = stream;
51
+ }
52
+
53
+ public ByteArrayInputStream
54
+ getStream()
55
+ {
56
+ return stream;
57
+ }
58
+
59
+ @Override
60
+ public void
61
+ run()
62
+ {
63
+ // don't do anything
64
+ }
65
+
66
+ @Override
67
+ public boolean
68
+ runAndReset()
69
+ {
70
+ // don't do anything
71
+ return true;
72
+ }
73
+
74
+ @Override
75
+ public void
76
+ set(Void v)
77
+ {
78
+ super.set(v);
79
+ }
80
+ }
81
+
82
+ public
83
+ NokogiriBlockingQueueInputStream()
84
+ {
85
+ queue = new LinkedBlockingQueue<Task>();
86
+ }
87
+
88
+ /**
89
+ * This method shouldn't be called unless the parser has finished parsing or
90
+ * threw an exception while doing so, otherwise, there'll be the protential
91
+ * that the read method will block indefinitely.
92
+ */
93
+ @Override
94
+ public synchronized void
95
+ close()
96
+ {
97
+ closed = true;
98
+ List<Task> tasks = new LinkedList<Task>();
99
+ queue.drainTo(tasks);
100
+ tasks.add(currentTask);
101
+ for (Task task : tasks) {
102
+ task.set(null);
103
+ }
104
+ }
105
+
106
+ /**
107
+ * Add @param stream to the end of the queue of data that will be returned by
108
+ * {@link #read()} and its variants. The method will @return a future whose
109
+ * {@link Future#get()} will block until the data in @param stream is read.
110
+ *
111
+ * Passing the special stream {@link #END} to this method, will cause
112
+ * {@link #read()} to return an eof indicator (i.e. -1) to the caller, after
113
+ * all the data inserted before {@link #END} is processed.
114
+ *
115
+ * @return
116
+ */
117
+ public synchronized Future<Void>
118
+ addChunk(ByteArrayInputStream stream) throws ClosedStreamException
119
+ {
120
+ if (closed) {
121
+ throw new ClosedStreamException("Cannot add a chunk to a closed stream");
122
+ }
123
+ Task task = new Task(stream);
124
+ queue.add(task);
125
+ return task;
126
+ }
127
+
128
+ /*
129
+ * (non-Javadoc)
130
+ *
131
+ * @see java.io.InputStream#read()
132
+ */
133
+ @Override
134
+ public int
135
+ read() throws IOException
136
+ {
137
+ if (currentTask == null || currentStream.available() == 0)
138
+ if (getNextTask() == -1) {
139
+ return -1;
140
+ }
141
+ return currentStream.read();
142
+ }
143
+
144
+ /*
145
+ * (non-Javadoc)
146
+ *
147
+ * @see java.io.InputStream#read(byte[], int, int)
148
+ */
149
+ @Override
150
+ public int
151
+ read(byte[] bytes, int off, int len)
152
+ {
153
+ if (currentTask == null || currentStream.available() == 0) {
154
+ if (getNextTask() == -1) {
155
+ currentTask.set(null);
156
+ return -1;
157
+ }
158
+ }
159
+ return currentStream.read(bytes, off, len);
160
+ }
161
+
162
+ protected int
163
+ getNextTask()
164
+ {
165
+ while (true) {
166
+ try {
167
+ if (currentTask != null) {
168
+ currentTask.set(null);
169
+ }
170
+ currentTask = queue.take();
171
+ currentStream = currentTask.getStream();
172
+ return currentStream.available() == 0 ? -1 : currentStream.available();
173
+ } catch (InterruptedException ex) {
174
+ // keep retrying to read
175
+ }
176
+ }
177
+ }
178
+ }
@@ -0,0 +1,99 @@
1
+ package nokogiri.internals;
2
+
3
+ import java.io.IOException;
4
+
5
+ import nokogiri.XmlDocument;
6
+
7
+ import org.apache.xerces.parsers.DOMParser;
8
+ import org.apache.xerces.parsers.XIncludeParserConfiguration;
9
+ import org.apache.xerces.xni.parser.XMLParserConfiguration;
10
+ import org.cyberneko.dtd.DTDConfiguration;
11
+ import org.w3c.dom.Document;
12
+ import org.xml.sax.Attributes;
13
+ import org.xml.sax.ContentHandler;
14
+ import org.xml.sax.InputSource;
15
+ import org.xml.sax.Locator;
16
+ import org.xml.sax.SAXException;
17
+
18
+ /**
19
+ * Sets up a Xerces/XNI DOM Parser for use with Nokogiri. Uses
20
+ * NekoDTD to parse the DTD into a tree of Nodes.
21
+ *
22
+ * @author Patrick Mahoney <pat@polycrystal.org>
23
+ */
24
+ public class NokogiriDomParser extends DOMParser
25
+ {
26
+ protected DOMParser dtd;
27
+ protected boolean xInclude;
28
+ protected XMLParserConfiguration config;
29
+
30
+ public
31
+ NokogiriDomParser(XMLParserConfiguration config)
32
+ {
33
+ super(config);
34
+ this.config = config;
35
+ initialize();
36
+ }
37
+
38
+ public
39
+ NokogiriDomParser(ParserContext.Options options)
40
+ {
41
+ xInclude = options.xInclude;
42
+ initialize();
43
+ }
44
+
45
+ protected void
46
+ initialize()
47
+ {
48
+ if (config == null) {
49
+ if (xInclude) {
50
+ config = new XIncludeParserConfiguration();
51
+ } else {
52
+ config = getXMLParserConfiguration();
53
+ }
54
+ }
55
+
56
+ DTDConfiguration dtdConfig = new DTDConfiguration();
57
+ dtd = new DOMParser(dtdConfig);
58
+
59
+ config.setDTDHandler(dtdConfig);
60
+ config.setDTDContentModelHandler(dtdConfig);
61
+ }
62
+
63
+ @Override
64
+ public void
65
+ parse(InputSource source) throws SAXException, IOException
66
+ {
67
+ dtd.reset();
68
+ if (xInclude) {
69
+ setEntityResolver(new NokogiriXInlcudeEntityResolver(source));
70
+ }
71
+ super.parse(source);
72
+ Document doc = getDocument();
73
+ if (doc == null) {
74
+ throw new RuntimeException("null document");
75
+ }
76
+
77
+ doc.setUserData(XmlDocument.DTD_RAW_DOCUMENT, dtd.getDocument(), null);
78
+ }
79
+
80
+ private static class NokogiriXInlcudeEntityResolver implements org.xml.sax.EntityResolver
81
+ {
82
+ InputSource source;
83
+ private
84
+ NokogiriXInlcudeEntityResolver(InputSource source)
85
+ {
86
+ this.source = source;
87
+ }
88
+
89
+ @Override
90
+ public InputSource
91
+ resolveEntity(String publicId, String systemId)
92
+ throws SAXException, IOException
93
+ {
94
+ if (systemId != null) { source.setSystemId(systemId); }
95
+ if (publicId != null) { source.setPublicId(publicId); }
96
+ return source;
97
+ }
98
+ }
99
+ }
@@ -0,0 +1,140 @@
1
+ package nokogiri.internals;
2
+
3
+ import java.io.ByteArrayInputStream;
4
+ import java.io.File;
5
+ import java.io.IOException;
6
+ import java.net.URI;
7
+
8
+ import nokogiri.internals.ParserContext.Options;
9
+
10
+ import org.jruby.Ruby;
11
+ import org.xml.sax.InputSource;
12
+ import org.xml.sax.SAXException;
13
+ import org.xml.sax.ext.EntityResolver2;
14
+
15
+ /**
16
+ * An entity resolver aware of the fact that the Ruby runtime can
17
+ * change directory but the JVM cannot. Thus any file based
18
+ * entity resolution that uses relative paths must be translated
19
+ * to be relative to the current directory of the Ruby runtime.
20
+ */
21
+ public class NokogiriEntityResolver implements EntityResolver2
22
+ {
23
+ protected final Ruby runtime;
24
+ private final NokogiriErrorHandler handler;
25
+ private final Options options;
26
+
27
+ public
28
+ NokogiriEntityResolver(Ruby runtime, NokogiriErrorHandler handler, Options options)
29
+ {
30
+ super();
31
+ this.runtime = runtime;
32
+ this.handler = handler;
33
+ this.options = options;
34
+ }
35
+
36
+ @Override
37
+ public InputSource
38
+ getExternalSubset(String name, String baseURI)
39
+ throws SAXException, IOException
40
+ {
41
+ return null;
42
+ }
43
+
44
+ @Override
45
+ public InputSource
46
+ resolveEntity(String publicId, String systemId)
47
+ throws SAXException, IOException
48
+ {
49
+ return resolveEntity(runtime, null, publicId, null, systemId);
50
+ }
51
+
52
+ @Override
53
+ public InputSource
54
+ resolveEntity(String name,
55
+ String publicId,
56
+ String baseURI,
57
+ String systemId)
58
+ throws SAXException, IOException
59
+ {
60
+ return resolveEntity(runtime, name, publicId, baseURI, systemId);
61
+ }
62
+
63
+ private static File
64
+ join(String parent, String child)
65
+ {
66
+ if (new File(parent).isFile()) {
67
+ parent = new File(parent).getParent();
68
+ }
69
+ return new File(parent, child);
70
+ }
71
+
72
+ private static InputSource
73
+ emptyInputSource(InputSource source)
74
+ {
75
+ source.setByteStream(new ByteArrayInputStream(new byte[0]));
76
+ return source;
77
+ }
78
+
79
+ private boolean
80
+ shouldLoadDtd()
81
+ {
82
+ return options.dtdLoad || options.dtdValid;
83
+ }
84
+
85
+ private void
86
+ addError(String errorMessage)
87
+ {
88
+ if (handler != null) { handler.addError(new Exception(errorMessage)); }
89
+ }
90
+
91
+ /**
92
+ * Create a file base input source taking into account the current
93
+ * directory of <code>runtime</code>.
94
+ * @throws SAXException
95
+ */
96
+ protected InputSource
97
+ resolveEntity(Ruby runtime, String name, String publicId, String baseURI, String systemId)
98
+ throws IOException, SAXException
99
+ {
100
+ InputSource s = new InputSource();
101
+ if (name.equals("[dtd]") && !shouldLoadDtd()) {
102
+ return emptyInputSource(s);
103
+ } else if (!name.equals("[dtd]") && !options.noEnt) {
104
+ return emptyInputSource(s);
105
+ }
106
+ String adjustedSystemId;
107
+ URI uri = URI.create(systemId);
108
+ if (options.noNet && uri.getHost() != null) {
109
+ addError("Attempt to load network entity " + systemId);
110
+ return emptyInputSource(s);
111
+ }
112
+ // if this is a url or absolute file name then use it
113
+ if (uri.isAbsolute() && !uri.isOpaque()) {
114
+ adjustedSystemId = uri.toURL().toString();
115
+ } else if (new File(uri.getPath()).isAbsolute()) {
116
+ adjustedSystemId = uri.getPath();
117
+ } else if (baseURI != null) {
118
+ URI baseuri = URI.create(baseURI);
119
+ if (options.noNet && baseuri.getHost() != null) {
120
+ addError("Attempt to load network entity " + systemId);
121
+ return emptyInputSource(s);
122
+ }
123
+ if (baseuri.getHost() == null) {
124
+ // this is a local file
125
+ adjustedSystemId = join(baseuri.getPath(), uri.getPath()).getCanonicalPath();
126
+ } else {
127
+ // this is a url, then resolve uri using baseuri
128
+ adjustedSystemId = baseuri.resolve(systemId).toURL().toString();
129
+ }
130
+ } else {
131
+ // baseURI is null we have to use the current working directory to resolve the entity
132
+ String pwd = runtime.getCurrentDirectory();
133
+ adjustedSystemId = join(pwd, uri.getPath()).getCanonicalPath();
134
+ }
135
+ s.setSystemId(adjustedSystemId);
136
+ s.setPublicId(publicId);
137
+ return s;
138
+ }
139
+
140
+ }
@@ -0,0 +1,65 @@
1
+ package nokogiri.internals;
2
+
3
+ import nokogiri.XmlSyntaxError;
4
+ import org.apache.xerces.xni.parser.XMLErrorHandler;
5
+ import org.jruby.Ruby;
6
+ import org.jruby.RubyException;
7
+ import org.jruby.exceptions.RaiseException;
8
+ import org.xml.sax.ErrorHandler;
9
+
10
+ import java.util.ArrayList;
11
+ import java.util.List;
12
+
13
+ /**
14
+ * Super class of error handlers.
15
+ *
16
+ * XMLErrorHandler is used by nokogiri.internals.HtmlDomParserContext since NekoHtml
17
+ * uses this type of the error handler.
18
+ *
19
+ * @author sergio
20
+ * @author Yoko Harada <yokolet@gmail.com>
21
+ */
22
+ public abstract class NokogiriErrorHandler implements ErrorHandler, XMLErrorHandler
23
+ {
24
+ private final Ruby runtime;
25
+ protected final List<RubyException> errors;
26
+ protected boolean noerror;
27
+ protected boolean nowarning;
28
+
29
+ public
30
+ NokogiriErrorHandler(Ruby runtime, boolean noerror, boolean nowarning)
31
+ {
32
+ this.runtime = runtime;
33
+ this.errors = new ArrayList<RubyException>(4);
34
+ this.noerror = noerror;
35
+ this.nowarning = nowarning;
36
+ }
37
+
38
+ public List<RubyException>
39
+ getErrors() { return errors; }
40
+
41
+ public void
42
+ addError(Exception ex)
43
+ {
44
+ addError(XmlSyntaxError.createXMLSyntaxError(runtime, ex));
45
+ }
46
+
47
+ public void
48
+ addError(RubyException ex)
49
+ {
50
+ errors.add(ex);
51
+ }
52
+
53
+ public void
54
+ addError(RaiseException ex)
55
+ {
56
+ addError(ex.getException());
57
+ }
58
+
59
+ protected boolean
60
+ usesNekoHtml(String domain)
61
+ {
62
+ return "http://cyberneko.org/html".equals(domain);
63
+ }
64
+
65
+ }