Nokogiri_precompiled_aarch64_dedshit 1.14.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +44 -0
- data/LICENSE-DEPENDENCIES.md +2224 -0
- data/LICENSE.md +9 -0
- data/README.md +287 -0
- data/bin/nokogiri +131 -0
- data/dependencies.yml +41 -0
- data/ext/java/nokogiri/Html4Document.java +157 -0
- data/ext/java/nokogiri/Html4ElementDescription.java +133 -0
- data/ext/java/nokogiri/Html4EntityLookup.java +63 -0
- data/ext/java/nokogiri/Html4SaxParserContext.java +289 -0
- data/ext/java/nokogiri/Html4SaxPushParser.java +213 -0
- data/ext/java/nokogiri/NokogiriService.java +613 -0
- data/ext/java/nokogiri/XmlAttr.java +154 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +119 -0
- data/ext/java/nokogiri/XmlCdata.java +60 -0
- data/ext/java/nokogiri/XmlComment.java +77 -0
- data/ext/java/nokogiri/XmlDocument.java +705 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +163 -0
- data/ext/java/nokogiri/XmlDtd.java +516 -0
- data/ext/java/nokogiri/XmlElement.java +44 -0
- data/ext/java/nokogiri/XmlElementContent.java +412 -0
- data/ext/java/nokogiri/XmlElementDecl.java +148 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +151 -0
- data/ext/java/nokogiri/XmlEntityReference.java +79 -0
- data/ext/java/nokogiri/XmlNamespace.java +193 -0
- data/ext/java/nokogiri/XmlNode.java +1938 -0
- data/ext/java/nokogiri/XmlNodeSet.java +463 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +79 -0
- data/ext/java/nokogiri/XmlReader.java +615 -0
- data/ext/java/nokogiri/XmlRelaxng.java +133 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +329 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +288 -0
- data/ext/java/nokogiri/XmlSchema.java +423 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +137 -0
- data/ext/java/nokogiri/XmlText.java +90 -0
- data/ext/java/nokogiri/XmlXpathContext.java +305 -0
- data/ext/java/nokogiri/XsltStylesheet.java +368 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +13 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +27 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +178 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +99 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +140 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +65 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +339 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +817 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +228 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +110 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +107 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +62 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +165 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +50 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +37 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +70 -0
- data/ext/java/nokogiri/internals/ParserContext.java +262 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +564 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +865 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +50 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +174 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +11 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +265 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +40 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +122 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +178 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +106 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +278 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +664 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +388 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +308 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +47 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +50 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +660 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +194 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +77 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +45 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +325 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +106 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +181 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +87 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +452 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +52 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +190 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +540 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1712 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +737 -0
- data/ext/nokogiri/depend +38 -0
- data/ext/nokogiri/extconf.rb +1086 -0
- data/ext/nokogiri/gumbo.c +594 -0
- data/ext/nokogiri/html4_document.c +167 -0
- data/ext/nokogiri/html4_element_description.c +294 -0
- data/ext/nokogiri/html4_entity_lookup.c +37 -0
- data/ext/nokogiri/html4_sax_parser_context.c +116 -0
- data/ext/nokogiri/html4_sax_push_parser.c +95 -0
- data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
- data/ext/nokogiri/nokogiri.c +265 -0
- data/ext/nokogiri/nokogiri.h +235 -0
- data/ext/nokogiri/test_global_handlers.c +42 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_cdata.c +57 -0
- data/ext/nokogiri/xml_comment.c +62 -0
- data/ext/nokogiri/xml_document.c +689 -0
- data/ext/nokogiri/xml_document_fragment.c +44 -0
- data/ext/nokogiri/xml_dtd.c +210 -0
- data/ext/nokogiri/xml_element_content.c +128 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_encoding_handler.c +104 -0
- data/ext/nokogiri/xml_entity_decl.c +112 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_namespace.c +186 -0
- data/ext/nokogiri/xml_node.c +2426 -0
- data/ext/nokogiri/xml_node_set.c +496 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_reader.c +794 -0
- data/ext/nokogiri/xml_relax_ng.c +164 -0
- data/ext/nokogiri/xml_sax_parser.c +316 -0
- data/ext/nokogiri/xml_sax_parser_context.c +283 -0
- data/ext/nokogiri/xml_sax_push_parser.c +166 -0
- data/ext/nokogiri/xml_schema.c +260 -0
- data/ext/nokogiri/xml_syntax_error.c +85 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_xpath_context.c +415 -0
- data/ext/nokogiri/xslt_stylesheet.c +363 -0
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +111 -0
- data/gumbo-parser/THANKS +27 -0
- data/gumbo-parser/src/Makefile +34 -0
- data/gumbo-parser/src/README.md +41 -0
- data/gumbo-parser/src/ascii.c +75 -0
- data/gumbo-parser/src/ascii.h +115 -0
- data/gumbo-parser/src/attribute.c +42 -0
- data/gumbo-parser/src/attribute.h +17 -0
- data/gumbo-parser/src/char_ref.c +22225 -0
- data/gumbo-parser/src/char_ref.h +29 -0
- data/gumbo-parser/src/char_ref.rl +2154 -0
- data/gumbo-parser/src/error.c +626 -0
- data/gumbo-parser/src/error.h +148 -0
- data/gumbo-parser/src/foreign_attrs.c +104 -0
- data/gumbo-parser/src/foreign_attrs.gperf +27 -0
- data/gumbo-parser/src/insertion_mode.h +33 -0
- data/gumbo-parser/src/macros.h +91 -0
- data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
- data/gumbo-parser/src/parser.c +4878 -0
- data/gumbo-parser/src/parser.h +41 -0
- data/gumbo-parser/src/replacement.h +33 -0
- data/gumbo-parser/src/string_buffer.c +103 -0
- data/gumbo-parser/src/string_buffer.h +68 -0
- data/gumbo-parser/src/string_piece.c +48 -0
- data/gumbo-parser/src/svg_attrs.c +174 -0
- data/gumbo-parser/src/svg_attrs.gperf +77 -0
- data/gumbo-parser/src/svg_tags.c +137 -0
- data/gumbo-parser/src/svg_tags.gperf +55 -0
- data/gumbo-parser/src/tag.c +223 -0
- data/gumbo-parser/src/tag_lookup.c +382 -0
- data/gumbo-parser/src/tag_lookup.gperf +170 -0
- data/gumbo-parser/src/tag_lookup.h +13 -0
- data/gumbo-parser/src/token_buffer.c +79 -0
- data/gumbo-parser/src/token_buffer.h +71 -0
- data/gumbo-parser/src/token_type.h +17 -0
- data/gumbo-parser/src/tokenizer.c +3463 -0
- data/gumbo-parser/src/tokenizer.h +112 -0
- data/gumbo-parser/src/tokenizer_states.h +339 -0
- data/gumbo-parser/src/utf8.c +245 -0
- data/gumbo-parser/src/utf8.h +164 -0
- data/gumbo-parser/src/util.c +66 -0
- data/gumbo-parser/src/util.h +34 -0
- data/gumbo-parser/src/vector.c +111 -0
- data/gumbo-parser/src/vector.h +45 -0
- data/lib/nokogiri/class_resolver.rb +67 -0
- data/lib/nokogiri/css/node.rb +54 -0
- data/lib/nokogiri/css/parser.rb +770 -0
- data/lib/nokogiri/css/parser.y +277 -0
- data/lib/nokogiri/css/parser_extras.rb +96 -0
- data/lib/nokogiri/css/syntax_error.rb +9 -0
- data/lib/nokogiri/css/tokenizer.rb +155 -0
- data/lib/nokogiri/css/tokenizer.rex +56 -0
- data/lib/nokogiri/css/xpath_visitor.rb +359 -0
- data/lib/nokogiri/css.rb +66 -0
- data/lib/nokogiri/decorators/slop.rb +44 -0
- data/lib/nokogiri/encoding_handler.rb +57 -0
- data/lib/nokogiri/extension.rb +32 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +48 -0
- data/lib/nokogiri/html4/builder.rb +37 -0
- data/lib/nokogiri/html4/document.rb +214 -0
- data/lib/nokogiri/html4/document_fragment.rb +54 -0
- data/lib/nokogiri/html4/element_description.rb +25 -0
- data/lib/nokogiri/html4/element_description_defaults.rb +572 -0
- data/lib/nokogiri/html4/encoding_reader.rb +121 -0
- data/lib/nokogiri/html4/entity_lookup.rb +15 -0
- data/lib/nokogiri/html4/sax/parser.rb +63 -0
- data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
- data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html4.rb +47 -0
- data/lib/nokogiri/html5/document.rb +168 -0
- data/lib/nokogiri/html5/document_fragment.rb +90 -0
- data/lib/nokogiri/html5/node.rb +98 -0
- data/lib/nokogiri/html5.rb +389 -0
- data/lib/nokogiri/jruby/dependencies.rb +3 -0
- data/lib/nokogiri/jruby/isorelax/isorelax/20030108/isorelax-20030108.jar +0 -0
- data/lib/nokogiri/jruby/net/sf/saxon/Saxon-HE/9.6.0-4/Saxon-HE-9.6.0-4.jar +0 -0
- data/lib/nokogiri/jruby/net/sourceforge/htmlunit/neko-htmlunit/2.63.0/neko-htmlunit-2.63.0.jar +0 -0
- data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
- data/lib/nokogiri/jruby/nu/validator/jing/20200702VNU/jing-20200702VNU.jar +0 -0
- data/lib/nokogiri/jruby/org/nokogiri/nekodtd/0.1.11.noko2/nekodtd-0.1.11.noko2.jar +0 -0
- data/lib/nokogiri/jruby/xalan/serializer/2.7.3/serializer-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xalan/xalan/2.7.3/xalan-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xerces/xercesImpl/2.12.2/xercesImpl-2.12.2.jar +0 -0
- data/lib/nokogiri/jruby/xml-apis/xml-apis/1.4.01/xml-apis-1.4.01.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +6 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +223 -0
- data/lib/nokogiri/version.rb +4 -0
- data/lib/nokogiri/xml/attr.rb +66 -0
- data/lib/nokogiri/xml/attribute_decl.rb +20 -0
- data/lib/nokogiri/xml/builder.rb +487 -0
- data/lib/nokogiri/xml/cdata.rb +13 -0
- data/lib/nokogiri/xml/character_data.rb +9 -0
- data/lib/nokogiri/xml/document.rb +471 -0
- data/lib/nokogiri/xml/document_fragment.rb +205 -0
- data/lib/nokogiri/xml/dtd.rb +34 -0
- data/lib/nokogiri/xml/element_content.rb +38 -0
- data/lib/nokogiri/xml/element_decl.rb +15 -0
- data/lib/nokogiri/xml/entity_decl.rb +21 -0
- data/lib/nokogiri/xml/entity_reference.rb +20 -0
- data/lib/nokogiri/xml/namespace.rb +58 -0
- data/lib/nokogiri/xml/node/save_options.rb +68 -0
- data/lib/nokogiri/xml/node.rb +1563 -0
- data/lib/nokogiri/xml/node_set.rb +447 -0
- data/lib/nokogiri/xml/notation.rb +19 -0
- data/lib/nokogiri/xml/parse_options.rb +213 -0
- data/lib/nokogiri/xml/pp/character_data.rb +21 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +4 -0
- data/lib/nokogiri/xml/processing_instruction.rb +11 -0
- data/lib/nokogiri/xml/reader.rb +105 -0
- data/lib/nokogiri/xml/relax_ng.rb +38 -0
- data/lib/nokogiri/xml/sax/document.rb +167 -0
- data/lib/nokogiri/xml/sax/parser.rb +125 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +21 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +6 -0
- data/lib/nokogiri/xml/schema.rb +73 -0
- data/lib/nokogiri/xml/searchable.rb +270 -0
- data/lib/nokogiri/xml/syntax_error.rb +72 -0
- data/lib/nokogiri/xml/text.rb +11 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +13 -0
- data/lib/nokogiri/xml/xpath.rb +21 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +27 -0
- data/lib/nokogiri/xslt.rb +65 -0
- data/lib/nokogiri.rb +120 -0
- data/lib/xsd/xmlparser/nokogiri.rb +106 -0
- metadata +391 -0
@@ -0,0 +1,133 @@
|
|
1
|
+
package nokogiri;
|
2
|
+
|
3
|
+
import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
|
4
|
+
|
5
|
+
import java.io.ByteArrayInputStream;
|
6
|
+
import java.io.IOException;
|
7
|
+
import java.io.InputStream;
|
8
|
+
import java.io.StringWriter;
|
9
|
+
import java.io.UnsupportedEncodingException;
|
10
|
+
|
11
|
+
import javax.xml.transform.Source;
|
12
|
+
import javax.xml.transform.TransformerConfigurationException;
|
13
|
+
import javax.xml.transform.TransformerException;
|
14
|
+
import javax.xml.transform.TransformerFactory;
|
15
|
+
import javax.xml.transform.dom.DOMSource;
|
16
|
+
import javax.xml.transform.stream.StreamResult;
|
17
|
+
import javax.xml.transform.stream.StreamSource;
|
18
|
+
|
19
|
+
import org.iso_relax.verifier.Schema;
|
20
|
+
import org.iso_relax.verifier.Verifier;
|
21
|
+
import org.iso_relax.verifier.VerifierConfigurationException;
|
22
|
+
import org.iso_relax.verifier.VerifierFactory;
|
23
|
+
import org.jruby.Ruby;
|
24
|
+
import org.jruby.RubyClass;
|
25
|
+
import org.jruby.anno.JRubyClass;
|
26
|
+
import org.jruby.runtime.ThreadContext;
|
27
|
+
import org.jruby.runtime.builtin.IRubyObject;
|
28
|
+
import org.w3c.dom.Document;
|
29
|
+
import org.xml.sax.ErrorHandler;
|
30
|
+
import org.xml.sax.SAXException;
|
31
|
+
|
32
|
+
/**
|
33
|
+
* Class for Nokogiri::XML::RelaxNG
|
34
|
+
*
|
35
|
+
* @author sergio
|
36
|
+
* @author Yoko Harada <yokolet@gmail.com>
|
37
|
+
*/
|
38
|
+
@JRubyClass(name = "Nokogiri::XML::RelaxNG", parent = "Nokogiri::XML::Schema")
|
39
|
+
public class XmlRelaxng extends XmlSchema
|
40
|
+
{
|
41
|
+
private static final long serialVersionUID = 1L;
|
42
|
+
private Verifier verifier;
|
43
|
+
|
44
|
+
public
|
45
|
+
XmlRelaxng(Ruby ruby, RubyClass klazz)
|
46
|
+
{
|
47
|
+
super(ruby, klazz);
|
48
|
+
}
|
49
|
+
|
50
|
+
private void
|
51
|
+
setVerifier(Verifier verifier)
|
52
|
+
{
|
53
|
+
this.verifier = verifier;
|
54
|
+
}
|
55
|
+
|
56
|
+
static XmlSchema
|
57
|
+
createSchemaInstance(ThreadContext context, RubyClass klazz, Source source, IRubyObject parseOptions)
|
58
|
+
{
|
59
|
+
Ruby runtime = context.getRuntime();
|
60
|
+
XmlRelaxng xmlRelaxng = (XmlRelaxng) NokogiriService.XML_RELAXNG_ALLOCATOR.allocate(runtime, klazz);
|
61
|
+
|
62
|
+
if (parseOptions == null) {
|
63
|
+
parseOptions = defaultParseOptions(context.getRuntime());
|
64
|
+
}
|
65
|
+
|
66
|
+
xmlRelaxng.setInstanceVariable("@errors", runtime.newEmptyArray());
|
67
|
+
xmlRelaxng.setInstanceVariable("@parse_options", parseOptions);
|
68
|
+
|
69
|
+
try {
|
70
|
+
Schema schema = xmlRelaxng.getSchema(source, context);
|
71
|
+
xmlRelaxng.setVerifier(schema.newVerifier());
|
72
|
+
return xmlRelaxng;
|
73
|
+
} catch (VerifierConfigurationException ex) {
|
74
|
+
throw context.getRuntime().newRuntimeError("Could not parse document: " + ex.getMessage());
|
75
|
+
}
|
76
|
+
}
|
77
|
+
|
78
|
+
private Schema
|
79
|
+
getSchema(Source source, ThreadContext context)
|
80
|
+
{
|
81
|
+
InputStream is;
|
82
|
+
VerifierFactory factory = new com.thaiopensource.relaxng.jarv.VerifierFactoryImpl();
|
83
|
+
if (source instanceof StreamSource) {
|
84
|
+
StreamSource ss = (StreamSource)source;
|
85
|
+
is = ss.getInputStream();
|
86
|
+
} else { //if (this.source instanceof DOMSource)
|
87
|
+
DOMSource ds = (DOMSource)source;
|
88
|
+
StringWriter xmlAsWriter = new StringWriter();
|
89
|
+
StreamResult result = new StreamResult(xmlAsWriter);
|
90
|
+
try {
|
91
|
+
TransformerFactory.newInstance().newTransformer().transform(ds, result);
|
92
|
+
} catch (TransformerConfigurationException ex) {
|
93
|
+
throw context.getRuntime()
|
94
|
+
.newRuntimeError("Could not parse document: " + ex.getMessage());
|
95
|
+
} catch (TransformerException ex) {
|
96
|
+
throw context.getRuntime()
|
97
|
+
.newRuntimeError("Could not parse document: " + ex.getMessage());
|
98
|
+
}
|
99
|
+
try {
|
100
|
+
is = new ByteArrayInputStream(xmlAsWriter.toString().getBytes("UTF-8"));
|
101
|
+
} catch (UnsupportedEncodingException ex) {
|
102
|
+
throw context.getRuntime()
|
103
|
+
.newRuntimeError("Could not parse document: " + ex.getMessage());
|
104
|
+
}
|
105
|
+
}
|
106
|
+
|
107
|
+
try {
|
108
|
+
return factory.compileSchema(is);
|
109
|
+
} catch (VerifierConfigurationException ex) {
|
110
|
+
throw context.getRuntime()
|
111
|
+
.newRuntimeError("Could not parse document: " + ex.getMessage());
|
112
|
+
} catch (SAXException ex) {
|
113
|
+
throw context.getRuntime()
|
114
|
+
.newRuntimeError("Could not parse document: " + ex.getMessage());
|
115
|
+
} catch (IOException ex) {
|
116
|
+
throw context.getRuntime().newIOError(ex.getClass() + ": " + ex.getMessage());
|
117
|
+
}
|
118
|
+
}
|
119
|
+
|
120
|
+
@Override
|
121
|
+
protected void
|
122
|
+
setErrorHandler(ErrorHandler errorHandler)
|
123
|
+
{
|
124
|
+
verifier.setErrorHandler(errorHandler);
|
125
|
+
}
|
126
|
+
|
127
|
+
@Override
|
128
|
+
protected void
|
129
|
+
validate(Document document) throws SAXException, IOException
|
130
|
+
{
|
131
|
+
verifier.verify(document);
|
132
|
+
}
|
133
|
+
}
|
@@ -0,0 +1,329 @@
|
|
1
|
+
package nokogiri;
|
2
|
+
|
3
|
+
import nokogiri.internals.*;
|
4
|
+
import org.apache.xerces.parsers.AbstractSAXParser;
|
5
|
+
import org.jruby.Ruby;
|
6
|
+
import org.jruby.RubyClass;
|
7
|
+
import org.jruby.RubyFixnum;
|
8
|
+
import org.jruby.anno.JRubyClass;
|
9
|
+
import org.jruby.anno.JRubyMethod;
|
10
|
+
import org.jruby.exceptions.RaiseException;
|
11
|
+
import org.jruby.runtime.Helpers;
|
12
|
+
import org.jruby.runtime.ThreadContext;
|
13
|
+
import org.jruby.runtime.builtin.IRubyObject;
|
14
|
+
import org.xml.sax.SAXException;
|
15
|
+
import org.xml.sax.SAXParseException;
|
16
|
+
|
17
|
+
import java.io.IOException;
|
18
|
+
import java.io.InputStream;
|
19
|
+
|
20
|
+
import static org.jruby.runtime.Helpers.invoke;
|
21
|
+
|
22
|
+
/**
|
23
|
+
* Base class for the SAX parsers.
|
24
|
+
*
|
25
|
+
* @author Patrick Mahoney <pat@polycrystal.org>
|
26
|
+
* @author Yoko Harada <yokolet@gmail.com>
|
27
|
+
*/
|
28
|
+
@JRubyClass(name = "Nokogiri::XML::SAX::ParserContext")
|
29
|
+
public class XmlSaxParserContext extends ParserContext
|
30
|
+
{
|
31
|
+
private static final long serialVersionUID = 1L;
|
32
|
+
|
33
|
+
protected static final String FEATURE_NAMESPACES =
|
34
|
+
"http://xml.org/sax/features/namespaces";
|
35
|
+
protected static final String FEATURE_NAMESPACE_PREFIXES =
|
36
|
+
"http://xml.org/sax/features/namespace-prefixes";
|
37
|
+
protected static final String FEATURE_LOAD_EXTERNAL_DTD =
|
38
|
+
"http://apache.org/xml/features/nonvalidating/load-external-dtd";
|
39
|
+
protected static final String FEATURE_CONTINUE_AFTER_FATAL_ERROR =
|
40
|
+
"http://apache.org/xml/features/continue-after-fatal-error";
|
41
|
+
|
42
|
+
protected AbstractSAXParser parser;
|
43
|
+
|
44
|
+
protected NokogiriHandler handler;
|
45
|
+
protected NokogiriErrorHandler errorHandler;
|
46
|
+
private boolean replaceEntities = true;
|
47
|
+
private boolean recovery = false;
|
48
|
+
|
49
|
+
public
|
50
|
+
XmlSaxParserContext(final Ruby ruby, RubyClass rubyClass)
|
51
|
+
{
|
52
|
+
super(ruby, rubyClass);
|
53
|
+
}
|
54
|
+
|
55
|
+
protected void
|
56
|
+
initialize(Ruby runtime)
|
57
|
+
{
|
58
|
+
try {
|
59
|
+
parser = createParser();
|
60
|
+
} catch (SAXException se) {
|
61
|
+
// Unexpected failure in XML subsystem
|
62
|
+
RaiseException ex = runtime.newRuntimeError(se.toString());
|
63
|
+
ex.initCause(se);
|
64
|
+
throw ex;
|
65
|
+
}
|
66
|
+
}
|
67
|
+
|
68
|
+
/**
|
69
|
+
* Create and return a copy of this object.
|
70
|
+
*
|
71
|
+
* @return a clone of this object
|
72
|
+
*/
|
73
|
+
@Override
|
74
|
+
public Object
|
75
|
+
clone() throws CloneNotSupportedException
|
76
|
+
{
|
77
|
+
return super.clone();
|
78
|
+
}
|
79
|
+
|
80
|
+
protected AbstractSAXParser
|
81
|
+
createParser() throws SAXException
|
82
|
+
{
|
83
|
+
XmlSaxParser parser = new XmlSaxParser();
|
84
|
+
parser.setFeature(FEATURE_NAMESPACE_PREFIXES, true);
|
85
|
+
parser.setFeature(FEATURE_LOAD_EXTERNAL_DTD, false);
|
86
|
+
return parser;
|
87
|
+
}
|
88
|
+
|
89
|
+
/**
|
90
|
+
* Create a new parser context that will parse the string
|
91
|
+
* <code>data</code>.
|
92
|
+
*/
|
93
|
+
@JRubyMethod(name = "memory", meta = true)
|
94
|
+
public static IRubyObject
|
95
|
+
parse_memory(ThreadContext context,
|
96
|
+
IRubyObject klazz,
|
97
|
+
IRubyObject data)
|
98
|
+
{
|
99
|
+
final Ruby runtime = context.runtime;
|
100
|
+
XmlSaxParserContext ctx = newInstance(runtime, (RubyClass) klazz);
|
101
|
+
ctx.initialize(runtime);
|
102
|
+
ctx.setStringInputSource(context, data, runtime.getNil());
|
103
|
+
return ctx;
|
104
|
+
}
|
105
|
+
|
106
|
+
/**
|
107
|
+
* Create a new parser context that will read from the file
|
108
|
+
* <code>data</code> and parse.
|
109
|
+
*/
|
110
|
+
@JRubyMethod(name = "file", meta = true)
|
111
|
+
public static IRubyObject
|
112
|
+
parse_file(ThreadContext context,
|
113
|
+
IRubyObject klazz,
|
114
|
+
IRubyObject data)
|
115
|
+
{
|
116
|
+
final Ruby runtime = context.runtime;
|
117
|
+
XmlSaxParserContext ctx = newInstance(runtime, (RubyClass) klazz);
|
118
|
+
ctx.initialize(context.getRuntime());
|
119
|
+
ctx.setInputSourceFile(context, data);
|
120
|
+
return ctx;
|
121
|
+
}
|
122
|
+
|
123
|
+
/**
|
124
|
+
* Create a new parser context that will read from the IO or
|
125
|
+
* StringIO <code>data</code> and parse.
|
126
|
+
*
|
127
|
+
* TODO: Currently ignores encoding <code>enc</code>.
|
128
|
+
*/
|
129
|
+
@JRubyMethod(name = "io", meta = true)
|
130
|
+
public static IRubyObject
|
131
|
+
parse_io(ThreadContext context,
|
132
|
+
IRubyObject klazz,
|
133
|
+
IRubyObject data,
|
134
|
+
IRubyObject encoding)
|
135
|
+
{
|
136
|
+
// check the type of the unused encoding to match behavior of CRuby
|
137
|
+
if (!(encoding instanceof RubyFixnum)) {
|
138
|
+
throw context.getRuntime().newTypeError("encoding must be kind_of String");
|
139
|
+
}
|
140
|
+
final Ruby runtime = context.runtime;
|
141
|
+
XmlSaxParserContext ctx = newInstance(runtime, (RubyClass) klazz);
|
142
|
+
ctx.initialize(runtime);
|
143
|
+
ctx.setIOInputSource(context, data, runtime.getNil());
|
144
|
+
return ctx;
|
145
|
+
}
|
146
|
+
|
147
|
+
/**
|
148
|
+
* Create a new parser context that will read from a raw input stream.
|
149
|
+
* Meant to be run in a separate thread by XmlSaxPushParser.
|
150
|
+
*/
|
151
|
+
static XmlSaxParserContext
|
152
|
+
parse_stream(final Ruby runtime, RubyClass klazz, InputStream stream)
|
153
|
+
{
|
154
|
+
XmlSaxParserContext ctx = newInstance(runtime, klazz);
|
155
|
+
ctx.initialize(runtime);
|
156
|
+
ctx.setInputSource(stream);
|
157
|
+
return ctx;
|
158
|
+
}
|
159
|
+
|
160
|
+
private static XmlSaxParserContext
|
161
|
+
newInstance(final Ruby runtime, final RubyClass klazz)
|
162
|
+
{
|
163
|
+
return (XmlSaxParserContext) NokogiriService.XML_SAXPARSER_CONTEXT_ALLOCATOR.allocate(runtime, klazz);
|
164
|
+
}
|
165
|
+
|
166
|
+
public final NokogiriHandler
|
167
|
+
getNokogiriHandler() { return handler; }
|
168
|
+
|
169
|
+
public final NokogiriErrorHandler
|
170
|
+
getNokogiriErrorHandler() { return errorHandler; }
|
171
|
+
|
172
|
+
/**
|
173
|
+
* Perform any initialization prior to parsing with the handler
|
174
|
+
* <code>handlerRuby</code>. Convenience hook for subclasses.
|
175
|
+
*/
|
176
|
+
protected void
|
177
|
+
preParse(Ruby runtime, IRubyObject handlerRuby, NokogiriHandler handler)
|
178
|
+
{
|
179
|
+
((XmlSaxParser) parser).setXmlDeclHandler(handler);
|
180
|
+
if (recovery) {
|
181
|
+
try {
|
182
|
+
parser.setFeature(FEATURE_CONTINUE_AFTER_FATAL_ERROR, true);
|
183
|
+
} catch (Exception e) {
|
184
|
+
// Unexpected failure in XML subsystem
|
185
|
+
throw runtime.newRuntimeError(e.getMessage());
|
186
|
+
}
|
187
|
+
}
|
188
|
+
}
|
189
|
+
|
190
|
+
protected void
|
191
|
+
postParse(Ruby runtime, IRubyObject handlerRuby, NokogiriHandler handler)
|
192
|
+
{
|
193
|
+
// noop
|
194
|
+
}
|
195
|
+
|
196
|
+
protected void
|
197
|
+
do_parse() throws SAXException, IOException
|
198
|
+
{
|
199
|
+
parser.parse(getInputSource());
|
200
|
+
}
|
201
|
+
|
202
|
+
protected static Options
|
203
|
+
defaultParseOptions(ThreadContext context)
|
204
|
+
{
|
205
|
+
return new ParserContext.Options(
|
206
|
+
RubyFixnum.fix2long(Helpers.invoke(context,
|
207
|
+
((RubyClass)context.getRuntime().getClassFromPath("Nokogiri::XML::ParseOptions"))
|
208
|
+
.getConstant("DEFAULT_XML"),
|
209
|
+
"to_i"))
|
210
|
+
);
|
211
|
+
}
|
212
|
+
|
213
|
+
@JRubyMethod
|
214
|
+
public IRubyObject
|
215
|
+
parse_with(ThreadContext context, IRubyObject handlerRuby)
|
216
|
+
{
|
217
|
+
final Ruby runtime = context.getRuntime();
|
218
|
+
|
219
|
+
if (!invoke(context, handlerRuby, "respond_to?", runtime.newSymbol("document")).isTrue()) {
|
220
|
+
throw runtime.newArgumentError("argument must respond_to document");
|
221
|
+
}
|
222
|
+
|
223
|
+
/* TODO: how should we pass in parse options? */
|
224
|
+
ParserContext.Options options = defaultParseOptions(context);
|
225
|
+
|
226
|
+
errorHandler = new NokogiriStrictErrorHandler(runtime, options.noError, options.noWarning);
|
227
|
+
handler = new NokogiriHandler(runtime, handlerRuby, errorHandler);
|
228
|
+
|
229
|
+
preParse(runtime, handlerRuby, handler);
|
230
|
+
parser.setContentHandler(handler);
|
231
|
+
parser.setErrorHandler(handler);
|
232
|
+
parser.setEntityResolver(new NokogiriEntityResolver(runtime, errorHandler, options));
|
233
|
+
|
234
|
+
try {
|
235
|
+
parser.setProperty("http://xml.org/sax/properties/lexical-handler", handler);
|
236
|
+
} catch (Exception ex) {
|
237
|
+
throw runtime.newRuntimeError("Problem while creating XML SAX Parser: " + ex.toString());
|
238
|
+
}
|
239
|
+
|
240
|
+
try {
|
241
|
+
try {
|
242
|
+
do_parse();
|
243
|
+
} catch (SAXParseException ex) {
|
244
|
+
// A bad document (<foo><bar></foo>) should call the
|
245
|
+
// error handler instead of raising a SAX exception.
|
246
|
+
|
247
|
+
// However, an EMPTY document should raise a RuntimeError.
|
248
|
+
// This is a bit kludgy, but AFAIK SAX doesn't distinguish
|
249
|
+
// between empty and bad whereas Nokogiri does.
|
250
|
+
String message = ex.getMessage();
|
251
|
+
if (message != null && message.contains("Premature end of file.") && stringDataSize < 1) {
|
252
|
+
throw runtime.newRuntimeError("couldn't parse document: " + message);
|
253
|
+
}
|
254
|
+
handler.error(ex);
|
255
|
+
}
|
256
|
+
} catch (SAXException ex) {
|
257
|
+
// Unexpected failure in XML subsystem
|
258
|
+
throw runtime.newRuntimeError(ex.getMessage());
|
259
|
+
} catch (IOException ex) {
|
260
|
+
throw runtime.newIOErrorFromException(ex);
|
261
|
+
}
|
262
|
+
|
263
|
+
postParse(runtime, handlerRuby, handler);
|
264
|
+
|
265
|
+
return runtime.getNil();
|
266
|
+
}
|
267
|
+
|
268
|
+
/**
|
269
|
+
* Can take a boolean assignment.
|
270
|
+
*
|
271
|
+
* @param context
|
272
|
+
* @param value
|
273
|
+
* @return
|
274
|
+
*/
|
275
|
+
@JRubyMethod(name = "replace_entities=")
|
276
|
+
public IRubyObject
|
277
|
+
set_replace_entities(ThreadContext context, IRubyObject value)
|
278
|
+
{
|
279
|
+
replaceEntities = value.isTrue();
|
280
|
+
return this;
|
281
|
+
}
|
282
|
+
|
283
|
+
@JRubyMethod(name = "replace_entities")
|
284
|
+
public IRubyObject
|
285
|
+
get_replace_entities(ThreadContext context)
|
286
|
+
{
|
287
|
+
return context.runtime.newBoolean(replaceEntities);
|
288
|
+
}
|
289
|
+
|
290
|
+
/**
|
291
|
+
* Can take a boolean assignment.
|
292
|
+
*
|
293
|
+
* @param context
|
294
|
+
* @param value
|
295
|
+
* @return
|
296
|
+
*/
|
297
|
+
@JRubyMethod(name = "recovery=")
|
298
|
+
public IRubyObject
|
299
|
+
set_recovery(ThreadContext context, IRubyObject value)
|
300
|
+
{
|
301
|
+
recovery = value.isTrue();
|
302
|
+
return this;
|
303
|
+
}
|
304
|
+
|
305
|
+
@JRubyMethod(name = "recovery")
|
306
|
+
public IRubyObject
|
307
|
+
get_recovery(ThreadContext context)
|
308
|
+
{
|
309
|
+
return context.runtime.newBoolean(recovery);
|
310
|
+
}
|
311
|
+
|
312
|
+
@JRubyMethod(name = "column")
|
313
|
+
public IRubyObject
|
314
|
+
column(ThreadContext context)
|
315
|
+
{
|
316
|
+
final Integer number = handler.getColumn();
|
317
|
+
if (number == null) { return context.getRuntime().getNil(); }
|
318
|
+
return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
|
319
|
+
}
|
320
|
+
|
321
|
+
@JRubyMethod(name = "line")
|
322
|
+
public IRubyObject
|
323
|
+
line(ThreadContext context)
|
324
|
+
{
|
325
|
+
final Integer number = handler.getLine();
|
326
|
+
if (number == null) { return context.getRuntime().getNil(); }
|
327
|
+
return RubyFixnum.newFixnum(context.getRuntime(), number.longValue());
|
328
|
+
}
|
329
|
+
}
|