nokogiri-backport 1.11.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +3 -0
- data/LICENSE-DEPENDENCIES.md +1682 -0
- data/LICENSE.md +9 -0
- data/README.md +272 -0
- data/bin/nokogiri +118 -0
- data/dependencies.yml +74 -0
- data/ext/java/nokogiri/EncodingHandler.java +124 -0
- data/ext/java/nokogiri/HtmlDocument.java +178 -0
- data/ext/java/nokogiri/HtmlElementDescription.java +148 -0
- data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
- data/ext/java/nokogiri/HtmlSaxParserContext.java +282 -0
- data/ext/java/nokogiri/HtmlSaxPushParser.java +222 -0
- data/ext/java/nokogiri/NokogiriService.java +597 -0
- data/ext/java/nokogiri/XmlAttr.java +162 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +129 -0
- data/ext/java/nokogiri/XmlCdata.java +82 -0
- data/ext/java/nokogiri/XmlComment.java +97 -0
- data/ext/java/nokogiri/XmlDocument.java +633 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +185 -0
- data/ext/java/nokogiri/XmlDtd.java +481 -0
- data/ext/java/nokogiri/XmlElement.java +68 -0
- data/ext/java/nokogiri/XmlElementContent.java +382 -0
- data/ext/java/nokogiri/XmlElementDecl.java +147 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +157 -0
- data/ext/java/nokogiri/XmlEntityReference.java +101 -0
- data/ext/java/nokogiri/XmlNamespace.java +199 -0
- data/ext/java/nokogiri/XmlNode.java +1684 -0
- data/ext/java/nokogiri/XmlNodeSet.java +434 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +100 -0
- data/ext/java/nokogiri/XmlReader.java +531 -0
- data/ext/java/nokogiri/XmlRelaxng.java +151 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +374 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +286 -0
- data/ext/java/nokogiri/XmlSchema.java +388 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +138 -0
- data/ext/java/nokogiri/XmlText.java +110 -0
- data/ext/java/nokogiri/XmlXpathContext.java +301 -0
- data/ext/java/nokogiri/XsltStylesheet.java +347 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +10 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +20 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +151 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +116 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +69 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +734 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +217 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +127 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +100 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +180 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +72 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +60 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
- data/ext/java/nokogiri/internals/ParserContext.java +259 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +488 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +778 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +73 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +168 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +274 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +119 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +159 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +37 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +93 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +252 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +639 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +367 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +295 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +40 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +630 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +173 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +76 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +42 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +293 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +93 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +79 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +166 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +76 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +402 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +51 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +179 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +507 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1745 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +685 -0
- data/ext/nokogiri/depend +477 -0
- data/ext/nokogiri/extconf.rb +836 -0
- data/ext/nokogiri/html_document.c +171 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +279 -0
- data/ext/nokogiri/html_element_description.h +10 -0
- data/ext/nokogiri/html_entity_lookup.c +32 -0
- data/ext/nokogiri/html_entity_lookup.h +8 -0
- data/ext/nokogiri/html_sax_parser_context.c +116 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/html_sax_push_parser.c +87 -0
- data/ext/nokogiri/html_sax_push_parser.h +9 -0
- data/ext/nokogiri/nokogiri.c +135 -0
- data/ext/nokogiri/nokogiri.h +130 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +62 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +69 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +622 -0
- data/ext/nokogiri/xml_document.h +23 -0
- data/ext/nokogiri/xml_document_fragment.c +48 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +202 -0
- data/ext/nokogiri/xml_dtd.h +10 -0
- data/ext/nokogiri/xml_element_content.c +123 -0
- data/ext/nokogiri/xml_element_content.h +10 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_element_decl.h +9 -0
- data/ext/nokogiri/xml_encoding_handler.c +79 -0
- data/ext/nokogiri/xml_encoding_handler.h +8 -0
- data/ext/nokogiri/xml_entity_decl.c +110 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +52 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +63 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
- data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
- data/ext/nokogiri/xml_namespace.c +111 -0
- data/ext/nokogiri/xml_namespace.h +14 -0
- data/ext/nokogiri/xml_node.c +1773 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +486 -0
- data/ext/nokogiri/xml_node_set.h +12 -0
- data/ext/nokogiri/xml_processing_instruction.c +56 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +657 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +179 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +305 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- data/ext/nokogiri/xml_sax_parser_context.c +262 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +159 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +276 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +64 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +52 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath_context.c +374 -0
- data/ext/nokogiri/xml_xpath_context.h +10 -0
- data/ext/nokogiri/xslt_stylesheet.c +263 -0
- data/ext/nokogiri/xslt_stylesheet.h +14 -0
- data/lib/isorelax.jar +0 -0
- data/lib/jing.jar +0 -0
- data/lib/nekodtd.jar +0 -0
- data/lib/nekohtml.jar +0 -0
- data/lib/nokogiri/css/node.rb +53 -0
- data/lib/nokogiri/css/parser.rb +751 -0
- data/lib/nokogiri/css/parser.y +272 -0
- data/lib/nokogiri/css/parser_extras.rb +94 -0
- data/lib/nokogiri/css/syntax_error.rb +8 -0
- data/lib/nokogiri/css/tokenizer.rb +154 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +260 -0
- data/lib/nokogiri/css.rb +28 -0
- data/lib/nokogiri/decorators/slop.rb +43 -0
- data/lib/nokogiri/html/builder.rb +36 -0
- data/lib/nokogiri/html/document.rb +322 -0
- data/lib/nokogiri/html/document_fragment.rb +50 -0
- data/lib/nokogiri/html/element_description.rb +24 -0
- data/lib/nokogiri/html/element_description_defaults.rb +672 -0
- data/lib/nokogiri/html/entity_lookup.rb +14 -0
- data/lib/nokogiri/html/sax/parser.rb +63 -0
- data/lib/nokogiri/html/sax/parser_context.rb +17 -0
- data/lib/nokogiri/html/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html.rb +38 -0
- data/lib/nokogiri/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +5 -0
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/version.rb +3 -0
- data/lib/nokogiri/xml/attr.rb +15 -0
- data/lib/nokogiri/xml/attribute_decl.rb +19 -0
- data/lib/nokogiri/xml/builder.rb +447 -0
- data/lib/nokogiri/xml/cdata.rb +12 -0
- data/lib/nokogiri/xml/character_data.rb +8 -0
- data/lib/nokogiri/xml/document.rb +290 -0
- data/lib/nokogiri/xml/document_fragment.rb +159 -0
- data/lib/nokogiri/xml/dtd.rb +33 -0
- data/lib/nokogiri/xml/element_content.rb +37 -0
- data/lib/nokogiri/xml/element_decl.rb +14 -0
- data/lib/nokogiri/xml/entity_decl.rb +20 -0
- data/lib/nokogiri/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/xml/namespace.rb +14 -0
- data/lib/nokogiri/xml/node/save_options.rb +62 -0
- data/lib/nokogiri/xml/node.rb +1240 -0
- data/lib/nokogiri/xml/node_set.rb +372 -0
- data/lib/nokogiri/xml/notation.rb +7 -0
- data/lib/nokogiri/xml/parse_options.rb +127 -0
- data/lib/nokogiri/xml/pp/character_data.rb +19 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +3 -0
- data/lib/nokogiri/xml/processing_instruction.rb +9 -0
- data/lib/nokogiri/xml/reader.rb +116 -0
- data/lib/nokogiri/xml/relax_ng.rb +37 -0
- data/lib/nokogiri/xml/sax/document.rb +172 -0
- data/lib/nokogiri/xml/sax/parser.rb +123 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +17 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +5 -0
- data/lib/nokogiri/xml/schema.rb +72 -0
- data/lib/nokogiri/xml/searchable.rb +239 -0
- data/lib/nokogiri/xml/syntax_error.rb +71 -0
- data/lib/nokogiri/xml/text.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
- data/lib/nokogiri/xml/xpath.rb +11 -0
- data/lib/nokogiri/xml/xpath_context.rb +17 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +26 -0
- data/lib/nokogiri/xslt.rb +57 -0
- data/lib/nokogiri.rb +144 -0
- data/lib/serializer.jar +0 -0
- data/lib/xalan.jar +0 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xml-apis.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +103 -0
- metadata +531 -0
@@ -0,0 +1,734 @@
|
|
1
|
+
/**
|
2
|
+
* (The MIT License)
|
3
|
+
*
|
4
|
+
* Copyright (c) 2008 - 2014:
|
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.io.ByteArrayInputStream;
|
36
|
+
import java.io.File;
|
37
|
+
import java.lang.reflect.InvocationTargetException;
|
38
|
+
import java.lang.reflect.Method;
|
39
|
+
import java.nio.ByteBuffer;
|
40
|
+
import java.nio.CharBuffer;
|
41
|
+
import java.nio.charset.Charset;
|
42
|
+
import java.util.List;
|
43
|
+
import java.util.Set;
|
44
|
+
import java.util.regex.Matcher;
|
45
|
+
import java.util.regex.Pattern;
|
46
|
+
|
47
|
+
import org.jruby.Ruby;
|
48
|
+
import org.jruby.RubyArray;
|
49
|
+
import org.jruby.RubyClass;
|
50
|
+
import org.jruby.RubyString;
|
51
|
+
import org.jruby.runtime.ThreadContext;
|
52
|
+
import org.jruby.runtime.builtin.IRubyObject;
|
53
|
+
import org.jruby.util.ByteList;
|
54
|
+
import org.w3c.dom.Attr;
|
55
|
+
import org.w3c.dom.DOMException;
|
56
|
+
import org.w3c.dom.Document;
|
57
|
+
import org.w3c.dom.Node;
|
58
|
+
import org.w3c.dom.NodeList;
|
59
|
+
|
60
|
+
import nokogiri.HtmlDocument;
|
61
|
+
import nokogiri.NokogiriService;
|
62
|
+
import nokogiri.XmlAttr;
|
63
|
+
import nokogiri.XmlCdata;
|
64
|
+
import nokogiri.XmlComment;
|
65
|
+
import nokogiri.XmlDocument;
|
66
|
+
import nokogiri.XmlDtd;
|
67
|
+
import nokogiri.XmlElement;
|
68
|
+
import nokogiri.XmlEntityReference;
|
69
|
+
import nokogiri.XmlNamespace;
|
70
|
+
import nokogiri.XmlNode;
|
71
|
+
import nokogiri.XmlProcessingInstruction;
|
72
|
+
import nokogiri.XmlText;
|
73
|
+
import nokogiri.XmlXpathContext;
|
74
|
+
|
75
|
+
/**
|
76
|
+
* A class for various utility methods.
|
77
|
+
*
|
78
|
+
* @author serabe
|
79
|
+
* @author Patrick Mahoney <pat@polycrystal.org>
|
80
|
+
* @author Yoko Harada <yokolet@gmail.com>
|
81
|
+
*/
|
82
|
+
public class NokogiriHelpers {
|
83
|
+
public static final String CACHED_NODE = "NOKOGIRI_CACHED_NODE";
|
84
|
+
public static final String ROOT_NODE_INVALID = "NOKOGIRI_ROOT_NODE_INVALID";
|
85
|
+
public static final String ENCODED_STRING = "NOKOGIRI_ENCODED_STRING";
|
86
|
+
|
87
|
+
public static XmlNode getCachedNode(Node node) {
|
88
|
+
return (XmlNode) node.getUserData(CACHED_NODE);
|
89
|
+
}
|
90
|
+
|
91
|
+
public static void clearCachedNode(Node node) {
|
92
|
+
node.setUserData(CACHED_NODE, null, null);
|
93
|
+
}
|
94
|
+
|
95
|
+
public static void clearXpathContext(Node node) {
|
96
|
+
if (node == null) return;
|
97
|
+
|
98
|
+
Node ownerDocument = node.getOwnerDocument();
|
99
|
+
if (ownerDocument == null) {
|
100
|
+
ownerDocument = node;
|
101
|
+
}
|
102
|
+
ownerDocument.setUserData(XmlXpathContext.XPATH_CONTEXT, null, null);
|
103
|
+
}
|
104
|
+
|
105
|
+
/**
|
106
|
+
* Get the XmlNode associated with the underlying
|
107
|
+
* <code>node</code>. Creates a new XmlNode (or appropriate subclass)
|
108
|
+
* or XmlNamespace wrapping <code>node</code> if there is no cached
|
109
|
+
* value.
|
110
|
+
*/
|
111
|
+
public static IRubyObject getCachedNodeOrCreate(Ruby runtime, Node node) {
|
112
|
+
if (node == null) return runtime.getNil();
|
113
|
+
if (node.getNodeType() == Node.ATTRIBUTE_NODE && isNamespace(node.getNodeName())) {
|
114
|
+
XmlDocument xmlDocument = (XmlDocument) node.getOwnerDocument().getUserData(CACHED_NODE);
|
115
|
+
if (!(xmlDocument instanceof HtmlDocument)) {
|
116
|
+
String prefix = getLocalNameForNamespace(((Attr) node).getName(), null);
|
117
|
+
String href = ((Attr) node).getValue();
|
118
|
+
XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefix, href);
|
119
|
+
if (xmlNamespace != null) return xmlNamespace;
|
120
|
+
return XmlNamespace.createFromAttr(runtime, (Attr) node);
|
121
|
+
}
|
122
|
+
}
|
123
|
+
XmlNode xmlNode = getCachedNode(node);
|
124
|
+
if (xmlNode == null) {
|
125
|
+
xmlNode = (XmlNode) constructNode(runtime, node);
|
126
|
+
node.setUserData(CACHED_NODE, xmlNode, null);
|
127
|
+
}
|
128
|
+
return xmlNode;
|
129
|
+
}
|
130
|
+
|
131
|
+
/**
|
132
|
+
* Construct a new XmlNode wrapping <code>node</code>. The proper
|
133
|
+
* subclass of XmlNode is chosen based on the type of
|
134
|
+
* <code>node</code>.
|
135
|
+
*/
|
136
|
+
public static IRubyObject constructNode(Ruby runtime, Node node) {
|
137
|
+
if (node == null) return runtime.getNil();
|
138
|
+
// this is slow; need a way to cache nokogiri classes/modules somewhere
|
139
|
+
switch (node.getNodeType()) {
|
140
|
+
case Node.ELEMENT_NODE:
|
141
|
+
XmlElement xmlElement = (XmlElement) NokogiriService.XML_ELEMENT_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Element"));
|
142
|
+
xmlElement.setNode(runtime, node);
|
143
|
+
return xmlElement;
|
144
|
+
case Node.ATTRIBUTE_NODE:
|
145
|
+
XmlAttr xmlAttr = (XmlAttr) NokogiriService.XML_ATTR_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Attr"));
|
146
|
+
xmlAttr.setNode(runtime, node);
|
147
|
+
return xmlAttr;
|
148
|
+
case Node.TEXT_NODE:
|
149
|
+
XmlText xmlText = (XmlText) NokogiriService.XML_TEXT_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Text"));
|
150
|
+
xmlText.setNode(runtime, node);
|
151
|
+
return xmlText;
|
152
|
+
case Node.COMMENT_NODE:
|
153
|
+
XmlComment xmlComment = (XmlComment) NokogiriService.XML_COMMENT_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Comment"));
|
154
|
+
xmlComment.setNode(runtime, node);
|
155
|
+
return xmlComment;
|
156
|
+
case Node.ENTITY_NODE:
|
157
|
+
return new XmlNode(runtime, getNokogiriClass(runtime, "Nokogiri::XML::EntityDecl"), node);
|
158
|
+
case Node.ENTITY_REFERENCE_NODE:
|
159
|
+
XmlEntityReference xmlEntityRef = (XmlEntityReference) NokogiriService.XML_ENTITY_REFERENCE_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::EntityReference"));
|
160
|
+
xmlEntityRef.setNode(runtime, node);
|
161
|
+
return xmlEntityRef;
|
162
|
+
case Node.PROCESSING_INSTRUCTION_NODE:
|
163
|
+
XmlProcessingInstruction xmlProcessingInstruction = (XmlProcessingInstruction) NokogiriService.XML_PROCESSING_INSTRUCTION_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::ProcessingInstruction"));
|
164
|
+
xmlProcessingInstruction.setNode(runtime, node);
|
165
|
+
return xmlProcessingInstruction;
|
166
|
+
case Node.CDATA_SECTION_NODE:
|
167
|
+
XmlCdata xmlCdata = (XmlCdata) NokogiriService.XML_CDATA_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::CDATA"));
|
168
|
+
xmlCdata.setNode(runtime, node);
|
169
|
+
return xmlCdata;
|
170
|
+
case Node.DOCUMENT_NODE:
|
171
|
+
XmlDocument xmlDocument = (XmlDocument) NokogiriService.XML_DOCUMENT_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Document"));
|
172
|
+
xmlDocument.setDocumentNode(runtime, (Document) node);
|
173
|
+
return xmlDocument;
|
174
|
+
case Node.DOCUMENT_TYPE_NODE:
|
175
|
+
XmlDtd xmlDtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
|
176
|
+
xmlDtd.setNode(runtime, node);
|
177
|
+
return xmlDtd;
|
178
|
+
default:
|
179
|
+
XmlNode xmlNode = (XmlNode) NokogiriService.XML_NODE_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Node"));
|
180
|
+
xmlNode.setNode(runtime, node);
|
181
|
+
return xmlNode;
|
182
|
+
}
|
183
|
+
}
|
184
|
+
|
185
|
+
public static RubyClass getNokogiriClass(Ruby ruby, String name) {
|
186
|
+
return NokogiriService.getNokogiriClassCache(ruby).get(name);
|
187
|
+
}
|
188
|
+
|
189
|
+
public static IRubyObject stringOrNil(Ruby runtime, String str) {
|
190
|
+
return str == null ? runtime.getNil() : convertString(runtime, str);
|
191
|
+
}
|
192
|
+
|
193
|
+
public static IRubyObject stringOrNil(Ruby runtime, CharSequence str) {
|
194
|
+
return str == null ? runtime.getNil() : convertString(runtime, str);
|
195
|
+
}
|
196
|
+
|
197
|
+
public static IRubyObject stringOrNil(Ruby runtime, byte[] bytes) {
|
198
|
+
return bytes == null ? runtime.getNil() : RubyString.newString(runtime, bytes);
|
199
|
+
}
|
200
|
+
|
201
|
+
public static IRubyObject stringOrBlank(Ruby runtime, String str) {
|
202
|
+
return str == null ? runtime.newString() : convertString(runtime, str);
|
203
|
+
}
|
204
|
+
|
205
|
+
public static RubyString convertString(Ruby runtime, String str) {
|
206
|
+
return RubyString.newUTF8String(runtime, str);
|
207
|
+
}
|
208
|
+
|
209
|
+
public static RubyString convertString(Ruby runtime, CharSequence str) {
|
210
|
+
return RubyString.newUTF8String(runtime, str);
|
211
|
+
}
|
212
|
+
|
213
|
+
/**
|
214
|
+
* Convert <code>s</code> to a RubyString, or if s is null or
|
215
|
+
* empty return RubyNil.
|
216
|
+
*/
|
217
|
+
public static IRubyObject nonEmptyStringOrNil(Ruby runtime, String s) {
|
218
|
+
if (s == null || s.length() == 0) return runtime.getNil();
|
219
|
+
return RubyString.newString(runtime, s);
|
220
|
+
}
|
221
|
+
|
222
|
+
/**
|
223
|
+
* Return the prefix of a qualified name like "prefix:local".
|
224
|
+
* Returns null if there is no prefix.
|
225
|
+
*/
|
226
|
+
public static String getPrefix(String qName) {
|
227
|
+
if (qName == null) return null;
|
228
|
+
|
229
|
+
final int pos = qName.indexOf(':');
|
230
|
+
return pos > 0 ? qName.substring(0, pos) : null;
|
231
|
+
}
|
232
|
+
|
233
|
+
/**
|
234
|
+
* Return the local part of a qualified name like "prefix:local".
|
235
|
+
* Returns <code>qName</code> if there is no prefix.
|
236
|
+
*/
|
237
|
+
public static String getLocalPart(String qName) {
|
238
|
+
if (qName == null) return null;
|
239
|
+
|
240
|
+
final int pos = qName.indexOf(':');
|
241
|
+
return pos > 0 ? qName.substring(pos + 1) : qName;
|
242
|
+
}
|
243
|
+
|
244
|
+
public static String getLocalNameForNamespace(String name, String defValue) {
|
245
|
+
String localName = getLocalPart(name);
|
246
|
+
return ("xmlns".equals(localName)) ? defValue : localName;
|
247
|
+
}
|
248
|
+
|
249
|
+
public static String rubyStringToString(IRubyObject str) {
|
250
|
+
if (str.isNil()) return null;
|
251
|
+
return str.convertToString().decodeString();
|
252
|
+
}
|
253
|
+
|
254
|
+
public static String rubyStringToString(RubyString str) {
|
255
|
+
return str.decodeString(); // if encoding UTF-8 will decode UTF-8
|
256
|
+
}
|
257
|
+
|
258
|
+
public static ByteArrayInputStream stringBytesToStream(final IRubyObject str) {
|
259
|
+
if (str instanceof RubyString || str.respondsTo("to_str")) {
|
260
|
+
final ByteList bytes = str.convertToString().getByteList();
|
261
|
+
return new ByteArrayInputStream(bytes.unsafeBytes(), bytes.begin(), bytes.length());
|
262
|
+
}
|
263
|
+
return null;
|
264
|
+
}
|
265
|
+
|
266
|
+
public static String getNodeCompletePath(Node node) {
|
267
|
+
|
268
|
+
Node cur, tmp, next;
|
269
|
+
|
270
|
+
String buffer = "";
|
271
|
+
|
272
|
+
cur = node;
|
273
|
+
|
274
|
+
do {
|
275
|
+
String name = "";
|
276
|
+
String sep = "?";
|
277
|
+
int occur = 0;
|
278
|
+
boolean generic = false;
|
279
|
+
|
280
|
+
if(cur.getNodeType() == Node.DOCUMENT_NODE) {
|
281
|
+
if(buffer.startsWith("/")) break;
|
282
|
+
|
283
|
+
sep = "/";
|
284
|
+
next = null;
|
285
|
+
} else if(cur.getNodeType() == Node.ELEMENT_NODE) {
|
286
|
+
generic = false;
|
287
|
+
sep = "/";
|
288
|
+
|
289
|
+
name = cur.getLocalName();
|
290
|
+
if (name == null) name = cur.getNodeName();
|
291
|
+
if(cur.getNamespaceURI() != null) {
|
292
|
+
if(cur.getPrefix() != null) {
|
293
|
+
name = cur.getPrefix() + ":" + name;
|
294
|
+
} else {
|
295
|
+
generic = true;
|
296
|
+
name = "*";
|
297
|
+
}
|
298
|
+
}
|
299
|
+
|
300
|
+
next = cur.getParentNode();
|
301
|
+
|
302
|
+
/*
|
303
|
+
* Thumbler index computation
|
304
|
+
*/
|
305
|
+
|
306
|
+
tmp = cur.getPreviousSibling();
|
307
|
+
|
308
|
+
while(tmp != null) {
|
309
|
+
if((tmp.getNodeType() == Node.ELEMENT_NODE) &&
|
310
|
+
(generic || fullNamesMatch(tmp, cur))) {
|
311
|
+
occur++;
|
312
|
+
}
|
313
|
+
tmp = tmp.getPreviousSibling();
|
314
|
+
}
|
315
|
+
|
316
|
+
if(occur == 0) {
|
317
|
+
tmp = cur.getNextSibling();
|
318
|
+
|
319
|
+
while(tmp != null && occur == 0) {
|
320
|
+
if((tmp.getNodeType() == Node.ELEMENT_NODE) &&
|
321
|
+
(generic || fullNamesMatch(tmp,cur))) {
|
322
|
+
occur++;
|
323
|
+
}
|
324
|
+
tmp = tmp.getNextSibling();
|
325
|
+
}
|
326
|
+
|
327
|
+
if(occur != 0) occur = 1;
|
328
|
+
|
329
|
+
} else {
|
330
|
+
occur++;
|
331
|
+
}
|
332
|
+
} else if(cur.getNodeType() == Node.COMMENT_NODE) {
|
333
|
+
sep = "/";
|
334
|
+
name = "comment()";
|
335
|
+
next = cur.getParentNode();
|
336
|
+
|
337
|
+
/*
|
338
|
+
* Thumbler index computation.
|
339
|
+
*/
|
340
|
+
|
341
|
+
tmp = cur.getPreviousSibling();
|
342
|
+
|
343
|
+
while(tmp != null) {
|
344
|
+
if(tmp.getNodeType() == Node.COMMENT_NODE) {
|
345
|
+
occur++;
|
346
|
+
}
|
347
|
+
tmp = tmp.getPreviousSibling();
|
348
|
+
}
|
349
|
+
|
350
|
+
if(occur == 0) {
|
351
|
+
tmp = cur.getNextSibling();
|
352
|
+
while(tmp != null && occur == 0) {
|
353
|
+
if(tmp.getNodeType() == Node.COMMENT_NODE) {
|
354
|
+
occur++;
|
355
|
+
}
|
356
|
+
tmp = tmp.getNextSibling();
|
357
|
+
}
|
358
|
+
if(occur != 0) occur = 1;
|
359
|
+
} else {
|
360
|
+
occur = 1;
|
361
|
+
}
|
362
|
+
|
363
|
+
} else if(cur.getNodeType() == Node.TEXT_NODE ||
|
364
|
+
cur.getNodeType() == Node.CDATA_SECTION_NODE) {
|
365
|
+
// I'm here. gist:129
|
366
|
+
// http://gist.github.com/144923
|
367
|
+
|
368
|
+
sep = "/";
|
369
|
+
name = "text()";
|
370
|
+
next = cur.getParentNode();
|
371
|
+
|
372
|
+
/*
|
373
|
+
* Thumbler index computation.
|
374
|
+
*/
|
375
|
+
|
376
|
+
tmp = cur.getPreviousSibling();
|
377
|
+
while(tmp != null) {
|
378
|
+
if(tmp.getNodeType() == Node.TEXT_NODE ||
|
379
|
+
tmp.getNodeType() == Node.CDATA_SECTION_NODE) {
|
380
|
+
occur++;
|
381
|
+
}
|
382
|
+
tmp = tmp.getPreviousSibling();
|
383
|
+
}
|
384
|
+
|
385
|
+
if(occur == 0) {
|
386
|
+
tmp = cur.getNextSibling();
|
387
|
+
|
388
|
+
while(tmp != null && occur == 0) {
|
389
|
+
if(tmp.getNodeType() == Node.TEXT_NODE ||
|
390
|
+
tmp.getNodeType() == Node.CDATA_SECTION_NODE) {
|
391
|
+
occur++;
|
392
|
+
}
|
393
|
+
tmp = tmp.getNextSibling();
|
394
|
+
}
|
395
|
+
} else {
|
396
|
+
occur++;
|
397
|
+
}
|
398
|
+
|
399
|
+
} else if(cur.getNodeType() == Node.PROCESSING_INSTRUCTION_NODE) {
|
400
|
+
sep = "/";
|
401
|
+
name = "processing-instruction('"+cur.getLocalName()+"')";
|
402
|
+
next = cur.getParentNode();
|
403
|
+
|
404
|
+
/*
|
405
|
+
* Thumbler index computation.
|
406
|
+
*/
|
407
|
+
|
408
|
+
tmp = cur.getParentNode();
|
409
|
+
|
410
|
+
while(tmp != null) {
|
411
|
+
if(tmp.getNodeType() == Node.PROCESSING_INSTRUCTION_NODE &&
|
412
|
+
tmp.getLocalName().equals(cur.getLocalName())) {
|
413
|
+
occur++;
|
414
|
+
}
|
415
|
+
tmp = tmp.getPreviousSibling();
|
416
|
+
}
|
417
|
+
|
418
|
+
if(occur == 0) {
|
419
|
+
tmp = cur.getNextSibling();
|
420
|
+
|
421
|
+
while(tmp != null && occur == 0) {
|
422
|
+
if(tmp.getNodeType() == Node.PROCESSING_INSTRUCTION_NODE &&
|
423
|
+
tmp.getLocalName().equals(cur.getLocalName())){
|
424
|
+
occur++;
|
425
|
+
}
|
426
|
+
tmp = tmp.getNextSibling();
|
427
|
+
}
|
428
|
+
|
429
|
+
if(occur != 0) {
|
430
|
+
occur = 1;
|
431
|
+
}
|
432
|
+
|
433
|
+
} else {
|
434
|
+
occur++;
|
435
|
+
}
|
436
|
+
|
437
|
+
} else if(cur.getNodeType() == Node.ATTRIBUTE_NODE) {
|
438
|
+
sep = "/@";
|
439
|
+
name = cur.getLocalName();
|
440
|
+
|
441
|
+
if(cur.getNamespaceURI() != null) {
|
442
|
+
if(cur.getPrefix() != null) {
|
443
|
+
name = cur.getPrefix() + ":" + name;
|
444
|
+
}
|
445
|
+
}
|
446
|
+
|
447
|
+
next = ((Attr) cur).getOwnerElement();
|
448
|
+
|
449
|
+
} else {
|
450
|
+
next = cur.getParentNode();
|
451
|
+
}
|
452
|
+
|
453
|
+
if(occur == 0){
|
454
|
+
buffer = sep+name+buffer;
|
455
|
+
} else {
|
456
|
+
buffer = sep+name+"["+occur+"]"+buffer;
|
457
|
+
}
|
458
|
+
|
459
|
+
cur = next;
|
460
|
+
|
461
|
+
} while(cur != null);
|
462
|
+
|
463
|
+
return buffer;
|
464
|
+
}
|
465
|
+
|
466
|
+
static boolean compareTwoNodes(Node m, Node n) {
|
467
|
+
return nodesAreEqual(m.getLocalName(), n.getLocalName()) &&
|
468
|
+
nodesAreEqual(m.getPrefix(), n.getPrefix());
|
469
|
+
}
|
470
|
+
|
471
|
+
private static boolean nodesAreEqual(Object a, Object b) {
|
472
|
+
return (((a == null) && (b == null)) ||
|
473
|
+
((a != null) && (b != null) && (b.equals(a))));
|
474
|
+
}
|
475
|
+
|
476
|
+
private static boolean fullNamesMatch(Node a, Node b) {
|
477
|
+
return a.getNodeName().equals(b.getNodeName());
|
478
|
+
}
|
479
|
+
|
480
|
+
private static final Pattern encoded_pattern = Pattern.compile("&|>|<| ");
|
481
|
+
private static final String[] encoded = {"&", ">", "<", " "};
|
482
|
+
private static final Pattern decoded_pattern = Pattern.compile("&|>|<|\r");
|
483
|
+
private static final String[] decoded = {"&", ">", "<", "\r"};
|
484
|
+
|
485
|
+
private static StringBuffer convert(Pattern ptn, CharSequence input, String[] oldChars, String[] newChars) {
|
486
|
+
Matcher matcher = ptn.matcher(input);
|
487
|
+
boolean result = matcher.find();
|
488
|
+
StringBuffer sb = new StringBuffer(input.length() + 8);
|
489
|
+
while (result) {
|
490
|
+
String matched = matcher.group();
|
491
|
+
String replacement = "";
|
492
|
+
for (int i=0; i<oldChars.length; i++) {
|
493
|
+
if (matched.contains(oldChars[i])) {
|
494
|
+
replacement = matched.replace(oldChars[i], newChars[i]);
|
495
|
+
break;
|
496
|
+
}
|
497
|
+
}
|
498
|
+
matcher.appendReplacement(sb, replacement);
|
499
|
+
result = matcher.find();
|
500
|
+
}
|
501
|
+
matcher.appendTail(sb);
|
502
|
+
return sb;
|
503
|
+
}
|
504
|
+
|
505
|
+
public static CharSequence encodeJavaString(CharSequence str) {
|
506
|
+
return convert(decoded_pattern, str, decoded, encoded);
|
507
|
+
}
|
508
|
+
|
509
|
+
public static CharSequence decodeJavaString(CharSequence str) {
|
510
|
+
return convert(encoded_pattern, str, encoded, decoded);
|
511
|
+
}
|
512
|
+
|
513
|
+
public static final String XMLNS_URI = "http://www.w3.org/2000/xmlns/";
|
514
|
+
public static boolean isNamespace(Node node) {
|
515
|
+
return (XMLNS_URI.equals(node.getNamespaceURI()) || isNamespace(node.getNodeName()));
|
516
|
+
}
|
517
|
+
|
518
|
+
public static boolean isNamespace(String nodeName) {
|
519
|
+
return (nodeName.startsWith("xmlns"));
|
520
|
+
}
|
521
|
+
|
522
|
+
public static boolean isNonDefaultNamespace(Node node) {
|
523
|
+
return (isNamespace(node) && ! "xmlns".equals(node.getNodeName()));
|
524
|
+
}
|
525
|
+
|
526
|
+
public static boolean isXmlBase(String attrName) {
|
527
|
+
return "xml:base".equals(attrName) || "xlink:href".equals(attrName);
|
528
|
+
}
|
529
|
+
|
530
|
+
public static boolean isBlank(IRubyObject obj) {
|
531
|
+
if ( !(obj instanceof XmlText) ) return false;
|
532
|
+
|
533
|
+
CharSequence content = ((XmlNode) obj).getContentImpl();
|
534
|
+
return content == null || isBlank(content);
|
535
|
+
}
|
536
|
+
|
537
|
+
public static boolean isBlank(CharSequence str) {
|
538
|
+
int len = str.length(); int beg = 0;
|
539
|
+
while ((beg < len) && (str.charAt(beg) <= ' ')) beg++;
|
540
|
+
return beg == len;
|
541
|
+
}
|
542
|
+
|
543
|
+
public static boolean isBlank(String str) {
|
544
|
+
return str.isEmpty() || isBlank((CharSequence) str);
|
545
|
+
}
|
546
|
+
|
547
|
+
public static boolean isNullOrEmpty(String str) {
|
548
|
+
return str == null || str.isEmpty();
|
549
|
+
}
|
550
|
+
|
551
|
+
public static CharSequence canonicalizeWhitespace(CharSequence str) {
|
552
|
+
final int len = str.length();
|
553
|
+
StringBuilder sb = new StringBuilder(len);
|
554
|
+
boolean newline_added = false;
|
555
|
+
for ( int i = 0; i < len; i++ ) {
|
556
|
+
char c = str.charAt(i);
|
557
|
+
if ( c == '\n' ) {
|
558
|
+
if ( ! newline_added ) {
|
559
|
+
sb.append(c); newline_added = true;
|
560
|
+
}
|
561
|
+
} else {
|
562
|
+
sb.append(c);
|
563
|
+
}
|
564
|
+
}
|
565
|
+
return sb;
|
566
|
+
}
|
567
|
+
|
568
|
+
public static String newQName(String newPrefix, Node node) {
|
569
|
+
String tagName = getLocalPart(node.getNodeName());
|
570
|
+
if (newPrefix == null) return tagName;
|
571
|
+
return newPrefix + ':' + tagName;
|
572
|
+
}
|
573
|
+
|
574
|
+
public static IRubyObject[] nodeListToRubyArray(Ruby runtime, NodeList nodes) {
|
575
|
+
IRubyObject[] array = new IRubyObject[nodes.getLength()];
|
576
|
+
for (int i = 0; i < nodes.getLength(); i++) {
|
577
|
+
array[i] = NokogiriHelpers.getCachedNodeOrCreate(runtime, nodes.item(i));
|
578
|
+
}
|
579
|
+
return array;
|
580
|
+
}
|
581
|
+
|
582
|
+
public static IRubyObject[] nodeListToArray(Ruby ruby, List<Node> nodes) {
|
583
|
+
IRubyObject[] result = new IRubyObject[nodes.size()];
|
584
|
+
for (int i = 0; i < result.length; i++) {
|
585
|
+
result[i] = NokogiriHelpers.getCachedNodeOrCreate(ruby, nodes.get(i));
|
586
|
+
}
|
587
|
+
return result;
|
588
|
+
}
|
589
|
+
|
590
|
+
public static RubyArray nodeArrayToRubyArray(Ruby ruby, Node[] nodes) {
|
591
|
+
RubyArray n = RubyArray.newArray(ruby, nodes.length);
|
592
|
+
for (int i = 0; i < nodes.length; i++) {
|
593
|
+
n.append(NokogiriHelpers.getCachedNodeOrCreate(ruby, nodes[i]));
|
594
|
+
}
|
595
|
+
return n;
|
596
|
+
}
|
597
|
+
|
598
|
+
public static String getValidEncodingOrNull(IRubyObject encoding) {
|
599
|
+
if (encoding.isNil()) return null; // charsetNames does not like contains(null)
|
600
|
+
String enc = rubyStringToString(encoding.convertToString());
|
601
|
+
if (CharsetNames.contains(enc)) return enc;
|
602
|
+
return null;
|
603
|
+
}
|
604
|
+
|
605
|
+
public static String getValidEncoding(IRubyObject encoding) {
|
606
|
+
String validEncoding = getValidEncodingOrNull(encoding);
|
607
|
+
if (validEncoding != null) return validEncoding;
|
608
|
+
return Charset.defaultCharset().name();
|
609
|
+
}
|
610
|
+
|
611
|
+
private static final Set<String> CharsetNames = Charset.availableCharsets().keySet();
|
612
|
+
|
613
|
+
public static String adjustSystemIdIfNecessary(String currentDir, String scriptFileName, String baseURI, String systemId) {
|
614
|
+
if (systemId == null) return systemId;
|
615
|
+
File file = new File(systemId);
|
616
|
+
if (file.isAbsolute()) return systemId;
|
617
|
+
String path = resolveSystemId(baseURI, systemId);
|
618
|
+
if (path != null) return path;
|
619
|
+
path = resolveSystemId(currentDir, systemId);
|
620
|
+
if (path != null) return path;
|
621
|
+
return resolveSystemId(scriptFileName, systemId);
|
622
|
+
}
|
623
|
+
|
624
|
+
private static String resolveSystemId(String baseName, String systemId) {
|
625
|
+
if (baseName == null || baseName.length() < 1) return null;
|
626
|
+
String parentName;
|
627
|
+
baseName = baseName.replace("%20", " ");
|
628
|
+
File base = new File(baseName);
|
629
|
+
if (base.isDirectory()) parentName = baseName;
|
630
|
+
else parentName = base.getParent();
|
631
|
+
if (parentName == null) return null;
|
632
|
+
if (parentName.toLowerCase().startsWith("file:")) parentName = parentName.substring("file:".length());
|
633
|
+
File dtdFile = new File(parentName + "/" + systemId);
|
634
|
+
if (dtdFile.exists()) return dtdFile.getPath();
|
635
|
+
return null;
|
636
|
+
}
|
637
|
+
|
638
|
+
private static final Charset UTF8 = Charset.forName("UTF-8");
|
639
|
+
|
640
|
+
public static boolean isUTF8(String encoding) {
|
641
|
+
if (encoding == null) return true; // no need to convert encoding
|
642
|
+
|
643
|
+
if ("UTF-8".equals(encoding)) return true;
|
644
|
+
return UTF8.aliases().contains(encoding);
|
645
|
+
}
|
646
|
+
|
647
|
+
public static ByteBuffer convertEncoding(Charset output_charset, CharSequence input_string) {
|
648
|
+
return output_charset.encode(CharBuffer.wrap(input_string)); // does replace implicitly on un-mappable characters
|
649
|
+
}
|
650
|
+
|
651
|
+
public static CharSequence convertEncodingByNKFIfNecessary(ThreadContext context, XmlDocument doc, CharSequence str) {
|
652
|
+
if (!(doc instanceof HtmlDocument)) return str;
|
653
|
+
String parsed_encoding = ((HtmlDocument)doc).getPraedEncoding();
|
654
|
+
if (parsed_encoding == null) return str;
|
655
|
+
String ruby_encoding = rubyStringToString(doc.getEncoding());
|
656
|
+
if (ruby_encoding == null) return str;
|
657
|
+
Charset encoding = Charset.forName(ruby_encoding);
|
658
|
+
if (Charset.forName(parsed_encoding).compareTo(encoding) == 0) return str;
|
659
|
+
if (str.length() == 0) return str; // no need to convert
|
660
|
+
return NokogiriHelpers.nkf(context, encoding, str);
|
661
|
+
}
|
662
|
+
|
663
|
+
private static final ByteList _Sw = new ByteList(new byte[] { '-','S','w' }, false);
|
664
|
+
private static final ByteList _Jw = new ByteList(new byte[] { '-','J','w' }, false);
|
665
|
+
private static final ByteList _Ew = new ByteList(new byte[] { '-','E','w' }, false);
|
666
|
+
private static final ByteList _Ww = new ByteList(new byte[] { '-','W','w' }, false);
|
667
|
+
|
668
|
+
// This method is used from HTML documents. HTML meta tag with encoding specification
|
669
|
+
// might appear after non-ascii characters are used. For example, a title tag before
|
670
|
+
// a meta tag. In such a case, Xerces encodes characters in UTF-8 without seeing meta tag.
|
671
|
+
// Nokogiri uses NKF library to convert characters correct encoding. This means the method
|
672
|
+
// works only for JIS/Shift_JIS/EUC-JP.
|
673
|
+
private static CharSequence nkf(ThreadContext context, Charset encoding, CharSequence str) {
|
674
|
+
final Ruby runtime = context.getRuntime();
|
675
|
+
final ByteList opt;
|
676
|
+
if (NokogiriHelpers.Shift_JIS.compareTo(encoding) == 0) opt = _Sw;
|
677
|
+
else if (NokogiriHelpers.ISO_2022_JP.compareTo(encoding) == 0) opt = _Jw;
|
678
|
+
else if (NokogiriHelpers.EUC_JP.compareTo(encoding) == 0) opt = _Ew;
|
679
|
+
else opt = _Ww; // should not come here. should be treated before this method.
|
680
|
+
|
681
|
+
Class nkfClass;
|
682
|
+
try {
|
683
|
+
// JRuby 1.7 and later
|
684
|
+
nkfClass = runtime.getClassLoader().loadClass("org.jruby.ext.nkf.RubyNKF");
|
685
|
+
} catch (ClassNotFoundException e1) {
|
686
|
+
return str;
|
687
|
+
}
|
688
|
+
Method nkf_method;
|
689
|
+
try {
|
690
|
+
nkf_method = nkfClass.getMethod("nkf", ThreadContext.class, IRubyObject.class, IRubyObject.class, IRubyObject.class);
|
691
|
+
RubyString r_str =
|
692
|
+
(RubyString)nkf_method.invoke(null, context, null, runtime.newString(opt), runtime.newString(str.toString()));
|
693
|
+
return NokogiriHelpers.rubyStringToString(r_str);
|
694
|
+
} catch (SecurityException e) {
|
695
|
+
return str;
|
696
|
+
} catch (NoSuchMethodException e) {
|
697
|
+
return str;
|
698
|
+
} catch (IllegalArgumentException e) {
|
699
|
+
return str;
|
700
|
+
} catch (IllegalAccessException e) {
|
701
|
+
return str;
|
702
|
+
} catch (InvocationTargetException e) {
|
703
|
+
return str;
|
704
|
+
}
|
705
|
+
}
|
706
|
+
|
707
|
+
private static final Charset Shift_JIS = Charset.forName("Shift_JIS");
|
708
|
+
private static final Charset ISO_2022_JP = Charset.forName("ISO-2022-JP"); // JIS
|
709
|
+
private static final Charset EUC_JP = Charset.forName("EUC-JP");
|
710
|
+
|
711
|
+
public static boolean shouldEncode(Node text) {
|
712
|
+
final Boolean encoded = (Boolean) text.getUserData(NokogiriHelpers.ENCODED_STRING);
|
713
|
+
return encoded == null || ! encoded;
|
714
|
+
}
|
715
|
+
|
716
|
+
public static boolean shouldDecode(Node text) {
|
717
|
+
return !shouldEncode(text);
|
718
|
+
}
|
719
|
+
|
720
|
+
public static NokogiriNamespaceCache getNamespaceCache(Node node) {
|
721
|
+
XmlDocument xmlDoc = (XmlDocument) getCachedNode(node.getOwnerDocument());
|
722
|
+
return xmlDoc.getNamespaceCache();
|
723
|
+
}
|
724
|
+
|
725
|
+
public static Node renameNode(Node node, String namespaceURI, String qualifiedName) throws DOMException {
|
726
|
+
Document doc = node.getOwnerDocument();
|
727
|
+
NokogiriNamespaceCache nsCache = getNamespaceCache(node);
|
728
|
+
Node result = doc.renameNode(node, namespaceURI, qualifiedName);
|
729
|
+
if (result != node) {
|
730
|
+
nsCache.replaceNode(node, result);
|
731
|
+
}
|
732
|
+
return result;
|
733
|
+
}
|
734
|
+
}
|