nokogiri 1.5.5.rc3-java → 1.5.6-java

data/ext/java/nokogiri/NokogiriService.java

@@ -100,6 +100,7 @@ public class NokogiriService implements BasicLibraryService {
         RubyModule htmlSaxModule = htmlModule.defineModuleUnder("SAX");
         RubyModule xsltModule = nokogiri.defineModuleUnder("XSLT");
 
+        createJavaLibraryVersionConstants(ruby, nokogiri);
         createNokogiriModule(ruby, nokogiri);
         createSyntaxErrors(ruby, nokogiri, xmlModule);
         RubyClass xmlNode = createXmlModule(ruby, xmlModule);
@@ -108,7 +109,12 @@ public class NokogiriService implements BasicLibraryService {
         createSaxModule(ruby, xmlSaxModule, htmlSaxModule);
         createXsltModule(ruby, xsltModule);
     }
-    
+
+    private void createJavaLibraryVersionConstants(Ruby ruby, RubyModule nokogiri) {
+        nokogiri.defineConstant("XERCES_VERSION", ruby.newString(org.apache.xerces.impl.Version.getVersion()));
+        nokogiri.defineConstant("NEKO_VERSION", ruby.newString(org.cyberneko.html.Version.getVersion()));
+    }
+
     private void createNokogiriModule(Ruby ruby, RubyModule nokogiri) {;
         RubyClass encHandler = nokogiri.defineClassUnder("EncodingHandler", ruby.getObject(), ENCODING_HANDLER_ALLOCATOR);
         encHandler.defineAnnotatedMethods(EncodingHandler.class);

data/ext/java/nokogiri/XmlDocument.java

@@ -64,6 +64,7 @@ import org.jruby.runtime.ThreadContext;
 import org.jruby.runtime.builtin.IRubyObject;
 import org.w3c.dom.Attr;
 import org.w3c.dom.Document;
+import org.w3c.dom.DocumentType;
 import org.w3c.dom.NamedNodeMap;
 import org.w3c.dom.Node;
 import org.w3c.dom.NodeList;
@@ -106,7 +107,7 @@ public class XmlDocument extends XmlNode {
 
     public XmlDocument(Ruby ruby, RubyClass klass, Document document) {
         super(ruby, klass, document);
-        nsCache = new NokogiriNamespaceCache();
+        initializeNamespaceCacheIfNecessary();
         createAndCacheNamespaces(ruby, document.getDocumentElement());
         stabilizeTextContent(document);
         setInstanceVariable("@decorators", ruby.getNil());
@@ -114,7 +115,7 @@ public class XmlDocument extends XmlNode {
     
     public void setDocumentNode(ThreadContext context, Node node) {
         super.setNode(context, node);
-        if (nsCache == null) nsCache = new NokogiriNamespaceCache();
+        initializeNamespaceCacheIfNecessary();
         Ruby runtime = context.getRuntime();
         if (node != null) {
             Document document = (Document)node;
@@ -195,6 +196,10 @@ public class XmlDocument extends XmlNode {
         return nsCache;
     }
     
+    public void initializeNamespaceCacheIfNecessary() {
+        if (nsCache == null) nsCache = new NokogiriNamespaceCache();
+    }
+    
     public void setNamespaceCache(NokogiriNamespaceCache nsCache) {
         this.nsCache = nsCache;
     }
@@ -284,7 +289,7 @@ public class XmlDocument extends XmlNode {
 
     @JRubyMethod
     public IRubyObject encoding(ThreadContext context) {
-        if (this.encoding == null) {
+        if (this.encoding == null || this.encoding.isNil()) {
             if (getDocument().getXmlEncoding() == null) {
                 this.encoding = context.getRuntime().getNil();
             } else {
@@ -292,7 +297,7 @@ public class XmlDocument extends XmlNode {
             }
         }
 
-        return this.encoding;
+        return this.encoding.isNil() ? this.encoding : this.encoding.asString().encode(context, context.getRuntime().newString("UTF-8"));
     }
 
     @JRubyMethod(meta = true)
@@ -434,8 +439,20 @@ public class XmlDocument extends XmlNode {
         IRubyObject dtd = (IRubyObject) node.getUserData(DTD_INTERNAL_SUBSET);
 
         if (dtd == null) {
-            if (getDocument().getDoctype() == null) dtd = context.getRuntime().getNil();
-            else dtd = XmlDtd.newFromInternalSubset(context.getRuntime(), getDocument());
+            Document document = getDocument();
+            if (document.getUserData(XmlDocument.DTD_RAW_DOCUMENT) != null) {
+                dtd = XmlDtd.newFromInternalSubset(context.getRuntime(), document);
+            } else if (document.getDoctype() != null) {
+                DocumentType docType = document.getDoctype();
+                dtd = XmlDtd.newEmpty(context.getRuntime(),
+                                      document,
+                                      context.getRuntime().newString(docType.getName()),
+                                      context.getRuntime().newString(docType.getPublicId()),
+                                      context.getRuntime().newString(docType.getSystemId()));
+            } else {
+                dtd = context.getRuntime().getNil();
+            }
+
             setInternalSubset(dtd);
         }
 
@@ -518,6 +535,7 @@ public class XmlDocument extends XmlNode {
     @JRubyMethod(meta=true)
     public static IRubyObject wrapJavaDocument(ThreadContext context, IRubyObject klazz, IRubyObject arg) {
         XmlDocument xmlDocument = (XmlDocument) NokogiriService.XML_DOCUMENT_ALLOCATOR.allocate(context.getRuntime(), getNokogiriClass(context.getRuntime(), "Nokogiri::XML::Document"));
+        RuntimeHelpers.invoke(context, xmlDocument, "initialize");
         Document document = (Document)arg.toJava(Document.class);
         xmlDocument.setDocumentNode(context, document);
         return xmlDocument;

data/ext/java/nokogiri/XmlDocumentFragment.java

@@ -33,7 +33,6 @@
 package nokogiri;
 
 import static nokogiri.internals.NokogiriHelpers.getLocalNameForNamespace;
-import static nokogiri.internals.NokogiriHelpers.getLocalPart;
 import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
 import static nokogiri.internals.NokogiriHelpers.getPrefix;
 import static nokogiri.internals.NokogiriHelpers.isNamespace;
@@ -93,7 +92,6 @@ public class XmlDocumentFragment extends XmlNode {
         if (args.length > 1 && args[1] instanceof RubyString) {
             args[1] = trim(context, doc, (RubyString)args[1]);
             if (XmlDocumentFragment.isTag((RubyString)args[1])) {
-                args[1] = RubyString.newString(context.getRuntime(), ignoreNamespaceIfNeeded(doc, rubyStringToString(args[1])));
                 args[1] = RubyString.newString(context.getRuntime(), addNamespaceDeclIfNeeded(doc, rubyStringToString(args[1])));
             }
         }
@@ -127,32 +125,10 @@ public class XmlDocumentFragment extends XmlNode {
         if (str.startsWith("<") && str.endsWith(">")) return true;
         return false;
     }
-    
+
     private static Pattern qname_pattern = Pattern.compile("[^</:>\\s]+:[^</:>=\\s]+");
     private static Pattern starttag_pattern = Pattern.compile("<[^</>]+>");
-    
-    private static String ignoreNamespaceIfNeeded(XmlDocument doc, String tags) {
-        if (doc.getDocument() == null) return tags;
-        Matcher matcher = qname_pattern.matcher(tags);
-        Map<String, String> rewriteTable = new HashMap<String, String>();
-        while(matcher.find()) {
-            String qName = matcher.group();
-            if (doc.getDocument().getDocumentElement() != null) {
-                NamedNodeMap nodeMap = doc.getDocument().getDocumentElement().getAttributes();
-                if (!isNamespaceDefined(qName, nodeMap)) {
-                    rewriteTable.put(qName, getLocalPart(qName));
-                }
-            } else {
-                rewriteTable.put(qName, getLocalPart(qName));
-            }
-        }
-        Set<String> keys = rewriteTable.keySet();
-        for (String key : keys) {
-            tags = tags.replace(key, rewriteTable.get(key));
-        }
-        return tags;
-    }
-    
+
     private static boolean isNamespaceDefined(String qName, NamedNodeMap nodeMap) {
         if (isNamespace(qName.intern())) return true;
         for (int i=0; i < nodeMap.getLength(); i++) {

data/ext/java/nokogiri/XmlDtd.java

@@ -129,9 +129,20 @@ public class XmlDtd extends XmlNode {
                                   IRubyObject name,
                                   IRubyObject external_id,
                                   IRubyObject system_id) {
-        Element placeHolder = doc.createElement("dtd_placeholder");
+
+        DocumentType placeholder = null;
+        if (doc.getDoctype() == null) {
+          String javaName = NokogiriHelpers.rubyStringToString(name);
+          String javaExternalId = NokogiriHelpers.rubyStringToString(external_id);
+          String javaSystemId = NokogiriHelpers.rubyStringToString(system_id);
+          placeholder = doc.getImplementation().createDocumentType(javaName, javaExternalId, javaSystemId);
+          doc.appendChild(placeholder);
+        } else {
+          placeholder = doc.getDoctype();
+        }
+        // FIXME: what if the document had a doc type, why are we here ?
         XmlDtd dtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
-        dtd.setNode(runtime, placeHolder);
+        dtd.setNode(runtime, placeholder);
         dtd.name = name;
         dtd.pubId = external_id;
         dtd.sysId = system_id;

data/ext/java/nokogiri/XmlElement.java

@@ -38,7 +38,6 @@ import org.jruby.Ruby;
 import org.jruby.RubyArray;
 import org.jruby.RubyClass;
 import org.jruby.anno.JRubyClass;
-import org.jruby.javasupport.util.RuntimeHelpers;
 import org.jruby.runtime.ThreadContext;
 import org.w3c.dom.Element;
 import org.w3c.dom.Node;
@@ -62,17 +61,9 @@ public class XmlElement extends XmlNode {
     
     @Override
     public void setNode(ThreadContext context, Node node) {
-        this.node = node;
-        if (node != null) {
-            resetCache();
-            if (node.getNodeType() != Node.DOCUMENT_NODE) {
-                doc = document(context);
-                setInstanceVariable("@document", doc);
-                if (doc != null) {
-                    RuntimeHelpers.invoke(context, doc, "decorate", this);
-                }
-            }
-        }
+      super.setNode(context, node);
+      if (doc != null)
+        setInstanceVariable("@document", doc);
     }
     
     @Override

data/ext/java/nokogiri/XmlEntityReference.java

@@ -1,7 +1,7 @@
 /**
  * (The MIT License)
  *
- * Copyright (c) 2008 - 2011:
+ * Copyright (c) 2008 - 2012:
  *
  * * {Aaron Patterson}[http://tenderlovemaking.com]
  * * {Mike Dalessio}[http://mike.daless.io]
@@ -34,7 +34,6 @@ package nokogiri;
 
 import static nokogiri.internals.NokogiriHelpers.getCachedNodeOrCreate;
 import static nokogiri.internals.NokogiriHelpers.rubyStringToString;
-
 import nokogiri.internals.SaveContextVisitor;
 
 import org.jruby.Ruby;
@@ -43,7 +42,6 @@ import org.jruby.anno.JRubyClass;
 import org.jruby.runtime.ThreadContext;
 import org.jruby.runtime.builtin.IRubyObject;
 import org.w3c.dom.Document;
-import org.w3c.dom.EntityReference;
 import org.w3c.dom.Node;
 
 /**
@@ -72,14 +70,21 @@ public class XmlEntityReference extends XmlNode {
         IRubyObject doc = args[0];
         IRubyObject name = args[1];
 
-        Document document = ((XmlNode) doc).getOwnerDocument();
-        Node node = document.createEntityReference(rubyStringToString(name));
-        setNode(context, node);
+        Document owner = ((XmlNode) doc).getOwnerDocument();
+        Node node = new NokogiriEntityReference(owner, rubyStringToString(name));
+        super.setNode(context, node);
+    }
+    
+    public void setNode(ThreadContext context, Node entityRef) {
+        Document owner = entityRef.getOwnerDocument();
+        String name = entityRef.getNodeName();
+        Node node = new NokogiriEntityReference(owner, name);
+        super.setNode(context, node);
     }
     
     @Override
     public void accept(ThreadContext context, SaveContextVisitor visitor) {
-        visitor.enter((EntityReference)node);
+        visitor.enterEntityReference((NokogiriEntityReference)node);
         Node child = node.getFirstChild();
         while (child != null) {
             IRubyObject nokoNode = getCachedNodeOrCreate(context.getRuntime(), child);
@@ -92,6 +97,25 @@ public class XmlEntityReference extends XmlNode {
             }
             child = child.getNextSibling();
         }
-        visitor.leave((EntityReference)node);
+        visitor.leaveEntityReference((NokogiriEntityReference)node);
+    }
+    
+    public class NokogiriEntityReference extends org.apache.xerces.dom.TextImpl {
+        // Nokogiri's EntityReference should quack like a org.w3c.dom.Text node.
+        // EntityReference node should not raise exception for names such as #xa.
+        // This is reported bug in issue#719.
+        // Also, EntityReference node should not bother xpath.
+        // For this purpose, Node type should be Node.TEXT_NODE.
+        public NokogiriEntityReference(Document owner, String name) {
+            super((org.apache.xerces.dom.DocumentImpl)owner, name);
+        }
+        
+        public short getNodeType() {
+            return Node.TEXT_NODE;
+        }
+
+        public String getNodeName() {
+            return getNodeValue();
+        }        
     }
 }

data/ext/java/nokogiri/XmlNamespace.java

@@ -98,7 +98,6 @@ public class XmlNamespace extends RubyObject {
         this.href = href;
         this.prefixString = prefixString;
         this.hrefString = hrefString;
-        this.attr.setUserData(CACHED_NODE, this, null);
         setInstanceVariable("@document", xmlDocument);
     }
     
@@ -115,6 +114,7 @@ public class XmlNamespace extends RubyObject {
         IRubyObject href_value = RubyString.newString(runtime, hrefValue);
         // check namespace cache
         XmlDocument xmlDocument = (XmlDocument)getCachedNodeOrCreate(runtime, attr.getOwnerDocument());
+        xmlDocument.initializeNamespaceCacheIfNecessary();
         XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefixValue, hrefValue);
         if (xmlNamespace != null) return xmlNamespace;
         
@@ -135,6 +135,7 @@ public class XmlNamespace extends RubyObject {
         Document document = owner.getOwnerDocument();
         // check namespace cache
         XmlDocument xmlDocument = (XmlDocument)getCachedNodeOrCreate(runtime, document);
+        xmlDocument.initializeNamespaceCacheIfNecessary();
         XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefixValue, hrefValue);
         if (xmlNamespace != null) return xmlNamespace;
 

data/ext/java/nokogiri/XmlNode.java

@@ -45,6 +45,7 @@ import java.io.InputStream;
 import java.nio.charset.CharacterCodingException;
 import java.nio.charset.Charset;
 import java.util.ArrayList;
+import java.util.Iterator;
 import java.util.List;
 
 import nokogiri.internals.HtmlDomParserContext;
@@ -210,22 +211,23 @@ public class XmlNode extends RubyObject {
      */
     public XmlNode(Ruby ruby, RubyClass cls, Node node) {
         super(ruby, cls);
-        this.node = node;
+        setNode(ruby.getCurrentContext(), node);
+    }
 
+    protected void decorate(Ruby ruby) {
         if (node != null) {
             resetCache();
 
             if (node.getNodeType() != Node.DOCUMENT_NODE) {
                 doc = document(ruby.getCurrentContext());
 
-                if (doc != null) {
+                if (doc != null && doc.isTrue()) {
                     RuntimeHelpers.invoke(ruby.getCurrentContext(), doc, "decorate", this);
                 }
             }
         }
-        
     }
-    
+
     /**
      * Create and return a copy of this object.
      *
@@ -448,6 +450,7 @@ public class XmlNode extends RubyObject {
     public void relink_namespace(ThreadContext context) {
         if (node instanceof Element) {
             Element e = (Element) node;
+            e.getOwnerDocument().setStrictErrorChecking(false);
             e.getOwnerDocument().renameNode(e, e.lookupNamespaceURI(e.getPrefix()), e.getNodeName());
 
             if (e.hasAttributes()) {
@@ -465,6 +468,9 @@ public class XmlNode extends RubyObject {
                     } else {
                         nsUri = attr.getNamespaceURI();
                     }
+                    if (!(nsUri == null || "".equals(nsUri))) {
+                        XmlNamespace.createFromAttr(context.getRuntime(), attr);
+                    }
                     e.getOwnerDocument().renameNode(attr, nsUri, nodeName);
                 }
             }
@@ -509,13 +515,8 @@ public class XmlNode extends RubyObject {
 
     public void setNode(ThreadContext context, Node node) {
         this.node = node;
-        
-        if (node != null) {
-            resetCache();
-            if (node.getNodeType() != Node.DOCUMENT_NODE) {
-                doc = document(context);
-            }
-        }
+
+        decorate(context.getRuntime());
 
         if (this instanceof XmlAttr) {
             ((XmlAttr)this).setNamespaceIfNecessary(context.getRuntime());
@@ -547,7 +548,7 @@ public class XmlNode extends RubyObject {
             str = NokogiriHelpers.getLocalPart(str);
         }
         if (str == null) str = "";
-        name = context.getRuntime().newString(str);
+        name = NokogiriHelpers.stringOrBlank(context.getRuntime(), str);
         return name;
     }
 
@@ -804,23 +805,46 @@ public class XmlNode extends RubyObject {
 
     @JRubyMethod(name = {"content", "text", "inner_text"})
     public IRubyObject content(ThreadContext context) {
-        if (content != null && content.isNil()) return content;
+        if (!node.hasChildNodes() && node.getNodeValue() == null &&
+            (node.getNodeType() == Node.TEXT_NODE || node.getNodeType() == Node.CDATA_SECTION_NODE))
+          return context.nil;
         String textContent;
-        if (content != null) textContent = rubyStringToString(content);
-        else if (this instanceof XmlDocument) {
+        if (this instanceof XmlDocument) {
             Node node = ((Document)this.node).getDocumentElement();
             if (node == null) {
                 textContent = "";
             } else {
-                textContent = ((Document)this.node).getDocumentElement().getTextContent().trim();
+                Node documentElement = ((Document)this.node).getDocumentElement();
+                StringBuffer buffer = new StringBuffer();
+                getTextContentRecursively(context, buffer, documentElement);
+                textContent = buffer.toString();
             }
         } else {
-            textContent = this.node.getTextContent();
+            StringBuffer buffer = new StringBuffer();
+            getTextContentRecursively(context, buffer, node);
+            textContent = buffer.toString();
         }
-        textContent = NokogiriHelpers.convertEncodingByNKFIfNecessary(context.getRuntime(), (XmlDocument)document(context), textContent);
-        String decodedText = null;
-        if (textContent != null) decodedText = NokogiriHelpers.decodeJavaString(textContent);
-        return stringOrNil(context.getRuntime(), decodedText);
+        NokogiriHelpers.convertEncodingByNKFIfNecessary(context.getRuntime(), (XmlDocument)document(context), textContent);
+        return stringOrNil(context.getRuntime(), textContent);
+    }
+
+    private void getTextContentRecursively(ThreadContext context, StringBuffer buffer, Node currentNode) {
+      String textContent = currentNode.getNodeValue();
+      if (textContent != null && NokogiriHelpers.shouldDecode(currentNode))
+        textContent = NokogiriHelpers.decodeJavaString(textContent);
+      if (textContent != null)
+        buffer.append(textContent);
+      NodeList children = currentNode.getChildNodes();
+      for (int i = 0; i < children.getLength(); i++) {
+        Node child = children.item(i);
+        if (hasTextContent(child))
+          getTextContentRecursively(context, buffer, child);
+      }
+    }
+
+    private boolean hasTextContent(Node child) {
+      return child.getNodeType() != Node.COMMENT_NODE &&
+          child.getNodeType() != Node.PROCESSING_INSTRUCTION_NODE;
     }
 
     @JRubyMethod
@@ -888,9 +912,7 @@ public class XmlNode extends RubyObject {
             String key = rubyStringToString(rbkey);
             Element element = (Element) node;
             String value = element.getAttribute(key);
-            if (value != null) {
-                return context.getRuntime().newString(value);
-            }
+            return nonEmptyStringOrNil(context.getRuntime(), value);
         }
         return context.getRuntime().getNil();
     }
@@ -1000,7 +1022,7 @@ public class XmlNode extends RubyObject {
         NokogiriNamespaceCache nsCache = xmlDocument.getNamespaceCache();
         String prefix = node.getPrefix();
         XmlNamespace namespace = nsCache.get(prefix == null ? "" : prefix, node.getNamespaceURI());
-        if (namespace == null || ((XmlNamespace) namespace).isEmpty()) {
+        if (namespace == null || namespace.isEmpty()) {
             return context.getRuntime().getNil();
         }
 
@@ -1023,10 +1045,10 @@ public class XmlNode extends RubyObject {
         if (doc instanceof HtmlDocument) return namespace_definitions;
         List<XmlNamespace> namespaces = ((XmlDocument)doc).getNamespaceCache().get(node);
         for (XmlNamespace namespace : namespaces) {
-            ((RubyArray)namespace_definitions).append(namespace);
+            namespace_definitions.append(namespace);
         }
 
-        return (RubyArray) namespace_definitions;
+        return namespace_definitions;
     }
 
     /**
@@ -1056,12 +1078,13 @@ public class XmlNode extends RubyObject {
     }
 
     protected void setContent(IRubyObject content) {
-        this.content = content;
         String javaContent = rubyStringToString(content);
         node.setTextContent(javaContent);
         if (javaContent.length() == 0) return;
         if (node.getNodeType() == Node.TEXT_NODE || node.getNodeType() == Node.CDATA_SECTION_NODE) return;
-        node.getFirstChild().setUserData(NokogiriHelpers.ENCODED_STRING, true, null);
+        if (node.getFirstChild() != null) {
+            node.getFirstChild().setUserData(NokogiriHelpers.ENCODED_STRING, true, null);
+        }
     }
 
     private void setContent(String content) {
@@ -1069,7 +1092,7 @@ public class XmlNode extends RubyObject {
         this.content = null;    // clear cache
     }
 
-    @JRubyMethod(name = "native_content=", visibility = Visibility.PRIVATE)
+    @JRubyMethod(name = "native_content=")
     public IRubyObject native_content_set(ThreadContext context, IRubyObject content) {
         setContent(content);
         return content;
@@ -1158,13 +1181,42 @@ public class XmlNode extends RubyObject {
             String key = rubyStringToString(rbkey);
             String val = rubyStringToString(rbval);
             Element element = (Element) node;
-            element.setAttribute(key, val);
+
+            int colonIndex = key.indexOf(":");
+            if (colonIndex > 0) {
+              String prefix = key.substring(0, colonIndex);
+              String uri = null;
+              if (prefix.equals("xml")) {
+                uri = "http://www.w3.org/XML/1998/namespace";
+              } else {
+                uri = findNamespaceHref(context, prefix);
+              }
+              element.setAttributeNS(uri, key, val);
+            } else {
+              element.setAttribute(key, val);
+            }
             return this;
         } else {
             return rbval;
         }
     }
 
+    private String findNamespaceHref(ThreadContext context, String prefix) {
+      XmlNode currentNode = this;
+      while(currentNode != document(context)) {
+        RubyArray namespaces = (RubyArray) currentNode.namespace_scopes(context);
+        Iterator iterator = namespaces.iterator();
+        while(iterator.hasNext()) {
+          XmlNamespace namespace = (XmlNamespace) iterator.next();
+          if (namespace.getPrefix().equals(prefix)) {
+            return namespace.getHref();
+          }
+        }
+        currentNode = (XmlNode) currentNode.parent(context);
+      }
+      return null;
+    }
+
     @JRubyMethod
     public IRubyObject parent(ThreadContext context) {
         /*