nokogiri 1.5.2-java → 1.5.3-java

data/ext/java/nokogiri/HtmlDocument.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]
@@ -55,6 +55,7 @@ import org.w3c.dom.NodeList;
  */
 @JRubyClass(name="Nokogiri::HTML::Document", parent="Nokogiri::XML::Document")
 public class HtmlDocument extends XmlDocument {
+    private String parsed_encoding = null;
 
     public HtmlDocument(Ruby ruby, RubyClass klazz) {
         super(ruby, klazz);
@@ -123,6 +124,14 @@ public class HtmlDocument extends XmlDocument {
             stabilzeAttrValue(children.item(i));
         }
     }
+    
+    public void setParsedEncoding(String encoding) {
+        parsed_encoding = encoding;
+    }
+    
+    public String getPraedEncoding() {
+        return parsed_encoding;
+    }
 
     /*
      * call-seq:

data/ext/java/nokogiri/XmlAttr.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]
@@ -96,6 +96,16 @@ public class XmlAttr extends XmlNode{
         Node attr = xmlDoc.getDocument().createAttribute(str);
         setNode(context, attr);
     }
+    
+    
+    // this method is called from XmlNode.setNode()
+    // if the node is attribute, and its name has prefix "xml"
+    // the default namespace should be registered for this attribute
+    void setNamespaceIfNecessary(Ruby runtime) {
+        if ("xml".equals(node.getPrefix())) {
+           XmlNamespace.createDefaultNamespace(runtime, node); 
+        }
+    }
 
     private boolean isHtmlBooleanAttr() {
         String name = node.getNodeName().toLowerCase();

data/ext/java/nokogiri/XmlDocument.java

@@ -128,6 +128,10 @@ public class XmlDocument extends XmlNode {
         this.encoding = encoding;
     }
     
+    public IRubyObject getEncoding() {
+        return encoding;
+    }
+    
     // not sure, but like attribute values, text value will be lost
     // unless it is referred once before this document is used.
     // this seems to happen only when the fragment is parsed from Node#in_context.

data/ext/java/nokogiri/XmlNamespace.java

@@ -36,6 +36,7 @@ import static nokogiri.internals.NokogiriHelpers.CACHED_NODE;
 import static nokogiri.internals.NokogiriHelpers.getCachedNodeOrCreate;
 import static nokogiri.internals.NokogiriHelpers.getLocalNameForNamespace;
 import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
+import static nokogiri.internals.NokogiriHelpers.stringOrNil;
 import nokogiri.internals.NokogiriHelpers;
 import nokogiri.internals.SaveContextVisitor;
 
@@ -155,6 +156,30 @@ public class XmlNamespace extends RubyObject {
         return namespace;
     }
     
+    // owner should be an Attr node
+    public static XmlNamespace createDefaultNamespace(Ruby runtime, Node owner) {
+        String prefixValue = owner.getPrefix();
+        String hrefValue = owner.getNamespaceURI();
+        Document document = owner.getOwnerDocument();
+        // check namespace cache
+        XmlDocument xmlDocument = (XmlDocument)getCachedNodeOrCreate(runtime, document);
+        XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefixValue, hrefValue);
+        if (xmlNamespace != null) return xmlNamespace;
+
+        // creating XmlNamespace instance
+        XmlNamespace namespace =
+            (XmlNamespace) NokogiriService.XML_NAMESPACE_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::Namespace"));
+
+        IRubyObject prefix = stringOrNil(runtime, prefixValue);
+        IRubyObject href = stringOrNil(runtime, hrefValue);
+        // initialize XmlNamespace object
+        namespace.init((Attr)owner, prefix, href, prefixValue, hrefValue, xmlDocument);
+        
+        // updating namespace cache
+        xmlDocument.getNamespaceCache().put(namespace, owner);
+        return namespace;
+    }
+    
     /**
      * Create and return a copy of this object.
      *

data/ext/java/nokogiri/XmlNode.java

@@ -42,13 +42,8 @@ import static nokogiri.internals.NokogiriHelpers.stringOrNil;
 
 import java.io.ByteArrayInputStream;
 import java.io.InputStream;
-import java.io.UnsupportedEncodingException;
-import java.nio.ByteBuffer;
-import java.nio.CharBuffer;
 import java.nio.charset.CharacterCodingException;
 import java.nio.charset.Charset;
-import java.nio.charset.CharsetDecoder;
-import java.nio.charset.CharsetEncoder;
 import java.util.ArrayList;
 import java.util.List;
 
@@ -479,6 +474,10 @@ public class XmlNode extends RubyObject {
                 doc = document(context);
             }
         }
+
+        if (this instanceof XmlAttr) {
+            ((XmlAttr)this).setNamespaceIfNecessary(context.getRuntime());
+        }
     }
 
     public void updateNodeNamespaceIfNecessary(ThreadContext context, XmlNamespace ns) {
@@ -763,6 +762,7 @@ public class XmlNode extends RubyObject {
         } else {
             textContent = this.node.getTextContent();
         }
+        textContent = NokogiriHelpers.convertEncodingByNKFIfNecessary(context.getRuntime(), (XmlDocument)document(context), textContent);
         String decodedText = null;
         if (textContent != null) decodedText = NokogiriHelpers.decodeJavaString(textContent);
         return stringOrNil(context.getRuntime(), decodedText);
@@ -1342,7 +1342,7 @@ public class XmlNode extends RubyObject {
          */
         if (parent.getNodeType() == Node.DOCUMENT_NODE &&
             otherNode.getNodeType() == Node.TEXT_NODE) {
-            Element e = ((Document)parent).createElement("text");
+            Element e = ((Document)parent).createElement("nokogiri_text_wrapper");
             e.appendChild(otherNode);
             otherNode = e;
         }

data/ext/java/nokogiri/XmlReader.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]
@@ -228,8 +228,15 @@ public class XmlReader extends RubyObject {
         if (!current.hasChildren) return null;
         StringBuffer sb = new StringBuffer();
         int currentDepth = (Integer)current.depth;
+        int inner = 0;
         for (ReaderNode node : nodeQueue) {
-            if (((Integer)node.depth) > currentDepth) sb.append(node.getString());
+            if (((Integer)node.depth) == currentDepth && node.getName().equals(current.getName())) {
+                inner++;
+            }
+            if (((Integer)node.depth) > currentDepth) {
+                sb.append(node.getString());
+            }
+            if (inner == 2) break;
         }
         return new String(sb);
     }
@@ -242,9 +249,17 @@ public class XmlReader extends RubyObject {
     private String getOuterXml(ArrayDeque<ReaderNode> nodeQueue, ReaderNode current) {
         if (current.depth < 0) return null;
         StringBuffer sb = new StringBuffer();
-        int initialDepth = (Integer)current.depth - 1;
+        int initialDepth = (Integer)current.depth;
+        int inner = 0;
         for (ReaderNode node : nodeQueue) {
-            if (((Integer)node.depth) > initialDepth) sb.append(node.getString());
+            if (((Integer)node.depth) >= initialDepth) {
+                if (((Integer)node.depth) == initialDepth && node.getName().equals(current.getName())) {
+                    inner++;
+                }
+
+                sb.append(node.getString());
+            }
+            if (inner == 2) break;
         }
         return new String(sb);
     }

data/ext/java/nokogiri/XmlSaxPushParser.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]
@@ -37,10 +37,15 @@ import static org.jruby.javasupport.util.RuntimeHelpers.invoke;
 
 import java.io.IOException;
 import java.io.InputStream;
-import java.nio.channels.ClosedChannelException;
+import java.io.OutputStream;
+import java.nio.channels.Channels;
+import java.nio.channels.Pipe;
+import java.util.concurrent.Callable;
+import java.util.concurrent.ExecutorService;
+import java.util.concurrent.Executors;
+import java.util.concurrent.FutureTask;
 
 import nokogiri.internals.ParserContext;
-import nokogiri.internals.PushInputStream;
 
 import org.jruby.Ruby;
 import org.jruby.RubyClass;
@@ -63,9 +68,12 @@ import org.jruby.runtime.builtin.IRubyObject;
 public class XmlSaxPushParser extends RubyObject {
     ParserContext.Options options;
     IRubyObject optionsRuby;
-    PushInputStream stream;
-    Thread reader;
-    Runner runner;
+    IRubyObject saxParser;
+    OutputStream ostream = null;
+    InputStream istream = null;
+    ParserTask parserTask = null;
+    FutureTask<XmlSaxParserContext> futureTask = null;
+    ExecutorService executor = null;
 
     public XmlSaxPushParser(Ruby ruby, RubyClass rubyClass) {
         super(ruby, rubyClass);
@@ -73,18 +81,12 @@ public class XmlSaxPushParser extends RubyObject {
 
     @JRubyMethod
     public IRubyObject initialize_native(final ThreadContext context,
-                                         IRubyObject _saxParser,
+                                         IRubyObject saxParser,
                                          IRubyObject fileName) {
-        optionsRuby = invoke(context,
-                             context.getRuntime().getClassFromPath("Nokogiri::XML::ParseOptions"),
-                             "new");
+        optionsRuby
+            = invoke(context, context.getRuntime().getClassFromPath("Nokogiri::XML::ParseOptions"), "new");
         options = new ParserContext.Options(0);
-        stream = new PushInputStream();
-
-        runner = new Runner(context, this, stream);
-        reader = new Thread(runner);
-        reader.start();
-
+        this.saxParser = saxParser;
         return this;
     }
 
@@ -110,74 +112,103 @@ public class XmlSaxPushParser extends RubyObject {
     @JRubyMethod
     public IRubyObject native_write(ThreadContext context, IRubyObject chunk,
                                     IRubyObject isLast) {
+        try {
+            initialize_task(context);
+        } catch (IOException e) {
+            throw context.getRuntime().newRuntimeError(e.getMessage());
+        }
         byte[] data = null;
         if (chunk instanceof RubyString || chunk.respondsTo("to_str")) {
             data = chunk.convertToString().getBytes();
-        } else {
-            XmlSyntaxError xmlSyntaxError = (XmlSyntaxError) NokogiriService.XML_SYNTAXERROR_ALLOCATOR.allocate(context.getRuntime(), getNokogiriClass(context.getRuntime(), "Nokogiri::XML::SyntaxError"));
+        } else { 
+            try {
+                terminateTask();
+            } catch (IOException e) {
+                throw context.getRuntime().newRuntimeError(e.getMessage());
+            }
+            XmlSyntaxError xmlSyntaxError = 
+                (XmlSyntaxError) NokogiriService.XML_SYNTAXERROR_ALLOCATOR.allocate(context.getRuntime(), getNokogiriClass(context.getRuntime(), "Nokogiri::XML::SyntaxError"));
             throw new RaiseException(xmlSyntaxError);
         }
 
-        int errorCount0 = runner.getErrorCount();
-
+        int errorCount0 = parserTask.getErrorCount();;
+        
         try {
-            stream.writeAndWaitForRead(data);
-        } catch (ClosedChannelException e) {
-            // ignore
+            if (isLast.isTrue()) {
+                IRubyObject document = invoke(context, this, "document");
+                invoke(context, document, "end_document");
+                terminateTask();
+            } else {
+                ostream.write(data);
+                Thread.currentThread().sleep(10);   // gives a reader a chance to work
+            }
         } catch (IOException e) {
-            throw context.getRuntime().newRuntimeError(e.toString());
+            throw context.getRuntime().newRuntimeError(e.getMessage());
+        } catch (InterruptedException e) {
+            throw context.getRuntime().newRuntimeError(e.getMessage());
         }
 
-        if (isLast.isTrue()) {
+        if (!options.recover && parserTask.getErrorCount() > errorCount0) {
             try {
-                stream.close();
+                terminateTask();
             } catch (IOException e) {
-                // ignore
-            }
-
-            for (;;) {
-                try {
-                    reader.join();
-                    break;
-                } catch (InterruptedException e) {
-                    // continue loop
-                }
+                throw context.getRuntime().newRuntimeError(e.getMessage());
             }
-        }
-
-        if (!options.recover && runner.getErrorCount() > errorCount0) {
-            throw new RaiseException(runner.getLastError(), true);
+            throw new RaiseException(parserTask.getLastError(), true);
         }
 
         return this;
     }
-
-    protected static class Runner implements Runnable {
-        protected ThreadContext context;
-        protected IRubyObject handler;
-        protected XmlSaxParserContext parser;
-
-        public Runner(ThreadContext context,
-                      IRubyObject handler,
-                      InputStream stream) {
+    
+    private void initialize_task(ThreadContext context) throws IOException {
+        if (futureTask == null || ostream == null || istream == null) {
+            Pipe pipe = Pipe.open();
+            Pipe.SinkChannel sink = pipe.sink();
+            ostream = Channels.newOutputStream(sink);
+            Pipe.SourceChannel source = pipe.source();
+            istream = Channels.newInputStream(source);
+
+            parserTask = new ParserTask(context, saxParser);
+            futureTask = new FutureTask<XmlSaxParserContext>(parserTask);
+            executor = Executors.newSingleThreadExecutor();
+            executor.submit(futureTask);
+        }
+    }
+    
+    private synchronized void terminateTask() throws IOException {
+        futureTask.cancel(true);
+        executor.shutdown();
+        ostream.close();
+        istream.close();
+        ostream = null;
+        istream = null;
+    }
+    
+    private class ParserTask implements Callable<XmlSaxParserContext> {
+        private ThreadContext context;
+        private IRubyObject handler;
+        private XmlSaxParserContext parser;
+        
+        private ParserTask(ThreadContext context, IRubyObject handler) {
             RubyClass klazz = getNokogiriClass(context.getRuntime(), "Nokogiri::XML::SAX::ParserContext");
-
             this.context = context;
             this.handler = handler;
-            this.parser = (XmlSaxParserContext) XmlSaxParserContext.parse_stream(context, klazz, stream);
+            this.parser = (XmlSaxParserContext) XmlSaxParserContext.parse_stream(context, klazz, istream);
         }
 
-        public void run() {
+        @Override
+        public XmlSaxParserContext call() throws Exception {
             parser.parse_with(context, handler);
+            return parser;
         }
-
-        public int getErrorCount() {
-            // check for null because thread may nto have started yet
+        
+        private synchronized int getErrorCount() {
+            // check for null because thread may not have started yet
             if (parser.getNokogiriHandler() == null) return 0;
             else return parser.getNokogiriHandler().getErrorCount();
         }
-
-        public RubyException getLastError() {
+        
+        private synchronized RubyException getLastError() {
             return (RubyException) parser.getNokogiriHandler().getLastError();
         }
     }

data/ext/java/nokogiri/XmlSyntaxError.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]
@@ -35,6 +35,7 @@ package nokogiri;
 import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
 import static nokogiri.internals.NokogiriHelpers.stringOrNil;
 
+import org.jruby.CompatVersion;
 import org.jruby.Ruby;
 import org.jruby.RubyClass;
 import org.jruby.RubyException;
@@ -108,12 +109,23 @@ public class XmlSyntaxError extends RubyException {
         return new XmlSyntaxError(runtime, klazz, e);
     }
 
-    @Override
-    @JRubyMethod(name = "to_s")
+    //@Override
+    //"to_s" method was branched in 1.8 and 1.9 since JRuby 1.6.6
+    // to support older version of JRuby, the annotation is commented out
+    @JRubyMethod(name = "to_s", compat = CompatVersion.RUBY1_8)
     public IRubyObject to_s(ThreadContext context) {
         if (exception != null && exception.getMessage() != null)
             return context.getRuntime().newString(exception.getMessage());
         else
             return super.to_s(context);  
     }
+    
+    //@Override
+    //"to_s" method was branched in 1.8 and 1.9 since JRuby 1.6.6
+    // to support older version of JRuby, the annotation is commented out
+    @JRubyMethod(name = "to_s", compat = CompatVersion.RUBY1_9)
+    public IRubyObject to_s19(ThreadContext context) {
+        return this.to_s(context);
+    }
+    
 }

data/ext/java/nokogiri/XmlXpathContext.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]
@@ -53,7 +53,6 @@ import org.jruby.RubyBoolean;
 import org.jruby.RubyClass;
 import org.jruby.RubyException;
 import org.jruby.RubyFloat;
-import org.jruby.RubyNumeric;
 import org.jruby.RubyObject;
 import org.jruby.RubyString;
 import org.jruby.anno.JRubyClass;
@@ -72,7 +71,7 @@ import org.w3c.dom.NodeList;
 @JRubyClass(name="Nokogiri::XML::XPathContext")
 public class XmlXpathContext extends RubyObject {
     private XmlNode context;
-    private static final XPath xpath = XPathFactory.newInstance().newXPath();;
+    private XPath xpath;
     
     public XmlXpathContext(Ruby ruby, RubyClass rubyClass) {
         super(ruby, rubyClass);
@@ -98,6 +97,7 @@ public class XmlXpathContext extends RubyObject {
     public static IRubyObject rbNew(ThreadContext context, IRubyObject klazz, IRubyObject node) {
         XmlNode xmlNode = (XmlNode)node;
         XmlXpathContext xmlXpathContext = (XmlXpathContext) NokogiriService.XML_XPATHCONTEXT_ALLOCATOR.allocate(context.getRuntime(), (RubyClass)klazz);
+        xmlXpathContext.xpath = XPathFactory.newInstance().newXPath();
         xmlXpathContext.setNode(xmlNode);
         return xmlXpathContext;
     }