nokogiri 1.5.0.beta.4-java → 1.5.0-java

data/ext/java/nokogiri/XmlDocumentFragment.java

@@ -45,8 +45,6 @@ import java.util.Set;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
-import nokogiri.internals.SaveContext;
-
 import org.jruby.Ruby;
 import org.jruby.RubyArray;
 import org.jruby.RubyClass;
@@ -57,6 +55,7 @@ import org.jruby.javasupport.util.RuntimeHelpers;
 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.NamedNodeMap;
 
 /**
@@ -92,6 +91,7 @@ public class XmlDocumentFragment extends XmlNode {
         
         // make wellformed fragment, ignore invalid namespace, or add appropriate namespace to parse
         if (args.length > 1 && args[1] instanceof RubyString) {
+            args[1] = trim(context, doc, (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])));
         }
@@ -107,6 +107,18 @@ public class XmlDocumentFragment extends XmlNode {
         RuntimeHelpers.invoke(context, fragment, "initialize", args);
         return fragment;
     }
+    
+    private static IRubyObject trim(ThreadContext context, XmlDocument xmlDocument, RubyString str) {
+        // checks whether schema is given. if exists, allows whitespace processing to a parser
+        Document document = (Document)xmlDocument.node;
+        if (document.getDoctype() != null) return str;
+        // strips trailing \n off forcefully
+        // not to return new object in case of no chomp needed, chomp! is used here.
+        IRubyObject result;
+        if (context.getRuntime().is1_9()) result = str.chomp_bang19(context);
+        else result = str.chomp_bang(context);
+        return result.isNil() ? str : result;
+    }
 
     private static Pattern qname_pattern = Pattern.compile("[^</:>\\s]+:[^</:>=\\s]+");
     private static Pattern starttag_pattern = Pattern.compile("<[^</>]+>");
@@ -208,10 +220,4 @@ public class XmlDocumentFragment extends XmlNode {
     public void relink_namespace(ThreadContext context) {
         ((XmlNodeSet) children(context)).relink_namespace(context);
     }
-
-    @Override
-    public void saveContent(ThreadContext context, SaveContext ctx) {
-        saveNodeListContent(context, (XmlNodeSet) children(context), ctx);
-    }
-
 }

data/ext/java/nokogiri/XmlDtd.java

@@ -37,7 +37,7 @@ import static nokogiri.internals.NokogiriHelpers.nonEmptyStringOrNil;
 import static nokogiri.internals.NokogiriHelpers.stringOrNil;
 import static org.jruby.javasupport.util.RuntimeHelpers.invoke;
 import nokogiri.internals.NokogiriHelpers;
-import nokogiri.internals.SaveContext;
+import nokogiri.internals.SaveContextVisitor;
 
 import org.apache.xerces.xni.QName;
 import org.cyberneko.dtd.DTDConfiguration;
@@ -98,43 +98,40 @@ public class XmlDtd extends XmlNode {
     public XmlDtd(Ruby ruby, RubyClass rubyClass) {
         super(ruby, rubyClass);
     }
+    
+    public void setNode(Ruby runtime, Node dtd) {
+        this.node = dtd;
+        notationClass = (RubyClass) runtime.getClassFromPath("Nokogiri::XML::Notation");
 
-    public XmlDtd(Ruby ruby) {
-        this(ruby, getNokogiriClass(ruby, "Nokogiri::XML::DTD"), null);
-    }
-
-    public XmlDtd(Ruby ruby, Node dtd) {
-        this(ruby, getNokogiriClass(ruby, "Nokogiri::XML::DTD"), dtd);
-    }
-
-    public XmlDtd(Ruby ruby, RubyClass rubyClass, Node dtd) {
-        super(ruby, rubyClass, dtd);
-        notationClass = (RubyClass)
-            ruby.getClassFromPath("Nokogiri::XML::Notation");
-
-        name = pubId = sysId = ruby.getNil();
+        name = pubId = sysId = runtime.getNil();
         if (dtd == null) return;
 
         // This is the dtd declaration stored in the document; it
         // contains the DTD name (root element) and public and system
-        // ids.  The actual declarations are in the NekoDTD 'dtd'
+        // ids. The actual declarations are in the NekoDTD 'dtd'
         // variable. I don't know of a way to consolidate the two.
 
         DocumentType otherDtd = dtd.getOwnerDocument().getDoctype();
         if (otherDtd != null) {
-            name = stringOrNil(ruby, otherDtd.getNodeName());
-            pubId = nonEmptyStringOrNil(ruby, otherDtd.getPublicId());
-            sysId = nonEmptyStringOrNil(ruby, otherDtd.getSystemId());
+            name = stringOrNil(runtime, otherDtd.getNodeName());
+            pubId = nonEmptyStringOrNil(runtime, otherDtd.getPublicId());
+            sysId = nonEmptyStringOrNil(runtime, otherDtd.getSystemId());
         }
     }
 
-    public static XmlDtd newEmpty(Ruby ruby,
+    public XmlDtd(Ruby ruby, RubyClass rubyClass, Node dtd) {
+        super(ruby, rubyClass, dtd);
+        setNode(ruby, dtd);
+    }
+
+    public static XmlDtd newEmpty(Ruby runtime,
                                   Document doc,
                                   IRubyObject name,
                                   IRubyObject external_id,
                                   IRubyObject system_id) {
         Element placeHolder = doc.createElement("dtd_placeholder");
-        XmlDtd dtd = new XmlDtd(ruby, placeHolder);
+        XmlDtd dtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
+        dtd.setNode(runtime, placeHolder);
         dtd.name = name;
         dtd.pubId = external_id;
         dtd.sysId = system_id;
@@ -155,38 +152,47 @@ public class XmlDtd extends XmlNode {
      * Document provided by NekoDTD.
      *
      */
-    public static XmlDtd newFromInternalSubset(Ruby ruby, Document doc) {
+    public static XmlDtd newFromInternalSubset(Ruby runtime, Document doc) {
         Object dtdTree_ = doc.getUserData(XmlDocument.DTD_RAW_DOCUMENT);
-        if (dtdTree_ == null)
-            return new XmlDtd(ruby);
+        if (dtdTree_ == null) {
+            XmlDtd xmlDtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
+            xmlDtd.setNode(runtime, null);
+            return xmlDtd;
+        }
 
         Node dtdTree = (Node) dtdTree_;
         Node dtd = getInternalSubset(dtdTree);
         if (dtd == null) {
-            return new XmlDtd(ruby);
+            XmlDtd xmlDtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
+            xmlDtd.setNode(runtime, null);
+            return xmlDtd;
         } else {
             // Import the node into doc so it has the correct owner document.
             dtd = doc.importNode(dtd, true);
-            return new XmlDtd(ruby, dtd);
+            XmlDtd xmlDtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
+            xmlDtd.setNode(runtime, dtd);
+            return xmlDtd;
         }
     }
 
-    public static IRubyObject newFromExternalSubset(Ruby ruby, Document doc) {
+    public static IRubyObject newFromExternalSubset(Ruby runtime, Document doc) {
         Object dtdTree_ = doc.getUserData(XmlDocument.DTD_RAW_DOCUMENT);
         if (dtdTree_ == null) {
-            return ruby.getNil();
+            return runtime.getNil();
         }
 
         Node dtdTree = (Node) dtdTree_;
         Node dtd = getExternalSubset(dtdTree);
         if (dtd == null) {
-            return ruby.getNil();
+            return runtime.getNil();
         } else if (!dtd.hasChildNodes()) {
-            return ruby.getNil();
+            return runtime.getNil();
         } else {
             // Import the node into doc so it has the correct owner document.
             dtd = doc.importNode(dtd, true);
-            return new XmlDtd(ruby, dtd);
+            XmlDtd xmlDtd = (XmlDtd) NokogiriService.XML_DTD_ALLOCATOR.allocate(runtime, getNokogiriClass(runtime, "Nokogiri::XML::DTD"));
+            xmlDtd.setNode(runtime, dtd);
+            return xmlDtd;
         }
     }
 
@@ -365,10 +371,7 @@ public class XmlDtd extends XmlNode {
         extractDecls(context, node.getFirstChild());
 
         // convert allDecls to a NodeSet
-        children =
-            new XmlNodeSet(runtime,
-                           getNokogiriClass(runtime, "Nokogiri::XML::NodeSet"),
-                           allDecls);
+        children = XmlNodeSet.newXmlNodeSet(context, allDecls);
 
         // add attribute decls as attributes to the matching element decl
         RubyArray keys = attributes.keys();
@@ -455,13 +458,12 @@ public class XmlDtd extends XmlNode {
         }
     }
     
-    public void saveContent(ThreadContext context, SaveContext ctx) {
-        ctx.append("<!DOCTYPE " + name + " ");
-        if (pubId != null) {
-            ctx.append("PUBLIC \"" + pubId + "\" \"" + sysId + "\">");
-        } else if (sysId != null) {
-            ctx.append("SYSTEM " + sysId);
-        }
+    @Override
+    public void accept(ThreadContext context, SaveContextVisitor visitor) {
+        // since we use nekoDTD to parse dtd, node might be ElementImpl type
+        // An external subset doesn't need to show up, so this method just see docType.
+        DocumentType docType = node.getOwnerDocument().getDoctype();
+        visitor.enter(docType);
+        visitor.leave(docType);
     }
-
 }

data/ext/java/nokogiri/XmlElement.java

@@ -32,9 +32,8 @@
 
 package nokogiri;
 
-import static nokogiri.internals.NokogiriHelpers.namedNodeMapToRubyArray;
 import static nokogiri.internals.NokogiriHelpers.rubyStringToString;
-import nokogiri.internals.SaveContext;
+import nokogiri.internals.SaveContextVisitor;
 
 import org.jruby.Ruby;
 import org.jruby.RubyArray;
@@ -94,8 +93,7 @@ public class XmlElement extends XmlNode {
             prefix.isNil() ? "xmlns" : "xmlns:" + rubyStringToString(prefix);
         element.setAttributeNS(uri, qName, rubyStringToString(href));
 
-        XmlNamespace ns = (XmlNamespace)
-            super.add_namespace_definition(context, prefix, href);
+        XmlNamespace ns = (XmlNamespace) super.add_namespace_definition(context, prefix, href);
         updateNodeNamespaceIfNecessary(context, ns);
 
         return ns;
@@ -174,49 +172,24 @@ public class XmlElement extends XmlNode {
             ((XmlNodeSet) children(context)).relink_namespace(context);
         }
     }
-
-    /**
-     * TODO: previous code handled elements with parent 'p' differently?.
-     */
+    
     @Override
-    public void saveContent(ThreadContext context, SaveContext ctx) {
-        Node firstChild = node.getFirstChild();
-        boolean empty = (firstChild == null);
-        short type = -1;
-        if (!empty) type = firstChild.getNodeType();
-        boolean inline = (!empty &&
-                          (type == Node.TEXT_NODE ||
-                           type == Node.CDATA_SECTION_NODE ||
-                           type == Node.ENTITY_REFERENCE_NODE));
-
-        if (empty) {
-            ctx.emptyTagStart(node.getNodeName());
-        } else if (inline) {
-            ctx.openTagInlineStart(node.getNodeName());
-        } else {
-            ctx.openTagStart(node.getNodeName());
-        }
-        
-        RubyArray attr_list = namedNodeMapToRubyArray(context.getRuntime(), node.getAttributes());
-        saveNodeListContent(context, attr_list, ctx);
-
-        if (empty) {
-            ctx.emptyTagEnd(node.getNodeName());
-            return;
-        } else if (inline) {
-            ctx.openTagInlineEnd();
-        } else {
-            ctx.openTagEnd();
-        }
-
-        saveNodeListContent(context, (XmlNodeSet) children(context), ctx);
-
-        if (inline) {
-            ctx.closeTagInline(node.getNodeName());
-        } else {
-            ctx.closeTag(node.getNodeName());
+    public void accept(ThreadContext context, SaveContextVisitor visitor) {
+        visitor.enter((Element)node);
+        XmlNodeSet xmlNodeSet = (XmlNodeSet) children(context);
+        if (xmlNodeSet.length() > 0) {
+            RubyArray array = (RubyArray) xmlNodeSet.to_a(context);
+            for(int i = 0; i < array.getLength(); i++) {
+                Object item = array.get(i);
+                if (item instanceof XmlNode) {
+                  XmlNode cur = (XmlNode) item;
+                  cur.accept(context, visitor);
+                } else if (item instanceof XmlNamespace) {
+                    XmlNamespace cur = (XmlNamespace)item;
+                    cur.accept(context, visitor);
+                }
+            }
         }
-
+        visitor.leave((Element)node);
     }
-
 }

data/ext/java/nokogiri/XmlElementDecl.java

@@ -54,12 +54,16 @@ import org.w3c.dom.Node;
 @JRubyClass(name="Nokogiri::XML::ElementDecl", parent="Nokogiri::XML::Node")
 public class XmlElementDecl extends XmlNode {
     RubyArray attrDecls;
-
     IRubyObject contentModel;
-
-    public XmlElementDecl(Ruby ruby, RubyClass klass) {
-        super(ruby, klass);
-        throw ruby.newRuntimeError("node required");
+    
+    public XmlElementDecl(Ruby ruby, RubyClass klazz) {
+        super(ruby, klazz);
+    }
+    
+    public void setNode(ThreadContext context, Node node) {
+        super.setNode(context, node);
+        attrDecls = RubyArray.newArray(context.getRuntime());
+        contentModel = context.getRuntime().getNil();
     }
 
     /**

data/ext/java/nokogiri/XmlEntityReference.java

@@ -32,14 +32,18 @@
 
 package nokogiri;
 
+import static nokogiri.internals.NokogiriHelpers.getCachedNodeOrCreate;
 import static nokogiri.internals.NokogiriHelpers.rubyStringToString;
 
+import nokogiri.internals.SaveContextVisitor;
+
 import org.jruby.Ruby;
 import org.jruby.RubyClass;
 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;
 
 /**
@@ -47,9 +51,10 @@ import org.w3c.dom.Node;
  * 
  * @author sergio
  * @author Patrick Mahoney <pat@polycrystal.org>
+ * @author Yoko Harada <yokolet@gmail.com>
  */
 @JRubyClass(name="Nokogiri::XML::EntityReference", parent="Nokogiri::XML::Node")
-public class XmlEntityReference extends XmlNode{
+public class XmlEntityReference extends XmlNode {
 
     public XmlEntityReference(Ruby ruby, RubyClass klazz) {
         super(ruby, klazz);
@@ -71,5 +76,22 @@ public class XmlEntityReference extends XmlNode{
         Node node = document.createEntityReference(rubyStringToString(name));
         setNode(context, node);
     }
-
+    
+    @Override
+    public void accept(ThreadContext context, SaveContextVisitor visitor) {
+        visitor.enter((EntityReference)node);
+        Node child = node.getFirstChild();
+        while (child != null) {
+            IRubyObject nokoNode = getCachedNodeOrCreate(context.getRuntime(), child);
+            if (nokoNode instanceof XmlNode) {
+                XmlNode cur = (XmlNode) nokoNode;
+                cur.accept(context, visitor);
+            } else if (nokoNode instanceof XmlNamespace) {
+                XmlNamespace cur = (XmlNamespace) nokoNode;
+                cur.accept(context, visitor);
+            }
+            child = child.getNextSibling();
+        }
+        visitor.leave((EntityReference)node);
+    }
 }

data/ext/java/nokogiri/XmlNamespace.java

@@ -32,9 +32,11 @@
 
 package nokogiri;
 
+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 nokogiri.internals.SaveContext;
+import nokogiri.internals.SaveContextVisitor;
 
 import org.jruby.Ruby;
 import org.jruby.RubyClass;
@@ -44,6 +46,8 @@ import org.jruby.anno.JRubyClass;
 import org.jruby.anno.JRubyMethod;
 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.Node;
 
 /**
@@ -54,37 +58,94 @@ import org.w3c.dom.Node;
  */
 @JRubyClass(name="Nokogiri::XML::Namespace")
 public class XmlNamespace extends RubyObject {
-
+    private Attr attr;
     private IRubyObject prefix;
     private IRubyObject href;
+    private String prefixString;
+    private String hrefString;
 
     public XmlNamespace(Ruby ruby, RubyClass klazz) {
         super(ruby, klazz);
     }
-
-    public XmlNamespace(Ruby ruby, String prefix, String href) {
-        this(ruby, getNokogiriClass(ruby, "Nokogiri::XML::Namespace"), prefix, href);
+    
+    public Node getNode() {
+        return attr;
     }
-
-    public XmlNamespace(Ruby ruby, RubyClass klazz, String prefix, String href) {
-        super(ruby, klazz);
-        this.prefix = (prefix == null) ? ruby.getNil() : RubyString.newString(ruby, prefix);
-        this.href = (href == null) ? ruby.getNil() : RubyString.newString(ruby, href);
+    
+    public String getPrefix() {
+        return prefixString;
     }
-
-    public XmlNamespace(Ruby ruby, IRubyObject prefix, IRubyObject href) {
-        this(ruby, getNokogiriClass(ruby, "Nokogiri::XML::Namespace"), prefix, href);
+    
+    public String getHref() {
+        return hrefString;
     }
 
-    public XmlNamespace(Ruby ruby, RubyClass klazz, IRubyObject prefix, IRubyObject href) {
-        super(ruby, klazz);
+    public void init(Attr attr, IRubyObject prefix, IRubyObject href, IRubyObject xmlDocument) {
+        init(attr, prefix, href, (String) prefix.toJava(String.class), (String) href.toJava(String.class), xmlDocument);
+    }
+    
+    public void init(Attr attr, IRubyObject prefix, IRubyObject href, String prefixString, String hrefString, IRubyObject xmlDocument) {
+        this.attr = attr;
         this.prefix = prefix;
         this.href = href;
+        this.prefixString = prefixString;
+        this.hrefString = hrefString;
+        this.attr.setUserData(CACHED_NODE, this, null);
+        setInstanceVariable("@document", xmlDocument);
     }
     
-    public void setDefinition(Ruby runtime, String prefix, String href) {
-        this.prefix = (prefix == null) ? runtime.getNil() : RubyString.newString(runtime, prefix);
-        this.href = (href == null) ? runtime.getNil() : RubyString.newString(runtime, href);
+    public static XmlNamespace createFromAttr(Ruby runtime, Attr attr) {
+        String prefixValue = getLocalNameForNamespace(attr.getName());
+        IRubyObject prefix_value;
+        if (prefixValue == null) {
+            prefix_value = runtime.getNil();
+            prefixValue = "";
+        } else {
+            prefix_value = RubyString.newString(runtime, prefixValue);
+        }
+        String hrefValue = attr.getValue();
+        IRubyObject href_value = RubyString.newString(runtime, hrefValue);
+        // check namespace cache
+        XmlDocument xmlDocument = (XmlDocument)getCachedNodeOrCreate(runtime, attr.getOwnerDocument());
+        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")); 
+        namespace.init(attr, prefix_value, href_value, prefixValue, hrefValue, xmlDocument);
+        
+        // updateing namespace cache
+        xmlDocument.getNamespaceCache().put(namespace, attr.getOwnerElement());
+        return namespace;
+    }
+    
+    public static XmlNamespace createFromPrefixAndHref(Node owner, IRubyObject prefix, IRubyObject href) {
+        String prefixValue = prefix.isNil() ? "" : (String) prefix.toJava(String.class);
+        String hrefValue = (String) href.toJava(String.class);
+        Ruby runtime = prefix.getRuntime();
+        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"));
+        String attrName = "xmlns";
+        if (!"".equals(prefixValue)) {
+            attrName = attrName + ":" + prefixValue;
+        }
+        Attr attrNode = document.createAttribute(attrName);
+        attrNode.setNodeValue(hrefValue);
+
+        // initialize XmlNamespace object
+        namespace.init(attrNode, prefix, href, prefixValue, hrefValue, xmlDocument);
+        
+        // updating namespace cache
+        xmlDocument.getNamespaceCache().put(namespace, owner);
+        return namespace;
     }
     
     /**
@@ -97,32 +158,26 @@ public class XmlNamespace extends RubyObject {
         return super.clone();
     }
 
-    public static XmlNamespace fromNode(Ruby ruby, Node node) {
-        String localName = getLocalNameForNamespace(node.getNodeName());
-        XmlNamespace namespace = (XmlNamespace) NokogiriService.XML_NAMESPACE_ALLOCATOR.allocate(ruby, getNokogiriClass(ruby, "Nokogiri::XML::Namespace"));
-        namespace.setDefinition(ruby, localName, node.getNodeValue());
-        return namespace;
-    }
-
     public boolean isEmpty() {
-        return this.prefix.isNil() && this.href.isNil();
-    }
-
-    public void setDocument(IRubyObject doc) {
-        this.setInstanceVariable("@document", doc);
+        return prefix.isNil() && href.isNil();
     }
 
     @JRubyMethod
     public IRubyObject href(ThreadContext context) {
-        return this.href;
+        return href;
     }
 
     @JRubyMethod
     public IRubyObject prefix(ThreadContext context) {
-        return this.prefix;
+        return prefix;
     }
     
-    public void saveContent(ThreadContext context, SaveContext ctx) {
-        ctx.append(" " + prefix + "=\"" + href + "\"");
+    public void accept(ThreadContext context, SaveContextVisitor visitor) {
+        String string = " " + prefix + "=\"" + href + "\"";
+        visitor.enter(string);
+        visitor.leave(string);
+        // is below better?
+        //visitor.enter(attr);
+        //visitor.leave(attr);
     }
 }