Re: XMLResource and OutputStream

[Vadim Gritsenko]

JC Tchitchiama wrote:
On Saturday 20 Dec 2003 6:20 pm, Vadim Gritsenko wrote:
 

JC Tchitchiama wrote:
   

.....
1) somebody has to lobby for the XMLResource interface to be extended with
a method OutputStream setContentAsOutputStream() throws XMLDBException
2) I have attached a patch file [for sake of discussion]. I know that the
implementation can be polished off with better OutputStream classes or
even a PipedInputStream/OutputStream combinaison or even some other
suggestions you may have. I have also left out the  IOException handling.
     

Hi,
I see couple of issues with the patch:
1. As far as I can see it will work only for small documents, up to 8Kb.
Once document is large than 8Kb, buffered out stream will write to pipe
stream more than once and it will override value of the this.content
several times, loosing data.
   

I am aware of the limitations you mentioned but I had to get the ball rolling.
 

Ok, fair enough. Let's roll it ;-)

One possible solution is use PipedOutputStream smartly and read the buffer out 
progressively. The trouble there is that if a failure appends then you have a 
partial document. 
Second possible solution  could to rely on the document size. If we somehow 
get the document size first before creating the BufferedOutputStream the we 
should be OK.

Any other thoughts ?
 

Sure. Every OutputStream has a close() method, and I think we should 
rely on this, as this is a standard contract in the java.io package. So, 
this patch can be changed to use new BufferedOutputStream(new 
ByteArrayOutputStream()), and set the resource's content when close() is 
called.

Thus, you will not have issues with the buffer size / document size, and 
buffer size can be reduced to 1024 or so.


2. Why trim is necessary, can you provide with more details - what
unwanted characters, what parsers?
   

I am xerces. When using KHexEdit I could see some charaters appended to the 
document that's why I trim the string.
 

Can you give specific example - what is added, what is code snippet 
which writes with these extra characters? I really don't like this 
trim() and I think that there is a bug elsewhere.

Vadim

Index: XMLResourceImpl.java
===================================================================
RCS file:
/home/cvspublic/xml-xindice/java/src/org/apache/xindice/client/xmldb/reso
urces/XMLResourceImpl.java,v retrieving revision 1.16
diff -u -r1.16 XMLResourceImpl.java
--- XMLResourceImpl.java        9 Aug 2003 05:01:55 -0000       1.16
+++ XMLResourceImpl.java        12 Nov 2003 11:31:09 -0000
@@ -78,6 +78,11 @@
import javax.xml.parsers.SAXParser;
import javax.xml.parsers.SAXParserFactory;
import java.io.StringReader;
+import java.io.OutputStream;
+import java.io.IOException;
+import java.io.PipedOutputStream;
+import java.io.BufferedOutputStream;
+
/**
* XMLResourceImpl provides an implementation to handle XML resources
@@ -100,6 +105,14 @@
   protected String content = null;
   /**
+     * buffer size generously big
+     *
+     */
+    public static final int BUFFER_SIZE = 8*1024;
+
+
+
+    /**
    * This is a SAX feature that controls how namespaces are reported in
SAX. * By default this feature is <em>on</em>.
    *
@@ -313,6 +326,40 @@
    */
   public ContentHandler setContentAsSAX() throws XMLDBException {
       return new SetContentHandler(this);
+    }
+
+    /**
+     * setContentAsOutputStream returns a OutputStream that can be used
to set the +     * content of the resource.
+     *
+     * @return The outputStream that is used to insert data into the
database. +     * @exception XMLDBException
+     */
+    public OutputStream setContentAsOutputStream() throws XMLDBException
{ +
+	PipedOutputStream pipedOut = new PipedOutputStream()
+	{
+		public void write(byte b[], int off, int len) throws IOException
+		{
+			try
+			{
+				/** WARNING it's important to trim the string
+			 	* because when XMLOutputter, DOMOutputter, XMLSerializer
+			 	* is used to populate the outputstream; they append
+			 	* charaters unwanted by most XML parsers.
+			 	*/
+				content = new String(b).trim();
+				bytes = null;
+			} catch (Exception e) {
+				//TODO log an exception here
+		    		//throw FaultCodes.createXMLDBException(e);
+			}
+
+		}
+
+	};
+
+	return new BufferedOutputStream(pipedOut, BUFFER_SIZE);
   }