[Snowball-discuss] Re: Possible memory leak in Snowballs Java stemmer

[Richard Boulton]

This is a multi-part message in MIME format.
--------------000000050708030209050502
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

I have applied the attached patch to fix this problem.  The java sources 
distributed on the snowball website now contain the updated code.

-- 
Richard

--------------000000050708030209050502
Content-Type: text/plain;
 name="diffs"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="diffs"

Index: SnowballProgram.java
===================================================================
RCS file: /home/cvs/snowball/website/net/sf/snowball/SnowballProgram.java,v
retrieving revision 1.2
diff -u -r1.2 SnowballProgram.java
--- SnowballProgram.java	28 Jan 2002 19:35:51 -0000	1.2
+++ SnowballProgram.java	7 Jun 2004 09:12:31 -0000
@@ -27,7 +27,15 @@
      */
     public String getCurrent()
     {
-	return current.toString();
+        String result = current.toString();
+        // Make a new StringBuffer.  If we reuse the old one, and a user of
+        // the library keeps a reference to the buffer returned (for example,
+        // by converting it to a String in a way which doesn't force a copy),
+        // the buffer size will not decrease, and we will risk wasting a large
+        // amount of memory.
+        // Thanks to Wolfram Esser for spotting this problem.
+        current = new StringBuffer();
+        return result;
     }
 
     // current string

--------------000000050708030209050502--