Java's built-in read/writeUTF() methods in ObjectInput/OutputStream use 'Modified UTF-8' for its encoding, which is cleverly designed to be almost exactly like UTF-8, until it's not and it bites you in the behind. Two new methods read/writeUnmodifiedUTF() may be used when real UTF-8 is required. At the current time Narya provides no further wiring-in of this functionality. See Whirled for a brilliant hack that does make use of it.

git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@6040 542714f4-19e9-0310-aa3c-eee0fc999fb1

src/java/com/threerings/io/ObjectInputStream.java

@@ -24,9 +24,11 @@ package com.threerings.io;
 import java.util.ArrayList;
 import java.util.HashMap;
 
+import java.io.ByteArrayInputStream;
 import java.io.DataInputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.io.InputStreamReader;
 
 import com.google.common.collect.Lists;
 import com.google.common.collect.Maps;
@@ -305,6 +307,27 @@ public class ObjectInputStream extends DataInputStream
         _streamer.readObject(_current, this, false);
     }
 
+    /**
+     * Read a string encoded as real UTF-8 (rather than the modified format handled by
+     * {link #readUTF}).
+     */
+    public String readUnmodifiedUTF ()
+        throws IOException
+    {
+        // find out how many raw bytes of UTF8 data there is
+        int utflen = readUnsignedShort();
+        // read precisely that many into a buffer
+        byte[] bbuf = new byte[utflen];
+        in.read(bbuf);
+    
+        // decode the UTF-8 stream into a character buffer
+        char[] cbuf = new char[utflen];
+        int read = new InputStreamReader(new ByteArrayInputStream(bbuf), "UTF-8").read(cbuf);
+    
+        // create and return the string given the number of decoded characters
+        return String.copyValueOf(cbuf, 0, read);
+    }
+
     @Override
     public String toString ()
     {

src/java/com/threerings/io/ObjectOutputStream.java

@@ -23,9 +23,11 @@ package com.threerings.io;
 
 import java.util.Map;
 
+import java.io.ByteArrayOutputStream;
 import java.io.DataOutputStream;
 import java.io.IOException;
 import java.io.OutputStream;
+import java.io.OutputStreamWriter;
 
 import com.google.common.collect.Maps;
 
@@ -289,6 +291,29 @@ public class ObjectOutputStream extends DataOutputStream
         _streamer.writeObject(_current, this, false);
     }
 
+    /**
+     * Write a string encoded as real UTF-8 (rather than the modified format handled by
+     * {link #writeUTF}).
+     */
+    public void writeUnmodifiedUTF (String str)
+        throws IOException
+    {
+        // prepare a buffer to accept the encoded UTF-8
+        ByteArrayOutputStream baos = new ByteArrayOutputStream();
+        OutputStreamWriter osw = new OutputStreamWriter(baos, "UTF-8");
+
+        // do the deed
+        osw.write(str);
+        osw.flush();
+
+        // now that we know how many bytes we'll need, write that number to the stream
+        writeShort(baos.size());
+
+        // then finally the bytes themselves
+        write(baos.toByteArray());
+    }
+
+
     /** Used to map classes to numeric codes and the {@link Streamer} instance used to write
      * them. */
     protected Map<Class<?>, ClassMapping> _classmap;