Patch to fix TagLib crash-inducing bug

[Umesh Shankar]

At the bottom of this message is a patch to TagLib to handle a bug when 
it converts old 3-character ID3v2 frame types to 4-character frame 
types. Basically, early in the function you had

frameID = header->FrameID();

then

updateFrame(header);

which *changes* the value of header->FrameID().

At this point, frameID holds an old, incorrect value. The result is that 
the true frameID is "APIC" or "COMM" or whatever, but the class returned 
is UnknownFrame. Naturally this causes a problem with applications that 
assume the type of the Frame class from the frame ID.  I added a line of 
code which re-assigns frameID = header->FrameID() after the update, 
which causes the right class to be instantiated.

This was discovered as an amaroK crash by Mark Kretschmann, who also 
supplied the test case. (Mark: the comment field seems to contain bogus 
data as well.)

I can't vouch for this being a total solution, since I haven't tested 
other similar cases, but it seemed to work for me in that amaroK did not 
crash.

Umesh

===================================================================
RCS file: /home/kde/kdesupport/taglib/mpeg/id3v2/id3v2framefactory.cpp,v
retrieving revision 1.30
diff -u -3 -p -r1.30 id3v2framefactory.cpp
--- mpeg/id3v2/id3v2framefactory.cpp    20 Oct 2004 23:34:55 -0000      1.30
+++ mpeg/id3v2/id3v2framefactory.cpp    16 May 2005 23:29:13 -0000
@@ -103,6 +103,8 @@ Frame *FrameFactory::createFrame(const B
      return 0;
    }

+  frameID = header->frameID();
+
    // This is where things get necissarily nasty.  Here we determine which
    // Frame subclass (or if none is found simply an Frame) based
    // on the frame ID.  Since there are a lot of possibilities, that means
us