Over 18 months ago there was a thread on this list about the proper
handling of accented characters in the Soundex encoder in commons-codec
but it never seemed to get resolved. In addition, there are still
failing unit tests that reference this issue in the current version of
the code. As someone who uses this code, I'd like to see all unit tests
passing, so I've done some investigation.
As a refresher, there were three options discussed for the behavior of
the Soundex codec when it sees an accented character:
1) Throw an IllegalArgumentException
2) Drop it silently
3) Replace it with the equivalent unaccented character
Right now the code drops it silently, but the unit tests are expecting
an IllegalArgumentException. The code in Soundex.map(char ch) seems to
be trying to throw this exception, but it will never happen because the
characters passed to it from Soundex.soundex are from a String that has
gone through SoundexUtils.clean(String str) which removes all characters
that fail a Character.isCharacter(char ch) check (accented chars fail
this check, I, erm, checked). This means if we want to throw an
IllegalArgumentException it must be done in SoundexUtils.clean, not
Soundex.map.
I think either behaviors 1 or 2 (drop silently, which is what we
currently do) would be easy to implement and then change the unit tests
to match the behavior so all unit tests on commons-codec pass.
If someone lets me know which behavior is desired I will submit a patch.
Note that behavior 2 only requires either removing the test cases or
changing them to expect the same encoding as an empty string.
References:
http://issues.apache.org/bugzilla/show_bug.cgi?id=29080
http://www.mail-archive.com/commons-dev@jakarta.apache.org/msg41974.html
Best,
Chris
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
