bug: fuzzystrmatch levenshtein is wrong

Marcin Mańk <marcin.mank@gmail.com>

From: marcin mank <marcin.mank@gmail.com>
To: PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2009-12-07T13:33:36Z
Lists: pgsql-hackers

Attachments

The current behavior of levenshtein(text,text,int,int,int) is wrong. Consider:

leki_dev=# select levenshtein('','a',2,4,5);
 levenshtein
-------------
           1
(1 row)


leki_dev=# select levenshtein('a','',2,4,5);
 levenshtein
-------------
           1
(1 row)


leki_dev=# select levenshtein('aa','a',2,4,5);
 levenshtein
-------------
           1
(1 row)


leki_dev=# select levenshtein('a','aa',2,4,5);
 levenshtein
-------------
           1
(1 row)

versus (after patch)

postgres=# select levenshtein('','a',2,4,5);
 levenshtein
-------------
           2
(1 row)

postgres=# select levenshtein('a','',2,4,5);
 levenshtein
-------------
           4
(1 row)

postgres=# select levenshtein('aa','a',2,4,5);
 levenshtein
-------------
           4
(1 row)

postgres=# select levenshtein('a','aa',2,4,5);
 levenshtein
-------------
           2
(1 row)

patch attached.

Greetings
Marcin Mańk