Levenshtein as a MySQL stored function
Edit distance in SQL works; O(n·m) will punish unfiltered scans.
Levenshtein counts edits between strings. As a stored function it is fine for small dictionaries and admin tools.
SELECT id, name, levenshtein(name, 'Open Query') AS dist
FROM organisations
HAVING dist <= 3
ORDER BY dist;Without a prefilter (LIKE, phonetic keys, blocking), this is a CPU-heavy full scan.
At scale
- External search
- Trigram indexes where available
- Normalise + block before expensive distance
Prototype in SQL. Do not put unfiltered Levenshtein on a public autocomplete hot path.