← All modules

How many single-character edits turn one string into another — plain Levenshtein, and the Damerau variant that counts a swapped pair as one edit — with a cut-off for when you only care whether it is close.

Version

0.3.0 github:ugexe

Depends

nothing outside the core

License

Artistic-2.0

Its own test suite

3 files, green

Checked

2026-09-14 against Raku++ 3.28.0 and Rakudo 2026.08

Where it lives

raku.land · source

Install it #

$ rakupp install Text::Levenshtein::Damerau

zef install Text::Levenshtein::Damerau writes the same store; either installer leaves the module usable by both engines.

What it is for #

"Did you mean commit?" needs a number for how far comit is from it, and edit distance is that number: the count of insertions, deletions and substitutions between two strings. Damerau's variant adds a fourth edit, the transposition of two adjacent characters, which is the typo people actually make — teh is one edit from the, not two. This distribution is both, as two subs, and it is what zef uses to suggest the module you probably meant.

Two distances and a cut-off #

use Text::Levenshtein::Damerau;

say ld('kitten', 'sitting'), ' ', dld('kitten', 'sitting');
say ld('abcd', 'acbd'), ' ', dld('abcd', 'acbd');
say ld('', 'abc'), ' ', dld('same', 'same');
say dld('rakudo', 'raku', 1).defined;

3 3
2 1
3 0
False

ld is Levenshtein and dld is Damerau-Levenshtein; they agree until a transposition is the cheapest edit, as on the second line, where swapping bc costs one edit under Damerau and two — a deletion and an insertion — under Levenshtein. Both take an optional third argument, a maximum.

The one thing to know #

The maximum does not clamp, it gives up. dld('rakudo', 'raku', 1) is not 2 and not 1: it is an undefined Int, because the distance passed the limit and the sub stopped counting. That is the efficient answer for a "close enough?" test over a long list of candidates — most comparisons stop early — and it means the result has to be checked with .defined before it is compared or sorted. A loop that does min over the distances will find the undefined ones sorting as zero, and pick the worst candidate as the best.