Change LevenshteinDistance to Damerau–Levenshtein distance? about fuzzysearch HOT 4 OPEN

Thinking about this a bit more, I'm reluctant to replace the current implementation with something twice as slow with vague real-world advantages. I think a better approach would be to make the edit distance function pluggable (with the current LevenshteinDistance function as a default).

I will investigate how this can be done without breaking changes (hopefully). And afterwards maybe you can contribute a DamerauLevenshteinDistance function? What do you think?

from fuzzysearch.

It would be nice to see some benchmarks comparing them.

from fuzzysearch.

I have implemented the Damerau–LevenshteinDistance(Optimal string alignment distance), and do some benchmark on them.

It seems like Damerau–LevenshteinDistance is 2x slower than LevenshteinDistance.Do you think it's an acceptable change?
If the performance is ok, I will clean the code and raise a pr

benchmark
code

from fuzzysearch.

Thinking about this a bit more, I'm reluctant to replace the current implementation with something twice as slow with vague real-world advantages. I think a better approach would be to make the edit distance function pluggable (with the current LevenshteinDistance function as a default).

I will investigate how this can be done without breaking changes (hopefully). And afterwards maybe you can contribute a DamerauLevenshteinDistance function? What do you think?

I think it's good idea.I'm looking forward that

from fuzzysearch.