For Python there are the following libraries:

• RapidFuzz provides various string metrics with a focus on makingthem as fast as possible.
• Levenshtein provides a couple string metrics and median implementations.For string metrics generally rapidfuzz should be te preferred choice. Opposed to all other libraries this library iscurrently GPLv2 licensed.
• python-Levenshtein this is an alias to the Levenshtein library thatonly exists for backwards compatibility reasons.
• JaroWinkler provides a fast implementation of Jaro and JaroWinkler similarity.This was placed in RapidFuzz at some point and so this largely exists for applications already using it.
• CyDifflib drop in replacement for difflib in the Python standard librarywhich is faster.

For C++ there are the following libraries:

• rapidfuzz-cpp provides various string metrics with a focus on makingthem as fast as possible. This should be used when the performance of the algorithms is more important thana slightly larger binary.

Similar to rust there will be an implementation focussing on binary size in the future.

For rust there are the following libraries:

• rapidfuzz-rs provides various string metrics with a focus on makingthem as fast as possible. This should be used when the performance of the algorithms is more important thana slightly larger binary.
• strsim-rs provides a lot of string metrics of rapidfuzz, but has a focus onkeeping the binary size as small as possible. This should be used when performance of the algorithms doesn't reallymatter too much, since the application only compares a small set of strings. An example for this would be suggestions ina CLI.

Repositories