Experiments in fuzzy-searching, edit distances and related fields. Copy from this code at your own risk.

NOTE: Since this repo is meant for experiments, I reserve the right to rebase it.

$ make

Your basic edit distance, with the common optimization to only keep part of the previous row (or column, depending on world-view) of the matrix, and supporting a max_k to reduce the number of matrix entries that needs to be calculated when you don't care about the true distance, only whether it's at least max_k or not.

Only supports extended-ASCII (octet) alphabet.

int levenshtein(const char *s1, int s1_len, const char *s2, int s2_len, int max_k);

If you specify -1 for the string lengths, strlen() will be called internally.

Specify -1 for max_k if you want the true distance returned in every case.

A buffer proportional to the length of the shortest input string is allocated on the stack, i.e space is in O(min(|s1|,|s2|)).

The function contains a fair amount of pre-checks which sacrifice a little bit of performance in the common case for robustness.

$ ./levenshtein 1 volvo volvvos
ed('volvo', 'volvvos', 1) at least 1
$ ./levenshtein 2 volvo volvvos
ed('volvo', 'volvvos', 2) at least 2
$ ./levenshtein 3 volvo volvvos
ed('volvo', 'volvvos', 3) = 2

Bog-standard implementation, building the full matrix, i.e space is in O(|s1|*|s2|)+|Σ|

Only supports extended-ASCII (octet) alphabet.

static int damerau_ref(const char *s1, int s1_len, const char *s2, int s2_len);

$ ./damerau CA AC
damerau('CA', 'AC') = 1
$ ./damerau AC ABC
damerau('AC', 'ABC') = 1
$ ./damerau CA ABC
damerau('CA', 'ABC') = 2

N/A