typodist

What

This program generates a list of possible typos of a given word.

The generated typos are sorted in a way that likely typos appear higher in the list of all typos than unlikely typos i.e. given test gest is a more common typo than lest because the t key on the keyboard is next to the g key but far from the l key and tset is even more likely since s and e where only swapped.

cargo run -- generate "test" | head -n 10

test
ttest
testt
teest
tesst
tets
etst
tset
twst
t3st

Different keyboard layouts and metrics can be selected

cargo run -- generate --help

typodist-generate 0.1.0
Matthias Schmitt <male.schmitt@posteo.de>

USAGE:
    typodist generate [FLAGS] [OPTIONS] <input>

FLAGS:
        --detailed         print detailed information about each generated typo (with metric distance)
    -h, --help             Prints help information
        --mobile-metric    
        --simple-metric    Use a simpler distance metric which treats insertion, deletion, substitution and
                           transposition equally
    -V, --version          Prints version information

OPTIONS:
    -l, --layout <layout>        Layout to use for distance metric. [default: QWERTZ]  [possible values: QWERTZ,
                                 QWERTZ_ANYSOFT_EXTRA]
    -m, --max-dist <max_dist>    maximum distance of the generated words [default: 1.1]

ARGS:
    <input>    input to generate "near" typos

How

Every generated typo is compared to the input and receives a ‘typo-likeliness’ score. This score is based on the Damerau–Levenshtein distance (Levenshtein distance + transpositions) but in addition to the edit distance of the word the distance of the characters on the keyboard are used to rate the likeliness of a insertion, substitution. Deletions and transposition get rated at a fixed cost but different metrics can be implemented.

Ideas

Instead of the distance of the characters it would be more accurate to transform the characters into actual key presses e.g. ‘A’ -> [shift, a] or ‘}’ -> [alt-gr, 8] (on qwertz) and then use the distance of the keys on the keyboard layout.

License

Licensed under the GNU Affero General Public License Version 3. This file may not be copied, modified, or distributed except according to those terms.