This is a small and fast command line tool that reads a two-column tab separated input from standard input and computes the shortest edit script (Myers' diff algorithm) to go from the string in column A to the string in column B. It also computed the edit distance (aka levenshtein distance).
It was written to build lemmatisers.
Install it using Rust's package manager:
cargo install sesdiff
No cargo/rust on your system yet? Do sudo apt install cargo on Debian/ubuntu based systems, brew install rust on mac, or use rustup.
This tool builds upon Dissimilar that provides the actual diff algorithm (will be downloaded and compiled in automatically).
$ sesdiff < input.tsv
Example input and output (reformatted for legibility, the first two columns correspond to the input). Output is in a four-column tab separated format:
hablaron hablar =[hablar]-[on] 2
contaron contar =[contar]-[on] 2
pidieron pedir =[p]-[i]+[e]=[di]-[eron]+[r] 6
говорим говорить =[говори]-[м]+[ть] 3
By default the full edit script will be provided in a simple language:
=[] - The text between brackets is identical in strings A and B-[] - The text between brackets is removed to get to string B+[] - The text between brackets is added to get to string BFor lemmatisation purposes, it makes sense for many languages to look at
suffixes (from right to left) and strip common prefixes. Pass the --suffix
option for that behaviour and output is now:
$ sesdiff --suffix < input.tsv
hablaron hablar -[on] 2
contaron contar -[on] 2
pidieron pedir -[eron]+[r]=[di]-[i]+[e] 6
говорим говорить -[м]+[ть] 3
There is also a --prefix option that strips common suffixes.
GNU General Public Licence v3