hwk

hwk (pronounced “hawk”) is a simple Haskell-based commandline text processing tool, somewhat similar to tools like awk, grep, sed. hwk applies composed pure Haskell functions to a list of lines of input, enabling text processing without having to remember an obscure DSL or awkward cli options. This tool can also help to encourage people to think functionally.

hwk was originally written by Lukas Martinelli in 2016-2017: see the original README file.

hwk is pretty similar to Hawk, so you may also want to try that for a different more sophisticated monadic implementation. Some of main differences are:

• hwk uses String for input for type simplicity, whereas hawk uses ByteString
• hawk has special options for controlling input and output field/lines delimiters, whereas in hwk everything is roughly just [String] -> [String] (more details below)
• by default hwk applies a function to the list of all the lines of stdin: hwk -l corresponds to hawk -m and hawk -a to hwk. hwk -a applies the function to the whole stdin.

Examples

Some simple use-cases are in the examples directory.

Change and append a string to each line:

$ seq 0 2 | hwk --line '(++ ".txt") . show . (+1) . int'
1.txt
2.txt
3.txt

or without line-mode: hwk 'map ((++ ".txt") . show . (+1) . int)'.

Sum all negative numbers:

$ seq -100 100 | hwk 'sum . filter (< 0) . ints'
-5050

The ints function transforms a list of strings into a list of ints

Factorials in your shell scripts!:

$ seq 10 12 | hwk --line 'let {fact 0 = 1; fact n = n * fact (n - 1)} in fact . int'
3628800
39916800
479001600

Extract data from a file:

$ cat /etc/passwd | hwk -l 'reverse . filter (/= "x") . take 3 . splitOn ":"' | head -3
0 root
1 bin
2 daemon

(uses splitOn from the extra library; -l is the short form of --line).

The argument passed to hwk must be a valid Haskell function: a function that takes a list of strings and returns a new list or a single value.

Check whether the input contains a certain string:

$ cat /etc/passwd | hwk --all 'bool "no" "yes" . isInfixOf "1000"'
yes

You can also type-check functions:

$ hwk --typecheck take
Int -> [a] -> [a]

or expressions:

$ hwk -t [1,2]
Num a => [a]

And evaluate expressions:

$ hwk -e '2 ^ 32 `div` 1024'
4194304

Run commands:

$ hwk --run getCurrentDirectory
/home/user/src

Configuration

hwk uses a Haskell configuration file ~/.config/hwk/Hwk.hs which provides the context for the hint evaluation of the supplied function. Hint (ghci) checks the current directory first when loading, so one can override the configuration on a directory basis.

The first time hwk is run it sets up ~/.config/hwk/Hwk.hs.

The default Hwk module configuration imports Prelude, Data.List, Data.Char, and System.FilePath into the hint interpreter.

You can add other modules to import to userModules or define your own functions to use in hwk expressions if you wish.

After a hwk version update you may need or wish to sync up your Hwk.hs file to take account of any new changes: a copy of the latest default Hwk.hs is also put in ~/.config/hwk/ with version suffix.

One can specify -c/--config-dir to load Hwk.hs from a different location.

Installation

Either use the install.sh script, or install by cabal-install or stack as described below:

install.sh script from source tree or git

Use stack unpack hwk or git clone https://github.com/juhp/hwk.

Then go to the source directory and run the install.sh script, which first runs stack install, then moves the binary installed by stack install to ~/.local/lib/hwk, and sets up a wrapper script ~/.local/bin/hwk which runs it.

If you wish you can change the resolver in stack.yaml first: it is also used to determine the resolver used by the created hwk wrapper script.

cabal

If you are on a Linux distro with a system installed ghc and Haskell libaries, you can install with cabal install to make use of them.

stack

Installing and running with stack is better if you do not have a system ghc and/or global system Haskell libraries installed.

If you prefer not to use install.sh in the source dir, you can install by hand: run stack install, and then run it with stack exec hwk ... using the same resolver.

How does hwk work?

• hwk use the hint library to apply haskell functions to input.
• By default it splits the input to a list of lines and applies the function to them
• Use -a or --all to apply a function to all the input, or -l/--line to map the function on each line separately, or -w/--words to map the function on each line of words,
• If you pass file arguments, their contents will be read and passed to the function.
• You can also typecheck the function or an expression with -t/--typecheck, evaluate an expr with -e/--eval, or -r/--run an IO statement.

Note the equivalences:

• hwk 'f . unlines' == hwk -a f
• hwk 'map f' == hwk -l f
• hwk -a 'f . lines' == hwk f
• hwk -l 'f . words' == hwk -w f
• hwk 'map (f . words)' == hwk -w f

Supported return types

The following return values are supported:

• String
• [String]
• [[String]]
• Int
• [Int]
• [[Int]]

Contribute

Open an issue or pull request at https://github.com/juhp/hwk to report problems or make suggestions and contributions.

Usage example contributions are also welcome.

Related/alternative projects

• https://github.com/gelisam/hawk
• https://github.com/bawolk/hsp
• https://en.wikipedia.org/wiki/AWK
• https://en.wikipedia.org/wiki/Sed
• https://code.google.com/p/pyp/