Pipe

Pipe is a Haskell library with operations for piping data through a pipeline of processes. It’s licensed under the 3-clause BSD license.

Example: MD5

Let’s say you want to quickly get the MD5 sum of a string. You know you have the md5sum binary on your system, so you don’t bother installing an additional library to do it. Instead you call the md5sum executable, writing a function which might look like the following:

import System.IO
import System.Process

getMD5 :: String -> IO String
getMD5 s = do
	(input, output, _, pid) <- runInteractiveCommand "md5sum"
	hPutStr input s
	result <- hGetContents output
	waitForProcess pid
	return result

Unless you’ve been messing around with System.Process lately, you probably had to check the docs for which exact function you wanted to use and what arguments to pass it.

Alternatively, you could use the Pipe library:

import System.Process.Pipe

getMD5 :: String -> IO String
getMD5 = pipeString [("md5sum",[])]

You should get the same result either way—but which code do you prefer? I find there’s too much noise in the standard way of doing things, so I wrote Pipe to make things simpler.

Obtaining

Assuming you have the cabal-install tool installed and working, the easiest way to obtain Pipe is with the cabal install pipe command.

Otherwise, you can download the source from Hackage, where you can also browse the documentation.

Failing that, the source package for the latest version, 2.1.2, is also available here:

Dependencies

The full list of dependencies, as listed in the .cabal file, is as follows:

base >= 4 && < 5
filepath >= 1.1 && < 2
process >= 1.0.1 && < 2
unix >= 2.3 && < 3 if not on a Windows platform