Literate Haskell support for Markdown

February 19, 2025 ยท View on GitHub

markdown-unlit is a custom unlit program. It can be used to extract Haskell code from Markdown files.

To use it with GHC, add

ghc-options: -pgmL markdown-unlit

to your cabal file.

Extended example

tl;dr markdown-unlit allows you to have a README.md, that at the same time is a literate Haskell program.

The following steps show you how to set things up, so that:

  • the Haskell code in your README.md gets syntax highlighted on GitHub
  • you can run your literate Haskell within GHCi
  • you can create a Cabal test-suite from your README.md (no broken code examples anymore, yay!)

The complete code of this example is provided in the example subdirectory.

1. Install markdown-unlit

$ cabal update && cabal install markdown-unlit

2. Create a README.md

# nifty-library: Do nifty things (effortlessly!)

Here is a basic example:

```haskell
main :: IO ()
main = putStrLn "That was easy!"
```

Code blocks with class haskell are syntax highlighted on GitHub (like so).

$ ln -s README.md README.lhs

4. Run your code

At this point we can load the code into GHCi:

$ ghci -pgmL markdown-unlit README.lhs
*Main> main
That was easy!

Or better yet, pipe the required flag into a .ghci file, and forget about it:

$ echo ':set -pgmL markdown-unlit' >> .ghci

$ ghci README.lhs
*Main> main
That was easy!

5. Create a Cabal test-suite

name:             nifty-library
version:          0.0.0
build-type:       Simple
cabal-version:    >= 1.8

library
  -- nothing here yet

test-suite readme
  type:           exitcode-stdio-1.0
  main-is:        README.lhs
  build-depends:  base, markdown-unlit
  ghc-options:    -pgmL markdown-unlit
  build-tool-depends: markdown-unlit:markdown-unlit

Run it like so:

$ cabal configure --enable-tests && cabal build && cabal test

Reordering code blocks

Code blocks that are tagged with top are moved to the beginning of the source file.

Example:

## Introduction

```haskell top
{-# LANGUAGE OverloadedStrings #-}
module MyModule where
```

## Working with textual data

```haskell top
import qualified Data.Text as Text
```
```haskell
-- |
-- >>> foo
-- 3
foo :: Int
foo = Text.length "foo"
```

## Working with binary data

```haskell top
import qualified Data.ByteString as ByteString
```
```haskell
-- |
-- >>> bar
-- 3
bar :: Int
bar = ByteString.length "foo"
```

$ doctest -pgmL markdown-unlit MyModule.lhs
Examples: 2  Tried: 2  Errors: 0  Failures: 0

More fine-grained control over the ordering

Optionally, top can be followed by :n where n is a non-negative integer. Code blocks with a smaller n move above code blocks with a larger n. top is an alias for top:0.

Customizing

By default, markdown-unlit extracts all code that is marked with haskell, unless it is marked with ignore as well. You can customize this by passing -optL <pattern> to GHC.

A simple pattern is just a class name, e.g.:

-optL foo

extracts all code that is marked with foo.

A class name can be negated by prepending it with a !, e.g.

-optL !foo

extracts all code, unless it is marked with foo.

You can use + to combine two patterns with AND, e.g.

-optL foo+bar

extracts all code that is marked with both foo and bar.

If -optL is given multiple times, the patterns are combined with OR, e.g.

-optL foo -optL bar

extracts all code that is either marked with foo or bar.

Development

Limitations

If you want to get any limitation lifted, open a ticket or send a pull request.

Contributing

Add tests for new code, and make sure that the test suite passes with your modifications.

cabal configure --enable-tests && cabal build && cabal test

Real world examples

That's it, have fun!