Generating a Permalink Slug in Haskell

One of the basic features of any blogging software is a function for generating “slugs” for your articles. A slug is normally a URL-friendly version of your article’s title that’s had any special characters and spaces stripped out, for example “My Awesome Article!!” might become “my-awesome-article”.

Generating a slug is a pretty straightforward exercise with regular expressions, but might not be quite so obvious in Haskell because of the rather arcane way that the regex libraries are provided.

Regex

Regular expressions in Haskell are weird. A Text.Regex.Base module provides an interface to a large variety of possible backends that do the heavy-lifting. The commonly-used functions in base are =~ and =~~, both polymorphic, meaning they behave differently depending on the type signature we specify for them. Read this beautifully written Haskell regex tutorial to understand this in more depth.

I’ve chosen to use the PCRE backend for Regex; a very fast module that is particularly suited to working with bytestrings. I’d install it on Archlinux like so (the cabal command is suited for any system):

sudo pacman -Sy pcre
cabal install regex-base regex-pcre

Another caveat in Haskell’s regex libraries is that no regex replace function is provided. Here’s one borrowed from spookylukey’s Haskell blog:

import Text.Regex.PCRE ( (=~~) )
import qualified Data.ByteString.Lazy.Char8 as B

{- | Replace using a regular expression. ByteString version. -}
regexReplace ::
    B.ByteString          -- ^ regular expression
    -> B.ByteString       -- ^ replacement text
    -> B.ByteString       -- ^ text to operate on
    -> B.ByteString
regexReplace regex replacement text = go text []
    where go str res =
              if B.null str
              then B.concat . reverse $ res
              else case (str =~~ regex) :: Maybe (B.ByteString, B.ByteString, B.ByteString) of
                     Nothing -> B.concat . reverse $ (str:res)
                     Just (bef, _ , aft) -> go aft (replacement:bef:res)

Bytestrings

It’s generally recommended for fast code to use the ByteString type rather than the Haskell’s built-in string, because the built-in string is considered slow. A Google search will reveal that massive memory usage and speed improvements in real-world applications are attributed to switching from strings to bytestrings.

Normally to use a bytestring in our code we’d have to write something like:

B.pack "my string"

However, if you have a reasonably recent version of GHC, you can specify the -XOverloadedStrings language option flag to define bytestrings the same way you define strings.

Generating Slugs

Now that we’ve got the basics in place, we can write a simple function to generate slugs for a blog:

import GHC.Unicode ( toLower )
import qualified Data.ByteString.Lazy.Char8 as B

makeSlug :: B.ByteString -> B.ByteString
makeSlug = regexReplace "[ _]" "-" 
         . regexReplace "[^a-z0-9_ ]+" "" 
         . B.map toLower

Save this to a file called Slug.hs and load up a GHCI prompt to test it out (remember to use -XOverloadedStrings!):

$ ghci -XOverloadedStrings
Prelude> :m Data.ByteString.Lazy.Char8
Prelude Data.ByteString.Lazy.Char8> :load Slug.hs
*Main Data.ByteString.Lazy.Char8> :set prompt "Prelude> "
Prelude> unpack $ makeSlug "My Awesome Article!!"
"my-awesome-article"
Prelude> unpack $ makeSlug "My Awesome w/ Numbers 789_2"
"my-awesome-w-numbers-789-2"

That’s the gist of it. The slug function itself can now be tweaked as desired.

Posted on February 11, 2010 from Silver Star