# Difference between revisions of "Data.List.Split"

From HaskellWiki

Line 8: | Line 8: | ||

** use a predicate on elements or sublists instead of giving explicit separators |
** use a predicate on elements or sublists instead of giving explicit separators |
||

** use approximate matching? |
** use approximate matching? |
||

+ | ** chunks of fixed length |
||

* how to split? |
* how to split? |
||

** discard the separators |
** discard the separators |
||

Line 64: | Line 65: | ||

| e == x = splitOn e xs |
| e == x = splitOn e xs |
||

| otherwise = let (h,t) = break f l in h:(splitEq e t) |
| otherwise = let (h,t) = break f l in h:(splitEq e t) |
||

+ | |||

+ | |||

+ | -- | split at regular intervals |
||

+ | splitEquidistant :: Int -> [a] -> [[a]] |
||

+ | splitEquidistant _ [] = [] |
||

+ | splitEquidistant n xs = y1 : split n y2 |
||

+ | where |
||

+ | (y1, y2) = splitAt n xs |
||

</haskell> |
</haskell> |

## Revision as of 19:41, 13 December 2008

A theoretical module which contains implementations/combinators for implementing every possible method of list-splitting known to man. This way no one has to argue about what the correct interface for split is, we can just have them all.

Some possible ways to split a list, to get your creative juices flowing:

- what to split on?
- single-element separator
- sublist separator
- use a list of possible separators instead of just one
- use a predicate on elements or sublists instead of giving explicit separators
- use approximate matching?
- chunks of fixed length

- how to split?
- discard the separators
- keep the separators with the preceding or following splits
- keep the separators as their own separate pieces of the result list
- what to do with separators at the beginning/end? create a blank split before/after, or not?

Add your implementations below! Once we converge on something good we can upload it to hackage.

```
{-# LANGUAGE ViewPatterns #-}
import Data.List (unfoldr)
-- intercalate :: [a] -> [[a]] -> [a]
-- intercalate x [a,b,c,x,y,z] = [a,x,b,x,c,x,x,y,x,z,x]
-- unintercalate :: [a] -> [a] -> [[a]]
-- unintercalate x [a,x,b,x,c,x,x,y,x,z,x] = [a,b,c,[],y,z]
-- unintercalate is the "inverse" of intercalate
match [] string = Just string
match (_:_) [] = Nothing
match (p:ps) (q:qs) | p == q = match ps qs
match (_:_) (_:_) | otherwise = Nothing
chopWith delimiter (match delimiter -> Just tail) = return ([], tail)
chopWith delimiter (c:cs) = chopWith delimiter cs >>= \(head, tail) ->
return (c:head, tail)
chopWith delimiter [] = Nothing
-- note: chopWith could be make 'more efficient' i.e. remove the >>=\-> bit
-- by adding an accumulator
unintercalate delimiter = unfoldr (chopWith delimiter)
-- > unintercalate "x" "axbxcxxyxzx"
-- ["a","b","c","","y","z"]
splitOn :: (a -> Bool) -> [a] -> [[a]]
splitOn _ [] = []
splitOn f l@(x:xs)
| f x = splitOn f xs
| otherwise = let (h,t) = break f l in h:(splitOn f t)
-- take the element who make predict true as delimiter
-- > splitOn even [1,3,5,6,7,3,3,2,1,1,1]
-- [[1,3,5],[7,3,3],[1,1,1]]
-- | like String split, except for any element that obeys Eq
splitEq :: Eq a -> [a] -> [[a]]
splitEq _ [] = []
splitEq e l@(x:xs)
| e == x = splitOn e xs
| otherwise = let (h,t) = break f l in h:(splitEq e t)
-- | split at regular intervals
splitEquidistant :: Int -> [a] -> [[a]]
splitEquidistant _ [] = []
splitEquidistant n xs = y1 : split n y2
where
(y1, y2) = splitAt n xs
```