Donald Bruce Stewart schrieb:
maeder:
-- | The 'splitBy' function takes a predicate and splits the input -- list at each element which satisfies the predicate. splitBy :: (a -> Bool) -> [a] -> [[a]] splitBy p s = case s of [] -> [] _ -> let (l, r) = break p s in case r of _ : t@(_:_) -> l : splitBy p t _ -> [l]
-- | The 'split' function splits the input list on each occurrence of -- the given element. split :: Eq a => a -> [a] -> [[a]] split c = splitBy (== c)
prop_lines_split xs = lines xs == split '\n' xs
prop_words_split xs = words xs == split ' ' xs
The last property is wrong. It must be: prop_words_split = words xs == filter (not . null) (split ' ' xs) Another property of my suggested functions: splitWords, splitLines, splitFields :: Eq a => a -> [a] -> [[a]] id = concat . intersperse [c] . splitFields c -- :: [a] -> [a] splitLines is equal to splitFields as long as there is no last seperator (or terminator) element. splitWords can be easily derived from splitLines or splitFields via "filter (not . null)". splitFields c = splitLines c . (++ [c]) Which of the two versions splitFields or splitLines should get the general name "split" is a matter of taste and of compatibility to the currently incompatible Data.PackedString or (your) Data.ByteString. If we cannot agree, I'ld suggest to keep (my) longer names, only. Cheers Christian