gistfile1.hs

import Data.Map (Map)
import Data.MultiSet (fromList, toOccurList)
import Control.Monad

-- http://lpaste.net/131001

data Pint something = Pint something | EmptyPint
                    deriving (Show, Eq)
data Drink = Water | Milk | Tea | Beer
           deriving (Show, Eq)

derek :: Pint Drink -> Pint Drink
derek EmptyPint = EmptyPint
derek (Pint Beer) = EmptyPint
derek (Pint x) = Pint x

-- main :: IO ()
-- main = do
--   putStrLn "Hello, World!"
--   putStrLn $ "Derek given a " ++ show (Pint Milk) ++ " gives you " ++
--     show ( derek (Pint Milk))
--   putStrLn $ "Derek given a " ++ show (Pint Beer) ++ " gives back a " ++
--     show ( derek (Pint Beer))

-- showEach . findMostFrequent 10 . countWordFrequency $ source


-- countWordFrequency :: String -> FreqCount String
-- countWordFrequency  = map (\l -> (length l, head l)) . group . sort . words

-- findMostFrequent :: Int -> FreqCount String -> FreqCount String
-- findMostFrequent howmany counts =
--   take howmany $
--   (sortBy (comparing (Down . fst))) $
--   counts



type FreqCount x = [(x,Int)]

showEach :: Show x => [x] -> IO ()
showEach = mapM_ (putStrLn . show)

normalizeText :: String -> String
normalizeText = filter (not . isPunctuation ) . map toLower

getWords :: String -> [String]
getWords = words . normalizeText

getWordPairs :: String -> [(String,String)]
getWordPairs source = [ (a, b) | a:b:_ <-  tails allWords ]
                      where allWords = getWords source

countFrequency :: Ord x => [x] -> FreqCount x
countFrequency = toOccurList . fromList

sortFrequencies :: (Ord x) => FreqCount (w,x) -> FreqCount (w,x)
sortFrequencies = sortBy (comparing (Down .snd))

countWordPairs :: String -> FreqCount (String,String)
countWordPairs = sortFrequencies . countFrequency . getWordPairs

main :: IO()
main = do
  source <- getContents
  showEach
    $ take 15
    $ countWordPairs source