Haskell: Pulling IP Addresses from Django Tracebacks

Spammer.hs

module Main where

import Control.Applicative
import Data.List
import System.Environment

data IpAddress = IpAddress String
               deriving (Show, Eq, Ord)

data IpTree = Leaf
          | Node IpAddress IpTree IpTree

cleanIp :: Char -> Bool
cleanIp '\'' = False
cleanIp '"' = False
cleanIp ',' = False
cleanIp _ = True

parseIp :: String -> IpAddress
parseIp line = IpAddress . filter cleanIp . last $ words line

filterIps :: String -> Bool
filterIps line = isInfixOf "REMOTE_ADDR" line

leadingZero :: String -> String
leadingZero n
  | length n < 3 = leadingZero $ '0' : n
  | otherwise = n

deLeadingZero :: String -> String
deLeadingZero ('0':xs) = deLeadingZero xs
deLeadingZero x = x

getAddress :: IpAddress -> String
getAddress (IpAddress address) = address

padIp :: (String -> String) -> String -> String
padIp f [] = []
padIp f ('.':xs) = '.' : padIp f xs
padIp f address = f takeAdd ++ (padIp f dropAdd)
  where takeAdd = takeWhile (/= '.') address
        dropAdd = dropWhile (/= '.') address

padding :: String -> String
padding address = padIp leadingZero address

depad :: String -> String
depad address = padIp deLeadingZero address

spammers :: String -> [IpAddress]
spammers = map parseIp . filter filterIps . lines 

ipOnly :: [IpAddress] -> [String]
ipOnly = map depad . sort . map padding . map getAddress

countUnique :: [String] -> [(String, Int)]
countUnique = map (\xs@(x:_) -> (x, length xs)) . group


flatten :: (String, Int) -> String
flatten (x, y) = x ++ " " ++ show y

main :: IO ()
main = do
       [f, g] <- getArgs
       input <- readFile f
       let output = unlines . map flatten . countUnique . ipOnly . spammers $ input
       writeFile g output