module Main where
import List
import Char

main :: IO ()
main = interact translate

translate :: String -> String
translate = display . reverse . sort . freqs . group . sort . words

freqs :: [[a]] -> [(Int,a)]
freqs xs = [(length x, head x) | x <- xs]

display :: [(Int,String)] -> String
display xs = unlines [snd x ++ "  " ++ show (fst x) | x <- xs]

-- additions

-- remove "hapax legomena", i.e. words that occur only once
notHapax :: [(Int,String)] -> [(Int,String)]
notHapax xs = [x | x <- xs, fst x > 1]

-- histogram from requency table
histogram :: [(Int,String)] -> String
histogram xs = 
  unlines [snd x ++ "\t\t"++ replicate (proportion xs (fst x)) '*' | x <- xs]

-- a number in proportion to maximal frequency, in percent
proportion :: [(Int,String)] -> Int -> Int
proportion xs i = div (50 * i) (fst (head xs))

-- leave only letter and spaces
normalize :: String -> String
normalize s = [toLower x | x <- s, isAlpha x || isSpace x]

-- remove HTML tags
unHTML :: String -> String
unHTML [] = []
unHTML s =
  let
    (s1,s2) = span (/= '<') s
    (_, s3) = span (/= '>') s2
  in s1 ++ unHTML (drop 1 s3)