src/Gargantext/Prelude.hs

   1 {-# OPTIONS_GHC -fno-warn-name-shadowing #-}
   2 {-# OPTIONS_GHC -fno-warn-type-defaults  #-}
   3 {-# LANGUAGE     NoImplicitPrelude       #-}
   4
   5 {-
   6 TODO: import head impossible from Protolude: why ?
   7 -}
   8
   9 module Gargantext.Prelude
  10   ( module Gargantext.Prelude
  11   , module Protolude
  12   , headMay
  13   )
  14   where
  15
  16 import Protolude ( Bool(True, False), Int, Double, Integer
  17                  , Fractional, Num, Maybe(Just,Nothing)
  18                  , Floating, Char, IO
  19                  , pure, (<$>), panic
  20                  , Ord, Integral, Foldable, RealFrac, Monad, filter
  21                  , reverse, map, zip, drop, take, zipWith
  22                  , sum, fromIntegral, length, fmap
  23                  , takeWhile, sqrt, undefined, identity
  24                  , abs, maximum, minimum, return, snd, truncate
  25                  , (+), (*), (/), (-), (.), (>=), ($), (**), (^)
  26                  )
  27
  28 -- TODO import functions optimized in Utils.Count
  29 -- import Protolude hiding (head, last, all, any, sum, product, length)
  30 -- import Gargantext.Utils.Count
  31
  32 import qualified Data.List     as L hiding (head, sum)
  33 import qualified Control.Monad as M
  34 import qualified Data.Map as Map
  35 import qualified Data.Vector as V
  36 import Safe (headMay)
  37
  38 --pf :: (a -> Bool) -> [a] -> [a]
  39 --pf = filter
  40
  41 pr :: [a] -> [a]
  42 pr = reverse
  43
  44 --pm :: (a -> b) -> [a] -> [b]
  45 --pm = map
  46
  47 map2 :: (t -> b) -> [[t]] -> [[b]]
  48 map2 fun = map (map fun)
  49
  50 pz :: [a] -> [b] -> [(a, b)]
  51 pz  = zip
  52
  53 pd :: Int -> [a] -> [a]
  54 pd  = drop
  55
  56 ptk :: Int -> [a] -> [a]
  57 ptk = take
  58
  59 pzw :: (a -> b -> c) -> [a] -> [b] -> [c]
  60 pzw = zipWith
  61
  62 -- Exponential Average
  63 eavg :: [Double] -> Double
  64 eavg (x:xs) = a*x + (1-a)*(eavg xs)
  65   where a = 0.70
  66 eavg [] = 0
  67
  68 -- Simple Average
  69 mean :: Fractional a => [a] -> a
  70 mean xs = if L.null xs then 0.0
  71                        else sum xs / fromIntegral (length xs)
  72
  73 sumMaybe :: Num a => [Maybe a] -> Maybe a
  74 sumMaybe = fmap sum . M.sequence
  75
  76 variance :: Floating a => [a] -> a
  77 variance xs = mean $ map (\x -> (x - m) ** 2) xs where
  78     m = mean xs
  79
  80 deviation :: [Double] -> Double
  81 deviation = sqrt . variance
  82
  83 movingAverage :: Fractional b => Int -> [b] -> [b]
  84 movingAverage steps xs = map mean $ chunkAlong steps 1 xs
  85
  86 ma :: [Double] -> [Double]
  87 ma = movingAverage 3
  88
  89
  90 -- | Function to split a range into chunks
  91 chunkAlong :: Int -> Int -> [a] -> [[a]]
  92 chunkAlong a b l = only (while  dropAlong)
  93     where
  94         only      = map (take a)
  95         while     = takeWhile  (\x -> length x >= a)
  96         dropAlong = L.scanl (\x _y -> drop b x) l ([1..] :: [Integer])
  97
  98 -- | Optimized version (Vector)
  99 chunkAlong' :: Int -> Int -> V.Vector a -> V.Vector (V.Vector a)
 100 chunkAlong' a b l = only (while  dropAlong)
 101     where
 102         only      = V.map (V.take a)
 103         while     = V.takeWhile  (\x -> V.length x >= a)
 104         dropAlong = V.scanl (\x _y -> V.drop b x) l (V.fromList [1..])
 105
 106 -- | TODO Inverse of chunk ? unchunkAlong ?
 107 unchunkAlong :: Int -> Int -> [[a]] -> [a]
 108 unchunkAlong = undefined
 109
 110
 111 -- splitAlong [2,3,4] ("helloworld" :: [Char]) == ["he", "llo", "worl", "d"]
 112 splitAlong :: [Int] -> [Char] -> [[Char]]
 113 splitAlong _ [] = [] -- No list? done
 114 splitAlong [] xs = [xs] -- No place to split at? Return the remainder
 115 splitAlong (x:xs) ys = take x ys : splitAlong xs (drop x ys) -- take until our split spot, recurse with next split spot and list remainder
 116
 117 takeWhileM :: (Monad m) => (a -> Bool) -> [m a] -> m [a]
 118 takeWhileM _ [] = return []
 119 takeWhileM p (a:as) = do
 120     v <- a
 121     if p v
 122         then do
 123             vs <- takeWhileM p as
 124             return (v:vs)
 125         else return []
 126
 127 -- SUMS
 128 -- To select the right algorithme according to the type:
 129 -- https://github.com/mikeizbicki/ifcxt
 130
 131 sumSimple :: Num a => [a] -> a
 132 sumSimple = L.foldl' (+) 0
 133
 134 -- | https://en.wikipedia.org/wiki/Kahan_summation_algorithm
 135 sumKahan :: Num a => [a] -> a
 136 sumKahan = snd . L.foldl' go (0,0)
 137     where
 138         go (c,t) i = ((t'-t)-y,t')
 139             where
 140                 y  = i-c
 141                 t' = t+y
 142
 143 -- | compute part of the dict
 144 count2map :: (Ord k, Foldable t) => t k -> Map.Map k Double
 145 count2map xs = Map.map (/ (fromIntegral (length xs))) (count2map' xs)
 146
 147 -- | insert in a dict
 148 count2map' :: (Ord k, Foldable t) => t k -> Map.Map k Double
 149 count2map' xs = L.foldl' (\x y -> Map.insertWith' (+) y 1 x) Map.empty xs
 150
 151
 152 trunc :: (RealFrac a, Integral c, Integral b) => b -> a -> c
 153 trunc n = truncate . (* 10^n)
 154
 155 trunc' :: Int -> Double -> Double
 156 trunc' n x = fromIntegral $ truncate $ (x * 10^n)
 157
 158
 159 bool2int :: Num a => Bool -> a
 160 bool2int b = case b of
 161                   True  -> 1
 162                   False -> 0
 163
 164 bool2double :: Bool -> Double
 165 bool2double bool = case bool of
 166                   True  -> 1.0
 167                   False -> 0.0
 168
 169
 170
 171 -- Normalizing && scaling data
 172 scale :: [Double] -> [Double]
 173 scale = scaleMinMax
 174
 175 scaleMinMax :: [Double] -> [Double]
 176 scaleMinMax xs = map (\x -> (x - mi / (ma - mi + 1) )) xs'
 177     where
 178         ma  = maximum xs'
 179         mi  = minimum xs'
 180         xs' = map abs xs
 181
 182 scaleNormalize :: [Double] -> [Double]
 183 scaleNormalize xs = map (\x -> (x - v / (m + 1))) xs'
 184     where
 185         v = variance xs'
 186         m = mean     xs'
 187         xs' = map abs xs
 188
 189
 190
 191 normalize :: [Double] -> [Double]
 192 normalize as = normalizeWith identity as
 193
 194 normalizeWith :: Fractional b => (a -> b) -> [a] -> [b]
 195 normalizeWith extract bs = map (\x -> x/(sum bs')) bs'
 196     where
 197         bs' = map extract bs
 198
 199 -- Zip functions to add
 200 zipFst :: ([b] -> [a]) -> [b] -> [(a, b)]
 201 zipFst  f xs = zip (f xs) xs
 202
 203 zipSnd :: ([a] -> [b]) -> [a] -> [(a, b)]
 204 zipSnd f xs = zip xs (f xs)