src/Gargantext/Prelude.hs

   1 {-|
   2 Module      : Gargantext.Prelude
   3 Description :
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Here is a longer description of this module, containing some
  11 commentary with @some markup@.
  12 -}
  13
  14 {-# OPTIONS_GHC -fno-warn-name-shadowing #-}
  15 {-# OPTIONS_GHC -fno-warn-type-defaults  #-}
  16
  17 {-# LANGUAGE     NoImplicitPrelude       #-}
  18
  19 module Gargantext.Prelude
  20   ( module Gargantext.Prelude
  21   , module Protolude
  22   , headMay
  23   , module Text.Show
  24   , module Text.Read
  25   , cs
  26   , module Data.Maybe
  27   )
  28   where
  29
  30 import Data.Maybe (isJust, fromJust, maybe)
  31 import Protolude ( Bool(True, False), Int, Double, Integer
  32                  , Fractional, Num, Maybe(Just,Nothing)
  33                  , Enum, Bounded, Float
  34                  , Floating, Char, IO
  35                  , pure, (<*>), (<$>), panic
  36                  , putStrLn
  37                  , head, flip
  38                  , Ord, Integral, Foldable, RealFrac, Monad, filter
  39                  , reverse, map, mapM, zip, drop, take, zipWith
  40                  , sum, fromIntegral, length, fmap, foldl, foldl'
  41                  , takeWhile, sqrt, undefined, identity
  42                  , abs, min, max, maximum, minimum, return, snd, truncate
  43                  , (+), (*), (/), (-), (.), ($), (&), (**), (^), (<), (>), log
  44                  , Eq, (==), (>=), (<=), (<>), (/=)
  45                  , (&&), (||), not, any
  46                  , fst, snd, toS
  47                  , elem, die, mod, div, const, either
  48                  , curry, uncurry, repeat
  49                  , otherwise, when
  50                  , undefined
  51                  , IO()
  52                  , compare
  53                  , on
  54                  )
  55
  56 -- TODO import functions optimized in Utils.Count
  57 -- import Protolude hiding (head, last, all, any, sum, product, length)
  58 -- import Gargantext.Utils.Count
  59 import qualified Data.List     as L hiding (head, sum)
  60 import qualified Control.Monad as M
  61
  62 import Data.Map (Map)
  63 import qualified Data.Map as M
  64
  65 import Data.Map.Strict (insertWith)
  66 import qualified Data.Vector as V
  67 import Safe (headMay)
  68 import Text.Show (Show(), show)
  69 import Text.Read (Read())
  70 import Data.String.Conversions (cs)
  71
  72 --pf :: (a -> Bool) -> [a] -> [a]
  73 --pf = filter
  74
  75 pr :: [a] -> [a]
  76 pr = reverse
  77
  78 --pm :: (a -> b) -> [a] -> [b]
  79 --pm = map
  80
  81 map2 :: (t -> b) -> [[t]] -> [[b]]
  82 map2 fun = map (map fun)
  83
  84 -- Exponential Average
  85 eavg :: [Double] -> Double
  86 eavg (x:xs) = a*x + (1-a)*(eavg xs)
  87   where a = 0.70
  88 eavg [] = 0
  89
  90 -- Simple Average
  91 mean :: Fractional a => [a] -> a
  92 mean xs = if L.null xs then 0.0
  93                        else sum xs / fromIntegral (length xs)
  94
  95 sumMaybe :: Num a => [Maybe a] -> Maybe a
  96 sumMaybe = fmap sum . M.sequence
  97
  98 variance :: Floating a => [a] -> a
  99 variance xs = mean $ map (\x -> (x - m) ** 2) xs where
 100     m = mean xs
 101
 102 deviation :: [Double] -> Double
 103 deviation = sqrt . variance
 104
 105 movingAverage :: Fractional b => Int -> [b] -> [b]
 106 movingAverage steps xs = map mean $ chunkAlong steps 1 xs
 107
 108 ma :: [Double] -> [Double]
 109 ma = movingAverage 3
 110
 111 -- | splitEvery n == chunkAlong n n
 112 splitEvery :: Int -> [a] -> [[a]]
 113 splitEvery _ [] = []
 114 splitEvery n xs =
 115   let (h,t) = L.splitAt n xs
 116   in  h : splitEvery n t
 117
 118 -- | Function to split a range into chunks
 119 chunkAlong :: Int -> Int -> [a] -> [[a]]
 120 chunkAlong a b l = only (while  dropAlong)
 121     where
 122         only      = map (take a)
 123         while     = takeWhile  (\x -> length x >= a)
 124         dropAlong = L.scanl (\x _y -> drop b x) l ([1..] :: [Integer])
 125
 126 -- | Optimized version (Vector)
 127 chunkAlong' :: Int -> Int -> V.Vector a -> V.Vector (V.Vector a)
 128 chunkAlong' a b l = only (while  dropAlong)
 129     where
 130         only      = V.map (V.take a)
 131         while     = V.takeWhile  (\x -> V.length x >= a)
 132         dropAlong = V.scanl (\x _y -> V.drop b x) l (V.fromList [1..])
 133
 134 -- | TODO Inverse of chunk ? unchunkAlong ?
 135 unchunkAlong :: Int -> Int -> [[a]] -> [a]
 136 unchunkAlong = undefined
 137
 138
 139 -- splitAlong [2,3,4] ("helloworld" :: [Char]) == ["he", "llo", "worl", "d"]
 140 splitAlong :: [Int] -> [Char] -> [[Char]]
 141 splitAlong _ [] = [] -- No list? done
 142 splitAlong [] xs = [xs] -- No place to split at? Return the remainder
 143 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
 144
 145 takeWhileM :: (Monad m) => (a -> Bool) -> [m a] -> m [a]
 146 takeWhileM _ [] = return []
 147 takeWhileM p (a:as) = do
 148     v <- a
 149     if p v
 150         then do
 151             vs <- takeWhileM p as
 152             return (v:vs)
 153         else return []
 154
 155 -- SUMS
 156 -- To select the right algorithme according to the type:
 157 -- https://github.com/mikeizbicki/ifcxt
 158
 159 sumSimple :: Num a => [a] -> a
 160 sumSimple = L.foldl' (+) 0
 161
 162 -- | https://en.wikipedia.org/wiki/Kahan_summation_algorithm
 163 sumKahan :: Num a => [a] -> a
 164 sumKahan = snd . L.foldl' go (0,0)
 165     where
 166         go (c,t) i = ((t'-t)-y,t')
 167             where
 168                 y  = i-c
 169                 t' = t+y
 170
 171 -- | compute part of the dict
 172 count2map :: (Ord k, Foldable t) => t k -> Map k Double
 173 count2map xs = M.map (/ (fromIntegral (length xs))) (count2map' xs)
 174
 175 -- | insert in a dict
 176 count2map' :: (Ord k, Foldable t) => t k -> Map k Double
 177 count2map' xs = L.foldl' (\x y -> insertWith (+) y 1 x) M.empty xs
 178
 179
 180 trunc :: (RealFrac a, Integral c, Integral b) => b -> a -> c
 181 trunc n = truncate . (* 10^n)
 182
 183 trunc' :: Int -> Double -> Double
 184 trunc' n x = fromIntegral $ truncate $ (x * 10^n)
 185
 186
 187 bool2int :: Num a => Bool -> a
 188 bool2int b = case b of
 189                   True  -> 1
 190                   False -> 0
 191
 192 bool2double :: Bool -> Double
 193 bool2double bool = case bool of
 194                   True  -> 1.0
 195                   False -> 0.0
 196
 197
 198
 199 -- Normalizing && scaling data
 200 scale :: [Double] -> [Double]
 201 scale = scaleMinMax
 202
 203 scaleMinMax :: [Double] -> [Double]
 204 scaleMinMax xs = map (\x -> (x - mi / (ma - mi + 1) )) xs'
 205     where
 206         ma  = maximum xs'
 207         mi  = minimum xs'
 208         xs' = map abs xs
 209
 210 scaleNormalize :: [Double] -> [Double]
 211 scaleNormalize xs = map (\x -> (x - v / (m + 1))) xs'
 212     where
 213         v = variance xs'
 214         m = mean     xs'
 215         xs' = map abs xs
 216
 217
 218
 219 normalize :: [Double] -> [Double]
 220 normalize as = normalizeWith identity as
 221
 222 normalizeWith :: Fractional b => (a -> b) -> [a] -> [b]
 223 normalizeWith extract bs = map (\x -> x/(sum bs')) bs'
 224     where
 225         bs' = map extract bs
 226
 227 -- Zip functions to add
 228 zipFst :: ([b] -> [a]) -> [b] -> [(a, b)]
 229 zipFst  f xs = zip (f xs) xs
 230
 231 zipSnd :: ([a] -> [b]) -> [a] -> [(a, b)]
 232 zipSnd f xs = zip xs (f xs)
 233
 234 -- Just
 235 unMaybe :: [Maybe a] -> [a]
 236 unMaybe = map fromJust . L.filter isJust
 237
 238 -- maximumWith
 239 maximumWith f = L.maximumBy (compare `on` f)
 240