fix for
issue #234 - the framebuffers for side-by-side and top/bottom had alpha channels. so when sprites with alpha were blended into the framebuffer it would blend the sprite pixel alpha with the fb pixel alpha
module Test where
import Control.Monad
import Data.Word
import qualified Data.IntSet as IS
data OP = Sum
| Mul
| Sub
deriving Show
genOps :: Int -> [[OP]]
genOps 1 = [[Sum], [Mul], [Sub]]
genOps n = [a : as | a <- [Sum, Mul, Sub], as <- genOps (n - 1)]
genPos :: Int -> Int -> [[Int]]
genPos m 1 = map (:[]) [-m..m - 1]
genPos m n = [a : as | a <- [-m..m - 1], as <- genPos m (n - 1)]
hash :: [Int] -> [OP] -> [Int] -> Int
hash poss op s = foldl applyOp s' (zip ss op)
where
applyOp v (n, Sum) = (v + n) `mod` 256
applyOp v (n, Mul) = (v * n) `mod` 256
applyOp v (n, Sub) = (v - n) `mod` 256
(s' : ss) = map (\p -> if p >= 0 then s !! p else s !! (l + p)) poss
l = length s
test = do
a <- liftM lines getContents
let w = minimum $ map length a
let opsNum = 4
let opsList = genOps (opsNum - 1)
let posList = genPos w opsNum
let target = length a
let wordsList = map (map fromEnum) a
let hashedSize = IS.size . IS.fromList
print $ length a
putStrLn . unlines . map show $ filter (\l -> fst l == length a) $ [(hs, (p, o)) | p <- posList, o <- opsList, let hs = hashedSize . map (hash p o) $ wordsList]
didIunderstand' = do
a <- liftM lines getContents
print $ length a
print . IS.size . IS.fromList . map (testHash . map fromEnum) $ a
where
testHash s = let l = length s in (
(s !! (l - 2) * s !! 1) + s !! (l - 1) - s !! 0
) `mod` 256