From afba6bce9c2e8ff77da2848f57f443088861db5d Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Tue, 19 Mar 2024 16:03:38 +1100
Subject: add benchmarking

---
 dedumi.hs | 74 ++-------------------------------------------------------------
 1 file changed, 2 insertions(+), 72 deletions(-)

(limited to 'dedumi.hs')

diff --git a/dedumi.hs b/dedumi.hs
index 3317a53..d26d48a 100644
--- a/dedumi.hs
+++ b/dedumi.hs
@@ -1,78 +1,8 @@
-{-# LANGUAGE DataKinds #-}
-{-# LANGUAGE DeriveGeneric #-}
-{-# LANGUAGE FlexibleInstances #-}
-{-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE OverloadedStrings #-}
-{-# LANGUAGE StandaloneDeriving #-}
-{-# LANGUAGE TypeApplications #-}
-{-# LANGUAGE TypeFamilies #-}
-{-# LANGUAGE TypeOperators #-}
-{-# OPTIONS_GHC -Wno-orphans #-}
-
 module Main where
 
-import Data.ByteString (ByteString)
-import qualified Data.ByteString as B
-import Data.Cuckoo
-import Data.FastQ
-import Data.Function
-import GHC.Prim (RealWorld)
-import GHC.TypeLits
-import Lens.Micro
+import Dedumi
 import Options.Generic
-import qualified Streamly.Data.Stream as S
-
-newtype NoLabel a = NoLabel {unNoLabel :: a} deriving (Generic)
-
-data Options w = Options
-  { umiLength :: w ::: Natural <?> "length of UMI prefix" <!> "8",
-    extraHashBases :: w ::: Natural <?> "extra hash bases to use for location proxy" <!> "4",
-    filterSize :: w ::: Natural <?> "Cuckoo filter size" <!> "200000000",
-    input1 :: w ::: NoLabel FilePath <?> "input fastq 1 path",
-    input2 :: w ::: NoLabel FilePath <?> "input fastq 2 path",
-    output1 :: w ::: NoLabel FilePath <?> "output fastq 1 path",
-    output2 :: w ::: NoLabel FilePath <?> "output fastq 2 path"
-  }
-  deriving (Generic)
-
-instance ParseFields a => ParseRecord (NoLabel a)
-
-instance ParseFields a => ParseFields (NoLabel a) where
-  parseFields msg _ _ def = fmap NoLabel (parseFields msg Nothing Nothing def)
-
-instance ParseRecord (Options Wrapped)
-
-instance CuckooFilterHash ByteString where
-  cuckooHash (Salt s) = saltedFnv1aByteString s
-  cuckooFingerprint (Salt s) = saltedSipHashByteString s
-  {-# INLINE cuckooHash #-}
-  {-# INLINE cuckooFingerprint #-}
-
-trim :: Int -> ReadPair -> ReadPair
-trim sz x =
-  x
-    & _1 . nucs %~ B.drop sz
-    & _2 . nucs %~ B.drop sz
-    & _1 . qual %~ B.drop sz
-    & _2 . qual %~ B.drop sz
-
-insert' :: (KnownNat b, KnownNat f) => Int -> CuckooFilter RealWorld b f ByteString -> ReadPair -> IO Bool
-insert' sz f x =
-  let y = B.take sz (x ^. _1 . nucs) <> B.take sz (x ^. _2 . nucs)
-   in member f y >>= \case
-        True -> pure False
-        False ->
-          insert f y >>= \case
-            True -> pure True
-            False -> error "filter full"
 
 main :: IO ()
-main = do
-  opts <- unwrapRecord "UMI deduplication"
-
-  f <- newCuckooFilter @4 @13 @ByteString 0 (filterSize opts)
-
-  parse (unNoLabel $ input1 opts) (unNoLabel $ input2 opts)
-    & S.filterM (insert' (fromIntegral $ umiLength opts + extraHashBases opts) f)
-    & fmap (trim . fromIntegral $ umiLength opts)
-    & unparse (unNoLabel $ output1 opts) (unNoLabel $ output2 opts)
+main =  unwrapRecord "UMI deduplication" >>= go
-- 
cgit v1.2.3