-- |
-- Module      : Streamly.Benchmark.Data.ParserD
-- Copyright   : (c) 2020 Composewell Technologies
--
-- License     : BSD-3-Clause
-- Maintainer  : streamly@composewell.com

{-# LANGUAGE CPP #-}
{-# LANGUAGE FlexibleContexts #-}
{-# LANGUAGE ScopedTypeVariables #-}

#ifdef __HADDOCK_VERSION__
#undef INSPECTION
#endif

#ifdef INSPECTION
{-# LANGUAGE TemplateHaskell #-}
{-# OPTIONS_GHC -fplugin Test.Inspection.Plugin #-}
#endif

module Main
  (
    main
  ) where

import Data.Functor.Identity (runIdentity)
import Data.Word (Word8)
import System.IO (Handle)
import Prelude hiding ()

import qualified Streamly.Prelude  as Stream
import qualified Streamly.Internal.Data.Array.Foreign as Array
import qualified Streamly.Internal.Data.Array.Stream.Foreign as ArrayStream
import qualified Streamly.Internal.FileSystem.Handle as Handle
import qualified Streamly.Internal.Unicode.Stream as Unicode

import Gauge hiding (env)
import Streamly.Benchmark.Common
import Streamly.Benchmark.Common.Handle

#ifdef INSPECTION
import Foreign.Storable (Storable)
import Streamly.Internal.Data.Stream.StreamD.Type (Step(..))
import Test.Inspection
#endif

-------------------------------------------------------------------------------
-- read chunked using toChunks
-------------------------------------------------------------------------------

-- | Get the last byte from a file bytestream.
toChunksLast :: Handle -> IO (Maybe Word8)
toChunksLast inh = do
    let s = Handle.toChunks inh
    larr <- Stream.last s
    return $ case larr of
        Nothing -> Nothing
        Just arr -> Array.getIndex arr (Array.length arr - 1)

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksLast
inspect $ 'toChunksLast `hasNoType` ''Step
#endif

-- | Count the number of bytes in a file.
toChunksSumLengths :: Handle -> IO Int
toChunksSumLengths inh =
    let s = Handle.toChunks inh
    in Stream.sum (Stream.map Array.length s)

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksSumLengths
inspect $ 'toChunksSumLengths `hasNoType` ''Step
#endif

-- | Sum the bytes in a file.
toChunksCountBytes :: Handle -> IO Word8
toChunksCountBytes inh = do
    let foldlArr' f z = runIdentity . Stream.foldl' f z . Array.toStream
    let s = Handle.toChunks inh
    Stream.foldl' (\acc arr -> acc + foldlArr' (+) 0 arr) 0 s

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksCountBytes
inspect $ 'toChunksCountBytes `hasNoType` ''Step
#endif

toChunksDecodeUtf8Arrays :: Handle -> IO ()
toChunksDecodeUtf8Arrays =
   Stream.drain . Unicode.decodeUtf8Arrays . Handle.toChunks

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksDecodeUtf8Arrays
-- inspect $ 'toChunksDecodeUtf8ArraysLenient `hasNoType` ''Step
#endif

-------------------------------------------------------------------------------
-- Splitting
-------------------------------------------------------------------------------

-- | Count the number of lines in a file.
toChunksSplitOnSuffix :: Handle -> IO Int
toChunksSplitOnSuffix =
    Stream.length . ArrayStream.splitOnSuffix 10 . Handle.toChunks

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksSplitOnSuffix
inspect $ 'toChunksSplitOnSuffix `hasNoType` ''Step
#endif

-- XXX use a word splitting combinator instead of splitOn and test it.
-- | Count the number of words in a file.
toChunksSplitOn :: Handle -> IO Int
toChunksSplitOn = Stream.length . ArrayStream.splitOn 32 . Handle.toChunks

#ifdef INSPECTION
inspect $ hasNoTypeClasses 'toChunksSplitOn
inspect $ 'toChunksSplitOn `hasNoType` ''Step
#endif

o_1_space_read_chunked :: BenchEnv -> [Benchmark]
o_1_space_read_chunked env =
    -- read using toChunks instead of read
    [ bgroup "reduce/toChunks"
        [ mkBench "Stream.last" env $ \inH _ ->
            toChunksLast inH
        -- Note: this cannot be fairly compared with GNU wc -c or wc -m as
        -- wc uses lseek to just determine the file size rather than reading
        -- and counting characters.
        , mkBench "Stream.sum . Stream.map Array.length" env $ \inH _ ->
            toChunksSumLengths inH
        , mkBench "splitOnSuffix" env $ \inH _ ->
            toChunksSplitOnSuffix inH
        , mkBench "splitOn" env $ \inH _ ->
            toChunksSplitOn inH
        , mkBench "countBytes" env $ \inH _ ->
            toChunksCountBytes inH
        , mkBenchSmall "decodeUtf8Arrays" env $ \inH _ ->
            toChunksDecodeUtf8Arrays inH
        ]
    ]

-------------------------------------------------------------------------------
-- copy with group/ungroup transformations
-------------------------------------------------------------------------------

-- | Lines and unlines
{-# NOINLINE copyChunksSplitInterposeSuffix #-}
copyChunksSplitInterposeSuffix :: Handle -> Handle -> IO ()
copyChunksSplitInterposeSuffix inh outh =
    Stream.fold (Handle.write outh)
        $ ArrayStream.interposeSuffix 10
        $ ArrayStream.splitOnSuffix 10
        $ Handle.toChunks inh

#ifdef INSPECTION
inspect $ hasNoTypeClassesExcept 'copyChunksSplitInterposeSuffix [''Storable]
inspect $ 'copyChunksSplitInterposeSuffix `hasNoType` ''Step
#endif

-- | Words and unwords
{-# NOINLINE copyChunksSplitInterpose #-}
copyChunksSplitInterpose :: Handle -> Handle -> IO ()
copyChunksSplitInterpose inh outh =
    Stream.fold (Handle.write outh)
        $ ArrayStream.interpose 32
        -- XXX this is not correct word splitting combinator
        $ ArrayStream.splitOn 32
        $ Handle.toChunks inh

#ifdef INSPECTION
inspect $ hasNoTypeClassesExcept 'copyChunksSplitInterpose [''Storable]
inspect $ 'copyChunksSplitInterpose `hasNoType` ''Step
#endif

o_1_space_copy_toChunks_group_ungroup :: BenchEnv -> [Benchmark]
o_1_space_copy_toChunks_group_ungroup env =
    [ bgroup "copy/toChunks/group-ungroup"
        [ mkBench "interposeSuffix . splitOnSuffix" env $ \inh outh ->
            copyChunksSplitInterposeSuffix inh outh
        , mkBenchSmall "interpose . splitOn" env $ \inh outh ->
            copyChunksSplitInterpose inh outh
        ]
    ]

-------------------------------------------------------------------------------
-- Driver
-------------------------------------------------------------------------------

moduleName :: String
moduleName = "Data.Array.Stream.Foreign"

main :: IO ()
main = do
    (value, cfg, benches) <- parseCLIOpts defaultStreamSize
    env <- mkHandleBenchEnv
    value `seq` runMode (mode cfg) cfg benches
        (allBenchmarks env)

    where

    allBenchmarks env =
        [ bgroup (o_1_space_prefix moduleName)
          ( o_1_space_read_chunked env
          ++ o_1_space_copy_toChunks_group_ungroup env
          )
        ]