6a4ea93 · text · git.iko.soy

git.iko.soy/text

on master help sign in

Transplant UTF-8 decoding benchmarks as of 44d20dca8f35

6a4ea93Bryan O'Sullivan15 years ago

tests/benchmarks/src/Data/Text/Benchmarks.hs+5−1

compare with master

unifiedsplit

@@ -39,7 +39,11 @@ benchmarks = do

39 -- Traditional benchmarks

40 bs <- sequence

41 [ Builder.benchmark

42− , DecodeUtf8.benchmark (tf "russian.txt")

42+ , DecodeUtf8.benchmark "html" (tf "libya-chinese.html")

43+ , DecodeUtf8.benchmark "xml" (tf "yiwiki.xml")

44+ , DecodeUtf8.benchmark "ascii" (tf "ascii.txt")

45+ , DecodeUtf8.benchmark "russian" (tf "russian.txt")

46+ , DecodeUtf8.benchmark "japanese" (tf "japanese.txt")

43 , EncodeUtf8.benchmark "επανάληψη 竺法蘭共譯"

47 , EncodeUtf8.benchmark "επανάληψη 竺法蘭共譯"

44 , Equality.benchmark (tf "japanese.txt")

48 , Equality.benchmark (tf "japanese.txt")

45 , FileRead.benchmark (tf "russian.txt")

49 , FileRead.benchmark (tf "russian.txt")

tests/benchmarks/src/Data/Text/Benchmarks/DecodeUtf8.hs+19−3

@@ -1,3 +1,5 @@

1+{-# LANGUAGE ForeignFunctionInterface #-}

1-- | Test decoding of UTF-8

3-- | Test decoding of UTF-8

2--

4--

3-- Tested in this benchmark:

5-- Tested in this benchmark:

@@ -16,7 +18,13 @@ module Data.Text.Benchmarks.DecodeUtf8

16 ( benchmark

18 ( benchmark

17 ) where

19 ) where

21+import Foreign.C.Types (CInt, CSize)

22+import Data.ByteString.Internal (ByteString(..))

23+import Foreign.Ptr (Ptr, plusPtr)

24+import Foreign.ForeignPtr (withForeignPtr)

25+import Data.Word (Word8)

26+import qualified Criterion as C

19−import Criterion (Benchmark, bgroup, bench, nf)

27+import Criterion (Benchmark, bgroup, nf)

20import qualified Codec.Binary.UTF8.Generic as U8

28import qualified Codec.Binary.UTF8.Generic as U8

21import qualified Data.ByteString as B

29import qualified Data.ByteString as B

22import qualified Data.ByteString.Lazy as BL

30import qualified Data.ByteString.Lazy as BL

@@ -25,12 +33,14 @@ import qualified Data.Text.Encoding as T

25import qualified Data.Text.Lazy as TL

33import qualified Data.Text.Lazy as TL

26import qualified Data.Text.Lazy.Encoding as TL

34import qualified Data.Text.Lazy.Encoding as TL

28−benchmark :: FilePath -> IO Benchmark

36+benchmark :: String -> FilePath -> IO Benchmark

29−benchmark fp = do

37+benchmark kind fp = do

30 bs <- B.readFile fp

38 bs <- B.readFile fp

31 lbs <- BL.readFile fp

39 lbs <- BL.readFile fp

40+ let bench name = C.bench (name ++ "+" ++ kind)

32 return $ bgroup "DecodeUtf8"

41 return $ bgroup "DecodeUtf8"

33 [ bench "Strict" $ nf T.decodeUtf8 bs

42 [ bench "Strict" $ nf T.decodeUtf8 bs

43+ , bench "IConv" $ iconv bs

34 , bench "StrictLength" $ nf (T.length . T.decodeUtf8) bs

44 , bench "StrictLength" $ nf (T.length . T.decodeUtf8) bs

35 , bench "StrictInitLength" $ nf (T.length . T.init . T.decodeUtf8) bs

45 , bench "StrictInitLength" $ nf (T.length . T.init . T.decodeUtf8) bs

36 , bench "Lazy" $ nf TL.decodeUtf8 lbs

46 , bench "Lazy" $ nf TL.decodeUtf8 lbs

@@ -41,3 +51,9 @@ benchmark fp = do

41 , bench "LazyStringUtf8" $ nf U8.toString lbs

51 , bench "LazyStringUtf8" $ nf U8.toString lbs

42 , bench "LazyStringUtf8Length" $ nf (length . U8.toString) lbs

52 , bench "LazyStringUtf8Length" $ nf (length . U8.toString) lbs

43 ]

53 ]

54+

55+iconv :: ByteString -> IO CInt

56+iconv (PS fp off len) = withForeignPtr fp $ \ptr ->

57+ time_iconv (ptr `plusPtr` off) (fromIntegral len)

58+

59+foreign import ccall unsafe time_iconv :: Ptr Word8 -> CSize -> IO CInt

tests/benchmarks/text-benchmarks.cabal+1

@@ -18,6 +18,7 @@ cabal-version: >=1.2

18executable text-benchmarks

19 hs-source-dirs: src ../..

20 c-sources: ../../cbits/cbits.c

21+ cbits/time_iconv.c

21 main-is: Data/Text/Benchmarks.hs

22 main-is: Data/Text/Benchmarks.hs

22 ghc-options: -Wall -O2

23 ghc-options: -Wall -O2

23 cpp-options: -DHAVE_DEEPSEQ

24 cpp-options: -DHAVE_DEEPSEQ