eed91e6 · text · git.iko.soy

git.iko.soy/text

on master help sign in

FileIndices → Search, and purify the benchmark

eed91e6Jasper Van der Jeugt15 years ago

tests/benchmarks/src/Data/Text/Benchmarks.hs+2−2

compare with master

unifiedsplit

@@ -13,7 +13,6 @@ import qualified Data.Text.Benchmarks.Builder as Builder

13import qualified Data.Text.Benchmarks.DecodeUtf8 as DecodeUtf8

14import qualified Data.Text.Benchmarks.EncodeUtf8 as EncodeUtf8

15import qualified Data.Text.Benchmarks.Equality as Equality

16−import qualified Data.Text.Benchmarks.FileIndices as FileIndices

17import qualified Data.Text.Benchmarks.FileRead as FileRead

16import qualified Data.Text.Benchmarks.FileRead as FileRead

18import qualified Data.Text.Benchmarks.FoldLines as FoldLines

17import qualified Data.Text.Benchmarks.FoldLines as FoldLines

19import qualified Data.Text.Benchmarks.HtmlCombinator as HtmlCombinator

18import qualified Data.Text.Benchmarks.HtmlCombinator as HtmlCombinator

@@ -21,6 +20,7 @@ import qualified Data.Text.Benchmarks.Ordering as Ordering

21import qualified Data.Text.Benchmarks.Pure as Pure

20import qualified Data.Text.Benchmarks.Pure as Pure

22import qualified Data.Text.Benchmarks.ReadNumbers as ReadNumbers

21import qualified Data.Text.Benchmarks.ReadNumbers as ReadNumbers

23import qualified Data.Text.Benchmarks.Replace as Replace

22import qualified Data.Text.Benchmarks.Replace as Replace

23+import qualified Data.Text.Benchmarks.Search as Search

24import qualified Data.Text.Benchmarks.WordCount as WordCount

26import qualified Data.Text.Benchmarks.Programs.Cut as Programs.Cut

@@ -42,7 +42,6 @@ benchmarks = do

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

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

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

45− , FileIndices.benchmark (tf "russian.txt") "принимая"

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

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

47 , FoldLines.benchmark (tf "russian.txt")

46 , FoldLines.benchmark (tf "russian.txt")

48 , HtmlCombinator.benchmark sink

47 , HtmlCombinator.benchmark sink

@@ -50,6 +49,7 @@ benchmarks = do

50 , Pure.benchmark (tf "japanese.txt")

49 , Pure.benchmark (tf "japanese.txt")

51 , ReadNumbers.benchmark (tf "numbers.txt")

50 , ReadNumbers.benchmark (tf "numbers.txt")

52 , Replace.benchmark (tf "russian.txt") sink "принимая" "своем"

51 , Replace.benchmark (tf "russian.txt") sink "принимая" "своем"

52+ , Search.benchmark (tf "russian.txt") "принимая"

53 , WordCount.benchmark (tf "russian.txt")

54 ]

tests/benchmarks/src/Data/Text/Benchmarks/FileIndices.hsdeleted−28

@@ -1,28 +0,0 @@

1−-- | Search for a pattern in a file, find the number of occurences

2−--

3−module Data.Text.Benchmarks.FileIndices

4− ( benchmark

5− ) where

6−

7−import Control.Exception (evaluate)

8−import Criterion (Benchmark, bench, bgroup)

9−import qualified Data.ByteString as B

10−import qualified Data.ByteString.Lazy as BL

11−import qualified Data.ByteString.Lazy.Search as BL

12−import qualified Data.Text.Lazy as TL

13−import qualified Data.Text.Lazy.Encoding as TL

14−import qualified Data.Text.Lazy.IO as TL

15−

16−benchmark :: FilePath -> TL.Text -> IO Benchmark

17−benchmark fp t = return $ bgroup "FileIndices"

18− [ bench "LazyText" $ TL.readFile fp >>= evaluate . text t

19− , bench "LazyByteString" $ BL.readFile fp >>= evaluate . byteString b

20− ]

21− where

22− b = B.concat $ BL.toChunks $ TL.encodeUtf8 t

23−

24−text :: TL.Text -> TL.Text -> Int

25−text needle = fromIntegral . TL.count needle

26−

27−byteString :: B.ByteString -> BL.ByteString -> Int

28−byteString needle = length . BL.indices needle

tests/benchmarks/src/Data/Text/Benchmarks/Search.hsnew file+44

@@ -0,0 +1,44 @@

1+-- | Search for a pattern in a file, find the number of occurences

2+--

3+module Data.Text.Benchmarks.Search

4+ ( benchmark

5+ ) where

7+import Criterion (Benchmark, bench, bgroup, whnf)

8+import qualified Data.ByteString as B

9+import qualified Data.ByteString.Lazy as BL

10+import qualified Data.ByteString.Lazy.Search as BL

11+import qualified Data.ByteString.Search as B

12+import qualified Data.Text as T

13+import qualified Data.Text.Encoding as T

14+import qualified Data.Text.IO as T

15+import qualified Data.Text.Lazy as TL

16+import qualified Data.Text.Lazy.IO as TL

17+

18+benchmark :: FilePath -> T.Text -> IO Benchmark

19+benchmark fp needleT = do

20+ b <- B.readFile fp

21+ bl <- BL.readFile fp

22+ t <- T.readFile fp

23+ tl <- TL.readFile fp

24+ return $ bgroup "FileIndices"

25+ [ bench "ByteString" $ whnf (byteString needleB) b

26+ , bench "LazyByteString" $ whnf (lazyByteString needleB) bl

27+ , bench "Text" $ whnf (text needleT) t

28+ , bench "LazyText" $ whnf (lazyText needleTL) tl

29+ ]

30+ where

31+ needleB = T.encodeUtf8 needleT

32+ needleTL = TL.fromChunks [needleT]

33+

34+byteString :: B.ByteString -> B.ByteString -> Int

35+byteString needle = length . B.indices needle

36+

37+lazyByteString :: B.ByteString -> BL.ByteString -> Int

38+lazyByteString needle = length . BL.indices needle

39+

40+text :: T.Text -> T.Text -> Int

41+text = T.count

42+

43+lazyText :: TL.Text -> TL.Text -> Int

44+lazyText needle = fromIntegral . TL.count needle