afb20f6 · text · git.iko.soy

tests/benchmarks/src/Data/Text/Benchmarks.hs+4−6

compare with master

unifiedsplit

@@ -15,14 +15,13 @@ import qualified Data.Text.Benchmarks.EncodeUtf8 as EncodeUtf8

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

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

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

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

19−import qualified Data.Text.Benchmarks.Ordering as Ordering

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

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

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

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

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

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

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

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

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

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

25

23

24+import qualified Data.Text.Benchmarks.Programs.BigTable as Programs.BigTable

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

25import qualified Data.Text.Benchmarks.Programs.Cut as Programs.Cut

27import qualified Data.Text.Benchmarks.Programs.Sort as Programs.Sort

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

28import qualified Data.Text.Benchmarks.Programs.StripTags as Programs.StripTags

27import qualified Data.Text.Benchmarks.Programs.StripTags as Programs.StripTags

@@ -40,12 +39,10 @@ benchmarks = do

40 bs <- sequence

39 bs <- sequence

41 [ Builder.benchmark

40 [ Builder.benchmark

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

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

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

42+ , EncodeUtf8.benchmark "επανάληψη 竺法蘭共譯"

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

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

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

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

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

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

47− , HtmlCombinator.benchmark sink

48− , Ordering.benchmark (tf "russian.txt")

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

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

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

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

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

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

@@ -55,7 +52,8 @@ benchmarks = do

55

52

56 -- Program-like benchmarks

53 -- Program-like benchmarks

57 ps <- bgroup "Programs" `fmap` sequence

54 ps <- bgroup "Programs" `fmap` sequence

55+ [ Programs.BigTable.benchmark sink

58− [ Programs.Cut.benchmark (tf "russian.txt") sink 20 40

56+ , Programs.Cut.benchmark (tf "russian.txt") sink 20 40

59 , Programs.Sort.benchmark (tf "russian.txt") sink

57 , Programs.Sort.benchmark (tf "russian.txt") sink

60 , Programs.StripTags.benchmark (tf "yiwiki.xml") sink

58 , Programs.StripTags.benchmark (tf "yiwiki.xml") sink

61 , Programs.Throughput.benchmark (tf "russian.txt") sink

59 , Programs.Throughput.benchmark (tf "russian.txt") sink

tests/benchmarks/src/Data/Text/Benchmarks/Builder.hs+4

@@ -1,5 +1,9 @@

1-- | Testing the internal builder monoid

2--

3+-- Tested in this benchmark:

4+--

5+-- * Concatenating many small strings using a builder

6+--

3{-# LANGUAGE OverloadedStrings #-}

7{-# LANGUAGE OverloadedStrings #-}

4module Data.Text.Benchmarks.Builder

8module Data.Text.Benchmarks.Builder

5 ( benchmark

9 ( benchmark

tests/benchmarks/src/Data/Text/Benchmarks/DecodeUtf8.hs+14

@@ -1,3 +1,17 @@

1+-- | Test decoding of UTF-8

2+--

3+-- Tested in this benchmark:

4+--

5+-- * Decoding bytes using UTF-8

6+--

7+-- In some tests:

8+--

9+-- * Taking the length of the result

10+--

11+-- * Taking the init of the result

12+--

13+-- The latter are used for testing stream fusion.

14+--

1module Data.Text.Benchmarks.DecodeUtf8

15module Data.Text.Benchmarks.DecodeUtf8

2 ( benchmark

16 ( benchmark

3 ) where

17 ) where

tests/benchmarks/src/Data/Text/Benchmarks/EncodeUtf8.hs+17−17

@@ -1,33 +1,33 @@

1+-- | UTF-8 encode a text

2+--

3+-- Tested in this benchmark:

4+--

5+-- * Replicating a string a number of times

6+--

7+-- * UTF-8 encoding it

8+--

1module Data.Text.Benchmarks.EncodeUtf8

9module Data.Text.Benchmarks.EncodeUtf8

2 ( benchmark

10 ( benchmark

3 ) where

11 ) where

4

12

5−import Criterion (Benchmark, bgroup, bench)

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

6−import System.IO (Handle, hPutStr)

7import qualified Data.ByteString as B

14import qualified Data.ByteString as B

8import qualified Data.ByteString.Lazy as BL

15import qualified Data.ByteString.Lazy as BL

9import qualified Data.Text as T

16import qualified Data.Text as T

10import qualified Data.Text.Encoding as T

17import qualified Data.Text.Encoding as T

11−import qualified Data.Text.IO as T

12import qualified Data.Text.Lazy as TL

18import qualified Data.Text.Lazy as TL

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

19import qualified Data.Text.Lazy.Encoding as TL

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

15

20

16−benchmark :: Handle -> String -> IO Benchmark

21+benchmark :: String -> IO Benchmark

22+benchmark string = do

17−benchmark sink string = return $ bgroup "EncodeUtf8"

23+ return $ bgroup "EncodeUtf8"

18− [ bench "String" $ hPutStr sink $ concat $ replicate k string

24+ [ bench "Text" $ whnf (B.length . T.encodeUtf8) text

19− , bench "Text" $ T.hPutStr sink $ T.replicate k text

20− , bench "TextLazy" $ TL.hPutStr sink $

21− TL.replicate (fromIntegral k) lazyText

22− , bench "TextByteString" $ B.hPutStr sink $

23− T.encodeUtf8 $ T.replicate k text

24− , bench "TextByteStringLazy" $ BL.hPutStr sink $

25− TL.encodeUtf8 $ TL.replicate (fromIntegral k) lazyText

25+ , bench "LazyText" $ whnf (BL.length . TL.encodeUtf8) lazyText

26− ]

26+ ]

27 where

28 -- The string in different formats

29− text = T.pack string

29+ text = T.replicate k $ T.pack string

30− lazyText = TL.pack string

30+ lazyText = TL.replicate (fromIntegral k) $ TL.pack string

31

32 -- Amount

33 k = 100000

tests/benchmarks/src/Data/Text/Benchmarks/Equality.hs+5−2

@@ -1,9 +1,12 @@

1-- | Compare a string with a copy of itself that is identical except

2-- for the last character.

3--

4+-- Tested in this benchmark:

5+--

6+-- * Comparison of strings (Eq instance)

7+--

4module Data.Text.Benchmarks.Equality

8module Data.Text.Benchmarks.Equality

5− (

6− benchmark

9+ ( benchmark

7 ) where

10 ) where

8

11

9import Criterion (Benchmark, bgroup, bench, whnf)

12import Criterion (Benchmark, bgroup, bench, whnf)

tests/benchmarks/src/Data/Text/Benchmarks/FileRead.hs+4

@@ -1,5 +1,9 @@

1-- | Benchmarks simple file reading

2--

3+-- Tested in this benchmark:

4+--

5+-- * Reading a file from the disk

6+--

3module Data.Text.Benchmarks.FileRead

7module Data.Text.Benchmarks.FileRead

4 ( benchmark

8 ( benchmark

5 ) where

9 ) where

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

@@ -1,6 +1,10 @@

1-- | Read a file line-by-line using handles, and perform a fold over the lines.

2-- The fold is used here to calculate the number of lines in the file.

3--

4+-- Tested in this benchmark:

5+--

6+-- * Buffered, line-based IO

7+--

4{-# LANGUAGE BangPatterns #-}

8{-# LANGUAGE BangPatterns #-}

5module Data.Text.Benchmarks.FoldLines

9module Data.Text.Benchmarks.FoldLines

6 ( benchmark

10 ( benchmark

@@ -14,7 +18,7 @@ import qualified Data.Text.IO as T

14

18

15benchmark :: FilePath -> IO Benchmark

19benchmark :: FilePath -> IO Benchmark

16benchmark fp = return $ bgroup "ReadLines"

20benchmark fp = return $ bgroup "ReadLines"

17− [ bench "Text" $ withHandle $ foldLinesT (\n _ -> n + 1) (0 :: Int)

21+ [ bench "Text" $ withHandle $ foldLinesT (\n _ -> n + 1) (0 :: Int)

18 , bench "ByteString" $ withHandle $ foldLinesB (\n _ -> n + 1) (0 :: Int)

22 , bench "ByteString" $ withHandle $ foldLinesB (\n _ -> n + 1) (0 :: Int)

19 ]

23 ]

20 where

24 where

tests/benchmarks/src/Data/Text/Benchmarks/HtmlCombinator.hs → tests/benchmarks/src/Data/Text/Benchmarks/Programs/BigTable.hsrenamed+8−2

@@ -1,7 +1,13 @@

1-- | Create a large HTML table and dump it to a handle

2--

3+-- Tested in this benchmark:

4+--

5+-- * Creating a large HTML document using a builder

6+--

7+-- * Writing to a handle

8+--

3{-# LANGUAGE OverloadedStrings #-}

9{-# LANGUAGE OverloadedStrings #-}

4−module Data.Text.Benchmarks.HtmlCombinator

10+module Data.Text.Benchmarks.Programs.BigTable

5 ( benchmark

11 ( benchmark

6 ) where

12 ) where

7

13

@@ -13,7 +19,7 @@ import System.IO (Handle)

13import qualified Data.Text as T

19import qualified Data.Text as T

14

20

15benchmark :: Handle -> IO Benchmark

21benchmark :: Handle -> IO Benchmark

16−benchmark sink = return $ bench "HtmlCombinator" $ do

22+benchmark sink = return $ bench "BigTable" $ do

17 hPutStr sink "Content-Type: text/html\n\n<table>"

23 hPutStr sink "Content-Type: text/html\n\n<table>"

18 hPutStr sink . toLazyText . makeTable =<< rows

24 hPutStr sink . toLazyText . makeTable =<< rows

19 hPutStr sink "</table>"

25 hPutStr sink "</table>"

tests/benchmarks/src/Data/Text/Benchmarks/Programs/Cut.hs+12

@@ -1,5 +1,17 @@

1-- | Cut into a file, selecting certain columns (e.g. columns 10 to 40)

2--

3+-- Tested in this benchmark:

4+--

5+-- * Reading the file

6+--

7+-- * Splitting into lines

8+--

9+-- * Taking a number of characters from the lines

10+--

11+-- * Joining the lines

12+--

13+-- * Writing back to a handle

14+--

3module Data.Text.Benchmarks.Programs.Cut

15module Data.Text.Benchmarks.Programs.Cut

4 ( benchmark

16 ( benchmark

5 ) where

17 ) where

tests/benchmarks/src/Data/Text/Benchmarks/Programs/Sort.hs+12

@@ -1,5 +1,17 @@

1-- | This benchmark sorts the lines of a file, like the @sort@ unix utility.

2--

3+-- Tested in this benchmark:

4+--

5+-- * Reading the file

6+--

7+-- * Splitting into lines

8+--

9+-- * Sorting the lines

10+--

11+-- * Joining the lines

12+--

13+-- * Writing back to a handle

14+--

3{-# LANGUAGE OverloadedStrings #-}

15{-# LANGUAGE OverloadedStrings #-}

4module Data.Text.Benchmarks.Programs.Sort

16module Data.Text.Benchmarks.Programs.Sort

5 ( benchmark

17 ( benchmark

tests/benchmarks/src/Data/Text/Benchmarks/Programs/StripTags.hs+8

@@ -2,6 +2,14 @@

2--

3-- This program was originally contributed by Petr Prokhorenkov.

4--

5+-- Tested in this benchmark:

6+--

7+-- * Reading the file

8+--

9+-- * Replacing text between HTML tags (<>) with whitespace

10+--

11+-- * Writing back to a handle

12+--

5{-# OPTIONS_GHC -fspec-constr-count=5 #-}

13{-# OPTIONS_GHC -fspec-constr-count=5 #-}

6module Data.Text.Benchmarks.Programs.StripTags

14module Data.Text.Benchmarks.Programs.StripTags

7 ( benchmark

15 ( benchmark

tests/benchmarks/src/Data/Text/Benchmarks/Programs/Throughput.hs+8

@@ -6,6 +6,14 @@

6-- do any actual encoding/decoding here, while String and Text do have UTF-8

7-- encoding/decoding.

8--

9+-- Tested in this benchmark:

10+--

11+-- * Reading the file

12+--

13+-- * Replacing text between HTML tags (<>) with whitespace

14+--

15+-- * Writing back to a handle

16+--

9module Data.Text.Benchmarks.Programs.Throughput

17module Data.Text.Benchmarks.Programs.Throughput

10 ( benchmark

18 ( benchmark

11 ) where

19 ) where

tests/benchmarks/src/Data/Text/Benchmarks/Pure.hs+4

@@ -1,5 +1,9 @@

1-- | Benchmarks various pure functions from the Text library

2--

3+-- Tested in this benchmark:

4+--

5+-- * Most pure functions defined the string types

6+--

3{-# LANGUAGE BangPatterns, GADTs, MagicHash #-}

7{-# LANGUAGE BangPatterns, GADTs, MagicHash #-}

4{-# OPTIONS_GHC -fno-warn-orphans #-}

8{-# OPTIONS_GHC -fno-warn-orphans #-}

5module Data.Text.Benchmarks.Pure

9module Data.Text.Benchmarks.Pure

tests/benchmarks/src/Data/Text/Benchmarks/ReadNumbers.hs+11−9

@@ -12,6 +12,10 @@

12-- The different benchmarks will only take into account the values they can

13-- parse.

14--

15+-- Tested in this benchmark:

16+--

17+-- * Lexing/parsing of different numerical types

18+--

15module Data.Text.Benchmarks.ReadNumbers

19module Data.Text.Benchmarks.ReadNumbers

16 ( benchmark

20 ( benchmark

17 ) where

21 ) where

@@ -40,14 +44,14 @@ benchmark fp = do

40 b <- B.lines `fmap` B.readFile fp

44 b <- B.lines `fmap` B.readFile fp

41 bl <- BL.lines `fmap` BL.readFile fp

45 bl <- BL.lines `fmap` BL.readFile fp

42 return $ bgroup "ReadNumbers"

46 return $ bgroup "ReadNumbers"

43− [ bench "DecimalString" $ whnf (int . string readDec) s

47+ [ bench "DecimalString" $ whnf (int . string readDec) s

44 , bench "HexadecimalString" $ whnf (int . string readHex) s

48 , bench "HexadecimalString" $ whnf (int . string readHex) s

45− , bench "DoubleString" $ whnf (double . string readFloat) s

49+ , bench "DoubleString" $ whnf (double . string readFloat) s

46

50

47− , bench "DecimalText" $ whnf (int . text (T.signed T.decimal)) t

51+ , bench "DecimalText" $ whnf (int . text (T.signed T.decimal)) t

48 , bench "HexadecimalText" $ whnf (int . text (T.signed T.hexadecimal)) t

52 , bench "HexadecimalText" $ whnf (int . text (T.signed T.hexadecimal)) t

49− , bench "DoubleText" $ whnf (double . text T.double) t

53+ , bench "DoubleText" $ whnf (double . text T.double) t

50− , bench "RationalText" $ whnf (double . text T.rational) t

54+ , bench "RationalText" $ whnf (double . text T.rational) t

51

55

52 , bench "DecimalLazyText" $

56 , bench "DecimalLazyText" $

53 whnf (int . text (TL.signed TL.decimal)) tl

57 whnf (int . text (TL.signed TL.decimal)) tl

@@ -58,10 +62,8 @@ benchmark fp = do

58 , bench "RationalLazyText" $

62 , bench "RationalLazyText" $

59 whnf (double . text TL.rational) tl

63 whnf (double . text TL.rational) tl

60

64

61− , bench "DecimalByteString" $

62− whnf (int . byteString B.readInt) b

65+ , bench "DecimalByteString" $ whnf (int . byteString B.readInt) b

63− , bench "DoubleByteString" $

64− whnf (double . byteString B.readDouble) b

66+ , bench "DoubleByteString" $ whnf (double . byteString B.readDouble) b

65

67

66 , bench "DecimalLazyByteString" $

68 , bench "DecimalLazyByteString" $

67 whnf (int . byteString BL.readInt) bl

69 whnf (int . byteString BL.readInt) bl

tests/benchmarks/src/Data/Text/Benchmarks/Replace.hs+4

@@ -1,5 +1,9 @@

1-- | Replace a string by another string

2--

3+-- Tested in this benchmark:

4+--

5+-- * Search and replace of a pattern in a text

6+--

3module Data.Text.Benchmarks.Replace

7module Data.Text.Benchmarks.Replace

4 ( benchmark

8 ( benchmark

5 ) where

9 ) where

tests/benchmarks/src/Data/Text/Benchmarks/Search.hs+4

@@ -1,5 +1,9 @@

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

2--

3+-- Tested in this benchmark:

4+--

5+-- * Searching all occurences of a pattern using library routines

6+--

3module Data.Text.Benchmarks.Search

7module Data.Text.Benchmarks.Search

4 ( benchmark

8 ( benchmark

5 ) where

9 ) where

tests/benchmarks/src/Data/Text/Benchmarks/WordFrequencies.hs+8

@@ -1,5 +1,13 @@

1-- | A word frequency count using the different string types

2--

3+-- Tested in this benchmark:

4+--

5+-- * Splitting into words

6+--

7+-- * Converting to lowercase

8+--

9+-- * Comparing: Eq/Ord instances

10+--

3module Data.Text.Benchmarks.WordFrequencies

11module Data.Text.Benchmarks.WordFrequencies

4 ( benchmark

12 ( benchmark

5 ) where

13 ) where