78449a3 · text · git.iko.soy

Data/Text/Search.hsnew file+44

compare with master

unifiedsplit

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

1+{-# LANGUAGE BangPatterns, ScopedTypeVariables #-}

2+

3+module Data.Text.Search where

4+

5+import qualified Data.Text as T ()

6+import qualified Data.Text.Array as A

7+import Data.Word (Word64)

8+import Data.Text.Internal (Text(..))

9+import Data.Bits ((.|.), (.&.))

10+import Data.Text.UnsafeShift (shiftL)

11+

12+indices :: Text -> Text -> [Int]

13+indices _needle@(Text narr noff nlen) _haystack@(Text harr hoff hlen)

14+ | ldiff < 0 = []

15+ | otherwise = outer 0

16+ where

17+ ldiff = hlen - nlen

18+ nlast = nlen - 1

19+ z = nindex nlast

20+ nindex k = A.unsafeIndex narr (noff+k)

21+ hindex k = A.unsafeIndex harr (hoff+k)

22+ (mask :: Word64, skip :: Int) = buildTable 0 0 (nlen-2)

23+ buildTable !i !msk !skp

24+ | i >= nlast = (msk .|. swizzle z, skp)

25+ | otherwise = buildTable (i+1) (msk .|. swizzle c) skp'

26+ where c = nindex i

27+ skp' | c == z = nlen - i - 2

28+ | otherwise = skp

29+ swizzle k = 1 `shiftL` (fromEnum k .&. 0x3f)

30+ outer !i

31+ | i > ldiff = []

32+ | c == z = if candidateMatch 0

33+ then i : outer (i + nlast)

34+ else if nextInPattern

35+ then outer (i + nlen)

36+ else outer (i + skip)

37+ | nextInPattern = outer (i + nlen)

38+ | otherwise = outer (i + 1)

39+ where c = hindex (i+nlast)

40+ candidateMatch j

41+ | j >= nlast = True

42+ | hindex (i+j) /= nindex j = False

43+ | otherwise = candidateMatch (j+1)

44+ nextInPattern = (mask .&. swizzle (hindex (i+nlen))) == 0

Data/Text/UnsafeShift.hs−2

@@ -45,14 +45,12 @@ instance UnsafeShift Word32 where

45 {-# INLINE shiftR #-}

46 shiftR (W32# x#) (I# i#) = W32# (x# `uncheckedShiftRL#` i#)

47

48−{-

49instance UnsafeShift Word64 where

48instance UnsafeShift Word64 where

50 {-# INLINE shiftL #-}

49 {-# INLINE shiftL #-}

51 shiftL (W64# x#) (I# i#) = W64# (x# `uncheckedShiftL64#` i#)

50 shiftL (W64# x#) (I# i#) = W64# (x# `uncheckedShiftL64#` i#)

52

51

53 {-# INLINE shiftR #-}

52 {-# INLINE shiftR #-}

54 shiftR (W64# x#) (I# i#) = W64# (x# `uncheckedShiftRL64#` i#)

53 shiftR (W64# x#) (I# i#) = W64# (x# `uncheckedShiftRL64#` i#)

55−-}

56

54

57instance UnsafeShift Int where

55instance UnsafeShift Int where

58 {-# INLINE shiftL #-}

56 {-# INLINE shiftL #-}

text.cabal+1

@@ -22,6 +22,7 @@ library

22 Data.Text.Foreign

23 Data.Text.Lazy

24 Data.Text.Lazy.Encoding

25+ Data.Text.Search

25 other-modules:

26 other-modules:

26 Data.Text.Array

27 Data.Text.Array

27 Data.Text.Encoding.Fusion

28 Data.Text.Encoding.Fusion