7537b75 · text · git.iko.soy

git.iko.soy/text

Rewrite find to use indices. New test is broken, alas.

--HG-- extra : convert_revision : 0f69b785d92da6da34b16c5e0e09d52e6a988355

commit7537b75a688ad5062a31d34a2c865d284c563cb9 authorBryan O'Sullivan <bos@serpentine.com> dateTue 6 Oct 2009, 18:38 (17 years ago) parent5f475ca

2 files changed+21−18

compare with master

unifiedsplit

Data/Text.hs+11−13 tests/Properties.hs+10−5

Data/Text.hs+11−13

@@ -1054,7 +1054,7 @@ filter :: (Char -> Bool) -> Text -> Text

10541054filter p t = unstream (S.filter p (stream t))

10551055{-# INLINE filter #-}

10561056

1057−-- /O(n*m)/ Find all non-overlapping instances of @needle@ in

1057+-- /O(n+m)/ Find all non-overlapping instances of @needle@ in

10581058-- @haystack@. The first element of the returned tuple is the prefix

10591059-- of @haystack@ before any matches of @needle@. The second element

10601060-- of the tuple is a list of non-overlapping substrings from

@@ -1071,20 +1071,18 @@ filter p t = unstream (S.filter p (stream t))

10711071--

10721072-- > concat (prefix : matches) == haystack

10731073-- > where (prefix, matches) = find needle haystack

1074+--

1075+-- In (unlikely) bad cases, this function's time complexity

1076+-- degenerates towards /O(n*m)/.

10741077find :: Text -> Text -> (Text, [Text])

1075−find pat@(Text _ _ plen) src@(Text sarr soff slen)

1076− | plen <= 0 = emptyError "find"

1077− | otherwise = (h,t)

1078+find pat src@(Text arr off len)

1079+ | null pat = emptyError "find"

1080+ | otherwise = case indices pat src of

1081+ [] -> (src, [])

1082+ (x:xs) -> (textP arr off x, go x xs)

10781083 where

1079− (h:t) = go 0 (search 0)

1080− go k (x:xs) = Text sarr k (x-k) : go x xs

1081− go _ [] = []

1082− search i

1083− | i >= slen = [slen] -- not found

1084− | pat `isPrefixOf` s = i : search (i+plen)

1085− | otherwise = search (i+d)

1086− where d = iter_ src i

1087− s = Text sarr (soff+i) (slen-i)

1084+ go !s (x:xs) = textP arr (s+off) (x-s) : go x xs

1085+ go s _ = [textP arr (s+off) (len-s)]

10881086{-# INLINE find #-}

10891087

10901088-------------------------------------------------------------------------------

tests/Properties.hs+10−5

@@ -2,7 +2,7 @@

22 TypeSynonymInstances #-}

33{-# OPTIONS_GHC -fno-enable-rewrite-rules #-}

5−import Test.QuickCheck

5+import Test.QuickCheck hiding (evaluate)

66import Text.Show.Functions ()

88import qualified Data.Bits as Bits (shiftL, shiftR)

@@ -15,7 +15,7 @@ import Data.Word (Word8, Word16, Word32)

1515import qualified Data.Text as T

1616import qualified Data.Text.Lazy as TL

1717import qualified Data.Text.Encoding as E

18−import Control.Exception (SomeException, try)

18+import Control.Exception (SomeException, evaluate, try)

1919import qualified Data.Text.Fusion as S

2020import qualified Data.Text.Fusion.Common as S

2121import Data.Text.Fusion.Size

@@ -38,8 +38,8 @@ import QuickCheckUtils (NotEmpty(..), small)

3838(=^=) :: (Eq a, Show a) => a -> a -> Bool

3939{-# NOINLINE (=^=) #-}

4040i =^= j = unsafePerformIO $ do

41− x <- try (return $! i)

42− y <- try (return $! j)

41+ x <- try (evaluate i)

42+ y <- try (evaluate j)

4343 case (x,y) of

4444 (Left (_ :: SomeException), Left (_ :: SomeException))

4545 -> return True

@@ -398,7 +398,11 @@ tl_inits = L.inits `eqP` (map unpackS . TL.inits)

398398t_tails = L.tails `eqP` (map unpackS . T.tails)

399399tl_tails = L.tails `eqP` (map unpackS . TL.tails)

400400

401−t_split_split p = T.split p `eq` Slow.split p

401+findSplit s t = let (x,xs) = T.find s t

402+ in x : L.map (T.drop (T.length s)) xs

403+

404+t_findSplit s = T.split s `eq` findSplit s

405+t_split_split s = T.split s `eq` Slow.split s

402406t_split_i (NotEmpty t) = id `eq` (T.intercalate t . T.split t)

403407tl_split_i (NotEmpty t) = id `eq` (TL.intercalate t . TL.split t)

404408t_splitTimes_i k (NotEmpty t) = id `eq` (T.intercalate t . T.splitTimes k t)

@@ -802,6 +806,7 @@ tests = [

802806 ],

803807

804808 testGroup "breaking many" [

809+ testProperty "t_findSplit" t_findSplit,

805810 testProperty "t_split_split" t_split_split,

806811 testProperty "t_split_i" t_split_i,

807812 testProperty "t_splitTimes_i" t_splitTimes_i,