@@ -3,6 +3,9 @@
3#if __GLASGOW_HASKELL__ >= 702
3#if __GLASGOW_HASKELL__ >= 702
4{-# LANGUAGE Trustworthy #-}
4{-# LANGUAGE Trustworthy #-}
6+#if __GLASGOW_HASKELL__ >= 708
7+{-# LANGUAGE TypeFamilies #-}
8-- Module : Data.Text
11-- Module : Data.Text
@@ -11,8 +14,7 @@
11-- (c) 2008, 2009 Tom Harper
14-- (c) 2008, 2009 Tom Harper
13-- License : BSD-style
16-- License : BSD-style
14−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,
17+-- Maintainer : bos@serpentine.com
16-- Stability : experimental
18-- Stability : experimental
17-- Portability : GHC
19-- Portability : GHC
@@ -30,8 +32,8 @@
31-- To use an extended and very rich family of functions for working
33-- To use an extended and very rich family of functions for working
32-- with Unicode text (including normalization, regular expressions,
34-- with Unicode text (including normalization, regular expressions,
33−-- non-standard encodings, text breaking, and locales), see the
35+-- non-standard encodings, text breaking, and locales), see
34−-- @text-icu@ package: <http://hackage.haskell.org/package/text-icu>
36+-- <http://hackage.haskell.org/package/text-icu the text-icu package >.
36module Data.Text
38module Data.Text
@@ -123,8 +125,11 @@ module Data.Text
124 -- ** Breaking strings
126 -- ** Breaking strings
127 , takeWhile
131 , takeWhile
128 , dropWhile
133 , dropWhile
129 , dropWhileEnd
134 , dropWhileEnd
130 , dropAround
135 , dropAround
@@ -187,82 +192,94 @@ module Data.Text
188 -- * Low level operations
193 -- * Low level operations
192import Prelude (Char, Bool(..), Int, Maybe(..), String,
198import Prelude (Char, Bool(..), Int, Maybe(..), String,
193 Eq(..), Ord(..), Ordering(..), (++),
199 Eq(..), Ord(..), Ordering(..), (++),
194− Read(..), Show(..),
200+ Read(..),
195− (&&), (||), (+), (-), (.), ($), ($!), (>>), (*),
201+ (&&), (||), (+), (-), (.), ($), ($!), (>>),
196− maxBound, not, return, otherwise, quot)
202+ not, return, otherwise, quot)
197#if defined(HAVE_DEEPSEQ)
203#if defined(HAVE_DEEPSEQ)
198−import Control.DeepSeq (NFData)
204+import Control.DeepSeq (NFData(rnf))
200#if defined(ASSERTS)
206#if defined(ASSERTS)
201import Control.Exception (assert)
207import Control.Exception (assert)
203import Data.Char (isSpace)
209import Data.Char (isSpace)
204−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))
210+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,
205−import Data.Data (mkNoRepType)
211+ Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))
206import Control.Monad (foldM)
212import Control.Monad (foldM)
213+import Control.Monad.ST (ST)
207import qualified Data.Text.Array as A
214import qualified Data.Text.Array as A
208import qualified Data.List as L
215import qualified Data.List as L
216+import Data.Binary (Binary(get, put))
209import Data.Monoid (Monoid(..))
217import Data.Monoid (Monoid(..))
218+#if MIN_VERSION_base(4,9,0)
219+import Data.Semigroup (Semigroup(..))
210import Data.String (IsString(..))
221import Data.String (IsString(..))
211−import qualified Data.Text.Fusion as S
222+import qualified Data.Text.Internal.Fusion as S
212−import qualified Data.Text.Fusion.Common as S
223+import qualified Data.Text.Internal.Fusion.Common as S
224+import Data.Text.Encoding (decodeUtf8', encodeUtf8)
213−import Data.Text.Fusion (stream, reverseStream, unstream)
225+import Data.Text.Internal.Fusion (stream, reverseStream, unstream)
214−import Data.Text.Private (span_)
226+import Data.Text.Internal.Private (span_)
215−import Data.Text.Internal (Text(..), empty, firstf, safe, text, textP)
227+import Data.Text.Internal (Text(..), empty, firstf, mul, safe, text)
228+import Data.Text.Show (singleton, unpack, unpackCString#)
216import qualified Prelude as P
229import qualified Prelude as P
217import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,
230import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,
218− unsafeHead, unsafeTail)
231+ reverseIter_, unsafeHead, unsafeTail)
219−import Data.Text.UnsafeChar (unsafeChr)
232+import Data.Text.Internal.Unsafe.Char (unsafeChr)
220−import qualified Data.Text.Util as U
233+import qualified Data.Text.Internal.Functions as F
221−import qualified Data.Text.Encoding.Utf16 as U16
234+import qualified Data.Text.Internal.Encoding.Utf16 as U16
222−import Data.Text.Search (indices)
235+import Data.Text.Internal.Search (indices)
223#if defined(__HADDOCK__)
236#if defined(__HADDOCK__)
224import Data.ByteString (ByteString)
237import Data.ByteString (ByteString)
225import qualified Data.Text.Lazy as L
238import qualified Data.Text.Lazy as L
226import Data.Int (Int64)
239import Data.Int (Int64)
241+import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)
228−#if __GLASGOW_HASKELL__ >= 702
242+#if __GLASGOW_HASKELL__ >= 708
229−import qualified GHC.CString as GHC
243+import qualified GHC.Exts as Exts
231−import qualified GHC.Base as GHC
245+#if MIN_VERSION_base(4,7,0)
246+import Text.Printf (PrintfArg, formatArg, formatString)
233−import GHC.Prim (Addr#)
235-- $strict
249-- $strict
237-- This package provides both strict and lazy 'Text' types. The
251-- This package provides both strict and lazy 'Text' types. The
238−-- strict type is provided by the 'Data.Text' package, while the lazy
252+-- strict type is provided by the "Data.Text" module, while the lazy
239−-- type is provided by the 'Data.Text.Lazy' package. Internally, the
253+-- type is provided by the "Data.Text.Lazy" module. Internally, the
240-- lazy @Text@ type consists of a list of strict chunks.
254-- lazy @Text@ type consists of a list of strict chunks.
242-- The strict 'Text' type requires that an entire string fit into
256-- The strict 'Text' type requires that an entire string fit into
243−-- memory at once. The lazy @Text@ type is capable of streaming
257+-- memory at once. The lazy 'Data.Text.Lazy.Text' type is capable of
244−-- strings that are larger than memory using a small memory footprint.
258+-- streaming strings that are larger than memory using a small memory
245−-- In many cases, the overhead of chunked streaming makes the lazy
259+-- footprint. In many cases, the overhead of chunked streaming makes
246−-- @Text@ type slower than its strict counterpart, but this is not
260+-- the lazy 'Data.Text.Lazy.Text' type slower than its strict
247−-- always the case. Sometimes, the time complexity of a function in
261+-- counterpart, but this is not always the case. Sometimes, the time
248−-- one module may be different from the other, due to their differing
262+-- complexity of a function in one module may be different from the
249−-- internal structures.
263+-- other, due to their differing internal structures.
251-- Each module provides an almost identical API, with the main
265-- Each module provides an almost identical API, with the main
252-- difference being that the strict module uses 'Int' values for
266-- difference being that the strict module uses 'Int' values for
253−-- lengths and counts, while the lazy module uses 'Int64' lengths.
267+-- lengths and counts, while the lazy module uses 'Data.Int.Int64'
255-- $replacement
270-- $replacement
257-- A 'Text' value is a sequence of Unicode scalar values, as defined
272-- A 'Text' value is a sequence of Unicode scalar values, as defined
258−-- in §3.9, definition D76 of the Unicode 5.2 standard:
274+-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35 §3.9, definition D76 of the Unicode 5.2 standard >.
259−-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35>. As
260−-- such, a 'Text' cannot contain values in the range U+D800 to U+DFFF
275+-- As such, a 'Text' cannot contain values in the range U+D800 to
261−-- inclusive. Haskell implementations admit all Unicode code points
276+-- U+DFFF inclusive. Haskell implementations admit all Unicode code
278+-- (<http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=13 §3.4, definition D10 >)
262−-- (§3.4, definition D10) as 'Char' values, including code points
279+-- as 'Char' values, including code points from this invalid range.
263−-- from this invalid range. This means that there are some 'Char'
280+-- This means that there are some 'Char' values that are not valid
264−-- values that are not valid Unicode scalar values, and the functions
281+-- Unicode scalar values, and the functions in this module must handle
265−-- in this module must handle those cases.
282+-- those cases.
267-- Within this module, many functions construct a 'Text' from one or
284-- Within this module, many functions construct a 'Text' from one or
268-- more 'Char' values. Those functions will substitute 'Char' values
285-- more 'Char' values. Those functions will substitute 'Char' values
@@ -276,8 +293,8 @@ import GHC.Prim (Addr#)
276-- range U+D800 through U+DFFF are used by UTF-16 to denote surrogate
293-- range U+D800 through U+DFFF are used by UTF-16 to denote surrogate
277-- code points, and so cannot be represented. The functions replace
294-- code points, and so cannot be represented. The functions replace
278-- invalid scalar values, instead of dropping them, as a security
295-- invalid scalar values, instead of dropping them, as a security
279−-- measure. For details, see Unicode Technical Report 36, §3.5:
296+-- measure. For details, see
280−-- <http://unicode.org/reports/tr36/#Deletion_of_Noncharacters>)
297+-- <http://unicode.org/reports/tr36/#Deletion_of_Noncharacters Unicode Technical Report 36, §3.5 >.)
282-- $fusion
299-- $fusion
@@ -295,11 +312,11 @@ import GHC.Prim (Addr#)
295-- > countChars = T.length . T.toUpper . E.decodeUtf8
312-- > countChars = T.length . T.toUpper . E.decodeUtf8
297-- From the type signatures involved, this looks like it should
314-- From the type signatures involved, this looks like it should
298−-- allocate one 'ByteString' value, and two 'Text' values. However,
315+-- allocate one 'Data.ByteString.ByteString' value, and two 'Text'
299−-- when a module is compiled with optimisation enabled under GHC, the
316+-- values. However, when a module is compiled with optimisation
300−-- two intermediate 'Text' values will be optimised away, and the
317+-- enabled under GHC, the two intermediate 'Text' values will be
301−-- function will be compiled down to a single loop over the source
318+-- optimised away, and the function will be compiled down to a single
302−-- 'ByteString'.
319+-- loop over the source 'Data.ByteString.ByteString'.
304-- Functions that can be fused by the compiler are documented with the
321-- Functions that can be fused by the compiler are documented with the
305-- phrase \"Subject to fusion\".
322-- phrase \"Subject to fusion\".
@@ -313,41 +330,82 @@ instance Eq Text where
313instance Ord Text where
330instance Ord Text where
314 compare = compareText
331 compare = compareText
316−instance Show Text where
317− showsPrec p ps r = showsPrec p (unpack ps) r
319instance Read Text where
333instance Read Text where
320 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
334 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
336+#if MIN_VERSION_base(4,9,0)
337+-- Semigroup orphan instances for older GHCs are provided by
338+-- 'semigroups` package
340+instance Semigroup Text where
322instance Monoid Text where
344instance Monoid Text where
323 mempty = empty
345 mempty = empty
346+#if MIN_VERSION_base(4,9,0)
347+ mappend = (<>) -- future-proof definition
324 mappend = append
349 mappend = append
325 mconcat = concat
351 mconcat = concat
327instance IsString Text where
353instance IsString Text where
328 fromString = pack
354 fromString = pack
356+#if __GLASGOW_HASKELL__ >= 708
357+instance Exts.IsList Text where
358+ type Item Text = Char
330#if defined(HAVE_DEEPSEQ)
363#if defined(HAVE_DEEPSEQ)
331−instance NFData Text
364+instance NFData Text where rnf !_ = ()
367+instance Binary Text where
368+ put t = put (encodeUtf8 t)
371+ case decodeUtf8' bs of
372+ P.Left exn -> P.fail (P.show exn)
373+ P.Right a -> P.return a
334−-- This instance preserves data abstraction at the cost of inefficiency.
375+-- | This instance preserves data abstraction at the cost of inefficiency.
335-- We omit reflection services for the sake of data abstraction.
376-- We omit reflection services for the sake of data abstraction.
337−-- This instance was created by copying the behavior of Data.Set and
378+-- This instance was created by copying the updated behavior of
379+-- @"Data.Set".@'Data.Set.Set' and @"Data.Map".@'Data.Map.Map'. If you
338−-- Data.Map. If you feel a mistake has been made, please feel free to
380+-- feel a mistake has been made, please feel free to submit
339−-- submit improvements.
381+-- improvements.
341−-- Original discussion is archived here:
383+-- The original discussion is archived here:
343−-- "could we get a Data instance for Data.Text.Text?"
344−-- http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93
384+-- <http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93 could we get a Data instance for Data.Text.Text? >
386+-- The followup discussion that changed the behavior of 'Data.Set.Set'
387+-- and 'Data.Map.Map' is archived here:
388+-- <http://markmail.org/message/trovdc6zkphyi3cr#query:+page:1+mid:a46der3iacwjcf6n+state:results Proposal: Allow gunfold for Data.Map, ... >
346instance Data Text where
390instance Data Text where
347 gfoldl f z txt = z pack `f` (unpack txt)
391 gfoldl f z txt = z pack `f` (unpack txt)
348− toConstr _ = P.error "Data.Text.Text.toConstr"
392+ toConstr _ = packConstr
393+ gunfold k z c = case constrIndex c of
349− gunfold _ _ = P.error "Data.Text.Text.gunfold"
395+ _ -> P.error "gunfold"
350− dataTypeOf _ = mkNoRepType "Data.Text.Text"
396+ dataTypeOf _ = textDataType
398+#if MIN_VERSION_base(4,7,0)
399+-- | Only defined for @base-4.7.0.0@ and later
400+instance PrintfArg Text where
401+ formatArg txt = formatString $ unpack txt
405+packConstr = mkConstr textDataType "pack" [] Prefix
407+textDataType :: DataType
408+textDataType = mkDataType "Data.Text.Text" [packConstr]
352-- | /O(n)/ Compare two 'Text' values lexicographically.
410-- | /O(n)/ Compare two 'Text' values lexicographically.
353compareText :: Text -> Text -> Ordering
411compareText :: Text -> Text -> Ordering
@@ -372,30 +430,6 @@ pack :: String -> Text
372pack = unstream . S.map safe . S.streamList
430pack = unstream . S.map safe . S.streamList
373{-# INLINE [1] pack #-}
431{-# INLINE [1] pack #-}
375−-- | /O(n)/ Convert a Text into a String. Subject to fusion.
376−unpack :: Text -> String
377−unpack = S.unstreamList . stream
378−{-# INLINE [1] unpack #-}
380−-- | /O(n)/ Convert a literal string into a Text. Subject to fusion.
381−unpackCString# :: Addr# -> Text
382−unpackCString# addr# = unstream (S.streamCString# addr#)
383−{-# NOINLINE unpackCString# #-}
385−{-# RULES "TEXT literal" forall a.
386− unstream (S.map safe (S.streamList (GHC.unpackCString# a)))
387− = unpackCString# a #-}
389−{-# RULES "TEXT literal UTF8" forall a.
390− unstream (S.map safe (S.streamList (GHC.unpackCStringUtf8# a)))
391− = unpackCString# a #-}
393−-- | /O(1)/ Convert a character into a Text. Subject to fusion.
394−-- Performs replacement on invalid scalar values.
395−singleton :: Char -> Text
396−singleton = unstream . S.singleton . safe
397−{-# INLINE [1] singleton #-}
399-- -----------------------------------------------------------------------------
433-- -----------------------------------------------------------------------------
400-- * Basic functions
434-- * Basic functions
@@ -426,12 +460,13 @@ append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)
426 | otherwise = overflowError "append"
460 | otherwise = overflowError "append"
428 len = len1+len2
462 len = len1+len2
463+ x :: ST s (A.MArray s)
430 arr <- A.new len
465 arr <- A.new len
431 A.copyI arr 0 arr1 off1 len1
466 A.copyI arr 0 arr1 off1 len1
432 A.copyI arr len1 arr2 off2 len
467 A.copyI arr len1 arr2 off2 len
433 return arr
468 return arr
434−{-# INLINE append #-}
469+{-# NOINLINE append #-}
437"TEXT append -> fused" [~1] forall t1 t2.
472"TEXT append -> fused" [~1] forall t1 t2.
@@ -451,8 +486,8 @@ head t = S.head (stream t)
451uncons :: Text -> Maybe (Char, Text)
486uncons :: Text -> Maybe (Char, Text)
452uncons t@(Text arr off len)
487uncons t@(Text arr off len)
453 | len <= 0 = Nothing
488 | len <= 0 = Nothing
489+ | otherwise = Just $ let !(Iter c d) = iter t 0
454− | otherwise = Just (c, textP arr (off+d) (len-d))
490+ in (c, text arr (off+d) (len-d))
455− where Iter c d = iter t 0
456{-# INLINE [1] uncons #-}
491{-# INLINE [1] uncons #-}
458-- | Lifted from Control.Arrow and specialized.
493-- | Lifted from Control.Arrow and specialized.
@@ -482,7 +517,7 @@ last (Text arr off len)
482tail :: Text -> Text
517tail :: Text -> Text
483tail t@(Text arr off len)
518tail t@(Text arr off len)
484 | len <= 0 = emptyError "tail"
519 | len <= 0 = emptyError "tail"
485− | otherwise = textP arr (off+d) (len-d)
520+ | otherwise = text arr (off+d) (len-d)
486 where d = iter_ t 0
521 where d = iter_ t 0
487{-# INLINE [1] tail #-}
522{-# INLINE [1] tail #-}
@@ -497,8 +532,8 @@ tail t@(Text arr off len)
497-- be non-empty. Subject to fusion.
532-- be non-empty. Subject to fusion.
498init :: Text -> Text
533init :: Text -> Text
499init (Text arr off len) | len <= 0 = emptyError "init"
534init (Text arr off len) | len <= 0 = emptyError "init"
500− | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)
535+ | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)
501− | otherwise = textP arr off (len-1)
536+ | otherwise = text arr off (len-1)
503 n = A.unsafeIndex arr (off+len-1)
538 n = A.unsafeIndex arr (off+len-1)
504{-# INLINE [1] init #-}
539{-# INLINE [1] init #-}
@@ -537,7 +572,9 @@ isSingleton = S.isSingleton . stream
537-- Subject to fusion.
572-- Subject to fusion.
538length :: Text -> Int
573length :: Text -> Int
539length t = S.length (stream t)
574length t = S.length (stream t)
540−{-# INLINE length #-}
575+{-# INLINE [0] length #-}
576+-- length needs to be phased after the compareN/length rules otherwise
577+-- it may inline before the rules have an opportunity to fire.
542-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
579-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
543-- Subject to fusion.
580-- Subject to fusion.
@@ -556,32 +593,32 @@ compareLength t n = S.compareLengthI (stream t) n
558"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
595"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
559− (==) (length t) n = compareLength t n == EQ
596+ eqInt (length t) n = compareLength t n == EQ
563"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
600"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
564− (/=) (length t) n = compareLength t n /= EQ
601+ neInt (length t) n = compareLength t n /= EQ
568"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
605"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
569− (<) (length t) n = compareLength t n == LT
606+ ltInt (length t) n = compareLength t n == LT
573"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
610"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
574− (<=) (length t) n = compareLength t n /= GT
611+ leInt (length t) n = compareLength t n /= GT
578"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
615"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
579− (>) (length t) n = compareLength t n == GT
616+ gtInt (length t) n = compareLength t n == GT
583"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
620"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
584− (>=) (length t) n = compareLength t n /= LT
621+ geInt (length t) n = compareLength t n /= LT
587-- -----------------------------------------------------------------------------
624-- -----------------------------------------------------------------------------
@@ -597,7 +634,7 @@ map f t = unstream (S.map (safe . f) (stream t))
597-- 'Text's and concatenates the list after interspersing the first
634-- 'Text's and concatenates the list after interspersing the first
598-- argument between each element of the list.
635-- argument between each element of the list.
599intercalate :: Text -> [Text] -> Text
636intercalate :: Text -> [Text] -> Text
600−intercalate t = concat . (U.intersperse t)
637+intercalate t = concat . (F.intersperse t)
601{-# INLINE intercalate #-}
638{-# INLINE intercalate #-}
603-- | /O(n)/ The 'intersperse' function takes a character and places it
640-- | /O(n)/ The 'intersperse' function takes a character and places it
@@ -612,22 +649,48 @@ reverse :: Text -> Text
612reverse t = S.reverse (stream t)
649reverse t = S.reverse (stream t)
613{-# INLINE reverse #-}
650{-# INLINE reverse #-}
615−-- | /O(m+n)/ Replace every occurrence of one substring with another.
652+-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in
653+-- @haystack@ with @replacement@.
655+-- This function behaves as though it was defined as follows:
658+-- replace needle replacement haystack =
659+-- 'intercalate' replacement ('splitOn' needle haystack)
662+-- As this suggests, each occurrence is replaced exactly once. So if
663+-- @needle@ occurs in @replacement@, that occurrence will /not/ itself
664+-- be replaced recursively:
666+-- > replace "oo" "foo" "oo" == "foo"
668+-- In cases where several instances of @needle@ overlap, only the
669+-- first one will be replaced:
671+-- > replace "ofo" "bar" "ofofo" == "barfo"
617-- In (unlikely) bad cases, this function's time complexity degrades
673-- In (unlikely) bad cases, this function's time complexity degrades
618-- towards /O(n*m)/.
674-- towards /O(n*m)/.
619−replace :: Text -> Text -> Text -> Text
675+replace :: Text
676+ -- ^ @needle@ to search for. If this string is empty, an
677+ -- error will occur.
679+ -- ^ @replacement@ to replace @needle@ with.
681+ -- ^ @haystack@ in which to search.
620replace needle@(Text _ _ neeLen)
683replace needle@(Text _ _ neeLen)
621 (Text repArr repOff repLen)
684 (Text repArr repOff repLen)
622 haystack@(Text hayArr hayOff hayLen)
685 haystack@(Text hayArr hayOff hayLen)
623 | neeLen == 0 = emptyError "replace"
686 | neeLen == 0 = emptyError "replace"
624 | L.null ixs = haystack
687 | L.null ixs = haystack
625 | len > 0 = Text (A.run x) 0 len
688 | len > 0 = Text (A.run x) 0 len
626− | len < 0 = overflowError "replace"
627 | otherwise = empty
689 | otherwise = empty
629 ixs = indices needle haystack
691 ixs = indices needle haystack
630− len = hayLen - (neeLen - repLen) * L.length ixs
692+ len = hayLen - (neeLen - repLen) `mul` L.length ixs
693+ x :: ST s (A.MArray s)
632 marr <- A.new len
695 marr <- A.new len
633 let loop (i:is) o d = do
696 let loop (i:is) o d = do
@@ -657,9 +720,8 @@ replace needle@(Text _ _ neeLen)
657-- /Note/: In some languages, case conversion is a locale- and
720-- /Note/: In some languages, case conversion is a locale- and
658-- context-dependent operation. The case conversion functions in this
721-- context-dependent operation. The case conversion functions in this
659-- module are /not/ locale sensitive. Programs that require locale
722-- module are /not/ locale sensitive. Programs that require locale
660−-- sensitivity should use appropriate versions of the case mapping
723+-- sensitivity should use appropriate versions of the
661−-- functions from the @text-icu@ package:
662−-- <http://hackage.haskell.org/package/text-icu>
724+-- <http://hackage.haskell.org/package/text-icu-0.6.3.7/docs/Data-Text-ICU.html#g:4 case mapping functions from the text-icu package >.
664-- | /O(n)/ Convert a string to folded case. Subject to fusion.
726-- | /O(n)/ Convert a string to folded case. Subject to fusion.
@@ -679,7 +741,7 @@ replace needle@(Text _ _ neeLen)
679-- instead of itself.
741-- instead of itself.
680toCaseFold :: Text -> Text
742toCaseFold :: Text -> Text
681toCaseFold t = unstream (S.toCaseFold (stream t))
743toCaseFold t = unstream (S.toCaseFold (stream t))
682−{-# INLINE [0] toCaseFold #-}
744+{-# INLINE toCaseFold #-}
684-- | /O(n)/ Convert a string to lower case, using simple case
746-- | /O(n)/ Convert a string to lower case, using simple case
685-- conversion. Subject to fusion.
747-- conversion. Subject to fusion.
@@ -839,6 +901,7 @@ concat ts = case ts' of
840 ts' = L.filter (not . null) ts
902 ts' = L.filter (not . null) ts
841 len = sumP "concat" $ L.map lengthWord16 ts'
903 len = sumP "concat" $ L.map lengthWord16 ts'
904+ go :: ST s (A.MArray s)
843 arr <- A.new len
906 arr <- A.new len
844 let step i (Text a o l) =
907 let step i (Text a o l) =
@@ -852,13 +915,13 @@ concatMap f = concat . foldr ((:) . f) []
852{-# INLINE concatMap #-}
915{-# INLINE concatMap #-}
854-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
917-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
855−-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.
918+-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.
856any :: (Char -> Bool) -> Text -> Bool
919any :: (Char -> Bool) -> Text -> Bool
857any p t = S.any p (stream t)
920any p t = S.any p (stream t)
858{-# INLINE any #-}
921{-# INLINE any #-}
860-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
923-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
861−-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.
924+-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.
862all :: (Char -> Bool) -> Text -> Bool
925all :: (Char -> Bool) -> Text -> Bool
863all p t = S.all p (stream t)
926all p t = S.all p (stream t)
864{-# INLINE all #-}
927{-# INLINE all #-}
@@ -949,10 +1012,10 @@ replicate n t@(Text a o l)
949 | n <= 0 || l <= 0 = empty
1012 | n <= 0 || l <= 0 = empty
950 | n == 1 = t
1013 | n == 1 = t
951 | isSingleton t = replicateChar n (unsafeHead t)
1014 | isSingleton t = replicateChar n (unsafeHead t)
952− | n <= maxBound `quot` l = Text (A.run x) 0 len
1015+ | otherwise = Text (A.run x) 0 len
953− | otherwise = overflowError "replicate"
955− len = l * n
1017+ len = l `mul` n
1018+ x :: ST s (A.MArray s)
957 arr <- A.new len
1020 arr <- A.new len
958 let loop !d !i | i >= n = return arr
1021 let loop !d !i | i >= n = return arr
@@ -1003,14 +1066,16 @@ take :: Int -> Text -> Text
1003take n t@(Text arr off len)
1066take n t@(Text arr off len)
1004 | n <= 0 = empty
1067 | n <= 0 = empty
1005 | n >= len = t
1068 | n >= len = t
1006− | otherwise = Text arr off (loop 0 0)
1069+ | otherwise = text arr off (iterN n t)
1009− | i >= len || cnt >= n = i
1010− | otherwise = loop (i+d) (cnt+1)
1011− where d = iter_ t i
1012{-# INLINE [1] take #-}
1070{-# INLINE [1] take #-}
1072+iterN :: Int -> Text -> Int
1073+iterN n t@(Text _arr _off len) = loop 0 0
1075+ | i >= len || cnt >= n = i
1076+ | otherwise = loop (i+d) (cnt+1)
1077+ where d = iter_ t i
1014{-# RULES
1079{-# RULES
1015"TEXT take -> fused" [~1] forall n t.
1080"TEXT take -> fused" [~1] forall n t.
1016 take n t = unstream (S.take n (stream t))
1081 take n t = unstream (S.take n (stream t))
@@ -1018,6 +1083,27 @@ take n t@(Text arr off len)
1018 unstream (S.take n (stream t)) = take n t
1083 unstream (S.take n (stream t)) = take n t
1086+-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after
1087+-- taking @n@ characters from the end of @t@.
1091+-- > takeEnd 3 "foobar" == "bar"
1092+takeEnd :: Int -> Text -> Text
1093+takeEnd n t@(Text arr off len)
1096+ | otherwise = text arr (off+i) (len-i)
1097+ where i = iterNEnd n t
1099+iterNEnd :: Int -> Text -> Int
1100+iterNEnd n t@(Text _arr _off len) = loop (len-1) n
1104+ | otherwise = loop (i+d) (m-1)
1105+ where d = reverseIter_ t i
1021-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
1107-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
1022-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
1108-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
1023-- is greater than the length of the 'Text'. Subject to fusion.
1109-- is greater than the length of the 'Text'. Subject to fusion.
@@ -1025,11 +1111,8 @@ drop :: Int -> Text -> Text
1025drop n t@(Text arr off len)
1111drop n t@(Text arr off len)
1026 | n <= 0 = t
1112 | n <= 0 = t
1027 | n >= len = empty
1113 | n >= len = empty
1028− | otherwise = loop 0 0
1114+ | otherwise = text arr (off+i) (len-i)
1030− | i >= len || cnt >= n = Text arr (off+i) (len-i)
1031− | otherwise = loop (i+d) (cnt+1)
1032− where d = iter_ t i
1115+ where i = iterN n t
1033{-# INLINE [1] drop #-}
1116{-# INLINE [1] drop #-}
1035{-# RULES
1118{-# RULES
@@ -1039,6 +1122,18 @@ drop n t@(Text arr off len)
1039 unstream (S.drop n (stream t)) = drop n t
1122 unstream (S.drop n (stream t)) = drop n t
1125+-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after
1126+-- dropping @n@ characters from the end of @t@.
1130+-- > dropEnd 3 "foobar" == "foo"
1131+dropEnd :: Int -> Text -> Text
1132+dropEnd n t@(Text arr off len)
1135+ | otherwise = text arr off (iterNEnd n t)
1042-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
1137-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
1043-- returns the longest prefix (possibly empty) of elements that
1138-- returns the longest prefix (possibly empty) of elements that
1044-- satisfy @p@. Subject to fusion.
1139-- satisfy @p@. Subject to fusion.
@@ -1046,7 +1141,7 @@ takeWhile :: (Char -> Bool) -> Text -> Text
1046takeWhile p t@(Text arr off len) = loop 0
1141takeWhile p t@(Text arr off len) = loop 0
1047 where loop !i | i >= len = t
1142 where loop !i | i >= len = t
1048 | p c = loop (i+d)
1143 | p c = loop (i+d)
1049− | otherwise = textP arr off i
1144+ | otherwise = text arr off i
1050 where Iter c d = iter t i
1145 where Iter c d = iter t i
1051{-# INLINE [1] takeWhile #-}
1146{-# INLINE [1] takeWhile #-}
@@ -1057,6 +1152,27 @@ takeWhile p t@(Text arr off len) = loop 0
1057 unstream (S.takeWhile p (stream t)) = takeWhile p t
1152 unstream (S.takeWhile p (stream t)) = takeWhile p t
1155+-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',
1156+-- returns the longest suffix (possibly empty) of elements that
1157+-- satisfy @p@. Subject to fusion.
1160+-- > takeWhileEnd (=='o') "foo" == "oo"
1161+takeWhileEnd :: (Char -> Bool) -> Text -> Text
1162+takeWhileEnd p t@(Text arr off len) = loop (len-1) len
1163+ where loop !i !l | l <= 0 = t
1164+ | p c = loop (i+d) (l+d)
1165+ | otherwise = text arr (off+l) (len-l)
1166+ where (c,d) = reverseIter t i
1167+{-# INLINE [1] takeWhileEnd #-}
1170+"TEXT takeWhileEnd -> fused" [~1] forall p t.
1171+ takeWhileEnd p t = S.reverse (S.takeWhile p (S.reverseStream t))
1172+"TEXT takeWhileEnd -> unfused" [1] forall p t.
1173+ S.reverse (S.takeWhile p (S.reverseStream t)) = takeWhileEnd p t
1060-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
1176-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
1061-- 'takeWhile' @p@ @t@. Subject to fusion.
1177-- 'takeWhile' @p@ @t@. Subject to fusion.
1062dropWhile :: (Char -> Bool) -> Text -> Text
1178dropWhile :: (Char -> Bool) -> Text -> Text
@@ -1075,8 +1191,9 @@ dropWhile p t@(Text arr off len) = loop 0 0
1077-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
1193-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
1078−-- dropping characters that fail the predicate @p@ from the end of
1194+-- dropping characters that satisfy the predicate @p@ from the end of
1079-- @t@. Subject to fusion.
1195-- @t@. Subject to fusion.
1080-- Examples:
1197-- Examples:
1082-- > dropWhileEnd (=='.') "foo..." == "foo"
1199-- > dropWhileEnd (=='.') "foo..." == "foo"
@@ -1096,7 +1213,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len
1098-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
1215-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
1099−-- dropping characters that fail the predicate @p@ from both the
1216+-- dropping characters that satisfy the predicate @p@ from both the
1100-- beginning and end of @t@. Subject to fusion.
1217-- beginning and end of @t@. Subject to fusion.
1101dropAround :: (Char -> Bool) -> Text -> Text
1218dropAround :: (Char -> Bool) -> Text -> Text
1102dropAround p = dropWhile p . dropWhileEnd p
1219dropAround p = dropWhile p . dropWhileEnd p
@@ -1131,13 +1248,8 @@ splitAt :: Int -> Text -> (Text, Text)
1131splitAt n t@(Text arr off len)
1248splitAt n t@(Text arr off len)
1132 | n <= 0 = (empty, t)
1249 | n <= 0 = (empty, t)
1133 | n >= len = (t, empty)
1250 | n >= len = (t, empty)
1134− | otherwise = (Text arr off k, Text arr (off+k) (len-k))
1251+ | otherwise = let k = iterN n t
1135− where k = loop 0 0
1252+ in (text arr off k, text arr (off+k) (len-k))
1137− | i >= len || cnt >= n = i
1138− | otherwise = loop (i+d) (cnt+1)
1139− where d = iter_ t i
1140−{-# INLINE splitAt #-}
1142-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
1254-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
1143-- a pair whose first element is the longest prefix (possibly empty)
1255-- a pair whose first element is the longest prefix (possibly empty)
@@ -1196,9 +1308,9 @@ tails t | null t = [empty]
1196-- copies to create substrings; they just construct new 'Text's that
1308-- copies to create substrings; they just construct new 'Text's that
1197-- are slices of the original.
1309-- are slices of the original.
1199−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first
1311+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'
1200−-- 'Text' argument, consuming the delimiter. An empty delimiter is
1312+-- argument (which cannot be empty), consuming the delimiter. An empty
1201−-- invalid, and will cause an error to be raised.
1313+-- delimiter is invalid, and will cause an error to be raised.
1203-- Examples:
1315-- Examples:
@@ -1211,16 +1323,23 @@ tails t | null t = [empty]
1211-- > intercalate s . splitOn s == id
1323-- > intercalate s . splitOn s == id
1212-- > splitOn (singleton c) == split (==c)
1324-- > splitOn (singleton c) == split (==c)
1326+-- (Note: the string @s@ to split on above cannot be empty.)
1214-- In (unlikely) bad cases, this function's time complexity degrades
1328-- In (unlikely) bad cases, this function's time complexity degrades
1215-- towards /O(n*m)/.
1329-- towards /O(n*m)/.
1216−splitOn :: Text -> Text -> [Text]
1330+splitOn :: Text
1331+ -- ^ String to split on. If this string is empty, an error
1217splitOn pat@(Text _ _ l) src@(Text arr off len)
1336splitOn pat@(Text _ _ l) src@(Text arr off len)
1218 | l <= 0 = emptyError "splitOn"
1337 | l <= 0 = emptyError "splitOn"
1219 | isSingleton pat = split (== unsafeHead pat) src
1338 | isSingleton pat = split (== unsafeHead pat) src
1220 | otherwise = go 0 (indices pat src)
1339 | otherwise = go 0 (indices pat src)
1222− go !s (x:xs) = textP arr (s+off) (x-s) : go (x+l) xs
1341+ go !s (x:xs) = text arr (s+off) (x-s) : go (x+l) xs
1223− go s _ = [textP arr (s+off) (len-s)]
1342+ go s _ = [text arr (s+off) (len-s)]
1224{-# INLINE [1] splitOn #-}
1343{-# INLINE [1] splitOn #-}
1226{-# RULES
1345{-# RULES
@@ -1312,7 +1431,7 @@ breakOn pat src@(Text arr off len)
1312 | null pat = emptyError "breakOn"
1431 | null pat = emptyError "breakOn"
1313 | otherwise = case indices pat src of
1432 | otherwise = case indices pat src of
1314 [] -> (src, empty)
1433 [] -> (src, empty)
1315− (x:_) -> (textP arr off x, textP arr (off+x) (len-x))
1434+ (x:_) -> (text arr off x, text arr (off+x) (len-x))
1316{-# INLINE breakOn #-}
1435{-# INLINE breakOn #-}
1318-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
1437-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
@@ -1354,7 +1473,7 @@ breakOnAll pat src@(Text arr off slen)
1354 | otherwise = L.map step (indices pat src)
1473 | otherwise = L.map step (indices pat src)
1356 step x = (chunk 0 x, chunk x (slen-x))
1475 step x = (chunk 0 x, chunk x (slen-x))
1357− chunk !n !l = textP arr (n+off) l
1476+ chunk !n !l = text arr (n+off) l
1358{-# INLINE breakOnAll #-}
1477{-# INLINE breakOnAll #-}
1360-------------------------------------------------------------------------------
1479-------------------------------------------------------------------------------
@@ -1421,7 +1540,7 @@ countChar c t = S.countChar c (stream t)
1421-- equivalent to a pair of 'unpack' operations.
1540-- equivalent to a pair of 'unpack' operations.
1422zip :: Text -> Text -> [(Char,Char)]
1541zip :: Text -> Text -> [(Char,Char)]
1423zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1542zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1424−{-# INLINE [0] zip #-}
1543+{-# INLINE zip #-}
1426-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
1545-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
1427-- given as the first argument, instead of a tupling function.
1546-- given as the first argument, instead of a tupling function.
@@ -1429,7 +1548,7 @@ zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1429zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1548zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1430zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))
1549zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))
1431 where g a b = safe (f a b)
1550 where g a b = safe (f a b)
1432−{-# INLINE [0] zipWith #-}
1551+{-# INLINE zipWith #-}
1434-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
1553-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
1435-- representing white space.
1554-- representing white space.
@@ -1554,7 +1673,7 @@ isInfixOf needle haystack
1554-- > fnordLength _ = -1
1673-- > fnordLength _ = -1
1555stripPrefix :: Text -> Text -> Maybe Text
1674stripPrefix :: Text -> Text -> Maybe Text
1556stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1675stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1557− | p `isPrefixOf` t = Just $! textP arr (off+plen) (len-plen)
1676+ | p `isPrefixOf` t = Just $! text arr (off+plen) (len-plen)
1558 | otherwise = Nothing
1677 | otherwise = Nothing
1560-- | /O(n)/ Find the longest non-empty common prefix of two strings
1679-- | /O(n)/ Find the longest non-empty common prefix of two strings
@@ -1574,8 +1693,8 @@ commonPrefixes t0@(Text arr0 off0 len0) t1@(Text arr1 off1 len1) = go 0 0
1575 go !i !j | i < len0 && j < len1 && a == b = go (i+d0) (j+d1)
1694 go !i !j | i < len0 && j < len1 && a == b = go (i+d0) (j+d1)
1576 | i > 0 = Just (Text arr0 off0 i,
1695 | i > 0 = Just (Text arr0 off0 i,
1577− textP arr0 (off0+i) (len0-i),
1696+ text arr0 (off0+i) (len0-i),
1578− textP arr1 (off1+j) (len1-j))
1697+ text arr1 (off1+j) (len1-j))
1579 | otherwise = Nothing
1698 | otherwise = Nothing
1580 where Iter a d0 = iter t0 i
1699 where Iter a d0 = iter t0 i
1581 Iter b d1 = iter t1 j
1700 Iter b d1 = iter t1 j
@@ -1600,7 +1719,7 @@ commonPrefixes t0@(Text arr0 off0 len0) t1@(Text arr1 off1 len1) = go 0 0
1600-- > quuxLength _ = -1
1719-- > quuxLength _ = -1
1601stripSuffix :: Text -> Text -> Maybe Text
1720stripSuffix :: Text -> Text -> Maybe Text
1602stripSuffix p@(Text _arr _off plen) t@(Text arr off len)
1721stripSuffix p@(Text _arr _off plen) t@(Text arr off len)
1603− | p `isSuffixOf` t = Just $! textP arr off (len-plen)
1722+ | p `isSuffixOf` t = Just $! text arr off (len-plen)
1604 | otherwise = Nothing
1723 | otherwise = Nothing
1606-- | Add a list of non-negative numbers. Errors out on overflow.
1725-- | Add a list of non-negative numbers. Errors out on overflow.
@@ -1629,6 +1748,7 @@ overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"
1629copy :: Text -> Text
1748copy :: Text -> Text
1630copy (Text arr off len) = Text (A.run go) 0 len
1749copy (Text arr off len) = Text (A.run go) 0 len
1751+ go :: ST s (A.MArray s)
1633 marr <- A.new len
1753 marr <- A.new len
1634 A.copyI marr 0 arr off len
1754 A.copyI marr 0 arr off len