@@ -1,48 +1,29 @@
1−{-# LANGUAGE BangPatterns, CPP, MagicHash, Rank2Types, UnboxedTuples #-}
1+{-# LANGUAGE BangPatterns #-}
22{-# OPTIONS_GHC -fno-warn-orphans #-}
3−#if __GLASGOW_HASKELL__ >= 702
4−{-# LANGUAGE Trustworthy #-}
5−#endif
6−#if __GLASGOW_HASKELL__ >= 708
7−{-# LANGUAGE TypeFamilies #-}
8−#endif
93
104-- |
115-- Module : Data.Text
12−-- Copyright : (c) 2009, 2010, 2011, 2012 Bryan O'Sullivan,
13−-- (c) 2009 Duncan Coutts,
14−-- (c) 2008, 2009 Tom Harper
6+-- Copyright : (c) Tom Harper 2008-2009,
7+-- (c) Bryan O'Sullivan 2009,
8+-- (c) Duncan Coutts 2009
159--
1610-- License : BSD-style
17−-- Maintainer : bos@serpentine.com
11+-- Maintainer : bos@serpentine.com, rtharper@aftereternity.co.uk,
12+-- duncan@haskell.org
1813-- Stability : experimental
1914-- Portability : GHC
2015--
21−-- A time and space-efficient implementation of Unicode text.
22−-- Suitable for performance critical use, both in terms of large data
23−-- quantities and high speed.
24−--
25−-- /Note/: Read below the synopsis for important notes on the use of
26−-- this module.
16+-- A time and space-efficient implementation of Unicode text using
17+-- packed Word16 arrays. Suitable for performance critical use, both
18+-- in terms of large data quantities and high speed.
2719--
2820-- This module is intended to be imported @qualified@, to avoid name
2921-- clashes with "Prelude" functions, e.g.
3022--
3123-- > import qualified Data.Text as T
32−--
33−-- To use an extended and very rich family of functions for working
34−-- with Unicode text (including normalization, regular expressions,
35−-- non-standard encodings, text breaking, and locales), see
36−-- <http://hackage.haskell.org/package/text-icu the text-icu package >.
3724
3825module Data.Text
3926 (
40− -- * Strict vs lazy types
41− -- $strict
42−
43− -- * Acceptable data
44− -- $replacement
45−
4627 -- * Fusion
4728 -- $fusion
4829
@@ -66,7 +47,6 @@ module Data.Text
6647 , init
6748 , null
6849 , length
69− , compareLength
7050
7151 -- * Transformations
7252 , map
@@ -81,7 +61,6 @@ module Data.Text
8161 , toCaseFold
8262 , toLower
8363 , toUpper
84− , toTitle
8564
8665 -- ** Justification
8766 , justifyLeft
@@ -118,6 +97,7 @@ module Data.Text
11897
11998 -- ** Generation and unfolding
12099 , replicate
100+ , replicateChar
121101 , unfoldr
122102 , unfoldrN
123103
@@ -125,11 +105,8 @@ module Data.Text
125105
126106 -- ** Breaking strings
127107 , take
128− , takeEnd
129108 , drop
130− , dropEnd
131109 , takeWhile
132− , takeWhileEnd
133110 , dropWhile
134111 , dropWhileEnd
135112 , dropAround
@@ -137,10 +114,8 @@ module Data.Text
137114 , stripStart
138115 , stripEnd
139116 , splitAt
140− , breakOn
141− , breakOnEnd
142− , break
143117 , span
118+ , break
144119 , group
145120 , groupBy
146121 , inits
@@ -148,8 +123,10 @@ module Data.Text
148123
149124 -- ** Breaking into many substrings
150125 -- $split
151− , splitOn
152126 , split
127+ , splitTimes
128+ , splitTimesEnd
129+ , splitWith
153130 , chunksOf
154131
155132 -- ** Breaking into lines and words
@@ -164,309 +141,136 @@ module Data.Text
164141 , isSuffixOf
165142 , isInfixOf
166143
167− -- ** View patterns
168− , stripPrefix
169− , stripSuffix
170− , commonPrefixes
171−
172144 -- * Searching
145+ , elem
173146 , filter
174− , breakOnAll
147+ , findAll
175148 , find
176149 , partition
177150
178151 -- , findSubstring
179−
152+
180153 -- * Indexing
181154 -- $index
182155 , index
183156 , findIndex
157+ , findIndices
158+ , elemIndex
159+ , elemIndices
184160 , count
185161
186− -- * Zipping
162+ -- * Zipping and unzipping
187163 , zip
188164 , zipWith
189165
190166 -- -* Ordered text
191167 -- , sort
192−
193− -- * Low level operations
194− , copy
195− , unpackCString#
196168 ) where
197169
198−import Prelude (Char, Bool(..), Int, Maybe(..), String,
199− Eq(..), Ord(..), Ordering(..), (++),
200− Read(..),
201− (&&), (||), (+), (-), (.), ($), ($!), (>>),
202− not, return, otherwise, quot)
203−#if defined(HAVE_DEEPSEQ)
204−import Control.DeepSeq (NFData(rnf))
205−#endif
206−#if defined(ASSERTS)
170+import Prelude (Char, Bool(..), Functor(..), Int, Maybe(..), String,
171+ Eq(..), Ord(..), (++),
172+ Read(..), Show(..),
173+ (&&), (||), (+), (-), (.), ($),
174+ fromIntegral, div, not, return, otherwise)
207175import Control.Exception (assert)
208−#endif
209176import Data.Char (isSpace)
210−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,
211− Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))
212−import Control.Monad (foldM)
213177import Control.Monad.ST (ST)
214178import qualified Data.Text.Array as A
215179import qualified Data.List as L
216−import Data.Binary (Binary(get, put))
217180import Data.Monoid (Monoid(..))
218−#if MIN_VERSION_base(4,9,0)
219−import Data.Semigroup (Semigroup(..))
220−#endif
181+import Data.Word (Word16)
221182import Data.String (IsString(..))
222−import qualified Data.Text.Internal.Fusion as S
223−import qualified Data.Text.Internal.Fusion.Common as S
224−import Data.Text.Encoding (decodeUtf8', encodeUtf8)
225−import Data.Text.Internal.Fusion (stream, reverseStream, unstream)
226−import Data.Text.Internal.Private (span_)
227−import Data.Text.Internal (Text(..), empty, firstf, mul, safe, text)
228−import Data.Text.Show (singleton, unpack, unpackCString#)
229−import qualified Prelude as P
230−import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,
231− reverseIter_, unsafeHead, unsafeTail)
232−import Data.Text.Internal.Unsafe.Char (unsafeChr)
233−import qualified Data.Text.Internal.Functions as F
234−import qualified Data.Text.Internal.Encoding.Utf16 as U16
235−import Data.Text.Internal.Search (indices)
236−#if defined(__HADDOCK__)
237−import Data.ByteString (ByteString)
238−import qualified Data.Text.Lazy as L
239−import Data.Int (Int64)
240−#endif
241−import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)
242−#if __GLASGOW_HASKELL__ >= 708
243−import qualified GHC.Exts as Exts
244−#endif
245−#if MIN_VERSION_base(4,7,0)
246−import Text.Printf (PrintfArg, formatArg, formatString)
247−#endif
248−
249−-- $strict
250−--
251−-- This package provides both strict and lazy 'Text' types. The
252−-- strict type is provided by the "Data.Text" module, while the lazy
253−-- type is provided by the "Data.Text.Lazy" module. Internally, the
254−-- lazy @Text@ type consists of a list of strict chunks.
255−--
256−-- The strict 'Text' type requires that an entire string fit into
257−-- memory at once. The lazy 'Data.Text.Lazy.Text' type is capable of
258−-- streaming strings that are larger than memory using a small memory
259−-- footprint. In many cases, the overhead of chunked streaming makes
260−-- the lazy 'Data.Text.Lazy.Text' type slower than its strict
261−-- counterpart, but this is not always the case. Sometimes, the time
262−-- complexity of a function in one module may be different from the
263−-- other, due to their differing internal structures.
264−--
265−-- Each module provides an almost identical API, with the main
266−-- difference being that the strict module uses 'Int' values for
267−-- lengths and counts, while the lazy module uses 'Data.Int.Int64'
268−-- lengths.
269183
270−-- $replacement
271−--
272−-- A 'Text' value is a sequence of Unicode scalar values, as defined
273−-- in
274−-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35 §3.9, definition D76 of the Unicode 5.2 standard >.
275−-- As such, a 'Text' cannot contain values in the range U+D800 to
276−-- U+DFFF inclusive. Haskell implementations admit all Unicode code
277−-- points
278−-- (<http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=13 §3.4, definition D10 >)
279−-- as 'Char' values, including code points from this invalid range.
280−-- This means that there are some 'Char' values that are not valid
281−-- Unicode scalar values, and the functions in this module must handle
282−-- those cases.
283−--
284−-- Within this module, many functions construct a 'Text' from one or
285−-- more 'Char' values. Those functions will substitute 'Char' values
286−-- that are not valid Unicode scalar values with the replacement
287−-- character \"�\" (U+FFFD). Functions that perform this
288−-- inspection and replacement are documented with the phrase
289−-- \"Performs replacement on invalid scalar values\".
290−--
291−-- (One reason for this policy of replacement is that internally, a
292−-- 'Text' value is represented as packed UTF-16 data. Values in the
293−-- range U+D800 through U+DFFF are used by UTF-16 to denote surrogate
294−-- code points, and so cannot be represented. The functions replace
295−-- invalid scalar values, instead of dropping them, as a security
296−-- measure. For details, see
297−-- <http://unicode.org/reports/tr36/#Deletion_of_Noncharacters Unicode Technical Report 36, §3.5 >.)
184+import qualified Data.Text.Fusion as S
185+import qualified Data.Text.Fusion.Common as S
186+import Data.Text.Fusion (stream, reverseStream, unstream)
187+
188+import Data.Text.Internal (Text(..), empty, text, textP)
189+import qualified Prelude as P
190+import Data.Text.Unsafe (iter, iter_, reverseIter, unsafeHead, unsafeTail)
191+import Data.Text.UnsafeChar (unsafeChr)
192+import qualified Data.Text.Encoding.Utf16 as U16
298193
299194-- $fusion
300195--
301196-- Most of the functions in this module are subject to /fusion/,
302197-- meaning that a pipeline of such functions will usually allocate at
303198-- most one 'Text' value.
304−--
305−-- As an example, consider the following pipeline:
306−--
307−-- > import Data.Text as T
308−-- > import Data.Text.Encoding as E
309−-- > import Data.ByteString (ByteString)
310−-- >
311−-- > countChars :: ByteString -> Int
312−-- > countChars = T.length . T.toUpper . E.decodeUtf8
313−--
314−-- From the type signatures involved, this looks like it should
315−-- allocate one 'Data.ByteString.ByteString' value, and two 'Text'
316−-- values. However, when a module is compiled with optimisation
317−-- enabled under GHC, the two intermediate 'Text' values will be
318−-- optimised away, and the function will be compiled down to a single
319−-- loop over the source 'Data.ByteString.ByteString'.
320−--
321−-- Functions that can be fused by the compiler are documented with the
322−-- phrase \"Subject to fusion\".
323199
324200instance Eq Text where
325− Text arrA offA lenA == Text arrB offB lenB
326− | lenA == lenB = A.equal arrA offA arrB offB lenA
327− | otherwise = False
201+ t1 == t2 = stream t1 == stream t2
328202 {-# INLINE (==) #-}
329203
330204instance Ord Text where
331− compare = compareText
205+ compare t1 t2 = compare (stream t1) (stream t2)
206+ {-# INLINE compare #-}
207+
208+instance Show Text where
209+ showsPrec p ps r = showsPrec p (unpack ps) r
332210
333211instance Read Text where
334212 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
335213
336−#if MIN_VERSION_base(4,9,0)
337−-- Semigroup orphan instances for older GHCs are provided by
338−-- 'semigroups` package
339−
340−instance Semigroup Text where
341− (<>) = append
342−#endif
343−
344214instance Monoid Text where
345215 mempty = empty
346−#if MIN_VERSION_base(4,9,0)
347− mappend = (<>) -- future-proof definition
348−#else
349216 mappend = append
350−#endif
351217 mconcat = concat
352218
353219instance IsString Text where
354220 fromString = pack
355221
356−#if __GLASGOW_HASKELL__ >= 708
357−instance Exts.IsList Text where
358− type Item Text = Char
359− fromList = pack
360− toList = unpack
361−#endif
362−
363−#if defined(HAVE_DEEPSEQ)
364−instance NFData Text where rnf !_ = ()
365−#endif
366−
367−instance Binary Text where
368− put t = put (encodeUtf8 t)
369− get = do
370− bs <- get
371− case decodeUtf8' bs of
372− P.Left exn -> P.fail (P.show exn)
373− P.Right a -> P.return a
374−
375−-- | This instance preserves data abstraction at the cost of inefficiency.
376−-- We omit reflection services for the sake of data abstraction.
377−--
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
380−-- feel a mistake has been made, please feel free to submit
381−-- improvements.
382−--
383−-- The original discussion is archived here:
384−-- <http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93 could we get a Data instance for Data.Text.Text? >
385−--
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, ... >
389−
390−instance Data Text where
391− gfoldl f z txt = z pack `f` (unpack txt)
392− toConstr _ = packConstr
393− gunfold k z c = case constrIndex c of
394− 1 -> k (z pack)
395− _ -> P.error "gunfold"
396− dataTypeOf _ = textDataType
397−
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
402−#endif
403−
404−packConstr :: Constr
405−packConstr = mkConstr textDataType "pack" [] Prefix
406−
407−textDataType :: DataType
408−textDataType = mkDataType "Data.Text.Text" [packConstr]
409−
410−-- | /O(n)/ Compare two 'Text' values lexicographically.
411−compareText :: Text -> Text -> Ordering
412−compareText ta@(Text _arrA _offA lenA) tb@(Text _arrB _offB lenB)
413− | lenA == 0 && lenB == 0 = EQ
414− | otherwise = go 0 0
415− where
416− go !i !j
417− | i >= lenA || j >= lenB = compare lenA lenB
418− | a < b = LT
419− | a > b = GT
420− | otherwise = go (i+di) (j+dj)
421− where Iter a di = iter ta i
422− Iter b dj = iter tb j
423−
424222-- -----------------------------------------------------------------------------
425223-- * Conversion to/from 'Text'
426224
427−-- | /O(n)/ Convert a 'String' into a 'Text'. Subject to
428−-- fusion. Performs replacement on invalid scalar values.
225+-- | /O(n)/ Convert a 'String' into a 'Text'. Subject to fusion.
429226pack :: String -> Text
430−pack = unstream . S.map safe . S.streamList
227+pack = unstream . S.streamList
431228{-# INLINE [1] pack #-}
432229
230+-- | /O(n)/ Convert a Text into a String. Subject to fusion.
231+unpack :: Text -> String
232+unpack = S.unstreamList . stream
233+{-# INLINE [1] unpack #-}
234+
235+-- | /O(1)/ Convert a character into a Text.
236+-- Subject to fusion.
237+singleton :: Char -> Text
238+singleton = unstream . S.singleton
239+{-# INLINE [1] singleton #-}
240+
433241-- -----------------------------------------------------------------------------
434242-- * Basic functions
435243
436244-- | /O(n)/ Adds a character to the front of a 'Text'. This function
437245-- is more costly than its 'List' counterpart because it requires
438−-- copying a new array. Subject to fusion. Performs replacement on
439−-- invalid scalar values.
246+-- copying a new array. Subject to fusion.
440247cons :: Char -> Text -> Text
441−cons c t = unstream (S.cons (safe c) (stream t))
248+cons c t = unstream (S.cons c (stream t))
442249{-# INLINE cons #-}
443250
444−infixr 5 `cons`
445−
446251-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
447252-- entire array in the process, unless fused. Subject to fusion.
448−-- Performs replacement on invalid scalar values.
449253snoc :: Text -> Char -> Text
450−snoc t c = unstream (S.snoc (stream t) (safe c))
254+snoc t c = unstream (S.snoc (stream t) c)
451255{-# INLINE snoc #-}
452256
453257-- | /O(n)/ Appends one 'Text' to the other by copying both of them
454258-- into a new 'Text'. Subject to fusion.
455259append :: Text -> Text -> Text
456−append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)
457− | len1 == 0 = b
458− | len2 == 0 = a
459− | len > 0 = Text (A.run x) 0 len
460− | otherwise = overflowError "append"
260+append (Text arr1 off1 len1) (Text arr2 off2 len2) = Text (A.run x) 0 len
461261 where
462262 len = len1+len2
463− x :: ST s (A.MArray s)
464263 x = do
465− arr <- A.new len
466− A.copyI arr 0 arr1 off1 len1
467− A.copyI arr len1 arr2 off2 len
264+ arr <- A.unsafeNew len :: ST s (A.MArray s Word16)
265+ copy arr1 off1 (len1+off1) arr 0
266+ copy arr2 off2 (len2+off2) arr len1
468267 return arr
469−{-# NOINLINE append #-}
268+ where
269+ copy arr i top arr' j
270+ | i >= top = return ()
271+ | otherwise = do A.unsafeWrite arr' j (arr `A.unsafeIndex` i)
272+ copy arr (i+1) top arr' (j+1)
273+{-# INLINE append #-}
470274
471275{-# RULES
472276"TEXT append -> fused" [~1] forall t1 t2.
@@ -486,14 +290,21 @@ head t = S.head (stream t)
486290uncons :: Text -> Maybe (Char, Text)
487291uncons t@(Text arr off len)
488292 | len <= 0 = Nothing
489− | otherwise = Just $ let !(Iter c d) = iter t 0
490− in (c, text arr (off+d) (len-d))
293+ | otherwise = Just (c, textP arr (off+d) (len-d))
294+ where (c,d) = iter t 0
491295{-# INLINE [1] uncons #-}
492296
493297-- | Lifted from Control.Arrow and specialized.
494298second :: (b -> c) -> (a,b) -> (a,c)
495299second f (a, b) = (a, f b)
496300
301+{-# RULES
302+"TEXT uncons -> fused" [~1] forall t.
303+ uncons t = fmap (second unstream) (S.uncons (stream t))
304+"TEXT uncons -> unfused" [1] forall t.
305+ fmap (second unstream) (S.uncons (stream t)) = uncons t
306+ #-}
307+
497308-- | /O(1)/ Returns the last character of a 'Text', which must be
498309-- non-empty. Subject to fusion.
499310last :: Text -> Char
@@ -517,7 +328,7 @@ last (Text arr off len)
517328tail :: Text -> Text
518329tail t@(Text arr off len)
519330 | len <= 0 = emptyError "tail"
520− | otherwise = text arr (off+d) (len-d)
331+ | otherwise = textP arr (off+d) (len-d)
521332 where d = iter_ t 0
522333{-# INLINE [1] tail #-}
523334
@@ -532,8 +343,8 @@ tail t@(Text arr off len)
532343-- be non-empty. Subject to fusion.
533344init :: Text -> Text
534345init (Text arr off len) | len <= 0 = emptyError "init"
535− | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)
536− | otherwise = text arr off (len-1)
346+ | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)
347+ | otherwise = textP arr off (len-1)
537348 where
538349 n = A.unsafeIndex arr (off+len-1)
539350{-# INLINE [1] init #-}
@@ -548,11 +359,7 @@ init (Text arr off len) | len <= 0 = emptyError "init"
548359-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to
549360-- fusion.
550361null :: Text -> Bool
551−null (Text _arr _off len) =
552−#if defined(ASSERTS)
553− assert (len >= 0) $
554−#endif
555− len <= 0
362+null (Text _arr _off len) = assert (len >= 0) $ len <= 0
556363{-# INLINE [1] null #-}
557364
558365{-# RULES
@@ -562,86 +369,31 @@ null (Text _arr _off len) =
562369 S.null (stream t) = null t
563370 #-}
564371
565−-- | /O(1)/ Tests whether a 'Text' contains exactly one character.
566−-- Subject to fusion.
567−isSingleton :: Text -> Bool
568−isSingleton = S.isSingleton . stream
569−{-# INLINE isSingleton #-}
570−
571372-- | /O(n)/ Returns the number of characters in a 'Text'.
572373-- Subject to fusion.
573374length :: Text -> Int
574375length t = S.length (stream t)
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.
578−
579−-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
580−-- Subject to fusion.
581−--
582−-- This function gives the same answer as comparing against the result
583−-- of 'length', but can short circuit if the count of characters is
584−-- greater than the number, and hence be more efficient.
585−compareLength :: Text -> Int -> Ordering
586−compareLength t n = S.compareLengthI (stream t) n
587−{-# INLINE [1] compareLength #-}
588−
589−{-# RULES
590−"TEXT compareN/length -> compareLength" [~1] forall t n.
591− compare (length t) n = compareLength t n
592− #-}
593−
594−{-# RULES
595−"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
596− eqInt (length t) n = compareLength t n == EQ
597− #-}
598−
599−{-# RULES
600−"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
601− neInt (length t) n = compareLength t n /= EQ
602− #-}
603−
604−{-# RULES
605−"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
606− ltInt (length t) n = compareLength t n == LT
607− #-}
608−
609−{-# RULES
610−"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
611− leInt (length t) n = compareLength t n /= GT
612− #-}
613−
614−{-# RULES
615−"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
616− gtInt (length t) n = compareLength t n == GT
617− #-}
618−
619−{-# RULES
620−"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
621− geInt (length t) n = compareLength t n /= LT
622− #-}
376+{-# INLINE length #-}
623377
624378-- -----------------------------------------------------------------------------
625379-- * Transformations
626380-- | /O(n)/ 'map' @f@ @t@ is the 'Text' obtained by applying @f@ to
627−-- each element of @t@. Subject to fusion. Performs replacement on
628−-- invalid scalar values.
381+-- each element of @t@. Subject to fusion.
629382map :: (Char -> Char) -> Text -> Text
630−map f t = unstream (S.map (safe . f) (stream t))
383+map f t = unstream (S.map f (stream t))
631384{-# INLINE [1] map #-}
632385
633386-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
634387-- 'Text's and concatenates the list after interspersing the first
635388-- argument between each element of the list.
636389intercalate :: Text -> [Text] -> Text
637−intercalate t = concat . (F.intersperse t)
390+intercalate t ts = unstream (S.intercalate (stream t) (L.map stream ts))
638391{-# INLINE intercalate #-}
639392
640393-- | /O(n)/ The 'intersperse' function takes a character and places it
641−-- between the characters of a 'Text'. Subject to fusion. Performs
642−-- replacement on invalid scalar values.
394+-- between the characters of a 'Text'. Subject to fusion.
643395intersperse :: Char -> Text -> Text
644−intersperse c t = unstream (S.intersperse (safe c) (stream t))
396+intersperse c t = unstream (S.intersperse c (stream t))
645397{-# INLINE intersperse #-}
646398
647399-- | /O(n)/ Reverse the characters of a string. Subject to fusion.
@@ -649,59 +401,13 @@ reverse :: Text -> Text
649401reverse t = S.reverse (stream t)
650402{-# INLINE reverse #-}
651403
652−-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in
653−-- @haystack@ with @replacement@.
654−--
655−-- This function behaves as though it was defined as follows:
656−--
657−-- @
658−-- replace needle replacement haystack =
659−-- 'intercalate' replacement ('splitOn' needle haystack)
660−-- @
661−--
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:
665−--
666−-- > replace "oo" "foo" "oo" == "foo"
667−--
668−-- In cases where several instances of @needle@ overlap, only the
669−-- first one will be replaced:
670−--
671−-- > replace "ofo" "bar" "ofofo" == "barfo"
672−--
673−-- In (unlikely) bad cases, this function's time complexity degrades
674−-- towards /O(n*m)/.
675−replace :: Text
676− -- ^ @needle@ to search for. If this string is empty, an
677− -- error will occur.
678− -> Text
679− -- ^ @replacement@ to replace @needle@ with.
680− -> Text
681− -- ^ @haystack@ in which to search.
404+-- | /O(m)*O(n)/ Replace every occurrence of one substring with another.
405+replace :: Text -- ^ Text to search for
406+ -> Text -- ^ Replacement text
407+ -> Text -- ^ Input text
682408 -> Text
683−replace needle@(Text _ _ neeLen)
684− (Text repArr repOff repLen)
685− haystack@(Text hayArr hayOff hayLen)
686− | neeLen == 0 = emptyError "replace"
687− | L.null ixs = haystack
688− | len > 0 = Text (A.run x) 0 len
689− | otherwise = empty
690− where
691− ixs = indices needle haystack
692− len = hayLen - (neeLen - repLen) `mul` L.length ixs
693− x :: ST s (A.MArray s)
694− x = do
695− marr <- A.new len
696− let loop (i:is) o d = do
697− let d0 = d + i - o
698− d1 = d0 + repLen
699− A.copyI marr d hayArr (hayOff+o) d0
700− A.copyI marr d0 repArr repOff d1
701− loop is (i + neeLen) d1
702− loop [] o d = A.copyI marr d hayArr (hayOff+o) len
703− loop ixs 0 0
704− return marr
409+replace s d = intercalate d . split s
410+{-# INLINE replace #-}
705411
706412-- ----------------------------------------------------------------------------
707413-- ** Case conversions (folds)
@@ -716,17 +422,10 @@ replace needle@(Text _ _ neeLen)
716422-- case conversion rules. As a result, these functions may map one
717423-- input character to two or three output characters. For examples,
718424-- see the documentation of each function.
719−--
720−-- /Note/: In some languages, case conversion is a locale- and
721−-- context-dependent operation. The case conversion functions in this
722−-- module are /not/ locale sensitive. Programs that require locale
723−-- sensitivity should use appropriate versions of the
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 >.
725425
726−-- | /O(n)/ Convert a string to folded case. Subject to fusion.
727−--
728−-- This function is mainly useful for performing caseless (also known
729−-- as case insensitive) string comparisons.
426+-- | /O(n)/ Convert a string to folded case. This function is mainly
427+-- useful for performing caseless (also known as case insensitive)
428+-- string comparisons.
730429--
731430-- A string @x@ is a caseless match for a string @y@ if and only if:
732431--
@@ -741,56 +440,27 @@ replace needle@(Text _ _ neeLen)
741440-- instead of itself.
742441toCaseFold :: Text -> Text
743442toCaseFold t = unstream (S.toCaseFold (stream t))
744−{-# INLINE toCaseFold #-}
443+{-# INLINE [0] toCaseFold #-}
745444
746445-- | /O(n)/ Convert a string to lower case, using simple case
747−-- conversion. Subject to fusion.
748−--
749−-- The result string may be longer than the input string. For
750−-- instance, \"İ\" (Latin capital letter I with dot above,
751−-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069)
752−-- followed by \" ̇\" (combining dot above, U+0307).
446+-- conversion. The result string may be longer than the input string.
447+-- For instance, \"İ\" (Latin capital letter I with dot above,
448+-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069) followed
449+-- by \" ̇\" (combining dot above, U+0307).
753450toLower :: Text -> Text
754451toLower t = unstream (S.toLower (stream t))
755452{-# INLINE toLower #-}
756453
757454-- | /O(n)/ Convert a string to upper case, using simple case
758−-- conversion. Subject to fusion.
759−--
760−-- The result string may be longer than the input string. For
761−-- instance, the German \"ß\" (eszett, U+00DF) maps to the
455+-- conversion. The result string may be longer than the input string.
456+-- For instance, the German \"ß\" (eszett, U+00DF) maps to the
762457-- two-letter sequence \"SS\".
763458toUpper :: Text -> Text
764459toUpper t = unstream (S.toUpper (stream t))
765460{-# INLINE toUpper #-}
766461
767−-- | /O(n)/ Convert a string to title case, using simple case
768−-- conversion. Subject to fusion.
769−--
770−-- The first letter of the input is converted to title case, as is
771−-- every subsequent letter that immediately follows a non-letter.
772−-- Every letter that immediately follows another letter is converted
773−-- to lower case.
774−--
775−-- The result string may be longer than the input string. For example,
776−-- the Latin small ligature fl (U+FB02) is converted to the
777−-- sequence Latin capital letter F (U+0046) followed by Latin small
778−-- letter l (U+006C).
779−--
780−-- /Note/: this function does not take language or culture specific
781−-- rules into account. For instance, in English, different style
782−-- guides disagree on whether the book name \"The Hill of the Red
783−-- Fox\" is correctly title cased—but this function will
784−-- capitalize /every/ word.
785−toTitle :: Text -> Text
786−toTitle t = unstream (S.toTitle (stream t))
787−{-# INLINE toTitle #-}
788−
789462-- | /O(n)/ Left-justify a string to the given length, using the
790−-- specified fill character on the right. Subject to fusion.
791−-- Performs replacement on invalid scalar values.
792−--
793−-- Examples:
463+-- specified fill character on the right. Subject to fusion. Examples:
794464--
795465-- > justifyLeft 7 'x' "foo" == "fooxxxx"
796466-- > justifyLeft 3 'x' "foobar" == "foobar"
@@ -809,10 +479,7 @@ justifyLeft k c t
809479 #-}
810480
811481-- | /O(n)/ Right-justify a string to the given length, using the
812−-- specified fill character on the left. Performs replacement on
813−-- invalid scalar values.
814−--
815−-- Examples:
482+-- specified fill character on the left. Examples:
816483--
817484-- > justifyRight 7 'x' "bar" == "xxxxbar"
818485-- > justifyRight 3 'x' "foobar" == "foobar"
@@ -823,11 +490,8 @@ justifyRight k c t
823490 where len = length t
824491{-# INLINE justifyRight #-}
825492
826−-- | /O(n)/ Center a string to the given length, using the specified
827−-- fill character on either side. Performs replacement on invalid
828−-- scalar values.
829−--
830−-- Examples:
493+-- | /O(n)/ Center a string to the given length, using the
494+-- specified fill character on either side. Examples:
831495--
832496-- > center 8 'x' "HS" = "xxxHSxxx"
833497center :: Int -> Char -> Text -> Text
@@ -836,7 +500,7 @@ center k c t
836500 | otherwise = replicateChar l c `append` t `append` replicateChar r c
837501 where len = length t
838502 d = k - len
839− r = d `quot` 2
503+ r = d `div` 2
840504 l = d - r
841505{-# INLINE center #-}
842506
@@ -854,12 +518,12 @@ transpose ts = P.map pack (L.transpose (P.map unpack ts))
854518-- (typically the left-identity of the operator), and a 'Text',
855519-- reduces the 'Text' using the binary operator, from left to right.
856520-- Subject to fusion.
857−foldl :: (a -> Char -> a) -> a -> Text -> a
521+foldl :: (b -> Char -> b) -> b -> Text -> b
858522foldl f z t = S.foldl f z (stream t)
859523{-# INLINE foldl #-}
860524
861525-- | /O(n)/ A strict version of 'foldl'. Subject to fusion.
862−foldl' :: (a -> Char -> a) -> a -> Text -> a
526+foldl' :: (b -> Char -> b) -> b -> Text -> b
863527foldl' f z t = S.foldl' f z (stream t)
864528{-# INLINE foldl' #-}
865529
@@ -878,12 +542,12 @@ foldl1' f t = S.foldl1' f (stream t)
878542-- (typically the right-identity of the operator), and a 'Text',
879543-- reduces the 'Text' using the binary operator, from right to left.
880544-- Subject to fusion.
881−foldr :: (Char -> a -> a) -> a -> Text -> a
545+foldr :: (Char -> b -> b) -> b -> Text -> b
882546foldr f z t = S.foldr f z (stream t)
883547{-# INLINE foldr #-}
884548
885549-- | /O(n)/ A variant of 'foldr' that has no starting value argument,
886−-- and thus must be applied to a non-empty 'Text'. Subject to
550+-- and thust must be applied to a non-empty 'Text'. Subject to
887551-- fusion.
888552foldr1 :: (Char -> Char -> Char) -> Text -> Char
889553foldr1 f t = S.foldr1 f (stream t)
@@ -892,36 +556,28 @@ foldr1 f t = S.foldr1 f (stream t)
892556-- -----------------------------------------------------------------------------
893557-- ** Special folds
894558
895−-- | /O(n)/ Concatenate a list of 'Text's.
559+-- | /O(n)/ Concatenate a list of 'Text's. Subject to fusion.
896560concat :: [Text] -> Text
897−concat ts = case ts' of
898− [] -> empty
899− [t] -> t
900− _ -> Text (A.run go) 0 len
901− where
902− ts' = L.filter (not . null) ts
903− len = sumP "concat" $ L.map lengthWord16 ts'
904− go :: ST s (A.MArray s)
905− go = do
906− arr <- A.new len
907− let step i (Text a o l) =
908− let !j = i + l in A.copyI arr i a o j >> return j
909− foldM step 0 ts' >> return arr
561+concat ts = unstream (S.concat (L.map stream ts))
562+{-# INLINE concat #-}
910563
911564-- | /O(n)/ Map a function over a 'Text' that results in a 'Text', and
912−-- concatenate the results.
565+-- concatenate the results. Subject to fusion.
566+--
567+-- Note: if in 'concatMap' @f@ @t@, @f@ is defined in terms of fusible
568+-- functions, it will also be fusible.
913569concatMap :: (Char -> Text) -> Text -> Text
914−concatMap f = concat . foldr ((:) . f) []
570+concatMap f t = unstream (S.concatMap (stream . f) (stream t))
915571{-# INLINE concatMap #-}
916572
917573-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
918−-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.
574+-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.
919575any :: (Char -> Bool) -> Text -> Bool
920576any p t = S.any p (stream t)
921577{-# INLINE any #-}
922578
923579-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
924−-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.
580+-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.
925581all :: (Char -> Bool) -> Text -> Bool
926582all p t = S.all p (stream t)
927583{-# INLINE all #-}
@@ -943,7 +599,6 @@ minimum t = S.minimum (stream t)
943599
944600-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
945601-- successive reduced values from the left. Subject to fusion.
946−-- Performs replacement on invalid scalar values.
947602--
948603-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
949604--
@@ -951,13 +606,11 @@ minimum t = S.minimum (stream t)
951606--
952607-- > last (scanl f z xs) == foldl f z xs.
953608scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
954−scanl f z t = unstream (S.scanl g z (stream t))
955− where g a b = safe (f a b)
609+scanl f z t = unstream (S.scanl f z (stream t))
956610{-# INLINE scanl #-}
957611
958612-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
959−-- value argument. Subject to fusion. Performs replacement on
960−-- invalid scalar values.
613+-- value argument. Subject to fusion.
961614--
962615-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
963616scanl1 :: (Char -> Char -> Char) -> Text -> Text
@@ -965,63 +618,48 @@ scanl1 f t | null t = empty
965618 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
966619{-# INLINE scanl1 #-}
967620
968−-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'. Performs
969−-- replacement on invalid scalar values.
621+-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'.
970622--
971623-- > scanr f v == reverse . scanl (flip f) v . reverse
972624scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
973−scanr f z = S.reverse . S.reverseScanr g z . reverseStream
974− where g a b = safe (f a b)
625+scanr f z = S.reverse . S.reverseScanr f z . reverseStream
975626{-# INLINE scanr #-}
976627
977628-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
978−-- value argument. Subject to fusion. Performs replacement on
979−-- invalid scalar values.
629+-- value argument. Subject to fusion.
980630scanr1 :: (Char -> Char -> Char) -> Text -> Text
981631scanr1 f t | null t = empty
982632 | otherwise = scanr f (last t) (init t)
983633{-# INLINE scanr1 #-}
984634
985−-- | /O(n)/ Like a combination of 'map' and 'foldl''. Applies a
635+-- | /O(n)/ Like a combination of 'map' and 'foldl'. Applies a
986636-- function to each element of a 'Text', passing an accumulating
987−-- parameter from left to right, and returns a final 'Text'. Performs
988−-- replacement on invalid scalar values.
637+-- parameter from left to right, and returns a final 'Text'.
989638mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
990−mapAccumL f z0 = S.mapAccumL g z0 . stream
991− where g a b = second safe (f a b)
992−{-# INLINE mapAccumL #-}
639+mapAccumL f s t = case uncons t of
640+ Nothing -> (s, empty)
641+ Just (x, xs) -> (s'', cons y ys)
642+ where (s', y ) = f s x
643+ (s'',ys) = mapAccumL f s' xs
993644
994645-- | The 'mapAccumR' function behaves like a combination of 'map' and
995−-- a strict 'foldr'; it applies a function to each element of a
996−-- 'Text', passing an accumulating parameter from right to left, and
997−-- returning a final value of this accumulator together with the new
998−-- 'Text'.
999−-- Performs replacement on invalid scalar values.
646+-- 'foldr'; it applies a function to each element of a 'Text', passing
647+-- an accumulating parameter from right to left, and returning a final
648+-- value of this accumulator together with the new 'Text'.
1000649mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
1001−mapAccumR f z0 = second reverse . S.mapAccumL g z0 . reverseStream
1002− where g a b = second safe (f a b)
1003−{-# INLINE mapAccumR #-}
650+mapAccumR f s t = case uncons t of
651+ Nothing -> (s, empty)
652+ Just (x, xs) -> (s'', cons y ys)
653+ where (s'',y ) = f s' x
654+ (s', ys) = mapAccumR f s xs
1004655
1005656-- -----------------------------------------------------------------------------
1006657-- ** Generating and unfolding 'Text's
1007658
1008659-- | /O(n*m)/ 'replicate' @n@ @t@ is a 'Text' consisting of the input
1009−-- @t@ repeated @n@ times.
660+-- @t@ repeated @n@ times. Subject to fusion.
1010661replicate :: Int -> Text -> Text
1011−replicate n t@(Text a o l)
1012− | n <= 0 || l <= 0 = empty
1013− | n == 1 = t
1014− | isSingleton t = replicateChar n (unsafeHead t)
1015− | otherwise = Text (A.run x) 0 len
1016− where
1017− len = l `mul` n
1018− x :: ST s (A.MArray s)
1019− x = do
1020− arr <- A.new len
1021− let loop !d !i | i >= n = return arr
1022− | otherwise = let m = d + l
1023− in A.copyI arr d a o m >> loop m (i+1)
1024− loop 0 0
662+replicate n t = unstream (S.replicateI (fromIntegral n) (S.stream t))
1025663{-# INLINE [1] replicate #-}
1026664
1027665{-# RULES
@@ -1032,8 +670,7 @@ replicate n t@(Text a o l)
1032670-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the
1033671-- value of every element. Subject to fusion.
1034672replicateChar :: Int -> Char -> Text
1035−replicateChar n c = unstream (S.replicateCharI n (safe c))
1036−{-# INLINE replicateChar #-}
673+replicateChar n c = unstream (S.replicateCharI n c)
1037674
1038675-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
1039676-- function is analogous to the List 'L.unfoldr'. 'unfoldr' builds a
@@ -1041,9 +678,9 @@ replicateChar n c = unstream (S.replicateCharI n (safe c))
1041678-- returns 'Nothing' if it is done producing the 'Text', otherwise
1042679-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
1043680-- string, and @b@ is the seed value for further production. Subject
1044−-- to fusion. Performs replacement on invalid scalar values.
681+-- to fusion.
1045682unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
1046−unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
683+unfoldr f s = unstream (S.unfoldr f s)
1047684{-# INLINE unfoldr #-}
1048685
1049686-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
@@ -1051,9 +688,9 @@ unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
1051688-- first argument to 'unfoldrN'. This function is more efficient than
1052689-- 'unfoldr' when the maximum length of the result is known and
1053690-- correct, otherwise its performance is similar to 'unfoldr'. Subject
1054−-- to fusion. Performs replacement on invalid scalar values.
691+-- to fusion.
1055692unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
1056−unfoldrN n f s = unstream (S.unfoldrN n (firstf safe . f) s)
693+unfoldrN n f s = unstream (S.unfoldrN n f s)
1057694{-# INLINE unfoldrN #-}
1058695
1059696-- -----------------------------------------------------------------------------
@@ -1066,16 +703,14 @@ take :: Int -> Text -> Text
1066703take n t@(Text arr off len)
1067704 | n <= 0 = empty
1068705 | n >= len = t
1069− | otherwise = text arr off (iterN n t)
706+ | otherwise = Text arr off (loop 0 0)
707+ where
708+ loop !i !cnt
709+ | i >= len || cnt >= n = i
710+ | otherwise = loop (i+d) (cnt+1)
711+ where d = iter_ t i
1070712{-# INLINE [1] take #-}
1071713
1072−iterN :: Int -> Text -> Int
1073−iterN n t@(Text _arr _off len) = loop 0 0
1074− where loop !i !cnt
1075− | i >= len || cnt >= n = i
1076− | otherwise = loop (i+d) (cnt+1)
1077− where d = iter_ t i
1078−
1079714{-# RULES
1080715"TEXT take -> fused" [~1] forall n t.
1081716 take n t = unstream (S.take n (stream t))
@@ -1083,36 +718,18 @@ iterN n t@(Text _arr _off len) = loop 0 0
1083718 unstream (S.take n (stream t)) = take n t
1084719 #-}
1085720
1086−-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after
1087−-- taking @n@ characters from the end of @t@.
1088−--
1089−-- Examples:
1090−--
1091−-- > takeEnd 3 "foobar" == "bar"
1092−takeEnd :: Int -> Text -> Text
1093−takeEnd n t@(Text arr off len)
1094− | n <= 0 = empty
1095− | n >= len = t
1096− | otherwise = text arr (off+i) (len-i)
1097− where i = iterNEnd n t
1098−
1099−iterNEnd :: Int -> Text -> Int
1100−iterNEnd n t@(Text _arr _off len) = loop (len-1) n
1101− where loop i !m
1102− | i <= 0 = 0
1103− | m <= 1 = i
1104− | otherwise = loop (i+d) (m-1)
1105− where d = reverseIter_ t i
1106−
1107721-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
1108−-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
1109−-- is greater than the length of the 'Text'. Subject to fusion.
722+-- 'Text' of length @n@, or the empty 'Text' if @n@ is greater than the
723+-- length of the 'Text'. Subject to fusion.
1110724drop :: Int -> Text -> Text
1111725drop n t@(Text arr off len)
1112726 | n <= 0 = t
1113727 | n >= len = empty
1114− | otherwise = text arr (off+i) (len-i)
1115− where i = iterN n t
728+ | otherwise = loop 0 0
729+ where loop !i !cnt
730+ | i >= len || cnt >= n = Text arr (off+i) (len-i)
731+ | otherwise = loop (i+d) (cnt+1)
732+ where d = iter_ t i
1116733{-# INLINE [1] drop #-}
1117734
1118735{-# RULES
@@ -1122,18 +739,6 @@ drop n t@(Text arr off len)
1122739 unstream (S.drop n (stream t)) = drop n t
1123740 #-}
1124741
1125−-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after
1126−-- dropping @n@ characters from the end of @t@.
1127−--
1128−-- Examples:
1129−--
1130−-- > dropEnd 3 "foobar" == "foo"
1131−dropEnd :: Int -> Text -> Text
1132−dropEnd n t@(Text arr off len)
1133− | n <= 0 = t
1134− | n >= len = empty
1135− | otherwise = text arr off (iterNEnd n t)
1136−
1137742-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
1138743-- returns the longest prefix (possibly empty) of elements that
1139744-- satisfy @p@. Subject to fusion.
@@ -1141,8 +746,8 @@ takeWhile :: (Char -> Bool) -> Text -> Text
1141746takeWhile p t@(Text arr off len) = loop 0
1142747 where loop !i | i >= len = t
1143748 | p c = loop (i+d)
1144− | otherwise = text arr off i
1145− where Iter c d = iter t i
749+ | otherwise = textP arr off i
750+ where (c,d) = iter t i
1146751{-# INLINE [1] takeWhile #-}
1147752
1148753{-# RULES
@@ -1152,27 +757,6 @@ takeWhile p t@(Text arr off len) = loop 0
1152757 unstream (S.takeWhile p (stream t)) = takeWhile p t
1153758 #-}
1154759
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.
1158−-- Examples:
1159−--
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 #-}
1168−
1169−{-# RULES
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
1174− #-}
1175−
1176760-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
1177761-- 'takeWhile' @p@ @t@. Subject to fusion.
1178762dropWhile :: (Char -> Bool) -> Text -> Text
@@ -1180,7 +764,7 @@ dropWhile p t@(Text arr off len) = loop 0 0
1180764 where loop !i !l | l >= len = empty
1181765 | p c = loop (i+d) (l+d)
1182766 | otherwise = Text arr (off+i) (len-l)
1183− where Iter c d = iter t i
767+ where (c,d) = iter t i
1184768{-# INLINE [1] dropWhile #-}
1185769
1186770{-# RULES
@@ -1191,9 +775,8 @@ dropWhile p t@(Text arr off len) = loop 0 0
1191775 #-}
1192776
1193777-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
1194−-- dropping characters that satisfy the predicate @p@ from the end of
778+-- dropping characters that fail the predicate @p@ from the end of
1195779-- @t@. Subject to fusion.
1196−--
1197780-- Examples:
1198781--
1199782-- > dropWhileEnd (=='.') "foo..." == "foo"
@@ -1213,7 +796,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len
1213796 #-}
1214797
1215798-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
1216−-- dropping characters that satisfy the predicate @p@ from both the
799+-- dropping characters that fail the predicate @p@ from both the
1217800-- beginning and end of @t@. Subject to fusion.
1218801dropAround :: (Char -> Bool) -> Text -> Text
1219802dropAround p = dropWhile p . dropWhileEnd p
@@ -1248,20 +831,28 @@ splitAt :: Int -> Text -> (Text, Text)
1248831splitAt n t@(Text arr off len)
1249832 | n <= 0 = (empty, t)
1250833 | n >= len = (t, empty)
1251− | otherwise = let k = iterN n t
1252− in (text arr off k, text arr (off+k) (len-k))
834+ | otherwise = (Text arr off k, Text arr (off+k) (len-k))
835+ where k = loop 0 0
836+ loop !i !cnt
837+ | i >= len || cnt >= n = i
838+ | otherwise = loop (i+d) (cnt+1)
839+ where d = iter_ t i
840+{-# INLINE splitAt #-}
1253841
1254−-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
1255−-- a pair whose first element is the longest prefix (possibly empty)
1256−-- of @t@ of elements that satisfy @p@, and whose second is the
1257−-- remainder of the list.
842+-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns a
843+-- pair whose first element is the longest prefix (possibly empty) of
844+-- @t@ of elements that satisfy @p@, and whose second is the remainder
845+-- of the list.
1258846span :: (Char -> Bool) -> Text -> (Text, Text)
1259−span p t = case span_ p t of
1260− (# hd,tl #) -> (hd,tl)
847+span p t@(Text arr off len) = (textP arr off k, textP arr (off+k) (len-k))
848+ where k = loop 0
849+ loop !i | i >= len || not (p c) = i
850+ | otherwise = loop (i+d)
851+ where (c,d) = iter t i
1261852{-# INLINE span #-}
1262853
1263−-- | /O(n)/ 'break' is like 'span', but the prefix returned is
1264−-- over elements that fail the predicate @p@.
854+-- | /O(n)/ 'break' is like 'span', but the prefix returned is over
855+-- elements that fail the predicate @p@.
1265856break :: (Char -> Bool) -> Text -> (Text, Text)
1266857break p = span (not . p)
1267858{-# INLINE break #-}
@@ -1273,7 +864,7 @@ groupBy p = loop
1273864 loop t@(Text arr off len)
1274865 | null t = []
1275866 | otherwise = text arr off n : loop (text arr (off+n) (len-n))
1276− where Iter c d = iter t 0
867+ where (c,d) = iter t 0
1277868 n = d + findAIndexOrEnd (not . p c) (Text arr (off+d) (len-d))
1278869
1279870-- | Returns the /array/ index (in units of 'Word16') at which a
@@ -1283,8 +874,8 @@ findAIndexOrEnd :: (Char -> Bool) -> Text -> Int
1283874findAIndexOrEnd q t@(Text _arr _off len) = go 0
1284875 where go !i | i >= len || q c = i
1285876 | otherwise = go (i+d)
1286− where Iter c d = iter t i
1287−
877+ where (c,d) = iter t i
878+
1288879-- | /O(n)/ Group characters in a string by equality.
1289880group :: Text -> [Text]
1290881group = groupBy (==)
@@ -1308,59 +899,107 @@ tails t | null t = [empty]
1308899-- copies to create substrings; they just construct new 'Text's that
1309900-- are slices of the original.
1310901
1311−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'
1312−-- argument (which cannot be empty), consuming the delimiter. An empty
1313−-- delimiter is invalid, and will cause an error to be raised.
902+-- | /O(m)*O(n)/ Break a 'Text' into pieces separated by the first
903+-- 'Text' argument, consuming the delimiter. An empty delimiter is
904+-- invalid, and will cause an error to be raised.
1314905--
1315906-- Examples:
1316907--
1317−-- > splitOn "\r\n" "a\r\nb\r\nd\r\ne" == ["a","b","d","e"]
1318−-- > splitOn "aaa" "aaaXaaaXaaaXaaa" == ["","X","X","X",""]
1319−-- > splitOn "x" "x" == ["",""]
1320−--
908+-- > split "\r\n" "a\r\nb\r\nd\r\ne" == ["a","b","d","e"]
909+-- > split "aaa" "aaaXaaaXaaaXaaa" == ["","X","X","X",""]
910+-- > split "x" "x" == ["",""]
911+--
1321912-- and
1322913--
1323−-- > intercalate s . splitOn s == id
1324−-- > splitOn (singleton c) == split (==c)
1325−--
1326−-- (Note: the string @s@ to split on above cannot be empty.)
1327−--
1328−-- In (unlikely) bad cases, this function's time complexity degrades
1329−-- towards /O(n*m)/.
1330−splitOn :: Text
1331− -- ^ String to split on. If this string is empty, an error
1332− -- will occur.
1333− -> Text
1334− -- ^ Input text.
1335− -> [Text]
1336−splitOn pat@(Text _ _ l) src@(Text arr off len)
1337− | l <= 0 = emptyError "splitOn"
1338− | isSingleton pat = split (== unsafeHead pat) src
1339− | otherwise = go 0 (indices pat src)
914+-- > intercalate s . split s == id
915+-- > split (singleton c) == splitWith (==c)
916+split :: Text -- ^ Text to split on
917+ -> Text -- ^ Input text
918+ -> [Text]
919+split pat src0
920+ | null pat = emptyError "split"
921+ | l == 1 = splitWith (== (unsafeHead pat)) src0
922+ | otherwise = go src0
1340923 where
1341− go !s (x:xs) = text arr (s+off) (x-s) : go (x+l) xs
1342− go s _ = [text arr (s+off) (len-s)]
1343−{-# INLINE [1] splitOn #-}
924+ l = length pat
925+ go src = search 0 src
926+ where
927+ search !n !s
928+ | null s = [src] -- not found
929+ | pat `isPrefixOf` s = take n src : go (drop l s)
930+ | otherwise = search (n+1) (unsafeTail s)
931+{-# INLINE [1] split #-}
1344932
1345933{-# RULES
1346−"TEXT splitOn/singleton -> split/==" [~1] forall c t.
1347− splitOn (singleton c) t = split (==c) t
934+"TEXT split/singleton -> splitWith/==" [~1] forall c t.
935+ split (singleton c) t = splitWith (==c) t
1348936 #-}
1349937
938+-- | /O(m)*O(n)/ Break a 'Text' into pieces at most @k@ times,
939+-- treating the first 'Text' argument as the delimiter to break on,
940+-- and consuming the delimiter. The last element of the list contains
941+-- the remaining text after the number of times to split has been
942+-- reached. A value of zero or less for @k@ causes no splitting to
943+-- occur. An empty delimiter is invalid, and will cause an error to be
944+-- raised.
945+--
946+-- Examples:
947+--
948+-- > splitTimes 0 "//" "a//b//c" == ["a//b//c"]
949+-- > splitTimes 2 ":" "a:b:c:d:e" == ["a","b","c:d:e"]
950+-- > splitTimes 100 "???" "a????b" == ["a","?b"]
951+--
952+-- and
953+--
954+-- > intercalate s . splitTimes k s == id
955+splitTimes :: Int -- ^ Maximum number of times to split
956+ -> Text -- ^ Text to split on
957+ -> Text -- ^ Input text
958+ -> [Text]
959+splitTimes k pat src0
960+ | k <= 0 = [src0]
961+ | null pat = emptyError "splitTimes"
962+ | otherwise = go k src0
963+ where
964+ l = length pat
965+ go !i src = search 0 src
966+ where
967+ search !n !s
968+ | i == 0 || null s = [src] -- not found or limit reached
969+ | pat `isPrefixOf` s = take n src : go (i-1) (drop l s)
970+ | otherwise = search (n+1) (unsafeTail s)
971+{-# INLINE splitTimes #-}
972+
973+-- | /O(m)*O(n)/ Break a 'Text' into pieces at most @k@ times, like
974+-- 'splitTimes', but start from the end of the input and work towards
975+-- the start.
976+--
977+-- Examples:
978+--
979+-- > splitTimes 2 "::" "a::b::c::d::e" == ["a","b","c::d::e"]
980+-- > splitTimesEnd 2 "::" "a::b::c::d::e" == ["a::b::c","d","e"]
981+splitTimesEnd :: Int -- ^ Maximum number of times to split
982+ -> Text -- ^ Text to split on
983+ -> Text -- ^ Input text
984+ -> [Text]
985+splitTimesEnd k pat src =
986+ L.reverse . L.map reverse $ splitTimes k (reverse pat) (reverse src)
987+{-# INLINE splitTimesEnd #-}
988+
1350989-- | /O(n)/ Splits a 'Text' into components delimited by separators,
1351990-- where the predicate returns True for a separator element. The
1352991-- resulting components do not contain the separators. Two adjacent
1353992-- separators result in an empty component in the output. eg.
1354993--
1355−-- > split (=='a') "aabbaca" == ["","","bb","c",""]
1356−-- > split (=='a') "" == [""]
1357−split :: (Char -> Bool) -> Text -> [Text]
1358−split _ t@(Text _off _arr 0) = [t]
1359−split p t = loop t
994+-- > splitWith (=='a') "aabbaca" == ["","","bb","c",""]
995+-- > splitWith (=='a') "" == [""]
996+splitWith :: (Char -> Bool) -> Text -> [Text]
997+splitWith _ t@(Text _off _arr 0) = [t]
998+splitWith p t = loop t
1360999 where loop s | null s' = [l]
13611000 | otherwise = l : loop (unsafeTail s')
1362− where (# l, s' #) = span_ (not . p) s
1363−{-# INLINE split #-}
1001+ where (l, s') = break p s
1002+{-# INLINE splitWith #-}
13641003
13651004-- | /O(n)/ Splits a 'Text' into components of length @k@. The last
13661005-- element may be shorter than the other chunks, depending on the
@@ -1379,14 +1018,22 @@ chunksOf k = go
13791018-- ----------------------------------------------------------------------------
13801019-- * Searching
13811020
1021+-------------------------------------------------------------------------------
1022+-- ** Searching by equality
1023+
1024+-- | /O(n)/ 'elem' is the 'Text' membership predicate.
1025+elem :: Char -> Text -> Bool
1026+elem c t = S.elem c (stream t)
1027+{-# INLINE elem #-}
1028+
13821029-------------------------------------------------------------------------------
13831030-- ** Searching with a predicate
13841031
1385−-- | /O(n)/ The 'find' function takes a predicate and a 'Text', and
1386−-- returns the first element matching the predicate, or 'Nothing' if
1387−-- there is no such element.
1032+-- | /O(n)/ The 'find' function takes a predicate and a 'Text',
1033+-- and returns the first element in matching the predicate, or 'Nothing'
1034+-- if there is no such element.
13881035find :: (Char -> Bool) -> Text -> Maybe Char
1389−find p t = S.findBy p (stream t)
1036+find p t = S.find p (stream t)
13901037{-# INLINE find #-}
13911038
13921039-- | /O(n)/ The 'partition' function takes a predicate and a 'Text',
@@ -1405,76 +1052,38 @@ filter :: (Char -> Bool) -> Text -> Text
14051052filter p t = unstream (S.filter p (stream t))
14061053{-# INLINE filter #-}
14071054
1408−-- | /O(n+m)/ Find the first instance of @needle@ (which must be
1409−-- non-'null') in @haystack@. The first element of the returned tuple
1410−-- is the prefix of @haystack@ before @needle@ is matched. The second
1411−-- is the remainder of @haystack@, starting with the match.
1055+-- /O(n*m)/ Find all non-overlapping instances of @needle@ in
1056+-- @haystack@. The first element of the returned tuple is the prefix
1057+-- of @haystack@ before any matches of @needle@. The second element
1058+-- of the tuple is a list of non-overlapping substrings from
1059+-- @haystack@. Each begins with @needle@ and ends just before the
1060+-- following match. If @needle@ is not present in @haystack@, the
1061+-- list will be empty.
14121062--
14131063-- Examples:
14141064--
1415−-- > breakOn "::" "a::b::c" ==> ("a", "::b::c")
1416−-- > breakOn "/" "foobar" ==> ("foobar", "")
1065+-- > findAll "::" "a::b::c" ==> ("a", ["::b", "::c"])
1066+-- > findAll "/" "foobar" ==> ("foobar", [])
14171067--
14181068-- Laws:
14191069--
1420−-- > append prefix match == haystack
1421−-- > where (prefix, match) = breakOn needle haystack
1422−--
1423−-- If you need to break a string by a substring repeatedly (e.g. you
1424−-- want to break on every instance of a substring), use 'breakOnAll'
1425−-- instead, as it has lower startup overhead.
1426−--
1427−-- In (unlikely) bad cases, this function's time complexity degrades
1428−-- towards /O(n*m)/.
1429−breakOn :: Text -> Text -> (Text, Text)
1430−breakOn pat src@(Text arr off len)
1431− | null pat = emptyError "breakOn"
1432− | otherwise = case indices pat src of
1433− [] -> (src, empty)
1434− (x:_) -> (text arr off x, text arr (off+x) (len-x))
1435−{-# INLINE breakOn #-}
1436−
1437−-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
1438−-- string.
1439−--
1440−-- The first element of the returned tuple is the prefix of @haystack@
1441−-- up to and including the last match of @needle@. The second is the
1442−-- remainder of @haystack@, following the match.
1443−--
1444−-- > breakOnEnd "::" "a::b::c" ==> ("a::b::", "c")
1445−breakOnEnd :: Text -> Text -> (Text, Text)
1446−breakOnEnd pat src = (reverse b, reverse a)
1447− where (a,b) = breakOn (reverse pat) (reverse src)
1448−{-# INLINE breakOnEnd #-}
1449−
1450−-- | /O(n+m)/ Find all non-overlapping instances of @needle@ in
1451−-- @haystack@. Each element of the returned list consists of a pair:
1452−--
1453−-- * The entire string prior to the /k/th match (i.e. the prefix)
1454−--
1455−-- * The /k/th match, followed by the remainder of the string
1456−--
1457−-- Examples:
1458−--
1459−-- > breakOnAll "::" ""
1460−-- > ==> []
1461−-- > breakOnAll "/" "a/b/c/"
1462−-- > ==> [("a", "/b/c/"), ("a/b", "/c/"), ("a/b/c", "/")]
1463−--
1464−-- In (unlikely) bad cases, this function's time complexity degrades
1465−-- towards /O(n*m)/.
1466−--
1467−-- The @needle@ parameter may not be empty.
1468−breakOnAll :: Text -- ^ @needle@ to search for
1469− -> Text -- ^ @haystack@ in which to search
1470− -> [(Text, Text)]
1471−breakOnAll pat src@(Text arr off slen)
1472− | null pat = emptyError "breakOnAll"
1473− | otherwise = L.map step (indices pat src)
1070+-- > concat (prefix : matches) == haystack
1071+-- > where (prefix, matches) = findAll needle haystack
1072+findAll :: Text -> Text -> (Text, [Text])
1073+findAll pat@(Text _ _ plen) src@(Text sarr soff slen)
1074+ | plen <= 0 = emptyError "findAll"
1075+ | otherwise = (h,t)
14741076 where
1475− step x = (chunk 0 x, chunk x (slen-x))
1476− chunk !n !l = text arr (n+off) l
1477−{-# INLINE breakOnAll #-}
1077+ (h:t) = go 0 (search 0)
1078+ go k (x:xs) = Text sarr k (x-k) : go x xs
1079+ go _ [] = []
1080+ search i
1081+ | i >= slen = [slen] -- not found
1082+ | pat `isPrefixOf` s = i : search (i+plen)
1083+ | otherwise = search (i+d)
1084+ where d = iter_ src i
1085+ s = Text sarr (soff+i) (slen-i)
1086+{-# INLINE findAll #-}
14781087
14791088-------------------------------------------------------------------------------
14801089-- ** Indexing 'Text's
@@ -1493,7 +1102,8 @@ breakOnAll pat src@(Text arr off slen)
14931102-- For example, suppose you have a string that you want to split on
14941103-- the substring @\"::\"@, such as @\"foo::bar::quux\"@. Instead of
14951104-- searching for the index of @\"::\"@ and taking the substrings
1496−-- before and after that index, you would instead use @breakOnAll \"::\"@.
1105+-- before and after that index, you would instead use @splitTimes 1
1106+-- "::"@.
14971107
14981108-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
14991109index :: Text -> Int -> Char
@@ -1507,17 +1117,42 @@ findIndex :: (Char -> Bool) -> Text -> Maybe Int
15071117findIndex p t = S.findIndex p (stream t)
15081118{-# INLINE findIndex #-}
15091119
1510−-- | /O(n+m)/ The 'count' function returns the number of times the
1120+-- | The 'findIndices' function extends 'findIndex', by returning the
1121+-- indices of all elements satisfying the predicate, in ascending
1122+-- order. Subject to fusion.
1123+findIndices :: (Char -> Bool) -> Text -> [Int]
1124+findIndices p t = S.findIndices p (stream t)
1125+{-# INLINE findIndices #-}
1126+
1127+-- | /O(n)/ The 'elemIndex' function returns the index of the first
1128+-- element in the given 'Text' which is equal to the query element, or
1129+-- 'Nothing' if there is no such element. Subject to fusion.
1130+elemIndex :: Char -> Text -> Maybe Int
1131+elemIndex c t = S.elemIndex c (stream t)
1132+{-# INLINE elemIndex #-}
1133+
1134+-- | /O(n)/ The 'elemIndices' function returns the index of every
1135+-- element in the given 'Text' which is equal to the query
1136+-- element. Subject to fusion.
1137+elemIndices :: Char -> Text -> [Int]
1138+elemIndices c t = S.elemIndices c (stream t)
1139+{-# INLINE elemIndices #-}
1140+
1141+-- | /O(n*m)/ The 'count' function returns the number of times the
15111142-- query string appears in the given 'Text'. An empty query string is
15121143-- invalid, and will cause an error to be raised.
1513−--
1514−-- In (unlikely) bad cases, this function's time complexity degrades
1515−-- towards /O(n*m)/.
15161144count :: Text -> Text -> Int
1517−count pat src
1518− | null pat = emptyError "count"
1519− | isSingleton pat = countChar (unsafeHead pat) src
1520− | otherwise = L.length (indices pat src)
1145+count pat src0
1146+ | null pat = emptyError "count"
1147+ | l == 1 = countChar (unsafeHead pat) src0
1148+ | otherwise = go 0 src0
1149+ where
1150+ l = length pat
1151+ go !n src = search src
1152+ where
1153+ search s | null s = n
1154+ | pat `isPrefixOf` s = go (n+1) (drop l s)
1155+ | otherwise = search (unsafeTail s)
15211156{-# INLINE [1] count #-}
15221157
15231158{-# RULES
@@ -1540,15 +1175,13 @@ countChar c t = S.countChar c (stream t)
15401175-- equivalent to a pair of 'unpack' operations.
15411176zip :: Text -> Text -> [(Char,Char)]
15421177zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1543−{-# INLINE zip #-}
1178+{-# INLINE [0] zip #-}
15441179
15451180-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
15461181-- given as the first argument, instead of a tupling function.
1547−-- Performs replacement on invalid scalar values.
15481182zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1549−zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))
1550− where g a b = safe (f a b)
1551−{-# INLINE zipWith #-}
1183+zipWith f t1 t2 = unstream (S.zipWith f (stream t1) (stream t2))
1184+{-# INLINE [0] zipWith #-}
15521185
15531186-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
15541187-- representing white space.
@@ -1564,7 +1197,7 @@ words t@(Text arr off len) = loop 0 0
15641197 then loop (start+1) (start+1)
15651198 else Text arr (start+off) (n-start) : loop (n+d) (n+d)
15661199 | otherwise = loop start (n+d)
1567− where Iter c d = iter t n
1200+ where (c,d) = iter t n
15681201{-# INLINE words #-}
15691202
15701203-- | /O(n)/ Breaks a 'Text' up into a list of 'Text's at
@@ -1574,10 +1207,9 @@ lines ps | null ps = []
15741207 | otherwise = h : if null t
15751208 then []
15761209 else lines (unsafeTail t)
1577− where (# h,t #) = span_ (/= '\n') ps
1210+ where (h,t) = span (/= '\n') ps
15781211{-# INLINE lines #-}
15791212
1580−{-
15811213-- | /O(n)/ Portably breaks a 'Text' up into a list of 'Text's at line
15821214-- boundaries.
15831215--
@@ -1585,6 +1217,7 @@ lines ps | null ps = []
15851217-- return immediately followed by a line feed, or a carriage return.
15861218-- This accounts for both Unix and Windows line ending conventions,
15871219-- and for the old convention used on Mac OS 9 and earlier.
1220+{-
15881221lines' :: Text -> [Text]
15891222lines' ps | null ps = []
15901223 | otherwise = h : case uncons t of
@@ -1611,7 +1244,8 @@ unwords = intercalate (singleton ' ')
16111244{-# INLINE unwords #-}
16121245
16131246-- | /O(n)/ The 'isPrefixOf' function takes two 'Text's and returns
1614−-- 'True' iff the first is a prefix of the second. Subject to fusion.
1247+-- 'True' iff the first is a prefix of the second. This function is
1248+-- subject to fusion.
16151249isPrefixOf :: Text -> Text -> Bool
16161250isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
16171251 alen <= blen && S.isPrefixOf (stream a) (stream b)
@@ -1620,6 +1254,8 @@ isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
16201254{-# RULES
16211255"TEXT isPrefixOf -> fused" [~1] forall s t.
16221256 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1257+"TEXT isPrefixOf -> unfused" [1] forall s t.
1258+ S.isPrefixOf (stream s) (stream t) = isPrefixOf s t
16231259 #-}
16241260
16251261-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
@@ -1632,124 +1268,13 @@ isSuffixOf a@(Text _aarr _aoff alen) b@(Text barr boff blen) =
16321268 | otherwise = Text barr (boff+d) alen
16331269{-# INLINE isSuffixOf #-}
16341270
1635−-- | /O(n+m)/ The 'isInfixOf' function takes two 'Text's and returns
1271+-- | /O(n)/ The 'isInfixOf' function takes two 'Text's and returns
16361272-- 'True' iff the first is contained, wholly and intact, anywhere
16371273-- within the second.
1638−--
1639−-- In (unlikely) bad cases, this function's time complexity degrades
1640−-- towards /O(n*m)/.
16411274isInfixOf :: Text -> Text -> Bool
1642−isInfixOf needle haystack
1643− | null needle = True
1644− | isSingleton needle = S.elem (unsafeHead needle) . S.stream $ haystack
1645− | otherwise = not . L.null . indices needle $ haystack
1646−{-# INLINE [1] isInfixOf #-}
1647−
1648−{-# RULES
1649−"TEXT isInfixOf/singleton -> S.elem/S.stream" [~1] forall n h.
1650− isInfixOf (singleton n) h = S.elem n (S.stream h)
1651− #-}
1652−
1653−-------------------------------------------------------------------------------
1654−-- * View patterns
1655−
1656−-- | /O(n)/ Return the suffix of the second string if its prefix
1657−-- matches the entire first string.
1658−--
1659−-- Examples:
1660−--
1661−-- > stripPrefix "foo" "foobar" == Just "bar"
1662−-- > stripPrefix "" "baz" == Just "baz"
1663−-- > stripPrefix "foo" "quux" == Nothing
1664−--
1665−-- This is particularly useful with the @ViewPatterns@ extension to
1666−-- GHC, as follows:
1667−--
1668−-- > {-# LANGUAGE ViewPatterns #-}
1669−-- > import Data.Text as T
1670−-- >
1671−-- > fnordLength :: Text -> Int
1672−-- > fnordLength (stripPrefix "fnord" -> Just suf) = T.length suf
1673−-- > fnordLength _ = -1
1674−stripPrefix :: Text -> Text -> Maybe Text
1675−stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1676− | p `isPrefixOf` t = Just $! text arr (off+plen) (len-plen)
1677− | otherwise = Nothing
1678−
1679−-- | /O(n)/ Find the longest non-empty common prefix of two strings
1680−-- and return it, along with the suffixes of each string at which they
1681−-- no longer match.
1682−--
1683−-- If the strings do not have a common prefix or either one is empty,
1684−-- this function returns 'Nothing'.
1685−--
1686−-- Examples:
1687−--
1688−-- > commonPrefixes "foobar" "fooquux" == Just ("foo","bar","quux")
1689−-- > commonPrefixes "veeble" "fetzer" == Nothing
1690−-- > commonPrefixes "" "baz" == Nothing
1691−commonPrefixes :: Text -> Text -> Maybe (Text,Text,Text)
1692−commonPrefixes t0@(Text arr0 off0 len0) t1@(Text arr1 off1 len1) = go 0 0
1693− where
1694− go !i !j | i < len0 && j < len1 && a == b = go (i+d0) (j+d1)
1695− | i > 0 = Just (Text arr0 off0 i,
1696− text arr0 (off0+i) (len0-i),
1697− text arr1 (off1+j) (len1-j))
1698− | otherwise = Nothing
1699− where Iter a d0 = iter t0 i
1700− Iter b d1 = iter t1 j
1701−
1702−-- | /O(n)/ Return the prefix of the second string if its suffix
1703−-- matches the entire first string.
1704−--
1705−-- Examples:
1706−--
1707−-- > stripSuffix "bar" "foobar" == Just "foo"
1708−-- > stripSuffix "" "baz" == Just "baz"
1709−-- > stripSuffix "foo" "quux" == Nothing
1710−--
1711−-- This is particularly useful with the @ViewPatterns@ extension to
1712−-- GHC, as follows:
1713−--
1714−-- > {-# LANGUAGE ViewPatterns #-}
1715−-- > import Data.Text as T
1716−-- >
1717−-- > quuxLength :: Text -> Int
1718−-- > quuxLength (stripSuffix "quux" -> Just pre) = T.length pre
1719−-- > quuxLength _ = -1
1720−stripSuffix :: Text -> Text -> Maybe Text
1721−stripSuffix p@(Text _arr _off plen) t@(Text arr off len)
1722− | p `isSuffixOf` t = Just $! text arr off (len-plen)
1723− | otherwise = Nothing
1724−
1725−-- | Add a list of non-negative numbers. Errors out on overflow.
1726−sumP :: String -> [Int] -> Int
1727−sumP fun = go 0
1728− where go !a (x:xs)
1729− | ax >= 0 = go ax xs
1730− | otherwise = overflowError fun
1731− where ax = a + x
1732− go a _ = a
1275+isInfixOf needle haystack = L.any (isPrefixOf needle) (tails haystack)
1276+{-# INLINE isInfixOf #-}
1277+-- TODO: a better implementation
17331278
17341279emptyError :: String -> a
1735−emptyError fun = P.error $ "Data.Text." ++ fun ++ ": empty input"
1736−
1737−overflowError :: String -> a
1738−overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"
1739−
1740−-- | /O(n)/ Make a distinct copy of the given string, sharing no
1741−-- storage with the original string.
1742−--
1743−-- As an example, suppose you read a large string, of which you need
1744−-- only a small portion. If you do not use 'copy', the entire original
1745−-- array will be kept alive in memory by the smaller string. Making a
1746−-- copy \"breaks the link\" to the original array, allowing it to be
1747−-- garbage collected if there are no other live references to it.
1748−copy :: Text -> Text
1749−copy (Text arr off len) = Text (A.run go) 0 len
1750− where
1751− go :: ST s (A.MArray s)
1752− go = do
1753− marr <- A.new len
1754− A.copyI marr 0 arr off len
1755− return marr
1280+emptyError fun = P.error ("Data.Text." ++ fun ++ ": empty input")