@@ -1,29 +1,48 @@
1−{-# LANGUAGE BangPatterns, CPP, Rank2Types #-}
1+{-# LANGUAGE BangPatterns, CPP, MagicHash, Rank2Types, UnboxedTuples #-}
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
39
410-- |
511-- Module : Data.Text
6−-- Copyright : (c) Tom Harper 2008-2009,
7−-- (c) Bryan O'Sullivan 2009,
8−-- (c) Duncan Coutts 2009
12+-- Copyright : (c) 2009, 2010, 2011, 2012 Bryan O'Sullivan,
13+-- (c) 2009 Duncan Coutts,
14+-- (c) 2008, 2009 Tom Harper
915--
1016-- License : BSD-style
11−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,
12−-- duncan@haskell.org
17+-- Maintainer : bos@serpentine.com
1318-- Stability : experimental
1419-- Portability : GHC
1520--
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.
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.
1927--
2028-- This module is intended to be imported @qualified@, to avoid name
2129-- clashes with "Prelude" functions, e.g.
2230--
2331-- > 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 >.
2437
2538module Data.Text
2639 (
40+ -- * Strict vs lazy types
41+ -- $strict
42+
43+ -- * Acceptable data
44+ -- $replacement
45+
2746 -- * Fusion
2847 -- $fusion
2948
@@ -47,6 +66,7 @@ module Data.Text
4766 , init
4867 , null
4968 , length
69+ , compareLength
5070
5171 -- * Transformations
5272 , map
@@ -61,6 +81,7 @@ module Data.Text
6181 , toCaseFold
6282 , toLower
6383 , toUpper
84+ , toTitle
6485
6586 -- ** Justification
6687 , justifyLeft
@@ -104,8 +125,11 @@ module Data.Text
104125
105126 -- ** Breaking strings
106127 , take
128+ , takeEnd
107129 , drop
130+ , dropEnd
108131 , takeWhile
132+ , takeWhileEnd
109133 , dropWhile
110134 , dropWhileEnd
111135 , dropAround
@@ -113,10 +137,10 @@ module Data.Text
113137 , stripStart
114138 , stripEnd
115139 , splitAt
116− , spanBy
140+ , breakOn
141+ , breakOnEnd
117142 , break
118− , breakEnd
119− , breakBy
143+ , span
120144 , group
121145 , groupBy
122146 , inits
@@ -124,8 +148,8 @@ module Data.Text
124148
125149 -- ** Breaking into many substrings
126150 -- $split
151+ , splitOn
127152 , split
128− , splitBy
129153 , chunksOf
130154
131155 -- ** Breaking into lines and words
@@ -141,164 +165,308 @@ module Data.Text
141165 , isInfixOf
142166
143167 -- ** View patterns
144− , prefixed
145− , suffixed
168+ , stripPrefix
169+ , stripSuffix
170+ , commonPrefixes
146171
147172 -- * Searching
148173 , filter
174+ , breakOnAll
149175 , find
150− , findBy
151− , partitionBy
176+ , partition
152177
153178 -- , findSubstring
154−
179+
155180 -- * Indexing
156181 -- $index
157182 , index
158183 , findIndex
159184 , count
160185
161− -- * Zipping and unzipping
186+ -- * Zipping
162187 , zip
163188 , zipWith
164189
165190 -- -* Ordered text
166191 -- , sort
192+
193+ -- * Low level operations
194+ , copy
195+ , unpackCString#
167196 ) where
168197
169−import Prelude (Char, Bool(..), Functor(..), Int, Maybe(..), String,
170− Eq(..), Ord(..), (++),
171− Read(..), Show(..),
172− (&&), (||), (+), (-), (.), ($), ($!), (>>), (*),
173− div, error, not, return, otherwise)
198+import Prelude (Char, Bool(..), Int, Maybe(..), String,
199+ Eq(..), Ord(..), Ordering(..), (++),
200+ Read(..),
201+ (&&), (||), (+), (-), (.), ($), ($!), (>>),
202+ not, return, otherwise, quot)
174203#if defined(HAVE_DEEPSEQ)
175−import Control.DeepSeq (NFData)
204+import Control.DeepSeq (NFData(rnf))
176205#endif
206+#if defined(ASSERTS)
177207import Control.Exception (assert)
178−import Data.Char (isSpace)
179−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))
180−#if __GLASGOW_HASKELL__ >= 612
181−import Data.Data (mkNoRepType)
182−#else
183−import Data.Data (mkNorepType)
184208#endif
209+import Data.Char (isSpace)
210+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,
211+ Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))
185212import Control.Monad (foldM)
186213import Control.Monad.ST (ST)
187214import qualified Data.Text.Array as A
188215import qualified Data.List as L
216+import Data.Binary (Binary(get, put))
189217import Data.Monoid (Monoid(..))
218+#if MIN_VERSION_base(4,9,0)
219+import Data.Semigroup (Semigroup(..))
220+#endif
190221import Data.String (IsString(..))
191−import qualified Data.Text.Fusion as S
192−import qualified Data.Text.Fusion.Common as S
193−import Data.Text.Fusion (stream, reverseStream, unstream)
194−import Data.Text.Internal (Text(..), empty, text, textP)
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#)
195229import qualified Prelude as P
196−import Data.Text.Unsafe (iter, iter_, reverseIter, unsafeHead, unsafeTail)
197−import Data.Text.UnsafeChar (unsafeChr)
198−import qualified Data.Text.Encoding.Utf16 as U16
199−import Data.Text.Search (indices)
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.
269+
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 >.)
200298
201299-- $fusion
202300--
203301-- Most of the functions in this module are subject to /fusion/,
204302-- meaning that a pipeline of such functions will usually allocate at
205303-- 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\".
206323
207324instance Eq Text where
208− t1 == t2 = stream t1 == stream t2
325+ Text arrA offA lenA == Text arrB offB lenB
326+ | lenA == lenB = A.equal arrA offA arrB offB lenA
327+ | otherwise = False
209328 {-# INLINE (==) #-}
210329
211330instance Ord Text where
212− compare t1 t2 = compare (stream t1) (stream t2)
213− {-# INLINE compare #-}
214−
215−instance Show Text where
216− showsPrec p ps r = showsPrec p (unpack ps) r
331+ compare = compareText
217332
218333instance Read Text where
219334 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
220335
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+
221344instance Monoid Text where
222345 mempty = empty
346+#if MIN_VERSION_base(4,9,0)
347+ mappend = (<>) -- future-proof definition
348+#else
223349 mappend = append
350+#endif
224351 mconcat = concat
225352
226353instance IsString Text where
227354 fromString = pack
228355
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+
229363#if defined(HAVE_DEEPSEQ)
230−instance NFData Text
364+instance NFData Text where rnf !_ = ()
231365#endif
232366
233−-- This instance preserves data abstraction at the cost of inefficiency.
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.
234376-- We omit reflection services for the sake of data abstraction.
235−--
236−-- This instance was created by copying the behavior of Data.Set and
237−-- Data.Map. If you feel a mistake has been made, please feel free to
238−-- submit improvements.
239377--
240−-- Original discussion is archived here:
241−
242−-- "could we get a Data instance for Data.Text.Text?"
243−-- http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93
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, ... >
244389
245390instance Data Text where
246391 gfoldl f z txt = z pack `f` (unpack txt)
247− toConstr _ = error "Data.Text.Text.toConstr"
248− gunfold _ _ = error "Data.Text.Text.gunfold"
249−#if __GLASGOW_HASKELL__ >= 612
250− dataTypeOf _ = mkNoRepType "Data.Text.Text"
251−#else
252− dataTypeOf _ = mkNorepType "Data.Text.Text"
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
253402#endif
254403
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+
255424-- -----------------------------------------------------------------------------
256425-- * Conversion to/from 'Text'
257426
258−-- | /O(n)/ Convert a 'String' into a 'Text'. Subject to fusion.
427+-- | /O(n)/ Convert a 'String' into a 'Text'. Subject to
428+-- fusion. Performs replacement on invalid scalar values.
259429pack :: String -> Text
260−pack = unstream . S.streamList
430+pack = unstream . S.map safe . S.streamList
261431{-# INLINE [1] pack #-}
262432
263−-- | /O(n)/ Convert a Text into a String. Subject to fusion.
264−unpack :: Text -> String
265−unpack = S.unstreamList . stream
266−{-# INLINE [1] unpack #-}
267−
268−-- | /O(1)/ Convert a character into a Text.
269−-- Subject to fusion.
270−singleton :: Char -> Text
271−singleton = unstream . S.singleton
272−{-# INLINE [1] singleton #-}
273−
274433-- -----------------------------------------------------------------------------
275434-- * Basic functions
276435
277436-- | /O(n)/ Adds a character to the front of a 'Text'. This function
278437-- is more costly than its 'List' counterpart because it requires
279−-- copying a new array. Subject to fusion.
438+-- copying a new array. Subject to fusion. Performs replacement on
439+-- invalid scalar values.
280440cons :: Char -> Text -> Text
281−cons c t = unstream (S.cons c (stream t))
441+cons c t = unstream (S.cons (safe c) (stream t))
282442{-# INLINE cons #-}
283443
444+infixr 5 `cons`
445+
284446-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
285447-- entire array in the process, unless fused. Subject to fusion.
448+-- Performs replacement on invalid scalar values.
286449snoc :: Text -> Char -> Text
287−snoc t c = unstream (S.snoc (stream t) c)
450+snoc t c = unstream (S.snoc (stream t) (safe c))
288451{-# INLINE snoc #-}
289452
290453-- | /O(n)/ Appends one 'Text' to the other by copying both of them
291454-- into a new 'Text'. Subject to fusion.
292455append :: Text -> Text -> Text
293−append (Text arr1 off1 len1) (Text arr2 off2 len2) = Text (A.run x) 0 len
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"
294461 where
295462 len = len1+len2
463+ x :: ST s (A.MArray s)
296464 x = do
297− arr <- A.unsafeNew len
298− copy arr 0 arr1 off1 len1
299− copy arr len1 arr2 off2 (len1+len2)
465+ arr <- A.new len
466+ A.copyI arr 0 arr1 off1 len1
467+ A.copyI arr len1 arr2 off2 len
300468 return arr
301−{-# INLINE append #-}
469+{-# NOINLINE append #-}
302470
303471{-# RULES
304472"TEXT append -> fused" [~1] forall t1 t2.
@@ -307,14 +475,6 @@ append (Text arr1 off1 len1) (Text arr2 off2 len2) = Text (A.run x) 0 len
307475 unstream (S.append (stream t1) (stream t2)) = append t1 t2
308476 #-}
309477
310−copy :: forall s. A.MArray s -> Int -> A.Array -> Int -> Int
311− -> ST s ()
312−copy dest i0 src j0 top = go i0 j0
313− where
314− go i j | i >= top = return ()
315− | otherwise = do A.unsafeWrite dest i (src `A.unsafeIndex` j)
316− go (i+1) (j+1)
317−
318478-- | /O(1)/ Returns the first character of a 'Text', which must be
319479-- non-empty. Subject to fusion.
320480head :: Text -> Char
@@ -326,21 +486,14 @@ head t = S.head (stream t)
326486uncons :: Text -> Maybe (Char, Text)
327487uncons t@(Text arr off len)
328488 | len <= 0 = Nothing
329− | otherwise = Just (c, textP arr (off+d) (len-d))
330− where (c,d) = iter t 0
489+ | otherwise = Just $ let !(Iter c d) = iter t 0
490+ in (c, text arr (off+d) (len-d))
331491{-# INLINE [1] uncons #-}
332492
333493-- | Lifted from Control.Arrow and specialized.
334494second :: (b -> c) -> (a,b) -> (a,c)
335495second f (a, b) = (a, f b)
336496
337−{-# RULES
338−"TEXT uncons -> fused" [~1] forall t.
339− uncons t = fmap (second unstream) (S.uncons (stream t))
340−"TEXT uncons -> unfused" [1] forall t.
341− fmap (second unstream) (S.uncons (stream t)) = uncons t
342− #-}
343−
344497-- | /O(1)/ Returns the last character of a 'Text', which must be
345498-- non-empty. Subject to fusion.
346499last :: Text -> Char
@@ -364,7 +517,7 @@ last (Text arr off len)
364517tail :: Text -> Text
365518tail t@(Text arr off len)
366519 | len <= 0 = emptyError "tail"
367− | otherwise = textP arr (off+d) (len-d)
520+ | otherwise = text arr (off+d) (len-d)
368521 where d = iter_ t 0
369522{-# INLINE [1] tail #-}
370523
@@ -379,8 +532,8 @@ tail t@(Text arr off len)
379532-- be non-empty. Subject to fusion.
380533init :: Text -> Text
381534init (Text arr off len) | len <= 0 = emptyError "init"
382− | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)
383− | otherwise = textP arr off (len-1)
535+ | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)
536+ | otherwise = text arr off (len-1)
384537 where
385538 n = A.unsafeIndex arr (off+len-1)
386539{-# INLINE [1] init #-}
@@ -395,7 +548,11 @@ init (Text arr off len) | len <= 0 = emptyError "init"
395548-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to
396549-- fusion.
397550null :: Text -> Bool
398−null (Text _arr _off len) = assert (len >= 0) $ len <= 0
551+null (Text _arr _off len) =
552+#if defined(ASSERTS)
553+ assert (len >= 0) $
554+#endif
555+ len <= 0
399556{-# INLINE [1] null #-}
400557
401558{-# RULES
@@ -415,27 +572,76 @@ isSingleton = S.isSingleton . stream
415572-- Subject to fusion.
416573length :: Text -> Int
417574length t = S.length (stream t)
418−{-# 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.
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+ #-}
419623
420624-- -----------------------------------------------------------------------------
421625-- * Transformations
422626-- | /O(n)/ 'map' @f@ @t@ is the 'Text' obtained by applying @f@ to
423−-- each element of @t@. Subject to fusion.
627+-- each element of @t@. Subject to fusion. Performs replacement on
628+-- invalid scalar values.
424629map :: (Char -> Char) -> Text -> Text
425−map f t = unstream (S.map f (stream t))
630+map f t = unstream (S.map (safe . f) (stream t))
426631{-# INLINE [1] map #-}
427632
428633-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
429634-- 'Text's and concatenates the list after interspersing the first
430635-- argument between each element of the list.
431636intercalate :: Text -> [Text] -> Text
432−intercalate t = concat . (L.intersperse t)
637+intercalate t = concat . (F.intersperse t)
433638{-# INLINE intercalate #-}
434639
435640-- | /O(n)/ The 'intersperse' function takes a character and places it
436−-- between the characters of a 'Text'. Subject to fusion.
641+-- between the characters of a 'Text'. Subject to fusion. Performs
642+-- replacement on invalid scalar values.
437643intersperse :: Char -> Text -> Text
438−intersperse c t = unstream (S.intersperse c (stream t))
644+intersperse c t = unstream (S.intersperse (safe c) (stream t))
439645{-# INLINE intersperse #-}
440646
441647-- | /O(n)/ Reverse the characters of a string. Subject to fusion.
@@ -443,13 +649,59 @@ reverse :: Text -> Text
443649reverse t = S.reverse (stream t)
444650{-# INLINE reverse #-}
445651
446−-- | /O(m+n)/ Replace every occurrence of one substring with another.
447−replace :: Text -- ^ Text to search for
448− -> Text -- ^ Replacement text
449− -> Text -- ^ Input text
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.
450680 -> Text
451−replace s d = intercalate d . split s
452−{-# INLINE replace #-}
681+ -- ^ @haystack@ in which to search.
682+ -> 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
453705
454706-- ----------------------------------------------------------------------------
455707-- ** Case conversions (folds)
@@ -464,10 +716,17 @@ replace s d = intercalate d . split s
464716-- case conversion rules. As a result, these functions may map one
465717-- input character to two or three output characters. For examples,
466718-- 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 >.
467725
468−-- | /O(n)/ Convert a string to folded case. This function is mainly
469−-- useful for performing caseless (also known as case insensitive)
470−-- string comparisons.
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.
471730--
472731-- A string @x@ is a caseless match for a string @y@ if and only if:
473732--
@@ -482,27 +741,56 @@ replace s d = intercalate d . split s
482741-- instead of itself.
483742toCaseFold :: Text -> Text
484743toCaseFold t = unstream (S.toCaseFold (stream t))
485−{-# INLINE [0] toCaseFold #-}
744+{-# INLINE toCaseFold #-}
486745
487746-- | /O(n)/ Convert a string to lower case, using simple case
488−-- conversion. The result string may be longer than the input string.
489−-- For instance, \"İ\" (Latin capital letter I with dot above,
490−-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069) followed
491−-- by \" ̇\" (combining dot above, U+0307).
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).
492753toLower :: Text -> Text
493754toLower t = unstream (S.toLower (stream t))
494755{-# INLINE toLower #-}
495756
496757-- | /O(n)/ Convert a string to upper case, using simple case
497−-- conversion. The result string may be longer than the input string.
498−-- For instance, the German \"ß\" (eszett, U+00DF) maps to the
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
499762-- two-letter sequence \"SS\".
500763toUpper :: Text -> Text
501764toUpper t = unstream (S.toUpper (stream t))
502765{-# INLINE toUpper #-}
503766
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+
504789-- | /O(n)/ Left-justify a string to the given length, using the
505−-- specified fill character on the right. Subject to fusion. Examples:
790+-- specified fill character on the right. Subject to fusion.
791+-- Performs replacement on invalid scalar values.
792+--
793+-- Examples:
506794--
507795-- > justifyLeft 7 'x' "foo" == "fooxxxx"
508796-- > justifyLeft 3 'x' "foobar" == "foobar"
@@ -521,7 +809,10 @@ justifyLeft k c t
521809 #-}
522810
523811-- | /O(n)/ Right-justify a string to the given length, using the
524−-- specified fill character on the left. Examples:
812+-- specified fill character on the left. Performs replacement on
813+-- invalid scalar values.
814+--
815+-- Examples:
525816--
526817-- > justifyRight 7 'x' "bar" == "xxxxbar"
527818-- > justifyRight 3 'x' "foobar" == "foobar"
@@ -532,8 +823,11 @@ justifyRight k c t
532823 where len = length t
533824{-# INLINE justifyRight #-}
534825
535−-- | /O(n)/ Center a string to the given length, using the
536−-- specified fill character on either side. Examples:
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:
537831--
538832-- > center 8 'x' "HS" = "xxxHSxxx"
539833center :: Int -> Char -> Text -> Text
@@ -542,7 +836,7 @@ center k c t
542836 | otherwise = replicateChar l c `append` t `append` replicateChar r c
543837 where len = length t
544838 d = k - len
545− r = d `div` 2
839+ r = d `quot` 2
546840 l = d - r
547841{-# INLINE center #-}
548842
@@ -560,12 +854,12 @@ transpose ts = P.map pack (L.transpose (P.map unpack ts))
560854-- (typically the left-identity of the operator), and a 'Text',
561855-- reduces the 'Text' using the binary operator, from left to right.
562856-- Subject to fusion.
563−foldl :: (b -> Char -> b) -> b -> Text -> b
857+foldl :: (a -> Char -> a) -> a -> Text -> a
564858foldl f z t = S.foldl f z (stream t)
565859{-# INLINE foldl #-}
566860
567861-- | /O(n)/ A strict version of 'foldl'. Subject to fusion.
568−foldl' :: (b -> Char -> b) -> b -> Text -> b
862+foldl' :: (a -> Char -> a) -> a -> Text -> a
569863foldl' f z t = S.foldl' f z (stream t)
570864{-# INLINE foldl' #-}
571865
@@ -584,12 +878,12 @@ foldl1' f t = S.foldl1' f (stream t)
584878-- (typically the right-identity of the operator), and a 'Text',
585879-- reduces the 'Text' using the binary operator, from right to left.
586880-- Subject to fusion.
587−foldr :: (Char -> b -> b) -> b -> Text -> b
881+foldr :: (Char -> a -> a) -> a -> Text -> a
588882foldr f z t = S.foldr f z (stream t)
589883{-# INLINE foldr #-}
590884
591885-- | /O(n)/ A variant of 'foldr' that has no starting value argument,
592−-- and thust must be applied to a non-empty 'Text'. Subject to
886+-- and thus must be applied to a non-empty 'Text'. Subject to
593887-- fusion.
594888foldr1 :: (Char -> Char -> Char) -> Text -> Char
595889foldr1 f t = S.foldr1 f (stream t)
@@ -600,14 +894,19 @@ foldr1 f t = S.foldr1 f (stream t)
600894
601895-- | /O(n)/ Concatenate a list of 'Text's.
602896concat :: [Text] -> Text
603−concat ts = Text (A.run go) 0 len
897+concat ts = case ts' of
898+ [] -> empty
899+ [t] -> t
900+ _ -> Text (A.run go) 0 len
604901 where
605− len = L.sum (L.map (\(Text _ _ l) -> l) ts)
902+ ts' = L.filter (not . null) ts
903+ len = sumP "concat" $ L.map lengthWord16 ts'
904+ go :: ST s (A.MArray s)
606905 go = do
607− arr <- A.unsafeNew len
608− let step i (Text a o l) = let j = i + l in copy arr i a o j >> return j
609− foldM step 0 ts >> return arr
610−{-# INLINE concat #-}
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
611910
612911-- | /O(n)/ Map a function over a 'Text' that results in a 'Text', and
613912-- concatenate the results.
@@ -616,13 +915,13 @@ concatMap f = concat . foldr ((:) . f) []
616915{-# INLINE concatMap #-}
617916
618917-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
619−-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.
918+-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.
620919any :: (Char -> Bool) -> Text -> Bool
621920any p t = S.any p (stream t)
622921{-# INLINE any #-}
623922
624923-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
625−-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.
924+-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.
626925all :: (Char -> Bool) -> Text -> Bool
627926all p t = S.all p (stream t)
628927{-# INLINE all #-}
@@ -644,6 +943,7 @@ minimum t = S.minimum (stream t)
644943
645944-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
646945-- successive reduced values from the left. Subject to fusion.
946+-- Performs replacement on invalid scalar values.
647947--
648948-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
649949--
@@ -651,11 +951,13 @@ minimum t = S.minimum (stream t)
651951--
652952-- > last (scanl f z xs) == foldl f z xs.
653953scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
654−scanl f z t = unstream (S.scanl f z (stream t))
954+scanl f z t = unstream (S.scanl g z (stream t))
955+ where g a b = safe (f a b)
655956{-# INLINE scanl #-}
656957
657958-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
658−-- value argument. Subject to fusion.
959+-- value argument. Subject to fusion. Performs replacement on
960+-- invalid scalar values.
659961--
660962-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
661963scanl1 :: (Char -> Char -> Char) -> Text -> Text
@@ -663,40 +965,42 @@ scanl1 f t | null t = empty
663965 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
664966{-# INLINE scanl1 #-}
665967
666−-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'.
968+-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'. Performs
969+-- replacement on invalid scalar values.
667970--
668971-- > scanr f v == reverse . scanl (flip f) v . reverse
669972scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
670−scanr f z = S.reverse . S.reverseScanr f z . reverseStream
973+scanr f z = S.reverse . S.reverseScanr g z . reverseStream
974+ where g a b = safe (f a b)
671975{-# INLINE scanr #-}
672976
673977-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
674−-- value argument. Subject to fusion.
978+-- value argument. Subject to fusion. Performs replacement on
979+-- invalid scalar values.
675980scanr1 :: (Char -> Char -> Char) -> Text -> Text
676981scanr1 f t | null t = empty
677982 | otherwise = scanr f (last t) (init t)
678983{-# INLINE scanr1 #-}
679984
680−-- | /O(n)/ Like a combination of 'map' and 'foldl'. Applies a
985+-- | /O(n)/ Like a combination of 'map' and 'foldl''. Applies a
681986-- function to each element of a 'Text', passing an accumulating
682−-- parameter from left to right, and returns a final 'Text'.
987+-- parameter from left to right, and returns a final 'Text'. Performs
988+-- replacement on invalid scalar values.
683989mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
684−mapAccumL f s t = case uncons t of
685− Nothing -> (s, empty)
686− Just (x, xs) -> (s'', cons y ys)
687− where (s', y ) = f s x
688− (s'',ys) = mapAccumL f s' xs
990+mapAccumL f z0 = S.mapAccumL g z0 . stream
991+ where g a b = second safe (f a b)
992+{-# INLINE mapAccumL #-}
689993
690994-- | The 'mapAccumR' function behaves like a combination of 'map' and
691−-- 'foldr'; it applies a function to each element of a 'Text', passing
692−-- an accumulating parameter from right to left, and returning a final
693−-- value of this accumulator together with the new 'Text'.
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.
6941000mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
695−mapAccumR f s t = case uncons t of
696− Nothing -> (s, empty)
697− Just (x, xs) -> (s'', cons y ys)
698− where (s'',y ) = f s' x
699− (s', ys) = mapAccumR f s xs
1001+mapAccumR f z0 = second reverse . S.mapAccumL g z0 . reverseStream
1002+ where g a b = second safe (f a b)
1003+{-# INLINE mapAccumR #-}
7001004
7011005-- -----------------------------------------------------------------------------
7021006-- ** Generating and unfolding 'Text's
@@ -705,17 +1009,18 @@ mapAccumR f s t = case uncons t of
7051009-- @t@ repeated @n@ times.
7061010replicate :: Int -> Text -> Text
7071011replicate n t@(Text a o l)
708− | n <= 0 || l <= 0 = empty
709− | n == 1 = t
710− | isSingleton t = replicateChar n (unsafeHead t)
711− | otherwise = Text (A.run x) 0 len
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
7121016 where
713− len = l * n
1017+ len = l `mul` n
1018+ x :: ST s (A.MArray s)
7141019 x = do
715− arr <- A.unsafeNew len
1020+ arr <- A.new len
7161021 let loop !d !i | i >= n = return arr
7171022 | otherwise = let m = d + l
718− in copy arr d a o m >> loop m (i+1)
1023+ in A.copyI arr d a o m >> loop m (i+1)
7191024 loop 0 0
7201025{-# INLINE [1] replicate #-}
7211026
@@ -727,7 +1032,7 @@ replicate n t@(Text a o l)
7271032-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the
7281033-- value of every element. Subject to fusion.
7291034replicateChar :: Int -> Char -> Text
730−replicateChar n c = unstream (S.replicateCharI n c)
1035+replicateChar n c = unstream (S.replicateCharI n (safe c))
7311036{-# INLINE replicateChar #-}
7321037
7331038-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
@@ -736,9 +1041,9 @@ replicateChar n c = unstream (S.replicateCharI n c)
7361041-- returns 'Nothing' if it is done producing the 'Text', otherwise
7371042-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
7381043-- string, and @b@ is the seed value for further production. Subject
739−-- to fusion.
1044+-- to fusion. Performs replacement on invalid scalar values.
7401045unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
741−unfoldr f s = unstream (S.unfoldr f s)
1046+unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
7421047{-# INLINE unfoldr #-}
7431048
7441049-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
@@ -746,9 +1051,9 @@ unfoldr f s = unstream (S.unfoldr f s)
7461051-- first argument to 'unfoldrN'. This function is more efficient than
7471052-- 'unfoldr' when the maximum length of the result is known and
7481053-- correct, otherwise its performance is similar to 'unfoldr'. Subject
749−-- to fusion.
1054+-- to fusion. Performs replacement on invalid scalar values.
7501055unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
751−unfoldrN n f s = unstream (S.unfoldrN n f s)
1056+unfoldrN n f s = unstream (S.unfoldrN n (firstf safe . f) s)
7521057{-# INLINE unfoldrN #-}
7531058
7541059-- -----------------------------------------------------------------------------
@@ -761,14 +1066,16 @@ take :: Int -> Text -> Text
7611066take n t@(Text arr off len)
7621067 | n <= 0 = empty
7631068 | n >= len = t
764− | otherwise = Text arr off (loop 0 0)
765− where
766− loop !i !cnt
767− | i >= len || cnt >= n = i
768− | otherwise = loop (i+d) (cnt+1)
769− where d = iter_ t i
1069+ | otherwise = text arr off (iterN n t)
7701070{-# INLINE [1] take #-}
7711071
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+
7721079{-# RULES
7731080"TEXT take -> fused" [~1] forall n t.
7741081 take n t = unstream (S.take n (stream t))
@@ -776,6 +1083,27 @@ take n t@(Text arr off len)
7761083 unstream (S.take n (stream t)) = take n t
7771084 #-}
7781085
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+
7791107-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
7801108-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
7811109-- is greater than the length of the 'Text'. Subject to fusion.
@@ -783,11 +1111,8 @@ drop :: Int -> Text -> Text
7831111drop n t@(Text arr off len)
7841112 | n <= 0 = t
7851113 | n >= len = empty
786− | otherwise = loop 0 0
787− where loop !i !cnt
788− | i >= len || cnt >= n = Text arr (off+i) (len-i)
789− | otherwise = loop (i+d) (cnt+1)
790− where d = iter_ t i
1114+ | otherwise = text arr (off+i) (len-i)
1115+ where i = iterN n t
7911116{-# INLINE [1] drop #-}
7921117
7931118{-# RULES
@@ -797,6 +1122,18 @@ drop n t@(Text arr off len)
7971122 unstream (S.drop n (stream t)) = drop n t
7981123 #-}
7991124
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+
8001137-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
8011138-- returns the longest prefix (possibly empty) of elements that
8021139-- satisfy @p@. Subject to fusion.
@@ -804,8 +1141,8 @@ takeWhile :: (Char -> Bool) -> Text -> Text
8041141takeWhile p t@(Text arr off len) = loop 0
8051142 where loop !i | i >= len = t
8061143 | p c = loop (i+d)
807− | otherwise = textP arr off i
808− where (c,d) = iter t i
1144+ | otherwise = text arr off i
1145+ where Iter c d = iter t i
8091146{-# INLINE [1] takeWhile #-}
8101147
8111148{-# RULES
@@ -815,6 +1152,27 @@ takeWhile p t@(Text arr off len) = loop 0
8151152 unstream (S.takeWhile p (stream t)) = takeWhile p t
8161153 #-}
8171154
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+
8181176-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
8191177-- 'takeWhile' @p@ @t@. Subject to fusion.
8201178dropWhile :: (Char -> Bool) -> Text -> Text
@@ -822,7 +1180,7 @@ dropWhile p t@(Text arr off len) = loop 0 0
8221180 where loop !i !l | l >= len = empty
8231181 | p c = loop (i+d) (l+d)
8241182 | otherwise = Text arr (off+i) (len-l)
825− where (c,d) = iter t i
1183+ where Iter c d = iter t i
8261184{-# INLINE [1] dropWhile #-}
8271185
8281186{-# RULES
@@ -833,8 +1191,9 @@ dropWhile p t@(Text arr off len) = loop 0 0
8331191 #-}
8341192
8351193-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
836−-- dropping characters that fail the predicate @p@ from the end of
1194+-- dropping characters that satisfy the predicate @p@ from the end of
8371195-- @t@. Subject to fusion.
1196+--
8381197-- Examples:
8391198--
8401199-- > dropWhileEnd (=='.') "foo..." == "foo"
@@ -854,7 +1213,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len
8541213 #-}
8551214
8561215-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
857−-- dropping characters that fail the predicate @p@ from both the
1216+-- dropping characters that satisfy the predicate @p@ from both the
8581217-- beginning and end of @t@. Subject to fusion.
8591218dropAround :: (Char -> Bool) -> Text -> Text
8601219dropAround p = dropWhile p . dropWhileEnd p
@@ -889,31 +1248,23 @@ splitAt :: Int -> Text -> (Text, Text)
8891248splitAt n t@(Text arr off len)
8901249 | n <= 0 = (empty, t)
8911250 | n >= len = (t, empty)
892− | otherwise = (Text arr off k, Text arr (off+k) (len-k))
893− where k = loop 0 0
894− loop !i !cnt
895− | i >= len || cnt >= n = i
896− | otherwise = loop (i+d) (cnt+1)
897− where d = iter_ t i
898−{-# INLINE splitAt #-}
1251+ | otherwise = let k = iterN n t
1252+ in (text arr off k, text arr (off+k) (len-k))
8991253
900−-- | /O(n)/ 'spanBy', applied to a predicate @p@ and text @t@, returns
1254+-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
9011255-- a pair whose first element is the longest prefix (possibly empty)
9021256-- of @t@ of elements that satisfy @p@, and whose second is the
9031257-- remainder of the list.
904−spanBy :: (Char -> Bool) -> Text -> (Text, Text)
905−spanBy p t@(Text arr off len) = (textP arr off k, textP arr (off+k) (len-k))
906− where k = loop 0
907− loop !i | i >= len || not (p c) = i
908− | otherwise = loop (i+d)
909− where (c,d) = iter t i
910−{-# INLINE spanBy #-}
911−
912−-- | /O(n)/ 'breakBy' is like 'spanBy', but the prefix returned is
1258+span :: (Char -> Bool) -> Text -> (Text, Text)
1259+span p t = case span_ p t of
1260+ (# hd,tl #) -> (hd,tl)
1261+{-# INLINE span #-}
1262+
1263+-- | /O(n)/ 'break' is like 'span', but the prefix returned is
9131264-- over elements that fail the predicate @p@.
914−breakBy :: (Char -> Bool) -> Text -> (Text, Text)
915−breakBy p = spanBy (not . p)
916−{-# INLINE breakBy #-}
1265+break :: (Char -> Bool) -> Text -> (Text, Text)
1266+break p = span (not . p)
1267+{-# INLINE break #-}
9171268
9181269-- | /O(n)/ Group characters in a string according to a predicate.
9191270groupBy :: (Char -> Char -> Bool) -> Text -> [Text]
@@ -922,7 +1273,7 @@ groupBy p = loop
9221273 loop t@(Text arr off len)
9231274 | null t = []
9241275 | otherwise = text arr off n : loop (text arr (off+n) (len-n))
925− where (c,d) = iter t 0
1276+ where Iter c d = iter t 0
9261277 n = d + findAIndexOrEnd (not . p c) (Text arr (off+d) (len-d))
9271278
9281279-- | Returns the /array/ index (in units of 'Word16') at which a
@@ -932,8 +1283,8 @@ findAIndexOrEnd :: (Char -> Bool) -> Text -> Int
9321283findAIndexOrEnd q t@(Text _arr _off len) = go 0
9331284 where go !i | i >= len || q c = i
9341285 | otherwise = go (i+d)
935− where (c,d) = iter t i
936−
1286+ where Iter c d = iter t i
1287+
9371288-- | /O(n)/ Group characters in a string by equality.
9381289group :: Text -> [Text]
9391290group = groupBy (==)
@@ -957,36 +1308,43 @@ tails t | null t = [empty]
9571308-- copies to create substrings; they just construct new 'Text's that
9581309-- are slices of the original.
9591310
960−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first
961−-- 'Text' argument, consuming the delimiter. An empty delimiter is
962−-- invalid, and will cause an error to be raised.
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.
9631314--
9641315-- Examples:
9651316--
966−-- > split "\r\n" "a\r\nb\r\nd\r\ne" == ["a","b","d","e"]
967−-- > split "aaa" "aaaXaaaXaaaXaaa" == ["","X","X","X",""]
968−-- > split "x" "x" == ["",""]
969−--
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+--
9701321-- and
9711322--
972−-- > intercalate s . split s == id
973−-- > split (singleton c) == splitBy (==c)
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.)
9741327--
9751328-- In (unlikely) bad cases, this function's time complexity degrades
9761329-- towards /O(n*m)/.
977−split :: Text -> Text -> [Text]
978−split pat@(Text _ _ l) src@(Text arr off len)
979− | l <= 0 = emptyError "split"
980− | isSingleton pat = splitBy (== unsafeHead pat) src
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
9811339 | otherwise = go 0 (indices pat src)
9821340 where
983− go !s (x:xs) = textP arr (s+off) (x-s) : go (x+l) xs
984− go s _ = [textP arr (s+off) (len-s)]
985−{-# INLINE [1] split #-}
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 #-}
9861344
9871345{-# RULES
988−"TEXT split/singleton -> splitBy/==" [~1] forall c t.
989− split (singleton c) t = splitBy (==c) t
1346+"TEXT splitOn/singleton -> split/==" [~1] forall c t.
1347+ splitOn (singleton c) t = split (==c) t
9901348 #-}
9911349
9921350-- | /O(n)/ Splits a 'Text' into components delimited by separators,
@@ -994,15 +1352,15 @@ split pat@(Text _ _ l) src@(Text arr off len)
9941352-- resulting components do not contain the separators. Two adjacent
9951353-- separators result in an empty component in the output. eg.
9961354--
997−-- > splitBy (=='a') "aabbaca" == ["","","bb","c",""]
998−-- > splitBy (=='a') "" == [""]
999−splitBy :: (Char -> Bool) -> Text -> [Text]
1000−splitBy _ t@(Text _off _arr 0) = [t]
1001−splitBy p t = loop t
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
10021360 where loop s | null s' = [l]
10031361 | otherwise = l : loop (unsafeTail s')
1004− where (l, s') = breakBy p s
1005−{-# INLINE splitBy #-}
1362+ where (# l, s' #) = span_ (not . p) s
1363+{-# INLINE split #-}
10061364
10071365-- | /O(n)/ Splits a 'Text' into components of length @k@. The last
10081366-- element may be shorter than the other chunks, depending on the
@@ -1024,21 +1382,21 @@ chunksOf k = go
10241382-------------------------------------------------------------------------------
10251383-- ** Searching with a predicate
10261384
1027−-- | /O(n)/ The 'findBy' function takes a predicate and a 'Text', and
1028−-- returns the first element in matching the predicate, or 'Nothing'
1029−-- if there is no such element.
1030−findBy :: (Char -> Bool) -> Text -> Maybe Char
1031−findBy p t = S.findBy p (stream t)
1032−{-# INLINE findBy #-}
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.
1388+find :: (Char -> Bool) -> Text -> Maybe Char
1389+find p t = S.findBy p (stream t)
1390+{-# INLINE find #-}
10331391
1034−-- | /O(n)/ The 'partitionBy' function takes a predicate and a 'Text',
1392+-- | /O(n)/ The 'partition' function takes a predicate and a 'Text',
10351393-- and returns the pair of 'Text's with elements which do and do not
10361394-- satisfy the predicate, respectively; i.e.
10371395--
1038−-- > partitionBy p t == (filter p t, filter (not . p) t)
1039−partitionBy :: (Char -> Bool) -> Text -> (Text, Text)
1040−partitionBy p t = (filter p t, filter (not . p) t)
1041−{-# INLINE partitionBy #-}
1396+-- > partition p t == (filter p t, filter (not . p) t)
1397+partition :: (Char -> Bool) -> Text -> (Text, Text)
1398+partition p t = (filter p t, filter (not . p) t)
1399+{-# INLINE partition #-}
10421400
10431401-- | /O(n)/ 'filter', applied to a predicate and a 'Text',
10441402-- returns a 'Text' containing those characters that satisfy the
@@ -1054,39 +1412,40 @@ filter p t = unstream (S.filter p (stream t))
10541412--
10551413-- Examples:
10561414--
1057−-- > break "::" "a::b::c" ==> ("a", "::b::c")
1058−-- > break "/" "foobar" ==> ("foobar", "")
1415+-- > breakOn "::" "a::b::c" ==> ("a", "::b::c")
1416+-- > breakOn "/" "foobar" ==> ("foobar", "")
10591417--
10601418-- Laws:
10611419--
10621420-- > append prefix match == haystack
1063−-- > where (prefix, match) = break needle haystack
1421+-- > where (prefix, match) = breakOn needle haystack
10641422--
10651423-- If you need to break a string by a substring repeatedly (e.g. you
1066−-- want to break on every instance of a substring), use 'find'
1424+-- want to break on every instance of a substring), use 'breakOnAll'
10671425-- instead, as it has lower startup overhead.
10681426--
10691427-- In (unlikely) bad cases, this function's time complexity degrades
10701428-- towards /O(n*m)/.
1071−break :: Text -> Text -> (Text, Text)
1072−break pat src@(Text arr off len)
1073− | null pat = emptyError "break"
1429+breakOn :: Text -> Text -> (Text, Text)
1430+breakOn pat src@(Text arr off len)
1431+ | null pat = emptyError "breakOn"
10741432 | otherwise = case indices pat src of
10751433 [] -> (src, empty)
1076− (x:_) -> (textP arr off x, textP arr (off+x) (len-x))
1077−{-# INLINE break #-}
1434+ (x:_) -> (text arr off x, text arr (off+x) (len-x))
1435+{-# INLINE breakOn #-}
10781436
1079−-- | /O(n+m)/ Similar to 'break', but searches from the end of the string.
1437+-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
1438+-- string.
10801439--
10811440-- The first element of the returned tuple is the prefix of @haystack@
10821441-- up to and including the last match of @needle@. The second is the
10831442-- remainder of @haystack@, following the match.
10841443--
1085−-- > breakEnd "::" "a::b::c" ==> ("a::b::", "c")
1086−breakEnd :: Text -> Text -> (Text, Text)
1087−breakEnd pat src = let (a,b) = break (reverse pat) (reverse src)
1088− in (reverse b, reverse a)
1089−{-# INLINE breakEnd #-}
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 #-}
10901449
10911450-- | /O(n+m)/ Find all non-overlapping instances of @needle@ in
10921451-- @haystack@. Each element of the returned list consists of a pair:
@@ -1097,25 +1456,25 @@ breakEnd pat src = let (a,b) = break (reverse pat) (reverse src)
10971456--
10981457-- Examples:
10991458--
1100−-- > find "::" ""
1459+-- > breakOnAll "::" ""
11011460-- > ==> []
1102−-- > find "/" "a/b/c/"
1461+-- > breakOnAll "/" "a/b/c/"
11031462-- > ==> [("a", "/b/c/"), ("a/b", "/c/"), ("a/b/c", "/")]
11041463--
11051464-- In (unlikely) bad cases, this function's time complexity degrades
11061465-- towards /O(n*m)/.
11071466--
11081467-- The @needle@ parameter may not be empty.
1109−find :: Text -- ^ @needle@ to search for
1110− -> Text -- ^ @haystack@ in which to search
1111− -> [(Text, Text)]
1112−find pat src@(Text arr off slen)
1113− | null pat = emptyError "find"
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"
11141473 | otherwise = L.map step (indices pat src)
11151474 where
11161475 step x = (chunk 0 x, chunk x (slen-x))
1117− chunk !n !l = textP arr (n+off) l
1118−{-# INLINE find #-}
1476+ chunk !n !l = text arr (n+off) l
1477+{-# INLINE breakOnAll #-}
11191478
11201479-------------------------------------------------------------------------------
11211480-- ** Indexing 'Text's
@@ -1134,7 +1493,7 @@ find pat src@(Text arr off slen)
11341493-- For example, suppose you have a string that you want to split on
11351494-- the substring @\"::\"@, such as @\"foo::bar::quux\"@. Instead of
11361495-- searching for the index of @\"::\"@ and taking the substrings
1137−-- before and after that index, you would instead use @find \"::\"@.
1496+-- before and after that index, you would instead use @breakOnAll \"::\"@.
11381497
11391498-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
11401499index :: Text -> Int -> Char
@@ -1181,13 +1540,15 @@ countChar c t = S.countChar c (stream t)
11811540-- equivalent to a pair of 'unpack' operations.
11821541zip :: Text -> Text -> [(Char,Char)]
11831542zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1184−{-# INLINE [0] zip #-}
1543+{-# INLINE zip #-}
11851544
11861545-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
11871546-- given as the first argument, instead of a tupling function.
1547+-- Performs replacement on invalid scalar values.
11881548zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1189−zipWith f t1 t2 = unstream (S.zipWith f (stream t1) (stream t2))
1190−{-# INLINE [0] zipWith #-}
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 #-}
11911552
11921553-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
11931554-- representing white space.
@@ -1203,7 +1564,7 @@ words t@(Text arr off len) = loop 0 0
12031564 then loop (start+1) (start+1)
12041565 else Text arr (start+off) (n-start) : loop (n+d) (n+d)
12051566 | otherwise = loop start (n+d)
1206− where (c,d) = iter t n
1567+ where Iter c d = iter t n
12071568{-# INLINE words #-}
12081569
12091570-- | /O(n)/ Breaks a 'Text' up into a list of 'Text's at
@@ -1213,7 +1574,7 @@ lines ps | null ps = []
12131574 | otherwise = h : if null t
12141575 then []
12151576 else lines (unsafeTail t)
1216− where (h,t) = spanBy (/= '\n') ps
1577+ where (# h,t #) = span_ (/= '\n') ps
12171578{-# INLINE lines #-}
12181579
12191580{-
@@ -1250,8 +1611,7 @@ unwords = intercalate (singleton ' ')
12501611{-# INLINE unwords #-}
12511612
12521613-- | /O(n)/ The 'isPrefixOf' function takes two 'Text's and returns
1253−-- 'True' iff the first is a prefix of the second. This function is
1254−-- subject to fusion.
1614+-- 'True' iff the first is a prefix of the second. Subject to fusion.
12551615isPrefixOf :: Text -> Text -> Bool
12561616isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
12571617 alen <= blen && S.isPrefixOf (stream a) (stream b)
@@ -1260,8 +1620,6 @@ isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
12601620{-# RULES
12611621"TEXT isPrefixOf -> fused" [~1] forall s t.
12621622 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1263−"TEXT isPrefixOf -> unfused" [1] forall s t.
1264− S.isPrefixOf (stream s) (stream t) = isPrefixOf s t
12651623 #-}
12661624
12671625-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
@@ -1295,13 +1653,14 @@ isInfixOf needle haystack
12951653-------------------------------------------------------------------------------
12961654-- * View patterns
12971655
1298−-- | /O(n)/ Returns the suffix of the second string if its prefix
1299−-- matches the first.
1656+-- | /O(n)/ Return the suffix of the second string if its prefix
1657+-- matches the entire first string.
13001658--
13011659-- Examples:
13021660--
1303−-- > prefixed "foo" "foobar" == Just "bar"
1304−-- > prefixed "foo" "quux" == Nothing
1661+-- > stripPrefix "foo" "foobar" == Just "bar"
1662+-- > stripPrefix "" "baz" == Just "baz"
1663+-- > stripPrefix "foo" "quux" == Nothing
13051664--
13061665-- This is particularly useful with the @ViewPatterns@ extension to
13071666-- GHC, as follows:
@@ -1310,20 +1669,44 @@ isInfixOf needle haystack
13101669-- > import Data.Text as T
13111670-- >
13121671-- > fnordLength :: Text -> Int
1313−-- > fnordLength (prefixed "fnord" -> Just suf) = T.length suf
1314−-- > fnordLength _ = -1
1315−prefixed :: Text -> Text -> Maybe Text
1316−prefixed p@(Text _arr _off plen) t@(Text arr off len)
1317− | p `isPrefixOf` t = Just $! textP arr (off+plen) (len-plen)
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)
13181677 | otherwise = Nothing
13191678
1320−-- | /O(n)/ Returns the prefix of the second string if its suffix
1321−-- matches the first.
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'.
13221685--
13231686-- Examples:
13241687--
1325−-- > suffixed "bar" "foobar" == Just "foo"
1326−-- > suffixed "foo" "quux" == Nothing
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
13271710--
13281711-- This is particularly useful with the @ViewPatterns@ extension to
13291712-- GHC, as follows:
@@ -1332,12 +1715,41 @@ prefixed p@(Text _arr _off plen) t@(Text arr off len)
13321715-- > import Data.Text as T
13331716-- >
13341717-- > quuxLength :: Text -> Int
1335−-- > quuxLength (suffixed "quux" -> Just pre) = T.length pre
1336−-- > quuxLength _ = -1
1337−suffixed :: Text -> Text -> Maybe Text
1338−suffixed p@(Text _arr _off plen) t@(Text arr off len)
1339− | p `isSuffixOf` t = Just $! textP arr off (len-plen)
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)
13401723 | otherwise = Nothing
13411724
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
1733+
13421734emptyError :: String -> a
1343−emptyError fun = P.error ("Data.Text." ++ fun ++ ": empty input")
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