@@ -1,48 +1,29 @@
1−{-# LANGUAGE BangPatterns, CPP, MagicHash, Rank2Types, UnboxedTuples #-}
1+{-# LANGUAGE BangPatterns #-}
2{-# OPTIONS_GHC -fno-warn-orphans #-}
2{-# OPTIONS_GHC -fno-warn-orphans #-}
3−#if __GLASGOW_HASKELL__ >= 702
4−{-# LANGUAGE Trustworthy #-}
6−#if __GLASGOW_HASKELL__ >= 708
7−{-# LANGUAGE TypeFamilies #-}
11-- Module : Data.Text
5-- Module : Data.Text
6+-- Copyright : (c) Tom Harper 2008-2009,
12−-- Copyright : (c) 2009, 2010, 2011, 2012 Bryan O'Sullivan,
7+-- (c) Bryan O'Sullivan 2009,
13−-- (c) 2009 Duncan Coutts,
8+-- (c) Duncan Coutts 2009
14−-- (c) 2008, 2009 Tom Harper
16-- License : BSD-style
10-- License : BSD-style
17−-- Maintainer : bos@serpentine.com
11+-- Maintainer : bos@serpentine.com, rtharper@aftereternity.co.uk,
18-- Stability : experimental
13-- Stability : experimental
19-- Portability : GHC
14-- Portability : GHC
21−-- A time and space-efficient implementation of Unicode text.
16+-- A time and space-efficient implementation of Unicode text using
22−-- Suitable for performance critical use, both in terms of large data
17+-- packed Word16 arrays. Suitable for performance critical use, both
23−-- quantities and high speed.
18+-- in terms of large data quantities and high speed.
25−-- /Note/: Read below the synopsis for important notes on the use of
28-- This module is intended to be imported @qualified@, to avoid name
20-- This module is intended to be imported @qualified@, to avoid name
29-- clashes with "Prelude" functions, e.g.
21-- clashes with "Prelude" functions, e.g.
31-- > import qualified Data.Text as T
23-- > import qualified Data.Text as T
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 >.
38module Data.Text
25module Data.Text
40− -- * Strict vs lazy types
46 -- * Fusion
27 -- * Fusion
47 -- $fusion
28 -- $fusion
@@ -66,7 +47,6 @@ module Data.Text
71 -- * Transformations
51 -- * Transformations
@@ -74,19 +54,12 @@ module Data.Text
74 , intersperse
54 , intersperse
75 , transpose
55 , transpose
79− -- ** Case conversion
58+ -- * Case conversion
81 , toCaseFold
60 , toCaseFold
91 -- * Folds
64 -- * Folds
@@ -125,32 +98,21 @@ module Data.Text
126 -- ** Breaking strings
99 -- ** Breaking strings
131 , takeWhile
102 , takeWhile
133 , dropWhile
103 , dropWhile
139 , splitAt
104 , splitAt
145 , groupBy
108 , groupBy
149 -- ** Breaking into many substrings
112 -- ** Breaking into many substrings
153− , chunksOf
114+ , splitWith
155 -- ** Breaking into lines and words
117 -- ** Breaking into lines and words
@@ -164,309 +126,137 @@ module Data.Text
164 , isSuffixOf
126 , isSuffixOf
165 , isInfixOf
127 , isInfixOf
172 -- * Searching
129 -- * Searching
176 , partition
133 , partition
178 -- , findSubstring
135 -- , findSubstring
180 -- * Indexing
137 -- * Indexing
183 , findIndex
139 , findIndex
186− -- * Zipping
145+ -- * Zipping and unzipping
188 , zipWith
147 , zipWith
190 -- -* Ordered text
149 -- -* Ordered text
191 -- , sort
150 -- , sort
193− -- * Low level operations
198−import Prelude (Char, Bool(..), Int, Maybe(..), String,
153+import Prelude (Char, Bool(..), Functor(..), Int, Maybe(..), String,
199− Eq(..), Ord(..), Ordering(..), (++),
154+ Eq(..), Ord(..), (++),
200− Read(..),
155+ Read(..), Show(..),
201− (&&), (||), (+), (-), (.), ($), ($!), (>>),
156+ (&&), (||), (+), (-), (.), ($),
202− not, return, otherwise, quot)
157+ not, return, otherwise)
203−#if defined(HAVE_DEEPSEQ)
204−import Control.DeepSeq (NFData(rnf))
207import Control.Exception (assert)
158import Control.Exception (assert)
209import Data.Char (isSpace)
159import 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)
213import Control.Monad.ST (ST)
160import Control.Monad.ST (ST)
214import qualified Data.Text.Array as A
161import qualified Data.Text.Array as A
215import qualified Data.List as L
162import qualified Data.List as L
216−import Data.Binary (Binary(get, put))
217import Data.Monoid (Monoid(..))
163import Data.Monoid (Monoid(..))
218−#if MIN_VERSION_base(4,9,0)
219−import Data.Semigroup (Semigroup(..))
164+import Data.Word (Word16)
221import Data.String (IsString(..))
165import 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)
241−import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)
242−#if __GLASGOW_HASKELL__ >= 708
243−import qualified GHC.Exts as Exts
245−#if MIN_VERSION_base(4,7,0)
246−import Text.Printf (PrintfArg, formatArg, formatString)
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.
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.
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'
270−-- $replacement
167+import qualified Data.Text.Fusion as S
271−--
168+import qualified Data.Text.Fusion.Common as S
272−-- A 'Text' value is a sequence of Unicode scalar values, as defined
169+import Data.Text.Fusion (stream, reverseStream, unstream)
274−-- <http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=35 §3.9, definition D76 of the Unicode 5.2 standard >.
171+import Data.Text.Internal (Text(..), empty, text, textP)
275−-- As such, a 'Text' cannot contain values in the range U+D800 to
172+import qualified Prelude as P
276−-- U+DFFF inclusive. Haskell implementations admit all Unicode code
173+import Data.Text.Unsafe (iter, iter_, unsafeHead, unsafeTail)
277−-- points
174+import Data.Text.UnsafeChar (unsafeChr)
278−-- (<http://www.unicode.org/versions/Unicode5.2.0/ch03.pdf#page=13 §3.4, definition D10 >)
175+import qualified Data.Text.Encoding.Utf16 as U16
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
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\".
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 >.)
299-- $fusion
177-- $fusion
301−-- Most of the functions in this module are subject to /fusion/,
179+-- Most of the functions in this module are subject to /array fusion/,
302−-- meaning that a pipeline of such functions will usually allocate at
180+-- meaning that a pipeline of functions will usually allocate at most
303−-- most one 'Text' value.
181+-- one 'Text' value.
305−-- As an example, consider the following pipeline:
307−-- > import Data.Text as T
308−-- > import Data.Text.Encoding as E
309−-- > import Data.ByteString (ByteString)
311−-- > countChars :: ByteString -> Int
312−-- > countChars = T.length . T.toUpper . E.decodeUtf8
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'.
321−-- Functions that can be fused by the compiler are documented with the
322−-- phrase \"Subject to fusion\".
324instance Eq Text where
183instance Eq Text where
325− Text arrA offA lenA == Text arrB offB lenB
184+ t1 == t2 = stream t1 == stream t2
326− | lenA == lenB = A.equal arrA offA arrB offB lenA
328 {-# INLINE (==) #-}
185 {-# INLINE (==) #-}
330instance Ord Text where
187instance Ord Text where
331− compare = compareText
188+ compare t1 t2 = compare (stream t1) (stream t2)
189+ {-# INLINE compare #-}
191+instance Show Text where
192+ showsPrec p ps r = showsPrec p (unpack ps) r
333instance Read Text where
194instance Read Text where
334 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
195 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
336−#if MIN_VERSION_base(4,9,0)
337−-- Semigroup orphan instances for older GHCs are provided by
338−-- 'semigroups` package
340−instance Semigroup Text where
344instance Monoid Text where
197instance Monoid Text where
345 mempty = empty
198 mempty = empty
346−#if MIN_VERSION_base(4,9,0)
347− mappend = (<>) -- future-proof definition
349 mappend = append
199 mappend = append
351 mconcat = concat
200 mconcat = concat
353instance IsString Text where
202instance IsString Text where
354 fromString = pack
203 fromString = pack
356−#if __GLASGOW_HASKELL__ >= 708
357−instance Exts.IsList Text where
358− type Item Text = Char
363−#if defined(HAVE_DEEPSEQ)
364−instance NFData Text where rnf !_ = ()
367−instance Binary Text where
368− put t = put (encodeUtf8 t)
371− case decodeUtf8' bs of
372− P.Left exn -> P.fail (P.show exn)
373− P.Right a -> P.return a
375−-- | This instance preserves data abstraction at the cost of inefficiency.
376−-- We omit reflection services for the sake of data abstraction.
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
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? >
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, ... >
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
395− _ -> P.error "gunfold"
396− dataTypeOf _ = textDataType
398−#if MIN_VERSION_base(4,7,0)
399−-- | Only defined for @base-4.7.0.0@ and later
400−instance PrintfArg Text where
401− formatArg txt = formatString $ unpack txt
405−packConstr = mkConstr textDataType "pack" [] Prefix
407−textDataType :: DataType
408−textDataType = mkDataType "Data.Text.Text" [packConstr]
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
417− | i >= lenA || j >= lenB = compare lenA lenB
420− | otherwise = go (i+di) (j+dj)
421− where Iter a di = iter ta i
422− Iter b dj = iter tb j
424-- -----------------------------------------------------------------------------
205-- -----------------------------------------------------------------------------
425-- * Conversion to/from 'Text'
206-- * Conversion to/from 'Text'
427−-- | /O(n)/ Convert a 'String' into a 'Text'. Subject to
208+-- | /O(n)/ Convert a 'String' into a 'Text'.
428−-- fusion. Performs replacement on invalid scalar values.
209+--
210+-- This function is subject to array fusion.
429pack :: String -> Text
211pack :: String -> Text
430−pack = unstream . S.map safe . S.streamList
212+pack = unstream . S.streamList
431{-# INLINE [1] pack #-}
213{-# INLINE [1] pack #-}
215+-- | /O(n)/ Convert a Text into a String.
216+-- Subject to array fusion.
217+unpack :: Text -> String
218+unpack = S.unstreamList . stream
219+{-# INLINE [1] unpack #-}
221+-- | /O(1)/ Convert a character into a Text.
222+-- Subject to array fusion.
223+singleton :: Char -> Text
224+singleton = unstream . S.singleton
225+{-# INLINE [1] singleton #-}
433-- -----------------------------------------------------------------------------
227-- -----------------------------------------------------------------------------
434-- * Basic functions
228-- * Basic functions
436-- | /O(n)/ Adds a character to the front of a 'Text'. This function
230-- | /O(n)/ Adds a character to the front of a 'Text'. This function
437-- is more costly than its 'List' counterpart because it requires
231-- is more costly than its 'List' counterpart because it requires
438−-- copying a new array. Subject to fusion. Performs replacement on
232+-- copying a new array. Subject to array fusion.
439−-- invalid scalar values.
440cons :: Char -> Text -> Text
233cons :: Char -> Text -> Text
441−cons c t = unstream (S.cons (safe c) (stream t))
234+cons c t = unstream (S.cons c (stream t))
442{-# INLINE cons #-}
235{-# INLINE cons #-}
446-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
237-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
447−-- entire array in the process, unless fused. Subject to fusion.
238+-- entire array in the process. Subject to array fusion.
448−-- Performs replacement on invalid scalar values.
449snoc :: Text -> Char -> Text
239snoc :: Text -> Char -> Text
450−snoc t c = unstream (S.snoc (stream t) (safe c))
240+snoc t c = unstream (S.snoc (stream t) c)
451{-# INLINE snoc #-}
241{-# INLINE snoc #-}
453-- | /O(n)/ Appends one 'Text' to the other by copying both of them
243-- | /O(n)/ Appends one 'Text' to the other by copying both of them
454−-- into a new 'Text'. Subject to fusion.
244+-- into a new 'Text'. Subject to array fusion.
455append :: Text -> Text -> Text
245append :: Text -> Text -> Text
456−append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)
246+append (Text arr1 off1 len1) (Text arr2 off2 len2) = Text (A.run x) 0 len
459− | len > 0 = Text (A.run x) 0 len
460− | otherwise = overflowError "append"
462 len = len1+len2
248 len = len1+len2
463− x :: ST s (A.MArray s)
465− arr <- A.new len
250+ arr <- A.unsafeNew len :: ST s (A.MArray s Word16)
466− A.copyI arr 0 arr1 off1 len1
251+ copy arr1 off1 (len1+off1) arr 0
467− A.copyI arr len1 arr2 off2 len
252+ copy arr2 off2 (len2+off2) arr len1
468 return arr
253 return arr
255+ copy arr i top arr' j
256+ | i >= top = return ()
257+ | otherwise = do A.unsafeWrite arr' j (arr `A.unsafeIndex` i)
258+ copy arr (i+1) top arr' (j+1)
469−{-# NOINLINE append #-}
259+{-# INLINE append #-}
472"TEXT append -> fused" [~1] forall t1 t2.
262"TEXT append -> fused" [~1] forall t1 t2.
@@ -476,26 +266,33 @@ append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)
478-- | /O(1)/ Returns the first character of a 'Text', which must be
268-- | /O(1)/ Returns the first character of a 'Text', which must be
479−-- non-empty. Subject to fusion.
269+-- non-empty. Subject to array fusion.
480head :: Text -> Char
270head :: Text -> Char
481head t = S.head (stream t)
271head t = S.head (stream t)
482{-# INLINE head #-}
272{-# INLINE head #-}
484-- | /O(1)/ Returns the first character and rest of a 'Text', or
274-- | /O(1)/ Returns the first character and rest of a 'Text', or
485−-- 'Nothing' if empty. Subject to fusion.
275+-- 'Nothing' if empty. Subject to array fusion.
486uncons :: Text -> Maybe (Char, Text)
276uncons :: Text -> Maybe (Char, Text)
487uncons t@(Text arr off len)
277uncons t@(Text arr off len)
488 | len <= 0 = Nothing
278 | len <= 0 = Nothing
489− | otherwise = Just $ let !(Iter c d) = iter t 0
490− in (c, text arr (off+d) (len-d))
279+ | otherwise = Just (c, textP arr (off+d) (len-d))
280+ where (c,d) = iter t 0
491−{-# INLINE [1] uncons #-}
281+{-# INLINE uncons #-}
493-- | Lifted from Control.Arrow and specialized.
283-- | Lifted from Control.Arrow and specialized.
494second :: (b -> c) -> (a,b) -> (a,c)
284second :: (b -> c) -> (a,b) -> (a,c)
495second f (a, b) = (a, f b)
285second f (a, b) = (a, f b)
288+"TEXT uncons -> fused" [~1] forall t.
289+ uncons t = fmap (second unstream) (S.uncons (stream t))
290+"TEXT uncons -> unfused" [1] forall t.
291+ fmap (second unstream) (S.uncons (stream t)) = uncons t
497-- | /O(1)/ Returns the last character of a 'Text', which must be
294-- | /O(1)/ Returns the last character of a 'Text', which must be
498−-- non-empty. Subject to fusion.
295+-- non-empty. Subject to array fusion.
499last :: Text -> Char
296last :: Text -> Char
500last (Text arr off len)
297last (Text arr off len)
501 | len <= 0 = emptyError "last"
298 | len <= 0 = emptyError "last"
@@ -513,11 +310,11 @@ last (Text arr off len)
515-- | /O(1)/ Returns all characters after the head of a 'Text', which
312-- | /O(1)/ Returns all characters after the head of a 'Text', which
516−-- must be non-empty. Subject to fusion.
313+-- must be non-empty. Subject to array fusion.
517tail :: Text -> Text
314tail :: Text -> Text
518tail t@(Text arr off len)
315tail t@(Text arr off len)
519 | len <= 0 = emptyError "tail"
316 | len <= 0 = emptyError "tail"
520− | otherwise = text arr (off+d) (len-d)
317+ | otherwise = textP arr (off+d) (len-d)
521 where d = iter_ t 0
318 where d = iter_ t 0
522{-# INLINE [1] tail #-}
319{-# INLINE [1] tail #-}
@@ -529,11 +326,11 @@ tail t@(Text arr off len)
531-- | /O(1)/ Returns all but the last character of a 'Text', which must
328-- | /O(1)/ Returns all but the last character of a 'Text', which must
532−-- be non-empty. Subject to fusion.
329+-- be non-empty. Subject to array fusion.
533init :: Text -> Text
330init :: Text -> Text
534init (Text arr off len) | len <= 0 = emptyError "init"
331init (Text arr off len) | len <= 0 = emptyError "init"
535− | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)
332+ | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)
536− | otherwise = text arr off (len-1)
333+ | otherwise = textP arr off (len-1)
538 n = A.unsafeIndex arr (off+len-1)
335 n = A.unsafeIndex arr (off+len-1)
539{-# INLINE [1] init #-}
336{-# INLINE [1] init #-}
@@ -545,14 +342,10 @@ init (Text arr off len) | len <= 0 = emptyError "init"
545 unstream (S.init (stream t)) = init t
342 unstream (S.init (stream t)) = init t
548−-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to
345+-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to array
549-- fusion.
346-- fusion.
550null :: Text -> Bool
347null :: Text -> Bool
551−null (Text _arr _off len) =
348+null (Text _arr _off len) = assert (len >= 0) $ len <= 0
556{-# INLINE [1] null #-}
349{-# INLINE [1] null #-}
@@ -562,171 +355,53 @@ null (Text _arr _off len) =
562 S.null (stream t) = null t
355 S.null (stream t) = null t
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 #-}
571-- | /O(n)/ Returns the number of characters in a 'Text'.
358-- | /O(n)/ Returns the number of characters in a 'Text'.
572−-- Subject to fusion.
359+-- Subject to array fusion.
573length :: Text -> Int
360length :: Text -> Int
574length t = S.length (stream t)
361length t = S.length (stream t)
575−{-# INLINE [0] length #-}
362+{-# INLINE 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.
579−-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
580−-- Subject to fusion.
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 #-}
590−"TEXT compareN/length -> compareLength" [~1] forall t n.
591− compare (length t) n = compareLength t n
595−"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
596− eqInt (length t) n = compareLength t n == EQ
600−"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
601− neInt (length t) n = compareLength t n /= EQ
605−"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
606− ltInt (length t) n = compareLength t n == LT
610−"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
611− leInt (length t) n = compareLength t n /= GT
615−"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
616− gtInt (length t) n = compareLength t n == GT
620−"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
621− geInt (length t) n = compareLength t n /= LT
624-- -----------------------------------------------------------------------------
364-- -----------------------------------------------------------------------------
625-- * Transformations
365-- * Transformations
626−-- | /O(n)/ 'map' @f@ @t@ is the 'Text' obtained by applying @f@ to
366+-- | /O(n)/ 'map' @f @xs is the 'Text' obtained by applying @f@ to
627−-- each element of @t@. Subject to fusion. Performs replacement on
367+-- each element of @xs@. Subject to array fusion.
628−-- invalid scalar values.
629map :: (Char -> Char) -> Text -> Text
368map :: (Char -> Char) -> Text -> Text
630−map f t = unstream (S.map (safe . f) (stream t))
369+map f t = unstream (S.map f (stream t))
631{-# INLINE [1] map #-}
370{-# INLINE [1] map #-}
633-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
372-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
634-- 'Text's and concatenates the list after interspersing the first
373-- 'Text's and concatenates the list after interspersing the first
635-- argument between each element of the list.
374-- argument between each element of the list.
636intercalate :: Text -> [Text] -> Text
375intercalate :: Text -> [Text] -> Text
637−intercalate t = concat . (F.intersperse t)
376+intercalate t ts = unstream (S.intercalate (stream t) (L.map stream ts))
638{-# INLINE intercalate #-}
377{-# INLINE intercalate #-}
640-- | /O(n)/ The 'intersperse' function takes a character and places it
379-- | /O(n)/ The 'intersperse' function takes a character and places it
641−-- between the characters of a 'Text'. Subject to fusion. Performs
380+-- between the characters of a 'Text'. Subject to array fusion.
642−-- replacement on invalid scalar values.
643intersperse :: Char -> Text -> Text
381intersperse :: Char -> Text -> Text
644−intersperse c t = unstream (S.intersperse (safe c) (stream t))
382+intersperse c t = unstream (S.intersperse c (stream t))
645{-# INLINE intersperse #-}
383{-# INLINE intersperse #-}
647−-- | /O(n)/ Reverse the characters of a string. Subject to fusion.
385+-- | /O(n)/ Reverse the characters of a string. Subject to array fusion.
648reverse :: Text -> Text
386reverse :: Text -> Text
649reverse t = S.reverse (stream t)
387reverse t = S.reverse (stream t)
650{-# INLINE reverse #-}
388{-# INLINE reverse #-}
652−-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in
653−-- @haystack@ with @replacement@.
655−-- This function behaves as though it was defined as follows:
658−-- replace needle replacement haystack =
659−-- 'intercalate' replacement ('splitOn' needle haystack)
662−-- As this suggests, each occurrence is replaced exactly once. So if
663−-- @needle@ occurs in @replacement@, that occurrence will /not/ itself
664−-- be replaced recursively:
666−-- > replace "oo" "foo" "oo" == "foo"
668−-- In cases where several instances of @needle@ overlap, only the
669−-- first one will be replaced:
671−-- > replace "ofo" "bar" "ofofo" == "barfo"
673−-- In (unlikely) bad cases, this function's time complexity degrades
676− -- ^ @needle@ to search for. If this string is empty, an
677− -- error will occur.
679− -- ^ @replacement@ to replace @needle@ with.
681− -- ^ @haystack@ in which to search.
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
691− ixs = indices needle haystack
692− len = hayLen - (neeLen - repLen) `mul` L.length ixs
693− x :: ST s (A.MArray s)
696− let loop (i:is) o d = do
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
706-- ----------------------------------------------------------------------------
390-- ----------------------------------------------------------------------------
707-- ** Case conversions (folds)
391-- ** Case conversions (folds)
711−-- When case converting 'Text' values, do not use combinators like
395+-- With Unicode text, it is incorrect to use combinators like @map
712−-- @map toUpper@ to case convert each character of a string
396+-- toUpper@ to case convert each character of a string individually.
713−-- individually, as this gives incorrect results according to the
397+-- Instead, use the whole-string case conversion functions from this
714−-- rules of some writing systems. The whole-string case conversion
398+-- module. For correctness in different writing systems, these
715−-- functions from this module, such as @toUpper@, obey the correct
716−-- case conversion rules. As a result, these functions may map one
717−-- input character to two or three output characters. For examples,
399+-- functions may map one input character to two or three output
718−-- see the documentation of each function.
400+-- characters.
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 >.
726−-- | /O(n)/ Convert a string to folded case. Subject to fusion.
402+-- | /O(n)/ Convert a string to folded case. This function is mainly
728−-- This function is mainly useful for performing caseless (also known
403+-- useful for performing caseless (or case insensitive) string
729−-- as case insensitive) string comparisons.
404+-- comparisons.
731-- A string @x@ is a caseless match for a string @y@ if and only if:
406-- A string @x@ is a caseless match for a string @y@ if and only if:
@@ -734,112 +409,31 @@ replace needle@(Text _ _ neeLen)
735-- The result string may be longer than the input string, and may
410-- The result string may be longer than the input string, and may
736-- differ from applying 'toLower' to the input string. For instance,
411-- differ from applying 'toLower' to the input string. For instance,
737−-- the Armenian small ligature \"ﬓ\" (men now, U+FB13) is case
412+-- the Armenian small ligature men now (U+FB13) is case folded to the
738−-- folded to the sequence \"մ\" (men, U+0574) followed by
739−-- \"ն\" (now, U+0576), while the Greek \"µ\" (micro sign,
413+-- bigram men now (U+0574 U+0576), while the micro sign (U+00B5) is
740−-- U+00B5) is case folded to \"μ\" (small letter mu, U+03BC)
414+-- case folded to the Greek small letter letter mu (U+03BC) instead of
741−-- instead of itself.
415+-- itself.
742toCaseFold :: Text -> Text
416toCaseFold :: Text -> Text
743toCaseFold t = unstream (S.toCaseFold (stream t))
417toCaseFold t = unstream (S.toCaseFold (stream t))
744−{-# INLINE toCaseFold #-}
418+{-# INLINE [0] toCaseFold #-}
746-- | /O(n)/ Convert a string to lower case, using simple case
420-- | /O(n)/ Convert a string to lower case, using simple case
747−-- conversion. Subject to fusion.
749−-- The result string may be longer than the input string. For
421+-- conversion. The result string may be longer than the input string.
750−-- instance, \"İ\" (Latin capital letter I with dot above,
422+-- For instance, the Latin capital letter I with dot above (U+0130)
751−-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069)
423+-- maps to the bigram Latin small letter i (U+0069) followed by
752−-- followed by \" ̇\" (combining dot above, U+0307).
424+-- combining dot above (U+0307).
753toLower :: Text -> Text
425toLower :: Text -> Text
754toLower t = unstream (S.toLower (stream t))
426toLower t = unstream (S.toLower (stream t))
755{-# INLINE toLower #-}
427{-# INLINE toLower #-}
757-- | /O(n)/ Convert a string to upper case, using simple case
429-- | /O(n)/ Convert a string to upper case, using simple case
758−-- conversion. Subject to fusion.
760−-- The result string may be longer than the input string. For
430+-- conversion. The result string may be longer than the input string.
761−-- instance, the German \"ß\" (eszett, U+00DF) maps to the
431+-- For instance, the German eszett (U+00DF) maps to the two-letter
762−-- two-letter sequence \"SS\".
432+-- sequence SS.
763toUpper :: Text -> Text
433toUpper :: Text -> Text
764toUpper t = unstream (S.toUpper (stream t))
434toUpper t = unstream (S.toUpper (stream t))
765{-# INLINE toUpper #-}
435{-# INLINE toUpper #-}
767−-- | /O(n)/ Convert a string to title case, using simple case
768−-- conversion. Subject to fusion.
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
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).
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 #-}
789−-- | /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.
795−-- > justifyLeft 7 'x' "foo" == "fooxxxx"
796−-- > justifyLeft 3 'x' "foobar" == "foobar"
797−justifyLeft :: Int -> Char -> Text -> Text
800− | otherwise = t `append` replicateChar (k-len) c
801− where len = length t
802−{-# INLINE [1] justifyLeft #-}
805−"TEXT justifyLeft -> fused" [~1] forall k c t.
806− justifyLeft k c t = unstream (S.justifyLeftI k c (stream t))
807−"TEXT justifyLeft -> unfused" [1] forall k c t.
808− unstream (S.justifyLeftI k c (stream t)) = justifyLeft k c t
811−-- | /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.
817−-- > justifyRight 7 'x' "bar" == "xxxxbar"
818−-- > justifyRight 3 'x' "foobar" == "foobar"
819−justifyRight :: Int -> Char -> Text -> Text
822− | otherwise = replicateChar (k-len) c `append` t
823− where len = length t
824−{-# INLINE justifyRight #-}
826−-- | /O(n)/ Center a string to the given length, using the specified
827−-- fill character on either side. Performs replacement on invalid
832−-- > center 8 'x' "HS" = "xxxHSxxx"
833−center :: Int -> Char -> Text -> Text
836− | otherwise = replicateChar l c `append` t `append` replicateChar r c
837− where len = length t
841−{-# INLINE center #-}
843-- | /O(n)/ The 'transpose' function transposes the rows and columns
437-- | /O(n)/ The 'transpose' function transposes the rows and columns
844-- of its 'Text' argument. Note that this function uses 'pack',
438-- of its 'Text' argument. Note that this function uses 'pack',
845-- 'unpack', and the list version of transpose, and is thus not very
439-- 'unpack', and the list version of transpose, and is thus not very
@@ -853,23 +447,26 @@ transpose ts = P.map pack (L.transpose (P.map unpack ts))
853-- | /O(n)/ 'foldl', applied to a binary operator, a starting value
447-- | /O(n)/ 'foldl', applied to a binary operator, a starting value
854-- (typically the left-identity of the operator), and a 'Text',
448-- (typically the left-identity of the operator), and a 'Text',
855-- reduces the 'Text' using the binary operator, from left to right.
449-- reduces the 'Text' using the binary operator, from left to right.
856−-- Subject to fusion.
450+-- Subject to array fusion.
857−foldl :: (a -> Char -> a) -> a -> Text -> a
451+foldl :: (b -> Char -> b) -> b -> Text -> b
858foldl f z t = S.foldl f z (stream t)
452foldl f z t = S.foldl f z (stream t)
859{-# INLINE foldl #-}
453{-# INLINE foldl #-}
861−-- | /O(n)/ A strict version of 'foldl'. Subject to fusion.
455+-- | /O(n)/ A strict version of 'foldl'.
456+-- Subject to array fusion.
862−foldl' :: (a -> Char -> a) -> a -> Text -> a
457+foldl' :: (b -> Char -> b) -> b -> Text -> b
863foldl' f z t = S.foldl' f z (stream t)
458foldl' f z t = S.foldl' f z (stream t)
864{-# INLINE foldl' #-}
459{-# INLINE foldl' #-}
866-- | /O(n)/ A variant of 'foldl' that has no starting value argument,
461-- | /O(n)/ A variant of 'foldl' that has no starting value argument,
867−-- and thus must be applied to a non-empty 'Text'. Subject to fusion.
462+-- and thus must be applied to a non-empty 'Text'. Subject to array
868foldl1 :: (Char -> Char -> Char) -> Text -> Char
464foldl1 :: (Char -> Char -> Char) -> Text -> Char
869foldl1 f t = S.foldl1 f (stream t)
465foldl1 f t = S.foldl1 f (stream t)
870{-# INLINE foldl1 #-}
466{-# INLINE foldl1 #-}
872−-- | /O(n)/ A strict version of 'foldl1'. Subject to fusion.
468+-- | /O(n)/ A strict version of 'foldl1'.
469+-- Subject to array fusion.
873foldl1' :: (Char -> Char -> Char) -> Text -> Char
470foldl1' :: (Char -> Char -> Char) -> Text -> Char
874foldl1' f t = S.foldl1' f (stream t)
471foldl1' f t = S.foldl1' f (stream t)
875{-# INLINE foldl1' #-}
472{-# INLINE foldl1' #-}
@@ -877,13 +474,13 @@ foldl1' f t = S.foldl1' f (stream t)
877-- | /O(n)/ 'foldr', applied to a binary operator, a starting value
474-- | /O(n)/ 'foldr', applied to a binary operator, a starting value
878-- (typically the right-identity of the operator), and a 'Text',
475-- (typically the right-identity of the operator), and a 'Text',
879-- reduces the 'Text' using the binary operator, from right to left.
476-- reduces the 'Text' using the binary operator, from right to left.
880−-- Subject to fusion.
477+-- Subject to array fusion.
881−foldr :: (Char -> a -> a) -> a -> Text -> a
478+foldr :: (Char -> b -> b) -> b -> Text -> b
882foldr f z t = S.foldr f z (stream t)
479foldr f z t = S.foldr f z (stream t)
883{-# INLINE foldr #-}
480{-# INLINE foldr #-}
885−-- | /O(n)/ A variant of 'foldr' that has no starting value argument,
482+-- | /O(n)/ A variant of 'foldr' that has no starting value argument, and
886−-- and thus must be applied to a non-empty 'Text'. Subject to
483+-- thust must be applied to a non-empty 'Text'. Subject to array
887-- fusion.
484-- fusion.
888foldr1 :: (Char -> Char -> Char) -> Text -> Char
485foldr1 :: (Char -> Char -> Char) -> Text -> Char
889foldr1 f t = S.foldr1 f (stream t)
486foldr1 f t = S.foldr1 f (stream t)
@@ -892,48 +489,40 @@ foldr1 f t = S.foldr1 f (stream t)
892-- -----------------------------------------------------------------------------
489-- -----------------------------------------------------------------------------
893-- ** Special folds
490-- ** Special folds
895−-- | /O(n)/ Concatenate a list of 'Text's.
492+-- | /O(n)/ Concatenate a list of 'Text's. Subject to array fusion.
896concat :: [Text] -> Text
493concat :: [Text] -> Text
897−concat ts = case ts' of
494+concat ts = unstream (S.concat (L.map stream ts))
898− [] -> empty
495+{-# INLINE concat #-}
900− _ -> Text (A.run go) 0 len
902− ts' = L.filter (not . null) ts
903− len = sumP "concat" $ L.map lengthWord16 ts'
904− go :: ST s (A.MArray s)
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
911-- | /O(n)/ Map a function over a 'Text' that results in a 'Text', and
497-- | /O(n)/ Map a function over a 'Text' that results in a 'Text', and
912−-- concatenate the results.
498+-- concatenate the results. This function is subject to array fusion.
500+-- Note: if in 'concatMap' @f@ @t@, @f@ is defined in terms of fusible
501+-- functions, it will also be fusible.
913concatMap :: (Char -> Text) -> Text -> Text
502concatMap :: (Char -> Text) -> Text -> Text
914−concatMap f = concat . foldr ((:) . f) []
503+concatMap f t = unstream (S.concatMap (stream . f) (stream t))
915{-# INLINE concatMap #-}
504{-# INLINE concatMap #-}
917-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
506-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
918−-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.
507+-- 'Text' @t@ satisifes the predicate @p@. Subject to array fusion.
919any :: (Char -> Bool) -> Text -> Bool
508any :: (Char -> Bool) -> Text -> Bool
920any p t = S.any p (stream t)
509any p t = S.any p (stream t)
921{-# INLINE any #-}
510{-# INLINE any #-}
923-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
512-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
924−-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.
513+-- 'Text' @t@ satisify the predicate @p@. Subject to array fusion.
925all :: (Char -> Bool) -> Text -> Bool
514all :: (Char -> Bool) -> Text -> Bool
926all p t = S.all p (stream t)
515all p t = S.all p (stream t)
927{-# INLINE all #-}
516{-# INLINE all #-}
929-- | /O(n)/ 'maximum' returns the maximum value from a 'Text', which
518-- | /O(n)/ 'maximum' returns the maximum value from a 'Text', which
930−-- must be non-empty. Subject to fusion.
519+-- must be non-empty. Subject to array fusion.
931maximum :: Text -> Char
520maximum :: Text -> Char
932maximum t = S.maximum (stream t)
521maximum t = S.maximum (stream t)
933{-# INLINE maximum #-}
522{-# INLINE maximum #-}
935-- | /O(n)/ 'minimum' returns the minimum value from a 'Text', which
524-- | /O(n)/ 'minimum' returns the minimum value from a 'Text', which
936−-- must be non-empty. Subject to fusion.
525+-- must be non-empty. Subject to array fusion.
937minimum :: Text -> Char
526minimum :: Text -> Char
938minimum t = S.minimum (stream t)
527minimum t = S.minimum (stream t)
939{-# INLINE minimum #-}
528{-# INLINE minimum #-}
@@ -942,8 +531,8 @@ minimum t = S.minimum (stream t)
942-- * Building 'Text's
531-- * Building 'Text's
944-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
533-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
945−-- successive reduced values from the left. Subject to fusion.
534+-- successive reduced values from the left. This function is subject
946−-- Performs replacement on invalid scalar values.
535+-- to array fusion.
948-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
537-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
@@ -951,13 +540,11 @@ minimum t = S.minimum (stream t)
952-- > last (scanl f z xs) == foldl f z xs.
541-- > last (scanl f z xs) == foldl f z xs.
953scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
542scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
954−scanl f z t = unstream (S.scanl g z (stream t))
543+scanl f z t = unstream (S.scanl f z (stream t))
955− where g a b = safe (f a b)
956{-# INLINE scanl #-}
544{-# INLINE scanl #-}
958-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
546-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
959−-- value argument. Subject to fusion. Performs replacement on
547+-- value argument. This function is subject to array fusion.
960−-- invalid scalar values.
962-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
549-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
963scanl1 :: (Char -> Char -> Char) -> Text -> Text
550scanl1 :: (Char -> Char -> Char) -> Text -> Text
@@ -965,75 +552,49 @@ scanl1 f t | null t = empty
965 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
552 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
966{-# INLINE scanl1 #-}
553{-# INLINE scanl1 #-}
968−-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'. Performs
555+-- | /O(n)/ 'scanr' is the right-to-left dual of 'scanl'.
969−-- replacement on invalid scalar values.
971-- > scanr f v == reverse . scanl (flip f) v . reverse
557-- > scanr f v == reverse . scanl (flip f) v . reverse
972scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
558scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
973−scanr f z = S.reverse . S.reverseScanr g z . reverseStream
559+scanr f z = S.reverse . S.reverseScanr f z . reverseStream
974− where g a b = safe (f a b)
975{-# INLINE scanr #-}
560{-# INLINE scanr #-}
977-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
562-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
978−-- value argument. Subject to fusion. Performs replacement on
563+-- value argument. This function is subject to array fusion.
979−-- invalid scalar values.
980scanr1 :: (Char -> Char -> Char) -> Text -> Text
564scanr1 :: (Char -> Char -> Char) -> Text -> Text
981scanr1 f t | null t = empty
565scanr1 f t | null t = empty
982 | otherwise = scanr f (last t) (init t)
566 | otherwise = scanr f (last t) (init t)
983{-# INLINE scanr1 #-}
567{-# INLINE scanr1 #-}
985−-- | /O(n)/ Like a combination of 'map' and 'foldl''. Applies a
569+-- | /O(n)/ Like a combination of 'map' and 'foldl'. Applies a
986-- function to each element of a 'Text', passing an accumulating
570-- function to each element of a 'Text', passing an accumulating
987−-- parameter from left to right, and returns a final 'Text'. Performs
571+-- parameter from left to right, and returns a final 'Text'.
988−-- replacement on invalid scalar values.
989mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
572mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
990−mapAccumL f z0 = S.mapAccumL g z0 . stream
573+mapAccumL f s t = case uncons t of
991− where g a b = second safe (f a b)
574+ Nothing -> (s, empty)
575+ Just (x, xs) -> (s'', cons y ys)
576+ where (s', y ) = f s x
992−{-# INLINE mapAccumL #-}
577+ (s'',ys) = mapAccumL f s' xs
994-- | The 'mapAccumR' function behaves like a combination of 'map' and
579-- | The 'mapAccumR' function behaves like a combination of 'map' and
995−-- a strict 'foldr'; it applies a function to each element of a
580+-- 'foldr'; it applies a function to each element of a 'Text', passing
996−-- 'Text', passing an accumulating parameter from right to left, and
581+-- an accumulating parameter from right to left, and returning a final
997−-- returning a final value of this accumulator together with the new
582+-- value of this accumulator together with the new 'Text'.
999−-- Performs replacement on invalid scalar values.
1000mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
583mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
1001−mapAccumR f z0 = second reverse . S.mapAccumL g z0 . reverseStream
584+mapAccumR f s t = case uncons t of
1002− where g a b = second safe (f a b)
585+ Nothing -> (s, empty)
586+ Just (x, xs) -> (s'', cons y ys)
587+ where (s'',y ) = f s' x
1003−{-# INLINE mapAccumR #-}
588+ (s', ys) = mapAccumR f s xs
1005-- -----------------------------------------------------------------------------
590-- -----------------------------------------------------------------------------
1006-- ** Generating and unfolding 'Text's
591-- ** Generating and unfolding 'Text's
1008−-- | /O(n*m)/ 'replicate' @n@ @t@ is a 'Text' consisting of the input
1009−-- @t@ repeated @n@ times.
1010−replicate :: Int -> Text -> Text
1011−replicate n t@(Text a o l)
1012− | n <= 0 || l <= 0 = empty
1014− | isSingleton t = replicateChar n (unsafeHead t)
1015− | otherwise = Text (A.run x) 0 len
1018− x :: ST s (A.MArray s)
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)
1025−{-# INLINE [1] replicate #-}
1028−"TEXT replicate/singleton -> replicateChar" [~1] forall n c.
1029− replicate n (singleton c) = replicateChar n c
1032−-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the
593+-- | /O(n)/ 'replicate' @n@ @c@ is a 'Text' of length @n@ with @c@ the
1033-- value of every element. Subject to fusion.
594-- value of every element. Subject to fusion.
1034−replicateChar :: Int -> Char -> Text
595+replicate :: Int -> Char -> Text
1035−replicateChar n c = unstream (S.replicateCharI n (safe c))
596+replicate n c = unstream (S.replicate n c)
1036−{-# INLINE replicateChar #-}
597+{-# INLINE replicate #-}
1038-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
599-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
1039-- function is analogous to the List 'L.unfoldr'. 'unfoldr' builds a
600-- function is analogous to the List 'L.unfoldr'. 'unfoldr' builds a
@@ -1041,9 +602,9 @@ replicateChar n c = unstream (S.replicateCharI n (safe c))
1041-- returns 'Nothing' if it is done producing the 'Text', otherwise
602-- returns 'Nothing' if it is done producing the 'Text', otherwise
1042-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
603-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
1043-- string, and @b@ is the seed value for further production. Subject
604-- string, and @b@ is the seed value for further production. Subject
1044−-- to fusion. Performs replacement on invalid scalar values.
605+-- to fusion.
1045unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
606unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
1046−unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
607+unfoldr f s = unstream (S.unfoldr f s)
1047{-# INLINE unfoldr #-}
608{-# INLINE unfoldr #-}
1049-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
610-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
@@ -1051,9 +612,9 @@ unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
1051-- first argument to 'unfoldrN'. This function is more efficient than
612-- first argument to 'unfoldrN'. This function is more efficient than
1052-- 'unfoldr' when the maximum length of the result is known and
613-- 'unfoldr' when the maximum length of the result is known and
1053-- correct, otherwise its performance is similar to 'unfoldr'. Subject
614-- correct, otherwise its performance is similar to 'unfoldr'. Subject
1054−-- to fusion. Performs replacement on invalid scalar values.
615+-- to fusion.
1055unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
616unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
1056−unfoldrN n f s = unstream (S.unfoldrN n (firstf safe . f) s)
617+unfoldrN n f s = unstream (S.unfoldrN n f s)
1057{-# INLINE unfoldrN #-}
618{-# INLINE unfoldrN #-}
1059-- -----------------------------------------------------------------------------
620-- -----------------------------------------------------------------------------
@@ -1066,16 +627,14 @@ take :: Int -> Text -> Text
1066take n t@(Text arr off len)
627take n t@(Text arr off len)
1067 | n <= 0 = empty
628 | n <= 0 = empty
1068 | n >= len = t
629 | n >= len = t
1069− | otherwise = text arr off (iterN n t)
630+ | otherwise = Text arr off (loop 0 0)
633+ | i >= len || cnt >= n = i
634+ | otherwise = loop (i+d) (cnt+1)
1070{-# INLINE [1] take #-}
636{-# INLINE [1] take #-}
1072−iterN :: Int -> Text -> Int
1073−iterN n t@(Text _arr _off len) = loop 0 0
1075− | i >= len || cnt >= n = i
1076− | otherwise = loop (i+d) (cnt+1)
1077− where d = iter_ t i
1079{-# RULES
638{-# RULES
1080"TEXT take -> fused" [~1] forall n t.
639"TEXT take -> fused" [~1] forall n t.
1081 take n t = unstream (S.take n (stream t))
640 take n t = unstream (S.take n (stream t))
@@ -1083,36 +642,18 @@ iterN n t@(Text _arr _off len) = loop 0 0
1083 unstream (S.take n (stream t)) = take n t
642 unstream (S.take n (stream t)) = take n t
1086−-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after
1087−-- taking @n@ characters from the end of @t@.
1091−-- > takeEnd 3 "foobar" == "bar"
1092−takeEnd :: Int -> Text -> Text
1093−takeEnd n t@(Text arr off len)
1096− | otherwise = text arr (off+i) (len-i)
1097− where i = iterNEnd n t
1099−iterNEnd :: Int -> Text -> Int
1100−iterNEnd n t@(Text _arr _off len) = loop (len-1) n
1104− | otherwise = loop (i+d) (m-1)
1105− where d = reverseIter_ t i
1107-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
645-- | /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@
646+-- 'Text' of length @n@, or the empty 'Text' if @n@ is greater than the
1109−-- is greater than the length of the 'Text'. Subject to fusion.
647+-- length of the 'Text'. Subject to fusion.
1110drop :: Int -> Text -> Text
648drop :: Int -> Text -> Text
1111drop n t@(Text arr off len)
649drop n t@(Text arr off len)
1112 | n <= 0 = t
650 | n <= 0 = t
1113 | n >= len = empty
651 | n >= len = empty
1114− | otherwise = text arr (off+i) (len-i)
652+ | otherwise = loop 0 0
1115− where i = iterN n t
653+ where loop !i !cnt
654+ | i >= len || cnt >= n = Text arr (off+i) (len-i)
655+ | otherwise = loop (i+d) (cnt+1)
1116{-# INLINE [1] drop #-}
657{-# INLINE [1] drop #-}
1118{-# RULES
659{-# RULES
@@ -1122,27 +663,15 @@ drop n t@(Text arr off len)
1122 unstream (S.drop n (stream t)) = drop n t
663 unstream (S.drop n (stream t)) = drop n t
1125−-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after
1126−-- dropping @n@ characters from the end of @t@.
1130−-- > dropEnd 3 "foobar" == "foo"
1131−dropEnd :: Int -> Text -> Text
1132−dropEnd n t@(Text arr off len)
1135− | otherwise = text arr off (iterNEnd n t)
1137−-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
666+-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text', returns
1138−-- returns the longest prefix (possibly empty) of elements that
667+-- the longest prefix (possibly empty) of elements that satisfy @p@.
1139−-- satisfy @p@. Subject to fusion.
668+-- This function is subject to array fusion.
1140takeWhile :: (Char -> Bool) -> Text -> Text
669takeWhile :: (Char -> Bool) -> Text -> Text
1141takeWhile p t@(Text arr off len) = loop 0
670takeWhile p t@(Text arr off len) = loop 0
1142 where loop !i | i >= len = t
671 where loop !i | i >= len = t
1143 | p c = loop (i+d)
672 | p c = loop (i+d)
1144− | otherwise = text arr off i
673+ | otherwise = textP arr off i
1145− where Iter c d = iter t i
674+ where (c,d) = iter t i
1146{-# INLINE [1] takeWhile #-}
675{-# INLINE [1] takeWhile #-}
1148{-# RULES
677{-# RULES
@@ -1152,35 +681,14 @@ takeWhile p t@(Text arr off len) = loop 0
1152 unstream (S.takeWhile p (stream t)) = takeWhile p t
681 unstream (S.takeWhile p (stream t)) = takeWhile p t
1155−-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',
1156−-- returns the longest suffix (possibly empty) of elements that
1157−-- satisfy @p@. Subject to fusion.
1160−-- > takeWhileEnd (=='o') "foo" == "oo"
1161−takeWhileEnd :: (Char -> Bool) -> Text -> Text
1162−takeWhileEnd p t@(Text arr off len) = loop (len-1) len
1163− where loop !i !l | l <= 0 = t
1164− | p c = loop (i+d) (l+d)
1165− | otherwise = text arr (off+l) (len-l)
1166− where (c,d) = reverseIter t i
1167−{-# INLINE [1] takeWhileEnd #-}
1170−"TEXT takeWhileEnd -> fused" [~1] forall p t.
1171− takeWhileEnd p t = S.reverse (S.takeWhile p (S.reverseStream t))
1172−"TEXT takeWhileEnd -> unfused" [1] forall p t.
1173− S.reverse (S.takeWhile p (S.reverseStream t)) = takeWhileEnd p t
1176−-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
684+-- | /O(n)/ 'dropWhile' @p@ @xs@ returns the suffix remaining after
1177−-- 'takeWhile' @p@ @t@. Subject to fusion.
685+-- 'takeWhile' @p@ @xs@. This function is subject to array fusion.
1178dropWhile :: (Char -> Bool) -> Text -> Text
686dropWhile :: (Char -> Bool) -> Text -> Text
1179dropWhile p t@(Text arr off len) = loop 0 0
687dropWhile p t@(Text arr off len) = loop 0 0
1180 where loop !i !l | l >= len = empty
688 where loop !i !l | l >= len = empty
1181 | p c = loop (i+d) (l+d)
689 | p c = loop (i+d) (l+d)
1182 | otherwise = Text arr (off+i) (len-l)
690 | otherwise = Text arr (off+i) (len-l)
1183− where Iter c d = iter t i
691+ where (c,d) = iter t i
1184{-# INLINE [1] dropWhile #-}
692{-# INLINE [1] dropWhile #-}
1186{-# RULES
694{-# RULES
@@ -1190,57 +698,6 @@ dropWhile p t@(Text arr off len) = loop 0 0
1190 unstream (S.dropWhile p (stream t)) = dropWhile p t
698 unstream (S.dropWhile p (stream t)) = dropWhile p t
1193−-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
1194−-- dropping characters that satisfy the predicate @p@ from the end of
1195−-- @t@. Subject to fusion.
1199−-- > dropWhileEnd (=='.') "foo..." == "foo"
1200−dropWhileEnd :: (Char -> Bool) -> Text -> Text
1201−dropWhileEnd p t@(Text arr off len) = loop (len-1) len
1202− where loop !i !l | l <= 0 = empty
1203− | p c = loop (i+d) (l+d)
1204− | otherwise = Text arr off l
1205− where (c,d) = reverseIter t i
1206−{-# INLINE [1] dropWhileEnd #-}
1209−"TEXT dropWhileEnd -> fused" [~1] forall p t.
1210− dropWhileEnd p t = S.reverse (S.dropWhile p (S.reverseStream t))
1211−"TEXT dropWhileEnd -> unfused" [1] forall p t.
1212− S.reverse (S.dropWhile p (S.reverseStream t)) = dropWhileEnd p t
1215−-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
1216−-- dropping characters that satisfy the predicate @p@ from both the
1217−-- beginning and end of @t@. Subject to fusion.
1218−dropAround :: (Char -> Bool) -> Text -> Text
1219−dropAround p = dropWhile p . dropWhileEnd p
1220−{-# INLINE [1] dropAround #-}
1222−-- | /O(n)/ Remove leading white space from a string. Equivalent to:
1224−-- > dropWhile isSpace
1225−stripStart :: Text -> Text
1226−stripStart = dropWhile isSpace
1227−{-# INLINE [1] stripStart #-}
1229−-- | /O(n)/ Remove trailing white space from a string. Equivalent to:
1231−-- > dropWhileEnd isSpace
1232−stripEnd :: Text -> Text
1233−stripEnd = dropWhileEnd isSpace
1234−{-# INLINE [1] stripEnd #-}
1236−-- | /O(n)/ Remove leading and trailing white space from a string.
1239−-- > dropAround isSpace
1240−strip :: Text -> Text
1241−strip = dropAround isSpace
1242−{-# INLINE [1] strip #-}
1244-- | /O(n)/ 'splitAt' @n t@ returns a pair whose first element is a
701-- | /O(n)/ 'splitAt' @n t@ returns a pair whose first element is a
1245-- prefix of @t@ of length @n@, and whose second is the remainder of
702-- prefix of @t@ of length @n@, and whose second is the remainder of
1246-- the string. It is equivalent to @('take' n t, 'drop' n t)@.
703-- the string. It is equivalent to @('take' n t, 'drop' n t)@.
@@ -1248,20 +705,28 @@ splitAt :: Int -> Text -> (Text, Text)
1248splitAt n t@(Text arr off len)
705splitAt n t@(Text arr off len)
1249 | n <= 0 = (empty, t)
706 | n <= 0 = (empty, t)
1250 | n >= len = (t, empty)
707 | n >= len = (t, empty)
1251− | otherwise = let k = iterN n t
708+ | otherwise = (Text arr off k, Text arr (off+k) (len-k))
1252− in (text arr off k, text arr (off+k) (len-k))
709+ where k = loop 0 0
711+ | i >= len || cnt >= n = i
712+ | otherwise = loop (i+d) (cnt+1)
714+{-# INLINE splitAt #-}
1254−-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
716+-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns a
1255−-- a pair whose first element is the longest prefix (possibly empty)
717+-- pair whose first element is the longest prefix (possibly empty) of
1256−-- of @t@ of elements that satisfy @p@, and whose second is the
718+-- @t@ of elements that satisfy @p@, and whose second is the remainder
1257−-- remainder of the list.
719+-- of the list.
1258span :: (Char -> Bool) -> Text -> (Text, Text)
720span :: (Char -> Bool) -> Text -> (Text, Text)
1259−span p t = case span_ p t of
721+span p t@(Text arr off len) = (textP arr off k, textP arr (off+k) (len-k))
1260− (# hd,tl #) -> (hd,tl)
722+ where k = loop 0
723+ loop !i | i >= len || not (p c) = i
724+ | otherwise = loop (i+d)
725+ where (c,d) = iter t i
1261{-# INLINE span #-}
726{-# INLINE span #-}
1263−-- | /O(n)/ 'break' is like 'span', but the prefix returned is
728+-- | /O(n)/ 'break' is like 'span', but the prefix returned is over
1264−-- over elements that fail the predicate @p@.
729+-- elements that fail the predicate @p@.
1265break :: (Char -> Bool) -> Text -> (Text, Text)
730break :: (Char -> Bool) -> Text -> (Text, Text)
1266break p = span (not . p)
731break p = span (not . p)
1267{-# INLINE break #-}
732{-# INLINE break #-}
@@ -1273,7 +738,7 @@ groupBy p = loop
1273 loop t@(Text arr off len)
738 loop t@(Text arr off len)
1274 | null t = []
739 | null t = []
1275 | otherwise = text arr off n : loop (text arr (off+n) (len-n))
740 | otherwise = text arr off n : loop (text arr (off+n) (len-n))
1276− where Iter c d = iter t 0
741+ where (c,d) = iter t 0
1277 n = d + findAIndexOrEnd (not . p c) (Text arr (off+d) (len-d))
742 n = d + findAIndexOrEnd (not . p c) (Text arr (off+d) (len-d))
1279-- | Returns the /array/ index (in units of 'Word16') at which a
744-- | Returns the /array/ index (in units of 'Word16') at which a
@@ -1283,8 +748,8 @@ findAIndexOrEnd :: (Char -> Bool) -> Text -> Int
1283findAIndexOrEnd q t@(Text _arr _off len) = go 0
748findAIndexOrEnd q t@(Text _arr _off len) = go 0
1284 where go !i | i >= len || q c = i
749 where go !i | i >= len || q c = i
1285 | otherwise = go (i+d)
750 | otherwise = go (i+d)
1286− where Iter c d = iter t i
751+ where (c,d) = iter t i
1288-- | /O(n)/ Group characters in a string by equality.
753-- | /O(n)/ Group characters in a string by equality.
1289group :: Text -> [Text]
754group :: Text -> [Text]
1290group = groupBy (==)
755group = groupBy (==)
@@ -1302,91 +767,59 @@ tails :: Text -> [Text]
1302tails t | null t = [empty]
767tails t | null t = [empty]
1303 | otherwise = t : tails (unsafeTail t)
768 | otherwise = t : tails (unsafeTail t)
1307−-- Splitting functions in this library do not perform character-wise
1308−-- copies to create substrings; they just construct new 'Text's that
1309−-- are slices of the original.
1311−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'
770+-- | /O(n)/ Break a 'Text' into pieces separated by the 'Char'
1312−-- argument (which cannot be empty), consuming the delimiter. An empty
771+-- argument, consuming the delimiter. I.e.
1313−-- delimiter is invalid, and will cause an error to be raised.
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" == ["",""]
773+-- > split '\n' "a\nb\nd\ne" == ["a","b","d","e"]
774+-- > split 'a' "aXaXaXa" == ["","X","X","X",""]
775+-- > split 'x' "x" == ["",""]
1323−-- > intercalate s . splitOn s == id
779+-- > intercalate (singleton c) . split c == id
1324−-- > splitOn (singleton c) == split (==c)
780+-- > split == splitWith . (==)
1326−-- (Note: the string @s@ to split on above cannot be empty.)
782+-- As for all splitting functions in this library, this function does
1327−--
783+-- not copy the substrings, it just constructs new 'Text's that are
1328−-- In (unlikely) bad cases, this function's time complexity degrades
784+-- slices of the original.
1329−-- towards /O(n*m)/.
785+split :: Char -> Text -> [Text]
1330−splitOn :: Text
786+split c = splitWith (==c)
1331− -- ^ String to split on. If this string is empty, an error
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)
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 #-}
787+{-# INLINE split #-}
1346−"TEXT splitOn/singleton -> split/==" [~1] forall c t.
1347− splitOn (singleton c) t = split (==c) t
1350-- | /O(n)/ Splits a 'Text' into components delimited by separators,
789-- | /O(n)/ Splits a 'Text' into components delimited by separators,
1351-- where the predicate returns True for a separator element. The
790-- where the predicate returns True for a separator element. The
1352-- resulting components do not contain the separators. Two adjacent
791-- resulting components do not contain the separators. Two adjacent
1353-- separators result in an empty component in the output. eg.
792-- separators result in an empty component in the output. eg.
1355−-- > split (=='a') "aabbaca" == ["","","bb","c",""]
794+-- > splitWith (=='a') "aabbaca" == ["","","bb","c",""]
1356−-- > split (=='a') "" == [""]
795+-- > splitWith (=='a') [] == []
1357−split :: (Char -> Bool) -> Text -> [Text]
796+splitWith :: (Char -> Bool) -> Text -> [Text]
1358−split _ t@(Text _off _arr 0) = [t]
797+splitWith _ (Text _off _arr 0) = []
1359−split p t = loop t
798+splitWith p t = loop t
1360 where loop s | null s' = [l]
799 where loop s | null s' = [l]
1361 | otherwise = l : loop (unsafeTail s')
800 | otherwise = l : loop (unsafeTail s')
1362− where (# l, s' #) = span_ (not . p) s
801+ where (l, s') = break p s
1363−{-# INLINE split #-}
802+{-# INLINE splitWith #-}
1365−-- | /O(n)/ Splits a 'Text' into components of length @k@. The last
1366−-- element may be shorter than the other chunks, depending on the
1367−-- length of the input. Examples:
1369−-- > chunksOf 3 "foobarbaz" == ["foo","bar","baz"]
1370−-- > chunksOf 4 "haskell.org" == ["hask","ell.","org"]
1371−chunksOf :: Int -> Text -> [Text]
1374− go t = case splitAt k t of
1375− (a,b) | null a -> []
1376− | otherwise -> a : go b
1377−{-# INLINE chunksOf #-}
1379-- ----------------------------------------------------------------------------
804-- ----------------------------------------------------------------------------
1380-- * Searching
805-- * Searching
807+-------------------------------------------------------------------------------
808+-- ** Searching by equality
810+-- | /O(n)/ 'elem' is the 'Text' membership predicate.
811+elem :: Char -> Text -> Bool
812+elem c t = S.elem c (stream t)
1382-------------------------------------------------------------------------------
815-------------------------------------------------------------------------------
1383-- ** Searching with a predicate
816-- ** Searching with a predicate
1385−-- | /O(n)/ The 'find' function takes a predicate and a 'Text', and
818+-- | /O(n)/ The 'find' function takes a predicate and a 'Text',
1386−-- returns the first element matching the predicate, or 'Nothing' if
819+-- and returns the first element in matching the predicate, or 'Nothing'
1387−-- there is no such element.
820+-- if there is no such element.
1388find :: (Char -> Bool) -> Text -> Maybe Char
821find :: (Char -> Bool) -> Text -> Maybe Char
1389−find p t = S.findBy p (stream t)
822+find p t = S.find p (stream t)
1390{-# INLINE find #-}
823{-# INLINE find #-}
1392-- | /O(n)/ The 'partition' function takes a predicate and a 'Text',
825-- | /O(n)/ The 'partition' function takes a predicate and a 'Text',
@@ -1398,103 +831,49 @@ partition :: (Char -> Bool) -> Text -> (Text, Text)
1398partition p t = (filter p t, filter (not . p) t)
831partition p t = (filter p t, filter (not . p) t)
1399{-# INLINE partition #-}
832{-# INLINE partition #-}
1401−-- | /O(n)/ 'filter', applied to a predicate and a 'Text',
834+-- | /O(n)/ Break a string on a substring, returning a pair of the
1402−-- returns a 'Text' containing those characters that satisfy the
835+-- part of the string prior to the match, and the rest of the string.
1404−filter :: (Char -> Bool) -> Text -> Text
1405−filter p t = unstream (S.filter p (stream t))
1406−{-# INLINE filter #-}
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.
1415−-- > breakOn "::" "a::b::c" ==> ("a", "::b::c")
1416−-- > breakOn "/" "foobar" ==> ("foobar", "")
1420−-- > append prefix match == haystack
1421−-- > where (prefix, match) = breakOn needle haystack
1423−-- If you need to break a string by a substring repeatedly (e.g. you
837+-- The following relationship holds:
1424−-- want to break on every instance of a substring), use 'breakOnAll'
1425−-- instead, as it has lower startup overhead.
1427−-- In (unlikely) bad cases, this function's time complexity degrades
839+-- > break (==c) l == breakSubstring (singleton c) l
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
1434− (x:_) -> (text arr off x, text arr (off+x) (len-x))
1435−{-# INLINE breakOn #-}
1437−-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
1440−-- The first element of the returned tuple is the prefix of @haystack@
841+-- For example, to tokenise a string, dropping delimiters:
1441−-- up to and including the last match of @needle@. The second is the
1442−-- remainder of @haystack@, following the match.
1444−-- > breakOnEnd "::" "a::b::c" ==> ("a::b::", "c")
843+-- > tokenise x y = h : if null t then [] else tokenise x (drop (length x) t)
1445−breakOnEnd :: Text -> Text -> (Text, Text)
844+-- > where (h,t) = breakSubstring x y
1446−breakOnEnd pat src = (reverse b, reverse a)
1447− where (a,b) = breakOn (reverse pat) (reverse src)
1448−{-# INLINE breakOnEnd #-}
1450−-- | /O(n+m)/ Find all non-overlapping instances of @needle@ in
1451−-- @haystack@. Each element of the returned list consists of a pair:
1453−-- * The entire string prior to the /k/th match (i.e. the prefix)
846+-- To skip to the first occurence of a string:
1455−-- * The /k/th match, followed by the remainder of the string
848+-- > snd (breakSubstring x y)
1457−-- Examples:
850+-- To take the parts of a string before a delimiter:
1459−-- > breakOnAll "::" ""
852+-- > fst (breakSubstring x y)
1461−-- > breakOnAll "/" "a/b/c/"
1462−-- > ==> [("a", "/b/c/"), ("a/b", "/c/"), ("a/b/c", "/")]
1464−-- In (unlikely) bad cases, this function's time complexity degrades
1465−-- towards /O(n*m)/.
1467−-- The @needle@ parameter may not be empty.
1468−breakOnAll :: Text -- ^ @needle@ to search for
854+breakSubstring :: Text -- ^ String to search for
1469− -> Text -- ^ @haystack@ in which to search
855+ -> Text -- ^ String to search in
1470− -> [(Text, Text)]
856+ -> (Text,Text) -- ^ Head and tail of string broken at substring
1471−breakOnAll pat src@(Text arr off slen)
857+
1472− | null pat = emptyError "breakOnAll"
858+breakSubstring pat src = search 0 src
1473− | otherwise = L.map step (indices pat src)
1475− step x = (chunk 0 x, chunk x (slen-x))
860+ search !n !s
1476− chunk !n !l = text arr (n+off) l
861+ | null s = (src,empty) -- not found
862+ | pat `isPrefixOf` s = (take n src,s)
863+ | otherwise = search (n+1) (unsafeTail s)
864+{-# INLINE breakSubstring #-}
866+-- | /O(n)/ 'filter', applied to a predicate and a 'Text',
867+-- returns a 'Text' containing those characters that satisfy the
869+filter :: (Char -> Bool) -> Text -> Text
870+filter p t = unstream (S.filter p (stream t))
1477−{-# INLINE breakOnAll #-}
871+{-# INLINE filter #-}
1479-------------------------------------------------------------------------------
874-------------------------------------------------------------------------------
1480-- ** Indexing 'Text's
875-- ** Indexing 'Text's
1484−-- If you think of a 'Text' value as an array of 'Char' values (which
1485−-- it is not), you run the risk of writing inefficient code.
1487−-- An idiom that is common in some languages is to find the numeric
1488−-- offset of a character or substring, then use that number to split
1489−-- or trim the searched string. With a 'Text' value, this approach
1490−-- would require two /O(n)/ operations: one to perform the search, and
1491−-- one to operate from wherever the search ended.
1493−-- For example, suppose you have a string that you want to split on
1494−-- the substring @\"::\"@, such as @\"foo::bar::quux\"@. Instead of
1495−-- searching for the index of @\"::\"@ and taking the substrings
1496−-- before and after that index, you would instead use @breakOnAll \"::\"@.
1498-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
877-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
1499index :: Text -> Int -> Char
878index :: Text -> Int -> Char
1500index t n = S.index (stream t) n
879index t n = S.index (stream t) n
@@ -1502,34 +881,39 @@ index t n = S.index (stream t) n
1503-- | /O(n)/ The 'findIndex' function takes a predicate and a 'Text'
882-- | /O(n)/ The 'findIndex' function takes a predicate and a 'Text'
1504-- and returns the index of the first element in the 'Text' satisfying
883-- and returns the index of the first element in the 'Text' satisfying
1505−-- the predicate. Subject to fusion.
884+-- the predicate. This function is subject to fusion.
1506findIndex :: (Char -> Bool) -> Text -> Maybe Int
885findIndex :: (Char -> Bool) -> Text -> Maybe Int
1507findIndex p t = S.findIndex p (stream t)
886findIndex p t = S.findIndex p (stream t)
1508{-# INLINE findIndex #-}
887{-# INLINE findIndex #-}
1510−-- | /O(n+m)/ The 'count' function returns the number of times the
889+-- | The 'findIndices' function extends 'findIndex', by returning the
1511−-- query string appears in the given 'Text'. An empty query string is
890+-- indices of all elements satisfying the predicate, in ascending
1512−-- invalid, and will cause an error to be raised.
891+-- order. This function is subject to fusion.
1513−--
892+findIndices :: (Char -> Bool) -> Text -> [Int]
1514−-- In (unlikely) bad cases, this function's time complexity degrades
893+findIndices p t = S.findIndices p (stream t)
1515−-- towards /O(n*m)/.
1516−count :: Text -> Text -> Int
1518− | null pat = emptyError "count"
1519− | isSingleton pat = countChar (unsafeHead pat) src
1520− | otherwise = L.length (indices pat src)
1521−{-# INLINE [1] count #-}
894+{-# INLINE findIndices #-}
1524−"TEXT count/singleton -> countChar" [~1] forall c t.
1525− count (singleton c) t = countChar c t
896+-- | /O(n)/ The 'elemIndex' function returns the index of the first
897+-- element in the given 'Text' which is equal to the query element, or
898+-- 'Nothing' if there is no such element. This function is subject to
900+elemIndex :: Char -> Text -> Maybe Int
901+elemIndex c t = S.elemIndex c (stream t)
902+{-# INLINE elemIndex #-}
904+-- | /O(n)/ The 'elemIndices' function returns the index of every
905+-- element in the given 'Text' which is equal to the query
906+-- element. This function is subject to fusion.
907+elemIndices :: Char -> Text -> [Int]
908+elemIndices c t = S.elemIndices c (stream t)
909+{-# INLINE elemIndices #-}
1528−-- | /O(n)/ The 'countChar' function returns the number of times the
911+-- | /O(n)/ The 'count' function returns the number of times the query
1529−-- query element appears in the given 'Text'. Subject to fusion.
912+-- element appears in the given 'Text'. This function is subject to
1530−countChar :: Char -> Text -> Int
914+count :: Char -> Text -> Int
1531−countChar c t = S.countChar c (stream t)
915+count c t = S.count c (stream t)
1532−{-# INLINE countChar #-}
916+{-# INLINE count #-}
1534-------------------------------------------------------------------------------
918-------------------------------------------------------------------------------
1535-- * Zipping
919-- * Zipping
@@ -1540,15 +924,13 @@ countChar c t = S.countChar c (stream t)
1540-- equivalent to a pair of 'unpack' operations.
924-- equivalent to a pair of 'unpack' operations.
1541zip :: Text -> Text -> [(Char,Char)]
925zip :: Text -> Text -> [(Char,Char)]
1542zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
926zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1543−{-# INLINE zip #-}
927+{-# INLINE [0] zip #-}
1545-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
929-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
1546-- given as the first argument, instead of a tupling function.
930-- given as the first argument, instead of a tupling function.
1547−-- Performs replacement on invalid scalar values.
1548zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
931zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1549−zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))
932+zipWith f t1 t2 = unstream (S.zipWith f (stream t1) (stream t2))
1550− where g a b = safe (f a b)
1551−{-# INLINE zipWith #-}
933+{-# INLINE [0] zipWith #-}
1553-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
935-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
1554-- representing white space.
936-- representing white space.
@@ -1564,7 +946,7 @@ words t@(Text arr off len) = loop 0 0
1564 then loop (start+1) (start+1)
946 then loop (start+1) (start+1)
1565 else Text arr (start+off) (n-start) : loop (n+d) (n+d)
947 else Text arr (start+off) (n-start) : loop (n+d) (n+d)
1566 | otherwise = loop start (n+d)
948 | otherwise = loop start (n+d)
1567− where Iter c d = iter t n
949+ where (c,d) = iter t n
1568{-# INLINE words #-}
950{-# INLINE words #-}
1570-- | /O(n)/ Breaks a 'Text' up into a list of 'Text's at
952-- | /O(n)/ Breaks a 'Text' up into a list of 'Text's at
@@ -1574,10 +956,9 @@ lines ps | null ps = []
1574 | otherwise = h : if null t
956 | otherwise = h : if null t
1576 else lines (unsafeTail t)
958 else lines (unsafeTail t)
1577− where (# h,t #) = span_ (/= '\n') ps
959+ where (h,t) = span (/= '\n') ps
1578{-# INLINE lines #-}
960{-# INLINE lines #-}
1581-- | /O(n)/ Portably breaks a 'Text' up into a list of 'Text's at line
962-- | /O(n)/ Portably breaks a 'Text' up into a list of 'Text's at line
1582-- boundaries.
963-- boundaries.
@@ -1585,6 +966,7 @@ lines ps | null ps = []
1585-- return immediately followed by a line feed, or a carriage return.
966-- return immediately followed by a line feed, or a carriage return.
1586-- This accounts for both Unix and Windows line ending conventions,
967-- This accounts for both Unix and Windows line ending conventions,
1587-- and for the old convention used on Mac OS 9 and earlier.
968-- and for the old convention used on Mac OS 9 and earlier.
1588lines' :: Text -> [Text]
970lines' :: Text -> [Text]
1589lines' ps | null ps = []
971lines' ps | null ps = []
1590 | otherwise = h : case uncons t of
972 | otherwise = h : case uncons t of
@@ -1611,7 +993,8 @@ unwords = intercalate (singleton ' ')
1611{-# INLINE unwords #-}
993{-# INLINE unwords #-}
1613-- | /O(n)/ The 'isPrefixOf' function takes two 'Text's and returns
995-- | /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.
996+-- 'True' iff the first is a prefix of the second. This function is
997+-- subject to fusion.
1615isPrefixOf :: Text -> Text -> Bool
998isPrefixOf :: Text -> Text -> Bool
1616isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
999isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
1617 alen <= blen && S.isPrefixOf (stream a) (stream b)
1000 alen <= blen && S.isPrefixOf (stream a) (stream b)
@@ -1620,6 +1003,8 @@ isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
1620{-# RULES
1003{-# RULES
1621"TEXT isPrefixOf -> fused" [~1] forall s t.
1004"TEXT isPrefixOf -> fused" [~1] forall s t.
1622 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1005 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1006+"TEXT isPrefixOf -> unfused" [1] forall s t.
1007+ S.isPrefixOf (stream s) (stream t) = isPrefixOf s t
1625-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
1010-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
@@ -1632,124 +1017,13 @@ isSuffixOf a@(Text _aarr _aoff alen) b@(Text barr boff blen) =
1632 | otherwise = Text barr (boff+d) alen
1017 | otherwise = Text barr (boff+d) alen
1633{-# INLINE isSuffixOf #-}
1018{-# INLINE isSuffixOf #-}
1635−-- | /O(n+m)/ The 'isInfixOf' function takes two 'Text's and returns
1020+-- | /O(n)/ The 'isInfixOf' function takes two 'Text's and returns
1636-- 'True' iff the first is contained, wholly and intact, anywhere
1021-- 'True' iff the first is contained, wholly and intact, anywhere
1637-- within the second.
1022-- within the second.
1639−-- In (unlikely) bad cases, this function's time complexity degrades
1640−-- towards /O(n*m)/.
1641isInfixOf :: Text -> Text -> Bool
1023isInfixOf :: Text -> Text -> Bool
1642−isInfixOf needle haystack
1024+isInfixOf needle haystack = L.any (isPrefixOf needle) (tails haystack)
1643− | null needle = True
1025+{-# INLINE isInfixOf #-}
1644− | isSingleton needle = S.elem (unsafeHead needle) . S.stream $ haystack
1026+-- TODO: a better implementation
1645− | otherwise = not . L.null . indices needle $ haystack
1646−{-# INLINE [1] isInfixOf #-}
1649−"TEXT isInfixOf/singleton -> S.elem/S.stream" [~1] forall n h.
1650− isInfixOf (singleton n) h = S.elem n (S.stream h)
1653−-------------------------------------------------------------------------------
1656−-- | /O(n)/ Return the suffix of the second string if its prefix
1657−-- matches the entire first string.
1661−-- > stripPrefix "foo" "foobar" == Just "bar"
1662−-- > stripPrefix "" "baz" == Just "baz"
1663−-- > stripPrefix "foo" "quux" == Nothing
1665−-- This is particularly useful with the @ViewPatterns@ extension to
1668−-- > {-# LANGUAGE ViewPatterns #-}
1669−-- > import Data.Text as T
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
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
1683−-- If the strings do not have a common prefix or either one is empty,
1684−-- this function returns '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
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
1702−-- | /O(n)/ Return the prefix of the second string if its suffix
1703−-- matches the entire first string.
1707−-- > stripSuffix "bar" "foobar" == Just "foo"
1708−-- > stripSuffix "" "baz" == Just "baz"
1709−-- > stripSuffix "foo" "quux" == Nothing
1711−-- This is particularly useful with the @ViewPatterns@ extension to
1714−-- > {-# LANGUAGE ViewPatterns #-}
1715−-- > import Data.Text as T
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
1725−-- | Add a list of non-negative numbers. Errors out on overflow.
1726−sumP :: String -> [Int] -> Int
1729− | ax >= 0 = go ax xs
1730− | otherwise = overflowError fun
1734emptyError :: String -> a
1028emptyError :: String -> a
1735−emptyError fun = P.error $ "Data.Text." ++ fun ++ ": empty input"
1029+emptyError fun = P.error ("Data.Text." ++ fun ++ ": empty input")
1737−overflowError :: String -> a
1738−overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"
1740−-- | /O(n)/ Make a distinct copy of the given string, sharing no
1741−-- storage with the original string.
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
1751− go :: ST s (A.MArray s)
1754− A.copyI marr 0 arr off len