@@ -1,32 +1,48 @@
1−{-# LANGUAGE BangPatterns, CPP, Rank2Types #-}
1+{-# LANGUAGE BangPatterns, CPP, MagicHash, Rank2Types, UnboxedTuples #-}
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 #-}
5-- Module : Data.Text
11-- Module : Data.Text
6−-- Copyright : (c) 2008, 2009 Tom Harper,
7−-- (c) 2009, 2010 Bryan O'Sullivan,
12+-- Copyright : (c) 2009, 2010, 2011, 2012 Bryan O'Sullivan,
8−-- (c) 2009 Duncan Coutts
13+-- (c) 2009 Duncan Coutts,
14+-- (c) 2008, 2009 Tom Harper
10-- License : BSD-style
16-- License : BSD-style
11−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,
17+-- Maintainer : bos@serpentine.com
13-- Stability : experimental
18-- Stability : experimental
14-- Portability : GHC
19-- Portability : GHC
16−-- A time and space-efficient implementation of Unicode text using
21+-- A time and space-efficient implementation of Unicode text.
17−-- packed Word16 arrays. Suitable for performance critical use, both
22+-- Suitable for performance critical use, both in terms of large data
18−-- in terms of large data quantities and high speed.
23+-- quantities and high speed.
25+-- /Note/: Read below the synopsis for important notes on the use of
20-- This module is intended to be imported @qualified@, to avoid name
28-- This module is intended to be imported @qualified@, to avoid name
21-- clashes with "Prelude" functions, e.g.
29-- clashes with "Prelude" functions, e.g.
23-- > import qualified Data.Text as T
31-- > 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 >.
25module Data.Text
38module Data.Text
27 -- * Strict vs lazy types
40 -- * Strict vs lazy types
28 -- $strict
41 -- $strict
30 -- * Fusion
46 -- * Fusion
31 -- $fusion
47 -- $fusion
@@ -65,6 +81,7 @@ module Data.Text
65 , toCaseFold
81 , toCaseFold
69 -- ** Justification
86 -- ** Justification
70 , justifyLeft
87 , justifyLeft
@@ -108,8 +125,11 @@ module Data.Text
109 -- ** Breaking strings
126 -- ** Breaking strings
112 , takeWhile
131 , takeWhile
113 , dropWhile
133 , dropWhile
114 , dropWhileEnd
134 , dropWhileEnd
115 , dropAround
135 , dropAround
@@ -147,6 +167,7 @@ module Data.Text
147 -- ** View patterns
167 -- ** View patterns
148 , stripPrefix
168 , stripPrefix
149 , stripSuffix
169 , stripSuffix
151 -- * Searching
172 -- * Searching
@@ -155,80 +176,125 @@ module Data.Text
155 , partition
176 , partition
157 -- , findSubstring
178 -- , findSubstring
159 -- * Indexing
180 -- * Indexing
160 -- $index
181 -- $index
162 , findIndex
183 , findIndex
165− -- * Zipping and unzipping
186+ -- * Zipping
167 , zipWith
188 , zipWith
169 -- -* Ordered text
190 -- -* Ordered text
170 -- , sort
191 -- , sort
193+ -- * Low level operations
173import Prelude (Char, Bool(..), Int, Maybe(..), String,
198import Prelude (Char, Bool(..), Int, Maybe(..), String,
174 Eq(..), Ord(..), Ordering(..), (++),
199 Eq(..), Ord(..), Ordering(..), (++),
175− Read(..), Show(..),
200+ Read(..),
176− (&&), (||), (+), (-), (.), ($), ($!), (>>), (*),
201+ (&&), (||), (+), (-), (.), ($), ($!), (>>),
177− div, maxBound, not, return, otherwise)
202+ not, return, otherwise, quot)
178#if defined(HAVE_DEEPSEQ)
203#if defined(HAVE_DEEPSEQ)
179−import Control.DeepSeq (NFData)
204+import Control.DeepSeq (NFData(rnf))
181#if defined(ASSERTS)
206#if defined(ASSERTS)
182import Control.Exception (assert)
207import Control.Exception (assert)
184import Data.Char (isSpace)
209import Data.Char (isSpace)
185−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))
210+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,
186−#if __GLASGOW_HASKELL__ >= 612
211+ Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))
187−import Data.Data (mkNoRepType)
189−import Data.Data (mkNorepType)
191import Control.Monad (foldM)
212import Control.Monad (foldM)
213+import Control.Monad.ST (ST)
192import qualified Data.Text.Array as A
214import qualified Data.Text.Array as A
193import qualified Data.List as L
215import qualified Data.List as L
216+import Data.Binary (Binary(get, put))
194import Data.Monoid (Monoid(..))
217import Data.Monoid (Monoid(..))
218+#if MIN_VERSION_base(4,9,0)
219+import Data.Semigroup (Semigroup(..))
195import Data.String (IsString(..))
221import Data.String (IsString(..))
196−import qualified Data.Text.Fusion as S
222+import qualified Data.Text.Internal.Fusion as S
197−import qualified Data.Text.Fusion.Common as S
223+import qualified Data.Text.Internal.Fusion.Common as S
224+import Data.Text.Encoding (decodeUtf8', encodeUtf8)
198−import Data.Text.Fusion (stream, reverseStream, unstream)
225+import Data.Text.Internal.Fusion (stream, reverseStream, unstream)
199−import Data.Text.Internal (Text(..), empty, text, textP)
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#)
200import qualified Prelude as P
229import qualified Prelude as P
201import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,
230import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,
202− unsafeHead, unsafeTail)
231+ reverseIter_, unsafeHead, unsafeTail)
203−import Data.Text.UnsafeChar (unsafeChr)
232+import Data.Text.Internal.Unsafe.Char (unsafeChr)
204−import qualified Data.Text.Util as U
233+import qualified Data.Text.Internal.Functions as F
205−import qualified Data.Text.Encoding.Utf16 as U16
234+import qualified Data.Text.Internal.Encoding.Utf16 as U16
206−import Data.Text.Search (indices)
235+import Data.Text.Internal.Search (indices)
207#if defined(__HADDOCK__)
236#if defined(__HADDOCK__)
208import Data.ByteString (ByteString)
237import Data.ByteString (ByteString)
209import qualified Data.Text.Lazy as L
238import qualified Data.Text.Lazy as L
210import Data.Int (Int64)
239import 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)
213-- $strict
249-- $strict
215-- This package provides both strict and lazy 'Text' types. The
251-- This package provides both strict and lazy 'Text' types. The
216−-- strict type is provided by the 'Data.Text' package, while the lazy
252+-- strict type is provided by the "Data.Text" module, while the lazy
217−-- type is provided by the 'Data.Text.Lazy' package. Internally, the
253+-- type is provided by the "Data.Text.Lazy" module. Internally, the
218-- lazy @Text@ type consists of a list of strict chunks.
254-- lazy @Text@ type consists of a list of strict chunks.
220-- The strict 'Text' type requires that an entire string fit into
256-- The strict 'Text' type requires that an entire string fit into
221−-- memory at once. The lazy @Text@ type is capable of streaming
257+-- memory at once. The lazy 'Data.Text.Lazy.Text' type is capable of
222−-- strings that are larger than memory using a small memory footprint.
258+-- streaming strings that are larger than memory using a small memory
223−-- In many cases, the overhead of chunked streaming makes the lazy
259+-- footprint. In many cases, the overhead of chunked streaming makes
224−-- @Text@ type slower than its strict counterpart, but this is not
260+-- the lazy 'Data.Text.Lazy.Text' type slower than its strict
225−-- always the case. Sometimes, the time complexity of a function in
261+-- counterpart, but this is not always the case. Sometimes, the time
226−-- one module may be different from the other, due to their differing
262+-- complexity of a function in one module may be different from the
227−-- internal structures.
263+-- other, due to their differing internal structures.
229-- Each module provides an almost identical API, with the main
265-- Each module provides an almost identical API, with the main
230-- difference being that the strict module uses 'Int' values for
266-- difference being that the strict module uses 'Int' values for
231−-- lengths and counts, while the lazy module uses 'Int64' lengths.
267+-- lengths and counts, while the lazy module uses 'Data.Int.Int64'
272+-- A 'Text' value is a sequence of Unicode scalar values, as defined
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
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
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 >.)
233-- $fusion
299-- $fusion
@@ -240,18 +306,19 @@ import Data.Int (Int64)
241-- > import Data.Text as T
307-- > import Data.Text as T
242-- > import Data.Text.Encoding as E
308-- > import Data.Text.Encoding as E
309+-- > import Data.ByteString (ByteString)
244-- > countChars :: ByteString -> Int
311-- > countChars :: ByteString -> Int
245-- > countChars = T.length . T.toUpper . E.decodeUtf8
312-- > countChars = T.length . T.toUpper . E.decodeUtf8
247-- From the type signatures involved, this looks like it should
314-- From the type signatures involved, this looks like it should
248−-- allocate one 'ByteString' value, and two 'Text' values. However,
315+-- allocate one 'Data.ByteString.ByteString' value, and two 'Text'
249−-- when a module is compiled with optimisation enabled under GHC, the
316+-- values. However, when a module is compiled with optimisation
250−-- two intermediate 'Text' values will be optimised away, and the
317+-- enabled under GHC, the two intermediate 'Text' values will be
251−-- function will be compiled down to a single loop over the source
318+-- optimised away, and the function will be compiled down to a single
252−-- 'ByteString'.
319+-- loop over the source 'Data.ByteString.ByteString'.
254−-- Functions that can be fused by the compiler are marked with the
321+-- Functions that can be fused by the compiler are documented with the
255-- phrase \"Subject to fusion\".
322-- phrase \"Subject to fusion\".
257instance Eq Text where
324instance Eq Text where
@@ -263,46 +330,83 @@ instance Eq Text where
263instance Ord Text where
330instance Ord Text where
264 compare = compareText
331 compare = compareText
266−instance Show Text where
267− showsPrec p ps r = showsPrec p (unpack ps) r
269instance Read Text where
333instance Read Text where
270 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
334 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]
336+#if MIN_VERSION_base(4,9,0)
337+-- Semigroup orphan instances for older GHCs are provided by
338+-- 'semigroups` package
340+instance Semigroup Text where
272instance Monoid Text where
344instance Monoid Text where
273 mempty = empty
345 mempty = empty
346+#if MIN_VERSION_base(4,9,0)
347+ mappend = (<>) -- future-proof definition
274 mappend = append
349 mappend = append
275 mconcat = concat
351 mconcat = concat
277instance IsString Text where
353instance IsString Text where
278 fromString = pack
354 fromString = pack
356+#if __GLASGOW_HASKELL__ >= 708
357+instance Exts.IsList Text where
358+ type Item Text = Char
280#if defined(HAVE_DEEPSEQ)
363#if defined(HAVE_DEEPSEQ)
281−instance NFData Text
364+instance NFData Text where rnf !_ = ()
367+instance Binary Text where
368+ put t = put (encodeUtf8 t)
371+ case decodeUtf8' bs of
372+ P.Left exn -> P.fail (P.show exn)
373+ P.Right a -> P.return a
284−-- This instance preserves data abstraction at the cost of inefficiency.
375+-- | This instance preserves data abstraction at the cost of inefficiency.
285-- We omit reflection services for the sake of data abstraction.
376-- We omit reflection services for the sake of data abstraction.
287−-- This instance was created by copying the behavior of Data.Set and
288−-- Data.Map. If you feel a mistake has been made, please feel free to
289−-- submit improvements.
291−-- Original discussion is archived here:
378+-- This instance was created by copying the updated behavior of
292−
379+-- @"Data.Set".@'Data.Set.Set' and @"Data.Map".@'Data.Map.Map'. If you
293−-- "could we get a Data instance for Data.Text.Text?"
380+-- feel a mistake has been made, please feel free to submit
383+-- The original discussion is archived here:
294−-- http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93
384+-- <http://groups.google.com/group/haskell-cafe/browse_thread/thread/b5bbb1b28a7e525d/0639d46852575b93 could we get a Data instance for Data.Text.Text? >
386+-- The followup discussion that changed the behavior of 'Data.Set.Set'
387+-- and 'Data.Map.Map' is archived here:
388+-- <http://markmail.org/message/trovdc6zkphyi3cr#query:+page:1+mid:a46der3iacwjcf6n+state:results Proposal: Allow gunfold for Data.Map, ... >
296instance Data Text where
390instance Data Text where
297 gfoldl f z txt = z pack `f` (unpack txt)
391 gfoldl f z txt = z pack `f` (unpack txt)
298− toConstr _ = P.error "Data.Text.Text.toConstr"
392+ toConstr _ = packConstr
393+ gunfold k z c = case constrIndex c of
299− gunfold _ _ = P.error "Data.Text.Text.gunfold"
395+ _ -> P.error "gunfold"
300−#if __GLASGOW_HASKELL__ >= 612
396+ dataTypeOf _ = textDataType
301− dataTypeOf _ = mkNoRepType "Data.Text.Text"
397+
302−#else
398+#if MIN_VERSION_base(4,7,0)
303− dataTypeOf _ = mkNorepType "Data.Text.Text"
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]
306-- | /O(n)/ Compare two 'Text' values lexicographically.
410-- | /O(n)/ Compare two 'Text' values lexicographically.
307compareText :: Text -> Text -> Ordering
411compareText :: Text -> Text -> Ordering
308compareText ta@(Text _arrA _offA lenA) tb@(Text _arrB _offB lenB)
412compareText ta@(Text _arrA _offA lenA) tb@(Text _arrB _offB lenB)
@@ -320,36 +424,30 @@ compareText ta@(Text _arrA _offA lenA) tb@(Text _arrB _offB lenB)
320-- -----------------------------------------------------------------------------
424-- -----------------------------------------------------------------------------
321-- * Conversion to/from 'Text'
425-- * Conversion to/from 'Text'
323−-- | /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.
324pack :: String -> Text
429pack :: String -> Text
325−pack = unstream . S.streamList
430+pack = unstream . S.map safe . S.streamList
326{-# INLINE [1] pack #-}
431{-# INLINE [1] pack #-}
328−-- | /O(n)/ Convert a Text into a String. Subject to fusion.
329−unpack :: Text -> String
330−unpack = S.unstreamList . stream
331−{-# INLINE [1] unpack #-}
333−-- | /O(1)/ Convert a character into a Text.
334−-- Subject to fusion.
335−singleton :: Char -> Text
336−singleton = unstream . S.singleton
337−{-# INLINE [1] singleton #-}
339-- -----------------------------------------------------------------------------
433-- -----------------------------------------------------------------------------
340-- * Basic functions
434-- * Basic functions
342-- | /O(n)/ Adds a character to the front of a 'Text'. This function
436-- | /O(n)/ Adds a character to the front of a 'Text'. This function
343-- is more costly than its 'List' counterpart because it requires
437-- is more costly than its 'List' counterpart because it requires
344−-- copying a new array. Subject to fusion.
438+-- copying a new array. Subject to fusion. Performs replacement on
439+-- invalid scalar values.
345cons :: Char -> Text -> Text
440cons :: Char -> Text -> Text
346−cons c t = unstream (S.cons c (stream t))
441+cons c t = unstream (S.cons (safe c) (stream t))
347{-# INLINE cons #-}
442{-# INLINE cons #-}
349-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
446-- | /O(n)/ Adds a character to the end of a 'Text'. This copies the
350-- entire array in the process, unless fused. Subject to fusion.
447-- entire array in the process, unless fused. Subject to fusion.
448+-- Performs replacement on invalid scalar values.
351snoc :: Text -> Char -> Text
449snoc :: Text -> Char -> Text
352−snoc t c = unstream (S.snoc (stream t) c)
450+snoc t c = unstream (S.snoc (stream t) (safe c))
353{-# INLINE snoc #-}
451{-# INLINE snoc #-}
355-- | /O(n)/ Appends one 'Text' to the other by copying both of them
453-- | /O(n)/ Appends one 'Text' to the other by copying both of them
@@ -362,12 +460,13 @@ append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)
362 | otherwise = overflowError "append"
460 | otherwise = overflowError "append"
364 len = len1+len2
462 len = len1+len2
463+ x :: ST s (A.MArray s)
366 arr <- A.new len
465 arr <- A.new len
367 A.copyI arr 0 arr1 off1 len1
466 A.copyI arr 0 arr1 off1 len1
368 A.copyI arr len1 arr2 off2 len
467 A.copyI arr len1 arr2 off2 len
369 return arr
468 return arr
370−{-# INLINE append #-}
469+{-# NOINLINE append #-}
373"TEXT append -> fused" [~1] forall t1 t2.
472"TEXT append -> fused" [~1] forall t1 t2.
@@ -387,8 +486,8 @@ head t = S.head (stream t)
387uncons :: Text -> Maybe (Char, Text)
486uncons :: Text -> Maybe (Char, Text)
388uncons t@(Text arr off len)
487uncons t@(Text arr off len)
389 | len <= 0 = Nothing
488 | len <= 0 = Nothing
489+ | otherwise = Just $ let !(Iter c d) = iter t 0
390− | otherwise = Just (c, textP arr (off+d) (len-d))
490+ in (c, text arr (off+d) (len-d))
391− where Iter c d = iter t 0
392{-# INLINE [1] uncons #-}
491{-# INLINE [1] uncons #-}
394-- | Lifted from Control.Arrow and specialized.
493-- | Lifted from Control.Arrow and specialized.
@@ -418,7 +517,7 @@ last (Text arr off len)
418tail :: Text -> Text
517tail :: Text -> Text
419tail t@(Text arr off len)
518tail t@(Text arr off len)
420 | len <= 0 = emptyError "tail"
519 | len <= 0 = emptyError "tail"
421− | otherwise = textP arr (off+d) (len-d)
520+ | otherwise = text arr (off+d) (len-d)
422 where d = iter_ t 0
521 where d = iter_ t 0
423{-# INLINE [1] tail #-}
522{-# INLINE [1] tail #-}
@@ -433,8 +532,8 @@ tail t@(Text arr off len)
433-- be non-empty. Subject to fusion.
532-- be non-empty. Subject to fusion.
434init :: Text -> Text
533init :: Text -> Text
435init (Text arr off len) | len <= 0 = emptyError "init"
534init (Text arr off len) | len <= 0 = emptyError "init"
436− | n >= 0xDC00 && n <= 0xDFFF = textP arr off (len-2)
535+ | n >= 0xDC00 && n <= 0xDFFF = text arr off (len-2)
437− | otherwise = textP arr off (len-1)
536+ | otherwise = text arr off (len-1)
439 n = A.unsafeIndex arr (off+len-1)
538 n = A.unsafeIndex arr (off+len-1)
440{-# INLINE [1] init #-}
539{-# INLINE [1] init #-}
@@ -473,7 +572,9 @@ isSingleton = S.isSingleton . stream
473-- Subject to fusion.
572-- Subject to fusion.
474length :: Text -> Int
573length :: Text -> Int
475length t = S.length (stream t)
574length t = S.length (stream t)
476−{-# 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.
478-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
579-- | /O(n)/ Compare the count of characters in a 'Text' to a number.
479-- Subject to fusion.
580-- Subject to fusion.
@@ -492,53 +593,55 @@ compareLength t n = S.compareLengthI (stream t) n
494"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
595"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.
495− (==) (length t) n = compareLength t n == EQ
596+ eqInt (length t) n = compareLength t n == EQ
499"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
600"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.
500− (/=) (length t) n = compareLength t n /= EQ
601+ neInt (length t) n = compareLength t n /= EQ
504"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
605"TEXT <N/length -> compareLength/==LT" [~1] forall t n.
505− (<) (length t) n = compareLength t n == LT
606+ ltInt (length t) n = compareLength t n == LT
509"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
610"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.
510− (<=) (length t) n = compareLength t n /= GT
611+ leInt (length t) n = compareLength t n /= GT
514"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
615"TEXT >N/length -> compareLength/==GT" [~1] forall t n.
515− (>) (length t) n = compareLength t n == GT
616+ gtInt (length t) n = compareLength t n == GT
519"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
620"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.
520− (>=) (length t) n = compareLength t n /= LT
621+ geInt (length t) n = compareLength t n /= LT
523-- -----------------------------------------------------------------------------
624-- -----------------------------------------------------------------------------
524-- * Transformations
625-- * Transformations
525-- | /O(n)/ 'map' @f@ @t@ is the 'Text' obtained by applying @f@ to
626-- | /O(n)/ 'map' @f@ @t@ is the 'Text' obtained by applying @f@ to
526−-- each element of @t@. Subject to fusion.
627+-- each element of @t@. Subject to fusion. Performs replacement on
628+-- invalid scalar values.
527map :: (Char -> Char) -> Text -> Text
629map :: (Char -> Char) -> Text -> Text
528−map f t = unstream (S.map f (stream t))
630+map f t = unstream (S.map (safe . f) (stream t))
529{-# INLINE [1] map #-}
631{-# INLINE [1] map #-}
531-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
633-- | /O(n)/ The 'intercalate' function takes a 'Text' and a list of
532-- 'Text's and concatenates the list after interspersing the first
634-- 'Text's and concatenates the list after interspersing the first
533-- argument between each element of the list.
635-- argument between each element of the list.
534intercalate :: Text -> [Text] -> Text
636intercalate :: Text -> [Text] -> Text
535−intercalate t = concat . (U.intersperse t)
637+intercalate t = concat . (F.intersperse t)
536{-# INLINE intercalate #-}
638{-# INLINE intercalate #-}
538-- | /O(n)/ The 'intersperse' function takes a character and places it
640-- | /O(n)/ The 'intersperse' function takes a character and places it
539−-- 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.
540intersperse :: Char -> Text -> Text
643intersperse :: Char -> Text -> Text
541−intersperse c t = unstream (S.intersperse c (stream t))
644+intersperse c t = unstream (S.intersperse (safe c) (stream t))
542{-# INLINE intersperse #-}
645{-# INLINE intersperse #-}
544-- | /O(n)/ Reverse the characters of a string. Subject to fusion.
647-- | /O(n)/ Reverse the characters of a string. Subject to fusion.
@@ -546,16 +649,59 @@ reverse :: Text -> Text
546reverse t = S.reverse (stream t)
649reverse t = S.reverse (stream t)
547{-# INLINE reverse #-}
650{-# INLINE reverse #-}
549−-- | /O(m+n)/ Replace every occurrence of one substring with another.
652+-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in
653+-- @haystack@ with @replacement@.
655+-- This function behaves as though it was defined as follows:
658+-- replace needle replacement haystack =
659+-- 'intercalate' replacement ('splitOn' needle haystack)
662+-- As this suggests, each occurrence is replaced exactly once. So if
663+-- @needle@ occurs in @replacement@, that occurrence will /not/ itself
664+-- be replaced recursively:
666+-- > replace "oo" "foo" "oo" == "foo"
668+-- In cases where several instances of @needle@ overlap, only the
669+-- first one will be replaced:
671+-- > replace "ofo" "bar" "ofofo" == "barfo"
551-- In (unlikely) bad cases, this function's time complexity degrades
673-- In (unlikely) bad cases, this function's time complexity degrades
552-- towards /O(n*m)/.
674-- towards /O(n*m)/.
553−replace :: Text -- ^ Text to search for
675+replace :: Text
554− -> Text -- ^ Replacement text
676+ -- ^ @needle@ to search for. If this string is empty, an
677+ -- error will occur.
555− -> Text -- ^ Input text
678+ -> Text
679+ -- ^ @replacement@ to replace @needle@ with.
557−replace s d = intercalate d . splitOn s
681+ -- ^ @haystack@ in which to search.
558−{-# INLINE replace #-}
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
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
560-- ----------------------------------------------------------------------------
706-- ----------------------------------------------------------------------------
561-- ** Case conversions (folds)
707-- ** Case conversions (folds)
@@ -574,12 +720,13 @@ replace s d = intercalate d . splitOn s
574-- /Note/: In some languages, case conversion is a locale- and
720-- /Note/: In some languages, case conversion is a locale- and
575-- context-dependent operation. The case conversion functions in this
721-- context-dependent operation. The case conversion functions in this
576-- module are /not/ locale sensitive. Programs that require locale
722-- module are /not/ locale sensitive. Programs that require locale
577−-- sensitivity should use appropriate versions of the case mapping
723+-- sensitivity should use appropriate versions of the
578−-- functions from the @text-icu@ package.
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 >.
580−-- | /O(n)/ Convert a string to folded case. This function is mainly
726+-- | /O(n)/ Convert a string to folded case. Subject to fusion.
581−-- useful for performing caseless (also known as case insensitive)
728+-- This function is mainly useful for performing caseless (also known
582−-- string comparisons.
729+-- as case insensitive) string comparisons.
584-- A string @x@ is a caseless match for a string @y@ if and only if:
731-- A string @x@ is a caseless match for a string @y@ if and only if:
@@ -594,27 +741,56 @@ replace s d = intercalate d . splitOn s
594-- instead of itself.
741-- instead of itself.
595toCaseFold :: Text -> Text
742toCaseFold :: Text -> Text
596toCaseFold t = unstream (S.toCaseFold (stream t))
743toCaseFold t = unstream (S.toCaseFold (stream t))
597−{-# INLINE [0] toCaseFold #-}
744+{-# INLINE toCaseFold #-}
599-- | /O(n)/ Convert a string to lower case, using simple case
746-- | /O(n)/ Convert a string to lower case, using simple case
747+-- conversion. Subject to fusion.
600−-- conversion. The result string may be longer than the input string.
749+-- The result string may be longer than the input string. For
601−-- For instance, \"İ\" (Latin capital letter I with dot above,
750+-- instance, \"İ\" (Latin capital letter I with dot above,
602−-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069) followed
751+-- U+0130) maps to the sequence \"i\" (Latin small letter i, U+0069)
603−-- by \" ̇\" (combining dot above, U+0307).
752+-- followed by \" ̇\" (combining dot above, U+0307).
604toLower :: Text -> Text
753toLower :: Text -> Text
605toLower t = unstream (S.toLower (stream t))
754toLower t = unstream (S.toLower (stream t))
606{-# INLINE toLower #-}
755{-# INLINE toLower #-}
608-- | /O(n)/ Convert a string to upper case, using simple case
757-- | /O(n)/ Convert a string to upper case, using simple case
758+-- conversion. Subject to fusion.
609−-- conversion. The result string may be longer than the input string.
760+-- The result string may be longer than the input string. For
610−-- For instance, the German \"ß\" (eszett, U+00DF) maps to the
761+-- instance, the German \"ß\" (eszett, U+00DF) maps to the
611-- two-letter sequence \"SS\".
762-- two-letter sequence \"SS\".
612toUpper :: Text -> Text
763toUpper :: Text -> Text
613toUpper t = unstream (S.toUpper (stream t))
764toUpper t = unstream (S.toUpper (stream t))
614{-# INLINE toUpper #-}
765{-# 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 #-}
616-- | /O(n)/ Left-justify a string to the given length, using the
789-- | /O(n)/ Left-justify a string to the given length, using the
617−-- 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.
619-- > justifyLeft 7 'x' "foo" == "fooxxxx"
795-- > justifyLeft 7 'x' "foo" == "fooxxxx"
620-- > justifyLeft 3 'x' "foobar" == "foobar"
796-- > justifyLeft 3 'x' "foobar" == "foobar"
@@ -633,7 +809,10 @@ justifyLeft k c t
635-- | /O(n)/ Right-justify a string to the given length, using the
811-- | /O(n)/ Right-justify a string to the given length, using the
636−-- specified fill character on the left. Examples:
812+-- specified fill character on the left. Performs replacement on
813+-- invalid scalar values.
638-- > justifyRight 7 'x' "bar" == "xxxxbar"
817-- > justifyRight 7 'x' "bar" == "xxxxbar"
639-- > justifyRight 3 'x' "foobar" == "foobar"
818-- > justifyRight 3 'x' "foobar" == "foobar"
@@ -644,8 +823,11 @@ justifyRight k c t
644 where len = length t
823 where len = length t
645{-# INLINE justifyRight #-}
824{-# INLINE justifyRight #-}
647−-- | /O(n)/ Center a string to the given length, using the
826+-- | /O(n)/ Center a string to the given length, using the specified
648−-- specified fill character on either side. Examples:
827+-- fill character on either side. Performs replacement on invalid
650-- > center 8 'x' "HS" = "xxxHSxxx"
832-- > center 8 'x' "HS" = "xxxHSxxx"
651center :: Int -> Char -> Text -> Text
833center :: Int -> Char -> Text -> Text
@@ -654,7 +836,7 @@ center k c t
654 | otherwise = replicateChar l c `append` t `append` replicateChar r c
836 | otherwise = replicateChar l c `append` t `append` replicateChar r c
655 where len = length t
837 where len = length t
656 d = k - len
838 d = k - len
657− r = d `div` 2
839+ r = d `quot` 2
658 l = d - r
840 l = d - r
659{-# INLINE center #-}
841{-# INLINE center #-}
@@ -701,7 +883,7 @@ foldr f z t = S.foldr f z (stream t)
701{-# INLINE foldr #-}
883{-# INLINE foldr #-}
703-- | /O(n)/ A variant of 'foldr' that has no starting value argument,
885-- | /O(n)/ A variant of 'foldr' that has no starting value argument,
704−-- 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
705-- fusion.
887-- fusion.
706foldr1 :: (Char -> Char -> Char) -> Text -> Char
888foldr1 :: (Char -> Char -> Char) -> Text -> Char
707foldr1 f t = S.foldr1 f (stream t)
889foldr1 f t = S.foldr1 f (stream t)
@@ -719,6 +901,7 @@ concat ts = case ts' of
720 ts' = L.filter (not . null) ts
902 ts' = L.filter (not . null) ts
721 len = sumP "concat" $ L.map lengthWord16 ts'
903 len = sumP "concat" $ L.map lengthWord16 ts'
904+ go :: ST s (A.MArray s)
723 arr <- A.new len
906 arr <- A.new len
724 let step i (Text a o l) =
907 let step i (Text a o l) =
@@ -732,13 +915,13 @@ concatMap f = concat . foldr ((:) . f) []
732{-# INLINE concatMap #-}
915{-# INLINE concatMap #-}
734-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
917-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the
735−-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.
918+-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.
736any :: (Char -> Bool) -> Text -> Bool
919any :: (Char -> Bool) -> Text -> Bool
737any p t = S.any p (stream t)
920any p t = S.any p (stream t)
738{-# INLINE any #-}
921{-# INLINE any #-}
740-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
923-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the
741−-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.
924+-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.
742all :: (Char -> Bool) -> Text -> Bool
925all :: (Char -> Bool) -> Text -> Bool
743all p t = S.all p (stream t)
926all p t = S.all p (stream t)
744{-# INLINE all #-}
927{-# INLINE all #-}
@@ -760,6 +943,7 @@ minimum t = S.minimum (stream t)
761-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
944-- | /O(n)/ 'scanl' is similar to 'foldl', but returns a list of
762-- successive reduced values from the left. Subject to fusion.
945-- successive reduced values from the left. Subject to fusion.
946+-- Performs replacement on invalid scalar values.
764-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
948-- > scanl f z [x1, x2, ...] == [z, z `f` x1, (z `f` x1) `f` x2, ...]
@@ -767,11 +951,13 @@ minimum t = S.minimum (stream t)
768-- > last (scanl f z xs) == foldl f z xs.
952-- > last (scanl f z xs) == foldl f z xs.
769scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
953scanl :: (Char -> Char -> Char) -> Char -> Text -> Text
770−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)
771{-# INLINE scanl #-}
956{-# INLINE scanl #-}
773-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
958-- | /O(n)/ 'scanl1' is a variant of 'scanl' that has no starting
774−-- value argument. Subject to fusion.
959+-- value argument. Subject to fusion. Performs replacement on
960+-- invalid scalar values.
776-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
962-- > scanl1 f [x1, x2, ...] == [x1, x1 `f` x2, ...]
777scanl1 :: (Char -> Char -> Char) -> Text -> Text
963scanl1 :: (Char -> Char -> Char) -> Text -> Text
@@ -779,15 +965,18 @@ scanl1 f t | null t = empty
779 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
965 | otherwise = scanl f (unsafeHead t) (unsafeTail t)
780{-# INLINE scanl1 #-}
966{-# INLINE scanl1 #-}
782−-- | /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.
784-- > scanr f v == reverse . scanl (flip f) v . reverse
971-- > scanr f v == reverse . scanl (flip f) v . reverse
785scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
972scanr :: (Char -> Char -> Char) -> Char -> Text -> Text
786−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)
787{-# INLINE scanr #-}
975{-# INLINE scanr #-}
789-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
977-- | /O(n)/ 'scanr1' is a variant of 'scanr' that has no starting
790−-- value argument. Subject to fusion.
978+-- value argument. Subject to fusion. Performs replacement on
979+-- invalid scalar values.
791scanr1 :: (Char -> Char -> Char) -> Text -> Text
980scanr1 :: (Char -> Char -> Char) -> Text -> Text
792scanr1 f t | null t = empty
981scanr1 f t | null t = empty
793 | otherwise = scanr f (last t) (init t)
982 | otherwise = scanr f (last t) (init t)
@@ -795,9 +984,11 @@ scanr1 f t | null t = empty
796-- | /O(n)/ Like a combination of 'map' and 'foldl''. Applies a
985-- | /O(n)/ Like a combination of 'map' and 'foldl''. Applies a
797-- function to each element of a 'Text', passing an accumulating
986-- function to each element of a 'Text', passing an accumulating
798−-- 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.
799mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
989mapAccumL :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
800−mapAccumL f z0 = S.mapAccumL f z0 . stream
990+mapAccumL f z0 = S.mapAccumL g z0 . stream
991+ where g a b = second safe (f a b)
801{-# INLINE mapAccumL #-}
992{-# INLINE mapAccumL #-}
803-- | The 'mapAccumR' function behaves like a combination of 'map' and
994-- | The 'mapAccumR' function behaves like a combination of 'map' and
@@ -805,8 +996,10 @@ mapAccumL f z0 = S.mapAccumL f z0 . stream
805-- 'Text', passing an accumulating parameter from right to left, and
996-- 'Text', passing an accumulating parameter from right to left, and
806-- returning a final value of this accumulator together with the new
997-- returning a final value of this accumulator together with the new
807-- 'Text'.
998-- 'Text'.
999+-- Performs replacement on invalid scalar values.
808mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
1000mapAccumR :: (a -> Char -> (a,Char)) -> a -> Text -> (a, Text)
809−mapAccumR f z0 = second reverse . S.mapAccumL f z0 . reverseStream
1001+mapAccumR f z0 = second reverse . S.mapAccumL g z0 . reverseStream
1002+ where g a b = second safe (f a b)
810{-# INLINE mapAccumR #-}
1003{-# INLINE mapAccumR #-}
812-- -----------------------------------------------------------------------------
1005-- -----------------------------------------------------------------------------
@@ -816,13 +1009,13 @@ mapAccumR f z0 = second reverse . S.mapAccumL f z0 . reverseStream
816-- @t@ repeated @n@ times.
1009-- @t@ repeated @n@ times.
817replicate :: Int -> Text -> Text
1010replicate :: Int -> Text -> Text
818replicate n t@(Text a o l)
1011replicate n t@(Text a o l)
819− | n <= 0 || l <= 0 = empty
1012+ | n <= 0 || l <= 0 = empty
820− | n == 1 = t
1013+ | n == 1 = t
821− | isSingleton t = replicateChar n (unsafeHead t)
1014+ | isSingleton t = replicateChar n (unsafeHead t)
822− | n <= maxBound `div` l = Text (A.run x) 0 len
1015+ | otherwise = Text (A.run x) 0 len
823− | otherwise = overflowError "replicate"
825− len = l * n
1017+ len = l `mul` n
1018+ x :: ST s (A.MArray s)
827 arr <- A.new len
1020 arr <- A.new len
828 let loop !d !i | i >= n = return arr
1021 let loop !d !i | i >= n = return arr
@@ -839,7 +1032,7 @@ replicate n t@(Text a o l)
839-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the
1032-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the
840-- value of every element. Subject to fusion.
1033-- value of every element. Subject to fusion.
841replicateChar :: Int -> Char -> Text
1034replicateChar :: Int -> Char -> Text
842−replicateChar n c = unstream (S.replicateCharI n c)
1035+replicateChar n c = unstream (S.replicateCharI n (safe c))
843{-# INLINE replicateChar #-}
1036{-# INLINE replicateChar #-}
845-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
1038-- | /O(n)/, where @n@ is the length of the result. The 'unfoldr'
@@ -848,9 +1041,9 @@ replicateChar n c = unstream (S.replicateCharI n c)
848-- returns 'Nothing' if it is done producing the 'Text', otherwise
1041-- returns 'Nothing' if it is done producing the 'Text', otherwise
849-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
1042-- 'Just' @(a,b)@. In this case, @a@ is the next 'Char' in the
850-- string, and @b@ is the seed value for further production. Subject
1043-- string, and @b@ is the seed value for further production. Subject
851−-- to fusion.
1044+-- to fusion. Performs replacement on invalid scalar values.
852unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
1045unfoldr :: (a -> Maybe (Char,a)) -> a -> Text
853−unfoldr f s = unstream (S.unfoldr f s)
1046+unfoldr f s = unstream (S.unfoldr (firstf safe . f) s)
854{-# INLINE unfoldr #-}
1047{-# INLINE unfoldr #-}
856-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
1049-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a 'Text' from a seed
@@ -858,9 +1051,9 @@ unfoldr f s = unstream (S.unfoldr f s)
858-- first argument to 'unfoldrN'. This function is more efficient than
1051-- first argument to 'unfoldrN'. This function is more efficient than
859-- 'unfoldr' when the maximum length of the result is known and
1052-- 'unfoldr' when the maximum length of the result is known and
860-- correct, otherwise its performance is similar to 'unfoldr'. Subject
1053-- correct, otherwise its performance is similar to 'unfoldr'. Subject
861−-- to fusion.
1054+-- to fusion. Performs replacement on invalid scalar values.
862unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
1055unfoldrN :: Int -> (a -> Maybe (Char,a)) -> a -> Text
863−unfoldrN n f s = unstream (S.unfoldrN n f s)
1056+unfoldrN n f s = unstream (S.unfoldrN n (firstf safe . f) s)
864{-# INLINE unfoldrN #-}
1057{-# INLINE unfoldrN #-}
866-- -----------------------------------------------------------------------------
1059-- -----------------------------------------------------------------------------
@@ -873,14 +1066,16 @@ take :: Int -> Text -> Text
873take n t@(Text arr off len)
1066take n t@(Text arr off len)
874 | n <= 0 = empty
1067 | n <= 0 = empty
875 | n >= len = t
1068 | n >= len = t
876− | otherwise = Text arr off (loop 0 0)
1069+ | otherwise = text arr off (iterN n t)
879− | i >= len || cnt >= n = i
880− | otherwise = loop (i+d) (cnt+1)
882{-# INLINE [1] take #-}
1070{-# INLINE [1] take #-}
1072+iterN :: Int -> Text -> Int
1073+iterN n t@(Text _arr _off len) = loop 0 0
1075+ | i >= len || cnt >= n = i
1076+ | otherwise = loop (i+d) (cnt+1)
1077+ where d = iter_ t i
884{-# RULES
1079{-# RULES
885"TEXT take -> fused" [~1] forall n t.
1080"TEXT take -> fused" [~1] forall n t.
886 take n t = unstream (S.take n (stream t))
1081 take n t = unstream (S.take n (stream t))
@@ -888,6 +1083,27 @@ take n t@(Text arr off len)
888 unstream (S.take n (stream t)) = take n t
1083 unstream (S.take n (stream t)) = take n t
1086+-- | /O(n)/ 'takeEnd' @n@ @t@ returns the suffix remaining after
1087+-- taking @n@ characters from the end of @t@.
1091+-- > takeEnd 3 "foobar" == "bar"
1092+takeEnd :: Int -> Text -> Text
1093+takeEnd n t@(Text arr off len)
1096+ | otherwise = text arr (off+i) (len-i)
1097+ where i = iterNEnd n t
1099+iterNEnd :: Int -> Text -> Int
1100+iterNEnd n t@(Text _arr _off len) = loop (len-1) n
1104+ | otherwise = loop (i+d) (m-1)
1105+ where d = reverseIter_ t i
891-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
1107-- | /O(n)/ 'drop' @n@, applied to a 'Text', returns the suffix of the
892-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
1108-- 'Text' after the first @n@ characters, or the empty 'Text' if @n@
893-- is greater than the length of the 'Text'. Subject to fusion.
1109-- is greater than the length of the 'Text'. Subject to fusion.
@@ -895,11 +1111,8 @@ drop :: Int -> Text -> Text
895drop n t@(Text arr off len)
1111drop n t@(Text arr off len)
896 | n <= 0 = t
1112 | n <= 0 = t
897 | n >= len = empty
1113 | n >= len = empty
898− | otherwise = loop 0 0
1114+ | otherwise = text arr (off+i) (len-i)
900− | i >= len || cnt >= n = Text arr (off+i) (len-i)
901− | otherwise = loop (i+d) (cnt+1)
902− where d = iter_ t i
1115+ where i = iterN n t
903{-# INLINE [1] drop #-}
1116{-# INLINE [1] drop #-}
905{-# RULES
1118{-# RULES
@@ -909,6 +1122,18 @@ drop n t@(Text arr off len)
909 unstream (S.drop n (stream t)) = drop n t
1122 unstream (S.drop n (stream t)) = drop n t
1125+-- | /O(n)/ 'dropEnd' @n@ @t@ returns the prefix remaining after
1126+-- dropping @n@ characters from the end of @t@.
1130+-- > dropEnd 3 "foobar" == "foo"
1131+dropEnd :: Int -> Text -> Text
1132+dropEnd n t@(Text arr off len)
1135+ | otherwise = text arr off (iterNEnd n t)
912-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
1137-- | /O(n)/ 'takeWhile', applied to a predicate @p@ and a 'Text',
913-- returns the longest prefix (possibly empty) of elements that
1138-- returns the longest prefix (possibly empty) of elements that
914-- satisfy @p@. Subject to fusion.
1139-- satisfy @p@. Subject to fusion.
@@ -916,7 +1141,7 @@ takeWhile :: (Char -> Bool) -> Text -> Text
916takeWhile p t@(Text arr off len) = loop 0
1141takeWhile p t@(Text arr off len) = loop 0
917 where loop !i | i >= len = t
1142 where loop !i | i >= len = t
918 | p c = loop (i+d)
1143 | p c = loop (i+d)
919− | otherwise = textP arr off i
1144+ | otherwise = text arr off i
920 where Iter c d = iter t i
1145 where Iter c d = iter t i
921{-# INLINE [1] takeWhile #-}
1146{-# INLINE [1] takeWhile #-}
@@ -927,6 +1152,27 @@ takeWhile p t@(Text arr off len) = loop 0
927 unstream (S.takeWhile p (stream t)) = takeWhile p t
1152 unstream (S.takeWhile p (stream t)) = takeWhile p t
1155+-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',
1156+-- returns the longest suffix (possibly empty) of elements that
1157+-- satisfy @p@. Subject to fusion.
1160+-- > takeWhileEnd (=='o') "foo" == "oo"
1161+takeWhileEnd :: (Char -> Bool) -> Text -> Text
1162+takeWhileEnd p t@(Text arr off len) = loop (len-1) len
1163+ where loop !i !l | l <= 0 = t
1164+ | p c = loop (i+d) (l+d)
1165+ | otherwise = text arr (off+l) (len-l)
1166+ where (c,d) = reverseIter t i
1167+{-# INLINE [1] takeWhileEnd #-}
1170+"TEXT takeWhileEnd -> fused" [~1] forall p t.
1171+ takeWhileEnd p t = S.reverse (S.takeWhile p (S.reverseStream t))
1172+"TEXT takeWhileEnd -> unfused" [1] forall p t.
1173+ S.reverse (S.takeWhile p (S.reverseStream t)) = takeWhileEnd p t
930-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
1176-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after
931-- 'takeWhile' @p@ @t@. Subject to fusion.
1177-- 'takeWhile' @p@ @t@. Subject to fusion.
932dropWhile :: (Char -> Bool) -> Text -> Text
1178dropWhile :: (Char -> Bool) -> Text -> Text
@@ -945,8 +1191,9 @@ dropWhile p t@(Text arr off len) = loop 0 0
947-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
1193-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after
948−-- dropping characters that fail the predicate @p@ from the end of
1194+-- dropping characters that satisfy the predicate @p@ from the end of
949-- @t@. Subject to fusion.
1195-- @t@. Subject to fusion.
950-- Examples:
1197-- Examples:
952-- > dropWhileEnd (=='.') "foo..." == "foo"
1199-- > dropWhileEnd (=='.') "foo..." == "foo"
@@ -966,7 +1213,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len
968-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
1215-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after
969−-- dropping characters that fail the predicate @p@ from both the
1216+-- dropping characters that satisfy the predicate @p@ from both the
970-- beginning and end of @t@. Subject to fusion.
1217-- beginning and end of @t@. Subject to fusion.
971dropAround :: (Char -> Bool) -> Text -> Text
1218dropAround :: (Char -> Bool) -> Text -> Text
972dropAround p = dropWhile p . dropWhileEnd p
1219dropAround p = dropWhile p . dropWhileEnd p
@@ -1001,24 +1248,16 @@ splitAt :: Int -> Text -> (Text, Text)
1001splitAt n t@(Text arr off len)
1248splitAt n t@(Text arr off len)
1002 | n <= 0 = (empty, t)
1249 | n <= 0 = (empty, t)
1003 | n >= len = (t, empty)
1250 | n >= len = (t, empty)
1004− | otherwise = (Text arr off k, Text arr (off+k) (len-k))
1251+ | otherwise = let k = iterN n t
1005− where k = loop 0 0
1252+ in (text arr off k, text arr (off+k) (len-k))
1007− | i >= len || cnt >= n = i
1008− | otherwise = loop (i+d) (cnt+1)
1009− where d = iter_ t i
1010−{-# INLINE splitAt #-}
1012-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
1254-- | /O(n)/ 'span', applied to a predicate @p@ and text @t@, returns
1013-- a pair whose first element is the longest prefix (possibly empty)
1255-- a pair whose first element is the longest prefix (possibly empty)
1014-- of @t@ of elements that satisfy @p@, and whose second is the
1256-- of @t@ of elements that satisfy @p@, and whose second is the
1015-- remainder of the list.
1257-- remainder of the list.
1016span :: (Char -> Bool) -> Text -> (Text, Text)
1258span :: (Char -> Bool) -> Text -> (Text, Text)
1017−span p t@(Text arr off len) = (textP arr off k, textP arr (off+k) (len-k))
1259+span p t = case span_ p t of
1018− where k = loop 0
1260+ (# hd,tl #) -> (hd,tl)
1019− loop !i | i >= len || not (p c) = i
1020− | otherwise = loop (i+d)
1021− where Iter c d = iter t i
1022{-# INLINE span #-}
1261{-# INLINE span #-}
1024-- | /O(n)/ 'break' is like 'span', but the prefix returned is
1263-- | /O(n)/ 'break' is like 'span', but the prefix returned is
@@ -1045,7 +1284,7 @@ findAIndexOrEnd q t@(Text _arr _off len) = go 0
1045 where go !i | i >= len || q c = i
1284 where go !i | i >= len || q c = i
1046 | otherwise = go (i+d)
1285 | otherwise = go (i+d)
1047 where Iter c d = iter t i
1286 where Iter c d = iter t i
1049-- | /O(n)/ Group characters in a string by equality.
1288-- | /O(n)/ Group characters in a string by equality.
1050group :: Text -> [Text]
1289group :: Text -> [Text]
1051group = groupBy (==)
1290group = groupBy (==)
@@ -1069,31 +1308,38 @@ tails t | null t = [empty]
1069-- copies to create substrings; they just construct new 'Text's that
1308-- copies to create substrings; they just construct new 'Text's that
1070-- are slices of the original.
1309-- are slices of the original.
1072−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first
1311+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'
1073−-- 'Text' argument, consuming the delimiter. An empty delimiter is
1312+-- argument (which cannot be empty), consuming the delimiter. An empty
1074−-- invalid, and will cause an error to be raised.
1313+-- delimiter is invalid, and will cause an error to be raised.
1076-- Examples:
1315-- Examples:
1078-- > splitOn "\r\n" "a\r\nb\r\nd\r\ne" == ["a","b","d","e"]
1317-- > splitOn "\r\n" "a\r\nb\r\nd\r\ne" == ["a","b","d","e"]
1079-- > splitOn "aaa" "aaaXaaaXaaaXaaa" == ["","X","X","X",""]
1318-- > splitOn "aaa" "aaaXaaaXaaaXaaa" == ["","X","X","X",""]
1080-- > splitOn "x" "x" == ["",""]
1319-- > splitOn "x" "x" == ["",""]
1084-- > intercalate s . splitOn s == id
1323-- > intercalate s . splitOn s == id
1085-- > splitOn (singleton c) == split (==c)
1324-- > splitOn (singleton c) == split (==c)
1326+-- (Note: the string @s@ to split on above cannot be empty.)
1087-- In (unlikely) bad cases, this function's time complexity degrades
1328-- In (unlikely) bad cases, this function's time complexity degrades
1088-- towards /O(n*m)/.
1329-- towards /O(n*m)/.
1089−splitOn :: Text -> Text -> [Text]
1330+splitOn :: Text
1331+ -- ^ String to split on. If this string is empty, an error
1090splitOn pat@(Text _ _ l) src@(Text arr off len)
1336splitOn pat@(Text _ _ l) src@(Text arr off len)
1091 | l <= 0 = emptyError "splitOn"
1337 | l <= 0 = emptyError "splitOn"
1092 | isSingleton pat = split (== unsafeHead pat) src
1338 | isSingleton pat = split (== unsafeHead pat) src
1093 | otherwise = go 0 (indices pat src)
1339 | otherwise = go 0 (indices pat src)
1095− go !s (x:xs) = textP arr (s+off) (x-s) : go (x+l) xs
1341+ go !s (x:xs) = text arr (s+off) (x-s) : go (x+l) xs
1096− go s _ = [textP arr (s+off) (len-s)]
1342+ go s _ = [text arr (s+off) (len-s)]
1097{-# INLINE [1] splitOn #-}
1343{-# INLINE [1] splitOn #-}
1099{-# RULES
1345{-# RULES
@@ -1113,7 +1359,7 @@ split _ t@(Text _off _arr 0) = [t]
1113split p t = loop t
1359split p t = loop t
1114 where loop s | null s' = [l]
1360 where loop s | null s' = [l]
1115 | otherwise = l : loop (unsafeTail s')
1361 | otherwise = l : loop (unsafeTail s')
1116− where (l, s') = break p s
1362+ where (# l, s' #) = span_ (not . p) s
1117{-# INLINE split #-}
1363{-# INLINE split #-}
1119-- | /O(n)/ Splits a 'Text' into components of length @k@. The last
1365-- | /O(n)/ Splits a 'Text' into components of length @k@. The last
@@ -1185,7 +1431,7 @@ breakOn pat src@(Text arr off len)
1185 | null pat = emptyError "breakOn"
1431 | null pat = emptyError "breakOn"
1186 | otherwise = case indices pat src of
1432 | otherwise = case indices pat src of
1187 [] -> (src, empty)
1433 [] -> (src, empty)
1188− (x:_) -> (textP arr off x, textP arr (off+x) (len-x))
1434+ (x:_) -> (text arr off x, text arr (off+x) (len-x))
1189{-# INLINE breakOn #-}
1435{-# INLINE breakOn #-}
1191-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
1437-- | /O(n+m)/ Similar to 'breakOn', but searches from the end of the
@@ -1210,9 +1456,9 @@ breakOnEnd pat src = (reverse b, reverse a)
1211-- Examples:
1457-- Examples:
1213−-- > find "::" ""
1459+-- > breakOnAll "::" ""
1214-- > ==> []
1460-- > ==> []
1215−-- > find "/" "a/b/c/"
1461+-- > breakOnAll "/" "a/b/c/"
1216-- > ==> [("a", "/b/c/"), ("a/b", "/c/"), ("a/b/c", "/")]
1462-- > ==> [("a", "/b/c/"), ("a/b", "/c/"), ("a/b/c", "/")]
1218-- In (unlikely) bad cases, this function's time complexity degrades
1464-- In (unlikely) bad cases, this function's time complexity degrades
@@ -1227,7 +1473,7 @@ breakOnAll pat src@(Text arr off slen)
1227 | otherwise = L.map step (indices pat src)
1473 | otherwise = L.map step (indices pat src)
1229 step x = (chunk 0 x, chunk x (slen-x))
1475 step x = (chunk 0 x, chunk x (slen-x))
1230− chunk !n !l = textP arr (n+off) l
1476+ chunk !n !l = text arr (n+off) l
1231{-# INLINE breakOnAll #-}
1477{-# INLINE breakOnAll #-}
1233-------------------------------------------------------------------------------
1479-------------------------------------------------------------------------------
@@ -1247,7 +1493,7 @@ breakOnAll pat src@(Text arr off slen)
1247-- For example, suppose you have a string that you want to split on
1493-- For example, suppose you have a string that you want to split on
1248-- the substring @\"::\"@, such as @\"foo::bar::quux\"@. Instead of
1494-- the substring @\"::\"@, such as @\"foo::bar::quux\"@. Instead of
1249-- searching for the index of @\"::\"@ and taking the substrings
1495-- searching for the index of @\"::\"@ and taking the substrings
1250−-- before and after that index, you would instead use @find \"::\"@.
1496+-- before and after that index, you would instead use @breakOnAll \"::\"@.
1252-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
1498-- | /O(n)/ 'Text' index (subscript) operator, starting from 0.
1253index :: Text -> Int -> Char
1499index :: Text -> Int -> Char
@@ -1294,13 +1540,15 @@ countChar c t = S.countChar c (stream t)
1294-- equivalent to a pair of 'unpack' operations.
1540-- equivalent to a pair of 'unpack' operations.
1295zip :: Text -> Text -> [(Char,Char)]
1541zip :: Text -> Text -> [(Char,Char)]
1296zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1542zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)
1297−{-# INLINE [0] zip #-}
1543+{-# INLINE zip #-}
1299-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
1545-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function
1300-- given as the first argument, instead of a tupling function.
1546-- given as the first argument, instead of a tupling function.
1547+-- Performs replacement on invalid scalar values.
1301zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1548zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text
1302−zipWith f t1 t2 = unstream (S.zipWith f (stream t1) (stream t2))
1549+zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))
1550+ where g a b = safe (f a b)
1303−{-# INLINE [0] zipWith #-}
1551+{-# INLINE zipWith #-}
1305-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
1553-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's
1306-- representing white space.
1554-- representing white space.
@@ -1326,7 +1574,7 @@ lines ps | null ps = []
1326 | otherwise = h : if null t
1574 | otherwise = h : if null t
1328 else lines (unsafeTail t)
1576 else lines (unsafeTail t)
1329− where (h,t) = span (/= '\n') ps
1577+ where (# h,t #) = span_ (/= '\n') ps
1330{-# INLINE lines #-}
1578{-# INLINE lines #-}
@@ -1363,8 +1611,7 @@ unwords = intercalate (singleton ' ')
1363{-# INLINE unwords #-}
1611{-# INLINE unwords #-}
1365-- | /O(n)/ The 'isPrefixOf' function takes two 'Text's and returns
1613-- | /O(n)/ The 'isPrefixOf' function takes two 'Text's and returns
1366−-- 'True' iff the first is a prefix of the second. This function is
1614+-- 'True' iff the first is a prefix of the second. Subject to fusion.
1367−-- subject to fusion.
1368isPrefixOf :: Text -> Text -> Bool
1615isPrefixOf :: Text -> Text -> Bool
1369isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
1616isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
1370 alen <= blen && S.isPrefixOf (stream a) (stream b)
1617 alen <= blen && S.isPrefixOf (stream a) (stream b)
@@ -1373,8 +1620,6 @@ isPrefixOf a@(Text _ _ alen) b@(Text _ _ blen) =
1373{-# RULES
1620{-# RULES
1374"TEXT isPrefixOf -> fused" [~1] forall s t.
1621"TEXT isPrefixOf -> fused" [~1] forall s t.
1375 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1622 isPrefixOf s t = S.isPrefixOf (stream s) (stream t)
1376−"TEXT isPrefixOf -> unfused" [1] forall s t.
1377− S.isPrefixOf (stream s) (stream t) = isPrefixOf s t
1380-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
1625-- | /O(n)/ The 'isSuffixOf' function takes two 'Text's and returns
@@ -1408,12 +1653,13 @@ isInfixOf needle haystack
1408-------------------------------------------------------------------------------
1653-------------------------------------------------------------------------------
1409-- * View patterns
1654-- * View patterns
1411−-- | /O(n)/ Returns the suffix of the second string if its prefix
1656+-- | /O(n)/ Return the suffix of the second string if its prefix
1412−-- matches the first.
1657+-- matches the entire first string.
1414-- Examples:
1659-- Examples:
1416-- > stripPrefix "foo" "foobar" == Just "bar"
1661-- > stripPrefix "foo" "foobar" == Just "bar"
1662+-- > stripPrefix "" "baz" == Just "baz"
1417-- > stripPrefix "foo" "quux" == Nothing
1663-- > stripPrefix "foo" "quux" == Nothing
1419-- This is particularly useful with the @ViewPatterns@ extension to
1665-- This is particularly useful with the @ViewPatterns@ extension to
@@ -1424,18 +1670,42 @@ isInfixOf needle haystack
1425-- > fnordLength :: Text -> Int
1671-- > fnordLength :: Text -> Int
1426-- > fnordLength (stripPrefix "fnord" -> Just suf) = T.length suf
1672-- > fnordLength (stripPrefix "fnord" -> Just suf) = T.length suf
1427−-- > fnordLength _ = -1
1673+-- > fnordLength _ = -1
1428stripPrefix :: Text -> Text -> Maybe Text
1674stripPrefix :: Text -> Text -> Maybe Text
1429stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1675stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1430− | p `isPrefixOf` t = Just $! textP arr (off+plen) (len-plen)
1676+ | p `isPrefixOf` t = Just $! text arr (off+plen) (len-plen)
1431 | otherwise = Nothing
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
1433−-- | /O(n)/ Returns the prefix of the second string if its suffix
1702+-- | /O(n)/ Return the prefix of the second string if its suffix
1434−-- matches the first.
1703+-- matches the entire first string.
1436-- Examples:
1705-- Examples:
1438-- > stripSuffix "bar" "foobar" == Just "foo"
1707-- > stripSuffix "bar" "foobar" == Just "foo"
1708+-- > stripSuffix "" "baz" == Just "baz"
1439-- > stripSuffix "foo" "quux" == Nothing
1709-- > stripSuffix "foo" "quux" == Nothing
1441-- This is particularly useful with the @ViewPatterns@ extension to
1711-- This is particularly useful with the @ViewPatterns@ extension to
@@ -1446,10 +1716,10 @@ stripPrefix p@(Text _arr _off plen) t@(Text arr off len)
1447-- > quuxLength :: Text -> Int
1717-- > quuxLength :: Text -> Int
1448-- > quuxLength (stripSuffix "quux" -> Just pre) = T.length pre
1718-- > quuxLength (stripSuffix "quux" -> Just pre) = T.length pre
1449−-- > quuxLength _ = -1
1719+-- > quuxLength _ = -1
1450stripSuffix :: Text -> Text -> Maybe Text
1720stripSuffix :: Text -> Text -> Maybe Text
1451stripSuffix p@(Text _arr _off plen) t@(Text arr off len)
1721stripSuffix p@(Text _arr _off plen) t@(Text arr off len)
1452− | p `isSuffixOf` t = Just $! textP arr off (len-plen)
1722+ | p `isSuffixOf` t = Just $! text arr off (len-plen)
1453 | otherwise = Nothing
1723 | otherwise = Nothing
1455-- | Add a list of non-negative numbers. Errors out on overflow.
1725-- | Add a list of non-negative numbers. Errors out on overflow.
@@ -1466,3 +1736,20 @@ emptyError fun = P.error $ "Data.Text." ++ fun ++ ": empty input"
1467overflowError :: String -> a
1737overflowError :: String -> a
1468overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"
1738overflowError 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