master...04dcb4a · text

git.iko.soy/text

13 files changed+321−58

since splitsnapshots

unifiedsplit

.hgtags+1 Data/Text.hs+13−1 Data/Text/Foreign.hs+1−1 Data/Text/Internal/Fusion/CaseMapping.hs+164−4 Data/Text/Internal/Fusion/Common.hs+10−9 Data/Text/Lazy.hs+15−6 Data/Text/Lazy/Read.hs+3−3 Data/Text/Read.hs+3−3 changelog.md+29−1 tests/Tests/Properties.hs+58−21 tests/Tests/QuickCheckUtils.hs+1−1 tests/text-tests.cabal+18−3 text.cabal+5−5

.hgtags+1

@@ -64,3 +64,4 @@ f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

6464631fd055e27fa10983e17e9ad36d1b285b85c661 1.2.1.3

65658221a68ed5ec37028357df9c6c25f52320e54ad9 1.2.2.0

6666ce1c5ac65eb9f5580114b4e380092b05a485e95d 1.2.2.1

67+d72c634a91ffe27ec5cd03921dd53efc8129d8f6 1.2.2.2

Data/Text.hs+13−1

@@ -60,6 +60,7 @@ module Data.Text

6060 , snoc

6161 , append

6262 , uncons

63+ , unsnoc

6364 , head

6465 , last

6566 , tail

@@ -545,6 +546,17 @@ init (Text arr off len) | len <= 0 = emptyError "init"

545546 unstream (S.init (stream t)) = init t

546547 #-}

547548

549+-- | /O(1)/ Returns all but the last character and the last character of a

550+-- 'Text', or 'Nothing' if empty.

551+unsnoc :: Text -> Maybe (Text, Char)

552+unsnoc (Text arr off len)

553+ | len <= 0 = Nothing

554+ | n < 0xDC00 || n > 0xDFFF = Just (text arr off (len-1), unsafeChr n)

555+ | otherwise = Just (text arr off (len-2), U16.chr2 n0 n)

556+ where n = A.unsafeIndex arr (off+len-1)

557+ n0 = A.unsafeIndex arr (off+len-2)

558+{-# INLINE [1] unsnoc #-}

559+

548560-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to

549561-- fusion.

550562null :: Text -> Bool

@@ -1099,8 +1111,8 @@ takeEnd n t@(Text arr off len)

10991111iterNEnd :: Int -> Text -> Int

11001112iterNEnd n t@(Text _arr _off len) = loop (len-1) n

11011113 where loop i !m

1114+ | m <= 0 = i+1

11021115 | i <= 0 = 0

1103− | m <= 1 = i

11041116 | otherwise = loop (i+d) (m-1)

11051117 where d = reverseIter_ t i

11061118

Data/Text/Foreign.hs+1−1

@@ -108,7 +108,7 @@ takeWord16 (I16 n) t@(Text arr off len)

108108 | n >= len || m >= len = t

109109 | otherwise = Text arr off m

110110 where

111− m | w < 0xDB00 || w > 0xD8FF = n

111+ m | w < 0xD800 || w > 0xDBFF = n

112112 | otherwise = n+1

113113 w = A.unsafeIndex arr (off+n-1)

114114

Data/Text/Internal/Fusion/CaseMapping.hs+164−4

@@ -1,10 +1,10 @@

11{-# LANGUAGE Rank2Types #-}

22-- AUTOMATICALLY GENERATED - DO NOT EDIT

33-- Generated by scripts/CaseMapping.hs

4−-- CaseFolding-8.0.0.txt

5−-- Date: 2015-01-13, 18:16:36 GMT [MD]

6−-- SpecialCasing-8.0.0.txt

7−-- Date: 2014-12-16, 23:08:04 GMT [MD]

4+-- CaseFolding-9.0.0.txt

5+-- Date: 2016-03-02, 18:54:54 GMT

6+-- SpecialCasing-9.0.0.txt

7+-- Date: 2016-03-02, 18:55:13 GMT

99module Data.Text.Internal.Fusion.CaseMapping where

1010import Data.Char

@@ -371,6 +371,24 @@ foldMapping '\x13fb' s = Yield '\x13f3' (CC s '\x0000' '\x0000')

371371foldMapping '\x13fc' s = Yield '\x13f4' (CC s '\x0000' '\x0000')

372372-- CHEROKEE SMALL LETTER MV

373373foldMapping '\x13fd' s = Yield '\x13f5' (CC s '\x0000' '\x0000')

374+-- CYRILLIC SMALL LETTER ROUNDED VE

375+foldMapping '\x1c80' s = Yield '\x0432' (CC s '\x0000' '\x0000')

376+-- CYRILLIC SMALL LETTER LONG-LEGGED DE

377+foldMapping '\x1c81' s = Yield '\x0434' (CC s '\x0000' '\x0000')

378+-- CYRILLIC SMALL LETTER NARROW O

379+foldMapping '\x1c82' s = Yield '\x043e' (CC s '\x0000' '\x0000')

380+-- CYRILLIC SMALL LETTER WIDE ES

381+foldMapping '\x1c83' s = Yield '\x0441' (CC s '\x0000' '\x0000')

382+-- CYRILLIC SMALL LETTER TALL TE

383+foldMapping '\x1c84' s = Yield '\x0442' (CC s '\x0000' '\x0000')

384+-- CYRILLIC SMALL LETTER THREE-LEGGED TE

385+foldMapping '\x1c85' s = Yield '\x0442' (CC s '\x0000' '\x0000')

386+-- CYRILLIC SMALL LETTER TALL HARD SIGN

387+foldMapping '\x1c86' s = Yield '\x044a' (CC s '\x0000' '\x0000')

388+-- CYRILLIC SMALL LETTER TALL YAT

389+foldMapping '\x1c87' s = Yield '\x0463' (CC s '\x0000' '\x0000')

390+-- CYRILLIC SMALL LETTER UNBLENDED UK

391+foldMapping '\x1c88' s = Yield '\xa64b' (CC s '\x0000' '\x0000')

374392-- LATIN SMALL LETTER H WITH LINE BELOW

375393foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

376394-- LATIN SMALL LETTER T WITH DIAERESIS

@@ -545,6 +563,8 @@ foldMapping '\x1ff6' s = Yield '\x03c9' (CC s '\x0342' '\x0000')

545563foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

546564-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

547565foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

566+-- LATIN CAPITAL LETTER SMALL CAPITAL I

567+foldMapping '\xa7ae' s = Yield '\x026a' (CC s '\x0000' '\x0000')

548568-- LATIN CAPITAL LETTER J WITH CROSSED-TAIL

549569foldMapping '\xa7b2' s = Yield '\x029d' (CC s '\x0000' '\x0000')

550570-- LATIN CAPITAL LETTER CHI

@@ -737,6 +757,78 @@ foldMapping '\xfb15' s = Yield '\x0574' (CC s '\x056b' '\x0000')

737757foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

738758-- ARMENIAN SMALL LIGATURE MEN XEH

739759foldMapping '\xfb17' s = Yield '\x0574' (CC s '\x056d' '\x0000')

760+-- OSAGE CAPITAL LETTER A

761+foldMapping '\x104b0' s = Yield '\x104d8' (CC s '\x0000' '\x0000')

762+-- OSAGE CAPITAL LETTER AI

763+foldMapping '\x104b1' s = Yield '\x104d9' (CC s '\x0000' '\x0000')

764+-- OSAGE CAPITAL LETTER AIN

765+foldMapping '\x104b2' s = Yield '\x104da' (CC s '\x0000' '\x0000')

766+-- OSAGE CAPITAL LETTER AH

767+foldMapping '\x104b3' s = Yield '\x104db' (CC s '\x0000' '\x0000')

768+-- OSAGE CAPITAL LETTER BRA

769+foldMapping '\x104b4' s = Yield '\x104dc' (CC s '\x0000' '\x0000')

770+-- OSAGE CAPITAL LETTER CHA

771+foldMapping '\x104b5' s = Yield '\x104dd' (CC s '\x0000' '\x0000')

772+-- OSAGE CAPITAL LETTER EHCHA

773+foldMapping '\x104b6' s = Yield '\x104de' (CC s '\x0000' '\x0000')

774+-- OSAGE CAPITAL LETTER E

775+foldMapping '\x104b7' s = Yield '\x104df' (CC s '\x0000' '\x0000')

776+-- OSAGE CAPITAL LETTER EIN

777+foldMapping '\x104b8' s = Yield '\x104e0' (CC s '\x0000' '\x0000')

778+-- OSAGE CAPITAL LETTER HA

779+foldMapping '\x104b9' s = Yield '\x104e1' (CC s '\x0000' '\x0000')

780+-- OSAGE CAPITAL LETTER HYA

781+foldMapping '\x104ba' s = Yield '\x104e2' (CC s '\x0000' '\x0000')

782+-- OSAGE CAPITAL LETTER I

783+foldMapping '\x104bb' s = Yield '\x104e3' (CC s '\x0000' '\x0000')

784+-- OSAGE CAPITAL LETTER KA

785+foldMapping '\x104bc' s = Yield '\x104e4' (CC s '\x0000' '\x0000')

786+-- OSAGE CAPITAL LETTER EHKA

787+foldMapping '\x104bd' s = Yield '\x104e5' (CC s '\x0000' '\x0000')

788+-- OSAGE CAPITAL LETTER KYA

789+foldMapping '\x104be' s = Yield '\x104e6' (CC s '\x0000' '\x0000')

790+-- OSAGE CAPITAL LETTER LA

791+foldMapping '\x104bf' s = Yield '\x104e7' (CC s '\x0000' '\x0000')

792+-- OSAGE CAPITAL LETTER MA

793+foldMapping '\x104c0' s = Yield '\x104e8' (CC s '\x0000' '\x0000')

794+-- OSAGE CAPITAL LETTER NA

795+foldMapping '\x104c1' s = Yield '\x104e9' (CC s '\x0000' '\x0000')

796+-- OSAGE CAPITAL LETTER O

797+foldMapping '\x104c2' s = Yield '\x104ea' (CC s '\x0000' '\x0000')

798+-- OSAGE CAPITAL LETTER OIN

799+foldMapping '\x104c3' s = Yield '\x104eb' (CC s '\x0000' '\x0000')

800+-- OSAGE CAPITAL LETTER PA

801+foldMapping '\x104c4' s = Yield '\x104ec' (CC s '\x0000' '\x0000')

802+-- OSAGE CAPITAL LETTER EHPA

803+foldMapping '\x104c5' s = Yield '\x104ed' (CC s '\x0000' '\x0000')

804+-- OSAGE CAPITAL LETTER SA

805+foldMapping '\x104c6' s = Yield '\x104ee' (CC s '\x0000' '\x0000')

806+-- OSAGE CAPITAL LETTER SHA

807+foldMapping '\x104c7' s = Yield '\x104ef' (CC s '\x0000' '\x0000')

808+-- OSAGE CAPITAL LETTER TA

809+foldMapping '\x104c8' s = Yield '\x104f0' (CC s '\x0000' '\x0000')

810+-- OSAGE CAPITAL LETTER EHTA

811+foldMapping '\x104c9' s = Yield '\x104f1' (CC s '\x0000' '\x0000')

812+-- OSAGE CAPITAL LETTER TSA

813+foldMapping '\x104ca' s = Yield '\x104f2' (CC s '\x0000' '\x0000')

814+-- OSAGE CAPITAL LETTER EHTSA

815+foldMapping '\x104cb' s = Yield '\x104f3' (CC s '\x0000' '\x0000')

816+-- OSAGE CAPITAL LETTER TSHA

817+foldMapping '\x104cc' s = Yield '\x104f4' (CC s '\x0000' '\x0000')

818+-- OSAGE CAPITAL LETTER DHA

819+foldMapping '\x104cd' s = Yield '\x104f5' (CC s '\x0000' '\x0000')

820+-- OSAGE CAPITAL LETTER U

821+foldMapping '\x104ce' s = Yield '\x104f6' (CC s '\x0000' '\x0000')

822+-- OSAGE CAPITAL LETTER WA

823+foldMapping '\x104cf' s = Yield '\x104f7' (CC s '\x0000' '\x0000')

824+-- OSAGE CAPITAL LETTER KHA

825+foldMapping '\x104d0' s = Yield '\x104f8' (CC s '\x0000' '\x0000')

826+-- OSAGE CAPITAL LETTER GHA

827+foldMapping '\x104d1' s = Yield '\x104f9' (CC s '\x0000' '\x0000')

828+-- OSAGE CAPITAL LETTER ZA

829+foldMapping '\x104d2' s = Yield '\x104fa' (CC s '\x0000' '\x0000')

830+-- OSAGE CAPITAL LETTER ZHA

831+foldMapping '\x104d3' s = Yield '\x104fb' (CC s '\x0000' '\x0000')

740832-- OLD HUNGARIAN CAPITAL LETTER A

741833foldMapping '\x10c80' s = Yield '\x10cc0' (CC s '\x0000' '\x0000')

742834-- OLD HUNGARIAN CAPITAL LETTER AA

@@ -839,4 +931,72 @@ foldMapping '\x10cb0' s = Yield '\x10cf0' (CC s '\x0000' '\x0000')

839931foldMapping '\x10cb1' s = Yield '\x10cf1' (CC s '\x0000' '\x0000')

840932-- OLD HUNGARIAN CAPITAL LETTER US

841933foldMapping '\x10cb2' s = Yield '\x10cf2' (CC s '\x0000' '\x0000')

934+-- ADLAM CAPITAL LETTER ALIF

935+foldMapping '\x1e900' s = Yield '\x1e922' (CC s '\x0000' '\x0000')

936+-- ADLAM CAPITAL LETTER DAALI

937+foldMapping '\x1e901' s = Yield '\x1e923' (CC s '\x0000' '\x0000')

938+-- ADLAM CAPITAL LETTER LAAM

939+foldMapping '\x1e902' s = Yield '\x1e924' (CC s '\x0000' '\x0000')

940+-- ADLAM CAPITAL LETTER MIIM

941+foldMapping '\x1e903' s = Yield '\x1e925' (CC s '\x0000' '\x0000')

942+-- ADLAM CAPITAL LETTER BA

943+foldMapping '\x1e904' s = Yield '\x1e926' (CC s '\x0000' '\x0000')

944+-- ADLAM CAPITAL LETTER SINNYIIYHE

945+foldMapping '\x1e905' s = Yield '\x1e927' (CC s '\x0000' '\x0000')

946+-- ADLAM CAPITAL LETTER PE

947+foldMapping '\x1e906' s = Yield '\x1e928' (CC s '\x0000' '\x0000')

948+-- ADLAM CAPITAL LETTER BHE

949+foldMapping '\x1e907' s = Yield '\x1e929' (CC s '\x0000' '\x0000')

950+-- ADLAM CAPITAL LETTER RA

951+foldMapping '\x1e908' s = Yield '\x1e92a' (CC s '\x0000' '\x0000')

952+-- ADLAM CAPITAL LETTER E

953+foldMapping '\x1e909' s = Yield '\x1e92b' (CC s '\x0000' '\x0000')

954+-- ADLAM CAPITAL LETTER FA

955+foldMapping '\x1e90a' s = Yield '\x1e92c' (CC s '\x0000' '\x0000')

956+-- ADLAM CAPITAL LETTER I

957+foldMapping '\x1e90b' s = Yield '\x1e92d' (CC s '\x0000' '\x0000')

958+-- ADLAM CAPITAL LETTER O

959+foldMapping '\x1e90c' s = Yield '\x1e92e' (CC s '\x0000' '\x0000')

960+-- ADLAM CAPITAL LETTER DHA

961+foldMapping '\x1e90d' s = Yield '\x1e92f' (CC s '\x0000' '\x0000')

962+-- ADLAM CAPITAL LETTER YHE

963+foldMapping '\x1e90e' s = Yield '\x1e930' (CC s '\x0000' '\x0000')

964+-- ADLAM CAPITAL LETTER WAW

965+foldMapping '\x1e90f' s = Yield '\x1e931' (CC s '\x0000' '\x0000')

966+-- ADLAM CAPITAL LETTER NUN

967+foldMapping '\x1e910' s = Yield '\x1e932' (CC s '\x0000' '\x0000')

968+-- ADLAM CAPITAL LETTER KAF

969+foldMapping '\x1e911' s = Yield '\x1e933' (CC s '\x0000' '\x0000')

970+-- ADLAM CAPITAL LETTER YA

971+foldMapping '\x1e912' s = Yield '\x1e934' (CC s '\x0000' '\x0000')

972+-- ADLAM CAPITAL LETTER U

973+foldMapping '\x1e913' s = Yield '\x1e935' (CC s '\x0000' '\x0000')

974+-- ADLAM CAPITAL LETTER JIIM

975+foldMapping '\x1e914' s = Yield '\x1e936' (CC s '\x0000' '\x0000')

976+-- ADLAM CAPITAL LETTER CHI

977+foldMapping '\x1e915' s = Yield '\x1e937' (CC s '\x0000' '\x0000')

978+-- ADLAM CAPITAL LETTER HA

979+foldMapping '\x1e916' s = Yield '\x1e938' (CC s '\x0000' '\x0000')

980+-- ADLAM CAPITAL LETTER QAAF

981+foldMapping '\x1e917' s = Yield '\x1e939' (CC s '\x0000' '\x0000')

982+-- ADLAM CAPITAL LETTER GA

983+foldMapping '\x1e918' s = Yield '\x1e93a' (CC s '\x0000' '\x0000')

984+-- ADLAM CAPITAL LETTER NYA

985+foldMapping '\x1e919' s = Yield '\x1e93b' (CC s '\x0000' '\x0000')

986+-- ADLAM CAPITAL LETTER TU

987+foldMapping '\x1e91a' s = Yield '\x1e93c' (CC s '\x0000' '\x0000')

988+-- ADLAM CAPITAL LETTER NHA

989+foldMapping '\x1e91b' s = Yield '\x1e93d' (CC s '\x0000' '\x0000')

990+-- ADLAM CAPITAL LETTER VA

991+foldMapping '\x1e91c' s = Yield '\x1e93e' (CC s '\x0000' '\x0000')

992+-- ADLAM CAPITAL LETTER KHA

993+foldMapping '\x1e91d' s = Yield '\x1e93f' (CC s '\x0000' '\x0000')

994+-- ADLAM CAPITAL LETTER GBE

995+foldMapping '\x1e91e' s = Yield '\x1e940' (CC s '\x0000' '\x0000')

996+-- ADLAM CAPITAL LETTER ZAL

997+foldMapping '\x1e91f' s = Yield '\x1e941' (CC s '\x0000' '\x0000')

998+-- ADLAM CAPITAL LETTER KPO

999+foldMapping '\x1e920' s = Yield '\x1e942' (CC s '\x0000' '\x0000')

1000+-- ADLAM CAPITAL LETTER SHA

1001+foldMapping '\x1e921' s = Yield '\x1e943' (CC s '\x0000' '\x0000')

8421002foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

Data/Text/Internal/Fusion/Common.hs+10−9

@@ -107,7 +107,7 @@ import Prelude (Bool(..), Char, Eq(..), Int, Integral, Maybe(..),

107107import qualified Data.List as L

108108import qualified Prelude as P

109109import Data.Bits (shiftL)

110−import Data.Char (isLetter)

110+import Data.Char (isLetter, isSpace)

111111import Data.Int (Int64)

112112import Data.Text.Internal.Fusion.Types

113113import Data.Text.Internal.Fusion.CaseMapping (foldMapping, lowerMapping, titleMapping,

@@ -462,15 +462,16 @@ toTitle (Stream next0 s0 len) = Stream next (CC (False :*: s0) '\0' '\0') len

462462 where

463463 next (CC (letter :*: s) '\0' _) =

464464 case next0 s of

465− Done -> Done

466− Skip s' -> Skip (CC (letter :*: s') '\0' '\0')

465+ Done -> Done

466+ Skip s' -> Skip (CC (letter :*: s') '\0' '\0')

467467 Yield c s'

468− | letter' -> if letter

469− then lowerMapping c (letter' :*: s')

470− else titleMapping c (letter' :*: s')

471− | otherwise -> Yield c (CC (letter' :*: s') '\0' '\0')

472− where letter' = isLetter c

473− next (CC s a b) = Yield a (CC s b '\0')

468+ | nonSpace -> if letter

469+ then lowerMapping c (nonSpace :*: s')

470+ else titleMapping c (letter' :*: s')

471+ | otherwise -> Yield c (CC (letter' :*: s') '\0' '\0')

472+ where nonSpace = P.not (isSpace c)

473+ letter' = isLetter c

474+ next (CC s a b) = Yield a (CC s b '\0')

474475{-# INLINE [0] toTitle #-}

475476

476477data Justify i s = Just1 !i !s

Data/Text/Lazy.hs+15−6

@@ -68,6 +68,7 @@ module Data.Text.Lazy

6868 , snoc

6969 , append

7070 , uncons

71+ , unsnoc

7172 , head

7273 , last

7374 , tail

@@ -475,9 +476,7 @@ fromStrict t = chunk t Empty

475476-- -----------------------------------------------------------------------------

476477-- * Basic functions

477478

478−-- | /O(n)/ Adds a character to the front of a 'Text'. This function

479−-- is more costly than its 'List' counterpart because it requires

480−-- copying a new array. Subject to fusion.

479+-- | /O(1)/ Adds a character to the front of a 'Text'. Subject to fusion.

481480cons :: Char -> Text -> Text

482481cons c t = Chunk (T.singleton c) t

483482{-# INLINE [1] cons #-}

@@ -561,6 +560,15 @@ init Empty = emptyError "init"

561560 unstream (S.init (stream t)) = init t

562561 #-}

563562

563+-- | /O(n\/c)/ Returns the 'init' and 'last' of a 'Text', or 'Nothing' if

564+-- empty.

565+--

566+-- * It is no faster than using 'init' and 'last'.

567+unsnoc :: Text -> Maybe (Text, Char)

568+unsnoc Empty = Nothing

569+unsnoc ts@(Chunk _ _) = Just (init ts, last ts)

570+{-# INLINE unsnoc #-}

571+

564572-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to

565573-- fusion.

566574null :: Text -> Bool

@@ -1153,9 +1161,10 @@ takeWhile p t0 = takeWhile' t0

11531161takeWhileEnd :: (Char -> Bool) -> Text -> Text

11541162takeWhileEnd p = takeChunk empty . L.reverse . toChunks

11551163 where takeChunk acc [] = acc

1156− takeChunk acc (t:ts) = if T.length t' < T.length t

1157− then (Chunk t' acc)

1158− else takeChunk (Chunk t' acc) ts

1164+ takeChunk acc (t:ts)

1165+ | T.lengthWord16 t' < T.lengthWord16 t

1166+ = chunk t' acc

1167+ | otherwise = takeChunk (Chunk t' acc) ts

11591168 where t' = T.takeWhileEnd p t

11601169{-# INLINE takeWhileEnd #-}

11611170

Data/Text/Lazy/Read.hs+3−3

@@ -68,9 +68,9 @@ decimal txt

6868 go n d = (n * 10 + fromIntegral (digitToInt d))

6969

7070-- | Read a hexadecimal integer, consisting of an optional leading

71−-- @\"0x\"@ followed by at least one decimal digit. Input is consumed

72−-- until a non-hex-digit or end of string is reached. This function

73−-- is case insensitive.

71+-- @\"0x\"@ followed by at least one hexadecimal digit. Input is

72+-- consumed until a non-hex-digit or end of string is reached.

73+-- This function is case insensitive.

7474--

7575-- This function does not handle leading sign characters. If you need

7676-- to handle signed input, use @'signed' 'hexadecimal'@.

Data/Text/Read.hs+3−3

@@ -67,9 +67,9 @@ decimal txt

6767 go n d = (n * 10 + fromIntegral (digitToInt d))

6868

6969-- | Read a hexadecimal integer, consisting of an optional leading

70−-- @\"0x\"@ followed by at least one decimal digit. Input is consumed

71−-- until a non-hex-digit or end of string is reached. This function

72−-- is case insensitive.

70+-- @\"0x\"@ followed by at least one hexadecimal digit. Input is

71+-- consumed until a non-hex-digit or end of string is reached.

72+-- This function is case insensitive.

7373--

7474-- This function does not handle leading sign characters. If you need

7575-- to handle signed input, use @'signed' 'hexadecimal'@.

changelog.md+29−1

@@ -1,10 +1,38 @@

1+1.2.3.0

3+* `unsnoc` implemented

5+* Bug fix: the lazy `takeWhileEnd` function violated the

6+ [lazy text invariant](https://github.com/bos/text/blob/1.2.3.0/Data/Text/Internal/Lazy.hs#L51).

8+* Spec compliance: `toCaseFold` now follows the Unicode 9.0 spec

9+ (updated from 8.0)

10+

11+

12+1.2.2.2

13+

14+* The `toTitle` function now correctly handles letters that

15+ immediately follow punctuation. Before, `"there's"` would turn into

16+ `"There'S"`. Now, it becomes `"There's"`.

17+

18+* The implementation of unstreaming is faster, resulting in operations

19+ such as `map` and `intersperse` speeding up by up to 30%, with

20+ smaller code generated.

21+

22+* The optimised length comparison function is now more likely to be

23+ used after some rewrite rule tweaking.

24+

25+* Bug fix: an off-by-one bug in `takeEnd` is fixed.

26+

27+* Bug fix: a logic error in `takeWord16` is fixed.

28+

1291.2.2.1

230

331* The switch to `integer-pure` in 1.2.2.0 was apparently mistaken.

432 The build flag has been renamed accordingly. Your army of diligent

533 maintainers apologizes for the churn.

634

7−* Spec compliance: toCaseFold now follows the Unicode 8.0 spec

35+* Spec compliance: `toCaseFold` now follows the Unicode 8.0 spec

836 (updated from 7.0)

937

1038* An STG lint error has been fixed

tests/Tests/Properties.hs+58−21

@@ -9,9 +9,9 @@ module Tests.Properties

99 ) where

1010

1111import Control.Applicative ((<$>), (<*>))

12−import Control.Arrow ((***), second)

12+import Control.Arrow ((***), first, second)

1313import Data.Bits ((.&.))

14−import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isUpper, ord)

14+import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isLetter, isUpper, ord)

1515import Data.Int (Int8, Int16, Int32, Int64)

1616import Data.Monoid (Monoid(..))

1717import Data.String (IsString(fromString))

@@ -23,6 +23,7 @@ import Data.Text.Internal.Search (indices)

2323import Data.Text.Lazy.Read as TL

2424import Data.Text.Read as T

2525import Data.Word (Word, Word8, Word16, Word32, Word64)

26+import Data.Maybe (mapMaybe)

2627import Numeric (showEFloat, showFFloat, showGFloat, showHex)

2728import Prelude hiding (replicate)

2829import Test.Framework (Test, testGroup)

@@ -30,6 +31,7 @@ import Test.Framework.Providers.QuickCheck2 (testProperty)

3031import Test.QuickCheck hiding ((.&.))

3132import Test.QuickCheck.Monadic

3233import Test.QuickCheck.Property (Property(..))

34+import Test.QuickCheck.Unicode (char)

3335import Tests.QuickCheckUtils

3436import Tests.Utils

3537import Text.Show.Functions ()

@@ -37,6 +39,7 @@ import qualified Control.Exception as Exception

3739import qualified Data.Bits as Bits (shiftL, shiftR)

3840import qualified Data.ByteString as B

3941import qualified Data.ByteString.Lazy as BL

42+import qualified Data.Char as C

4043import qualified Data.List as L

4144import qualified Data.Text as T

4245import qualified Data.Text.Encoding as E

@@ -60,10 +63,10 @@ t_pack_unpack = (T.unpack . T.pack) `eq` id

6063tl_pack_unpack = (TL.unpack . TL.pack) `eq` id

6164t_stream_unstream = (S.unstream . S.stream) `eq` id

6265tl_stream_unstream = (SL.unstream . SL.stream) `eq` id

63−t_reverse_stream t = (S.reverse . S.reverseStream) t == t

64−t_singleton c = [c] == (T.unpack . T.singleton) c

65−tl_singleton c = [c] == (TL.unpack . TL.singleton) c

66−tl_unstreamChunks x = f 11 x == f 1000 x

66+t_reverse_stream t = (S.reverse . S.reverseStream) t === t

67+t_singleton c = [c] === (T.unpack . T.singleton) c

68+tl_singleton c = [c] === (TL.unpack . TL.singleton) c

69+tl_unstreamChunks x = f 11 x === f 1000 x

6770 where f n = SL.unstreamChunks n . S.streamList

6871tl_chunk_unchunk = (TL.fromChunks . TL.toChunks) `eq` id

6972tl_from_to_strict = (TL.fromStrict . TL.toStrict) `eq` id

@@ -74,13 +77,13 @@ encodeL1 = B.pack . map (fromIntegral . fromEnum) . T.unpack

7477encodeLazyL1 :: TL.Text -> BL.ByteString

7578encodeLazyL1 = BL.fromChunks . map encodeL1 . TL.toChunks

7679

77−t_ascii t = E.decodeASCII (E.encodeUtf8 a) == a

80+t_ascii t = E.decodeASCII (E.encodeUtf8 a) === a

7881 where a = T.map (\c -> chr (ord c `mod` 128)) t

79−tl_ascii t = EL.decodeASCII (EL.encodeUtf8 a) == a

82+tl_ascii t = EL.decodeASCII (EL.encodeUtf8 a) === a

8083 where a = TL.map (\c -> chr (ord c `mod` 128)) t

81−t_latin1 t = E.decodeLatin1 (encodeL1 a) == a

84+t_latin1 t = E.decodeLatin1 (encodeL1 a) === a

8285 where a = T.map (\c -> chr (ord c `mod` 256)) t

83−tl_latin1 t = EL.decodeLatin1 (encodeLazyL1 a) == a

86+tl_latin1 t = EL.decodeLatin1 (encodeLazyL1 a) === a

8487 where a = TL.map (\c -> chr (ord c `mod` 256)) t

8588t_utf8 = forAll genUnicode $ (E.decodeUtf8 . E.encodeUtf8) `eq` id

8689t_utf8' = forAll genUnicode $ (E.decodeUtf8' . E.encodeUtf8) `eq` (id . Right)

@@ -111,7 +114,7 @@ t_utf8_undecoded = forAll genUnicode $ \t ->

111114 let b = E.encodeUtf8 t

112115 ls = concatMap (leftover . E.encodeUtf8 . T.singleton) . T.unpack $ t

113116 leftover = (++ [B.empty]) . init . tail . B.inits

114− in (map snd . feedChunksOf 1 E.streamDecodeUtf8) b == ls

117+ in (map snd . feedChunksOf 1 E.streamDecodeUtf8) b === ls

115118

116119data Badness = Solo | Leading | Trailing

117120 deriving (Eq, Show)

@@ -201,8 +204,8 @@ t_mconcat = unsquare $

201204 mconcat `eq` (unpackS . mconcat . L.map T.pack)

202205tl_mconcat = unsquare $

203206 mconcat `eq` (unpackS . mconcat . L.map TL.pack)

204−t_mempty = mempty == (unpackS (mempty :: T.Text))

205−tl_mempty = mempty == (unpackS (mempty :: TL.Text))

207+t_mempty = mempty === (unpackS (mempty :: T.Text))

208+tl_mempty = mempty === (unpackS (mempty :: TL.Text))

206209t_IsString = fromString `eqP` (T.unpack . fromString)

207210tl_IsString = fromString `eqP` (TL.unpack . fromString)

208211

@@ -229,6 +232,13 @@ sf_uncons p = (uncons . L.filter p) `eqP`

229232 (fmap (second unpackS) . S.uncons . S.filter p)

230233t_uncons = uncons `eqP` (fmap (second unpackS) . T.uncons)

231234tl_uncons = uncons `eqP` (fmap (second unpackS) . TL.uncons)

235+

236+unsnoc xs@(_:_) = Just (init xs, last xs)

237+unsnoc [] = Nothing

238+

239+t_unsnoc = unsnoc `eqP` (fmap (first unpackS) . T.unsnoc)

240+tl_unsnoc = unsnoc `eqP` (fmap (first unpackS) . TL.unsnoc)

241+

232242s_head = head `eqP` S.head

233243sf_head p = (head . L.filter p) `eqP` (S.head . S.filter p)

234244t_head = head `eqP` T.head

@@ -326,6 +336,19 @@ t_toUpper_upper t = p (T.toUpper t) >= p t

326336 where p = T.length . T.filter isUpper

327337tl_toUpper_upper t = p (TL.toUpper t) >= p t

328338 where p = TL.length . TL.filter isUpper

339+t_toTitle_title t = all (<= 1) (caps w)

340+ where caps = fmap (T.length . T.filter isUpper) . T.words . T.toTitle

341+ -- TIL: there exist uppercase-only letters

342+ w = T.filter (\c -> if C.isUpper c then C.toLower c /= c else True) t

343+t_toTitle_1stNotLower = and . notLow . T.toTitle . T.filter stable

344+ where notLow = mapMaybe (fmap (not . isLower) . (T.find isLetter)) . T.words

345+ -- Surprise! The Spanish/Portuguese ordinal indicators changed

346+ -- from category Ll (letter, lowercase) to Lo (letter, other)

347+ -- in Unicode 7.0

348+ -- Oh, and there exist lowercase-only letters (see previous test)

349+ stable c = if isLower c

350+ then C.toUpper c /= c

351+ else c /= '\170' && c /= '\186'

329352

330353justifyLeft k c xs = xs ++ L.replicate (k - length xs) c

331354justifyRight m n xs = L.replicate (m - length xs) n ++ xs

@@ -514,12 +537,20 @@ s_takeWhile p = L.takeWhile p `eqP` (unpackS . S.takeWhile p)

514537s_takeWhile_s p = L.takeWhile p `eqP` (unpackS . S.unstream . S.takeWhile p)

515538sf_takeWhile q p = (L.takeWhile p . L.filter q) `eqP`

516539 (unpackS . S.takeWhile p . S.filter q)

540+noMatch = do

541+ c <- char

542+ d <- suchThat char (/= c)

543+ return (c,d)

517544t_takeWhile p = L.takeWhile p `eqP` (unpackS . T.takeWhile p)

518545tl_takeWhile p = L.takeWhile p `eqP` (unpackS . TL.takeWhile p)

519546t_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

520547 (unpackS . T.takeWhileEnd p)

548+t_takeWhileEnd_null t = forAll noMatch $ \(c,d) -> T.null $

549+ T.takeWhileEnd (==d) (T.snoc t c)

521550tl_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

522551 (unpackS . TL.takeWhileEnd p)

552+tl_takeWhileEnd_null t = forAll noMatch $ \(c,d) -> TL.null $

553+ TL.takeWhileEnd (==d) (TL.snoc t c)

523554s_dropWhile p = L.dropWhile p `eqP` (unpackS . S.dropWhile p)

524555s_dropWhile_s p = L.dropWhile p `eqP` (unpackS . S.unstream . S.dropWhile p)

525556sf_dropWhile q p = (L.dropWhile p . L.filter q) `eqP`

@@ -720,7 +751,7 @@ tl_indices (NotEmpty s) = lazyIndices s `eq` S.indices s

720751 conc = T.concat . TL.toChunks

721752t_indices_occurs = unsquare $ \(NotEmpty t) ts ->

722753 let s = T.intercalate t ts

723− in Slow.indices t s == indices t s

754+ in Slow.indices t s === indices t s

724755

725756-- Bit shifts.

726757shiftL w = forAll (choose (0,width-1)) $ \k -> Bits.shiftL w k == U.shiftL w k

@@ -812,18 +843,18 @@ tb_formatRealFloat_double (a::Double) = tb_formatRealFloat a

812843-- Reading.

813844

814845t_decimal (n::Int) s =

815− T.signed T.decimal (T.pack (show n) `T.append` t) == Right (n,t)

846+ T.signed T.decimal (T.pack (show n) `T.append` t) === Right (n,t)

816847 where t = T.dropWhile isDigit s

817848tl_decimal (n::Int) s =

818− TL.signed TL.decimal (TL.pack (show n) `TL.append` t) == Right (n,t)

849+ TL.signed TL.decimal (TL.pack (show n) `TL.append` t) === Right (n,t)

819850 where t = TL.dropWhile isDigit s

820851t_hexadecimal m s ox =

821− T.hexadecimal (T.concat [p, T.pack (showHex n ""), t]) == Right (n,t)

852+ T.hexadecimal (T.concat [p, T.pack (showHex n ""), t]) === Right (n,t)

822853 where t = T.dropWhile isHexDigit s

823854 p = if ox then "0x" else ""

824855 n = getPositive m :: Int

825856tl_hexadecimal m s ox =

826− TL.hexadecimal (TL.concat [p, TL.pack (showHex n ""), t]) == Right (n,t)

857+ TL.hexadecimal (TL.concat [p, TL.pack (showHex n ""), t]) === Right (n,t)

827858 where t = TL.dropWhile isHexDigit s

828859 p = if ox then "0x" else ""

829860 n = getPositive m :: Int

@@ -867,11 +898,11 @@ tl_write_read_line e m b t = write_read head TL.filter TL.hPutStrLn

867898

868899t_dropWord16 m t = dropWord16 m t `T.isSuffixOf` t

869900t_takeWord16 m t = takeWord16 m t `T.isPrefixOf` t

870−t_take_drop_16 m t = T.append (takeWord16 n t) (dropWord16 n t) == t

901+t_take_drop_16 m t = T.append (takeWord16 n t) (dropWord16 n t) === t

871902 where n = small m

872903t_use_from t = monadicIO $ assert . (==t) =<< run (useAsPtr t fromPtr)

873904

874−t_copy t = T.copy t == t

905+t_copy t = T.copy t === t

875906

876907-- Regression tests.

877908s_filter_eq s = S.filter p t == S.streamList (filter p s)

@@ -967,6 +998,8 @@ tests =

967998 testProperty "sf_uncons" sf_uncons,

968999 testProperty "t_uncons" t_uncons,

9691000 testProperty "tl_uncons" tl_uncons,

1001+ testProperty "t_unsnoc" t_unsnoc,

1002+ testProperty "tl_unsnoc" tl_unsnoc,

9701003 testProperty "s_head" s_head,

9711004 testProperty "sf_head" sf_head,

9721005 testProperty "t_head" t_head,

@@ -1030,7 +1063,9 @@ tests =

10301063 testProperty "tl_toLower_lower" tl_toLower_lower,

10311064 testProperty "t_toUpper_length" t_toUpper_length,

10321065 testProperty "t_toUpper_upper" t_toUpper_upper,

1033− testProperty "tl_toUpper_upper" tl_toUpper_upper

1066+ testProperty "tl_toUpper_upper" tl_toUpper_upper,

1067+ testProperty "t_toTitle_title" t_toTitle_title,

1068+ testProperty "t_toTitle_1stNotLower" t_toTitle_1stNotLower

10341069 ],

10351070

10361071 testGroup "justification" [

@@ -1147,7 +1182,9 @@ tests =

11471182 testProperty "t_takeWhile" t_takeWhile,

11481183 testProperty "tl_takeWhile" tl_takeWhile,

11491184 testProperty "t_takeWhileEnd" t_takeWhileEnd,

1185+ testProperty "t_takeWhileEnd_null" t_takeWhileEnd_null,

11501186 testProperty "tl_takeWhileEnd" tl_takeWhileEnd,

1187+ testProperty "tl_takeWhileEnd_null" tl_takeWhileEnd_null,

11511188 testProperty "sf_dropWhile" sf_dropWhile,

11521189 testProperty "s_dropWhile" s_dropWhile,

11531190 testProperty "s_dropWhile_s" s_dropWhile_s,

tests/Tests/QuickCheckUtils.hs+1−1

@@ -106,7 +106,7 @@ smallArbitrary = sized $ \n -> resize (smallish n) arbitrary

106106 where smallish = round . (sqrt :: Double -> Double) . fromIntegral . abs

107107

108108instance Arbitrary T.Text where

109− arbitrary = T.pack `fmap` arbitrary

109+ arbitrary = T.pack `fmap` string

110110 shrink = map T.pack . shrink . T.unpack

111111

112112instance Arbitrary TL.Text where

tests/text-tests.cabal+18−3

@@ -20,6 +20,11 @@ flag hpc

2020 default: False

2121 manual: True

2222

23+flag bytestring-builder

24+ description: Depend on the bytestring-builder package for backwards compatibility.

25+ default: False

26+ manual: False

27+

2328executable text-tests

2429 main-is: Tests.hs

2530

@@ -48,16 +53,21 @@ executable text-tests

4853 HUnit >= 1.2,

4954 QuickCheck >= 2.7,

5055 base == 4.*,

51− bytestring,

5256 deepseq,

5357 directory,

54− quickcheck-unicode,

58+ quickcheck-unicode >= 1.0.1.0,

5559 random,

5660 test-framework >= 0.4,

5761 test-framework-hunit >= 0.2,

5862 test-framework-quickcheck2 >= 0.2,

5963 text-tests

6064

65+ if flag(bytestring-builder)

66+ build-depends: bytestring >= 0.9 && < 0.10.4,

67+ bytestring-builder >= 0.10.4

68+ else

69+ build-depends: bytestring >= 0.10.4

70+

6171executable text-tests-stdio

6272 main-is: Tests/IO.hs

6373

@@ -137,7 +147,12 @@ library

137147 array,

138148 base == 4.*,

139149 binary,

140− bytestring,

141150 deepseq,

142151 ghc-prim,

143152 integer-gmp

153+

154+ if flag(bytestring-builder)

155+ build-depends: bytestring >= 0.9 && < 0.10.4,

156+ bytestring-builder >= 0.10.4

157+ else

158+ build-depends: bytestring >= 0.10.4

text.cabal+5−5

@@ -1,7 +1,7 @@

11name: text

2−version: 1.2.2.1

3−homepage: https://github.com/bos/text

4−bug-reports: https://github.com/bos/text/issues

2+version: 1.2.3.0

3+homepage: https://github.com/haskell/text

4+bug-reports: https://github.com/haskell/text/issues

55synopsis: An efficient packed Unicode text type.

66description:

77 .

@@ -180,7 +180,7 @@ test-suite tests

180180 deepseq,

181181 directory,

182182 ghc-prim,

183− quickcheck-unicode,

183+ quickcheck-unicode >= 1.0.1.0,

184184 random,

185185 test-framework >= 0.4,

186186 test-framework-hunit >= 0.2,

@@ -201,7 +201,7 @@ test-suite tests

201201

202202source-repository head

203203 type: git

204− location: https://github.com/bos/text

204+ location: https://github.com/haskell/text

205205

206206source-repository head

207207 type: mercurial