2b2cb08 · text · git.iko.soy

git.iko.soy/text

Reduce the amount of code generated for string literals

Previously, every Text literal generated a big wad of inefficient code: {-# LANGUAGE OverloadedStrings #-} foo :: Text foo = "foo" This would first convert to a String, then to a Text. To make matters worse, the code for conversion from String to Text was inlined at every site where a string literal occurred (expected, but undesired, behaviour). In this change, we introduce a direct conversion from Addr# to each of the Text types, and we ensure that uses of these never result in excessive code generation.

commit2b2cb084c4689c06f1a7851ff8eb1e412eb02c1b authorBryan O'Sullivan <bos@serpentine.com> dateSat 7 Apr 2012, 21:33 (15 years ago) parent0d6d979

3 files changed+69−6

compare with master

unifiedsplit

Data/Text.hs+17−2 Data/Text/Fusion/Common.hs+35−2 Data/Text/Lazy.hs+17−2

Data/Text.hs+17−2

@@ -1,9 +1,9 @@

1−{-# LANGUAGE BangPatterns, CPP, Rank2Types, UnboxedTuples #-}

1+{-# LANGUAGE BangPatterns, CPP, MagicHash, Rank2Types, UnboxedTuples #-}

22{-# OPTIONS_GHC -fno-warn-orphans #-}

44-- |

55-- Module : Data.Text

99--

@@ -222,6 +222,8 @@ import Data.ByteString (ByteString)

222222import qualified Data.Text.Lazy as L

223223import Data.Int (Int64)

224224#endif

225+import qualified GHC.CString as GHC

226+import GHC.Prim (Addr#)

225227

226228-- $strict

227229--

@@ -372,6 +374,19 @@ unpack :: Text -> String

372374unpack = S.unstreamList . stream

373375{-# INLINE [1] unpack #-}

374376

377+-- | /O(n)/ Convert a literal string into a Text.

378+unpackCString# :: Addr# -> Text

379+unpackCString# addr# = unstream (S.streamCString# addr#)

380+{-# NOINLINE unpackCString# #-}

381+

382+{-# RULES "TEXT literal" forall a.

383+ unstream (S.streamList (L.map safe (GHC.unpackCString# a)))

384+ = unpackCString# a #-}

385+

386+{-# RULES "TEXT literal UTF8" forall a.

387+ unstream (S.streamList (L.map safe (GHC.unpackCStringUtf8# a)))

388+ = unpackCString# a #-}

389+

375390-- | /O(1)/ Convert a character into a Text. Subject to fusion.

376391-- Performs replacement on invalid scalar values.

377392singleton :: Char -> Text

Data/Text/Fusion/Common.hs+35−2

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

1−{-# LANGUAGE BangPatterns, Rank2Types #-}

1+{-# LANGUAGE BangPatterns, MagicHash, Rank2Types #-}

22-- |

33-- Module : Data.Text.Fusion.Common

4−-- Copyright : (c) Bryan O'Sullivan 2009

4+-- Copyright : (c) Bryan O'Sullivan 2009, 2012

55--

66-- License : BSD-style

77-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

@@ -17,6 +17,7 @@ module Data.Text.Fusion.Common

1717 singleton

1818 , streamList

1919 , unstreamList

20+ , streamCString#

2021

2122 -- * Basic interface

2223 , cons

@@ -104,10 +105,13 @@ import Prelude (Bool(..), Char, Eq(..), Int, Integral, Maybe(..),

104105 (&&), fromIntegral, otherwise)

105106import qualified Data.List as L

106107import qualified Prelude as P

108+import Data.Bits (shiftL)

107109import Data.Int (Int64)

108110import Data.Text.Fusion.Internal

109111import Data.Text.Fusion.CaseMapping (foldMapping, lowerMapping, upperMapping)

110112import Data.Text.Fusion.Size

113+import GHC.Prim (Addr#, chr#, indexCharOffAddr#, ord#)

114+import GHC.Types (Char(..), Int(..))

111115

112116singleton :: Char -> Stream Char

113117singleton c = Stream next False 1

@@ -131,6 +135,35 @@ unstreamList (Stream next s0 _len) = unfold s0

131135

132136{-# RULES "STREAM streamList/unstreamList fusion" forall s. streamList (unstreamList s) = s #-}

133137

138+-- | Stream the UTF-8-like packed encoding used by GHC to represent

139+-- constant strings in generated code.

140+--

141+-- This encoding uses the byte sequence "\xc0\x80" to represent NUL,

142+-- and the string is NUL-terminated.

143+streamCString# :: Addr# -> Stream Char

144+streamCString# addr = Stream step 0 unknownSize

145+ where

146+ step !i

147+ | b == 0 = Done

148+ | b <= 0x7f = Yield (C# b#) (i+1)

149+ | b <= 0xdf = let !c = chr $ ((b-0xc0) `shiftL` 6) + next 1

150+ in Yield c (i+2)

151+ | b <= 0xef = let !c = chr $ ((b-0xe0) `shiftL` 12) +

152+ (next 1 `shiftL` 6) +

153+ next 2

154+ in Yield c (i+3)

155+ | otherwise = let !c = chr $ ((b-0xf0) `shiftL` 18) +

156+ (next 1 `shiftL` 12) +

157+ (next 2 `shiftL` 6) +

158+ next 3

159+ in Yield c (i+4)

160+ where b = I# (ord# b#)

161+ next n = I# (ord# (at# (i+n))) - 0x80

162+ b# = at# i

163+ at# (I# i#) = indexCharOffAddr# addr i#

164+ chr (I# i#) = C# (chr# i#)

165+{-# INLINE [0] streamCString# #-}

166+

134167-- ----------------------------------------------------------------------------

135168-- * Basic stream functions

136169

Data/Text/Lazy.hs+17−2

@@ -1,8 +1,8 @@

11{-# OPTIONS_GHC -fno-warn-orphans #-}

2−{-# LANGUAGE BangPatterns, CPP #-}

2+{-# LANGUAGE BangPatterns, MagicHash, CPP #-}

33-- |

44-- Module : Data.Text.Lazy

66--

77-- License : BSD-style

88-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

@@ -217,6 +217,8 @@ import Data.Text.Lazy.Internal (Text(..), chunk, empty, foldlChunks, foldrChunks

217217import Data.Text.Internal (firstf, safe, textP)

218218import qualified Data.Text.Util as U

219219import Data.Text.Lazy.Search (indices)

220+import qualified GHC.CString as GHC

221+import GHC.Prim (Addr#)

220222

221223-- $fusion

222224--

@@ -351,6 +353,19 @@ unpack :: Text -> String

351353unpack t = S.unstreamList (stream t)

352354{-# INLINE [1] unpack #-}

353355

356+-- | /O(n)/ Convert a literal string into a Text.

357+unpackCString# :: Addr# -> Text

358+unpackCString# addr# = unstream (S.streamCString# addr#)

359+{-# NOINLINE unpackCString# #-}

360+

361+{-# RULES "TEXT literal" forall a.

362+ unstream (S.streamList (L.map safe (GHC.unpackCString# a)))

363+ = unpackCString# a #-}

364+

365+{-# RULES "TEXT literal UTF8" forall a.

366+ unstream (S.streamList (L.map safe (GHC.unpackCStringUtf8# a)))

367+ = unpackCString# a #-}

368+

354369-- | /O(1)/ Convert a character into a Text. Subject to fusion.

355370-- Performs replacement on invalid scalar values.

356371singleton :: Char -> Text