diff --git a/Data-Rope.cabal b/Data-Rope.cabal --- a/Data-Rope.cabal +++ b/Data-Rope.cabal @@ -1,5 +1,5 @@ Name: Data-Rope -Version: 0.1 +Version: 0.2 Synopsis: Ropes, an alternative to (Byte)Strings. Description: Ropes : an alternative to Strings, a time and space-efficient representation of character diff --git a/Data/Rope.lhs b/Data/Rope.lhs --- a/Data/Rope.lhs +++ b/Data/Rope.lhs @@ -46,10 +46,13 @@ -- * Indexing 'Rope's index, elemIndex, + elemIndex', elemIndices, -- * Input and Output readFile, hGet, + hGetLine, + hGetContents, hPut, hPutStrLn, hPutStr, @@ -72,7 +75,7 @@ import qualified Data.ByteString.Internal as B import Data.String import Data.Word - +import Control.Exception (finally) #if __GLASGOW_HASKELL__ >=608 import GHC.IO (IO(IO),unsafePerformIO,unsafeInterleaveIO) import GHC.Base (realWorld#) @@ -83,11 +86,13 @@ import System.IO.Unsafe #endif import Data.Maybe -import GHC.Conc +import GHC.Conc (par) import System.Posix (openFd,closeFd,defaultFileFlags,getFileStatus, fileSize,OpenMode(..)) -import System.IO hiding (hPutStrLn,hPutStr,putStrLn,putStr,readFile) +import System.IO (hGetBuf,hPutBuf,hSeek,hTell,hFileSize, + hPutChar, stdout, hClose) + import qualified Prelude(length,map,splitAt) import Prelude hiding (length,null,head,tail,last,init,map,reverse, foldl,foldr, @@ -298,7 +303,7 @@ -} -- | O(1) Tests whether a 'Rope' is empty. null::Rope->Bool -null x=length x>0 +null x=length x==0 -- | O(1) Length of a 'Rope'. length::Rope->Int @@ -626,12 +631,13 @@ index_ (Concat{r,l}) i_ |i_>=length l = index_ r (i_-length l) |otherwise = index_ l i_ + {- index_ (File{handle,position}) i_= unsafePerformIO $ alloca $ \cc->do hSeek handle AbsoluteSeek $ fromIntegral $ position+i_ hGetBuf handle cc 1 peek cc - +-} -- | O(n) returns the index of the first element equal to the query element. This implementation -- uses memchr at leaves, and explores the rope in parallel (with 'GHC.Conc.par'). @@ -664,6 +670,36 @@ x<-peek ptr if x==w then return $ Just i else loop (ptr`plusPtr`1) (i+1) l +-- | O(n) Same as 'elemIndex', but explores the 'Rope' sequentially. Useful for +-- 'Rope's loaded lazily with 'readFile'. + +elemIndex'::Word8->Rope->Maybe Int +elemIndex' w rope= + elemIndex_ 0 rope + where + elemIndex_ i (String{contents,offset,length_})= + inlinePerformIO $ withForeignPtr contents $ \con->do + ptr<-memchr (con`plusPtr`offset) w (fromIntegral length_) + return $ if ptr==nullPtr then Nothing else Just $ i+offset+(ptr`minusPtr`con) + + elemIndex_ i (Concat{l,r})= + case elemIndex_ i l of + Nothing->elemIndex_ (i+length l) r + a->a + + elemIndex_ i (File{handle,position,length_})= + unsafePerformIO $ allocaBytes length_ $ \c->do + hSeek handle AbsoluteSeek $ fromIntegral position + l<-hGetBuf handle c length_ + x<-loop c 0 l + return $ x>>=(return.(+i)) + + loop ptr i l + | i>=l = return Nothing + | otherwise = do + x<-peek ptr + if x==w then return $ Just i else loop (ptr`plusPtr`1) (i+1) l + -- | O(n) returns the list of all positions where the queried elements occurs in the 'Rope'. -- This implementation uses memchr. @@ -921,8 +957,7 @@ putStrLn::Rope->IO() putStrLn=hPutStrLn stdout - - +-- | fromPtr encapsule un pointeur, un offset et une longueur dans un @Rope@.. fromPtr::ForeignPtr Word8->Int->Int->IO Rope fromPtr fp offset s | s<=leafSize = withForeignPtr fp $ \p->do @@ -983,6 +1018,7 @@ fromPtr fp 0 (fromIntegral s) + -- | Strict hGet. The whole rope is constructed. hGet::Handle->Int->IO Rope hGet handle length_=do @@ -1012,5 +1048,55 @@ bb<-buildRope handle (position+fromIntegral a) b return $ Concat { length_=len,l=aa,r=bb,sizeC=size aa+size bb } +-- | Returns the next line in the input 'Handle'. If you need to iterate 'hGetLine', +-- it may be more efficient to first @mmap@ the file using 'readFile', or even load +-- it with then iterate +-- @'breakByte' 0x0a@ : 'hGetLine' allocates a buffer to read the file +-- and may waste most of this space if the lines are shorter than the standard buffer +-- size of this module. +hGetLine::Handle->IO Rope +hGetLine h= + hGetLine' empty + + where + hGetLine' x=do + contents<-mallocForeignPtrBytes leafSize + withForeignPtr contents $ \con->do + l<-hGetBuf h con leafSize + let readBytes=min l leafSize + if readBytes<=0 then return x else do + ptr<-memchr con 0x0a (fromIntegral readBytes) + if ptr==nullPtr then do + i0<-mallocForeignPtr + withForeignPtr i0 $ \i->poke i readBytes + let y=String { contents,i0,offset=0,length_=readBytes } + hGetLine' (x`append`y) + + else do + + i0<-mallocForeignPtr + let lineLen=ptr`minusPtr`con + y=String { contents,i0,offset=0,length_=lineLen } + withForeignPtr i0 $ \i->poke i lineLen + hSeek h RelativeSeek (fromIntegral $ lineLen-readBytes+1) + return $ x`append`y + +-- | Reads the contents of a file handle strictly, then closes it. + +hGetContents::Handle->IO Rope +hGetContents h=finally (hGetContents' empty) (hClose h) + + where + hGetContents' r=do + contents<-mallocForeignPtrBytes leafSize + withForeignPtr contents $ \l->do + x<-hGetBuf h l leafSize + i0<-mallocForeignPtr + withForeignPtr i0 $ \i->poke i x + let y=String { contents,i0,offset=0,length_=x } + + if x==0 then return r else + if xRope +singleton x=(R.singleton).w2c + + +\end{code} diff --git a/Data/Rope/Internals.hsc b/Data/Rope/Internals.hsc --- a/Data/Rope/Internals.hsc +++ b/Data/Rope/Internals.hsc @@ -7,7 +7,7 @@ import Foreign.Storable import Foreign.C.Types import Data.Word - +import GHC.Base foreign import ccall unsafe "sys/mman.h mmap" c_mmap:: Ptr a->CSize->CInt->CInt->CInt->CInt->IO (Ptr Word8) @@ -32,3 +32,10 @@ c_MAP_FILE::CInt c_MAP_FILE= #const MAP_FILE + + +w2c::Word8->Char +w2c=unsafeChr.fromIntegral + +c2w::Char->Word8 +c2w=fromIntegral.ord