Use Haskell ByteString to Read/Write File

Haskell的list类型应用很广泛，但不是做所有事情都适合用。List类型从头部取一个元素，或从头部插入一个元素的时间复杂度是O(1)，如果随机访问任意位置的元素，则时间复杂度为O(n)。List类型如果求它的长度用length函数，则时间复杂度为O(n)，所以如果判断一个List对象是否为空，最好用null函数，而不是判断length等于0。

 

Haskell在读写文件的时候，尤其大的文件的时候，最好不采用List数据类型。Haskell提供的Data.ByteString类型，可以用来比较有效率地读写大文件。

 

下面是一个应用Data.ByteString来分割合并文件的例子：

 

 import qualified Data.ByteString.Lazy as BS import System.Environment (getArgs, getProgName ) import System.IO (IOMode(ReadMode), hClose, openFile, hFileSize, hGetLine,readFile,writeFile,appendFile, Handle(..) ) import Data.String (fromString) import Data.Int import Control.Exception (handle) main = do progName <- getProgName args <- getArgs case args of -- two args, divide file ([_,_]) -> do let divideFileName = args !! 0 let numberOfPart = read (args !! 1 ) :: Integer divideFile divideFileName numberOfPart -- three args, merge file ([_,_,_]) -> do let option = args !! 0 if option == "-m" then do let partFileName = args !! 1 let outFileName = args !! 2 mergeFile partFileName outFileName else printUsage progName -- otherwise print usage (_) -> printUsage progName {- partFileName is gemerated when dividing file; outFileName is the fileName for merged file -} mergeFile partFileName outFileName = do names<-partFileNames partFileName mergeFile' names outFileName {- fileNames is get from partFile which is generated when dividing file -} mergeFile' fileNames outFileName = case fileNames of (fileName:fs) -> do content <- BS.readFile fileName BS.appendFile outFileName content mergeFile' fs outFileName _ -> return () partFileNames :: FilePath ->IO [FilePath] partFileNames partFileName = do fileHandle <- openFile partFileName ReadMode result <- parseFileName fileHandle hClose fileHandle return result parseFileName::Handle->IO [FilePath] parseFileName fileHandle = handle ( eofEncuntered ) $ do line<-hGetLine fileHandle let fileName = line2FileName line fileNames<-parseFileName fileHandle return (fileName:fileNames) {- the second word in the line is file names -} line2FileName::String->String line2FileName s = words s !! 1 eofEncuntered::IOError ->IO [FilePath] eofEncuntered e = return [] printUsage progName = do mapM_ putStrLn lines where lines = [progName ++ " <divide-file-name> <number-of-parts> ", progName ++ " -m <parts-file-name> <out-file-name> "] filePartList :: Integer -> Integer -> [Integer] filePartList fileTotalSize numberOfPart = let eachSize = fileTotalSize `div` numberOfPart remain = fileTotalSize `mod` numberOfPart in if remain > 0 then filePartList' fileTotalSize (eachSize+1) else filePartList' fileTotalSize eachSize where filePartList' fileTotalSize eachSize | fileTotalSize <= eachSize = [fileTotalSize] | otherwise = eachSize : (filePartList' (fileTotalSize - eachSize) eachSize) divideFile :: FilePath -> Integer -> IO () divideFile divideFileName numberOfPart = do fileTotalSize <- fileSize divideFileName content<-BS.readFile divideFileName let fileSizeList = filePartList fileTotalSize numberOfPart let fileMapList = zip [0..] fileSizeList saveIndividualFile divideFileName fileMapList content saveDivideInfo divideFileName fileMapList saveDivideInfo :: FilePath -> [(Integer,Integer)] -> IO () saveDivideInfo fileName fileMapList = writeFile (fileName++".parts") (strLst2Str (divideInfo fileMapList)) where divideInfo fileMapList=case fileMapList of ((fileIndex,fileSize):fm) -> ((show fileIndex) ++" "++ fileName ++ (show fileIndex) ++ " "++(show fileSize) ): (divideInfo fm) [] -> [] strLst2Str :: [String] -> String strLst2Str (x:xs) = x++('/n':(strLst2Str xs)) strLst2Str [] = "" saveIndividualFile :: FilePath -> [(Integer,Integer)] ->BS.ByteString->IO () saveIndividualFile fileName fileMapList content = case fileMapList of ((fileIndex,fileSize):fm) -> do let prefix = BS.take (read(show(fileSize))::Int64) content let postfix = BS.drop (read(show(fileSize))::Int64) content BS.writeFile (fileName ++ (show fileIndex)) prefix saveIndividualFile fileName fm postfix [] -> return () fileSize :: FilePath -> IO Integer fileSize fileName = do fileHandle <- openFile fileName ReadMode size <- hFileSize fileHandle hClose fileHandle return size