首先 - 道歉。这是我编译的第一个 Haskell 代码。我正在直接从 Real World Haskell 第 24 章编译一些代码。该代码使用在另一个源文件中实现的 MapReduce 引擎来计算一行中的单词数。这是代码：
module Main where
import Control.Monad (forM_)
import Data.Int (Int64)
import qualified Data.ByteString.Lazy.Char8 as LB
import System.Environment (getArgs)
import LineChunks (chunkedReadWith)
import MapReduce (mapReduce, rnf)
lineCount :: [LB.ByteString] -> Int64
lineCount = mapReduce rdeepseq (LB.count '\n')
rdeepseq sum
main :: IO ()
main = do
args <- getArgs
forM_ args $ \path -> do
numLines <- chunkedReadWith lineCount path
putStrLn $ path ++ ": " ++ show numLines
这段代码编译得很好，我得到了一个 LineCount.exe。
现在，我应该如何实际使用它来计算文件中的行数？我有一个文件“测试”，其中包含一些测试文本。但是当我这样做时：
LineCount test
在命令行上，我得到：
Exception: test: hGetBufSome: illegal operation (handle is closed)
有什么问题？
这是另一个文件中的更多代码：
module LineChunks
(
chunkedReadWith
) where
import Control.Exception (bracket, finally)
import Control.Monad (forM, liftM)
import Control.Parallel.Strategies (NFData, rdeepseq)
import Data.Int (Int64)
import qualified Data.ByteString.Lazy.Char8 as LB
import GHC.Conc (numCapabilities)
import System.IO
data ChunkSpec = CS {
chunkOffset :: !Int64
, chunkLength :: !Int64
} deriving (Eq, Show)
withChunks :: (NFData a) =>
(FilePath -> IO [ChunkSpec])
-> ([LB.ByteString] -> a)
-> FilePath
-> IO a
withChunks chunkFunc process path = do
(chunks, handles) <- chunkedRead chunkFunc path
let r = process chunks
(rdeepseq r `seq` return r) `finally` mapM_ hClose handles
chunkedReadWith :: (NFData a) =>
([LB.ByteString] -> a) -> FilePath -> IO a
chunkedReadWith func path =
withChunks (lineChunks (numCapabilities * 4)) func path
{-- /snippet withChunks --}
{-- snippet chunkedRead --}
chunkedRead :: (FilePath -> IO [ChunkSpec])
-> FilePath
-> IO ([LB.ByteString], [Handle])
chunkedRead chunkFunc path = do
chunks <- chunkFunc path
liftM unzip . forM chunks $ \spec -> do
h <- openFile path ReadMode
hSeek h AbsoluteSeek (fromIntegral (chunkOffset spec))
chunk <- LB.take (chunkLength spec) `liftM` LB.hGetContents h
return (chunk, h)
{-- /snippet chunkedRead --}
{-- snippet lineChunks --}
lineChunks :: Int -> FilePath -> IO [ChunkSpec]
lineChunks numChunks path = do
bracket (openFile path ReadMode) hClose $ \h -> do
totalSize <- fromIntegral `liftM` hFileSize h
let chunkSize = totalSize `div` fromIntegral numChunks
findChunks offset = do
let newOffset = offset + chunkSize
hSeek h AbsoluteSeek (fromIntegral newOffset)
let findNewline off = do
eof <- hIsEOF h
if eof
then return [CS offset (totalSize - offset)]
else do
bytes <- LB.hGet h 4096
case LB.elemIndex '\n' bytes of
Just n -> do
chunks@(c:_) <- findChunks (off + n + 1)
let coff = chunkOffset c
return (CS offset (coff - offset):chunks)
Nothing -> findNewline (off + LB.length bytes)
findNewline newOffset
findChunks 0
{-- /snippet lineChunks --}
-- Ensure that a series of ChunkSpecs is contiguous and
-- non-overlapping.
prop_contig (CS o l:cs@(CS o' _:_)) | o + l == o' = prop_contig cs
| otherwise = False
prop_contig _ = True