git-annex/Remote/Directory/LegacyChunked.hs
Joey Hess dadbb510f6
incremental hashing for fileRetriever
It uses tailVerify to hash the file while it's being written.

This is able to sometimes avoid a separate checksum step. Although
if the file gets written quickly enough, tailVerify may not see it
get created before the write finishes, and the checksum still happens.

Testing with the directory special remote, incremental checksumming did
not happen. But then I disabled the copy CoW probing, and it did work.
What's going on with that is the CoW probe creates an empty file on
failure, then deletes it, and then the file is created again. tailVerify
will open the first, empty file, and so fails to read the content that
gets written to the file that replaces it.

The directory special remote really ought to be able to avoid needing to
use tailVerify, and while other special remotes could do things that
cause similar problems, they probably don't. And if they do, it just
means the checksum doesn't get done incrementally.

Sponsored-by: Dartmouth College's DANDI project
2021-08-13 15:43:29 -04:00

126 lines
4.8 KiB
Haskell

{- Legacy chunksize support for directory special remote.
-
- Can be removed eventually.
-
- Copyright 2011-2012 Joey Hess <id@joeyh.name>
-
- Licensed under the GNU AGPL version 3 or higher.
-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE RankNTypes #-}
module Remote.Directory.LegacyChunked where
import qualified Data.ByteString.Lazy as L
import qualified Data.ByteString as S
import qualified System.FilePath.ByteString as P
import Annex.Common
import Utility.FileMode
import Remote.Helper.Special
import qualified Remote.Helper.Chunked.Legacy as Legacy
import Annex.Tmp
import Utility.Metered
import Utility.Directory.Create
import qualified Utility.RawFilePath as R
withCheckedFiles :: (FilePath -> IO Bool) -> FilePath -> (FilePath -> Key -> [FilePath]) -> Key -> ([FilePath] -> IO Bool) -> IO Bool
withCheckedFiles _ [] _locations _ _ = return False
withCheckedFiles check d locations k a = go $ locations d k
where
go [] = return False
go (f:fs) = do
let chunkcount = f ++ Legacy.chunkCount
ifM (check chunkcount)
( do
chunks <- Legacy.listChunks f <$> readFile chunkcount
ifM (allM check chunks)
( a chunks , return False )
, do
chunks <- Legacy.probeChunks f check
if null chunks
then go fs
else a chunks
)
withStoredFiles :: FilePath -> (FilePath -> Key -> [FilePath]) -> Key -> ([FilePath] -> IO Bool) -> IO Bool
withStoredFiles = withCheckedFiles doesFileExist
{- Splits a ByteString into chunks and writes to dests, obeying configured
- chunk size (not to be confused with the L.ByteString chunk size). -}
storeLegacyChunked :: MeterUpdate -> ChunkSize -> [FilePath] -> L.ByteString -> IO [FilePath]
storeLegacyChunked _ _ [] _ = error "bad storeLegacyChunked call"
storeLegacyChunked meterupdate chunksize alldests@(firstdest:_) b
| L.null b = do
-- always write at least one file, even for empty
L.writeFile firstdest b
return [firstdest]
| otherwise = storeLegacyChunked' meterupdate chunksize alldests (L.toChunks b) []
storeLegacyChunked' :: MeterUpdate -> ChunkSize -> [FilePath] -> [S.ByteString] -> [FilePath] -> IO [FilePath]
storeLegacyChunked' _ _ [] _ _ = error "ran out of dests"
storeLegacyChunked' _ _ _ [] c = return $ reverse c
storeLegacyChunked' meterupdate chunksize (d:dests) bs c = do
bs' <- withFile d WriteMode $
feed zeroBytesProcessed chunksize bs
storeLegacyChunked' meterupdate chunksize dests bs' (d:c)
where
feed _ _ [] _ = return []
feed bytes sz (l:ls) h = do
let len = S.length l
let s = fromIntegral len
if s <= sz || sz == chunksize
then do
S.hPut h l
let bytes' = addBytesProcessed bytes len
meterupdate bytes'
feed bytes' (sz - s) ls h
else return (l:ls)
storeHelper :: FilePath -> (RawFilePath -> RawFilePath -> IO ()) -> Key -> ([FilePath] -> IO [FilePath]) -> FilePath -> FilePath -> IO ()
storeHelper repotop finalizer key storer tmpdir destdir = do
void $ liftIO $ tryIO $ createDirectoryUnder
(toRawFilePath repotop)
(toRawFilePath tmpdir)
Legacy.storeChunks key tmpdir destdir storer recorder (legacyFinalizer finalizer)
where
recorder f s = do
let f' = toRawFilePath f
void $ tryIO $ allowWrite f'
writeFile f s
void $ tryIO $ preventWrite f'
store :: FilePath -> ChunkSize -> (RawFilePath -> RawFilePath -> IO ()) -> Key -> L.ByteString -> MeterUpdate -> FilePath -> FilePath -> IO ()
store repotop chunksize finalizer k b p = storeHelper repotop finalizer k $ \dests ->
storeLegacyChunked p chunksize dests b
{- Need to get a single ByteString containing every chunk.
- Done very innefficiently, by writing to a temp file.
- :/ This is legacy code..
-}
retrieve :: (RawFilePath -> Key -> [RawFilePath]) -> RawFilePath -> Retriever
retrieve locations d basek p miv c = withOtherTmp $ \tmpdir -> do
showLongNote "This remote uses the deprecated chunksize setting. So this will be quite slow."
let tmp = tmpdir P.</> keyFile basek <> ".directorylegacy.tmp"
let tmp' = fromRawFilePath tmp
let go = \k sink -> do
liftIO $ void $ withStoredFiles (fromRawFilePath d) (legacyLocations locations) k $ \fs -> do
forM_ fs $
S.appendFile tmp' <=< S.readFile
return True
b <- liftIO $ L.readFile tmp'
liftIO $ removeWhenExistsWith R.removeLink tmp
sink b
byteRetriever go basek p miv c
checkKey :: RawFilePath -> (RawFilePath -> Key -> [RawFilePath]) -> Key -> Annex Bool
checkKey d locations k = liftIO $
withStoredFiles (fromRawFilePath d) (legacyLocations locations) k $
-- withStoredFiles checked that it exists
const $ return True
legacyFinalizer :: (RawFilePath -> RawFilePath -> IO ()) -> (FilePath -> FilePath -> IO ())
legacyFinalizer f = \a b -> f (toRawFilePath a) (toRawFilePath b)
legacyLocations :: (RawFilePath -> Key -> [RawFilePath]) -> (FilePath -> Key -> [FilePath])
legacyLocations locations = \f k ->
map fromRawFilePath $ locations (toRawFilePath f) k