handle multiple VURL checksums in one pass

git-annex fsck and some other commands that verify the content of a key
were using the non-incremental verification interface. But for VURL
urls, that interface is innefficient because when there are multiple
equivilant keys, it has to separately read and checksum for each key in
turn until one matches. It's more efficient for those to use the
incremental interface, since the file can be read a single time.

There's no real downside to using the incremental interface when available.

Note that more speedup could be had for VURL, if it was able to
calculate the checksum a single time and then compare with the
equivilant keys checksums. When the equivilant keys use the same type of
checksum.

Sponsored-by: k0ld on Patreon

Annex/Verify.hs

@@ -12,6 +12,7 @@ module Annex.Verify (
 	shouldVerify,
 	verifyKeyContentPostRetrieval,
 	verifyKeyContent,
+	verifyKeyContent',
 	Verification(..),
 	unVerified,
 	warnUnverifiableInsecure,
@@ -97,16 +98,32 @@ verifyKeyContentPostRetrieval rsp v verification k f = case (rsp, verification)
 				resumeVerifyKeyContent k f iv
 			_ -> verifyKeyContent k f
 
+-- When possible, does an incremental verification, because that can be
+-- faster. Eg, the VURL backend can need to try multiple checksums and only
+-- with an incremental verification does it avoid reading files twice.
 verifyKeyContent :: Key -> RawFilePath -> Annex Bool
 verifyKeyContent k f = verifyKeySize k f <&&> verifyKeyContent' k f
 
+-- Does not verify size.
 verifyKeyContent' :: Key -> RawFilePath -> Annex Bool
 verifyKeyContent' k f = 
 	Backend.maybeLookupBackendVariety (fromKey keyVariety k) >>= \case
 		Nothing -> return True
-		Just b -> case Types.Backend.verifyKeyContent b of
-			Nothing -> return True
-			Just verifier -> verifier k f
+		Just b -> case (Types.Backend.verifyKeyContentIncrementally b, Types.Backend.verifyKeyContent b) of
+			(Nothing, Nothing) -> return True
+			(Just mkiv, mverifier) -> do
+				iv <- mkiv k
+				showAction (UnquotedString (descIncrementalVerifier iv))
+				res <- liftIO $ catchDefaultIO Nothing $
+					withBinaryFile (fromRawFilePath f) ReadMode $ \h -> do
+						feedIncrementalVerifier h iv
+						finalizeIncrementalVerifier iv
+				case res of
+					Just res' -> return res'
+					Nothing -> case mverifier of
+						Nothing -> return True
+						Just verifier -> verifier k f
+			(Nothing, Just verifier) -> verifier k f
 
 resumeVerifyKeyContent :: Key -> RawFilePath -> IncrementalVerifier -> Annex Bool
 resumeVerifyKeyContent k f iv = liftIO (positionIncrementalVerifier iv) >>= \case
@@ -132,17 +149,18 @@ resumeVerifyKeyContent k f iv = liftIO (positionIncrementalVerifier iv) >>= \cas
 			liftIO $ catchDefaultIO (Just False) $
 				withBinaryFile (fromRawFilePath f) ReadMode $ \h -> do
 					hSeek h AbsoluteSeek endpos
-					feedincremental h
+					feedIncrementalVerifier h iv
 					finalizeIncrementalVerifier iv
 	
-	feedincremental h = do
-		b <- S.hGetSome h chunk
-		if S.null b
-			then return ()
-			else do
-				updateIncrementalVerifier iv b
-				feedincremental h
-
+feedIncrementalVerifier :: Handle -> IncrementalVerifier -> IO ()
+feedIncrementalVerifier h iv = do
+	b <- S.hGetSome h chunk
+	if S.null b
+		then return ()
+		else do
+			updateIncrementalVerifier iv b
+			feedIncrementalVerifier h iv
+  where
 	chunk = 65536
 
 verifyKeySize :: Key -> RawFilePath -> Annex Bool

Command/Fsck.hs

@@ -1,6 +1,6 @@
 {- git-annex command
  -
- - Copyright 2010-2022 Joey Hess <id@joeyh.name>
+ - Copyright 2010-2023 Joey Hess <id@joeyh.name>
  -
  - Licensed under the GNU AGPL version 3 or higher.
  -}
@@ -16,6 +16,7 @@ import qualified Remote
 import qualified Types.Backend
 import qualified Backend
 import Annex.Content
+import Annex.Verify
 #ifndef mingw32_HOST_OS
 import Annex.Version
 import Annex.Content.Presence
@@ -524,17 +525,14 @@ checkBackendRemote backend key remote ai localcopy =
 	checkBackendOr (badContentRemote remote localcopy) backend key localcopy ai
 
 checkBackendOr :: (Key -> Annex String) -> Backend -> Key -> RawFilePath -> ActionItem -> Annex Bool
-checkBackendOr bad backend key file ai =
-	case Types.Backend.verifyKeyContent backend of
-		Just verifier -> do
-			ok <- verifier key file
-			unless ok $ do
-				msg <- bad key
-				warning $ actionItemDesc ai
-					<> ": Bad file content; "
-					<> UnquotedString msg
-			return ok
-		Nothing -> return True
+checkBackendOr bad backend key file ai = do
+	ok <- verifyKeyContent' key file
+	unless ok $ do
+		msg <- bad key
+		warning $ actionItemDesc ai
+			<> ": Bad file content; "
+			<> UnquotedString msg
+	return ok
 
 {- Check, if there are InodeCaches recorded for a key, that one of them
  - matches the object file. There are situations where the InodeCache