importfeed: fix bug while also speeding up 12x!

* Fix bug that could make git-annex importfeed not see recently recorded
  state when configured with annex.alwayscommit=false.
* importfeed: Made "checking known urls" phase run 12 times faster.

The massive speedup is because it no longer queries for metadata
accompanying each url. Instead it processes the whole git-annex branch and
checks all metadata files for feed item ids, and uses any it finds.

This could result in a behavior change, in an unlikely situation: If a feed
id is recorded in a key's metadata, but the url gets removed, the old code
would not see that item id and would re-download it if it finds an url for
it in a feed, while the new code will see the item id. I don't think
the old behavior was intentional, and it may be that the new behavior is
better. Not gonna worry about this.

CHANGELOG

@@ -15,6 +15,9 @@ git-annex (8.20210331) UNRELEASED; urgency=medium
   * Fix bug caused by recent optimisations that could make git-annex not
     see recently recorded status information when configured with
     annex.alwayscommit=false.
+  * Fix bug that could make git-annex importfeed not see recently recorded
+    state when configured with annex.alwayscommit=false.
+  * importfeed: Made "checking known urls" phase run 12 times faster.
 
  -- Joey Hess <id@joeyh.name>  Thu, 01 Apr 2021 12:17:26 -0400
 

Command/ImportFeed.hs

@@ -20,7 +20,6 @@ import Data.Time.Format
 import Data.Time.Calendar
 import Data.Time.LocalTime
 import qualified Data.Text as T
-import Control.Concurrent.Async
 import qualified System.FilePath.ByteString as P
 
 import Command
@@ -45,10 +44,8 @@ import Annex.MetaData
 import Annex.FileMatcher
 import Command.AddUrl (addWorkTree)
 import Annex.UntrustedFilePath
-import qualified Git.Ref
 import qualified Annex.Branch
 import Logs
-import Git.CatFile (catObjectStream)
 
 cmd :: Command
 cmd = notBareRepo $
@@ -125,52 +122,38 @@ getCache opttemplate = ifM (Annex.getState Annex.force)
 	( ret S.empty S.empty
 	, do
 		showStart "importfeed" "checking known urls" (SeekInput [])
-		(is, us) <- unzip <$> knownItems
+		(us, is) <- knownItems
 		showEndOk
-		ret (S.fromList us) (S.fromList (concat is))
+		ret (S.fromList us) (S.fromList is)
 	)
   where
 	tmpl = Utility.Format.gen $ fromMaybe defaultTemplate opttemplate
 	ret us is = return $ Cache us is tmpl
 
-knownItems :: Annex [([ItemId], URLString)]
+{- Scan all url logs and metadata logs in the branch and find urls
-knownItems = do
+ - and ItemIds that are already known. -}
-	g <- Annex.gitRepo
+knownItems :: Annex ([URLString], [ItemId])
-	config <- Annex.getGitConfig
+knownItems = Annex.Branch.overBranchFileContents select (go [] [])
-	catObjectStream g $ \catfeeder catcloser catreader -> do
-		rt <- liftIO $ async $ reader catreader []
-		overKnownUrls (feeder config catfeeder catcloser)
-		liftIO (wait rt)
   where
-	feeder config catfeeder catcloser urlreader = urlreader >>= \case
+	select f
-		Just (k, us) -> do
+		| isUrlLog f = Just ()
-			forM_ us $ \u ->
+		| isMetaDataLog f = Just ()
-				let logf = metaDataLogFile config k
+		| otherwise = Nothing
-				    ref = Git.Ref.branchFileRef Annex.Branch.fullname logf
-				in liftIO $ catfeeder (u, ref)
-			feeder config catfeeder catcloser urlreader
-		Nothing -> liftIO catcloser
 
-	reader catreader c = catreader >>= \case
+	go uc ic reader = reader >>= \case
-		Just (u, Just mdc) ->
+		Just ((), f, Just content)
-			let !itemids = S.toList $ S.filter (/= noneValue) $
+			| isUrlLog f -> case parseUrlLog content of
-				S.map (decodeBS . fromMetaValue) $
+				[] -> go uc ic reader
-					currentMetaDataValues itemIdField $
+				us -> go (us++uc) ic reader
-						parseCurrentMetaData mdc
+			| isMetaDataLog f ->
-			in reader catreader ((itemids,u):c)
+				let s = currentMetaDataValues itemIdField $
-		Just (u, Nothing) -> reader catreader (([],u):c)
+					parseCurrentMetaData content
-		Nothing -> return c
+				in if S.null s
-
+					then go uc ic reader
-overKnownUrls :: (Annex (Maybe (Key, [URLString])) -> Annex a) -> Annex a
+					else go uc (map (decodeBS . fromMetaValue) (S.toList s)++ic) reader
-overKnownUrls a = Annex.Branch.overBranchFileContents urlLogFileKey (a . go)
+			| otherwise -> go uc ic reader
-  where
+		Just ((), _, Nothing) -> go uc ic reader
-	go reader = reader >>= \case
+		Nothing -> return (uc, ic)
-		Just (k, _, Just content) ->
-			case parseUrlLog content of
-				[] -> go reader
-				us -> return (Just (k, us))
-		Just (_, _, Nothing) -> go reader
-		Nothing -> return Nothing
 
 findDownloads :: URLString -> Feed -> [ToDownload]
 findDownloads u f = catMaybes $ map mk (feedItems f)

doc/todo/hiding_a_repository.mdwn

@@ -155,8 +155,8 @@ later write.
 > * [[bugs/git-annex_branch_caching_bug]] was a problem, now fixed.
 > * Any other similar direct accesses of the branch, not going through
 >   Annex.Branch, also need to be fixed (and may be missing journal files
->   already?) Command.ImportFeed.knownItems is one. Command.Log behavior
+>   already?) Most fixed now. Command.Log behavior needs to be
->   needs to be investigated, may be ok.
+>   investigated still.
 > 
 > * Need to implement regardingPrivateUUID and privateUUIDsKnown,
 >   which need to look at the git config to find the private uuids.