handle keys with extensions consistently in all locales

Fix some cases where handling of keys with extensions varied depending on
the locale.

A filename with a unicode extension would before generate a key with an
extension in a unicode locale, but not in LANG=C, because the extension
was not all alphanumeric. Also the the length of the extension could be
counted differently depending on the locale.

In a non-unicode locale, git-annex migrate would see that the extension
was not all alphanumeric and want to "upgrade" it. Now that doesn't happen.

As far as backwards compatability, this does mean that unicode
extensions are counted by the number of bytes, not number of characters.
So, if someone is using unicode extensions, they may find git-annex
stops using them when adding files, because their extensions are too
long. Keys already in their repo with the "too long" extensions will
still work though, so this only prevents adding the same content with
the same extension generating the same key. Documented this by
documenting that annex.maxextensionlength is a number of bytes.

Also, if a filename has an extension that is not valid utf-8 and the
locale is utf-8, the extension will be allowed now, and an old
git-annex, in the same locale would not, and would also want to
"upgrade" that.

Backend/Hash.hs

@@ -1,6 +1,6 @@
 {- git-annex hashing backends
  -
- - Copyright 2011-2019 Joey Hess <id@joeyh.name>
+ - Copyright 2011-2020 Joey Hess <id@joeyh.name>
  -
  - Licensed under the GNU AGPL version 3 or higher.
  -}
@@ -24,7 +24,9 @@ import Utility.Metered
 import qualified Data.ByteString as S
 import qualified Data.ByteString.Char8 as S8
 import qualified Data.ByteString.Lazy as L
+import qualified System.FilePath.ByteString as P
 import Data.Char
+import Data.Word
 import Control.DeepSeq
 import Control.Exception (evaluate)
 
@@ -104,22 +106,22 @@ keyValueE hash source meterupdate =
   where
 	addE k = do
 		maxlen <- annexMaxExtensionLength <$> Annex.getGitConfig
-		let ext = selectExtension maxlen (keyFilename source)
+		let ext = selectExtension maxlen (toRawFilePath (keyFilename source))
 		return $ Just $ alterKey k $ \d -> d
-			{ keyName = keyName d <> encodeBS ext
+			{ keyName = keyName d <> ext
 			, keyVariety = hashKeyVariety hash (HasExt True)
 			}
 
-selectExtension :: Maybe Int -> FilePath -> String
+selectExtension :: Maybe Int -> RawFilePath -> S.ByteString
 selectExtension maxlen f
 	| null es = ""
-	| otherwise = intercalate "." ("":es)
+	| otherwise = S.intercalate "." ("":es)
   where
-	es = filter (not . null) $ reverse $
-		take 2 $ filter (all validInExtension) $
+	es = filter (not . S.null) $ reverse $
+		take 2 $ filter (S.all validInExtension) $
 		takeWhile shortenough $
-		reverse $ splitc '.' $ takeExtensions f
-	shortenough e = length e <= fromMaybe maxExtensionLen maxlen
+		reverse $ S.split (fromIntegral (ord '.')) (P.takeExtensions f)
+	shortenough e = S.length e <= fromMaybe maxExtensionLen maxlen
 
 maxExtensionLen :: Int
 maxExtensionLen = 4 -- long enough for "jpeg"
@@ -152,11 +154,13 @@ checkKeyChecksum hash key file = catchIOErrorType HardwareFault hwfault $ do
 keyHash :: Key -> S.ByteString
 keyHash = fst . splitKeyNameExtension
 
-validInExtension :: Char -> Bool
+validInExtension :: Word8 -> Bool
 validInExtension c
-	| isAlphaNum c = True
-	| c == '.' = True
-	| otherwise = False
+	| c >= 48 && c <= 57 = True -- numbers
+	| c >= 65 && c <= 90 = True -- A-Z
+	| c >= 97 && c <= 122 = True -- a-z
+	| c <= 127 = False -- other ascii, spaces, punctuation, control chars
+	| otherwise = True -- utf8 is allowed, also other encodings
 
 {- Upgrade keys that have the \ prefix on their hash due to a bug, or
  - that contain non-alphanumeric characters in their extension.
@@ -168,7 +172,7 @@ validInExtension c
 needsUpgrade :: Key -> Bool
 needsUpgrade key = or
 	[ "\\" `S8.isPrefixOf` keyHash key
-	, any (not . validInExtension) (decodeBS $ snd $ splitKeyNameExtension key)
+	, S.any (not . validInExtension) (snd $ splitKeyNameExtension key)
 	, not (hasExt (fromKey keyVariety key)) && keyHash key /= fromKey keyName key
 	]
 
@@ -188,7 +192,7 @@ trivialMigrate' oldkey newbackend afile maxextlen
 		AssociatedFile Nothing -> Nothing
 		AssociatedFile (Just file) -> Just $ alterKey oldkey $ \d -> d
 			{ keyName = keyHash oldkey 
-				<> encodeBS' (selectExtension maxextlen (fromRawFilePath file))
+				<> selectExtension maxextlen file
 			, keyVariety = newvariety
 			}
 	{- Upgrade to fix bad previous migration that created a

CHANGELOG

@@ -25,6 +25,10 @@ git-annex (8.20200221) UNRELEASED; urgency=medium
     the remote cannot be fetched from by git, so git fetch --all
     will not try to use it.
   * Makefile: Support newer versions of cabal that use the new-build system.
+  * Fix some cases where handling of keys with extensions varied depending
+    on the locale.
+  * annex.maxextensionlength used to be the number of characters, not
+    bytes, when in a utf-8 locale. It's now always the number of bytes.
 
  -- Joey Hess <id@joeyh.name>  Wed, 19 Feb 2020 12:48:58 -0400
 

doc/git-annex.mdwn

@@ -871,11 +871,11 @@ Like other git commands, git-annex is configured via `.git/config`.
 
 * `annex.maxextensionlength`
 
-  Maximum length of what is considered a filename extension when adding a
-  file to a backend that preserves filename extensions. The default length
-  is 4, which allows extensions like "jpeg". The dot before the extension
-  is not counted part of its length.  At most two extensions at the end of
-  a filename will be preserved, e.g. .gz or .tar.gz .
+  Maximum length, in bytes, of what is considered a filename extension when
+  adding a file to a backend that preserves filename extensions. The
+  default length is 4, which allows extensions like "jpeg". The dot before
+  the extension is not counted part of its length. At most two extensions
+  at the end of a filename will be preserved, e.g. .gz or .tar.gz .
 
 * `annex.diskreserve`