pre-commit hook

6 files changed, 55 insertions, 15 deletions

diff --git a/Commands.hs b/Commands.hs
index 015a2d822..b6bc6dfc2 100644
--- a/Commands.hs
+++ b/Commands.hs
@@ -68,7 +68,7 @@ doSubCmd cmdname start param = do
 {- A subcommand can broadly want one of several kinds of input parameters.
  - This allows a first stage of filtering before starting a subcommand. -}
 data SubCmdWants = FilesInGit | FilesNotInGit | FilesMissing
-	| Description | Keys | Tempfile
+	| Description | Keys | Tempfile | FilesToBeCommitted
 
 data SubCommand = Command {
 	subcmdname :: String,
@@ -91,7 +91,9 @@ subCmds =  [
 	, (Command "unannex"	unannexStart	FilesInGit
 		"undo accidential add command")
 	, (Command "fix"	fixStart	FilesInGit
-		"fix up files' symlinks to point to annexed content")
+		"fix up symlinks to point to annexed content")
+	, (Command "pre-commit"	fixStart	FilesToBeCommitted
+		"fix up symlinks before they are committed")
 	, (Command "fromkey"	fromKeyStart	FilesMissing
 		"adds a file using a specific key")
 	, (Command "dropkey"	dropKeyStart	Keys
@@ -130,7 +132,7 @@ usage = usageInfo header options ++ "\nSubcommands:\n" ++ cmddescs
 		cmddescs = unlines $ map (\c -> indent $ showcmd c) subCmds
 		showcmd c =
 			(subcmdname c) ++
-			(pad 10 (subcmdname c)) ++
+			(pad 11 (subcmdname c)) ++
 			(descWanted (subcmdwants c)) ++
 			(pad 13 (descWanted (subcmdwants c))) ++
 			(subcmddesc c)
@@ -161,6 +163,17 @@ findWanted FilesMissing params repo = do
 			if (e) then return False else return True
 findWanted Description params _ = do
 	return $ [unwords params]
+findWanted FilesToBeCommitted params repo = do
+	files <- mapM gitcached params
+	return $ foldl (++) [] files
+	where
+		gitcached p = do
+			-- ask git for files staged for commit that
+			-- are being added, moved, or changed (but not deleted)
+			fs0 <- Git.pipeRead repo ["diff", "--cached", 
+				"--name-only", "--diff-filter=ACMRT",
+				"-z", "HEAD", p]
+			return $ filter (not . null) $ split "\0" fs0
 findWanted _ params _ = return params
 
 {- Parses command line and returns two lists of actions to be 
@@ -360,6 +373,7 @@ initPerform description = do
 	u <- getUUID g
 	describeUUID u description
 	liftIO $ gitAttributes g
+	liftIO $ gitPreCommitHook g
 	return $ Just $ initCleanup
 initCleanup :: Annex Bool
 initCleanup = do
diff --git a/Core.hs b/Core.hs
index 80bf56cc4..254bcec51 100644
--- a/Core.hs
+++ b/Core.hs
@@ -73,6 +73,21 @@ gitAttributes repo = do
 			Git.run repo ["commit", "-m", "git-annex setup", 
 					attributes]
 
+{- set up a git pre-commit hook, if one is not already present -}
+gitPreCommitHook :: Git.Repo -> IO ()
+gitPreCommitHook repo = do
+	let hook = (Git.workTree repo) ++ "/" ++ (Git.dir repo) ++
+		"/hooks/pre-commit"
+	exists <- doesFileExist hook
+	if (exists)
+		then putStrLn $ "pre-commit hook (" ++ hook ++ ") already exists, not configuring"
+		else do
+			writeFile hook $ "#!/bin/sh\n" ++
+				"# automatically configured by git-annex\n" ++ 
+				"git annex pre-commit .\n"
+			p <- getPermissions hook
+			setPermissions hook $ p {executable = True}
+
 {- Checks if a given key is currently present in the annexLocation.
  -
  - This can be run against a remote repository to check the key there. -}
diff --git a/debian/changelog b/debian/changelog
index fd1fddafd..d0922c1c5 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -4,6 +4,9 @@ git-annex (0.02) UNRELEASED; urgency=low
     file content when dropping files.
   * New move subcommand, that can move files from or to a remote.
   * New fromkey subcommand, for registering urls, etc.
+  * git-annex init will now set up a pre-commit hook that fixes up symlinks
+    before they are committed, to ensure that moving symlinks around does not
+    break them.
   * Add remote.annex-ignore git config setting to allow completly disabling
     a given remote.
   * --from/--to can be used to control the remote repository that git-annex
diff --git a/doc/git-annex.mdwn b/doc/git-annex.mdwn
index 417ff7e58..f7bb64988 100644
--- a/doc/git-annex.mdwn
+++ b/doc/git-annex.mdwn
@@ -107,9 +107,13 @@ Many git-annex subcommands will stage changes for later `git commit` by you.
   Fixes up symlinks that have become broken to again point to annexed content.
   This is useful to run if you have been moving the symlinks around.
 
-  You do not normally need to run this by hand since `git-annex init`
-  installs a pre-commit hook that automatically fixes up symlinks when
-  they are committed.
+* pre-commit [path ...]
+
+  Fixes up symlinks that are staged as part of a commit, to ensure they
+  point to annexed content. 
+
+  This is meant to be called from git's pre-commit hook. `git annex init`
+  automatically creates a pre-commit hook using this.
 
 * fromkey file
 
@@ -202,7 +206,7 @@ decscriptions. You may edit it.
 `.git-annex/*.log` is where git-annex records its content tracking
 information. These files should be committed to git.
 
-`.git-annex/.gitattributes` is configured to use git's union merge driver
+`.gitattributes` is configured to use git's union merge driver
 to avoid conflicts when merging files in the `.git-annex` directory.
 
 # AUTHOR
diff --git a/doc/todo/symlink_farming_commit_hook.mdwn b/doc/todo/symlink_farming_commit_hook.mdwn
index af03beb70..3e93cb34b 100644
--- a/doc/todo/symlink_farming_commit_hook.mdwn
+++ b/doc/todo/symlink_farming_commit_hook.mdwn
@@ -10,3 +10,5 @@ up.
 back to git-annex. If you want to have your own shell script in the post-commit
 hook, just make it call `git annex` with no parameters. git-annex will detect
 when it's run from a git hook and do the necessary fixups.
+
+[[done]]
diff --git a/doc/walkthrough.mdwn b/doc/walkthrough.mdwn
index ce68a77f7..b49a00f19 100644
--- a/doc/walkthrough.mdwn
+++ b/doc/walkthrough.mdwn
@@ -43,23 +43,25 @@ if you are using a centralized bare repository.
 	add debian.iso ok
 	# git commit -a -m added
 
-Notice you commit at the end, this checks in git-annex's record of the
-files but not their actual, large, content.
+When you add a file to the annex and commit it, only a symlink to
+the annexed content is committed. The content itself is stored in
+git-annex's backend.
 
 ## renaming files
 
 	# cd ~/annex
 	# git mv big_file my_cool_big_file
 	# mkdir iso
-	# git mv debian.iso iso
-	# git annex fix .
-	fix iso/debian.iso ok
+	# git mv debian.iso iso/
 	# git commit -m moved
 
 You can use any normal git operations to move files around, or even
-make copies or delete them. `git-annex fix` needs to be run if a file
-is moved into a different directory, in order to fix up the symlink 
-pointing to the file's content.
+make copies or delete them.
+
+Notice that, since annexed files are represented by symlinks,
+the symlink will break when the file is moved into a subdirectory.
+But, git-annex will fix this up for you when you commit --
+it has a pre-commit hook that watches for and corrects broken symlinks.
 
 ## getting file content