Merge branch 'upstream' into improve_referenced_objects_performance

3 files changed, 60 insertions, 5 deletions

diff --git a/git-fat b/git-fat
index 7eaef13..aa6a45d 100755
--- a/git-fat
+++ b/git-fat
@@ -135,6 +135,13 @@ class GitFat(object):
         self.magiclens = [magiclen(enc) for enc in [self.encode_v1, self.encode_v2]] # All prior versions
     def setup(self):
         mkdir_p(self.objdir)
+    def is_init_done(self):
+        return gitconfig_get('filter.fat.clean') or gitconfig_get('filter.fat.smudge')
+    def assert_init_done(self):
+        if not self.is_init_done():
+            sys.stderr.write('fatal: git-fat is not yet configured in this repository.\n')
+            sys.stderr.write('Run "git fat init" to configure.\n')
+            sys.exit(1)
     def get_rsync(self):
         cfgpath   = os.path.join(self.gitroot,'.gitfat')
         remote    = gitconfig_get('rsync.remote', file=cfgpath)
@@ -388,6 +395,7 @@ class GitFat(object):
             sys.exit(p.returncode)
     def checkout(self, show_orphans=False):
         'Update any stale files in the present working tree'
+        self.assert_init_done()
         for digest, fname in self.orphan_files():
             objpath = os.path.join(self.objdir, digest)
             if os.access(objpath, os.R_OK):
@@ -450,9 +458,27 @@ class GitFat(object):
             fname = os.path.join(self.objdir, obj)
             print('%10d %s' % (os.stat(fname).st_size, obj))
             os.remove(fname)
+
+    def cmd_verify(self):
+        """Print details of git-fat objects with incorrect data hash"""
+        corrupted_objects = []
+        for obj in self.catalog_objects():
+            fname = os.path.join(self.objdir, obj)
+            h = hashlib.new('sha1')
+            for block in readblocks(open(fname)):
+                h.update(block)
+            data_hash = h.hexdigest()
+            if obj != data_hash:
+                corrupted_objects.append((obj, data_hash))
+        if corrupted_objects:
+            print('Corrupted objects: %d' % len(corrupted_objects))
+            for obj, data_hash in corrupted_objects:
+                print('%s data hash is %s' % (obj, data_hash))
+            sys.exit(1)
+
     def cmd_init(self):
         self.setup()
-        if gitconfig_get('filter.fat.clean') or gitconfig_get('filter.fat.smudge'):
+        if self.is_init_done():
             print('Git fat already configured, check configuration in .git/config')
         else:
             gitconfig_set('filter.fat.clean', 'git-fat filter-clean')
@@ -575,6 +601,8 @@ if __name__ == '__main__':
         fat.cmd_pull(sys.argv[2:])
     elif cmd == 'gc':
         fat.cmd_gc()
+    elif cmd == 'verify':
+        fat.cmd_verify()
     elif cmd == 'checkout':
         fat.cmd_checkout(sys.argv[2:])
     elif cmd == 'find':
@@ -582,4 +610,4 @@ if __name__ == '__main__':
     elif cmd == 'index-filter':
         fat.cmd_index_filter(sys.argv[2:])
     else:
-        print('Usage: git fat [init|status|push|pull|gc|checkout|find|index-filter]', file=sys.stderr)
+        print('Usage: git fat [init|status|push|pull|gc|verify|checkout|find|index-filter]', file=sys.stderr)
diff --git a/test-retroactive.sh b/test-retroactive.sh
index dd02367..51a38ec 100755
--- a/test-retroactive.sh
+++ b/test-retroactive.sh
@@ -1,8 +1,11 @@
 #!/bin/sh -ex
 
+fullpath() { echo "`pwd`/$1"; }
+
 git init retro
 cd retro
 cp /usr/share/dict/words words.big
+chmod u+w words.big
 git add words.big
 git commit -m'Add big file without using git-fat'
 sort words.big > sorted.big
@@ -25,7 +28,7 @@ git commit -am'Import big files into git-fat'
 git log --stat
 
 git fat find 10000 | awk '{print $1}' > fat-files
-git filter-branch --index-filter "git fat index-filter $(realpath fat-files) --manage-gitattributes" --tag-name-filter cat -- --all
+git filter-branch --index-filter "git fat index-filter $(fullpath fat-files) --manage-gitattributes" --tag-name-filter cat -- --all
 
 git log --stat
 git checkout HEAD^
@@ -37,14 +40,14 @@ ls -al
 git checkout master
 cat > .gitfat <<EOF
 [rsync]
-remote = $(realpath ../retro-store)
+remote = $(fullpath ../retro-store)
 EOF
 git add .gitfat
 git commit -m'Add .gitfat for local push'
 git fat push
 
 cd ..
-git clone file:///$(realpath retro) retro-clone
+git clone file:///$(fullpath retro) retro-clone
 cd retro-clone
 git fat init
 git fat pull
diff --git a/test.sh b/test.sh
index e9c9163..0ee63ea 100755
--- a/test.sh
+++ b/test.sh
@@ -2,6 +2,9 @@
 # Any copyright is dedicated to the Public Domain.
 # http://creativecommons.org/publicdomain/zero/1.0/
 
+# Clear out repos and fat store from prior test runs
+rm -fR fat-test fat-test2 /tmp/fat-store
+
 git init fat-test
 cd fat-test
 git fat init
@@ -29,6 +32,19 @@ git fat push
 cd ..
 git clone fat-test fat-test2
 cd fat-test2
+# checkout and pull should fail in repo not yet init'ed for git-fat
+git fat checkout && true
+if [ $? -eq 0 ]
+then
+    echo 'ERROR: "git fat checkout" in uninitialised repo should fail'
+    exit 1
+fi
+git fat pull -- 'a.fa*' && true
+if [ $? -eq 0 ]
+then
+    echo 'ERROR: "git fat pull" in uninitialised repo should fail'
+    exit 1
+fi
 git fat init
 git fat pull -- 'a.fa*'
 cat a.fat
@@ -38,3 +54,11 @@ git commit -m'add d with normal content'
 rm d
 git fat pull
 
+# Check verify command finds corrupt object
+mv .git/fat/objects/6ecec2e21d3033e7ba53e2db63f69dbd3a011fa8 \
+   .git/fat/objects/6ecec2e21d3033e7ba53e2db63f69dbd3a011fa8.bak
+echo "Not the right data" > .git/fat/objects/6ecec2e21d3033e7ba53e2db63f69dbd3a011fa8
+git fat verify && true
+if [ $? -eq 0 ]; then echo "Verify did not detect invalid object"; exit 1; fi
+mv .git/fat/objects/6ecec2e21d3033e7ba53e2db63f69dbd3a011fa8.bak \
+   .git/fat/objects/6ecec2e21d3033e7ba53e2db63f69dbd3a011fa8