gvfs: allow corrupt objects to be re-downloaded

As of 9e59b38 (object-file: emit corruption errors when detected,
2022-12-14), Git will loudly complain about corrupt objects.

That is fine, as long as the idea isn't to re-download locally-corrupted
objects. But that's exactly what we want to do in VFS for Git via the
`read-object` hook, as per the `GitCorruptObjectTests` code
added in microsoft/VFSForGit@2db0c030eb25 (New
features: [...] -  GVFS can now recover from corrupted git object files
[...] , 2018-02-16).

So let's support precisely that, and add a regression test that ensures
that re-downloading corrupt objects via the `read-object` hook works.

While at it, avoid the XOR operator to flip the bits, when we actually
want to make sure that they are turned off: Use the AND-NOT operator for
that purpose.

Helped-by: Matthew John Cheetham <mjcheetham@outlook.com>
Helped-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

Author: dscho

commit.c

@@ -566,9 +566,14 @@ int repo_parse_commit_internal(struct repository *r,
 	int flags = OBJECT_INFO_LOOKUP_REPLACE | OBJECT_INFO_SKIP_FETCH_OBJECT |
 		    OBJECT_INFO_DIE_IF_CORRUPT;
 
-	/* But the GVFS Protocol _does_ support missing commits! */
+	/*
+	 * But the GVFS Protocol _does_ support missing commits!
+	 * And the idea with VFS for Git is to re-download corrupted objects,
+	 * not to fail!
+	 */
 	if (gvfs_config_is_set(GVFS_MISSING_OK))
-		flags ^= OBJECT_INFO_SKIP_FETCH_OBJECT;
+		flags &= ~(OBJECT_INFO_SKIP_FETCH_OBJECT |
+			   OBJECT_INFO_DIE_IF_CORRUPT);
 
 	if (!item)
 		return -1;

object-store.c

@@ -7,6 +7,7 @@
 #include "dir.h"
 #include "environment.h"
 #include "gettext.h"
+#include "gvfs.h"
 #include "gvfs-helper-client.h"
 #include "hex.h"
 #include "hook.h"
@@ -1107,6 +1108,9 @@ void *repo_read_object_file(struct repository *r,
 	unsigned flags = OBJECT_INFO_DIE_IF_CORRUPT | OBJECT_INFO_LOOKUP_REPLACE;
 	void *data;
 
+	if (gvfs_config_is_set(GVFS_MISSING_OK))
+		flags &= ~OBJECT_INFO_DIE_IF_CORRUPT;
+
 	oi.typep = type;
 	oi.sizep = size;
 	oi.contentp = &data;

t/t1060-object-corruption.sh

@@ -3,6 +3,7 @@
 test_description='see how we handle various forms of corruption'
 
 . ./test-lib.sh
+. "$TEST_DIRECTORY"/lib-diff-data.sh
 
 # convert "1234abcd" to ".git/objects/12/34abcd"
 obj_to_file() {
@@ -62,6 +63,35 @@ test_expect_success 'streaming a corrupt blob fails' '
 	)
 '
 
+test_expect_success PERL 'truncated objects can be re-retrieved via GVFS' '
+	git init truncated &&
+	COPYING_test_data >truncated/COPYING &&
+	git -C truncated add COPYING &&
+	test_tick &&
+	git -C truncated commit -m initial COPYING &&
+
+	# set up the `read-object` hook so that it overwrites the corrupt object
+	mkdir -p truncated/.git/hooks &&
+	sed -e "1s|/usr/bin/perl|$PERL_PATH|" \
+	  -e "s/system/unlink \".git\/objects\/\" . substr(\$sha1, 0, 2) . \"\/\" . substr(\$sha1, 2); &/" \
+	  <$TEST_DIRECTORY/t0410/read-object \
+	  >truncated/.git/hooks/read-object &&
+	chmod +x truncated/.git/hooks/read-object &&
+
+	# ensure that the parent repository has a copy of the object, from
+	# where the `read-object` can read it
+	sha="$(git hash-object -w truncated/COPYING)" &&
+	file=$(obj_to_file $sha) &&
+	size=$(test_file_size $file) &&
+	chmod u+w truncated/$file &&
+	test-tool truncate truncated/$file $(($size-8)) &&
+
+	rm truncated/COPYING &&
+	test_must_fail git -C truncated reset --hard &&
+	git -C truncated -c core.gvfs=4 -c core.virtualizeObjects \
+		reset --hard
+'
+
 test_expect_success 'getting type of a corrupt blob fails' '
 	(
 		cd bit-error &&