Bug: git archive not doing export-subst when using --prefix= option

Bug: git archive not doing export-subst when using --prefix= option

[Ulrik Sverdrup]

In one of my small projects, I just tried to add an export-subst format to
my version file to have the git version in the released tarball. However it
seems git-archive --prefix="tarballname-version/" does not work well
together with export-subst.

I have very old git! (v1.5.4.4). So please help me and try to reproduce
this with newer versions:

The repo is publicly available at

git clone git://repo.or.cz/dragbox.git

The file in the repo having an export substitution format is
Dragbox/version.py.in

Now try to export the git tree with git-archive, with and without --prefix=
We will do some hackery and grep the binary output.

	$ git archive HEAD | grep --text "git_version"
	git_version = "9c3f6c0af7e67a354cac9a24e24e2057d17778ef"
	$ git archive --prefix="dragbox-git/" HEAD | grep --text "git_version"
	git_version = "$Format:%H$"


I noticed in another even smaller repository that it would not behave the
same but equally spurious; it would only perform the substitution for some
prefixes and not for others (for example with or without ending slash)

Regards
Ulrik Sverdrup

[PATCH] git-archive: ignore prefix when checking file attribute

[René Scharfe]

Ulrik Sverdrup noticed that git-archive doesn't correctly apply the attribute
export-subst when the option --prefix is given, too.

When it checked if a file has the attribute turned on, git-archive would try
to look up the full path -- including the prefix -- in .gitattributes.  That's
wrong, as the prefix doesn't need to have any relation to any existing
directories, tracked or not.

This patch makes git-archive ignore the prefix when looking up if value of the
attribute export-subst for a file.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
---
Ulrik, does this patch fix the problem for you?  It applies to v1.5.4.4 and
v1.5.5 equally.

 archive-tar.c       |    6 ++++--
 archive-zip.c       |    6 ++++--
 t/t5000-tar-tree.sh |   15 ++++++++++++++-
 3 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/archive-tar.c b/archive-tar.c
index 30aa2e2..4add802 100644
--- a/archive-tar.c
+++ b/archive-tar.c
@@ -17,6 +17,7 @@ static time_t archive_time;
 static int tar_umask = 002;
 static int verbose;
 static const struct commit *commit;
+static size_t base_len;
 
 /* writes out the whole block, but only if it is full */
 static void write_if_needed(void)
@@ -251,8 +252,8 @@ static int write_tar_entry(const unsigned char *sha1,
 		buffer = NULL;
 		size = 0;
 	} else {
-		buffer = sha1_file_to_archive(path.buf, sha1, mode, &type,
-		                              &size, commit);
+		buffer = sha1_file_to_archive(path.buf + base_len, sha1, mode,
+				&type, &size, commit);
 		if (!buffer)
 			die("cannot read %s", sha1_to_hex(sha1));
 	}
@@ -272,6 +273,7 @@ int write_tar_archive(struct archiver_args *args)
 	archive_time = args->time;
 	verbose = args->verbose;
 	commit = args->commit;
+	base_len = args->base ? strlen(args->base) : 0;
 
 	if (args->commit_sha1)
 		write_global_extended_header(args->commit_sha1);
diff --git a/archive-zip.c b/archive-zip.c
index 74e30f6..18c0f87 100644
--- a/archive-zip.c
+++ b/archive-zip.c
@@ -13,6 +13,7 @@ static int verbose;
 static int zip_date;
 static int zip_time;
 static const struct commit *commit;
+static size_t base_len;
 
 static unsigned char *zip_dir;
 static unsigned int zip_dir_size;
@@ -197,8 +198,8 @@ static int write_zip_entry(const unsigned char *sha1,
 		if (S_ISREG(mode) && zlib_compression_level != 0)
 			method = 8;
 		result = 0;
-		buffer = sha1_file_to_archive(path, sha1, mode, &type, &size,
-		                              commit);
+		buffer = sha1_file_to_archive(path + base_len, sha1, mode,
+				&type, &size, commit);
 		if (!buffer)
 			die("cannot read %s", sha1_to_hex(sha1));
 		crc = crc32(crc, buffer, size);
@@ -321,6 +322,7 @@ int write_zip_archive(struct archiver_args *args)
 	zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
 	verbose = args->verbose;
 	commit = args->commit;
+	base_len = args->base ? strlen(args->base) : 0;
 
 	if (args->base && plen > 0 && args->base[plen - 1] == '/') {
 		char *base = xstrdup(args->base);
diff --git a/t/t5000-tar-tree.sh b/t/t5000-tar-tree.sh
index dca2067..fa62b6a 100755
--- a/t/t5000-tar-tree.sh
+++ b/t/t5000-tar-tree.sh
@@ -109,9 +109,10 @@ test_expect_success \
     'diff -r a c/prefix/a'
 
 test_expect_success \
-    'create an archive with a substfiles' \
+    'create archives with substfiles' \
     'echo "substfile?" export-subst >a/.gitattributes &&
      git archive HEAD >f.tar &&
+     git archive --prefix=prefix/ HEAD >g.tar &&
      rm a/.gitattributes'
 
 test_expect_success \
@@ -127,6 +128,18 @@ test_expect_success \
 '
 
 test_expect_success \
+    'extract substfiles from archive with prefix' \
+    '(mkdir g && cd g && $TAR xf -) <g.tar'
+
+test_expect_success \
+     'validate substfile contents from archive with prefix' \
+     'git log --max-count=1 "--pretty=format:A${SUBSTFORMAT}O" HEAD \
+      >g/prefix/a/substfile1.expected &&
+      diff g/prefix/a/substfile1.expected g/prefix/a/substfile1 &&
+      diff a/substfile2 g/prefix/a/substfile2
+'
+
+test_expect_success \
     'git archive --format=zip' \
     'git archive --format=zip HEAD >d.zip'
 

Re: [PATCH] git-archive: ignore prefix when checking file attribute

[Ulrik Sverdrup]

2008/4/9, René Scharfe <rene.scharfe@lsrfire.ath.cx>:
> Ulrik Sverdrup noticed that git-archive doesn't correctly apply the attribute
>  export-subst when the option --prefix is given, too.
>
>  When it checked if a file has the attribute turned on, git-archive would try
>  to look up the full path -- including the prefix -- in .gitattributes.  That's
>  wrong, as the prefix doesn't need to have any relation to any existing
>  directories, tracked or not.
>
>  This patch makes git-archive ignore the prefix when looking up if value of the
>  attribute export-subst for a file.
>
>  Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
>  ---
>  Ulrik, does this patch fix the problem for you?  It applies to v1.5.4.4 and
>  v1.5.5 equally.


Thanks for the patch René!

First I reproduced my problem on vanilla git v1.5.5, and it is still
there as in v.1.5.4.4.
I applied this patch and that fixed my issue, testing both tar and zip.

Ulrik Sverdrup

Re: [PATCH] git-archive: ignore prefix when checking file attribute

[Junio C Hamano]

Thanks.  Will apply as a fix.

In the longer term, we probably should teach attributes.c to optionally
read from a tree (introduce git_attribute_set_tree() call upfront, or
something like that, but I haven't thought through the details yet), so
that archive can ignore what happens to be checked out.