Improve robustness of diff rename detection

Under some strange circumstances, diffs can end up listing files
that we can't actually open successfully.  Instead of aborting
the git_diff_find_similar, this makes it so that those files just
won't be considered as valid rename/copy targets instead.

1 files changed, 12 insertions, 5 deletions

diff --git a/src/diff_tform.c b/src/diff_tform.c
index 201a0e896..de2d5f616 100644
--- a/src/diff_tform.c
+++ b/src/diff_tform.c
@@ -386,8 +386,12 @@ static int similarity_calc(
 
 		/* TODO: apply wd-to-odb filters to file data if necessary */
 
-		if (!(error = git_buf_joinpath(
-				&path, git_repository_workdir(diff->repo), file->path)))
+		if ((error = git_buf_joinpath(
+				 &path, git_repository_workdir(diff->repo), file->path)) < 0)
+			return error;
+
+		/* if path is not a regular file, just skip this item */
+		if (git_path_isfile(path.ptr))
 			error = opts->metric->file_signature(
 				&cache[file_idx], file, path.ptr, opts->metric->payload);
 
@@ -398,8 +402,11 @@ static int similarity_calc(
 
 		/* TODO: add max size threshold a la diff? */
 
-		if ((error = git_blob_lookup(&blob, diff->repo, &file->oid)) < 0)
-			return error;
+		if (git_blob_lookup(&blob, diff->repo, &file->oid) < 0) {
+			/* if lookup fails, just skip this item in similarity calc */
+			giterr_clear();
+			return 0;
+		}
 
 		blobsize = git_blob_rawsize(blob);
 		if (!git__is_sizet(blobsize)) /* ? what to do ? */
@@ -437,7 +444,7 @@ static int similarity_measure(
 		return -1;
 	if (!cache[b_idx] && similarity_calc(diff, opts, b_idx, cache) < 0)
 		return -1;
-	
+
 	/* some metrics may not wish to process this file (too big / too small) */
 	if (!cache[a_idx] || !cache[b_idx])
 		return 0;