Update diff handling of untracked directories

When diff encounters an untracked directory, there was a shortcut
that it took which is not compatible with core git.  This makes
the default behavior no longer take that shortcut and instead look
inside the untracked directory to see if there are any untracked
files within it.  If there are not, then the directory is treated
as an ignore directory instead of an untracked directory.  This
has implications for the git_status APIs.

include/git2/diff.h

@@ -124,6 +124,13 @@ typedef enum {
 	 *  adds all files under the directory as IGNORED entries, too.
 	 */
 	GIT_DIFF_RECURSE_IGNORED_DIRS = (1 << 18),
+	/** For an untracked directory, diff can immediately label it UNTRACKED,
+	 *  but this differs from core Git which scans underneath for untracked
+	 *  or ignored files and marks the directory ignored unless it contains
+	 *  untracked files under it.  That search can be slow.  This flag makes
+	 *  diff skip ahead and immediately report the directory as untracked.
+	 */
+	GIT_DIFF_FAST_UNTRACKED_DIRS = (1 << 19),
 } git_diff_option_t;
 
 /**

include/git2/submodule.h

@@ -103,20 +103,20 @@ typedef enum {
  * * WD_UNTRACKED      - wd contains untracked files
  */
 typedef enum {
-	 GIT_SUBMODULE_STATUS_IN_HEAD           = (1u << 0),
+	GIT_SUBMODULE_STATUS_IN_HEAD           = (1u << 0),
-	 GIT_SUBMODULE_STATUS_IN_INDEX          = (1u << 1),
+	GIT_SUBMODULE_STATUS_IN_INDEX          = (1u << 1),
-	 GIT_SUBMODULE_STATUS_IN_CONFIG         = (1u << 2),
+	GIT_SUBMODULE_STATUS_IN_CONFIG         = (1u << 2),
-	 GIT_SUBMODULE_STATUS_IN_WD             = (1u << 3),
+	GIT_SUBMODULE_STATUS_IN_WD             = (1u << 3),
-	 GIT_SUBMODULE_STATUS_INDEX_ADDED       = (1u << 4),
+	GIT_SUBMODULE_STATUS_INDEX_ADDED       = (1u << 4),
-	 GIT_SUBMODULE_STATUS_INDEX_DELETED     = (1u << 5),
+	GIT_SUBMODULE_STATUS_INDEX_DELETED     = (1u << 5),
-	 GIT_SUBMODULE_STATUS_INDEX_MODIFIED    = (1u << 6),
+	GIT_SUBMODULE_STATUS_INDEX_MODIFIED    = (1u << 6),
-	 GIT_SUBMODULE_STATUS_WD_UNINITIALIZED  = (1u << 7),
+	GIT_SUBMODULE_STATUS_WD_UNINITIALIZED  = (1u << 7),
-	 GIT_SUBMODULE_STATUS_WD_ADDED          = (1u << 8),
+	GIT_SUBMODULE_STATUS_WD_ADDED          = (1u << 8),
-	 GIT_SUBMODULE_STATUS_WD_DELETED        = (1u << 9),
+	GIT_SUBMODULE_STATUS_WD_DELETED        = (1u << 9),
-	 GIT_SUBMODULE_STATUS_WD_MODIFIED       = (1u << 10),
+	GIT_SUBMODULE_STATUS_WD_MODIFIED       = (1u << 10),
-	 GIT_SUBMODULE_STATUS_WD_INDEX_MODIFIED = (1u << 11),
+	GIT_SUBMODULE_STATUS_WD_INDEX_MODIFIED = (1u << 11),
-	 GIT_SUBMODULE_STATUS_WD_WD_MODIFIED    = (1u << 12),
+	GIT_SUBMODULE_STATUS_WD_WD_MODIFIED    = (1u << 12),
-	 GIT_SUBMODULE_STATUS_WD_UNTRACKED      = (1u << 13),
+	GIT_SUBMODULE_STATUS_WD_UNTRACKED      = (1u << 13),
 } git_submodule_status_t;
 
 #define GIT_SUBMODULE_STATUS__IN_FLAGS \

src/diff.c

@@ -698,56 +698,58 @@ static bool entry_is_prefixed(
 			item->path[pathlen] == '/');
 }
 
-static int handle_unmatched_new_directory(
+static int diff_scan_inside_untracked_dir(
-	git_diff_list *diff, diff_in_progress *info, git_delta_t *delta)
+	git_diff_list *diff, diff_in_progress *info, git_delta_t *delta_type)
 {
 	int error = 0;
-	const git_index_entry *nitem = info->nitem;
+	git_buf base = GIT_BUF_INIT;
-	bool contains_oitem = entry_is_prefixed(diff, info->oitem, nitem);
+	bool is_ignored;
-	bool recurse_into_dir =
-		(*delta == GIT_DELTA_UNTRACKED &&
-		 DIFF_FLAG_IS_SET(diff, GIT_DIFF_RECURSE_UNTRACKED_DIRS)) ||
-		(*delta == GIT_DELTA_IGNORED &&
-		 DIFF_FLAG_IS_SET(diff, GIT_DIFF_RECURSE_IGNORED_DIRS));
 
-	/* do not advance into directories that contain a .git file */
+	*delta_type = GIT_DELTA_IGNORED;
-	if (!contains_oitem && recurse_into_dir) {
+	git_buf_sets(&base, info->nitem->path);
-		git_buf *full = NULL;
-		if (git_iterator_current_workdir_path(&full, info->new_iter) < 0)
-			return -1;
-		if (git_path_contains_dir(full, DOT_GIT))
-			recurse_into_dir = false;
-	}
 
-	/* if directory is ignored, remember ignore_prefix */
+	/* advance into untracked directory */
-	if ((contains_oitem || recurse_into_dir) &&
+	if ((error = git_iterator_advance_into(&info->nitem, info->new_iter)) < 0) {
-		*delta == GIT_DELTA_UNTRACKED &&
-		git_iterator_current_is_ignored(info->new_iter))
-	{
-		git_buf_sets(&info->ignore_prefix, info->nitem->path);
-		*delta = GIT_DELTA_IGNORED;
 
-		/* skip recursion if we've just learned this is ignored */
+		/* skip ahead if empty */
-		if (DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_RECURSE_IGNORED_DIRS))
-			recurse_into_dir = false;
-	}
-
-	if (contains_oitem || recurse_into_dir) {
-		/* advance into directory */
-		error = git_iterator_advance_into(&info->nitem, info->new_iter);
-
-		/* if directory is empty, can't advance into it, so skip */
 		if (error == GIT_ENOTFOUND) {
 			giterr_clear();
 			error = git_iterator_advance(&info->nitem, info->new_iter);
-
-			git_buf_clear(&info->ignore_prefix);
 		}
 
-		/* return UNMODIFIED to tell caller not to create a new record */
+		return error;
-		*delta = GIT_DELTA_UNMODIFIED;
 	}
 
+	/* look for actual untracked file */
+	while (!diff->pfxcomp(info->nitem->path, git_buf_cstr(&base))) {
+		is_ignored = git_iterator_current_is_ignored(info->new_iter);
+
+		/* need to recurse into non-ignored directories */
+		if (!is_ignored && S_ISDIR(info->nitem->mode)) {
+			if ((error = git_iterator_advance_into(
+					 &info->nitem, info->new_iter)) < 0)
+				break;
+			continue;
+		}
+
+		/* found a non-ignored item - treat parent dir as untracked */
+		if (!is_ignored) {
+			*delta_type = GIT_DELTA_UNTRACKED;
+			break;
+		}
+
+		if ((error = git_iterator_advance(&info->nitem, info->new_iter)) < 0)
+			break;
+	}
+
+	/* finish off scan */
+	while (!diff->pfxcomp(info->nitem->path, git_buf_cstr(&base))) {
+		if ((error = git_iterator_advance(&info->nitem, info->new_iter)) < 0)
+			break;
+	}
+
+	git_buf_free(&base);
+
 	return error;
 }
 
@@ -757,36 +759,116 @@ static int handle_unmatched_new_item(
 	int error = 0;
 	const git_index_entry *nitem = info->nitem;
 	git_delta_t delta_type = GIT_DELTA_UNTRACKED;
+	bool contains_oitem;
 
-	/* check if contained in ignored parent directory */
+	/* check if this is a prefix of the other side */
-	if (git_buf_len(&info->ignore_prefix) &&
+	contains_oitem = entry_is_prefixed(diff, info->oitem, nitem);
-		diff->pfxcomp(nitem->path, git_buf_cstr(&info->ignore_prefix)) == 0)
-		delta_type = GIT_DELTA_IGNORED;
 
-	if (S_ISDIR(nitem->mode)) {
+	/* check if this is contained in an ignored parent directory */
-		error = handle_unmatched_new_directory(diff, info, &delta_type);
+	if (git_buf_len(&info->ignore_prefix)) {
-
+		if (diff->pfxcomp(nitem->path, git_buf_cstr(&info->ignore_prefix)) == 0)
-		if (error || delta_type == GIT_DELTA_UNMODIFIED)
+			delta_type = GIT_DELTA_IGNORED;
-			return error;
+		else
+			git_buf_clear(&info->ignore_prefix);
 	}
 
-	/* In core git, the next two "else if" clauses are effectively
+	if (S_ISDIR(nitem->mode)) {
-	 * reversed -- i.e. when an untracked file contained in an
+		bool recurse_into_dir = contains_oitem;
-	 * ignored directory is individually ignored, it shows up as an
+
-	 * ignored file in the diff list, even though other untracked
+		/* if not already inside an ignored dir, check if this is ignored */
-	 * files in the same directory are skipped completely.
+		if (delta_type != GIT_DELTA_IGNORED &&
+			git_iterator_current_is_ignored(info->new_iter))
+		{
+			delta_type = GIT_DELTA_IGNORED;
+			git_buf_sets(&info->ignore_prefix, nitem->path);
+		}
+
+		/* check if user requests recursion into this type of dir */
+		recurse_into_dir = contains_oitem ||
+			(delta_type == GIT_DELTA_UNTRACKED &&
+			 DIFF_FLAG_IS_SET(diff, GIT_DIFF_RECURSE_UNTRACKED_DIRS)) ||
+			(delta_type == GIT_DELTA_IGNORED &&
+			 DIFF_FLAG_IS_SET(diff, GIT_DIFF_RECURSE_IGNORED_DIRS));
+
+		/* do not advance into directories that contain a .git file */
+		if (recurse_into_dir) {
+			git_buf *full = NULL;
+			if (git_iterator_current_workdir_path(&full, info->new_iter) < 0)
+				return -1;
+			if (full && git_path_contains_dir(full, DOT_GIT))
+				recurse_into_dir = false;
+		}
+
+		/* still have to look into untracked directories to match core git -
+		 * with no untracked files, directory is treated as ignored
+		 */
+		if (!recurse_into_dir &&
+			delta_type == GIT_DELTA_UNTRACKED &&
+			DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_FAST_UNTRACKED_DIRS))
+		{
+			git_diff_delta *last;
+
+			/* attempt to insert record for this directory */
+			if ((error = diff_delta__from_one(diff, delta_type, nitem)) < 0)
+				return error;
+
+			/* if delta wasn't created (because of rules), just skip ahead */
+			last = diff_delta__last_for_item(diff, nitem);
+			if (!last)
+				return git_iterator_advance(&info->nitem, info->new_iter);
+
+			/* iterate into dir looking for an actual untracked file */
+			if (diff_scan_inside_untracked_dir(diff, info, &delta_type) < 0)
+				return -1;
+
+			/* it iteration changed delta type, the update the record */
+			if (delta_type == GIT_DELTA_IGNORED) {
+				last->status = GIT_DELTA_IGNORED;
+
+				/* remove the record if we don't want ignored records */
+				if (DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_INCLUDE_IGNORED)) {
+					git_vector_pop(&diff->deltas);
+					git__free(last);
+				}
+			}
+
+			return 0;
+		}
+
+		/* try to advance into directory if necessary */
+		if (recurse_into_dir) {
+			error = git_iterator_advance_into(&info->nitem, info->new_iter);
+
+			/* if real error or no error, proceed with iteration */
+			if (error != GIT_ENOTFOUND)
+				return error;
+			giterr_clear();
+
+			/* if directory is empty, can't advance into it, so either skip
+			 * it or ignore it
+			 */
+			if (contains_oitem)
+				return git_iterator_advance(&info->nitem, info->new_iter);
+			delta_type = GIT_DELTA_IGNORED;
+		}
+	}
+
+	/* In core git, the next two checks are effectively reversed --
+	 * i.e. when an file contained in an ignored directory is explicitly
+	 * ignored, it shows up as an ignored file in the diff list, even though
+	 * other untracked files in the same directory are skipped completely.
 	 *
-	 * To me, this is odd.  If the directory is ignored and the file
+	 * To me, this seems odd.  If the directory is ignored and the file is
-	 * is untracked, we should skip it consistently, regardless of
+	 * untracked, we should skip it consistently, regardless of whether it
-	 * whether it happens to match a pattern in the ignore file.
+	 * happens to match a pattern in the ignore file.
 	 *
-	 * To match the core git behavior, just reverse the following
+	 * To match the core git behavior, reverse the following two if checks
-	 * two "else if" cases so that individual file ignores are
+	 * so that individual file ignores are checked before container
-	 * checked before container directory exclusions are used to
+	 * directory exclusions are used to skip the file.
-	 * skip the file.
 	 */
 	else if (delta_type == GIT_DELTA_IGNORED &&
-			 DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_RECURSE_IGNORED_DIRS))
+		DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_RECURSE_IGNORED_DIRS))
+		/* item contained in ignored directory, so skip over it */
 		return git_iterator_advance(&info->nitem, info->new_iter);
 
 	else if (git_iterator_current_is_ignored(info->new_iter))
@@ -795,15 +877,16 @@ static int handle_unmatched_new_item(
 	else if (info->new_iter->type != GIT_ITERATOR_TYPE_WORKDIR)
 		delta_type = GIT_DELTA_ADDED;
 
+	/* Actually create the record for this item if necessary */
 	if ((error = diff_delta__from_one(diff, delta_type, nitem)) < 0)
 		return error;
 
-	/* if we are generating TYPECHANGE records then check for that
+	/* If user requested TYPECHANGE records, then check for that instead of
-	 * instead of just generating an ADDED/UNTRACKED record
+	 * just generating an ADDED/UNTRACKED record
 	 */
 	if (delta_type != GIT_DELTA_IGNORED &&
 		DIFF_FLAG_IS_SET(diff, GIT_DIFF_INCLUDE_TYPECHANGE_TREES) &&
-		entry_is_prefixed(diff, info->oitem, nitem))
+		contains_oitem)
 	{
 		/* this entry was prefixed with a tree - make TYPECHANGE */
 		git_diff_delta *last = diff_delta__last_for_item(diff, nitem);

src/vector.c

@@ -277,15 +277,13 @@ void git_vector_swap(git_vector *a, git_vector *b)
 
 int git_vector_resize_to(git_vector *v, size_t new_length)
 {
-	if (new_length <= v->length)
-		return 0;
-
 	if (new_length > v->_alloc_size &&
 		resize_vector(v, new_length) < 0)
 		return -1;
 
-	memset(&v->contents[v->length], 0,
+	if (new_length > v->length)
-		sizeof(void *) * (new_length - v->length));
+		memset(&v->contents[v->length], 0,
+			sizeof(void *) * (new_length - v->length));
 
 	v->length = new_length;
 

tests-clar/status/status_helpers.c

@@ -40,7 +40,8 @@ int cb_status__single(const char *p, unsigned int s, void *payload)
 {
 	status_entry_single *data = (status_entry_single *)payload;
 
-	GIT_UNUSED(p);
+	if (data->debug)
+		fprintf(stderr, "%02d: %s (%04x)\n", data->count, p, s);
 
 	data->count++;
 	data->status = s;

tests-clar/status/status_helpers.h

@@ -24,6 +24,7 @@ extern int cb_status__count(const char *p, unsigned int s, void *payload);
 typedef struct {
 	int count;
 	unsigned int status;
+	bool debug;
 } status_entry_single;
 
 /* cb_status__single takes payload of "status_entry_single *" */

tests-clar/status/worktree.c

@@ -258,9 +258,8 @@ void test_status_worktree__ignores(void)
 
 static int cb_status__check_592(const char *p, unsigned int s, void *payload)
 {
-	GIT_UNUSED(payload);
+	if (s != GIT_STATUS_WT_DELETED ||
-
+		(payload != NULL && strcmp(p, (const char *)payload) != 0))
-	if (s != GIT_STATUS_WT_DELETED || (payload != NULL && strcmp(p, (const char *)payload) != 0))
 		return -1;
 
 	return 0;

tests-clar/submodule/status.c

@@ -383,3 +383,30 @@ void test_submodule_status__iterator(void)
 
 	cl_git_pass(git_status_foreach_ext(g_repo, &opts, confirm_submodule_status, &exp));
 }
+
+void test_submodule_status__untracked_dirs_containing_ignored_files(void)
+{
+	git_buf path = GIT_BUF_INIT;
+	unsigned int status, expected;
+	git_submodule *sm;
+
+	cl_git_pass(git_buf_joinpath(&path, git_repository_path(g_repo), "modules/sm_unchanged/info/exclude"));
+	cl_git_append2file(git_buf_cstr(&path), "\n*.ignored\n");
+
+	cl_git_pass(git_buf_joinpath(&path, git_repository_workdir(g_repo), "sm_unchanged/directory"));
+	cl_git_pass(git_futils_mkdir(git_buf_cstr(&path), NULL, 0755, 0));
+	cl_git_pass(git_buf_joinpath(&path, git_buf_cstr(&path), "i_am.ignored"));
+	cl_git_mkfile(git_buf_cstr(&path), "ignored this file, please\n");
+
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_unchanged"));
+	cl_git_pass(git_submodule_status(&status, sm));
+
+	cl_assert(GIT_SUBMODULE_STATUS_IS_UNMODIFIED(status));
+
+	expected = GIT_SUBMODULE_STATUS_IN_HEAD |
+		GIT_SUBMODULE_STATUS_IN_INDEX |
+		GIT_SUBMODULE_STATUS_IN_CONFIG |
+		GIT_SUBMODULE_STATUS_IN_WD;
+
+	cl_assert(status == expected);
+}