Merge pull request #2241 from libgit2/rb/stash-skip-submodules

Improve stash and checkout for ignored + untracked items

7 files changed, 211 insertions, 107 deletions

diff --git a/src/checkout.c b/src/checkout.c
index 0b385226b..bc976b854 100644
--- a/src/checkout.c
+++ b/src/checkout.c
@@ -56,6 +56,7 @@ typedef struct {
 	git_vector conflicts;
 	git_buf path;
 	size_t workdir_len;
+	git_buf tmp;
 	unsigned int strategy;
 	int can_symlink;
 	bool reload_submodules;
@@ -259,21 +260,41 @@ static int checkout_action_no_wd(
 	return checkout_action_common(action, data, delta, NULL);
 }
 
+static bool wd_item_is_removable(git_iterator *iter, const git_index_entry *wd)
+{
+	git_buf *full = NULL;
+
+	if (wd->mode != GIT_FILEMODE_TREE)
+		return true;
+	if (git_iterator_current_workdir_path(&full, iter) < 0)
+		return true;
+	return !full || !git_path_contains(full, DOT_GIT);
+}
+
+static int checkout_queue_remove(checkout_data *data, const char *path)
+{
+	char *copy = git_pool_strdup(&data->pool, path);
+	GITERR_CHECK_ALLOC(copy);
+	return git_vector_insert(&data->removes, copy);
+}
+
+/* note that this advances the iterator over the wd item */
 static int checkout_action_wd_only(
 	checkout_data *data,
 	git_iterator *workdir,
-	const git_index_entry *wd,
+	const git_index_entry **wditem,
 	git_vector *pathspec)
 {
 	int error = 0;
 	bool remove = false;
 	git_checkout_notify_t notify = GIT_CHECKOUT_NOTIFY_NONE;
+	const git_index_entry *wd = *wditem;
 
 	if (!git_pathspec__match(
 			pathspec, wd->path,
 			(data->strategy & GIT_CHECKOUT_DISABLE_PATHSPEC_MATCH) != 0,
 			git_iterator_ignore_case(workdir), NULL, NULL))
-		return 0;
+		return git_iterator_advance(wditem, workdir);
 
 	/* check if item is tracked in the index but not in the checkout diff */
 	if (data->index != NULL) {
@@ -303,24 +324,49 @@ static int checkout_action_wd_only(
 		}
 	}
 
-	if (notify != GIT_CHECKOUT_NOTIFY_NONE)
-		/* found in index */;
-	else if (git_iterator_current_is_ignored(workdir)) {
-		notify = GIT_CHECKOUT_NOTIFY_IGNORED;
-		remove = ((data->strategy & GIT_CHECKOUT_REMOVE_IGNORED) != 0);
-	}
-	else {
-		notify = GIT_CHECKOUT_NOTIFY_UNTRACKED;
-		remove = ((data->strategy & GIT_CHECKOUT_REMOVE_UNTRACKED) != 0);
-	}
+	if (notify != GIT_CHECKOUT_NOTIFY_NONE) {
+		/* if we found something in the index, notify and advance */
+		if ((error = checkout_notify(data, notify, NULL, wd)) != 0)
+			return error;
+
+		if (remove && wd_item_is_removable(workdir, wd))
+			error = checkout_queue_remove(data, wd->path);
+
+		if (!error)
+			error = git_iterator_advance(wditem, workdir);
+	} else {
+		/* untracked or ignored - can't know which until we advance through */
+		bool over = false, removable = wd_item_is_removable(workdir, wd);
+		git_iterator_status_t untracked_state;
+
+		/* copy the entry for issuing notification callback later */
+		git_index_entry saved_wd = *wd;
+		git_buf_sets(&data->tmp, wd->path);
+		saved_wd.path = data->tmp.ptr;
+
+		error = git_iterator_advance_over_with_status(
+			wditem, &untracked_state, workdir);
+		if (error == GIT_ITEROVER)
+			over = true;
+		else if (error < 0)
+			return error;
 
-	error = checkout_notify(data, notify, NULL, wd);
+		if (untracked_state == GIT_ITERATOR_STATUS_IGNORED) {
+			notify = GIT_CHECKOUT_NOTIFY_IGNORED;
+			remove = ((data->strategy & GIT_CHECKOUT_REMOVE_IGNORED) != 0);
+		} else {
+			notify = GIT_CHECKOUT_NOTIFY_UNTRACKED;
+			remove = ((data->strategy & GIT_CHECKOUT_REMOVE_UNTRACKED) != 0);
+		}
 
-	if (!error && remove) {
-		char *path = git_pool_strdup(&data->pool, wd->path);
-		GITERR_CHECK_ALLOC(path);
+		if ((error = checkout_notify(data, notify, NULL, &saved_wd)) != 0)
+			return error;
 
-		error = git_vector_insert(&data->removes, path);
+		if (remove && removable)
+			error = checkout_queue_remove(data, saved_wd.path);
+
+		if (!error && over) /* restore ITEROVER if needed */
+			error = GIT_ITEROVER;
 	}
 
 	return error;
@@ -554,11 +600,8 @@ static int checkout_action(
 			}
 
 			/* case 1 - handle wd item (if it matches pathspec) */
-			error = checkout_action_wd_only(data, workdir, wd, pathspec);
-			if (error)
-				goto done;
-			if ((error = git_iterator_advance(wditem, workdir)) < 0 &&
-				error != GIT_ITEROVER)
+			error = checkout_action_wd_only(data, workdir, wditem, pathspec);
+			if (error && error != GIT_ITEROVER)
 				goto done;
 			continue;
 		}
@@ -619,10 +662,8 @@ static int checkout_remaining_wd_items(
 {
 	int error = 0;
 
-	while (wd && !error) {
-		if (!(error = checkout_action_wd_only(data, workdir, wd, spec)))
-			error = git_iterator_advance(&wd, workdir);
-	}
+	while (wd && !error)
+		error = checkout_action_wd_only(data, workdir, &wd, spec);
 
 	if (error == GIT_ITEROVER)
 		error = 0;
@@ -1853,6 +1894,7 @@ static void checkout_data_clear(checkout_data *data)
 	data->pfx = NULL;
 
 	git_buf_free(&data->path);
+	git_buf_free(&data->tmp);
 
 	git_index_free(data->index);
 	data->index = NULL;
diff --git a/src/diff.c b/src/diff.c
index fd881c6f6..4b6fbe25a 100644
--- a/src/diff.c
+++ b/src/diff.c
@@ -784,72 +784,6 @@ static bool entry_is_prefixed(
 			item->path[pathlen] == '/');
 }
 
-static int diff_scan_inside_untracked_dir(
-	git_diff *diff, diff_in_progress *info, git_delta_t *delta_type)
-{
-	int error = 0;
-	git_buf base = GIT_BUF_INIT;
-	bool is_ignored;
-
-	*delta_type = GIT_DELTA_IGNORED;
-	git_buf_sets(&base, info->nitem->path);
-
-	/* advance into untracked directory */
-	if ((error = git_iterator_advance_into(&info->nitem, info->new_iter)) < 0) {
-
-		/* skip ahead if empty */
-		if (error == GIT_ENOTFOUND) {
-			giterr_clear();
-			error = git_iterator_advance(&info->nitem, info->new_iter);
-		}
-
-		goto done;
-	}
-
-	/* look for actual untracked file */
-	while (info->nitem != NULL &&
-		   !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)) {
-			error = git_iterator_advance_into(&info->nitem, info->new_iter);
-
-			if (!error)
-				continue;
-			else if (error == GIT_ENOTFOUND) {
-				error = 0;
-				is_ignored = true; /* treat empty as ignored */
-			} else
-				break; /* real error, must stop */
-		}
-
-		/* 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 (info->nitem != NULL &&
-		   !diff->pfxcomp(info->nitem->path, git_buf_cstr(&base))) {
-		if ((error = git_iterator_advance(&info->nitem, info->new_iter)) < 0)
-			break;
-	}
-
-done:
-	git_buf_free(&base);
-
-	if (error == GIT_ITEROVER)
-		error = 0;
-
-	return error;
-}
-
 static int handle_unmatched_new_item(
 	git_diff *diff, diff_in_progress *info)
 {
@@ -905,6 +839,7 @@ static int handle_unmatched_new_item(
 			DIFF_FLAG_ISNT_SET(diff, GIT_DIFF_ENABLE_FAST_UNTRACKED_DIRS))
 		{
 			git_diff_delta *last;
+			git_iterator_status_t untracked_state;
 
 			/* attempt to insert record for this directory */
 			if ((error = diff_delta__from_one(diff, delta_type, nitem)) != 0)
@@ -916,11 +851,14 @@ static int handle_unmatched_new_item(
 				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;
+			if ((error = git_iterator_advance_over_with_status(
+					&info->nitem, &untracked_state, info->new_iter)) < 0 &&
+				error != GIT_ITEROVER)
+				return error;
 
-			/* it iteration changed delta type, the update the record */
-			if (delta_type == GIT_DELTA_IGNORED) {
+			/* if we found nothing or just ignored items, update the record */
+			if (untracked_state == GIT_ITERATOR_STATUS_IGNORED ||
+				untracked_state == GIT_ITERATOR_STATUS_EMPTY) {
 				last->status = GIT_DELTA_IGNORED;
 
 				/* remove the record if we don't want ignored records */
diff --git a/src/iterator.c b/src/iterator.c
index 63c14f962..ef27fa71f 100644
--- a/src/iterator.c
+++ b/src/iterator.c
@@ -1528,3 +1528,77 @@ int git_iterator_current_workdir_path(git_buf **path, git_iterator *iter)
 
 	return 0;
 }
+
+int git_iterator_advance_over_with_status(
+	const git_index_entry **entryptr,
+	git_iterator_status_t *status,
+	git_iterator *iter)
+{
+	int error = 0;
+	workdir_iterator *wi = (workdir_iterator *)iter;
+	char *base = NULL;
+	const git_index_entry *entry;
+
+	*status = GIT_ITERATOR_STATUS_NORMAL;
+
+	if (iter->type != GIT_ITERATOR_TYPE_WORKDIR)
+		return git_iterator_advance(entryptr, iter);
+	if ((error = git_iterator_current(&entry, iter)) < 0)
+		return error;
+
+	if (!S_ISDIR(entry->mode)) {
+		if (git_ignore__lookup(
+				&wi->ignores, wi->fi.entry.path, &wi->is_ignored) < 0)
+			wi->is_ignored = true;
+		if (wi->is_ignored)
+			*status = GIT_ITERATOR_STATUS_IGNORED;
+		return git_iterator_advance(entryptr, iter);
+	}
+
+	*status = GIT_ITERATOR_STATUS_EMPTY;
+
+	base = git__strdup(entry->path);
+	GITERR_CHECK_ALLOC(base);
+
+	/* scan inside directory looking for a non-ignored item */
+	while (entry && !iter->prefixcomp(entry->path, base)) {
+		if (git_ignore__lookup(
+				&wi->ignores, wi->fi.entry.path, &wi->is_ignored) < 0)
+			wi->is_ignored = true;
+
+		/* if we found an explicitly ignored item, then update from
+		 * EMPTY to IGNORED
+		 */
+		if (wi->is_ignored)
+			*status = GIT_ITERATOR_STATUS_IGNORED;
+		else if (S_ISDIR(entry->mode)) {
+			error = git_iterator_advance_into(&entry, iter);
+
+			if (!error)
+				continue;
+			else if (error == GIT_ENOTFOUND) {
+				error = 0;
+				wi->is_ignored = true; /* mark empty directories as ignored */
+			} else
+				break; /* real error, stop here */
+		} else {
+			/* we found a non-ignored item, treat parent as untracked */
+			*status = GIT_ITERATOR_STATUS_NORMAL;
+			break;
+		}
+
+		if ((error = git_iterator_advance(&entry, iter)) < 0)
+			break;
+	}
+
+	/* wrap up scan back to base directory */
+	while (entry && !iter->prefixcomp(entry->path, base))
+		if ((error = git_iterator_advance(&entry, iter)) < 0)
+			break;
+
+	*entryptr = entry;
+	git__free(base);
+
+	return error;
+}
+
diff --git a/src/iterator.h b/src/iterator.h
index 751e139d0..ba9c1e486 100644
--- a/src/iterator.h
+++ b/src/iterator.h
@@ -258,4 +258,23 @@ extern int git_iterator_current_workdir_path(
 /* Return index pointer if index iterator, else NULL */
 extern git_index *git_iterator_get_index(git_iterator *iter);
 
+typedef enum {
+	GIT_ITERATOR_STATUS_NORMAL = 0,
+	GIT_ITERATOR_STATUS_IGNORED = 1,
+	GIT_ITERATOR_STATUS_EMPTY = 2
+} git_iterator_status_t;
+
+/* Advance over a directory and check if it contains no files or just
+ * ignored files.
+ *
+ * In a tree or the index, all directories will contain files, but in the
+ * working directory it is possible to have an empty directory tree or a
+ * tree that only contains ignored files.  Many Git operations treat these
+ * cases specially.  This advances over a directory (presumably an
+ * untracked directory) but checks during the scan if there are any files
+ * and any non-ignored files.
+ */
+extern int git_iterator_advance_over_with_status(
+	const git_index_entry **entry, git_iterator_status_t *status, git_iterator *iter);
+
 #endif
diff --git a/src/stash.c b/src/stash.c
index d20e29b80..86e0a627c 100644
--- a/src/stash.c
+++ b/src/stash.c
@@ -178,7 +178,8 @@ static int stash_update_index_from_diff(
 			break;
 
 		case GIT_DELTA_UNTRACKED:
-			if (data->include_untracked)
+			if (data->include_untracked &&
+				delta->new_file.mode != GIT_FILEMODE_TREE)
 				add_path = delta->new_file.path;
 			break;
 
diff --git a/tests/stash/save.c b/tests/stash/save.c
index f06c1fb71..87c6d7e0f 100644
--- a/tests/stash/save.c
+++ b/tests/stash/save.c
@@ -148,6 +148,25 @@ void test_stash_save__can_include_untracked_files(void)
 	assert_blob_oid("refs/stash^3:just.ignore", NULL);
 }
 
+void test_stash_save__untracked_skips_ignored(void)
+{
+	cl_git_append2file("stash/.gitignore", "bundle/vendor/\n");
+	cl_must_pass(p_mkdir("stash/bundle", 0777));
+	cl_must_pass(p_mkdir("stash/bundle/vendor", 0777));
+	cl_git_mkfile("stash/bundle/vendor/blah", "contents\n");
+
+	cl_assert(git_path_exists("stash/when")); /* untracked */
+	cl_assert(git_path_exists("stash/just.ignore")); /* ignored */
+	cl_assert(git_path_exists("stash/bundle/vendor/blah")); /* ignored */
+
+	cl_git_pass(git_stash_save(
+		&stash_tip_oid, repo, signature, NULL, GIT_STASH_INCLUDE_UNTRACKED));
+
+	cl_assert(!git_path_exists("stash/when"));
+	cl_assert(git_path_exists("stash/bundle/vendor/blah"));
+	cl_assert(git_path_exists("stash/just.ignore"));
+}
+
 void test_stash_save__can_include_untracked_and_ignored_files(void)
 {
 	cl_git_pass(git_stash_save(&stash_tip_oid, repo, signature, NULL, GIT_STASH_INCLUDE_UNTRACKED | GIT_STASH_INCLUDE_IGNORED));
@@ -342,7 +361,7 @@ void test_stash_save__can_stage_normal_then_stage_untracked(void)
 
 void test_stash_save__including_untracked_without_any_untracked_file_creates_an_empty_tree(void)
 {
-	cl_git_pass(p_unlink("stash/when"));
+	cl_must_pass(p_unlink("stash/when"));
 
 	assert_status(repo, "what", GIT_STATUS_WT_MODIFIED | GIT_STATUS_INDEX_MODIFIED);
 	assert_status(repo, "how", GIT_STATUS_INDEX_MODIFIED);
@@ -354,3 +373,18 @@ void test_stash_save__including_untracked_without_any_untracked_file_creates_an_
 
 	assert_object_oid("stash^3^{tree}", EMPTY_TREE, GIT_OBJ_TREE);
 }
+
+void test_stash_save__skip_submodules(void)
+{
+	git_repository *untracked_repo;
+	cl_git_pass(git_repository_init(&untracked_repo, "stash/untracked_repo", false));
+	cl_git_mkfile("stash/untracked_repo/content", "stuff");
+	git_repository_free(untracked_repo);
+
+	assert_status(repo, "untracked_repo/", GIT_STATUS_WT_NEW);
+
+	cl_git_pass(git_stash_save(
+		&stash_tip_oid, repo, signature, NULL, GIT_STASH_INCLUDE_UNTRACKED));
+
+	assert_status(repo, "untracked_repo/", GIT_STATUS_WT_NEW);
+}
diff --git a/tests/stash/stash_helpers.c b/tests/stash/stash_helpers.c
index 8b7d685f8..ff683eced 100644
--- a/tests/stash/stash_helpers.c
+++ b/tests/stash/stash_helpers.c
@@ -42,15 +42,11 @@ void assert_status(
 	int status_flags)
 {
 	unsigned int status;
-	int error;
 
-	error = git_status_file(&status, repo, path);
-
-	if (status_flags < 0) {
-		cl_assert_equal_i(status_flags, error);
-		return;
+	if (status_flags < 0)
+		cl_assert_equal_i(status_flags, git_status_file(&status, repo, path));
+	else {
+		cl_git_pass(git_status_file(&status, repo, path));
+		cl_assert_equal_i((unsigned int)status_flags, status);
 	}
-
-	cl_assert_equal_i(0, error);
-	cl_assert_equal_i((unsigned int)status_flags, status);
 }