submodule: teach rm to remove submodules unless they contain a git directory

Currently using "git rm" on a submodule - populated or not - fails with
this error:

	fatal: git rm: '<submodule path>': Is a directory

This made sense in the past as there was no way to remove a submodule
without possibly removing unpushed parts of the submodule's history
contained in its .git directory too, so erroring out here protected the
user from possible loss of data.

But submodules cloned with a recent git version do not contain the .git
directory anymore, they use a gitfile to point to their git directory
which is safely stored inside the superproject's .git directory. The work
tree of these submodules can safely be removed without losing history, so
let's teach git to do so.

Using rm on an unpopulated submodule now removes the empty directory from
the work tree and the gitlink from the index. If the submodule's directory
is missing from the work tree, it will still be removed from the index.

Using rm on a populated submodule using a gitfile will apply the usual
checks for work tree modification adapted to submodules (unless forced).
For a submodule that means that the HEAD is the same as recorded in the
index, no tracked files are modified and no untracked files that aren't
ignored are present in the submodules work tree (ignored files are deemed
expendable and won't stop a submodule's work tree from being removed).
That logic has to be applied in all nested submodules too.

Using rm on a submodule which has its .git directory inside the work trees
top level directory will just error out like it did before to protect the
repository, even when forced. In the future git could either provide a
message informing the user to convert the submodule to use a gitfile or
even attempt to do the conversion itself, but that is not part of this
change.

Signed-off-by: Jens Lehmann <[email protected]>
Signed-off-by: Junio C Hamano <[email protected]>

Author: jlehmann

Documentation/git-rm.txt

@@ -107,6 +107,21 @@ as well as modifications of existing paths.
 Typically you would first remove all tracked files from the working
 tree using this command:
 
+Submodules
+~~~~~~~~~~
+Only submodules using a gitfile (which means they were cloned
+with a git version 1.7.8 or newer) will be removed from the work
+tree, as their repository lives inside the .git directory of the
+superproject. If a submodule (or one of those nested inside it)
+still uses a .git directory, `git rm` will fail - no matter if forced
+or not - to protect the submodule's history.
+
+A submodule is considered up-to-date when the HEAD is the same as
+recorded in the index, no tracked files are modified and no untracked
+files that aren't ignored are present in the submodules work tree.
+Ignored files are deemed expendable and won't stop a submodule's work
+tree from being removed.
+
 ----------------
 git ls-files -z | xargs -0 rm -f
 ----------------

builtin/rm.c

@@ -9,6 +9,7 @@
 #include "cache-tree.h"
 #include "tree-walk.h"
 #include "parse-options.h"
+#include "submodule.h"
 
 static const char * const builtin_rm_usage[] = {
 	"git rm [options] [--] <file>...",
@@ -17,9 +18,58 @@ static const char * const builtin_rm_usage[] = {
 
 static struct {
 	int nr, alloc;
-	const char **name;
+	struct {
+		const char *name;
+		char is_submodule;
+	} *entry;
 } list;
 
+static int get_ours_cache_pos(const char *path, int pos)
+{
+	int i = -pos - 1;
+
+	while ((i < active_nr) && !strcmp(active_cache[i]->name, path)) {
+		if (ce_stage(active_cache[i]) == 2)
+			return i;
+		i++;
+	}
+	return -1;
+}
+
+static int check_submodules_use_gitfiles(void)
+{
+	int i;
+	int errs = 0;
+
+	for (i = 0; i < list.nr; i++) {
+		const char *name = list.entry[i].name;
+		int pos;
+		struct cache_entry *ce;
+		struct stat st;
+
+		pos = cache_name_pos(name, strlen(name));
+		if (pos < 0) {
+			pos = get_ours_cache_pos(name, pos);
+			if (pos < 0)
+				continue;
+		}
+		ce = active_cache[pos];
+
+		if (!S_ISGITLINK(ce->ce_mode) ||
+		    (lstat(ce->name, &st) < 0) ||
+		    is_empty_dir(name))
+			continue;
+
+		if (!submodule_uses_gitfile(name))
+			errs = error(_("submodule '%s' (or one of its nested "
+				     "submodules) uses a .git directory\n"
+				     "(use 'rm -rf' if you really want to remove "
+				     "it including all of its history)"), name);
+	}
+
+	return errs;
+}
+
 static int check_local_mod(unsigned char *head, int index_only)
 {
 	/*
@@ -37,15 +87,26 @@ static int check_local_mod(unsigned char *head, int index_only)
 		struct stat st;
 		int pos;
 		struct cache_entry *ce;
-		const char *name = list.name[i];
+		const char *name = list.entry[i].name;
 		unsigned char sha1[20];
 		unsigned mode;
 		int local_changes = 0;
 		int staged_changes = 0;
 
 		pos = cache_name_pos(name, strlen(name));
-		if (pos < 0)
-			continue; /* removing unmerged entry */
+		if (pos < 0) {
+			/*
+			 * Skip unmerged entries except for populated submodules
+			 * that could lose history when removed.
+			 */
+			pos = get_ours_cache_pos(name, pos);
+			if (pos < 0)
+				continue;
+
+			if (!S_ISGITLINK(active_cache[pos]->ce_mode) ||
+			    is_empty_dir(name))
+				continue;
+		}
 		ce = active_cache[pos];
 
 		if (lstat(ce->name, &st) < 0) {
@@ -58,9 +119,10 @@ static int check_local_mod(unsigned char *head, int index_only)
 			/* if a file was removed and it is now a
 			 * directory, that is the same as ENOENT as
 			 * far as git is concerned; we do not track
-			 * directories.
+			 * directories unless they are submodules.
 			 */
-			continue;
+			if (!S_ISGITLINK(ce->ce_mode))
+				continue;
 		}
 
 		/*
@@ -80,8 +142,11 @@ static int check_local_mod(unsigned char *head, int index_only)
 
 		/*
 		 * Is the index different from the file in the work tree?
+		 * If it's a submodule, is its work tree modified?
 		 */
-		if (ce_match_stat(ce, &st, 0))
+		if (ce_match_stat(ce, &st, 0) ||
+		    (S_ISGITLINK(ce->ce_mode) &&
+		     !ok_to_remove_submodule(ce->name)))
 			local_changes = 1;
 
 		/*
@@ -115,10 +180,18 @@ static int check_local_mod(unsigned char *head, int index_only)
 				errs = error(_("'%s' has changes staged in the index\n"
 					     "(use --cached to keep the file, "
 					     "or -f to force removal)"), name);
-			if (local_changes)
-				errs = error(_("'%s' has local modifications\n"
-					     "(use --cached to keep the file, "
-					     "or -f to force removal)"), name);
+			if (local_changes) {
+				if (S_ISGITLINK(ce->ce_mode) &&
+				    !submodule_uses_gitfile(name)) {
+					errs = error(_("submodule '%s' (or one of its nested "
+						     "submodules) uses a .git directory\n"
+						     "(use 'rm -rf' if you really want to remove "
+						     "it including all of its history)"), name);
+				} else
+					errs = error(_("'%s' has local modifications\n"
+						     "(use --cached to keep the file, "
+						     "or -f to force removal)"), name);
+			}
 		}
 	}
 	return errs;
@@ -173,8 +246,9 @@ int cmd_rm(int argc, const char **argv, const char *prefix)
 		struct cache_entry *ce = active_cache[i];
 		if (!match_pathspec(pathspec, ce->name, ce_namelen(ce), 0, seen))
 			continue;
-		ALLOC_GROW(list.name, list.nr + 1, list.alloc);
-		list.name[list.nr++] = ce->name;
+		ALLOC_GROW(list.entry, list.nr + 1, list.alloc);
+		list.entry[list.nr].name = ce->name;
+		list.entry[list.nr++].is_submodule = S_ISGITLINK(ce->ce_mode);
 	}
 
 	if (pathspec) {
@@ -215,14 +289,17 @@ int cmd_rm(int argc, const char **argv, const char *prefix)
 			hashclr(sha1);
 		if (check_local_mod(sha1, index_only))
 			exit(1);
+	} else if (!index_only) {
+		if (check_submodules_use_gitfiles())
+			exit(1);
 	}
 
 	/*
 	 * First remove the names from the index: we won't commit
 	 * the index unless all of them succeed.
 	 */
 	for (i = 0; i < list.nr; i++) {
-		const char *path = list.name[i];
+		const char *path = list.entry[i].name;
 		if (!quiet)
 			printf("rm '%s'\n", path);
 
@@ -244,7 +321,25 @@ int cmd_rm(int argc, const char **argv, const char *prefix)
 	if (!index_only) {
 		int removed = 0;
 		for (i = 0; i < list.nr; i++) {
-			const char *path = list.name[i];
+			const char *path = list.entry[i].name;
+			if (list.entry[i].is_submodule) {
+				if (is_empty_dir(path)) {
+					if (!rmdir(path)) {
+						removed = 1;
+						continue;
+					}
+				} else {
+					struct strbuf buf = STRBUF_INIT;
+					strbuf_addstr(&buf, path);
+					if (!remove_dir_recursively(&buf, 0)) {
+						removed = 1;
+						strbuf_release(&buf);
+						continue;
+					}
+					strbuf_release(&buf);
+					/* Fallthrough and let remove_path() fail. */
+				}
+			}
 			if (!remove_path(path)) {
 				removed = 1;
 				continue;

submodule.c

@@ -758,6 +758,86 @@ unsigned is_submodule_modified(const char *path, int ignore_untracked)
 	return dirty_submodule;
 }
 
+int submodule_uses_gitfile(const char *path)
+{
+	struct child_process cp;
+	const char *argv[] = {
+		"submodule",
+		"foreach",
+		"--quiet",
+		"--recursive",
+		"test -f .git",
+		NULL,
+	};
+	struct strbuf buf = STRBUF_INIT;
+	const char *git_dir;
+
+	strbuf_addf(&buf, "%s/.git", path);
+	git_dir = read_gitfile(buf.buf);
+	if (!git_dir) {
+		strbuf_release(&buf);
+		return 0;
+	}
+	strbuf_release(&buf);
+
+	/* Now test that all nested submodules use a gitfile too */
+	memset(&cp, 0, sizeof(cp));
+	cp.argv = argv;
+	cp.env = local_repo_env;
+	cp.git_cmd = 1;
+	cp.no_stdin = 1;
+	cp.no_stderr = 1;
+	cp.no_stdout = 1;
+	cp.dir = path;
+	if (run_command(&cp))
+		return 0;
+
+	return 1;
+}
+
+int ok_to_remove_submodule(const char *path)
+{
+	struct stat st;
+	ssize_t len;
+	struct child_process cp;
+	const char *argv[] = {
+		"status",
+		"--porcelain",
+		"-u",
+		"--ignore-submodules=none",
+		NULL,
+	};
+	struct strbuf buf = STRBUF_INIT;
+	int ok_to_remove = 1;
+
+	if ((lstat(path, &st) < 0) || is_empty_dir(path))
+		return 1;
+
+	if (!submodule_uses_gitfile(path))
+		return 0;
+
+	memset(&cp, 0, sizeof(cp));
+	cp.argv = argv;
+	cp.env = local_repo_env;
+	cp.git_cmd = 1;
+	cp.no_stdin = 1;
+	cp.out = -1;
+	cp.dir = path;
+	if (start_command(&cp))
+		die("Could not run 'git status --porcelain -uall --ignore-submodules=none' in submodule %s", path);
+
+	len = strbuf_read(&buf, cp.out, 1024);
+	if (len > 2)
+		ok_to_remove = 0;
+	close(cp.out);
+
+	if (finish_command(&cp))
+		die("'git status --porcelain -uall --ignore-submodules=none' failed in submodule %s", path);
+
+	strbuf_release(&buf);
+	return ok_to_remove;
+}
+
 static int find_first_merges(struct object_array *result, const char *path,
 		struct commit *a, struct commit *b)
 {

submodule.h

@@ -27,6 +27,8 @@ int fetch_populated_submodules(int num_options, const char **options,
 			       const char *prefix, int command_line_option,
 			       int quiet);
 unsigned is_submodule_modified(const char *path, int ignore_untracked);
+int submodule_uses_gitfile(const char *path);
+int ok_to_remove_submodule(const char *path);
 int merge_submodule(unsigned char result[20], const char *path, const unsigned char base[20],
 		    const unsigned char a[20], const unsigned char b[20], int search);
 int find_unpushed_submodules(unsigned char new_sha1[20], const char *remotes_name,