repository: Introduce "reserved names"

A repository can have multiple "reserved names" now, not just
a single "short name" for the repository folder itself.  Refactor
to include a git_repository__reserved_names that returns all the
reserved names for a repository.

src/path.c

@@ -1351,30 +1351,31 @@ static bool verify_dotgit_hfs(const char *path, size_t len)
 
 GIT_INLINE(bool) verify_dotgit_ntfs(git_repository *repo, const char *path, size_t len)
 {
-	const char *shortname = NULL;
-	size_t i, start, shortname_len = 0;
-
-	/* See if the repo has a custom shortname (not "GIT~1") */
-	if (repo &&
-		(shortname = git_repository__8dot3_name(repo)) &&
-		shortname != git_repository__8dot3_default)
-		shortname_len = strlen(shortname);
-
-	if (len >= 4 && strncasecmp(path, ".git", 4) == 0)
-		start = 4;
-	else if (len >= git_repository__8dot3_default_len &&
-		strncasecmp(path, git_repository__8dot3_default, git_repository__8dot3_default_len) == 0)
-		start = git_repository__8dot3_default_len;
-	else if (shortname_len && len >= shortname_len &&
-		strncasecmp(path, shortname, shortname_len) == 0)
-		start = shortname_len;
-	else
+	git_buf *reserved = git_repository__reserved_names_win32;
+	size_t reserved_len = git_repository__reserved_names_win32_len;
+	size_t start = 0, i;
+
+	if (repo)
+		git_repository__reserved_names(&reserved, &reserved_len, repo, true);
+
+	for (i = 0; i < reserved_len; i++) {
+		git_buf *r = &reserved[i];
+
+		if (len >= r->size &&
+			strncasecmp(path, r->ptr, r->size) == 0) {
+			start = r->size;
+			break;
+		}
+	}
+
+	if (!start)
 		return true;
 
-	/* Reject paths beginning with ".git\" */
+	/* Reject paths like ".git\" */
 	if (path[start] == '\\')
 		return false;
 
+	/* Reject paths like '.git ' or '.git.' */
 	for (i = start; i < len; i++) {
 		if (path[i] != ' ' && path[i] != '.')
 			return true;

src/repository.c

@@ -37,8 +37,16 @@
 
 #define GIT_REPO_VERSION 0
 
-const char *git_repository__8dot3_default = "GIT~1";
-size_t git_repository__8dot3_default_len = 5;
+git_buf git_repository__reserved_names_win32[] = {
+	{ DOT_GIT, 0, CONST_STRLEN(DOT_GIT) },
+	{ GIT_DIR_SHORTNAME, 0, CONST_STRLEN(GIT_DIR_SHORTNAME) }
+};
+size_t git_repository__reserved_names_win32_len = 2;
+
+git_buf git_repository__reserved_names_posix[] = {
+	{ DOT_GIT, 0, CONST_STRLEN(DOT_GIT) },
+};
+size_t git_repository__reserved_names_posix_len = 1;
 
 static void set_odb(git_repository *repo, git_odb *odb)
 {
@@ -110,6 +118,8 @@ void git_repository__cleanup(git_repository *repo)
 
 void git_repository_free(git_repository *repo)
 {
+	size_t i;
+
 	if (repo == NULL)
 		return;
 
@@ -120,10 +130,12 @@ void git_repository_free(git_repository *repo)
 	git_diff_driver_registry_free(repo->diff_drivers);
 	repo->diff_drivers = NULL;
 
+	for (i = 0; i < repo->reserved_names.size; i++)
+		git_buf_free(git_array_get(repo->reserved_names, i));
+
 	git__free(repo->path_repository);
 	git__free(repo->workdir);
 	git__free(repo->namespace);
-	git__free(repo->name_8dot3);
 
 	git__memzero(repo, sizeof(*repo));
 	git__free(repo);
@@ -153,18 +165,26 @@ static bool valid_repository_path(git_buf *repository_path)
 static git_repository *repository_alloc(void)
 {
 	git_repository *repo = git__calloc(1, sizeof(git_repository));
-	if (!repo)
-		return NULL;
 
-	if (git_cache_init(&repo->objects) < 0) {
-		git__free(repo);
-		return NULL;
-	}
+	if (repo == NULL ||
+		git_cache_init(&repo->objects) < 0)
+		goto on_error;
+
+	git_array_init_to_size(repo->reserved_names, 4);
+	if (!repo->reserved_names.ptr)
+		goto on_error;
 
 	/* set all the entries in the cvar cache to `unset` */
 	git_repository__cvar_cache_clear(repo);
 
 	return repo;
+
+on_error:
+	if (repo)
+		git_cache_free(&repo->objects);
+
+	git__free(repo);
+	return NULL;
 }
 
 int git_repository_new(git_repository **out)
@@ -314,6 +334,7 @@ static int read_gitfile(git_buf *path_out, const char *file_path)
 static int find_repo(
 	git_buf *repo_path,
 	git_buf *parent_path,
+	git_buf *link_path,
 	const char *start_path,
 	uint32_t flags,
 	const char *ceiling_dirs)
@@ -356,9 +377,14 @@ static int find_repo(
 				git_buf repo_link = GIT_BUF_INIT;
 
 				if (!(error = read_gitfile(&repo_link, path.ptr))) {
-					if (valid_repository_path(&repo_link))
+					if (valid_repository_path(&repo_link)) {
 						git_buf_swap(repo_path, &repo_link);
 
+						if (link_path)
+							error = git_buf_put(link_path, 
+								path.ptr, path.size);
+					}
+
 					git_buf_free(&repo_link);
 					break;
 				}
@@ -445,13 +471,16 @@ int git_repository_open_ext(
 	const char *ceiling_dirs)
 {
 	int error;
-	git_buf path = GIT_BUF_INIT, parent = GIT_BUF_INIT;
+	git_buf path = GIT_BUF_INIT, parent = GIT_BUF_INIT,
+		link_path = GIT_BUF_INIT;
 	git_repository *repo;
 
 	if (repo_ptr)
 		*repo_ptr = NULL;
 
-	error = find_repo(&path, &parent, start_path, flags, ceiling_dirs);
+	error = find_repo(
+		&path, &parent, &link_path, start_path, flags, ceiling_dirs);
+
 	if (error < 0 || !repo_ptr)
 		return error;
 
@@ -461,6 +490,11 @@ int git_repository_open_ext(
 	repo->path_repository = git_buf_detach(&path);
 	GITERR_CHECK_ALLOC(repo->path_repository);
 
+	if (link_path.size) {
+		repo->path_gitlink = git_buf_detach(&link_path);
+		GITERR_CHECK_ALLOC(repo->path_gitlink);
+	}
+
 	if ((flags & GIT_REPOSITORY_OPEN_BARE) != 0)
 		repo->is_bare = 1;
 	else {
@@ -512,7 +546,7 @@ int git_repository_discover(
 
 	git_buf_sanitize(out);
 
-	return find_repo(out, NULL, start_path, flags, ceiling_dirs);
+	return find_repo(out, NULL, NULL, start_path, flags, ceiling_dirs);
 }
 
 static int load_config(
@@ -797,27 +831,88 @@ const char *git_repository_get_namespace(git_repository *repo)
 	return repo->namespace;
 }
 
-const char *git_repository__8dot3_name(git_repository *repo)
+#ifdef GIT_WIN32
+static int reserved_names_add8dot3(git_repository *repo, const char *path)
 {
-	if (!repo->has_8dot3) {
-		repo->has_8dot3 = 1;
+	char *name = git_win32_path_8dot3_name(path);
+	const char *def = GIT_DIR_SHORTNAME;
+	size_t name_len, def_len = CONST_STRLEN(GIT_DIR_SHORTNAME);
+	git_buf *buf;
 
-#ifdef GIT_WIN32
+	if (!name)
+		return 0;
+
+	name_len = strlen(name);
+
+	if (name_len == def_len && memcmp(name, def, def_len) == 0) {
+		git__free(name);
+		return 0;
+	}
+
+	if ((buf = git_array_alloc(repo->reserved_names)) == NULL)
+		return -1;
+
+	git_buf_attach(buf, name, name_len);
+	return true;
+}
+
+bool git_repository__reserved_names(
+	git_buf **out, size_t *outlen, git_repository *repo, bool include_ntfs)
+{
+	GIT_UNUSED(include_ntfs);
+
+	if (repo->reserved_names.size == 0) {
+		git_buf *buf;
+		size_t i;
+
+		/* Add the static defaults */
+		for (i = 0; i < git_repository__reserved_names_win32_len; i++) {
+			if ((buf = git_array_alloc(repo->reserved_names)) == NULL)
+				goto on_error;
+
+			buf->ptr = git_repository__reserved_names_win32[i].ptr;
+			buf->size = git_repository__reserved_names_win32[i].size;
+		}
+
+		/* Try to add any repo-specific reserved names */
 		if (!repo->is_bare) {
-			repo->name_8dot3 = git_win32_path_8dot3_name(repo->path_repository);
+			const char *reserved_path = repo->path_gitlink ?
+				repo->path_gitlink : repo->path_repository;
 
-			/* We anticipate the 8.3 name is "GIT~1", so use a static for
-			 * easy testing in the common case */
-			if (repo->name_8dot3 &&
-				strcasecmp(repo->name_8dot3, git_repository__8dot3_default) == 0)
-				repo->has_8dot3_default = 1;
+			if (reserved_names_add8dot3(repo, reserved_path) < 0)
+				goto on_error;
 		}
-#endif
 	}
 
-	return repo->has_8dot3_default ?
-		git_repository__8dot3_default : repo->name_8dot3;
+	*out = repo->reserved_names.ptr;
+	*outlen = repo->reserved_names.size;
+
+	return true;
+
+	/* Always give good defaults, even on OOM */
+on_error:
+	*out = git_repository__reserved_names_win32;
+	*outlen = git_repository__reserved_names_win32_len;
+
+	return false;
 }
+#else
+bool git_repository__reserved_names(
+	git_buf **out, size_t *outlen, git_repository *repo, bool include_ntfs)
+{
+	GIT_UNUSED(repo);
+
+	if (include_ntfs) {
+		*out = git_repository__reserved_names_win32;
+		*outlen = git_repository__reserved_names_win32_len;
+	} else {
+		*out = git_repository__reserved_names_posix;
+		*outlen = git_repository__reserved_names_posix_len;
+	}
+
+	return true;
+}
+#endif
 
 static int check_repositoryformatversion(git_config *config)
 {

src/repository.h

@@ -14,6 +14,7 @@
 #include "git2/object.h"
 #include "git2/config.h"
 
+#include "array.h"
 #include "cache.h"
 #include "refs.h"
 #include "buffer.h"
@@ -27,6 +28,9 @@
 #define GIT_DIR_MODE 0755
 #define GIT_BARE_DIR_MODE 0777
 
+/* Default DOS-compatible 8.3 "short name" for a git repository, "GIT~1" */
+#define GIT_DIR_SHORTNAME "GIT~1"
+
 /** Cvar cache identifiers */
 typedef enum {
 	GIT_CVAR_AUTO_CRLF = 0, /* core.autocrlf */
@@ -124,13 +128,13 @@ struct git_repository {
 	git_diff_driver_registry *diff_drivers;
 
 	char *path_repository;
+	char *path_gitlink;
 	char *workdir;
 	char *namespace;
-	char *name_8dot3;
 
-	unsigned is_bare:1,
-		has_8dot3:1,
-		has_8dot3_default:1;
+	git_array_t(git_buf) reserved_names;
+
+	unsigned is_bare:1;
 	unsigned int lru_counter;
 
 	git_atomic attr_session_key;
@@ -185,19 +189,24 @@ int git_repository__set_orig_head(git_repository *repo, const git_oid *orig_head
 
 int git_repository__cleanup_files(git_repository *repo, const char *files[], size_t files_len);
 
-/*
- * Gets the DOS-compatible 8.3 "short name".  This will return only the
- * short name for the repository directory (ie, "git~1" for ".git").  This
- * will always return a pointer to `git_repository__8dot3_default` when
- * "GIT~1" is the short name.  This will return NULL for bare repositories,
- * and systems that do not have a short name.
- */
-const char *git_repository__8dot3_name(git_repository *repo);
+/* The default "reserved names" for a repository */
+extern git_buf git_repository__reserved_names_win32[];
+extern size_t git_repository__reserved_names_win32_len;
+
+extern git_buf git_repository__reserved_names_posix[];
+extern size_t git_repository__reserved_names_posix_len;
 
-/* The default DOS-compatible 8.3 "short name" for a git repository,
- * "GIT~1".
+/*
+ * Gets any "reserved names" in the repository.  This will return paths
+ * that should not be allowed in the repository (like ".git") to avoid
+ * conflicting with the repository path, or with alternate mechanisms to
+ * the repository path (eg, "GIT~1").  Every attempt will be made to look
+ * up all possible reserved names - if there was a conflict for the shortname
+ * GIT~1, for example, this function will try to look up the alternate
+ * shortname.  If that fails, this function returns false, but out and outlen
+ * will still be populated with good defaults.
  */
-extern const char *git_repository__8dot3_default;
-extern size_t git_repository__8dot3_default_len;
+bool git_repository__reserved_names(
+	git_buf **out, size_t *outlen, git_repository *repo, bool include_ntfs);
 
 #endif

tests/repo/reservedname.c

+#include "clar_libgit2.h"
+#include "../submodule/submodule_helpers.h"
+#include "repository.h"
+
+void test_repo_reservedname__cleanup(void)
+{
+	cl_git_sandbox_cleanup();
+}
+
+void test_repo_reservedname__includes_shortname_on_win32(void)
+{
+	git_repository *repo;
+	git_buf *reserved;
+	size_t reserved_len;
+
+	repo = cl_git_sandbox_init("nasty");
+	cl_assert(git_repository__reserved_names(&reserved, &reserved_len, repo, false));
+
+#ifdef GIT_WIN32
+	cl_assert_equal_i(2, reserved_len);
+	cl_assert_equal_s(".git", reserved[0].ptr);
+	cl_assert_equal_s("GIT~1", reserved[1].ptr);
+#else
+	cl_assert_equal_i(1, reserved_len);
+	cl_assert_equal_s(".git", reserved[0].ptr);
+#endif
+}
+
+void test_repo_reservedname__includes_shortname_when_requested(void)
+{
+	git_repository *repo;
+	git_buf *reserved;
+	size_t reserved_len;
+
+	repo = cl_git_sandbox_init("nasty");
+	cl_assert(git_repository__reserved_names(&reserved, &reserved_len, repo, true));
+
+	cl_assert_equal_i(2, reserved_len);
+	cl_assert_equal_s(".git", reserved[0].ptr);
+	cl_assert_equal_s("GIT~1", reserved[1].ptr);
+}
+
+/* Ensures that custom shortnames are included: creates a GIT~1 so that the
+ * .git folder itself will have to be named GIT~2
+ */
+void test_repo_reservedname__custom_shortname_recognized(void)
+{
+#ifdef GIT_WIN32
+	git_repository *repo;
+	git_buf *reserved;
+	size_t reserved_len;
+
+	if (!cl_sandbox_supports_8dot3())
+		clar__skip();
+
+	repo = cl_git_sandbox_init("nasty");
+
+	cl_must_pass(p_rename("nasty/.git", "nasty/_temp"));
+	cl_git_write2file("nasty/git~1", "", 0, O_RDWR|O_CREAT, 0666);
+	cl_must_pass(p_rename("nasty/_temp", "nasty/.git"));
+
+	cl_assert(git_repository__reserved_names(&reserved, &reserved_len, repo, true));
+
+	cl_assert_equal_i(3, reserved_len);
+	cl_assert_equal_s(".git", reserved[0].ptr);
+	cl_assert_equal_s("GIT~1", reserved[1].ptr);
+	cl_assert_equal_s("GIT~2", reserved[2].ptr);
+#endif
+}
+
+/* When looking at the short name for a submodule, we need to prevent
+ * people from overwriting the `.git` file in the submodule working
+ * directory itself.  We don't want to look at the actual repository
+ * path, since it will be in the super's repository above us, and
+ * typically named with the name of our subrepository.  Consequently,
+ * preventing access to the short name of the actual repository path
+ * would prevent us from creating files with the same name as the
+ * subrepo.  (Eg, a submodule named "libgit2" could not contain a file
+ * named "libgit2", which would be unfortunate.)
+ */
+void test_repo_reservedname__submodule_pointer(void)
+{
+#ifdef GIT_WIN32
+	git_repository *super_repo, *sub_repo;
+	git_submodule *sub;
+	git_buf *sub_reserved;
+	size_t sub_reserved_len;
+
+	if (!cl_sandbox_supports_8dot3())
+		clar__skip();
+
+	super_repo = setup_fixture_submod2();
+
+	assert_submodule_exists(super_repo, "sm_unchanged");
+
+	cl_git_pass(git_submodule_lookup(&sub, super_repo, "sm_unchanged"));
+	cl_git_pass(git_submodule_open(&sub_repo, sub));
+
+	cl_assert(git_repository__reserved_names(&sub_reserved, &sub_reserved_len, sub_repo, true));
+
+	cl_assert_equal_i(2, sub_reserved_len);
+	cl_assert_equal_s(".git", sub_reserved[0].ptr);
+	cl_assert_equal_s("GIT~1", sub_reserved[1].ptr);
+
+	git_submodule_free(sub);
+	git_repository_free(sub_repo);
+#endif
+}