Merge pull request #1767 from libgit2/win32-bigger-utf8-buffer

Bigger buffer for utf-8 parsing in win32

10 files changed, 103 insertions, 86 deletions

diff --git a/src/fileops.c b/src/fileops.c
index 7f8418d7a..aea8075d5 100644
--- a/src/fileops.c
+++ b/src/fileops.c
@@ -58,9 +58,9 @@ int git_futils_creat_locked(const char *path, const mode_t mode)
 	int fd;
 
 #ifdef GIT_WIN32
-	wchar_t buf[GIT_WIN_PATH];
+	git_win32_path buf;
 
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path_from_c(buf, path);
 	fd = _wopen(buf, O_WRONLY | O_CREAT | O_TRUNC |
 		O_EXCL | O_BINARY | O_CLOEXEC, mode);
 #else
diff --git a/src/path.c b/src/path.c
index b81675b49..a753a734d 100644
--- a/src/path.c
+++ b/src/path.c
@@ -8,7 +8,6 @@
 #include "path.h"
 #include "posix.h"
 #ifdef GIT_WIN32
-#include "win32/dir.h"
 #include "win32/posix.h"
 #else
 #include <dirent.h>
@@ -486,24 +485,26 @@ bool git_path_is_empty_dir(const char *path)
 {
 	git_buf pathbuf = GIT_BUF_INIT;
 	HANDLE hFind = INVALID_HANDLE_VALUE;
-	wchar_t wbuf[GIT_WIN_PATH];
+	git_win32_path wbuf;
 	WIN32_FIND_DATAW ffd;
 	bool retval = true;
 
 	if (!git_path_isdir(path)) return false;
 
 	git_buf_printf(&pathbuf, "%s\\*", path);
-	git__utf8_to_16(wbuf, GIT_WIN_PATH, git_buf_cstr(&pathbuf));
+	git_win32_path_from_c(wbuf, git_buf_cstr(&pathbuf));
 
 	hFind = FindFirstFileW(wbuf, &ffd);
 	if (INVALID_HANDLE_VALUE == hFind) {
 		giterr_set(GITERR_OS, "Couldn't open '%s'", path);
+		git_buf_free(&pathbuf);
 		return false;
 	}
 
 	do {
 		if (!git_path_is_dot_or_dotdotW(ffd.cFileName)) {
 			retval = false;
+			break;
 		}
 	} while (FindNextFileW(hFind, &ffd) != 0);
 
diff --git a/src/transports/winhttp.c b/src/transports/winhttp.c
index 95e422dc0..8decd8d51 100644
--- a/src/transports/winhttp.c
+++ b/src/transports/winhttp.c
@@ -893,7 +893,7 @@ static int winhttp_connect(
 	const char *url)
 {
 	wchar_t *ua = L"git/1.0 (libgit2 " WIDEN(LIBGIT2_VERSION) L")";
-	wchar_t host[GIT_WIN_PATH];
+	git_win32_path host;
 	int32_t port;
 	const char *default_port = "80";
 	int ret;
@@ -920,7 +920,7 @@ static int winhttp_connect(
 		return -1;
 
 	/* Prepare host */
-	git__utf8_to_16(host, GIT_WIN_PATH, t->host);
+	git_win32_path_from_c(host, t->host);
 
 	/* Establish session */
 	t->session = WinHttpOpen(
@@ -934,7 +934,7 @@ static int winhttp_connect(
 		giterr_set(GITERR_OS, "Failed to init WinHTTP");
 		return -1;
 	}
-	
+
 	/* Establish connection */
 	t->connection = WinHttpConnect(
 			t->session,
diff --git a/src/win32/dir.c b/src/win32/dir.c
index 8c51d8378..f7859b73f 100644
--- a/src/win32/dir.c
+++ b/src/win32/dir.c
@@ -5,8 +5,7 @@
  * a Linking Exception. For full terms see the included COPYING file.
  */
 #define GIT__WIN32_NO_WRAP_DIR
-#include "dir.h"
-#include "utf-conv.h"
+#include "posix.h"
 
 static int init_filter(char *filter, size_t n, const char *dir)
 {
@@ -25,36 +24,32 @@ static int init_filter(char *filter, size_t n, const char *dir)
 
 git__DIR *git__opendir(const char *dir)
 {
-	char filter[GIT_WIN_PATH];
-	wchar_t filter_w[GIT_WIN_PATH];
+	git_win32_path_as_utf8 filter;
+	git_win32_path filter_w;
 	git__DIR *new = NULL;
+	size_t dirlen;
 
 	if (!dir || !init_filter(filter, sizeof(filter), dir))
 		return NULL;
 
-	new = git__calloc(1, sizeof(*new));
+	dirlen = strlen(dir);
+
+	new = git__calloc(sizeof(*new) + dirlen + 1, 1);
 	if (!new)
 		return NULL;
+	memcpy(new->dir, dir, dirlen);
 
-	new->dir = git__strdup(dir);
-	if (!new->dir)
-		goto fail;
-
-	git__utf8_to_16(filter_w, GIT_WIN_PATH, filter);
+	git_win32_path_from_c(filter_w, filter);
 	new->h = FindFirstFileW(filter_w, &new->f);
 
 	if (new->h == INVALID_HANDLE_VALUE) {
 		giterr_set(GITERR_OS, "Could not open directory '%s'", dir);
-		goto fail;
+		git__free(new);
+		return NULL;
 	}
 
 	new->first = 1;
 	return new;
-
-fail:
-	git__free(new->dir);
-	git__free(new);
-	return NULL;
 }
 
 int git__readdir_ext(
@@ -80,7 +75,7 @@ int git__readdir_ext(
 	if (wcslen(d->f.cFileName) >= sizeof(entry->d_name))
 		return -1;
 
-	git__utf16_to_8(entry->d_name, d->f.cFileName);
+	git_win32_path_to_c(entry->d_name, d->f.cFileName);
 	entry->d_ino = 0;
 
 	*result = entry;
@@ -101,8 +96,8 @@ struct git__dirent *git__readdir(git__DIR *d)
 
 void git__rewinddir(git__DIR *d)
 {
-	char filter[GIT_WIN_PATH];
-	wchar_t filter_w[GIT_WIN_PATH];
+	git_win32_path_as_utf8 filter;
+	git_win32_path filter_w;
 
 	if (!d)
 		return;
@@ -116,7 +111,7 @@ void git__rewinddir(git__DIR *d)
 	if (!init_filter(filter, sizeof(filter), d->dir))
 		return;
 
-	git__utf8_to_16(filter_w, GIT_WIN_PATH, filter);
+	git_win32_path_from_c(filter_w, filter);
 	d->h = FindFirstFileW(filter_w, &d->f);
 
 	if (d->h == INVALID_HANDLE_VALUE)
@@ -134,8 +129,7 @@ int git__closedir(git__DIR *d)
 		FindClose(d->h);
 		d->h = INVALID_HANDLE_VALUE;
 	}
-	git__free(d->dir);
-	d->dir = NULL;
+
 	git__free(d);
 	return 0;
 }
diff --git a/src/win32/dir.h b/src/win32/dir.h
index 7696d468e..24d48f6ba 100644
--- a/src/win32/dir.h
+++ b/src/win32/dir.h
@@ -11,15 +11,15 @@
 
 struct git__dirent {
 	int d_ino;
-	char d_name[261];
+	git_win32_path_as_utf8 d_name;
 };
 
 typedef struct {
 	HANDLE h;
 	WIN32_FIND_DATAW f;
 	struct git__dirent entry;
-	char *dir;
 	int first;
+	char dir[GIT_FLEX_ARRAY];
 } git__DIR;
 
 extern git__DIR *git__opendir(const char *);
diff --git a/src/win32/findfile.c b/src/win32/findfile.c
index 9d9051bff..a1c11fcfb 100644
--- a/src/win32/findfile.c
+++ b/src/win32/findfile.c
@@ -23,11 +23,11 @@ int git_win32__expand_path(struct git_win32__path *s_root, const wchar_t *templ)
 	return s_root->len ? 0 : -1;
 }
 
-static int win32_path_utf16_to_8(git_buf *path_utf8, const wchar_t *path_utf16)
+static int win32_path_to_8(git_buf *path_utf8, const wchar_t *path)
 {
 	char temp_utf8[GIT_PATH_MAX];
 
-	git__utf16_to_8(temp_utf8, path_utf16);
+	git__utf16_to_8(temp_utf8, GIT_PATH_MAX, path);
 	git_path_mkposix(temp_utf8);
 
 	return git_buf_sets(path_utf8, temp_utf8);
@@ -53,7 +53,7 @@ int git_win32__find_file(
 	if (*filename == '/' || *filename == '\\')
 		filename++;
 
-	git__utf8_to_16(file_utf16 + root->len - 1, alloc_len, filename);
+	git__utf8_to_16(file_utf16 + root->len - 1, alloc_len - root->len, filename);
 
 	/* check access */
 	if (_waccess(file_utf16, F_OK) < 0) {
@@ -61,7 +61,7 @@ int git_win32__find_file(
 		return GIT_ENOTFOUND;
 	}
 
-	win32_path_utf16_to_8(path, file_utf16);
+	win32_path_to_8(path, file_utf16);
 	git__free(file_utf16);
 
 	return 0;
@@ -113,7 +113,7 @@ static int win32_find_git_in_path(git_buf *buf, const wchar_t *gitexe)
 			/* replace "bin\\" or "cmd\\" with "etc\\" */
 			wcscpy(&root.path[root.len - 4], L"etc\\");
 
-			win32_path_utf16_to_8(buf, root.path);
+			win32_path_to_8(buf, root.path);
 			return 0;
 		}
 	}
@@ -146,7 +146,7 @@ static int win32_find_git_in_registry(
 			wcscat(path16.path, L"etc\\");
 			path16.len += 4;
 
-			win32_path_utf16_to_8(buf, path16.path);
+			win32_path_to_8(buf, path16.path);
 		}
 
 		RegCloseKey(hKey);
@@ -168,7 +168,7 @@ static int win32_find_existing_dirs(
 			path16.path[0] != L'%' &&
 			!_waccess(path16.path, F_OK))
 		{
-			win32_path_utf16_to_8(&buf, path16.path);
+			win32_path_to_8(&buf, path16.path);
 
 			if (buf.size)
 				git_buf_join(out, GIT_PATH_LIST_SEPARATOR, out->ptr, buf.ptr);
diff --git a/src/win32/posix.h b/src/win32/posix.h
index c49c2175c..5f924a026 100644
--- a/src/win32/posix.h
+++ b/src/win32/posix.h
@@ -8,7 +8,9 @@
 #define INCLUDE_posix__w32_h__
 
 #include "common.h"
+#include "../posix.h"
 #include "utf-conv.h"
+#include "dir.h"
 
 GIT_INLINE(int) p_link(const char *old, const char *new)
 {
@@ -20,9 +22,9 @@ GIT_INLINE(int) p_link(const char *old, const char *new)
 
 GIT_INLINE(int) p_mkdir(const char *path, mode_t mode)
 {
-	wchar_t buf[GIT_WIN_PATH];
+	git_win32_path buf;
 	GIT_UNUSED(mode);
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path_from_c(buf, path);
 	return _wmkdir(buf);
 }
 
diff --git a/src/win32/posix_w32.c b/src/win32/posix_w32.c
index 036632e2a..9d5b8d877 100644
--- a/src/win32/posix_w32.c
+++ b/src/win32/posix_w32.c
@@ -16,8 +16,8 @@
 
 int p_unlink(const char *path)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	_wchmod(buf, 0666);
 	return _wunlink(buf);
 }
@@ -59,10 +59,11 @@ static int do_lstat(
 	const char *file_name, struct stat *buf, int posix_enotdir)
 {
 	WIN32_FILE_ATTRIBUTE_DATA fdata;
-	wchar_t fbuf[GIT_WIN_PATH], lastch;
+	git_win32_path fbuf;
+	wchar_t lastch;
 	int flen;
 
-	flen = git__utf8_to_16(fbuf, GIT_WIN_PATH, file_name);
+	flen = git_win32_path_from_c(fbuf, file_name);
 
 	/* truncate trailing slashes */
 	for (; flen > 0; --flen) {
@@ -108,10 +109,10 @@ static int do_lstat(
 		 * the length of the path pointed to, which we expect everywhere else
 		 */
 		if (S_ISLNK(fMode)) {
-			char target[GIT_WIN_PATH];
+			git_win32_path_as_utf8 target;
 			int readlink_result;
 
-			readlink_result = p_readlink(file_name, target, GIT_WIN_PATH);
+			readlink_result = p_readlink(file_name, target, sizeof(target));
 
 			if (readlink_result == -1)
 				return -1;
@@ -165,7 +166,7 @@ int p_readlink(const char *link, char *target, size_t target_len)
 	static fpath_func pGetFinalPath = NULL;
 	HANDLE hFile;
 	DWORD dwRet;
-	wchar_t link_w[GIT_WIN_PATH];
+	git_win32_path link_w;
 	wchar_t* target_w;
 	int error = 0;
 
@@ -188,7 +189,7 @@ int p_readlink(const char *link, char *target, size_t target_len)
 		}
 	}
 
-	git__utf8_to_16(link_w, GIT_WIN_PATH, link);
+	git_win32_path_from_c(link_w, link);
 
 	hFile = CreateFileW(link_w,			// file to open
 			GENERIC_READ,			// open for reading
@@ -254,10 +255,10 @@ int p_symlink(const char *old, const char *new)
 
 int p_open(const char *path, int flags, ...)
 {
-	wchar_t buf[GIT_WIN_PATH];
+	git_win32_path buf;
 	mode_t mode = 0;
 
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path_from_c(buf, path);
 
 	if (flags & O_CREAT) {
 		va_list arg_list;
@@ -272,8 +273,8 @@ int p_open(const char *path, int flags, ...)
 
 int p_creat(const char *path, mode_t mode)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	return _wopen(buf, _O_WRONLY | _O_CREAT | _O_TRUNC | _O_BINARY, mode);
 }
 
@@ -299,7 +300,7 @@ int p_getcwd(char *buffer_out, size_t size)
 
 int p_stat(const char* path, struct stat* buf)
 {
-	char target[GIT_WIN_PATH];
+	git_win32_path_as_utf8 target;
 	int error = 0;
 
 	error = do_lstat(path, buf, 0);
@@ -307,7 +308,7 @@ int p_stat(const char* path, struct stat* buf)
 	/* We need not do this in a loop to unwind chains of symlinks since
 	 * p_readlink calls GetFinalPathNameByHandle which does it for us. */
 	if (error >= 0 && S_ISLNK(buf->st_mode) &&
-		(error = p_readlink(path, target, GIT_WIN_PATH)) >= 0)
+		(error = p_readlink(path, target, sizeof(target))) >= 0)
 		error = do_lstat(target, buf, 0);
 
 	return error;
@@ -315,23 +316,23 @@ int p_stat(const char* path, struct stat* buf)
 
 int p_chdir(const char* path)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	return _wchdir(buf);
 }
 
 int p_chmod(const char* path, mode_t mode)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	return _wchmod(buf, mode);
 }
 
 int p_rmdir(const char* path)
 {
 	int error;
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 
 	error = _wrmdir(buf);
 
@@ -347,24 +348,24 @@ int p_rmdir(const char* path)
 
 int p_hide_directory__w32(const char *path)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	return (SetFileAttributesW(buf, FILE_ATTRIBUTE_HIDDEN) != 0) ? 0 : -1;
 }
 
 char *p_realpath(const char *orig_path, char *buffer)
 {
 	int ret;
-	wchar_t orig_path_w[GIT_WIN_PATH];
-	wchar_t buffer_w[GIT_WIN_PATH];
+	git_win32_path orig_path_w;
+	git_win32_path buffer_w;
 
-	git__utf8_to_16(orig_path_w, GIT_WIN_PATH, orig_path);
+	git_win32_path_from_c(orig_path_w, orig_path);
 
 	/* Implicitly use GetCurrentDirectory which can be a threading issue */
-	ret = GetFullPathNameW(orig_path_w, GIT_WIN_PATH, buffer_w, NULL);
+	ret = GetFullPathNameW(orig_path_w, GIT_WIN_PATH_UTF16, buffer_w, NULL);
 
 	/* According to MSDN, a return value equals to zero means a failure. */
-	if (ret == 0 || ret > GIT_WIN_PATH)
+	if (ret == 0 || ret > GIT_WIN_PATH_UTF16)
 		buffer = NULL;
 
 	else if (GetFileAttributesW(buffer_w) == INVALID_FILE_ATTRIBUTES) {
@@ -448,18 +449,18 @@ int p_setenv(const char* name, const char* value, int overwrite)
 
 int p_access(const char* path, mode_t mode)
 {
-	wchar_t buf[GIT_WIN_PATH];
-	git__utf8_to_16(buf, GIT_WIN_PATH, path);
+	git_win32_path buf;
+	git_win32_path_from_c(buf, path);
 	return _waccess(buf, mode);
 }
 
 int p_rename(const char *from, const char *to)
 {
-	wchar_t wfrom[GIT_WIN_PATH];
-	wchar_t wto[GIT_WIN_PATH];
+	git_win32_path wfrom;
+	git_win32_path wto;
 
-	git__utf8_to_16(wfrom, GIT_WIN_PATH, from);
-	git__utf8_to_16(wto, GIT_WIN_PATH, to);
+	git_win32_path_from_c(wfrom, from);
+	git_win32_path_from_c(wto, to);
 	return MoveFileExW(wfrom, wto, MOVEFILE_REPLACE_EXISTING | MOVEFILE_COPY_ALLOWED) ? 0 : -1;
 }
 
@@ -513,10 +514,10 @@ p_gmtime_r (const time_t *timer, struct tm *result)
 #else
 #define DELTA_EPOCH_IN_MICROSECS  11644473600000000ULL
 #endif
- 
+
 #ifndef _TIMEZONE_DEFINED
 #define _TIMEZONE_DEFINED
-struct timezone 
+struct timezone
 {
 	int  tz_minuteswest; /* minutes W of Greenwich */
 	int  tz_dsttime;     /* type of dst correction */
diff --git a/src/win32/utf-conv.c b/src/win32/utf-conv.c
index c06f3a8c2..d4dbfbab9 100644
--- a/src/win32/utf-conv.c
+++ b/src/win32/utf-conv.c
@@ -70,12 +70,12 @@ void git__utf8_to_16(wchar_t *dest, size_t length, const char *src)
 }
 #endif
 
-int git__utf8_to_16(wchar_t *dest, size_t length, const char *src)
+int git__utf8_to_16(wchar_t * dest, size_t dest_size, const char *src)
 {
-	return MultiByteToWideChar(CP_UTF8, 0, src, -1, dest, (int)length);
+	return MultiByteToWideChar(CP_UTF8, 0, src, -1, dest, (int)dest_size);
 }
 
-int git__utf16_to_8(char *out, const wchar_t *input)
+int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src)
 {
-	return WideCharToMultiByte(CP_UTF8, 0, input, -1, out, GIT_WIN_PATH, NULL, NULL);
+	return WideCharToMultiByte(CP_UTF8, 0, src, -1, dest, (int)dest_size, NULL, NULL);
 }
diff --git a/src/win32/utf-conv.h b/src/win32/utf-conv.h
index 6cc9205f7..3af77580e 100644
--- a/src/win32/utf-conv.h
+++ b/src/win32/utf-conv.h
@@ -4,16 +4,35 @@
  * This file is part of libgit2, distributed under the GNU GPL v2 with
  * a Linking Exception. For full terms see the included COPYING file.
  */
+#ifndef INCLUDE_git_utfconv_h__
+#define INCLUDE_git_utfconv_h__
 
 #include <wchar.h>
+#include "common.h"
 
-#ifndef INCLUDE_git_utfconv_h__
-#define INCLUDE_git_utfconv_h__
+/* Maximum characters in a Windows path plus one for NUL byte */
+#define GIT_WIN_PATH_UTF16 (260 + 1)
 
-#define GIT_WIN_PATH (260 + 1)
+/* Maximum bytes necessary to convert a full-length UTF16 path to UTF8 */
+#define GIT_WIN_PATH_UTF8  (260 * 4 + 1)
 
-int git__utf8_to_16(wchar_t *dest, size_t length, const char *src);
-int git__utf16_to_8(char *dest, const wchar_t *src);
+typedef wchar_t git_win32_path[GIT_WIN_PATH_UTF16];
 
-#endif
+typedef char git_win32_path_as_utf8[GIT_WIN_PATH_UTF8];
 
+/* dest_size is the size of dest in wchar_t's */
+int git__utf8_to_16(wchar_t * dest, size_t dest_size, const char *src);
+/* dest_size is the size of dest in char's */
+int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src);
+
+GIT_INLINE(int) git_win32_path_from_c(git_win32_path dest, const char *src)
+{
+	return git__utf8_to_16(dest, GIT_WIN_PATH_UTF16, src);
+}
+
+GIT_INLINE(int) git_win32_path_to_c(git_win32_path_as_utf8 dest, const wchar_t *src)
+{
+	return git__utf16_to_8(dest, GIT_WIN_PATH_UTF8, src);
+}
+
+#endif