1 files changed, 126 insertions, 43 deletions

diff --git a/src/path.c b/src/path.c
index 750dd3ef7..5beab97ed 100644
--- a/src/path.c
+++ b/src/path.c
@@ -9,10 +9,10 @@
 #include "posix.h"
 #ifdef GIT_WIN32
 #include "win32/posix.h"
+#include "win32/w32_util.h"
 #else
 #include <dirent.h>
 #endif
-#include <stdarg.h>
 #include <stdio.h>
 #include <ctype.h>
 
@@ -436,8 +436,12 @@ int git_path_walk_up(
 	while (scan >= stop) {
 		error = cb(data, &iter);
 		iter.ptr[scan] = oldc;
-		if (error < 0)
+
+		if (error) {
+			giterr_set_after_callback(error);
 			break;
+		}
+
 		scan = git_buf_rfind_next(&iter, '/');
 		if (scan >= 0) {
 			scan++;
@@ -483,33 +487,33 @@ bool git_path_isfile(const char *path)
 
 bool git_path_is_empty_dir(const char *path)
 {
-	HANDLE hFind = INVALID_HANDLE_VALUE;
-	git_win32_path wbuf;
-	int wbufsz;
-	WIN32_FIND_DATAW ffd;
-	bool retval = true;
-
-	if (!git_path_isdir(path))
-		return false;
-
-	wbufsz = git_win32_path_from_c(wbuf, path);
-	if (!wbufsz || wbufsz + 2 > GIT_WIN_PATH_UTF16)
-		return false;
-	memcpy(&wbuf[wbufsz - 1], L"\\*", 3 * sizeof(wchar_t));
-
-	hFind = FindFirstFileW(wbuf, &ffd);
-	if (INVALID_HANDLE_VALUE == hFind)
-		return false;
-
-	do {
-		if (!git_path_is_dot_or_dotdotW(ffd.cFileName)) {
-			retval = false;
-			break;
+	git_win32_path filter_w;
+	bool empty = false;
+
+	if (git_win32__findfirstfile_filter(filter_w, path)) {
+		WIN32_FIND_DATAW findData;
+		HANDLE hFind = FindFirstFileW(filter_w, &findData);
+
+		/* If the find handle was created successfully, then it's a directory */
+		if (hFind != INVALID_HANDLE_VALUE) {
+			empty = true;
+
+			do {
+				/* Allow the enumeration to return . and .. and still be considered
+				 * empty. In the special case of drive roots (i.e. C:\) where . and
+				 * .. do not occur, we can still consider the path to be an empty
+				 * directory if there's nothing there. */
+				if (!git_path_is_dot_or_dotdotW(findData.cFileName)) {
+					empty = false;
+					break;
+				}
+			} while (FindNextFileW(hFind, &findData));
+
+			FindClose(hFind);
 		}
-	} while (FindNextFileW(hFind, &ffd) != 0);
+	}
 
-	FindClose(hFind);
-	return retval;
+	return empty;
 }
 
 #else
@@ -528,7 +532,9 @@ bool git_path_is_empty_dir(const char *path)
 	if (!git_path_isdir(path))
 		return false;
 
-	if (!(error = git_buf_sets(&dir, path)))
+	if ((error = git_buf_sets(&dir, path)) != 0)
+		giterr_clear();
+	else
 		error = git_path_direach(&dir, 0, path_found_entry, NULL);
 
 	git_buf_free(&dir);
@@ -619,7 +625,7 @@ int git_path_find_dir(git_buf *dir, const char *path, const char *base)
 
 	/* call dirname if this is not a directory */
 	if (!error) /* && git_path_isdir(dir->ptr) == false) */
-		error = git_path_dirname_r(dir, dir->ptr);
+		error = (git_path_dirname_r(dir, dir->ptr) < 0) ? -1 : 0;
 
 	if (!error)
 		error = git_path_to_dir(dir);
@@ -777,8 +783,10 @@ int git_path_iconv(git_path_iconv_t *ic, char **in, size_t *inlen)
 		!git_path_has_non_ascii(*in, *inlen))
 		return 0;
 
+	git_buf_clear(&ic->buf);
+
 	while (1) {
-		if (git_buf_grow(&ic->buf, wantlen) < 0)
+		if (git_buf_grow(&ic->buf, wantlen + 1) < 0)
 			return -1;
 
 		nfc    = ic->buf.ptr   + ic->buf.size;
@@ -791,8 +799,11 @@ int git_path_iconv(git_path_iconv_t *ic, char **in, size_t *inlen)
 		if (rv != (size_t)-1)
 			break;
 
+		/* if we cannot convert the data (probably because iconv thinks
+		 * it is not valid UTF-8 source data), then use original data
+		 */
 		if (errno != E2BIG)
-			goto fail;
+			return 0;
 
 		/* make space for 2x the remaining data to be converted
 		 * (with per retry overhead to avoid infinite loops)
@@ -815,6 +826,64 @@ fail:
 	return -1;
 }
 
+static const char *nfc_file = "\xC3\x85\x73\x74\x72\xC3\xB6\x6D.XXXXXX";
+static const char *nfd_file = "\x41\xCC\x8A\x73\x74\x72\x6F\xCC\x88\x6D.XXXXXX";
+
+/* Check if the platform is decomposing unicode data for us.  We will
+ * emulate core Git and prefer to use precomposed unicode data internally
+ * on these platforms, composing the decomposed unicode on the fly.
+ *
+ * This mainly happens on the Mac where HDFS stores filenames as
+ * decomposed unicode.  Even on VFAT and SAMBA file systems, the Mac will
+ * return decomposed unicode from readdir() even when the actual
+ * filesystem is storing precomposed unicode.
+ */
+bool git_path_does_fs_decompose_unicode(const char *root)
+{
+	git_buf path = GIT_BUF_INIT;
+	int fd;
+	bool found_decomposed = false;
+	char tmp[6];
+
+	/* Create a file using a precomposed path and then try to find it
+	 * using the decomposed name.  If the lookup fails, then we will mark
+	 * that we should precompose unicode for this repository.
+	 */
+	if (git_buf_joinpath(&path, root, nfc_file) < 0 ||
+		(fd = p_mkstemp(path.ptr)) < 0)
+		goto done;
+	p_close(fd);
+
+	/* record trailing digits generated by mkstemp */
+	memcpy(tmp, path.ptr + path.size - sizeof(tmp), sizeof(tmp));
+
+	/* try to look up as NFD path */
+	if (git_buf_joinpath(&path, root, nfd_file) < 0)
+		goto done;
+	memcpy(path.ptr + path.size - sizeof(tmp), tmp, sizeof(tmp));
+
+	found_decomposed = git_path_exists(path.ptr);
+
+	/* remove temporary file (using original precomposed path) */
+	if (git_buf_joinpath(&path, root, nfc_file) < 0)
+		goto done;
+	memcpy(path.ptr + path.size - sizeof(tmp), tmp, sizeof(tmp));
+
+	(void)p_unlink(path.ptr);
+
+done:
+	git_buf_free(&path);
+	return found_decomposed;
+}
+
+#else
+
+bool git_path_does_fs_decompose_unicode(const char *root)
+{
+	GIT_UNUSED(root);
+	return false;
+}
+
 #endif
 
 #if defined(__sun) || defined(__GNU__)
@@ -848,6 +917,9 @@ int git_path_direach(
 
 	if ((dir = opendir(path->ptr)) == NULL) {
 		giterr_set(GITERR_OS, "Failed to open directory '%s'", path->ptr);
+		if (errno == ENOENT)
+			return GIT_ENOTFOUND;
+
 		return -1;
 	}
 
@@ -867,7 +939,7 @@ int git_path_direach(
 		if ((error = git_path_iconv(&ic, &de_path, &de_len)) < 0)
 			break;
 #endif
-				
+
 		if ((error = git_buf_put(path, de_path, de_len)) < 0)
 			break;
 
@@ -875,8 +947,8 @@ int git_path_direach(
 
 		git_buf_truncate(path, wd_len); /* restore path */
 
-		if (error) {
-			error = GIT_EUSER;
+		if (error != 0) {
+			giterr_set_after_callback(error);
 			break;
 		}
 	}
@@ -1043,15 +1115,8 @@ int git_path_dirload_with_stat(
 		}
 
 		if (S_ISDIR(ps->st.st_mode)) {
-			if ((error = git_buf_joinpath(&full, full.ptr, ".git")) < 0)
-				break;
-
-			if (p_access(full.ptr, F_OK) == 0) {
-				ps->st.st_mode = GIT_FILEMODE_COMMIT;
-			} else {
-				ps->path[ps->path_len++] = '/';
-				ps->path[ps->path_len] = '\0';
-			}
+			ps->path[ps->path_len++] = '/';
+			ps->path[ps->path_len] = '\0';
 		}
 	}
 
@@ -1062,3 +1127,21 @@ int git_path_dirload_with_stat(
 
 	return error;
 }
+
+int git_path_from_url_or_path(git_buf *local_path_out, const char *url_or_path)
+{
+	int error;
+
+	/* If url_or_path begins with file:// treat it as a URL */
+	if (!git__prefixcmp(url_or_path, "file://")) {
+		if ((error = git_path_fromurl(local_path_out, url_or_path)) < 0) {
+			return error;
+		}
+	} else { /* We assume url_or_path is already a path */
+		if ((error = git_buf_sets(local_path_out, url_or_path)) < 0) {
+			return error;
+		}
+	}
+
+	return 0;
+}