From: Michihiro NAKAJIMA <ggcueroad@gmail.com>
Date: Thu, 8 Sep 2011 06:18:18 +0000 (-0400)
Subject: Fix issue 174.
X-Git-Tag: v3.0.0a~69
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=057322dfa8692c886eecf062be9fd80c862b7858;p=thirdparty%2Flibarchive.git

Fix issue 174.
Correctly handle Windows style full-path names, which are
leading \\.\ or \\?\ or \\?\\UNC\ or \\?\Volume{GUID}\ or C:\,
in archive_write_disk_windows.c.

SVN-Revision: 3692
---

diff --git a/libarchive/archive_write_disk_windows.c b/libarchive/archive_write_disk_windows.c
index 863dcdc9e..1dcf9d0dd 100644
--- a/libarchive/archive_write_disk_windows.c
+++ b/libarchive/archive_write_disk_windows.c
@@ -1864,6 +1864,22 @@ check_symlinks(struct archive_write_disk *a)
 	return (ARCHIVE_OK);
 }
 
+static int
+guidword(wchar_t *p, int n)
+{
+	int i;
+
+	for (i = 0; i < n; i++) {
+		if ((*p >= L'0' && *p <= L'9') ||
+		    (*p >= L'a' && *p <= L'f') ||
+		    (*p >= L'A' && *p <= L'F'))
+			p++;
+		else
+			return (-1);
+	}
+	return (0);
+}
+
 /*
  * Canonicalize the pathname.  In particular, this strips duplicate
  * '\' characters, '.' elements, and trailing '\'.  It also raises an
@@ -1873,20 +1889,95 @@ check_symlinks(struct archive_write_disk *a)
 static int
 cleanup_pathname(struct archive_write_disk *a)
 {
-	wchar_t *dest, *src, *p;
+	wchar_t *dest, *src, *p, *top;
 	wchar_t separator = L'\0';
 
-	dest = src = a->name;
-	if (*src == L'\0') {
+	p = a->name;
+	if (*p == L'\0') {
 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 		    "Invalid empty pathname");
 		return (ARCHIVE_FAILED);
 	}
 
-	for (p = a->name; *p != L'\0'; p++) {
+	/* Replace '/' by '\' */
+	for (; *p != L'\0'; p++) {
 		if (*p == L'/')
 			*p = L'\\';
-		/* Rewrite the path name if its character is a unusable. */
+	}
+	p = a->name;
+
+	/* Skip leading "\\.\" or "\\?\" or "\\?\UNC\" or
+	 * "\\?\Volume{GUID}\"
+	 * (absolute path prefixes used by Windows API) */
+	if (p[0] == L'\\' && p[1] == L'\\' &&
+	    (p[2] == L'.' || p[2] == L'?') && p[3] ==  L'\\')
+	{
+		/* A path begin with "\\?\UNC\" */
+		if (p[2] == L'?' &&
+		    (p[4] == L'U' || p[4] == L'u') &&
+		    (p[5] == L'N' || p[5] == L'n') &&
+		    (p[6] == L'C' || p[6] == L'c') &&
+		    p[7] == L'\\')
+			p += 8;
+		/* A path begin with "\\?\Volume{GUID}\" */
+		else if (p[2] == L'?' &&
+		    (p[4] == L'V' || p[4] == L'v') &&
+		    (p[5] == L'O' || p[5] == L'o') &&
+		    (p[6] == L'L' || p[6] == L'l') &&
+		    (p[7] == L'U' || p[7] == L'u') &&
+		    (p[8] == L'M' || p[8] == L'm') &&
+		    (p[9] == L'E' || p[9] == L'e') &&
+		    p[10] == L'{') {
+			if (guidword(p+11, 8) == 0 && p[19] == L'-' &&
+			    guidword(p+20, 4) == 0 && p[24] == L'-' &&
+			    guidword(p+25, 4) == 0 && p[29] == L'-' &&
+			    guidword(p+30, 4) == 0 && p[34] == L'-' &&
+			    guidword(p+35, 12) == 0 && p[47] == L'}' &&
+			    p[48] == L'\\')
+				p += 49;
+			else
+				p += 4;
+		/* A path begin with "\\.\PhysicalDriveX" */
+		} else if (p[2] == L'.' &&
+		    (p[4] == L'P' || p[4] == L'p') &&
+		    (p[5] == L'H' || p[5] == L'h') &&
+		    (p[6] == L'Y' || p[6] == L'y') &&
+		    (p[7] == L'S' || p[7] == L's') &&
+		    (p[8] == L'I' || p[8] == L'i') &&
+		    (p[9] == L'C' || p[9] == L'c') &&
+		    (p[9] == L'A' || p[9] == L'a') &&
+		    (p[9] == L'L' || p[9] == L'l') &&
+		    (p[9] == L'D' || p[9] == L'd') &&
+		    (p[9] == L'R' || p[9] == L'r') &&
+		    (p[9] == L'I' || p[9] == L'i') &&
+		    (p[9] == L'V' || p[9] == L'v') &&
+		    (p[9] == L'E' || p[9] == L'e') &&
+		    (p[10] >= L'0' && p[10] <= L'9') &&
+		    p[11] == L'\0') {
+			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
+			    "Path is a physical drive name");
+			return (ARCHIVE_FAILED);
+		} else
+			p += 4;
+	}
+
+	/* Skip leading drive letter from archives created
+	 * on Windows. */
+	if (((p[0] >= L'a' && p[0] <= L'z') ||
+	     (p[0] >= L'A' && p[0] <= L'Z')) &&
+		 p[1] == L':') {
+		if (p[2] == L'\0') {
+			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
+			    "Path is a drive name");
+			return (ARCHIVE_FAILED);
+		}
+		if (p[2] == L'\\')
+			p += 2;
+	}
+
+	top = dest = src = p;
+	/* Rewrite the path name if its character is a unusable. */
+	for (; *p != L'\0'; p++) {
 		if (*p == L':' || *p == L'*' || *p == L'?' || *p == L'"' ||
 		    *p == L'<' || *p == L'>' || *p == L'|')
 			*p = L'_';
@@ -1915,7 +2006,8 @@ cleanup_pathname(struct archive_write_disk *a)
 			} else if (src[1] == L'.') {
 				if (src[2] == L'\\' || src[2] == L'\0') {
 					/* Conditionally warn about '..' */
-					if (a->flags & ARCHIVE_EXTRACT_SECURE_NODOTDOT) {
+					if (a->flags &
+					    ARCHIVE_EXTRACT_SECURE_NODOTDOT) {
 						archive_set_error(&a->archive,
 						    ARCHIVE_ERRNO_MISC,
 						    "Path contains '..'");
@@ -1949,7 +2041,7 @@ cleanup_pathname(struct archive_write_disk *a)
 	 * We've just copied zero or more path elements, not including the
 	 * final '\'.
 	 */
-	if (dest == a->name) {
+	if (dest == top) {
 		/*
 		 * Nothing got copied.  The path must have been something
 		 * like '.' or '\' or './' or '/././././/./'.
diff --git a/libarchive/test/test_write_disk.c b/libarchive/test/test_write_disk.c
index 9a4475599..60bcdc24e 100644
--- a/libarchive/test/test_write_disk.c
+++ b/libarchive/test/test_write_disk.c
@@ -195,13 +195,13 @@ static void create_reg_file_win(struct archive_entry *ae, const char *msg)
 {
 	static const char data[]="abcdefghijklmnopqrstuvwxyz";
 	struct archive *ad;
-	struct stat st;
-	char *p, *fname;
+	struct _stat st;
+	wchar_t *p, *fname;
 	size_t l;
 
 	/* Write the entry to disk. */
 	assert((ad = archive_write_disk_new()) != NULL);
-        archive_write_disk_set_options(ad, ARCHIVE_EXTRACT_TIME);
+	archive_write_disk_set_options(ad, ARCHIVE_EXTRACT_TIME);
 	failure("%s", msg);
 	archive_entry_set_size(ae, sizeof(data));
 	archive_entry_set_mtime(ae, 123456789, 0);
@@ -211,25 +211,38 @@ static void create_reg_file_win(struct archive_entry *ae, const char *msg)
 	assertEqualInt(0, archive_write_free(ad));
 
 	/* Test the entries on disk. */
-	l = strlen(archive_entry_pathname(ae));
-	fname = malloc(l + 1);
+	l = wcslen(archive_entry_pathname_w(ae));
+	fname = malloc((l + 1) * sizeof(wchar_t));
 	assert(NULL != fname);
-	strcpy(fname, archive_entry_pathname(ae));
+	wcscpy(fname, archive_entry_pathname_w(ae));
+	p = fname;
+	/* Skip leading drive letter from archives created
+	 * on Windows. */
+	if (((p[0] >= L'a' && p[0] <= L'z') ||
+	     (p[0] >= L'A' && p[0] <= L'Z')) &&
+		 p[1] == L':' && p[2] == L'\\') {
+		p += 3;
+	}
 	/* Replace unusable characters in Windows to '_' */
-	for (p = fname; *p != '\0'; p++)
-		if (*p == ':' || *p == '*' || *p == '?' ||
-		    *p == '"' || *p == '<' || *p == '>' || *p == '|')
+	for (; *p != L'\0'; p++)
+		if (*p == L':' || *p == L'*' || *p == L'?' ||
+		    *p == L'"' || *p == L'<' || *p == L'>' || *p == L'|')
 			*p = '_';
-	assert(0 == stat(fname, &st));
+	assert(0 == _wstat(fname, &st));
 	failure("st.st_mode=%o archive_entry_mode(ae)=%o",
 	    st.st_mode, archive_entry_mode(ae));
 	assertEqualInt(st.st_size, sizeof(data));
+	free(fname);
 }
 #endif /* _WIN32 && !__CYGWIN__ */
 
 DEFINE_TEST(test_write_disk)
 {
 	struct archive_entry *ae;
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	wchar_t *fullpath;
+	DWORD l;
+#endif
 
 	/* Force the umask to something predictable. */
 	assertUmask(UMASK);
@@ -293,7 +306,7 @@ DEFINE_TEST(test_write_disk)
 #if defined(_WIN32) && !defined(__CYGWIN__)
 	/* A file with unusable characters in its file name. */
 	assert((ae = archive_entry_new()) != NULL);
-	archive_entry_copy_pathname(ae, "f:i*l?e\"f<i>l|e");
+	archive_entry_copy_pathname_w(ae, L"f:i*l?e\"f<i>l|e");
 	archive_entry_set_mode(ae, S_IFREG | 0755);
 	create_reg_file_win(ae, "Test creating a regular file"
 	    " with unusable characters in its file name");
@@ -301,10 +314,36 @@ DEFINE_TEST(test_write_disk)
 
 	/* A file with unusable characters in its directory name. */
 	assert((ae = archive_entry_new()) != NULL);
-	archive_entry_copy_pathname(ae, "d:i*r?e\"c<t>o|ry/file1");
+	archive_entry_copy_pathname_w(ae, L"d:i*r?e\"c<t>o|ry/file1");
+	archive_entry_set_mode(ae, S_IFREG | 0755);
+	create_reg_file_win(ae, "Test creating a regular file"
+	    " with unusable characters in its file name");
+	archive_entry_free(ae);
+
+	/* A full-path file with unusable characters in its file name. */
+	assert((l = GetCurrentDirectoryW(0, NULL)) != 0);
+	assert((fullpath = malloc((l + 20) * sizeof(wchar_t))) != NULL);
+	assert((l = GetCurrentDirectoryW(l, fullpath)) != 0);
+	wcscat(fullpath, L"\\f:i*l?e\"f<i>l|e");
+	assert((ae = archive_entry_new()) != NULL);
+	archive_entry_copy_pathname_w(ae, fullpath);
+	archive_entry_set_mode(ae, S_IFREG | 0755);
+	create_reg_file_win(ae, "Test creating a regular file"
+	    " with unusable characters in its file name");
+	archive_entry_free(ae);
+	free(fullpath);
+
+	/* A full-path file with unusable characters in its directory name. */
+	assert((l = GetCurrentDirectoryW(0, NULL)) != 0);
+	assert((fullpath = malloc((l + 30) * sizeof(wchar_t))) != NULL);
+	assert((l = GetCurrentDirectoryW(l, fullpath)) != 0);
+	wcscat(fullpath, L"\\d:i*r?e\"c<t>o|ry/file1");
+	assert((ae = archive_entry_new()) != NULL);
+	archive_entry_copy_pathname_w(ae, fullpath);
 	archive_entry_set_mode(ae, S_IFREG | 0755);
 	create_reg_file_win(ae, "Test creating a regular file"
 	    " with unusable characters in its file name");
 	archive_entry_free(ae);
+	free(fullpath);
 #endif /* _WIN32 && !__CYGWIN__ */
 }