X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fpaths.c;h=62347c9fbd25027aec8ad183208e1941ebb85792;hp=9becb2128185d6dddc06c5251332c9ca4bf53bca;hb=eeb96703721a2ba6c64ec63f2c330515e8992b4f;hpb=ce7a3173855e654bf1aa52f1b7d7e676667192cd

diff --git a/src/paths.c b/src/paths.c
index 9becb212..62347c9f 100644
--- a/src/paths.c
+++ b/src/paths.c
@@ -33,93 +33,130 @@
 
 /* Like the basename() function, but does not modify @path; it just returns a
  * pointer to it.  This assumes the path separator is the
- * OS_PREFERRED_PATH_SEPARATOR. */
+ * OS_PREFERRED_PATH_SEPARATOR.  */
 const tchar *
 path_basename(const tchar *path)
 {
 	return path_basename_with_len(path, tstrlen(path));
 }
 
-/* Like path_basename(), but take an explicit string length. */
+/* Like path_basename(), but take an explicit string length.  */
 const tchar *
 path_basename_with_len(const tchar *path, size_t len)
 {
-	const tchar *p = &path[len] - 1;
-
-	/* Trailing slashes. */
-	while (1) {
-		if (p == path - 1)
-			return T("");
-		if (*p != OS_PREFERRED_PATH_SEPARATOR)
-			break;
-		p--;
-	}
+	const tchar *p = &path[len];
+
+	do {
+		if (p == path)
+			return &path[len];
+	} while (*--p == OS_PREFERRED_PATH_SEPARATOR);
 
-	while ((p != path - 1) && *p != OS_PREFERRED_PATH_SEPARATOR)
-		p--;
+	do {
+		if (p == path)
+			return &path[0];
+	} while (*--p != OS_PREFERRED_PATH_SEPARATOR);
 
-	return p + 1;
+	return ++p;
 }
 
 
-/*
- * Returns a pointer to the part of @path following the first colon in the last
- * path component, or NULL if the last path component does not contain a colon.
- */
+/* Returns a pointer to the part of @path following the first colon in the last
+ * path component, or NULL if the last path component does not contain a colon
+ * or has no characters following the first colon.  */
 const tchar *
 path_stream_name(const tchar *path)
 {
 	const tchar *base = path_basename(path);
 	const tchar *stream_name = tstrchr(base, T(':'));
-	if (!stream_name)
+	if (stream_name == NULL || *(stream_name + 1) == T('\0'))
 		return NULL;
 	else
 		return stream_name + 1;
 }
 
-
-/* Duplicate a path; return empty string for NULL input. */
+/* Duplicate a path; return empty string for NULL input.  */
 tchar *
 canonicalize_fs_path(const tchar *fs_path)
 {
-	if (!fs_path)
+	if (fs_path == NULL)
 		fs_path = T("");
 	return TSTRDUP(fs_path);
 }
 
-/* 
- * canonicalize_wim_path - Given a user-provided path to a file within a WIM
+/* Collapse and translate path separators, and strip trailing slashes.  Doesn't
+ * add or delete a leading slash.
+ *
+ * @in may alias @out.
+ */
+void
+do_canonicalize_path(const tchar *in, tchar *out)
+{
+	tchar *orig_out = out;
+
+	while (*in) {
+		if (is_any_path_separator(*in)) {
+			/* Collapse multiple path separators into one  */
+			*out++ = WIM_PATH_SEPARATOR;
+			do {
+				in++;
+			} while (is_any_path_separator(*in));
+		} else {
+			/* Copy non-path-separator character  */
+			*out++ = *in++;
+		}
+	}
+
+	/* Remove trailing slash if existent  */
+	if (out - orig_out > 1 && *(out - 1) == WIM_PATH_SEPARATOR)
+		--out;
+
+	*out = T('\0');
+}
+
+/*
+ * canonicalize_wim_path() - Given a user-provided path to a file within a WIM
  * image, translate it into a "canonical" path.
  *
- * To do this, we translate all supported path separators
- * (is_any_path_separator()) into the WIM_PATH_SEPARATOR, and strip any leading
- * and trailing slashes.  The returned string is allocated.  Note that there
- * still may be consecutive path separators within the string.  Furthermore, the
- * string may be empty, which indicates the root dentry of the WIM image.
+ * - Translate both types of slash into a consistent type (WIM_PATH_SEPARATOR).
+ * - Collapse path separators.
+ * - Add leading slash if missing.
+ * - Strip trailing slashes.
+ *
+ * Examples (with WIM_PATH_SEPARATOR == '/'):
+ *
+ * 		=> /		[ either NULL or empty string ]
+ * /		=> /
+ * \		=> /
+ * hello	=> /hello
+ * \hello	=> /hello
+ * \hello	=> /hello
+ * /hello/	=> /hello
+ * \hello/	=> /hello
+ * /hello//1	=> /hello/1
+ * \\hello\\1\\	=> /hello/1
  */
 tchar *
 canonicalize_wim_path(const tchar *wim_path)
 {
-	tchar *p;
-	tchar *canonical_path;
-
-	if (wim_path == NULL) {
-		wim_path = T("");
-	} else {
-		while (is_any_path_separator(*wim_path))
-			wim_path++;
-	}
-	canonical_path = TSTRDUP(wim_path);
-	if (canonical_path) {
-		for (p = canonical_path; *p; p++)
-			if (is_any_path_separator(*p))
-				*p = WIM_PATH_SEPARATOR;
-		for (p = tstrchr(canonical_path, T('\0')) - 1;
-		     p >= canonical_path && *p == WIM_PATH_SEPARATOR;
-		     p--)
-		{
-			*p = T('\0');
-		}
-	}
-	return canonical_path;
+	const tchar *in;
+	tchar *out;
+	tchar *result;
+
+	in = wim_path;
+	if (!in)
+		in = T("");
+
+	result = MALLOC((1 + tstrlen(in) + 1) * sizeof(result[0]));
+	if (!result)
+		return NULL;
+
+	out = result;
+
+	/* Add leading slash if missing  */
+	if (!is_any_path_separator(*in))
+		*out++ = WIM_PATH_SEPARATOR;
+
+	do_canonicalize_path(in, out);
+
+	return result;
 }