simplify and improve canonicalize_name

 - Replace scan and memmove with simple automaton
 - Convert forward to back slashes, lots of file systems treat them
   equivalenty, this saves us a few more checks down the road
 - Remove './' path components. We can saveley remove them in a string
   processing step instead of throwing an error. Also they actually appear
   often in tar balls, possibly not under user control.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

2 files changed, 41 insertions, 43 deletions

diff --git a/include/util.h b/include/util.h
index 28d21bc..d5b20dd 100644
--- a/include/util.h
+++ b/include/util.h
@@ -6,9 +6,9 @@
 #include <stdint.h>
 
 /*
-  Removes all preceeding and trailing slashes, shortens all sequences of
-  slashes to a single slash and returns failure state if one of the path
-  components is '..' or '.'.
+  Convert back to forward slashed, remove all preceeding and trailing slashes,
+  collapse all sequences of slashes, remove all path components that are '.'
+  and returns failure state if one of the path components is '..'.
 
   Returns 0 on success.
 */
diff --git a/lib/util/canonicalize_name.c b/lib/util/canonicalize_name.c
index 6703b66..57e4d66 100644
--- a/lib/util/canonicalize_name.c
+++ b/lib/util/canonicalize_name.c
@@ -1,56 +1,54 @@
 /* SPDX-License-Identifier: GPL-3.0-or-later */
-#include <string.h>
-
 #include "util.h"
 
-int canonicalize_name(char *filename)
+static void normalize_slashes(char *filename)
 {
-	char *ptr = filename;
-	int i;
-
-	while (*ptr == '/')
-		++ptr;
-
-	if (ptr != filename) {
-		memmove(filename, ptr, strlen(ptr) + 1);
-		ptr = filename;
-	}
-
-	while (*ptr != '\0') {
-		if (*ptr == '/') {
-			for (i = 0; ptr[i] == '/'; ++i)
-				;
-
-			if (i > 1)
-				memmove(ptr + 1, ptr + i, strlen(ptr + i) + 1);
-		}
-
-		if (ptr[0] == '/' && ptr[1] == '\0') {
-			*ptr = '\0';
-			break;
+	char *dst = filename, *src = filename;
+
+	while (*src == '/' || *src == '\\')
+		++src;
+
+	while (*src != '\0') {
+		if (*src == '/' || *src == '\\') {
+			while (*src == '/' || *src == '\\')
+				++src;
+			if (*src == '\0')
+				break;
+			*(dst++) = '/';
+		} else {
+			*(dst++) = *(src++);
 		}
-
-		++ptr;
 	}
 
-	ptr = filename;
+	*dst = '\0';
+}
 
-	while (*ptr != '\0') {
-		if (ptr[0] == '.') {
-			if (ptr[1] == '/' || ptr[1] == '\0')
-				return -1;
+int canonicalize_name(char *filename)
+{
+	char *dst = filename, *src = filename;
 
-			if (ptr[1] == '.' &&
-			    (ptr[2] == '/' || ptr[2] == '\0')) {
-				return -1;
+	normalize_slashes(filename);
+
+	while (*src != '\0') {
+		if (src[0] == '.') {
+			if (src[1] == '\0')
+				break;
+			if (src[1] == '/') {
+				src += 2;
+				continue;
 			}
+			if (src[1] == '.' && (src[2] == '/' || src[2] == '\0'))
+				return -1;
 		}
 
-		while (*ptr != '\0' && *ptr != '/')
-			++ptr;
-		if (*ptr == '/')
-			++ptr;
+		while (*src != '\0' && *src != '/')
+			*(dst++) = *(src++);
+
+		if (*src == '/')
+			*(dst++) = *(src++);
 	}
 
+	*dst = '\0';
+	normalize_slashes(filename);
 	return 0;
 }