From 8e716325366e3dd9d3417c824d74fe5072fd0ac8 Mon Sep 17 00:00:00 2001
From: Lorenzooone <lollo.lollo.rbiz@gmail.com>
Date: Mon, 14 Aug 2023 18:12:00 +0200
Subject: [PATCH] Fix segfaults at EOF for scan_includes, and add token/newline
 checks (#1068)

---
 tools/scan_includes.c | 36 +++++++++++++++++++++++++++++++-----
 1 file changed, 31 insertions(+), 5 deletions(-)

diff --git a/tools/scan_includes.c b/tools/scan_includes.c
index e57ddc358..826af56cc 100644
--- a/tools/scan_includes.c
+++ b/tools/scan_includes.c
@@ -42,29 +42,48 @@ void scan_file(const char *filename, bool strict) {
 
 	for (char *ptr = contents; ptr && ptr - contents < size; ptr++) {
 		bool is_incbin = false, is_include = false;
+		ptr = strpbrk(ptr, ";\"Ii");
+		if (!ptr) {
+			break;
+		}
 		switch (*ptr) {
 		case ';':
-			ptr = strchr(ptr, '\n');
+			// Skip comments until the end of the line
+			ptr = strchr(ptr + 1, '\n');
 			if (!ptr) {
 				fprintf(stderr, "%s: no newline at end of file\n", filename);
+				goto done;
 			}
 			break;
 		case '"':
-			ptr++;
-			ptr = strchr(ptr, '"');
+			// Skip string iteral until the closing quote
+			ptr = strchr(ptr + 1, '"');
 			if (ptr) {
 				ptr++;
 			} else {
 				fprintf(stderr, "%s: unterminated string\n", filename);
+				goto done;
 			}
 			break;
 		case 'I':
 		case 'i':
+			/* empty statement between the label and the variable declaration */;
+			// Check that an INCLUDE/INCBIN starts as its own token
+			char before = ptr > contents ? *(ptr - 1) : '\n';
+			if (before != ' ' && before != '\t' && before != '\n' && before != ':') {
+				break;
+			}
 			is_incbin = !strncmp(ptr, "INCBIN", 6) || !strncmp(ptr, "incbin", 6);
 			is_include = !strncmp(ptr, "INCLUDE", 7) || !strncmp(ptr, "include", 7);
 			if (is_incbin || is_include) {
-				ptr = strchr(ptr, '"');
-				if (ptr) {
+				// Check that an INCLUDE/INCBIN ends as its own token
+				ptr += is_include ? 7 : 6;
+				if (*ptr != ' ' && *ptr != '\t' && *ptr != '\n' && *ptr != '"') {
+					break;
+				}
+				ptr += strspn(ptr, " \t");
+				if (*ptr == '"') {
+					// Print the file path and recursively scan INCLUDEs
 					ptr++;
 					char *include_path = ptr;
 					size_t length = strcspn(ptr, "\"");
@@ -74,12 +93,19 @@ void scan_file(const char *filename, bool strict) {
 					if (is_include) {
 						scan_file(include_path, strict);
 					}
+				} else {
+					fprintf(stderr, "%s: no file path after INC%s\n", filename, is_include ? "LUDE" : "BIN");
+					// Continue to process a comment
+					if (*ptr == ';') {
+						ptr--;
+					}
 				}
 			}
 			break;
 		}
 	}
 
+done:
 	free(contents);
 }