listamsb: sanity check end-of-line addresses.

3 files changed, 92 insertions, 8 deletions

diff --git a/listamsb.1 b/listamsb.1
index a21fb5a..a80a17d 100644
--- a/listamsb.1
+++ b/listamsb.1
@@ -27,7 +27,7 @@ level margin: \\n[rst2man-indent\\n[rst2man-indent-level]]
 .\" new: \\n[rst2man-indent\\n[rst2man-indent-level]]
 .in \\n[rst2man-indent\\n[rst2man-indent-level]]u
 ..
-.TH "LISTAMSB" 1 "2025-02-25" "0.2.1" "Urchlay's Atari 8-bit Tools"
+.TH "LISTAMSB" 1 "2025-02-26" "0.2.1" "Urchlay's Atari 8-bit Tools"
 .SH NAME
 listamsb \- List the source of a tokenized Atari Microsoft BASIC program
 .SH SYNOPSIS
@@ -116,6 +116,22 @@ unexpected EOF, file truncated?
 A tokenized AMSB file always ends with three null bytes (\fI$00\fP).
 This file doesn\(aqt. Probably the rest of the file is missing.
 .IP \(bu 2
+line \fIN\fP: EOL address \fIA\fP too (low|high)
+.sp
+Each line begins with the address of the next line. If the address
+is below \fI$0700\fP, it\(aqs below the minimum value of \fIMEMLO\fP, which
+is impossible. If it\(aqs above \fI$bc1f\fP, it would be in the display
+list, screen memory, or OS ROM (also impossible). File is
+corrupt, or not an AMSB file.
+.IP \(bu 2
+line \fIN\fP: EOL address \fIA\fP <= previous \fIB\fP
+.sp
+Corrupt file, or not an AMSB file.
+.IP \(bu 2
+line \fIN\fP EOL address doesn\(aqt match actual line length \fIL\fP
+.sp
+Same as above: corrupt, or not AMSB.
+.IP \(bu 2
 line number out of order
 .sp
 Probably means the file is corrupted, or isn\(aqt really an AMSB file.
diff --git a/listamsb.c b/listamsb.c
index 91e7d1c..b139dd5 100644
--- a/listamsb.c
+++ b/listamsb.c
@@ -5,18 +5,24 @@
 
 #include "amsbtok.h"
 
+/* this should always be defined in <stdio.h>, but just in case... */
+#ifndef BUFSIZ
+#define BUFSIZ 4096
+#endif
+
+/* range for one-byte tokens */
 #define MIN_STD_TOK 0x80 /* END */
 #define MAX_STD_TOK 0xf8 /* <   */
 
+/* range for 2nd byte of two-byte tokens (1st is always 0xff) */
 #define MIN_EXT_TOK 0xa3 /* SGN */
 #define MAX_EXT_TOK 0xc5 /* STACK */
 
-#ifndef BUFSIZ
-#define BUFSIZ 4096
-#endif
-
+/* good old Atari EOL character */
 #define EOL 0x9b
 
+/* every MS BASIC I ever saw had the same line number limit. it's
+   kind of arbitrary: why not allow the full range, up to 65535? */
 #define MAX_LINENO 63999
 
 /* a program bigger than this can't possibly fit into memory,
@@ -31,6 +37,14 @@
 #define MIN_PROGLEN 5
 #define EMPTY_PROGLEN 2
 
+/* an EOL address below this has to be an error, since
+   this is the lowest MEMLO can ever be. */
+#define MIN_PTR 0x0700
+
+/* an EOL address higher than this has to be an error, since
+	it would overlap the GR.0 display list or the ROMs at $c000 */
+#define MAX_PTR 0xbc1f
+
 const char *self;
 
 char pipe_command[BUFSIZ + 1] = { "a8cat" };
@@ -121,12 +135,15 @@ void unknown_token(int lineno, unsigned char byte, int ext) {
 
 int next_line(void) {
 	static int last_lineno = -1;
-	int ptr, lineno, was_ff, in_string;
+	static int last_ptr = -1;
+	int ptr, lineno, was_ff, in_string, offset, len;
 	unsigned char byte;
 
+	offset = bytes_read;
+
 	/* pointer to last token on the line, offset by whatever MEMLO
 		happened to be when the file was SAVEd. 0 means this is the
-		last line, otherwise we don't use its value. */
+		last line. */
 	ptr = read_word();
 	if(!ptr) {
 		if(verbose)
@@ -136,7 +153,25 @@ int next_line(void) {
 
 	lineno = read_word();
 	if(verbose)
-		fprintf(stderr, "found line number %d\n", lineno);
+		fprintf(stderr, "found line %d, offset %d, end-of-line %d\n", lineno, offset, ptr);
+
+	if(ptr < MIN_PTR) {
+		fprintf(stderr, "%s: line %d: EOL address $%04x too low (<$%04x)\n",
+			self, lineno, ptr, MIN_PTR);
+		warnings++;
+	} else if(ptr >= MAX_PTR) {
+		fprintf(stderr, "%s: line %d: EOL address $%04x too high (>$%04x)\n",
+			self, lineno, ptr, MAX_PTR);
+		warnings++;
+	}
+
+	if(last_ptr != -1) {
+		if(ptr <= last_ptr) {
+			fprintf(stderr, "%s: line %d: EOL address $%04x <= previous $%04x\n",
+				self, lineno, ptr, last_ptr);
+			warnings++;
+		}
+	}
 
 	if(lineno <= last_lineno) {
 		fprintf(stderr, "%s: line number out of order (%d <= %d)\n",
@@ -214,6 +249,23 @@ int next_line(void) {
 		}
 	}
 
+	len = bytes_read - offset;
+
+	if(verbose) {
+		fprintf(stderr, "      line %d length: %d\n", lineno, len);
+	}
+
+	if(last_ptr != -1) {
+		int plen = ptr - last_ptr;
+		if(len != plen) {
+			fprintf(stderr, "%s: line %d: EOL address doesn't match actual line length %d\n",
+					self, lineno, len);
+			warnings++;
+		}
+	}
+
+	last_ptr = ptr;
+
 	putc(EOL, outfile);
 
 	return 1;
diff --git a/listamsb.rst b/listamsb.rst
index acf3718..37eb0ea 100644
--- a/listamsb.rst
+++ b/listamsb.rst
@@ -100,6 +100,22 @@ continues processing.
   A tokenized AMSB file always ends with three null bytes (*$00*\).
   This file doesn't. Probably the rest of the file is missing.
 
+- line *N*: EOL address *A* too (low|high)
+
+  Each line begins with the address of the next line. If the address
+  is below *$0700*, it's below the minimum value of *MEMLO*, which
+  is impossible. If it's above *$bc1f*, it would be in the display
+  list, screen memory, or OS ROM (also impossible). File is
+  corrupt, or not an AMSB file.
+
+- line *N*: EOL address *A* <= previous *B*
+
+  Corrupt file, or not an AMSB file.
+
+- line *N* EOL address doesn't match actual line length *L*
+
+  Same as above: corrupt, or not AMSB.
+
 - line number out of order
 
   Probably means the file is corrupted, or isn't really an AMSB file.