Improve & document filename & size sanity-checking.

3 files changed, 76 insertions, 34 deletions

diff --git a/src/extract.c b/src/extract.c
index dc6a4f1..af3efeb 100644
--- a/src/extract.c
+++ b/src/extract.c
@@ -19,41 +19,23 @@ u16 dpeek(int addr) {
 	return mem[addr] | (mem[addr + 1] << 8);
 }
 
-void fix_filename(void) {
-	char *p;
-
-	if(!out_filename) return;
-
-	if(strlen(out_filename) > 12) {
-		fprintf(stderr, "%s: filename in ALF header not null-terminated, fixing.\n", self);
-		out_filename[12] = '\0';
-	}
-
-	if(opts.lowercase) {
-		for(p = out_filename; *p; p++)
-			*p = tolower(*p);
-	}
+static void bad_atari_filename(const char *msg) {
+	char fn[50] = "";
+	char xbuf[5];
+	int i;
+	u8 c;
 
-	if(!opts.keepdot) {
-		for(p = out_filename; *p; p++)
-			if(p[0] == '.' && !p[1])
-				*p = '\0';
+	for(i = 0; (c = (u8)out_filename[i]) && i < 12; i++) {
+		if(c < ' ' || c > '|') {
+			/* not printable, insert a hex escape */
+			sprintf(xbuf, "$%02X", c);
+			strcat(fn, xbuf);
+		} else {
+			strncat(fn, (char *)&c, 1);
+		}
 	}
-}
-
-void make_backup(void) {
-	/* up to 12-char FILENAME.EXT, plus a ~, plus null terminator = 14 */
-	char backup[14];
 
-	strncpy(backup, out_filename, 13);
-	strncat(backup, "~", 13);
-
-	/* silently ignore errors! */
-	rename(out_filename, backup);
-}
-
-static void bad_atari_filename(const char *msg) {
-	fprintf(stderr, "%s: bad Atari filename: %s\n", self, msg);
+	fprintf(stderr, "%s: bad Atari filename \"%s\": %s\n", self, fn, msg);
 }
 
 /* note to self: it's tempting to use isalpha(), isprint(), etc
@@ -93,6 +75,41 @@ static void sanity_check_filename(void) {
 		bad_atari_filename("invalid characters. corrupt ALF file?");
 }
 
+void fix_filename(void) {
+	char *p;
+
+	if(!out_filename) return;
+
+	if(strlen(out_filename) > 12) {
+		fprintf(stderr, "%s: filename in ALF header not null-terminated, fixing.\n", self);
+		out_filename[12] = '\0';
+	}
+
+	sanity_check_filename();
+
+	if(opts.lowercase) {
+		for(p = out_filename; *p; p++)
+			*p = tolower(*p);
+	}
+
+	if(!opts.keepdot) {
+		for(p = out_filename; *p; p++)
+			if(p[0] == '.' && !p[1])
+				*p = '\0';
+	}
+}
+
+void make_backup(void) {
+	/* up to 12-char FILENAME.EXT, plus a ~, plus null terminator = 14 */
+	char backup[14];
+
+	strncpy(backup, out_filename, 13);
+	strncat(backup, "~", 13);
+
+	/* silently ignore errors! */
+	rename(out_filename, backup);
+}
+
 void extract_alf(void) {
 	/* get ready to call fake 6502 stuff. set up memory like the Atari. */
 	dpoke(MEMTOP, 0xbc1f);
@@ -100,7 +117,6 @@ void extract_alf(void) {
 	while(read_alf_header()) {
 		out_filename = (char *)(mem + alf_hdr_filename);
 
-		sanity_check_filename();
 		fix_filename();
 
 		if(!file_wanted(out_filename)) {
diff --git a/src/unalf.1 b/src/unalf.1
index 9303f0b..b2f918f 100644
--- a/src/unalf.1
+++ b/src/unalf.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 "UNALF" 1 "2025-11-14" "0.1.0" "Urchlay's Atari 8-bit Tools"
+.TH "UNALF" 1 "2025-11-15" "0.1.0" "Urchlay's Atari 8-bit Tools"
 .SH NAME
 unalf \- extract Atari 8-bit ALF archives
 .\" RST source for unalf(1) man page. Convert with:
@@ -266,6 +266,19 @@ If you have an ALF file that does this, try it
 with the Atari \fBUNALF14.COM\fP and see if it has the same problem.
 If it does, you\(aqve found a bug in \fBunalf\fP, please send me the
 \&.alf file via email (see \fBAUTHOR\fP, below).
+.TP
+.B \fBbad Atari filename\fP \fI<filename>\fP \fI<reason>\fP
+The filename stored in the ALF header doesn\(aqt follow the rules for
+Atari DOS filenames. \fI<reason>\fP will be something like "doesn\(aqt start
+with A\-Z" or "invalid character". The filename will be printed with
+any unprintable characters as hex values (e.g. \fB$01\fP).
+.TP
+.B \fBheader #<n> (compressed|original) size is...\fP
+Followed by "impossibly large", "suspiciously large", or "too large
+to fit on a floppy disk". May indicate a corrupt archive, or someone
+really might have created an ALF file with files this big... though
+"impossibly large" means >16MB. \fBunalf\fP can\(aqt extract a file
+that big.
 .UNINDENT
 .SH NOTES
 .sp
diff --git a/src/unalf.rst b/src/unalf.rst
index b829bf9..2e716a1 100644
--- a/src/unalf.rst
+++ b/src/unalf.rst
@@ -221,6 +221,19 @@ Besides the standard error messages such as "no such file or directory":
   If it does, you've found a bug in **unalf**, please send me the
   .alf file via email (see **AUTHOR**, below).
 
+**bad Atari filename** *<filename>* *<reason>*
+  The filename stored in the ALF header doesn't follow the rules for
+  Atari DOS filenames. *<reason>* will be something like "doesn't start
+  with A-Z" or "invalid character". The filename will be printed with
+  any unprintable characters as hex values (e.g. **$01**).
+
+**header #<n> (compressed|original) size is...**
+  Followed by "impossibly large", "suspiciously large", or "too large
+  to fit on a floppy disk". May indicate a corrupt archive, or someone
+  really might have created an ALF file with files this big... though
+  "impossibly large" means >=16MB. **unalf** can't extract a file
+  that big.
+
 NOTES
 =====