Better text file detection for -aa; convert tabs with -a or -aa.

1 files changed, 25 insertions, 10 deletions

diff --git a/src/io.c b/src/io.c
index 2323700..1896df9 100644
--- a/src/io.c
+++ b/src/io.c
@@ -5,6 +5,7 @@
 #include "addrs.h"
 
 static int headers_read = 0;
+static int convert_eols = 0;
 
 static void die_arc(void) {
 	fprintf(stderr, "%s: this is an ARC file, not ALF\n", self);
@@ -104,27 +105,41 @@ void readblock(void) {
 	dpoke(buf_len_l, bytes);
 }
 
-int is_text_file(char *fn) {
-	if(globmatch("*.txt", fn)) return 1;
-	if(globmatch("*.doc", fn)) return 1;
-	if(globmatch("*.lst", fn)) return 1;
-	return 0;
+static int is_printable(u8 c) {
+	return (c == 0x9b || (c >= ' ' && c <= 124));
+}
+
+static int is_text_file(u8 *buf) {
+	return is_printable(buf[0]) && is_printable(buf[1]);
 }
 
-/* mirror of readblock() */
+/* mirror of readblock(), plus EOL conversion if needed. With -a,
+   a file is considered text if its first 2 bytes are printable ATASCII,
+   including EOLs. With -aa, all files are converted. */
 void writeblock(void) {
 	int i, bytes, len, bufadr;
 	u8 *buf;
-	extern char *out_filename;
 
 	bufadr = dpeek(buf_adr_l);
 	buf = mem + bufadr;
 	len = dpeek(buf_len_l);
 
-	if(opts.txtconv) {
-		if(opts.txtconv > 1 || is_text_file(out_filename))
-		for(i = 0; i < len; i++)
+	if(new_file) {
+		if(opts.txtconv > 1) {
+			convert_eols = 1;
+		} else if(opts.txtconv == 1 && len > 1) {
+			convert_eols = is_text_file(buf);
+		} else {
+			convert_eols = 0;
+		}
+	}
+	new_file = 0;
+
+	if(convert_eols) {
+		for(i = 0; i < len; i++) {
 			if(buf[i] == 0x9b) buf[i] = '\n';
+			if(buf[i] == 0x7f) buf[i] = '\t';
+		}
 	}
 
 	// fprintf(stderr, "writeblock, bufadr = $%04x, len = $%04x\n", bufadr, len);