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

6 files changed, 50 insertions, 29 deletions

diff --git a/src/extract.c b/src/extract.c
index 1b9cf72..21d2bb4 100644
--- a/src/extract.c
+++ b/src/extract.c
@@ -8,6 +8,7 @@
 #include "addrs.h"
 
 int bad_checksum, bad_checksum_count = 0;
+int new_file = 0;
 char *out_filename;
 
 void dpoke(int addr, u16 value) {
@@ -159,6 +160,7 @@ void extract_alf(void) {
 		}
 
 		bad_checksum = 0;
+		new_file = 1;
 		uncrunch_file();
 		if(bad_checksum) bad_checksum_count++;
 
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);
diff --git a/src/unalf.1 b/src/unalf.1
index 5d25618..34d1571 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 "2025-11-17" "0.1.0" "Urchlay's Atari 8-bit Tools"
+.TH "UNALF" 1 "2025-11-18" "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:
@@ -56,22 +56,24 @@ below for details.
 .INDENT 0.0
 .TP
 .B  \-a
-Convert text file line endings from Atari EOL to UNIX newline. Text
-files are detected by name: \fB*.txt\fP, \fB*.doc\fP, \fB*.lst\fP files
-are considered text.
+Convert text file line endings and tabs from ATASCII to ASCII. Text
+files are detected by looking at the first 2 bytes of the extracted
+file. If both are printable ASCII (or ATASCII EOL, aka \fB$9B\fP), the
+file is considered text.
 .sp
-Note that \fIonly\fP line\-endings are converted. Other ATASCII characters
-are left alone. If you need anything more in\-depth, use \fBa8cat\fP(1).
+Note that \fIonly\fP line\-endings and tabs are converted. Other ATASCII
+characters are left alone. If you need anything more in\-depth, use
+\fBa8cat\fP(1).
 .UNINDENT
-.\" convert EOLs in text files.
+.\" convert EOLs and tabs in text files.
 .
 .INDENT 0.0
 .TP
-.BI \-a\fB a
+.B \fB\-aa\fP
 Convert line endings in all extracted files. This will corrupt any
 executables or non\-text data files, so use with caution.
 .UNINDENT
-.\" convert EOLs in ALL files.
+.\" convert EOLs and tabs in ALL files.
 .
 .INDENT 0.0
 .TP
diff --git a/src/unalf.h b/src/unalf.h
index 422183a..12654ec 100644
--- a/src/unalf.h
+++ b/src/unalf.h
@@ -45,6 +45,7 @@ unsigned int getquad(int offs);
 
 /* extract.c */
 extern int bad_checksum_count;
+extern int new_file;
 void extract_alf(void);
 void dpoke(int addr, u16 value);
 u16 dpeek(int addr);
diff --git a/src/unalf.rst b/src/unalf.rst
index e23debd..4c3a658 100644
--- a/src/unalf.rst
+++ b/src/unalf.rst
@@ -44,20 +44,22 @@ OPTIONS
 =======
 
 -a
-  Convert text file line endings from Atari EOL to UNIX newline. Text
-  files are detected by name: **\*.txt**, **\*.doc**, **\*.lst** files
-  are considered text.
+  Convert text file line endings and tabs from ATASCII to ASCII. Text
+  files are detected by looking at the first 2 bytes of the extracted
+  file. If both are printable ASCII (or ATASCII EOL, aka **$9B**), the
+  file is considered text.
 
-  Note that *only* line-endings are converted. Other ATASCII characters
-  are left alone. If you need anything more in-depth, use **a8cat**\(1).
+  Note that *only* line-endings and tabs are converted. Other ATASCII
+  characters are left alone. If you need anything more in-depth, use
+  **a8cat**\(1).
 
-.. convert EOLs in text files.
+.. convert EOLs and tabs in text files.
 
--aa
+**-aa**
   Convert line endings in all extracted files. This will corrupt any
   executables or non-text data files, so use with caution.
 
-.. convert EOLs in ALL files.
+.. convert EOLs and tabs in ALL files.
 
 -d output-dir
   Write extracted files to this directory, which will be created if it
diff --git a/src/usage.c b/src/usage.c
index 554aa01..260a7c6 100644
--- a/src/usage.c
+++ b/src/usage.c
@@ -1,6 +1,5 @@
 const char *usage_msg[] = {
-	"  -a: convert EOLs in text files.",
-	"  -aa: convert EOLs in ALL files.",
+	"  -a: convert EOLs and tabs in text files.",
 	"  -d output-dir: set output directory (created if needed).",
 	"  -e: extract files (redundant; this is the default action).",
 	"  -h, --help: show this help message.",