listbas: add support for Turbo BASIC XL.

6 files changed, 296 insertions, 23 deletions

diff --git a/Makefile b/Makefile
index b8e29c7..7d33a91 100644
--- a/Makefile
+++ b/Makefile
@@ -63,8 +63,8 @@ vxrefbas: bas.o
 
 cxrefbas: bas.o bcdfp.o linetab.o
 
-listbas: listbas.c bas.o bcdfp.o tokens.o atables.o
-	$(CC) $(CFLAGS) -o listbas listbas.c bas.o bcdfp.o tokens.o atables.o -lm
+listbas: listbas.c bas.o bcdfp.o tokens.o atables.o turbo_tokens.c
+	$(CC) $(CFLAGS) -o listbas listbas.c bas.o bcdfp.o tokens.o atables.o turbo_tokens.c -lm
 
 bas.o: bas.c bas.h
 
diff --git a/listbas.1 b/listbas.1
index ba293f2..f8e4219 100644
--- a/listbas.1
+++ b/listbas.1
@@ -27,12 +27,12 @@ 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 "LISTBAS" 1 "2024-07-14" "0.2.1" "Urchlay's Atari 8-bit Tools"
+.TH "LISTBAS" 1 "2024-07-16" "0.2.1" "Urchlay's Atari 8-bit Tools"
 .SH NAME
 listbas \- List the source of a tokenized Atari 8-bit BASIC program
 .SH SYNOPSIS
 .sp
-listbas [\fB\-a\fP | \fB\-d\fP | \fB\-m\fP | \fB\-x\fP | \fB\-U\fP] [\fB\-b\fP] [\fB\-u\fP] [\fB\-i\fP] [\fB\-l\fP] [\fB\-n\fP | \fB\-C\fP] [\fB\-v\fP] [\fB\-c\fP \fIcolors\fP] \fBinput\-file\fP
+listbas [\fB\-a\fP | \fB\-d\fP | \fB\-m\fP | \fB\-x\fP | \fB\-U\fP] [\fB\-B\fP] [\fB\-u\fP] [\fB\-i\fP] [\fB\-l\fP] [\fB\-n\fP | \fB\-C\fP] [\fB\-v\fP] [\fB\-c\fP \fIcolors\fP] \fBinput\-file\fP
 .SH DESCRIPTION
 .sp
 \fBlistbas\fP acts like the \fILIST\fP command in BASIC. It reads a
@@ -77,6 +77,30 @@ because you can\(aqt tell what the dots are supposed to represent.
 .SS Other options
 .INDENT 0.0
 .TP
+.B \fB\-b\fP
+Set the BASIC the program was written in. Choices are:
+.INDENT 7.0
+.TP
+.B \fB\-ba\fP
+Program is Atari BASIC; this is the default.
+.TP
+.B \fB\-ba+\fP
+Program is OSS BASIC/A+.
+.TP
+.B \fB\-bt\fP
+Program is Turbo BASIC XL.
+.TP
+.B \fB\-bxl\fP
+Program is OSS BASIC XL.
+.TP
+.B \fB\-bxe\fP
+Program is OSS BASIC XE.
+.UNINDENT
+.sp
+If you see lots of "bad token XX" messages, or if the code just doesn\(aqt
+make any sense, you\(aqre using the wrong BASIC option. \fBwhichbas\fP(1)
+can (usually) detect the BASIC a program was written in.
+.TP
 .B \fB\-i\fP
 Include the immediate mode command (line 32768) in the output.
 .TP
@@ -94,7 +118,7 @@ No color. Has no effect if \fB\-a\fP is in effect, since this
 mode doesn\(aqt support color anyway. Disabling color does not
 disable reverse video.
 .TP
-.B \fB\-b\fP
+.B \fB\-B\fP
 Use bold for color output. This may make it easier to read on
 some terminals. Or, it may hurt your eyes...
 .TP
diff --git a/listbas.c b/listbas.c
index dcd303b..bfe7615 100644
--- a/listbas.c
+++ b/listbas.c
@@ -11,7 +11,20 @@
 #include "bas.h"
 #include "bcdfp.h"
 #include "tokens.h"
+#include "turbo_tokens.h"
+/*
+#include "aplus_tokens.h"
+#include "bxl_tokens.h"
+#include "bxe_tokens.h"
+*/
 #include "atables.h"
+#include "whichbas.h"
+
+#define B_ATARI SRET_ATARI
+#define B_TURBO SRET_TURBO
+#define B_APLUS SRET_APLUS
+#define B_BXL   SRET_BXL
+#define B_BXE   SRET_BXE
 
 #define COLOR_FMT "\x1b[%d;3%dm"  /* 1st %d is 1 for bold, 2nd is color */
 
@@ -50,6 +63,11 @@
 #define M_MAG     3   /* -m      */
 #define M_DOTS    4   /* -d      */
 
+const char *cmd_tokens[256];
+const char  *op_tokens[256];
+
+int bas_type = B_ATARI;
+
 int output_mode = M_UTF8;
 
 int bold =        0;  /* 1 with -b */
@@ -99,14 +117,37 @@ void parse_color_scheme(const char *arg) {
 	color_varnames  = parse_color(arg[7]);
 }
 
+int get_bas_type(char *arg) {
+	if(arg[0] == 't')
+		return B_TURBO;
+
+	if(arg[0] == 'a') {
+		if(arg[1] == '+')
+			return B_APLUS;
+		else
+			return B_ATARI;
+	}
+
+	if(arg[0] == 'x') {
+		if(arg[1] == 'l')
+			return B_BXL;
+		else if(arg[1] == 'e')
+			return B_BXE;
+	}
+
+	fprintf(stderr, "%s: Invalid BASIC type for -b option: %s\n", self, arg);
+	exit(1);
+}
+
 void print_help(void) {
-	printf("Usage: %s [-a|-d|-m|-x|-U] [-b] [-i] [-l] [-u] [-n|-C] [-v] [-c *colors*] <inputfile>\n", self);
+	printf("Usage: %s [-a|-d|-m|-x|-U] [-B] [-i] [-l] [-u] [-n|-C] [-v] [-c *colors*] <inputfile>\n", self);
+	printf("  -b <XX>: set BASIC type. XX is: a = atari, t = turbo, xl, xe, a+.\n");
 	printf("  -U: output ATASCII as Unicode/UTF-8 (this is the default).\n");
 	printf("  -a: output raw ATASCII.\n");
 	printf("  -d: use dots instead of Unicode/UTF-8.\n");
 	printf("  -m: magazine style listing (see a8cat(1)).\n");
 	printf("  -x: XL international character set (UTF-8).\n");
-	printf("  -b: use bold for color output.\n");
+	printf("  -B: use bold for color output.\n");
 	printf("  -i: show immediate mode command (line 32768).\n");
 	printf("  -l: don't print line numbers.\n");
 	printf("  -C: enable color syntax highlighting (this is the default).\n");
@@ -121,7 +162,7 @@ void parse_args(int argc, char **argv, int from_env) {
 
 	optind = 1;
 
-	while( (opt = getopt(argc, argv, "UCviamnbdhxulc:")) != -1) {
+	while( (opt = getopt(argc, argv, "b:UCviamnBdhxulc:")) != -1) {
 		switch(opt) {
 			case 'U': output_mode = M_UTF8;    break;
 			case 'a': output_mode = M_ATASCII; break;
@@ -130,11 +171,12 @@ void parse_args(int argc, char **argv, int from_env) {
 			case 'x': output_mode = M_UTF8_I;  break;
 			case 'v': verbose =     1; break;
 			case 'i': immediate =   1; break;
-			case 'b': bold =        1; break;
+			case 'B': bold =        1; break;
 			case 'u': underline =   1; break;
 			case 'C': color =       1; break;
 			case 'n': color =       0; break;
 			case 'l': skip_lineno = 1; break;
+			case 'b': bas_type = get_bas_type(optarg); break;
 			case 'c': parse_color_scheme(optarg); break;
 			case 'h': print_help(); exit(0);
 			default:
@@ -389,7 +431,7 @@ CALLBACK(print_cmd) {
 	if(tok == CMD_ILET) return;
 
 	if(color) color_on(color_cmd);
-	if(tok > last_command || (!(name = commands[tok]))) {
+	if((!(name = cmd_tokens[tok]))) {
 		fprintf(outfh, "(bad cmd token $%02x) ", tok);
 		badtok = 1;
 	} else {
@@ -398,10 +440,70 @@ CALLBACK(print_cmd) {
 	if(color) color_off();
 }
 
+void aplus_op_color_on(unsigned char tok) {
+}
+
+void op_color_on(unsigned char tok) {
+	if(!color) return;
+
+	if(bas_type == B_APLUS) {
+		aplus_op_color_on(tok);
+		return;
+	}
+
+	/* common ops for Atari, Turbo, BXL, BXE */
+	if(tok <= last_operator) {
+		if(tok > 0x3c)
+			color_on(color_func);
+		else if(tok == OP_UMINUS || tok == OP_UPLUS)
+			color_on(color_const); /* show leading sign in same color as the number */
+		else if((tok >= 0x17 && tok <= 0x1b) || (tok >= 0x28 && tok <= 0x2a))
+			color_on(color_cmd);
+		else
+			color_on(color_op);
+		return;
+	}
+
+	if(bas_type == B_TURBO) {
+		switch(tok) {
+			case 0x55:
+			case 0x58:
+			case 0x59: /* pseudo-func */
+			case 0x5b:
+			case 0x5c:
+			case 0x5e:
+			case 0x5f: /* pseudo-func */
+			case 0x60: /* pseudo-func */
+			case 0x63: /* pseudo-func */
+			case 0x64:
+			case 0x65:
+			case 0x6b:
+			case 0x6c: /* pseudo-func */
+			case 0x6d: /* pseudo-func */
+				color_on(color_func);
+				return;
+			case 0x5a:
+			case 0x5d:
+			case 0x61:
+			case 0x62:
+			case 0x6a:
+				color_on(color_cmd);
+				return;
+			default:
+				color_on(color_op);
+		}
+	}
+}
+
 CALLBACK(print_op) {
 	const char *name;
 
 	switch(tok) {
+		case OP_HEXCONST:
+			if(color) color_on(color_op);
+			outchr('$');
+			if(color) color_off();
+			/* fall thru */
 		case OP_NUMCONST:
 			print_number(pos + 1);
 			return;
@@ -413,17 +515,9 @@ CALLBACK(print_op) {
 		default: break;
 	}
 
-	if(color) {
-		if(tok > 0x3c)
-			color_on(color_func);
-		else if(tok == OP_UMINUS || tok == OP_UPLUS)
-			color_on(color_const); /* show leading sign in same color as the number */
-		else if((tok >= 0x17 && tok <= 0x1b) || (tok >= 0x28 && tok <= 0x2a))
-			color_on(color_cmd);
-		else
-			color_on(color_op);
-	}
-	if(tok > last_operator || (!(name = operators[tok]))) {
+	if(color) op_color_on(tok);
+
+	if((!(name = op_tokens[tok]))) {
 		fprintf(outfh, "(bad op token $%02x)", tok);
 		badtok = 1;
 	} else {
@@ -496,12 +590,58 @@ void list(void) {
 	walk_code(0, 32767 + immediate);
 }
 
+void init_bas_tables() {
+	memmove(cmd_tokens, commands, (last_command + 1) * sizeof(char *));
+	memmove(op_tokens, operators, (last_operator + 1) * sizeof(char *));
+}
+
+void init_aplus_tables() {
+	die("BASIC A+ not supported yet!");
+	/*
+	memmove(cmd_tokens, aplus_commands, (last_aplus_command + 1) * sizeof(char *));
+	memmove(op_tokens, aplus_operators, (last_aplus_operator + 1) * sizeof(char *));
+	*/
+}
+
+void init_turbo_tables() {
+	memmove(cmd_tokens + last_command  + 1, turbo_cmds, turbo_cmd_size);
+	memmove(op_tokens  + last_operator + 1, turbo_ops,  turbo_ops_size);
+}
+
+void init_bxl_tables() {
+	die("BASIC XL not supported yet!");
+}
+
+void init_bxe_tables() {
+	die("BASIC XE not supported yet!");
+}
+
+void init_token_tables() {
+	if(bas_type == B_APLUS) {
+		init_aplus_tables();
+		return;
+	}
+
+	init_bas_tables();
+
+	if(bas_type == B_TURBO)
+		init_turbo_tables();
+	else if(bas_type == B_BXL)
+		init_bxl_tables();
+	else if(bas_type == B_BXE)
+		init_bxe_tables();
+}
+
 int main(int argc, char **argv) {
 	set_self(*argv);
 	parse_general_args(argc, argv, print_help);
 	parse_env_args();
 	parse_args(argc, argv, 0);
 
+	if(bas_type != B_ATARI) allow_hex_const = 1;
+
+	init_token_tables();
+
 	readfile();
 	parse_header();
 
diff --git a/listbas.rst b/listbas.rst
index 63eb575..518dd5f 100644
--- a/listbas.rst
+++ b/listbas.rst
@@ -11,7 +11,7 @@ List the source of a tokenized Atari 8-bit BASIC program
 SYNOPSIS
 ========
 
-listbas [**-a** | **-d** | **-m** | **-x** | **-U**] [**-b**] [**-u**] [**-i**] [**-l**] [**-n** | **-C**] [**-v**] [**-c** *colors*] **input-file**
+listbas [**-a** | **-d** | **-m** | **-x** | **-U**] [**-B**] [**-u**] [**-i**] [**-l**] [**-n** | **-C**] [**-v**] [**-c** *colors*] **input-file**
 
 DESCRIPTION
 ===========
@@ -61,6 +61,28 @@ characters.
 Other options
 -------------
 
+**-b**
+  Set the BASIC the program was written in. Choices are:
+
+  **-ba**
+    Program is Atari BASIC; this is the default.
+
+  **-ba+**
+    Program is OSS BASIC/A+.
+
+  **-bt**
+    Program is Turbo BASIC XL.
+
+  **-bxl**
+    Program is OSS BASIC XL.
+
+  **-bxe**
+    Program is OSS BASIC XE.
+
+  If you see lots of "bad token XX" messages, or if the code just doesn't
+  make any sense, you're using the wrong BASIC option. **whichbas**\(1)
+  can (usually) detect the BASIC a program was written in.
+
 **-i**
   Include the immediate mode command (line 32768) in the output.
 
@@ -78,7 +100,7 @@ Other options
   mode doesn't support color anyway. Disabling color does not
   disable reverse video.
 
-**-b**
+**-B**
   Use bold for color output. This may make it easier to read on
   some terminals. Or, it may hurt your eyes...
 
diff --git a/turbo_tokens.c b/turbo_tokens.c
new file mode 100644
index 0000000..2278e21
--- /dev/null
+++ b/turbo_tokens.c
@@ -0,0 +1,82 @@
+/* these start at token 0x38 */
+const char *turbo_cmds[] = {
+	"DPOKE", /* $38 */
+	"MOVE",
+	"-MOVE",
+	"*F",
+	"REPEAT",
+	"UNTIL",
+	"WHILE",
+	"WEND",
+	"ELSE",  /* $40 */
+	"ENDIF",
+	"BPUT",
+	"BGET",
+	"FILLTO",
+	"DO",
+	"LOOP",
+	"EXIT",
+	"DIR",   /* $48 */
+	"LOCK",
+	"UNLOCK",
+	"RENAME",
+	"DELETE",
+	"PAUSE",
+	"TIME$=",
+	"PROC",
+	"EXEC",   /* $50 */
+	"ENDPROC",
+	"FCOLOR",
+	"*L",
+	"------------------------------",
+	"RENUM",
+	"DEL",
+	"DUMP",
+	"TRACE",  /* $58 */
+	"TEXT",
+	"BLOAD",
+	"BRUN",
+	"GO#",
+	"#",
+	"*B",
+	"PAINT",
+	"CLS",    /* $58 */
+	"DSOUND",
+	"CIRCLE",
+	"%PUT",
+	"%GET"    /* $64 */
+};
+
+const int turbo_cmd_size = sizeof(turbo_cmds);
+
+/* these start at token 0x55 */
+const char *turbo_ops[] = {
+	"DPEEK", /* $55 */
+	"&",
+	"!",
+	"INSTR", /* $58 */
+	"INKEY$",
+	" EXOR ",
+	"HEX$",
+	"DEC",
+	" DIV ",
+	"FRAC",
+	"TIME$",
+	"TIME",  /* $60 */
+	" MOD ",
+	" EXEC ",
+	"RND",
+	"RAND",
+	"TRUNC",
+	"%0",
+	"%1",
+	"%2",    /* $68 */
+	"%3",
+	" GO# ",
+	"UINSTR",
+	"ERR",
+	"ERL",   /* $6D */
+};
+
+const int turbo_ops_size = sizeof(turbo_ops);
+
diff --git a/turbo_tokens.h b/turbo_tokens.h
new file mode 100644
index 0000000..7696d52
--- /dev/null
+++ b/turbo_tokens.h
@@ -0,0 +1,5 @@
+extern const char *turbo_cmds[] ;
+extern const int turbo_cmd_size ;
+
+extern const char *turbo_ops[] ;
+extern const int turbo_ops_size ;