From d7de9038eb0599232a8840ac48e97d9bb4462ab0 Mon Sep 17 00:00:00 2001
From: "B. Watson" <urchlay@slackware.uk>
Date: Mon, 8 Jul 2024 17:23:12 -0400
Subject: whichbas: improve BXL/BXE string array detection. Still not 100%
 perfect.

---
 whichbas.c | 37 +++++++++++++++++++++++++++++--------
 1 file changed, 29 insertions(+), 8 deletions(-)

diff --git a/whichbas.c b/whichbas.c
index 2b39a44..780af6a 100644
--- a/whichbas.c
+++ b/whichbas.c
@@ -313,7 +313,7 @@ CALLBACK(handle_op) {
 	   DIM A$(10,10) is illegal in Atari/Turbo.
 	   partial: this only works if the first dimension is either a
 	   constant or a scalar variable (not an array element or an
-	   expression).
+	   expression). fortunately most programs use constants in DIM.
 	 */
 	if(tok == OP_DIM_STR_LPAR) {
 		int str2d = 0;
@@ -373,11 +373,26 @@ CALLBACK(handle_op) {
 			}
 			break;
 		case 0x59: /* INKEY$ (0 arg pseudo-func) in TB, string array separator semicolon in BXL/BXE */
-			/* TODO: it might be better to detect INKEY$ by looking for end of line
-			   or statement, or OP_ARR_COMMA. INKEY$ can't be followed by anything else,
-			   AFAICT, but I'm not *that* certain yet... */
-			if(nexttok == OP_NUMCONST || nexttok == OP_HEXCONST || nexttok >= 0x80) {
-				/* INKEY$ may not be followed by a numeric constant or a variable of any kind */
+			/* partial: ...but pretty good. we *can't* check nexttok == OP_GRP_RPAR, because
+			   VAL(INKEY$) or ASC(INKEY$) are legit Turbo code.
+			   This can fail to catch A$(X;Y) if X and Y are both complex expressions. */
+			if(nexttok == OP_EOS || nexttok == OP_EOL) {
+				/* the semicolon can't be the last token on the line (needs at least
+				   a right-paren), but INKEY$ can. */
+				remove_type(BT_BXL_BXE);
+			} else if(pos == last_cmd_pos + 1) {
+				/* INKEY$ can be the first operator after the command, e.g if the command
+				   is IF. The semicolon cannot. */
+				remove_type(BT_BXL_BXE);
+			} else if(last_op_tok == OP_STR_ASSIGN) {
+				/* catches A$=INKEY$, for what that's worth. */
+				remove_type(BT_BXL_BXE);
+			} else if(last_op_tok == OP_NUMCONST || last_op_tok == OP_HEXCONST || last_op_tok == OP_STRCONST || last_op_tok >= 0x80) {
+				/* INKEY$ can't directly follow a constant or a variable. */
+				remove_type(BT_TURBO);
+			} else if(nexttok == OP_NUMCONST || nexttok == OP_HEXCONST || nexttok >= 0x80 || nexttok == OP_GRP_LPAR) {
+				/* INKEY$ may not be followed by a numeric constant or a variable of any kind, or
+				   a parenthesized expression. */
 				remove_type(BT_TURBO);
 			}
 			break;
@@ -587,9 +602,15 @@ void check_variables(void) {
 			bas_type = BT_TURBO;
 			print_result();
 		}
-		/* TODO: check for BXL/BXE string arrays, if I can ever
-		   figure out how to do it. */
 	}
+
+	/* I was hoping to check for BXL/BXE string arrays here. However,
+	   looking at a SAVEd file, they look identical to regular string
+	   variables (variable type $80, rest of the VVTP entry all $00).
+	   When the program's actually in memory, BXL/BXE sets the
+	   variable type byte to $91 for DIMed string array var and
+	   $81 (same as Atari/Turbo) for a regular DIMed string var.
+	   Unfortunately in the SAVE file, it's always $80. */
 }
 
 int main(int argc, char **argv) {
-- 
cgit v1.2.3