Improve parsing of symbols.

In particular, don't confuse `4..` with a symbol.

TODO

@@ -4,10 +4,6 @@ listing format errors: ignore whitespace of input
 
 documentation: print supported directives
 
-register symbols: %3+1 is the same as %4 (pdf page 3-9 aka 35),
-but where precisely register symbols can be used, and how,
-isn't specified.
-
 ---------------------------------------
 I was not able to locate a Macro-11 language reference manual any more
 recent than for RT11 version *3*, so I used that plus my recollection

parse.c

@@ -743,23 +743,23 @@ char           *get_symbol(
 {
     int             len;
     char           *symcp;
-    int             digits = 0;
+    int             start_digit = 0;
+    int             not_digits = 0;
 
     cp = skipwhite(cp);                /* Skip leading whitespace */
 
     if (!issym((unsigned char)*cp))
         return NULL;
 
-    digits = 0;
     if (isdigit((unsigned char)*cp))
-        digits = 2;                    /* Think about digit count */
+        start_digit = 1;
 
     for (symcp = cp + 1; issym((unsigned char)*symcp); symcp++) {
         if (!isdigit((unsigned char)*symcp)) /* Not a digit? */
-            digits--;                  /* Make a note. */
+            not_digits++;                    /* Make a note. */
     }
 
-    if (digits == 2)
+    if (start_digit && not_digits == 0)
         return NULL;                   /* Not a symbol, it's a digit string */
 
     if (endp)
@@ -780,9 +780,9 @@ char           *get_symbol(
     if (islocal) {
         *islocal = 0;
 
-        /* Turn to local label format */
+        /* Check if local label format */
-        if (digits == 1) {
+        if (start_digit) {
-            if (symcp[len - 1] == '$') {
+            if (not_digits == 1 && symcp[len - 1] == '$') {
                 char           *newsym = memcheck(malloc(32));  /* Overkill */
 
                 sprintf(newsym, "%ld$%d", strtol(symcp, NULL, 10), lsb);
@@ -792,8 +792,7 @@ char           *get_symbol(
                 }
                 free(symcp);
                 symcp = newsym;
-                if (islocal)
+                *islocal = SYMBOLFLAG_LOCAL;
-                    *islocal = SYMBOLFLAG_LOCAL;
                 lsb_used++;
             } else {
                 free(symcp);
@@ -801,8 +800,8 @@ char           *get_symbol(
             }
         }
     } else {
-        /* disallow local label format */
+        /* Disallow local label format */
-        if (isdigit((unsigned char)*symcp)) {
+        if (start_digit) {
             free(symcp);
             return NULL;
         }

tests/RunTests

@@ -33,6 +33,7 @@ TESTS="test-asciz \
     test-reg \
     test-reloc \
     test-rept \
+    test-syntax \
     test-ua-pl \
     test-undef \
     test-word-comma"

tests/test-syntax.lst.ok

@@ -0,0 +1,64 @@
+       1                                ;;;;
+       2                                ;
+       3                                ; Test some aspects of syntax.
+       4                                ;
+       5                                ; Reference MACRO11 does very weird things with these errors.
+       6                                ; Fortunately I don't plan to produce exactly the same results in case of errors.
+       7                                ; It seems to recognise an operand where there is none (or none yet)...
+       8                                ;
+       9                                ; AQ   37 000022  012767  000004  177772          mov     #4..,r0
+      10                                ; AQ   38 000030  012767  000011  177772          mov     #9..,r0
+      11                                ; AQ   39 000036  012767  000004  000000G         mov     #4.$,r0
+      12                                ; AQU  40 000044  012767  000000  177772          mov     #4$.,r0
+      13                                ;      41
+      14                                ; A    42 000052  012767  000004  177772          mov     #4..
+      15                                ; A    43 000060  012767  000004  000000G         mov     #4.$
+      16                                ; AU   44 000066  012767  000000  177772          mov     #4$.
+      17                                
+test-syntax.mac:18: ***ERROR Invalid syntax (comma expected)
+      18                                	mov	#4..,r0		; bad syntax
+test-syntax.mac:19: ***ERROR Invalid syntax (comma expected)
+      19                                	mov	#4$.,r0		; bad syntax
+test-syntax.mac:20: ***ERROR Invalid syntax (comma expected)
+      20                                	mov	#4.$,r0		; bad syntax
+test-syntax.mac:21: ***ERROR Invalid syntax (comma expected)
+      21                                	mov	#4$$,r0		; bad syntax
+      22                                
+test-syntax.mac:23: ***ERROR Invalid syntax (comma expected)
+      23                                	mov	#4..		; bad syntax
+test-syntax.mac:24: ***ERROR Invalid syntax (comma expected)
+      24                                	mov	#4$.		; bad syntax
+test-syntax.mac:25: ***ERROR Invalid syntax (comma expected)
+      25                                	mov	#4.$		; bad syntax
+test-syntax.mac:26: ***ERROR Invalid syntax (comma expected)
+      26                                	mov	#4$$		; bad syntax
+      27                                
+      28                                ;; page 2-4:
+      29                                ;; Multiple expressions used in the operand field of a MACRO-11 statement
+      30                                ;; must be separated by a comma;
+      31                                ;; multiple symbols similarly must be delimited by a valid separator
+      32                                ;; (a comma, tab, and/or space).
+      33                                ;; When the operator field contains an op code, associated operands are
+      34                                ;; always expressions, ...
+      35                                
+      36        000001                  a=1
+test-syntax.mac:37: ***ERROR Invalid syntax (comma expected)
+      37                                        mov     #4 r0
+test-syntax.mac:38: ***ERROR Invalid syntax (comma expected)
+      38                                        mov     a r0
+      39                                
+      40                                ;; page 2-3:
+      41                                ;; An operator is terminated by a space, tab, or any non-Radix-50 character,
+      42 000000 017700  000001'                 mov@a,r0
+      42                                
+
+
+Symbol table
+
+.       000004R      001 A     = 000001           
+
+
+Program sections:
+
+. ABS.  000000    000   (RW,I,GBL,ABS,OVR,NOSAV)
+        000004    001   (RW,I,LCL,REL,CON,NOSAV)

tests/test-syntax.mac

@@ -0,0 +1,42 @@
+;;;;
+;
+; Test some aspects of syntax.
+;
+; Reference MACRO11 does very weird things with these errors.
+; Fortunately I don't plan to produce exactly the same results in case of errors.
+; It seems to recognise an operand where there is none (or none yet)...
+;
+; AQ   37 000022  012767  000004  177772          mov     #4..,r0
+; AQ   38 000030  012767  000011  177772          mov     #9..,r0
+; AQ   39 000036  012767  000004  000000G         mov     #4.$,r0
+; AQU  40 000044  012767  000000  177772          mov     #4$.,r0
+;      41
+; A    42 000052  012767  000004  177772          mov     #4..
+; A    43 000060  012767  000004  000000G         mov     #4.$
+; AU   44 000066  012767  000000  177772          mov     #4$.
+
+	mov	#4..,r0		; bad syntax
+	mov	#4$.,r0		; bad syntax
+	mov	#4.$,r0		; bad syntax
+	mov	#4$$,r0		; bad syntax
+
+	mov	#4..		; bad syntax
+	mov	#4$.		; bad syntax
+	mov	#4.$		; bad syntax
+	mov	#4$$		; bad syntax
+
+;; page 2-4:
+;; Multiple expressions used in the operand field of a MACRO-11 statement
+;; must be separated by a comma;
+;; multiple symbols similarly must be delimited by a valid separator
+;; (a comma, tab, and/or space).
+;; When the operator field contains an op code, associated operands are
+;; always expressions, ...
+
+a=1
+        mov     #4 r0
+        mov     a r0
+
+;; page 2-3:
+;; An operator is terminated by a space, tab, or any non-Radix-50 character,
+        mov@a,r0