Disable writing to arbitrary elements of SYMTAB.

9 files changed, 41 insertions, 53 deletions

diff --git a/NEWS b/NEWS
index a25c30d7..b543b0fe 100644
--- a/NEWS
+++ b/NEWS
@@ -18,6 +18,9 @@ Changes from 4.2.x to 5.0.0
 4. PROCINFO["platform"] yields a string indicating the platform for
    which gawk was compiled.
 
+5. Writing to elements of SYMTAB that are not variable names now
+   causes a fatal error.
+
 Changes from 4.2.1 to 4.2.2
 ---------------------------
 
diff --git a/doc/ChangeLog b/doc/ChangeLog
index f31aae7f..261388a2 100644
--- a/doc/ChangeLog
+++ b/doc/ChangeLog
@@ -1,3 +1,9 @@
+2018-11-29         Arnold D. Robbins     <arnold@skeeve.com>
+
+	* gawktexi.in (Auto-set): Document that you can no longer use
+	arbitrary indices in SYMTAB.
+	* gawk.1: Ditto.
+
 2018-11-27         Arnold D. Robbins     <arnold@skeeve.com>
 
 	* gawktexi.in (Other Versions): Document GoAWK, an awk interpreter
diff --git a/doc/gawk.1 b/doc/gawk.1
index 5caff79b..4964cb3d 100644
--- a/doc/gawk.1
+++ b/doc/gawk.1
@@ -13,7 +13,7 @@
 .		if \w'\(rq' .ds rq "\(rq
 .	\}
 .\}
-.TH GAWK 1 "Nov 26 2018" "Free Software Foundation" "Utility Commands"
+.TH GAWK 1 "Nov 29 2018" "Free Software Foundation" "Utility Commands"
 .SH NAME
 gawk \- pattern scanning and processing language
 .SH SYNOPSIS
@@ -1425,7 +1425,8 @@ You may not use the
 .B delete
 statement with the
 .B SYMTAB
-array.
+array, nor assign to elements with an index that is
+not a variable name.
 .TP
 .B TEXTDOMAIN
 The text domain of the \*(AK program; used to find the localized
diff --git a/doc/gawktexi.in b/doc/gawktexi.in
index 9a21913b..cc667534 100644
--- a/doc/gawktexi.in
+++ b/doc/gawktexi.in
@@ -15253,7 +15253,8 @@ if an element in @code{SYMTAB} is an array.
 Also, you may not use the @code{delete} statement with the
 @code{SYMTAB} array.
 
-You may use an index for @code{SYMTAB} that is not a predefined identifier:
+Prior to @value{PVERSION} 5.0 of @command{gawk}, you could
+use an index for @code{SYMTAB} that was not a predefined identifier:
 
 @example
 SYMTAB["xxx"] = 5
@@ -15261,9 +15262,8 @@ print SYMTAB["xxx"]
 @end example
 
 @noindent
-This works as expected: in this case @code{SYMTAB} acts just like
-a regular array.  The only difference is that you can't then delete
-@code{SYMTAB["xxx"]}.
+This no longer works, instead producing a fatal error, as it led
+to rampant confusion.
 
 @cindex Schorr, Andrew
 The @code{SYMTAB} array is more interesting than it looks. Andrew Schorr
diff --git a/interpret.h b/interpret.h
index 4381a929..6a1a08b4 100644
--- a/interpret.h
+++ b/interpret.h
@@ -348,16 +348,19 @@ uninitialized_scalar:
 			 * 	3. Values that awk code stuck into SYMTAB not related to variables (Node_value)
 			 * For 1, since we are giving it a value, we have to change the type to Node_var.
 			 * For 1 and 2, we have to step through the Node_var to get to the value.
-			 * For 3, we just us the value we got from assoc_lookup(), above.
+			 * For 3, we fatal out. This avoids confusion on things like
+			 * SYMTAB["a foo"] = 42	# variable with a space in its name?
 			 */
 			if (t1 == func_table)
 				fatal(_("cannot assign to elements of FUNCTAB"));
-			else if (   t1 == symbol_table
-				 && (   (*lhs)->type == Node_var
+			else if (t1 == symbol_table) {
+				if ((   (*lhs)->type == Node_var
 				     || (*lhs)->type == Node_var_new)) {
-				update_global_values();		/* make sure stuff like NF, NR, are up to date */
-				(*lhs)->type = Node_var;	/* in case was Node_var_new */
-				lhs = & ((*lhs)->var_value);	/* extra level of indirection */
+					update_global_values();		/* make sure stuff like NF, NR, are up to date */
+					(*lhs)->type = Node_var;	/* in case was Node_var_new */
+					lhs = & ((*lhs)->var_value);	/* extra level of indirection */
+				} else
+					fatal(_("cannot assign to arbitrary elements of SYMTAB"));
 			}
 
 			assert(set_idx == NULL);
@@ -658,22 +661,25 @@ mod:
 			/*
 			 * Changing something in FUNCTAB is not allowed.
 			 *
-			 * SYMTAB is a little more messy.  Three kinds of values may
-			 * be stored in SYMTAB:
+			 * SYMTAB is a little more messy.  Three possibilities for SYMTAB:
 			 * 	1. Variables that don"t yet have a value (Node_var_new)
 			 * 	2. Variables that have a value (Node_var)
 			 * 	3. Values that awk code stuck into SYMTAB not related to variables (Node_value)
 			 * For 1, since we are giving it a value, we have to change the type to Node_var.
 			 * For 1 and 2, we have to step through the Node_var to get to the value.
-			 * For 3, we just us the value we got from assoc_lookup(), above.
+			 * For 3, we fatal out. This avoids confusion on things like
+			 * SYMTAB["a foo"] = 42	# variable with a space in its name?
 			 */
 			if (t1 == func_table)
 				fatal(_("cannot assign to elements of FUNCTAB"));
-			else if (   t1 == symbol_table
-				 && (   (*lhs)->type == Node_var
+			else if (t1 == symbol_table) {
+				if ((   (*lhs)->type == Node_var
 				     || (*lhs)->type == Node_var_new)) {
-				(*lhs)->type = Node_var;	/* in case was Node_var_new */
-				lhs = & ((*lhs)->var_value);	/* extra level of indirection */
+					update_global_values();		/* make sure stuff like NF, NR, are up to date */
+					(*lhs)->type = Node_var;	/* in case was Node_var_new */
+					lhs = & ((*lhs)->var_value);	/* extra level of indirection */
+				} else
+					fatal(_("cannot assign to arbitrary elements of SYMTAB"));
 			}
 
 			unref(*lhs);
diff --git a/test/Makefile.am b/test/Makefile.am
index e1ecc649..a44a1944 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -2101,8 +2101,7 @@ charasbytes:
 
 symtab6:
 	@echo $@
-	@$(AWK) -d__$@ -f "$(srcdir)"/$@.awk
-	@grep -v '^ENVIRON' __$@ | grep -v '^PROCINFO' > _$@ ; rm __$@
+	@$(AWK) -f "$(srcdir)"/$@.awk > _$@ 2>&1 || echo EXIT CODE: $$? >>_$@
 	@-$(CMP) "$(srcdir)"/$@.ok _$@ && rm -f _$@
 
 symtab8:
diff --git a/test/Makefile.in b/test/Makefile.in
index 1e3fe2ee..4c195884 100644
--- a/test/Makefile.in
+++ b/test/Makefile.in
@@ -2549,8 +2549,7 @@ charasbytes:
 
 symtab6:
 	@echo $@
-	@$(AWK) -d__$@ -f "$(srcdir)"/$@.awk
-	@grep -v '^ENVIRON' __$@ | grep -v '^PROCINFO' > _$@ ; rm __$@
+	@$(AWK) -f "$(srcdir)"/$@.awk > _$@ 2>&1 || echo EXIT CODE: $$? >>_$@
 	@-$(CMP) "$(srcdir)"/$@.ok _$@ && rm -f _$@
 
 symtab8:
diff --git a/test/symtab6.ok b/test/symtab6.ok
index 7de717a0..23a1633d 100644
--- a/test/symtab6.ok
+++ b/test/symtab6.ok
@@ -1,28 +1,2 @@
-ARGC: 1
-ARGIND: 0
-ARGV: array, 1 elements
-BINMODE: 0
-CONVFMT: "%.6g"
-ERRNO: ""
-FIELDWIDTHS: ""
-FILENAME: ""
-FNR: 0
-FPAT: "[^[:space:]]+"
-FS: " "
-FUNCTAB: array, 41 elements
-IGNORECASE: 0
-LINT: 0
-NF: 0
-NR: 0
-OFMT: "%.6g"
-OFS: " "
-ORS: "\n"
-PREC: 53
-RLENGTH: 0
-ROUNDMODE: "N"
-RS: "\n"
-RSTART: 0
-RT: ""
-SUBSEP: "\034"
-SYMTAB: array, 29 elements
-TEXTDOMAIN: "messages"
+gawk: ./symtab6.awk:1: fatal: cannot assign to arbitrary elements of SYMTAB
+EXIT CODE: 2
diff --git a/test/symtab7.ok b/test/symtab7.ok
index 28328831..37de1a49 100644
--- a/test/symtab7.ok
+++ b/test/symtab7.ok
@@ -1,2 +1,2 @@
-30
-40
+gawk: symtab7.awk:4: (FILENAME=- FNR=1) fatal: cannot assign to arbitrary elements of SYMTAB
+EXIT CODE: 2