Disallow gawk builtin/keyword as variable assignment.

5 files changed, 50 insertions, 16 deletions

diff --git a/ChangeLog b/ChangeLog
index 28603eda..1772c68d 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,12 @@
+Tue May 31 23:01:00 2011  John Haque      <j.eh@mchsi.com>
+
+	* re.c (check_bracket_exp): Use mem* routines instead of str*
+	for searching.
+	* main.c (arg_assign): Disallow builtin or user-defined function
+	as the name of a variable.
+	* awkgram.y (check_special): Rework so can be called from
+	arg_assign.
+
 Tue May 31 22:23:41 2011  Arnold D. Robbins  <arnold@skeeve.com>
 
 	In order to attain the goal of having ranges act like they are
diff --git a/awkgram.c b/awkgram.c
index b588bc80..4edec572 100644
--- a/awkgram.c
+++ b/awkgram.c
@@ -6180,9 +6180,6 @@ retry:
 			break_allowed++;
 		if (tokentab[mid].flags & CONTINUE)
 			continue_allowed++;
-		if ((do_traditional && (tokentab[mid].flags & GAWKX))
-		    || (do_posix && (tokentab[mid].flags & NOT_POSIX)))
-			goto out;
 
 		switch (class) {
 		case LEX_INCLUDE:
@@ -8646,8 +8643,12 @@ check_special(const char *name)
 			high = mid - 1;
 		else if (i > 0)		/* token > mid */
 			low = mid + 1;
-		else
+		else {
+			if ((do_traditional && (tokentab[mid].flags & GAWKX))
+					|| (do_posix && (tokentab[mid].flags & NOT_POSIX)))
+				return -1;
 			return mid;
+		}
 	}
 	return -1;
 }
diff --git a/awkgram.y b/awkgram.y
index d5b2a294..6b28b520 100644
--- a/awkgram.y
+++ b/awkgram.y
@@ -3490,9 +3490,6 @@ retry:
 			break_allowed++;
 		if (tokentab[mid].flags & CONTINUE)
 			continue_allowed++;
-		if ((do_traditional && (tokentab[mid].flags & GAWKX))
-		    || (do_posix && (tokentab[mid].flags & NOT_POSIX)))
-			goto out;
 
 		switch (class) {
 		case LEX_INCLUDE:
@@ -5956,8 +5953,12 @@ check_special(const char *name)
 			high = mid - 1;
 		else if (i > 0)		/* token > mid */
 			low = mid + 1;
-		else
+		else {
+			if ((do_traditional && (tokentab[mid].flags & GAWKX))
+					|| (do_posix && (tokentab[mid].flags & NOT_POSIX)))
+				return -1;
 			return mid;
+		}
 	}
 	return -1;
 }
diff --git a/main.c b/main.c
index 034de705..b30924b6 100644
--- a/main.c
+++ b/main.c
@@ -1137,6 +1137,7 @@ arg_assign(char *arg, int initing)
 	NODE *var;
 	NODE *it;
 	NODE **lhs;
+	long save_FNR;
 
 	if (! initing && disallow_var_assigns)
 		return FALSE;	/* --exec */
@@ -1155,6 +1156,12 @@ arg_assign(char *arg, int initing)
 
 	*cp++ = '\0';
 
+	/* avoid false source indications in a fatal message */
+	source = NULL;
+	sourceline = 0;
+	save_FNR = FNR;
+	FNR = 0;
+
 	/* first check that the variable name has valid syntax */
 	badvar = FALSE;
 	if (! isalpha((unsigned char) arg[0]) && arg[0] != '_')
@@ -1174,6 +1181,15 @@ arg_assign(char *arg, int initing)
 			lintwarn(_("`%s' is not a variable name, looking for file `%s=%s'"),
 				arg, arg, cp);
 	} else {
+		if (check_special(arg) >= 0)
+			fatal(_("cannot use gawk builtin `%s' as variable name"), arg);
+
+		if (! initing) {
+			var = lookup(arg);
+			if (var != NULL && var->type == Node_func)
+				fatal(_("cannot use function `%s' as variable name"), arg); 
+		}
+
 		/*
 		 * BWK awk expands escapes inside assignments.
 		 * This makes sense, so we do it too.
@@ -1214,7 +1230,7 @@ arg_assign(char *arg, int initing)
 
 	if (! initing)
 		*--cp = '=';	/* restore original text of ARGV */
-
+	FNR = save_FNR;
 	return ! badvar;
 }
 
diff --git a/re.c b/re.c
index 1f220b3e..a617831a 100644
--- a/re.c
+++ b/re.c
@@ -541,16 +541,20 @@ check_bracket_exp(char *s, size_t length)
 	int i;
 	int found = FALSE;
 	char save;
-	char *sp, *sp2;
+	char *sp, *sp2, *end;
 	int len;
 	int count = 0;
 
+	if (length == 0)
+		return;
+
+	end = s + length;
 	save = s[length];
 	s[length] = '\0';
 	sp = s;
 
 again:
-	sp = sp2 = strchr(sp, '[');
+	sp = sp2 = memchr(sp, '[', (end - sp));
 	if (sp == NULL)
 		goto done;
 
@@ -569,8 +573,10 @@ again:
 			warning(_("range of the form `[%c-%c]' is locale dependant"),
 					sp[-1], sp[1]);
 		}
-		if (count == 0)
+		if (count == 0) {
+			sp++;	/* skip past ']' */
 			break;
+		}
 	}
 
 	if (count > 0) {	/* bad regex, give up */
@@ -580,8 +586,11 @@ again:
 	/* sp2 has start */
 
 	for (i = 0; classes[i].name != NULL; i++) {
+		if (classes[i].warned)
+			continue;
 		len = classes[i].len;
-		if (strncmp(sp2, classes[i].name, len) == 0) {
+		if (   len == (sp - sp2)
+		    && memcmp(sp2, classes[i].name, len) == 0) {
 			found = TRUE;
 			break;
 		}
@@ -593,9 +602,7 @@ again:
 		classes[i].warned = TRUE;
 	}
 
-	if (*sp != '\0') {
-		if (*sp == ']')
-			sp++;
+	if (*sp < end) {
 		found = FALSE;
 		goto again;
 	}