fixed bugs of not handling <NL> properly in local/global variable declarations.

qse/include/qse/awk/awk.h

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.h 213 2009-06-26 13:05:19Z hyunghwan.chung $
+ * $Id: awk.h 214 2009-06-27 02:50:54Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -609,6 +609,7 @@ enum qse_awk_errnum_t
 	QSE_AWK_EDUPLCL,        /* duplicate local variable name */
 	QSE_AWK_EBADPAR,        /* not a valid parameter name */
 	QSE_AWK_EBADVAR,        /* not a valid variable name */
+	QSE_AWK_EVARMS,         /* variable name missing */
 	QSE_AWK_EUNDEF,         /* undefined identifier */
 	QSE_AWK_ELVALUE,        /* l-value required */
 	QSE_AWK_EGBLTM,         /* too many global variables */

qse/lib/awk/err.c

@@ -1,5 +1,5 @@
 /*
- * $Id: err.c 213 2009-06-26 13:05:19Z hyunghwan.chung $
+ * $Id: err.c 214 2009-06-27 02:50:54Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -99,6 +99,7 @@ const qse_char_t* qse_awk_dflerrstr (qse_awk_t* awk, qse_awk_errnum_t errnum)
 		QSE_T("duplicate local variable '${0}'"),
 		QSE_T("'${0}' not a valid parameter name"),
 		QSE_T("'${0}' not a valid variable name"),
+		QSE_T("variable name missing"),
 		QSE_T("undefined identifier '${0}'"),
 		QSE_T("l-value required"),
 		QSE_T("too many global variables"),

qse/lib/awk/parse.c

@@ -1,5 +1,5 @@
 /*
- * $Id: parse.c 213 2009-06-26 13:05:19Z hyunghwan.chung $
+ * $Id: parse.c 214 2009-06-27 02:50:54Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -387,10 +387,14 @@ static global_t gtab[] =
 		qse_awk_seterror ((awk), (code), (line), &errarg); \
 	} while (0)
 
+#define MATCH_TERMINATOR_NORMAL(awk) \
+	(MATCH((awk),TOKEN_SEMICOLON) || MATCH((awk),TOKEN_NEWLINE))
+
+#define MATCH_TERMINATOR_RBRACE(awk) \
+	((awk->option & QSE_AWK_NEWLINE) && MATCH((awk),TOKEN_RBRACE))
+
 #define MATCH_TERMINATOR(awk) \
-	(MATCH((awk),TOKEN_SEMICOLON) || \
+	(MATCH_TERMINATOR_NORMAL(awk) || MATCH_TERMINATOR_RBRACE(awk))
-	 MATCH((awk),TOKEN_NEWLINE) || \
-	 ((awk->option & QSE_AWK_NEWLINE) && MATCH((awk),TOKEN_RBRACE)))
 
 qse_size_t qse_awk_getmaxdepth (qse_awk_t* awk, int type)
 {
@@ -592,7 +596,7 @@ exit_parse:
 static qse_awk_t* parse_progunit (qse_awk_t* awk)
 {
 	/*
-	gbl xxx, xxxx;
+	global xxx, xxxx;
 	BEGIN { action }
 	END { action }
 	pattern { action }
@@ -1204,6 +1208,12 @@ static qse_awk_nde_t* parse_block (
 	{
 		while (1) 
 		{
+			/* skip new lines before local declaration in a block*/
+			while (MATCH(awk,TOKEN_NEWLINE))
+			{
+				if (get_token(awk) <= -1) return QSE_NULL;
+			}
+
 			if (!MATCH(awk,TOKEN_LOCAL)) break;
 
 			if (get_token(awk) <= -1) 
@@ -1624,6 +1634,14 @@ int qse_awk_delgbl (
 
 static qse_awk_t* collect_globals (qse_awk_t* awk)
 {
+	if (MATCH(awk,TOKEN_NEWLINE))
+	{
+		/* special check if the first name is on the 
+		 * same line when QSE_AWK_NEWLINE is on */
+		SETERR (awk, QSE_AWK_EVARMS);
+		return QSE_NULL;
+	}
+
 	while (1) 
 	{
 		if (!MATCH(awk,TOKEN_IDENT)) 
@@ -1640,7 +1658,19 @@ static qse_awk_t* collect_globals (qse_awk_t* awk)
 
 		if (get_token(awk) <= -1) return QSE_NULL;
 
-		if (MATCH_TERMINATOR(awk)) break;
+		if (MATCH_TERMINATOR_NORMAL(awk)) 
+		{
+			/* skip a terminator (;, <NL>) */
+			if (get_token(awk) <= -1) return QSE_NULL;
+			break;
+		}
+
+		/*
+		 * unlike collect_locals(), the right brace cannot
+		 * terminate a global declaration as it can never be
+		 * placed within a block. 
+		 * so do not perform MATCH_TERMINATOR_RBRACE(awk))
+		 */
 
 		if (!MATCH(awk,TOKEN_COMMA)) 
 		{
@@ -1648,11 +1678,13 @@ static qse_awk_t* collect_globals (qse_awk_t* awk)
 			return QSE_NULL;
 		}
 
+		do
+		{
 			if (get_token(awk) <= -1) return QSE_NULL;
 		} 
+		while (MATCH(awk,TOKEN_NEWLINE));
+	}
 
-	/* skip a semicolon */
-	if (get_token(awk) <= -1) return QSE_NULL;
 
 	return awk;
 }
@@ -1660,11 +1692,19 @@ static qse_awk_t* collect_globals (qse_awk_t* awk)
 static qse_awk_t* collect_locals (
 	qse_awk_t* awk, qse_size_t nlcls, qse_bool_t istop)
 {
-	qse_xstr_t lcl;
+	if (MATCH(awk,TOKEN_NEWLINE))
-	qse_size_t n;
+	{
+		/* special check if the first name is on the 
+		 * same line when QSE_AWK_NEWLINE is on */
+		SETERR (awk, QSE_AWK_EVARMS);
+		return QSE_NULL;
+	}
 
 	while (1) 
 	{
+		qse_xstr_t lcl;
+		qse_size_t n;
+
 		if (!MATCH(awk,TOKEN_IDENT)) 
 		{
 			SETERRTOK (awk, QSE_AWK_EBADVAR);
@@ -1741,7 +1781,18 @@ static qse_awk_t* collect_locals (
 
 		if (get_token(awk) <= -1) return QSE_NULL;
 
-		if (MATCH_TERMINATOR(awk)) break;
+		if (MATCH_TERMINATOR_NORMAL(awk)) 
+		{
+			/* skip the terminator (;, <NL>) */
+			if (get_token(awk) <= -1) return QSE_NULL;
+			break;
+		}
+
+		if (MATCH_TERMINATOR_RBRACE(awk))
+		{
+			/* should not skip } */
+			break;
+		}
 
 		if (!MATCH(awk,TOKEN_COMMA))
 		{
@@ -1749,11 +1800,12 @@ static qse_awk_t* collect_locals (
 			return QSE_NULL;
 		}
 
+		do
+		{
 			if (get_token(awk) <= -1) return QSE_NULL;
 		}
-
+		while (MATCH(awk,TOKEN_NEWLINE));
-	/* skip a semicolon */
+	}
-	if (get_token(awk) <= -1) return QSE_NULL;
 
 	return awk;
 }
@@ -1924,18 +1976,20 @@ static qse_awk_nde_t* parse_statement_nb (qse_awk_t* awk, qse_size_t line)
 
 	if (nde == QSE_NULL) return QSE_NULL;
 
-	/* check if a statement ends with a semicolon */
+	if (MATCH_TERMINATOR_NORMAL(awk))
-	if (MATCH_TERMINATOR(awk)) 
 	{
-		/* eat up the semicolon or a new line and read in the next token
+		/* check if a statement ends with a semicolon or <NL> */
-		 * when QSE_AWK_NEWLINE is set, a statement may end with }.
+		if (get_token(awk) <= -1)
-		 * it should not be eaten up here. */
-		if (!MATCH(awk,TOKEN_RBRACE) && get_token(awk) <= -1) 
 		{
 			if (nde != QSE_NULL) qse_awk_clrpt (awk, nde);
 			return QSE_NULL;
 		}
 	}
+	else if (MATCH_TERMINATOR_RBRACE(awk))
+	{
+		/* do not skip the right brace as a statement terminator. 
+		 * is there anything to do here? */
+	}
 	else
 	{
 		if (nde != QSE_NULL) qse_awk_clrpt (awk, nde);
@@ -5992,4 +6046,3 @@ int qse_awk_putsrcstrx (
 
 	return 0;
 }
-

qse/regress/awk/lang-001.awk

@@ -0,0 +1,3 @@
+# cannot use function name as a parameter name
+function f(f) { print f; }
+BEGIN { f("hello"); }

qse/regress/awk/lang-002.awk

@@ -0,0 +1,7 @@
+function f(f)
+{
+	print f;
+	f("my hello");
+}
+
+BEGIN { f(10); }

qse/regress/awk/lang-003.awk

@@ -0,0 +1,3 @@
+# should print 50
+function fn(f) { f = 20; }
+BEGIN { f = 50; fn(100); print f; }

qse/regress/awk/lang-004.awk

@@ -0,0 +1,3 @@
+# A function and a named variable cannot have the same name.
+function a () { }
+BEGIN { a = 20; }

qse/regress/awk/lang-005.awk

@@ -0,0 +1,7 @@
+function a (a) { print a; } 
+
+BEGIN {
+	local a;
+	a = 20; 
+	a (1000); 
+}

qse/regress/awk/lang-006.awk

@@ -0,0 +1,2 @@
+global a;
+function a () { }

qse/regress/awk/lang-007.awk

@@ -0,0 +1,4 @@
+function fn () { a = 20; return a;}
+global a;
+BEGIN { a = 30; print fn (); print a; }
+

qse/regress/awk/lang-008.awk

@@ -0,0 +1,16 @@
+global x;
+BEGIN {
+	x = 1;
+	{
+		local x;
+		x = 2;
+		{
+			local x;
+			x = 3;
+			print x;
+		}
+		print x;
+	}
+	print x;
+}
+  

qse/regress/awk/lang-009.awk

@@ -0,0 +1,9 @@
+function a (a) { print a; } 
+
+BEGIN {
+	local a;
+	a = 20; 
+}
+
+END { a (1000); }
+

qse/regress/awk/regress.sh

@@ -23,6 +23,7 @@ print_usage()
 ###################
 
 QSEAWK=${QSEAWK:=../../cmd/awk/qseawk}
+TMPFILE="${TMPFILE:=./regress.temp}"
 
 PROGS="
 	cou-001.awk/cou.dat//
@@ -81,6 +82,16 @@ PROGS="
 	emp-026.awk/emp.dat//
 	emp-027.awk/emp.dat//
 
+	lang-001.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-002.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-003.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-004.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-005.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-006.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-007.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-008.awk///--implicit=off --explicit=on --newline=on -o-
+	lang-009.awk/lang-009.awk//--implicit=off --explicit=on --newline=on -o-
+
 	quicksort.awk/quicksort.dat//
 	quicksort2.awk/quicksort2.dat//
 	asm.awk/asm.s/asm.dat/
@@ -96,13 +107,24 @@ PROGS="
 	exit 1;
 }
 
-for prog in ${PROGS}
+echo "${PROGS}" > "${TMPFILE}"
+
+while read prog
 do
+	[ -z "${prog}" ] && continue
+
 	script="`echo ${prog} | cut -d/ -f1`"
 	datafile="`echo ${prog} | cut -d/ -f2`"
 	redinfile="`echo ${prog} | cut -d/ -f3`"
 	awkopts="`echo ${prog} | cut -d/ -f4`"
 
+	[ -z "${script}" ] && continue
+	[ -f "${script}" ] || 
+	{
+		echo_so "${script} not found"
+		continue
+	}
+
 	if [ -n "${redinfile}" ]
 	then
 		echo_so "${QSEAWK} ${awkopts} -f ${script} ${datafile} < ${redinfile}"
@@ -111,6 +133,9 @@ do
 		echo_so "${QSEAWK} ${awkopts} -f ${script} ${datafile}"
 		${QSEAWK} ${awkopts} -f ${script} ${datafile} 
 	fi
-done
+
+done < "${TMPFILE}"
+
+rm -f "${TMPFILE}"
 
 exit 0