enhanced split(),ENVIRON,ARGV in awk to build a numeric string if a string can be converted to a number

qse/cmd/awk/awk.c

@@ -292,7 +292,7 @@ static qse_htb_walk_t set_global (
 
 static int apply_fs_and_gvm (qse_awk_rtx_t* rtx, struct arg_t* arg)
 {
-	if (arg->fs != QSE_NULL)
+	if (arg->fs)
 	{
 		qse_awk_val_t* fs;
 
@@ -306,7 +306,7 @@ static int apply_fs_and_gvm (qse_awk_rtx_t* rtx, struct arg_t* arg)
 		qse_awk_rtx_refdownval (rtx, fs);
 	}
 
-	if (arg->gvm != QSE_NULL)
+	if (arg->gvm)
 	{
 		/* set the value of user-defined global variables 
 		 * to a runtime context */

qse/include/qse/awk/Awk.hpp

@@ -718,10 +718,10 @@ public:
 		int setInt (Run* r, long_t v);
 		int setFlt (flt_t v);
 		int setFlt (Run* r, flt_t v);
-		int setStr (const char_t* str, size_t len);
+		int setStr (const char_t* str, size_t len, bool numeric = false);
-		int setStr (Run* r, const char_t* str, size_t len);
+		int setStr (Run* r, const char_t* str, size_t len, bool numeric = false);
-		int setStr (const char_t* str);
+		int setStr (const char_t* str, bool numeric = false);
-		int setStr (Run* r, const char_t* str);
+		int setStr (Run* r, const char_t* str, bool numeric = false);
 
 		int setIndexedVal (
 			const Index& idx,
@@ -758,25 +758,29 @@ public:
 		int setIndexedStr (
 			const Index&  idx,
 			const char_t* str,
-			size_t        len
+			size_t        len,
+			bool  numeric = false
 		);
 
 		int setIndexedStr (
 			Run*          r,
 			const Index&  idx,
 			const char_t* str,
-			size_t        len
+			size_t        len,
+			bool  numeric = false
 		);
 
 		int setIndexedStr (
 			const Index&  idx,
-			const char_t* str
+			const char_t* str,
+			bool  numeric = false
 		);
 
 		int setIndexedStr (
 			Run*          r,
 			const Index&  idx,
-			const char_t* str
+			const char_t* str,
+			bool  numeric = false
 		);
 
 		///

qse/include/qse/awk/awk.h

@@ -2358,6 +2358,18 @@ QSE_EXPORT qse_awk_val_t* qse_awk_rtx_makestrval2 (
 	qse_size_t        len2
 );
 
+
+/**
+ * The qse_awk_rtx_makenstrvalwithstr() function creates a numeric string 
+ * value from a null-terminated string. A numeric string is a string value 
+ * whose one of the header fields \b nstr is 1.
+ * \return value on success, #QSE_NULL on failure
+ */
+QSE_EXPORT qse_awk_val_t* qse_awk_rtx_makenstrvalwithstr (
+	qse_awk_rtx_t* rtx,
+	const qse_char_t* str
+);
+
 /**
  * The qse_awk_rtx_makenstrvalwithcstr() function creates a numeric string 
  * value. A numeric string is a string value whose one of the header fields

qse/lib/awk/Awk.cpp

@@ -573,7 +573,7 @@ int Awk::Value::setFlt (Run* r, flt_t v)
 	return n;
 }
 
-int Awk::Value::setStr (const char_t* str, size_t len)
+int Awk::Value::setStr (const char_t* str, size_t len, bool numeric)
 {
 	if (this->run == QSE_NULL) 
 	{
@@ -581,13 +581,19 @@ int Awk::Value::setStr (const char_t* str, size_t len)
 		 * set an error number for the same reason */
 		return -1; 
 	}
-	return setStr (this->run, str, len);
+	return setStr (this->run, str, len, numeric);
 }
 
-int Awk::Value::setStr (Run* r, const char_t* str, size_t len)
+int Awk::Value::setStr (Run* r, const char_t* str, size_t len, bool numeric)
 {
 	val_t* tmp;
-	tmp = qse_awk_rtx_makestrval (r->rtx, str, len);
+
+	cstr_t cstr;
+	cstr.ptr = str;
+	cstr.len = len;
+
+	tmp = numeric? qse_awk_rtx_makenstrvalwithcstr (r->rtx, &cstr):
+	               qse_awk_rtx_makestrvalwithcstr (r->rtx, &cstr);
 	if (tmp == QSE_NULL)
 	{
 		r->awk->retrieveError (r);
@@ -599,16 +605,17 @@ int Awk::Value::setStr (Run* r, const char_t* str, size_t len)
 	return n;
 }
 
-int Awk::Value::setStr (const char_t* str)
+int Awk::Value::setStr (const char_t* str, bool numeric)
 {
 	if (this->run == QSE_NULL) return -1;
-	return setStr (this->run, str);
+	return setStr (this->run, str, numeric);
 }
 
-int Awk::Value::setStr (Run* r, const char_t* str)
+int Awk::Value::setStr (Run* r, const char_t* str, bool numeric)
 {
 	val_t* tmp;
-	tmp = qse_awk_rtx_makestrvalwithstr (r->rtx, str);
+	tmp = numeric? qse_awk_rtx_makenstrvalwithstr (r->rtx, str):
+	               qse_awk_rtx_makestrvalwithstr (r->rtx, str);
 	if (tmp == QSE_NULL) 
 	{
 		r->awk->retrieveError (r);
@@ -733,17 +740,23 @@ int Awk::Value::setIndexedFlt (Run* r, const Index& idx, flt_t v)
 	return n;
 }
 
-int Awk::Value::setIndexedStr (const Index& idx, const char_t* str, size_t len)
+int Awk::Value::setIndexedStr (const Index& idx, const char_t* str, size_t len, bool numeric)
 {
 	if (run == QSE_NULL) return -1;
-	return setIndexedStr (run, idx, str, len);
+	return setIndexedStr (run, idx, str, len, numeric);
 }
 
 int Awk::Value::setIndexedStr (
-	Run* r, const Index& idx, const char_t* str, size_t len)
+	Run* r, const Index& idx, const char_t* str, size_t len, bool numeric)
 {
 	val_t* tmp;
-	tmp = qse_awk_rtx_makestrval (r->rtx, str, len);
+
+	cstr_t cstr;
+	cstr.ptr = str;
+	cstr.len = len;
+
+	tmp = numeric? qse_awk_rtx_makenstrvalwithcstr (r->rtx, &cstr):
+	               qse_awk_rtx_makestrvalwithcstr (r->rtx, &cstr);
 	if (tmp == QSE_NULL) 
 	{
 		r->awk->retrieveError (r);
@@ -757,16 +770,17 @@ int Awk::Value::setIndexedStr (
 	return n;
 }
 
-int Awk::Value::setIndexedStr (const Index& idx, const char_t* str)
+int Awk::Value::setIndexedStr (const Index& idx, const char_t* str, bool numeric)
 {
 	if (run == QSE_NULL) return -1;
-	return setIndexedStr (run, idx, str);
+	return setIndexedStr (run, idx, str, numeric);
 }
 
-int Awk::Value::setIndexedStr (Run* r, const Index& idx, const char_t* str)
+int Awk::Value::setIndexedStr (Run* r, const Index& idx, const char_t* str, bool numeric)
 {
 	val_t* tmp;
-	tmp = qse_awk_rtx_makestrvalwithstr (r->rtx, str);
+	tmp = numeric? qse_awk_rtx_makenstrvalwithstr (r->rtx, str):
+	               qse_awk_rtx_makestrvalwithstr (r->rtx, str);
 	if (tmp == QSE_NULL)
 	{
 		r->awk->retrieveError (r);
@@ -1412,7 +1426,7 @@ int Awk::dispatch_function (Run* run, const fnc_info_t* fi)
 					}
 					else
 					{
-						xx = args[i].setStr (run, QSE_T(""), 0);
+						xx = args[i].setStr (run, QSE_T(""), (size_t)0);
 					}
 					break;
 				}

qse/lib/awk/StdAwk.cpp

@@ -225,7 +225,7 @@ int StdAwk::build_argcv (Run* run)
 		if (argv.setIndexedStr (
 			Value::IntIndex(i), 
 			this->runarg.ptr[i].ptr, 
-			this->runarg.ptr[i].len) <= -1) return -1;
+			this->runarg.ptr[i].len, true) <= -1) return -1;
 	}
 		
 	run->setGlobal (this->gbl_argc, (long_t)this->runarg.len);
@@ -297,7 +297,8 @@ int StdAwk::__build_environ (Run* run, void* envptr)
 			*eq = QSE_WT('=');
 		#endif
 
-			v_env.setIndexedStr (Value::Index (kptr, klen), vptr);
+			// numeric string
+			v_env.setIndexedStr (Value::Index (kptr, klen), vptr, true);
 
 		#if ((defined(QSE_ENV_CHAR_IS_MCHAR) && defined(QSE_CHAR_IS_MCHAR)) || \
 		     (defined(QSE_ENV_CHAR_IS_WCHAR) && defined(QSE_CHAR_IS_WCHAR)))

qse/lib/awk/fnc.c

@@ -766,8 +766,10 @@ static int fnc_split (qse_awk_rtx_t* run, const qse_awk_fnc_info_t* fi)
 
 		QSE_ASSERT ((tok.ptr != QSE_NULL && tok.len > 0) || tok.len == 0);
 
-		/* create the field string */
+		/* create the field string - however, the split function must
-		t2 = qse_awk_rtx_makestrvalwithcstr (run, &tok);
+		 * create a numeric string if the string is a number */
+		/*t2 = qse_awk_rtx_makestrvalwithcstr (run, &tok);*/
+		t2 = qse_awk_rtx_makenstrvalwithcstr (run, &tok);
 		if (t2 == QSE_NULL) goto oops;
 
 		/* put it into the map */

qse/lib/awk/run.c

@@ -4371,7 +4371,7 @@ static int __cmp_str_str (
 
 	if (ls->nstr == 0 || rs->nstr == 0)
 	{
-		/* nother are definitely a string */
+		/* both are definitely strings */
 		return (rtx->gbl.ignorecase)?
 			qse_strxncasecmp (ls->val.ptr, ls->val.len, rs->val.ptr, rs->val.len):
 			qse_strxncmp (ls->val.ptr, ls->val.len, rs->val.ptr, rs->val.len);

qse/lib/awk/std.c

@@ -1814,7 +1814,10 @@ static int build_argcv (
 	{
 		for (argc = 1, p = icf; *p; p++, argc++) 
 		{
-			v_tmp = qse_awk_rtx_makestrvalwithstr (rtx, *p);
+			/* the argument must compose a numeric string if it can be.
+			 * so call qse_awk_rtx_makenstrvalwithstr() instead of
+			 * qse_awk_rtx_makestrvalwithstr(). */
+			v_tmp = qse_awk_rtx_makenstrvalwithstr (rtx, *p);
 			if (v_tmp == QSE_NULL)
 			{
 				qse_awk_rtx_refdownval (rtx, v_argv);
@@ -1941,7 +1944,10 @@ static int __build_environ (
 			*eq = QSE_WT('=');
 		#endif
 
-			v_tmp = qse_awk_rtx_makestrvalwithstr (rtx, vptr);
+			/* the string in ENVIRON should be a numeric string if
+			 * it can be converted to a number. call makenstrval()
+			 * instead of makestrval() */
+			v_tmp = qse_awk_rtx_makenstrvalwithstr (rtx, vptr);
 			if (v_tmp == QSE_NULL)
 			{
 		#if ((defined(QSE_ENV_CHAR_IS_MCHAR) && defined(QSE_CHAR_IS_MCHAR)) || \

qse/lib/awk/val.c

@@ -406,6 +406,14 @@ init:
 	return (qse_awk_val_t*)val;
 }
 
+qse_awk_val_t* qse_awk_rtx_makenstrvalwithstr (qse_awk_rtx_t* rtx, const qse_char_t* str)
+{
+	qse_cstr_t cstr;
+	cstr.ptr = str;
+	cstr.len = qse_strlen(str);
+	return qse_awk_rtx_makenstrvalwithcstr (rtx, &cstr);
+}
+
 qse_awk_val_t* qse_awk_rtx_makenstrvalwithcstr (qse_awk_rtx_t* rtx, const qse_cstr_t* str)
 {
 	int x;