implemented the option fourth parameter to match() in awk to support submatch retrieval

qse/lib/awk/fnc.c

@@ -66,7 +66,7 @@ static qse_awk_fnc_t sysfnctab[] =
 	{ {QSE_T("toupper"),  7}, 0, { {1,     1, QSE_NULL},      qse_awk_fnc_toupper,  0 }, QSE_NULL},
 	{ {QSE_T("gsub"),     4}, 0, { {2,     3, QSE_T("xvr")},  qse_awk_fnc_gsub,     0 }, QSE_NULL},
 	{ {QSE_T("sub"),      3}, 0, { {2,     3, QSE_T("xvr")},  qse_awk_fnc_sub,      0 }, QSE_NULL},
-	{ {QSE_T("match"),    5}, 0, { {2,     3, QSE_T("vxv")}, qse_awk_fnc_match,    0 }, QSE_NULL},
+	{ {QSE_T("match"),    5}, 0, { {2,     4, QSE_T("vxvr")}, qse_awk_fnc_match,    0 }, QSE_NULL},
 	{ {QSE_T("sprintf"),  7}, 0, { {1, A_MAX, QSE_NULL},      qse_awk_fnc_sprintf,  0 }, QSE_NULL},
 
 	/* math functions */
@@ -1001,7 +1001,7 @@ static int __substitute (qse_awk_rtx_t* rtx, qse_awk_int_t max_count)
 		{
 			n = qse_awk_matchrex (
 				rtx->awk, rex, rtx->gbl.ignorecase, 
-				&s2, &cur, &mat, &errnum
+				&s2, &cur, &mat, QSE_NULL, &errnum
 			);
 		}
 		else n = 0;
@@ -1159,15 +1159,17 @@ int qse_awk_fnc_sub (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 int qse_awk_fnc_match (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 {
 	qse_size_t nargs;
-	qse_awk_val_t* a0, * a1;
+	qse_awk_val_t* a0, * a1, * a3;
 	qse_char_t* str0;
 	qse_size_t len0;
 	qse_awk_int_t idx, start = 1;
+	qse_awk_val_t* x0 = QSE_NULL, * x1 = QSE_NULL, * x2 = QSE_NULL;
 	int n;
-	qse_cstr_t mat;
+	qse_cstr_t mat, submat[9];
+	qse_str_t* tmpbuf = QSE_NULL;
 
 	nargs = qse_awk_rtx_getnargs (rtx);
-	QSE_ASSERT (nargs >= 2 && nargs <= 3);
+	QSE_ASSERT (nargs >= 2 && nargs <= 4);
 	
 	a0 = qse_awk_rtx_getarg (rtx, 0);
 	a1 = qse_awk_rtx_getarg (rtx, 1);
@@ -1181,6 +1183,8 @@ int qse_awk_fnc_match (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 		 * it is treated as a match start index */
 		n = qse_awk_rtx_valtoint (rtx, a2, &start);
 		if (n <= -1) return -1;
+
+		if (nargs >= 4) a3 = qse_awk_rtx_getarg (rtx, 3);
 	}
 
 	str0 = qse_awk_rtx_getvalstr (rtx, a0, &len0);
@@ -1189,6 +1193,7 @@ int qse_awk_fnc_match (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 	if (start == 0) start = 1;
 	else if (start < 0) start = len0 + start + 1;
 
+	QSE_MEMSET (submat, 0, QSE_SIZEOF(submat));
 	if (start > len0 || start <= 0) n = 0;
 	else
 	{
@@ -1198,7 +1203,7 @@ int qse_awk_fnc_match (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 		tmp.ptr = str0 + start - 1;
 		tmp.len = len0 - start + 1;
 
-		n = qse_awk_rtx_matchrex (rtx, a1, &tmp, &tmp, &mat);
+		n = qse_awk_rtx_matchrex (rtx, a1, &tmp, &tmp, &mat, (nargs >= 4? submat: QSE_NULL));
 		if (n <= -1) return -1;
 	}
 
@@ -1207,43 +1212,90 @@ int qse_awk_fnc_match (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 	/* RSTART: 0 on no match */
 	idx = (n == 0)? 0: ((qse_awk_int_t)(mat.ptr-str0) + 1);
 
-	a0 = qse_awk_rtx_makeintval (rtx, idx);
+	x0 = qse_awk_rtx_makeintval (rtx, idx);
-	if (a0 == QSE_NULL) return -1;
+	if (!x0) goto oops;
-
+	qse_awk_rtx_refupval (rtx, x0);
-	qse_awk_rtx_refupval (rtx, a0);
 
 	/* RLENGTH: -1 on no match */
-	a1 = qse_awk_rtx_makeintval (rtx, 
+	x1 = qse_awk_rtx_makeintval (rtx, 
 		((n == 0)? (qse_awk_int_t)-1: (qse_awk_int_t)mat.len));
-	if (a1 == QSE_NULL)
+	if (!x1) goto oops;
+	qse_awk_rtx_refupval (rtx, x1);
+
+	if (nargs >= 4)
 	{
-		qse_awk_rtx_refdownval (rtx, a0);
+		const qse_cstr_t* subsep;
-		return -1;
+		qse_awk_int_t submatcount;
+		qse_size_t i, xlen;
+		qse_awk_val_t* tv;
+
+		tmpbuf = qse_str_open (qse_awk_rtx_getmmgr(rtx), 0, 64);
+		if (!tmpbuf) goto oops;
+
+		x2 = qse_awk_rtx_makemapval (rtx);
+		if (!x2) goto oops;
+		qse_awk_rtx_refupval (rtx, x2);
+
+		submatcount =0;
+		subsep = qse_awk_rtx_getsubsep (rtx);
+		for (i = 0; i < QSE_COUNTOF(submat); i++)
+		{
+			if (!submat[i].ptr) break;
+
+			submatcount++;
+
+			if (qse_str_fmt (tmpbuf, QSE_T("%d"), (int)submatcount) == (qse_size_t)-1 ||
+			    qse_str_ncat (tmpbuf, subsep->ptr, subsep->len) == (qse_size_t)-1) goto oops;
+			xlen = QSE_STR_LEN(tmpbuf);
+			if (qse_str_ncat (tmpbuf, QSE_T("start"), 5) == (qse_size_t)-1) goto oops;
+
+			tv = qse_awk_rtx_makeintval (rtx, submat[i].ptr - str0 + 1);
+			if (!tv) goto oops;
+			if (!qse_awk_rtx_setmapvalfld (rtx, x2, QSE_STR_PTR(tmpbuf), QSE_STR_LEN(tmpbuf), tv))
+			{
+				qse_awk_rtx_refupval (rtx, tv);
+				qse_awk_rtx_refdownval (rtx, tv);
+				goto oops;
 			}
 
-	qse_awk_rtx_refupval (rtx, a1);
+			if (qse_str_setlen (tmpbuf, xlen) == (qse_size_t)-1 ||
+			    qse_str_ncat (tmpbuf, QSE_T("length"), 6) == (qse_size_t)-1) goto oops;
 
-	if (qse_awk_rtx_setgbl (rtx, QSE_AWK_GBL_RSTART, a0) <= -1)
+			tv = qse_awk_rtx_makeintval (rtx, submat[i].len);
+			if (!tv) goto oops;
+			if (!qse_awk_rtx_setmapvalfld (rtx, x2, QSE_STR_PTR(tmpbuf), QSE_STR_LEN(tmpbuf), tv))
 			{
-		qse_awk_rtx_refdownval (rtx, a1);
+				qse_awk_rtx_refupval (rtx, tv);
-		qse_awk_rtx_refdownval (rtx, a0);
+				qse_awk_rtx_refdownval (rtx, tv);
-		return -1;
+				goto oops;
+			}
+		}
+		/* the caller of this function must be able to get the submatch count by
+		 * dividing the array size by 2 */
+
+		if (qse_awk_rtx_setrefval (rtx, (qse_awk_val_ref_t*)a3, x2) <= -1) goto oops;
 	}
 
-	if (qse_awk_rtx_setgbl (rtx, QSE_AWK_GBL_RLENGTH, a1) <= -1)
+	if (qse_awk_rtx_setgbl (rtx, QSE_AWK_GBL_RSTART, x0) <= -1 ||
+	    qse_awk_rtx_setgbl (rtx, QSE_AWK_GBL_RLENGTH, x1) <= -1)
 	{
-		qse_awk_rtx_refdownval (rtx, a1);
+		goto oops;
-		qse_awk_rtx_refdownval (rtx, a0);
-		return -1;
 	}
 
-/* TOOD: set RMATCH to hold the mainmatch and submatches inside () */
+	qse_awk_rtx_setretval (rtx, x0);
 
-	qse_awk_rtx_setretval (rtx, a0);
+	if (tmpbuf) qse_str_close (tmpbuf);
-
+	if (x2) qse_awk_rtx_refdownval (rtx, x2);
-	qse_awk_rtx_refdownval (rtx, a1);
+	qse_awk_rtx_refdownval (rtx, x1);
-	qse_awk_rtx_refdownval (rtx, a0);
+	qse_awk_rtx_refdownval (rtx, x0);
 	return 0;
+
+oops:
+	if (tmpbuf) qse_str_close (tmpbuf);
+	if (x2) qse_awk_rtx_refdownval (rtx, x2);
+	if (x1) qse_awk_rtx_refdownval (rtx, x1);
+	if (x0) qse_awk_rtx_refdownval (rtx, x0);
+	return -1;
 }
 
 int qse_awk_fnc_sprintf (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)

qse/lib/awk/misc.c

@@ -906,7 +906,7 @@ qse_char_t* qse_awk_rtx_strxntokbyrex (
 	{
 		n = qse_awk_matchrex (
 			rtx->awk, rex, rtx->gbl.ignorecase,
-			&s, &cursub, &match, errnum);
+			&s, &cursub, &match, QSE_NULL, errnum);
 		if (n == -1) return QSE_NULL;
 		if (n == 0)
 		{
@@ -1248,12 +1248,13 @@ static int matchtre (
 int qse_awk_matchrex (
 	qse_awk_t* awk, void* code, int icase,
 	const qse_cstr_t* str, const qse_cstr_t* substr,
-	qse_cstr_t* match, qse_awk_errnum_t* errnum)
+	qse_cstr_t* match, qse_cstr_t submat[9], qse_awk_errnum_t* errnum)
 {
 #if defined(USE_REX)
 	int x;
 	qse_rex_errnum_t err;
 
+	/* submatch is not supported */
 	x = qse_matchrex (
 		awk->mmgr, awk->opt.depth.s.rex_match, code, 
 		(icase? QSE_REX_IGNORECASE: 0), str, substr, match, &err);
@@ -1266,7 +1267,7 @@ int qse_awk_matchrex (
 	x = matchtre (
 		awk, code,
 		((str->ptr == substr->ptr)? opt: (opt | QSE_TRE_NOTBOL)),
-		substr, match, QSE_NULL, errnum
+		substr, match, submat, errnum
 	);
 	return x;
 #endif
@@ -1295,7 +1296,7 @@ void qse_awk_freerex (qse_awk_t* awk, void* code, void* icode)
 
 int qse_awk_rtx_matchrex (
 	qse_awk_rtx_t* rtx, qse_awk_val_t* val,
-	const qse_cstr_t* str, const qse_cstr_t* substr, qse_cstr_t* match)
+	const qse_cstr_t* str, const qse_cstr_t* substr, qse_cstr_t* match, qse_cstr_t submat[9])
 {
 	void* code;
 	int icase, x;
@@ -1329,6 +1330,7 @@ int qse_awk_rtx_matchrex (
 	}
 	
 #if defined(USE_REX)
+	/* submatch not supported */
 	x = qse_matchrex (
 		rtx->awk->mmgr, rtx->awk->opt.depth.s.rex_match,
 		code, (icase? QSE_REX_IGNORECASE: 0),
@@ -1338,7 +1340,7 @@ int qse_awk_rtx_matchrex (
 	x = matchtre (
 		rtx->awk, code,
 		((str->ptr == substr->ptr)? QSE_TRE_BACKTRACKING: (QSE_TRE_BACKTRACKING | QSE_TRE_NOTBOL)),
-		substr, match, QSE_NULL, &awkerr
+		substr, match, submat, &awkerr
 	);
 	if (x <= -1) qse_awk_rtx_seterrnum (rtx, awkerr, QSE_NULL);
 #endif

qse/lib/awk/misc.h

@@ -81,7 +81,7 @@ int qse_awk_buildrex (
 int qse_awk_matchrex (
 	qse_awk_t* awk, void* code, int icase,
 	const qse_cstr_t* str, const qse_cstr_t* substr,
-	qse_cstr_t* match, qse_awk_errnum_t* errnum
+	qse_cstr_t* match, qse_cstr_t submat[9], qse_awk_errnum_t* errnum
 );
 
 void qse_awk_freerex (qse_awk_t* awk, void* code, void* icode);
@@ -89,7 +89,7 @@ void qse_awk_freerex (qse_awk_t* awk, void* code, void* icode);
 int qse_awk_rtx_matchrex (
 	qse_awk_rtx_t* rtx, qse_awk_val_t* val,
 	const qse_cstr_t* str, const qse_cstr_t* substr,
-	qse_cstr_t* match
+	qse_cstr_t* match, qse_cstr_t submat[9]
 );
 
 #if defined(__cplusplus)

qse/lib/awk/parse.c

@@ -3367,28 +3367,30 @@ static int fold_constants_for_binop (
 				if (((qse_awk_nde_int_t*)right)->val == 0)
 				{
 					qse_awk_seterrnum (awk, QSE_AWK_EDIVBY0, QSE_NULL);
-					return QSE_NULL;
+					fold = -2; /* error */
 				}
-
+				else if (INT_BINOP_INT(left,%,right))
-				if (INT_BINOP_INT(left,%,right))
 				{
 					folded->r = (qse_awk_flt_t)((qse_awk_nde_int_t*)left)->val / 
 					            (qse_awk_flt_t)((qse_awk_nde_int_t*)right)->val;
 					fold = QSE_AWK_NDE_FLT;
-					break;
 				}
-
+				else
+				{
 					folded->l = INT_BINOP_INT(left,/,right);
+				}
 				break;
 
 			case QSE_AWK_BINOP_IDIV:
 				if (((qse_awk_nde_int_t*)right)->val == 0)
 				{
 					qse_awk_seterrnum (awk, QSE_AWK_EDIVBY0, QSE_NULL);
-					return QSE_NULL;
+					fold = -2; /* error */
 				}
-
+				else
+				{
 					folded->l = INT_BINOP_INT(left,/,right);
+				}
 				break;
 
 			case QSE_AWK_BINOP_MOD:
@@ -3396,7 +3398,7 @@ static int fold_constants_for_binop (
 				break;
 
 			default:
-				fold = -1;
+				fold = -1; /* no folding */
 				break;
 		}
 	}
@@ -3700,6 +3702,9 @@ static qse_awk_nde_t* parse_binary (
 
 				break;
 
+			case -2:
+				goto oops;
+
 			default:
 			{
 				qse_awk_nde_t* tmp;

qse/lib/awk/rio.c

@@ -239,7 +239,7 @@ static QSE_INLINE int match_long_rs (
 	ret = qse_awk_matchrex (
 		run->awk, run->gbl.rs[run->gbl.ignorecase], 
 		run->gbl.ignorecase, QSE_STR_XSTR(buf), QSE_STR_XSTR(buf),
-		&match, &errnum);
+		&match, QSE_NULL, &errnum);
 	if (ret <= -1)
 	{
 		qse_awk_rtx_seterrnum (run, errnum, QSE_NULL);

qse/lib/awk/run.c

@@ -293,7 +293,7 @@ QSE_INLINE qse_awk_val_t* qse_awk_rtx_getarg (qse_awk_rtx_t* run, qse_size_t idx
 
 QSE_INLINE qse_awk_val_t* qse_awk_rtx_getgbl (qse_awk_rtx_t* run, int id)
 {
-	QSE_ASSERT (id >= 0 && id < (int)QSE_LDA_SIZE(run->awk->parse.gbls));
+	QSE_ASSERT (id >= 0 && id < (int)QSE_ARR_SIZE(run->awk->parse.gbls));
 	return RTX_STACK_GBL (run, id);
 }
 
@@ -687,7 +687,7 @@ QSE_INLINE void qse_awk_rtx_setretval (
 QSE_INLINE int qse_awk_rtx_setgbl (
 	qse_awk_rtx_t* rtx, int id, qse_awk_val_t* val)
 {
-	QSE_ASSERT (id >= 0 && id < (int)QSE_LDA_SIZE(rtx->awk->parse.gbls));
+	QSE_ASSERT (id >= 0 && id < (int)QSE_ARR_SIZE(rtx->awk->parse.gbls));
 	return set_global (rtx, id, QSE_NULL, val, 0);
 }
 
@@ -3234,7 +3234,7 @@ static qse_awk_val_t* eval_expression (qse_awk_rtx_t* rtx, qse_awk_nde_t* nde)
 			vs.len = ((qse_awk_val_str_t*)rtx->inrec.d0)->val.len;
 		}
 
-		n = qse_awk_rtx_matchrex (rtx, v, &vs, &vs, QSE_NULL);
+		n = qse_awk_rtx_matchrex (rtx, v, &vs, &vs, QSE_NULL, QSE_NULL);
 		if (n <= -1)
 		{
 			ADJERR_LOC (rtx, &nde->loc);
@@ -5041,7 +5041,7 @@ static qse_awk_val_t* eval_binop_match0 (
 	out.ptr = qse_awk_rtx_getvalstr (rtx, left, &out.len);
 	if (out.ptr == QSE_NULL) return QSE_NULL;
 
-	n = qse_awk_rtx_matchrex (rtx, right, &out, &out, QSE_NULL);
+	n = qse_awk_rtx_matchrex (rtx, right, &out, &out, QSE_NULL, QSE_NULL);
 	qse_awk_rtx_freevalstr (rtx, left, out.ptr);
 
 	if (n <= -1)