work in progress. trying to accept the encoding name in str::tombs() and str::frommbs()

qse/include/qse/awk/awk.h

@@ -2892,12 +2892,15 @@ QSE_EXPORT void qse_awk_rtx_freevalstr (
 );
 
 
-QSE_EXPORT qse_mchar_t* qse_awk_rtx_getvalmbs (
+QSE_EXPORT qse_mchar_t* qse_awk_rtx_getvalmbswithcmgr (
 	qse_awk_rtx_t*       rtx, /**< runtime context */
 	const qse_awk_val_t* val, /**< value to convert */
-	qse_size_t*          len  /**< result length */
+	qse_size_t*          len, /**< result length */
+	qse_cmgr_t*          cmgr
 );
 
+#define qse_awk_rtx_getvalmbs(rtx,val,len) qse_awk_rtx_getvalmbswithcmgr(rtx, val, len, qse_awk_rtx_getcmgr(rtx))
+
 QSE_EXPORT void qse_awk_rtx_freevalmbs (
 	qse_awk_rtx_t*       rtx, /**< runtime context */
 	const qse_awk_val_t* val, /**< value to convert */

qse/lib/awk/mod-str.c

@@ -27,6 +27,7 @@
 #include "mod-str.h"
 #include <qse/cmn/str.h>
 #include <qse/cmn/chr.h>
+#include <qse/cmn/mbwc.h>
 #include "../cmn/mem-prv.h"
 #include "fnc.h"
 #include "val.h"
@@ -357,13 +358,32 @@ static int fnc_tocharcode (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 
 static int fnc_frommbs (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 {
-	/* str::frommbs(B"byte-string") */
+	/* str::frommbs(B"byte-string" [, "encoding-name"]) */
 	qse_awk_val_t* a0, * r;
+	qse_cmgr_t* cmgr = qse_awk_rtx_getcmgr(rtx);
+
+	if (qse_awk_rtx_getnargs(rtx) >= 2)
+	{
+		qse_awk_val_t* a1;
+		qse_cstr_t enc;
+
+		a1 = qse_awk_rtx_getarg(rtx, 1);
+		enc.ptr = qse_awk_rtx_getvalstr(rtx, a1, &enc.len);
+		if (!enc.ptr) return -1;
+		cmgr = (enc.len == qse_strlen(enc.ptr))? qse_findcmgr(enc.ptr): QSE_NULL;
+		qse_awk_rtx_freevalstr (rtx, a1, enc.ptr);
+
+		if (!cmgr) 
+		{
+			/* if the encoding name is not known, return a zero-length string */
+			r = qse_awk_rtx_makestrval(rtx, QSE_NULL, 0); /* this never fails for length 0 */
+			goto done;
+		}
+	}
 
 	a0 = qse_awk_rtx_getarg(rtx, 0);
 	switch (QSE_AWK_RTX_GETVALTYPE(rtx, a0))
 	{
-
 		case QSE_AWK_VAL_STR:
 			r = a0;
 			break;
@@ -380,14 +400,34 @@ static int fnc_frommbs (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 		}
 	}
 
+done:
 	qse_awk_rtx_setretval (rtx, r);
 	return 0;
 }
 
 static int fnc_tombs (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 {
-	/* str::tombs("string") */
+	/* str::tombs("string", [, "encoding-name"]) */
 	qse_awk_val_t* a0, * r;
+	qse_cmgr_t* cmgr = qse_awk_rtx_getcmgr(rtx);
+
+	if (qse_awk_rtx_getnargs(rtx) >= 2)
+	{
+		qse_awk_val_t* a1;
+		qse_cstr_t enc;
+		a1 = qse_awk_rtx_getarg(rtx, 1);
+		enc.ptr = qse_awk_rtx_getvalstr(rtx, a1, &enc.len);
+		if (!enc.ptr) return -1;
+		cmgr = (enc.len == qse_strlen(enc.ptr))? qse_findcmgr(enc.ptr): QSE_NULL;
+		qse_awk_rtx_freevalstr (rtx, a1, enc.ptr);
+
+		if (!cmgr) 
+		{
+			/* if the encoding name is not known, return a zero-length string */
+			r = qse_awk_rtx_makembsval(rtx, QSE_NULL, 0); /* this never fails for length 0 */
+			goto done;
+		}
+	}
 
 	a0 = qse_awk_rtx_getarg(rtx, 0);
 	switch (QSE_AWK_RTX_GETVALTYPE(rtx, a0))
@@ -399,7 +439,7 @@ static int fnc_tombs (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 		default:
 		{
 			qse_mcstr_t str;
-			str.ptr = qse_awk_rtx_getvalmbs(rtx, a0, &str.len);
+			str.ptr = qse_awk_rtx_getvalmbswithcmgr(rtx, a0, &str.len, cmgr);
 			if (!str.ptr) return -1;
 			r = qse_awk_rtx_makembsvalwithmxstr(rtx, &str);
 			qse_awk_rtx_freevalmbs (rtx, a0, str.ptr);
@@ -408,6 +448,7 @@ static int fnc_tombs (qse_awk_rtx_t* rtx, const qse_awk_fnc_info_t* fi)
 		}
 	}
 
+done:
 	qse_awk_rtx_setretval (rtx, r);
 	return 0;
 }
@@ -501,7 +542,7 @@ static fnctab_t fnctab[] =
 {
 	/* keep this table sorted for binary search in query(). */
 	{ QSE_T("fromcharcode"), { { 0, A_MAX, QSE_NULL },  fnc_fromcharcode,      0 } },
-	{ QSE_T("frommbs"),      { { 1, 1, QSE_NULL },      fnc_frommbs,           0 } },
+	{ QSE_T("frommbs"),      { { 1, 2, QSE_NULL },      fnc_frommbs,           0 } },
 	{ QSE_T("gsub"),         { { 2, 3, QSE_T("xvr")},   qse_awk_fnc_gsub,      0 } },
 	{ QSE_T("index"),        { { 2, 3, QSE_NULL },      qse_awk_fnc_index,     0 } },
 	{ QSE_T("isalnum"),      { { 1, 1, QSE_NULL },      fnc_isalnum,           0 } },
@@ -528,7 +569,7 @@ static fnctab_t fnctab[] =
 	{ QSE_T("substr"),       { { 2, 3, QSE_NULL },      qse_awk_fnc_substr,    0 } },
 	{ QSE_T("tocharcode"),   { { 1, 2, QSE_NULL },      fnc_tocharcode,        0 } },
 	{ QSE_T("tolower"),      { { 1, 1, QSE_NULL },      qse_awk_fnc_tolower,   0 } },
-	{ QSE_T("tombs"),        { { 1, 1, QSE_NULL },      fnc_tombs,             0 } },
+	{ QSE_T("tombs"),        { { 1, 2, QSE_NULL },      fnc_tombs,             0 } },
 	{ QSE_T("tonum"),        { { 1, 2, QSE_NULL },      fnc_tonum,             0 } },
 	{ QSE_T("toupper"),      { { 1, 1, QSE_NULL },      qse_awk_fnc_toupper,   0 } },
 	{ QSE_T("trim"),         { { 1, 2, QSE_NULL },      fnc_trim,              0 } }

qse/lib/awk/val.c

@@ -36,9 +36,11 @@
 
 static qse_awk_val_nil_t awk_nil = { QSE_AWK_VAL_NIL, 0, 1, 0, 0 };
 static qse_awk_val_str_t awk_zls = { QSE_AWK_VAL_STR, 0, 1, 0, 0, { QSE_T(""), 0 } };
+static qse_awk_val_mbs_t awk_zlm = { QSE_AWK_VAL_MBS, 0, 1, 0, 0, { QSE_MT(""), 0 } };
 
 qse_awk_val_t* qse_awk_val_nil = (qse_awk_val_t*)&awk_nil;
 qse_awk_val_t* qse_awk_val_zls = (qse_awk_val_t*)&awk_zls; 
+qse_awk_val_t* qse_awk_val_zlm = (qse_awk_val_t*)&awk_zlm;
 
 qse_awk_val_t* qse_getawknilval (void)
 {
@@ -167,10 +169,13 @@ qse_awk_val_t* qse_awk_rtx_makestrvalwithxstr (qse_awk_rtx_t* rtx, const qse_cst
 {
 	qse_awk_val_str_t* val = QSE_NULL;
 	qse_size_t rlen = str->len;
-
 #ifdef ENABLE_FEATURE_SCACHE
 	qse_size_t i;
+#endif
 
+	if (rlen <= 0) return qse_awk_val_zls;
+
+#ifdef ENABLE_FEATURE_SCACHE
 	i = rlen / FEATURE_SCACHE_BLOCK_UNIT;
 	if (i < QSE_COUNTOF(rtx->scache_count))
 	{
@@ -381,7 +386,11 @@ qse_awk_val_t* qse_awk_rtx_makenstrvalwithxstr (qse_awk_rtx_t* rtx, const qse_cs
 qse_awk_val_t* qse_awk_rtx_makembsval (qse_awk_rtx_t* rtx, const qse_mchar_t* ptr, qse_size_t len)
 {
 	qse_awk_val_mbs_t* val = QSE_NULL;
-	qse_size_t xsz  = len * QSE_SIZEOF(*ptr);
+	qse_size_t xsz;
+
+	if (len <= 0) return qse_awk_val_zlm;
+
+	xsz = len * QSE_SIZEOF(*ptr);
 
 	val = (qse_awk_val_mbs_t*)qse_awk_rtx_callocmem(rtx, QSE_SIZEOF(qse_awk_val_mbs_t) + xsz + QSE_SIZEOF(*ptr));
 	if (!val) return QSE_NULL;
@@ -1635,6 +1644,7 @@ void qse_awk_rtx_freevalstr (qse_awk_rtx_t* rtx, const qse_awk_val_t* v, qse_cha
 }
 
 
+#if 0
 qse_mchar_t* qse_awk_rtx_getvalmbs (qse_awk_rtx_t* rtx, const qse_awk_val_t* v, qse_size_t* len)
 {
 	if (QSE_AWK_RTX_GETVALTYPE(rtx, v) == QSE_AWK_VAL_MBS)
@@ -1647,6 +1657,20 @@ qse_mchar_t* qse_awk_rtx_getvalmbs (qse_awk_rtx_t* rtx, const qse_awk_val_t* v,
 		return qse_awk_rtx_valtombsdup(rtx, v, len);
 	}
 }
+#endif
+
+qse_mchar_t* qse_awk_rtx_getvalmbswithcmgr (qse_awk_rtx_t* rtx, const qse_awk_val_t* v, qse_size_t* len, qse_cmgr_t* cmgr)
+{
+	if (QSE_AWK_RTX_GETVALTYPE(rtx, v) == QSE_AWK_VAL_MBS)
+	{
+		if (len) *len = ((qse_awk_val_mbs_t*)v)->val.len;
+		return ((qse_awk_val_mbs_t*)v)->val.ptr;
+	}
+	else
+	{
+		return qse_awk_rtx_valtombsdup(rtx, v, len); /* TODO: */
+	}
+}
 
 void qse_awk_rtx_freevalmbs (qse_awk_rtx_t* rtx, const qse_awk_val_t* v, qse_mchar_t* str)
 {

qse/lib/awk/val.h

@@ -66,7 +66,7 @@ struct qse_awk_val_rchunk_t
  * add a field to indicate if a value is static.
  * 
 
-#define IS_STATICVAL(val) ((val) == QSE_NULL || (val) == qse_awk_val_nil || (val) == qse_awk_val_zls)
+#define IS_STATICVAL(val) ((val) == QSE_NULL || (val) == qse_awk_val_nil || (val) == qse_awk_val_zls || (val) == qse_awk_val_zlm)
 */
 #define IS_STATICVAL(val) ((val)->stat)