added many charcter handling code

lib/utl.c

@@ -1,6 +1,4 @@
 /*
- * $Id$
- *
     Copyright (c) 2016-2018 Chung, Hyung-Hwan. All rights reserved.
 
     Redistribution and use in source and binary forms, with or without
@@ -848,43 +846,149 @@ HCL_INLINE int hcl_conv_ucstr_to_bcstr_with_cmgr (const hcl_uch_t* ucs, hcl_oow_
 
 /* ----------------------------------------------------------------------- */
 
-static hcl_cmgr_t utf8_cmgr =
+static hcl_cmgr_t builtin_cmgr[] =
 {
-	hcl_utf8_to_uc,
-	hcl_uc_to_utf8
+	/* keep the order aligned with hcl_cmgr_id_t values in <hcl-utl.h> */
+	{ hcl_utf8_to_uc,  hcl_uc_to_utf8 },
+	{ hcl_utf16_to_uc, hcl_uc_to_utf16 },
+	{ hcl_mb8_to_uc,   hcl_uc_to_mb8 }
 };
 
-hcl_cmgr_t* hcl_get_utf8_cmgr (void)
+hcl_cmgr_t* hcl_get_cmgr_by_id (hcl_cmgr_id_t id)
 {
-	return &utf8_cmgr;
+	return &builtin_cmgr[id];
 }
 
+static struct
+{
+	const hcl_bch_t* name;
+	hcl_cmgr_id_t     id;
+} builtin_cmgr_tab[] =
+{
+	{ "utf8",    HCL_CMGR_UTF8 },
+	{ "utf16",   HCL_CMGR_UTF16 },
+	{ "mb8",     HCL_CMGR_MB8 }
+};
+
+
+hcl_cmgr_t* hcl_get_cmgr_by_bcstr (const hcl_bch_t* name)
+{
+	if (name)
+	{
+		hcl_oow_t i;
+
+		for (i = 0; i < HCL_COUNTOF(builtin_cmgr_tab); i++)
+		{
+			if (hcl_comp_bcstr(name, builtin_cmgr_tab[i].name) == 0)
+			{
+				return &builtin_cmgr[builtin_cmgr_tab[i].id];
+			}
+		 }
+	}
+
+	return HCL_NULL;
+}
+
+hcl_cmgr_t* hcl_get_cmgr_by_ucstr (const hcl_uch_t* name)
+{
+	if (name)
+	{
+		hcl_oow_t i;
+
+		for (i = 0; i < HCL_COUNTOF(builtin_cmgr_tab); i++)
+		{
+			if (hcl_comp_ucstr_bcstr(name, builtin_cmgr_tab[i].name) == 0)
+			{
+				return &builtin_cmgr[builtin_cmgr_tab[i].id];
+			}
+		 }
+	}
+
+	return HCL_NULL;
+}
+
+/* ----------------------------------------------------------------------- */
+
 int hcl_conv_utf8_to_uchars (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
 {
 	/* the source is length bound */
-	return hcl_conv_bchars_to_uchars_with_cmgr(bcs, bcslen, ucs, ucslen, &utf8_cmgr, 0);
+	return hcl_conv_bchars_to_uchars_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_UTF8], 0);
 }
 
 int hcl_conv_uchars_to_utf8 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
 {
 	/* length bound */
-	return hcl_conv_uchars_to_bchars_with_cmgr(ucs, ucslen, bcs, bcslen, &utf8_cmgr);
+	return hcl_conv_uchars_to_bchars_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_UTF8]);
 }
 
 int hcl_conv_utf8_to_ucstr (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
 {
 	/* null-terminated. */
-	return hcl_conv_bcstr_to_ucstr_with_cmgr(bcs, bcslen, ucs, ucslen, &utf8_cmgr, 0);
+	return hcl_conv_bcstr_to_ucstr_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_UTF8], 0);
 }
 
 int hcl_conv_ucstr_to_utf8 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
 {
 	/* null-terminated */
-	return hcl_conv_ucstr_to_bcstr_with_cmgr(ucs, ucslen, bcs, bcslen, &utf8_cmgr);
+	return hcl_conv_ucstr_to_bcstr_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_UTF8]);
 }
 
 /* ----------------------------------------------------------------------- */
 
+int hcl_conv_utf16_to_uchars (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
+{
+	/* the source is length bound */
+	return hcl_conv_bchars_to_uchars_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_UTF16], 0);
+}
+
+int hcl_conv_uchars_to_utf16 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
+{
+	/* length bound */
+	return hcl_conv_uchars_to_bchars_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_UTF16]);
+}
+
+int hcl_conv_utf16_to_ucstr (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
+{
+	/* null-terminated. */
+	return hcl_conv_bcstr_to_ucstr_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_UTF16], 0);
+}
+
+int hcl_conv_ucstr_to_utf16 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
+{
+	/* null-terminated */
+	return hcl_conv_ucstr_to_bcstr_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_UTF16]);
+}
+
+/* ----------------------------------------------------------------------- */
+
+int hcl_conv_mb8_to_uchars (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
+{
+	/* the source is length bound */
+	return hcl_conv_bchars_to_uchars_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_MB8], 0);
+}
+
+int hcl_conv_uchars_to_mb8 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
+{
+	/* length bound */
+	return hcl_conv_uchars_to_bchars_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_MB8]);
+}
+
+int hcl_conv_mb8_to_ucstr (const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
+{
+	/* null-terminated. */
+	return hcl_conv_bcstr_to_ucstr_with_cmgr(bcs, bcslen, ucs, ucslen, &builtin_cmgr[HCL_CMGR_MB8], 0);
+}
+
+int hcl_conv_ucstr_to_mb8 (const hcl_uch_t* ucs, hcl_oow_t* ucslen, hcl_bch_t* bcs, hcl_oow_t* bcslen)
+{
+	/* null-terminated */
+	return hcl_conv_ucstr_to_bcstr_with_cmgr(ucs, ucslen, bcs, bcslen, &builtin_cmgr[HCL_CMGR_MB8]);
+}
+
+/* ----------------------------------------------------------------------- */
+
+/* ----------------------------------------------------------------------- */
+
 int hcl_convbtouchars (hcl_t* hcl, const hcl_bch_t* bcs, hcl_oow_t* bcslen, hcl_uch_t* ucs, hcl_oow_t* ucslen)
 {
 	/* length bound */