From 59910e65d2ce102e002ac1eb08b47a04ea9d8c55 Mon Sep 17 00:00:00 2001
From: NishiOwO <nishi@nishi.boats>
Date: Thu, 9 Oct 2025 10:46:09 +0000
Subject: [PATCH] add utf32 function

git-svn-id: http://svn2.nishi.boats/svn/milsko/trunk@239 b9cfdab3-6d41-4d17-bbe4-086880011989
---
 GNUmakefile          |  2 +-
 include/Mw/Milsko.h  |  1 +
 include/Mw/Unicode.h | 28 +++++++++++++++++++++
 src/unicode.c        | 60 ++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 90 insertions(+), 1 deletion(-)
 create mode 100644 include/Mw/Unicode.h
 create mode 100644 src/unicode.c

diff --git a/GNUmakefile b/GNUmakefile
index 3668732..e2969cc 100644
--- a/GNUmakefile
+++ b/GNUmakefile
@@ -26,7 +26,7 @@ L_CFLAGS = $(DEPINC) $(CFLAGS) -fPIC -D_MILSKO
 L_LDFLAGS = $(LDFLAGS)
 L_LIBS = $(LIBS)
 
-L_OBJS = src/core.o src/default.o src/draw.o src/lowlevel.o src/font.o src/boldfont.o src/error.o
+L_OBJS = src/core.o src/default.o src/draw.o src/lowlevel.o src/font.o src/boldfont.o src/error.o src/unicode.o
 L_OBJS += src/external/ds.o src/external/image.o
 L_OBJS += src/widget/window.o src/widget/button.o src/widget/frame.o src/widget/menu.o src/widget/submenu.o src/widget/image.o src/widget/scrollbar.o src/widget/checkbox.o src/widget/label.o src/widget/entry.o
 L_OBJS += src/cursor/default.o src/cursor/cross.o src/cursor/text.o
diff --git a/include/Mw/Milsko.h b/include/Mw/Milsko.h
index 8ea962d..de6450b 100644
--- a/include/Mw/Milsko.h
+++ b/include/Mw/Milsko.h
@@ -17,6 +17,7 @@
 #include <Mw/Error.h>
 #include <Mw/Cursor.h>
 #include <Mw/Constants.h>
+#include <Mw/Unicode.h>
 
 #include <Mw/Widget/Window.h>
 #include <Mw/Widget/Menu.h>
diff --git a/include/Mw/Unicode.h b/include/Mw/Unicode.h
new file mode 100644
index 0000000..62794f3
--- /dev/null
+++ b/include/Mw/Unicode.h
@@ -0,0 +1,28 @@
+/* $Id$ */
+/*!
+ * %file Mw/Unicode.h
+ * %brief Handles UTF8 stuff
+ */
+#ifndef __MW_UNICODE_H__
+#define __MW_UNICODE_H__
+
+#include <Mw/MachDep.h>
+#include <Mw/TypeDefs.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*!
+ * %brief Converts UTF-8 to UTF-32
+ * %brief input Input
+ * %brief output Output
+ * %return Bytes this multibyte takes
+ */
+MWDECL int MwUTF8ToUTF32(const char* input, int32_t* output);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/src/unicode.c b/src/unicode.c
new file mode 100644
index 0000000..940a941
--- /dev/null
+++ b/src/unicode.c
@@ -0,0 +1,60 @@
+/* $Id$ */
+#include <Mw/Milsko.h>
+
+/* this code was taken from my old code... :) */
+
+#define CAST_I32(x) ((int)(x))
+
+static int utf8_count(unsigned char c) {
+	if(c < 0x80) {
+		return 1;
+	}
+	if(0xc2 <= c && c < 0xe0) {
+		return 2;
+	}
+	if(0xe0 <= c && c < 0xf0) {
+		return 3;
+	}
+	if(0xf0 <= c && c < 0xf8) {
+		return 4;
+	}
+	return 0;
+}
+
+static int utf8_later(unsigned char c) {
+	return 0x80 <= c && c < 0xc0;
+}
+
+int MwUTF8ToUTF32(const char* input, int32_t* output) {
+	const unsigned char* inbuf = (const unsigned char*)input;
+	int		     b	   = utf8_count(inbuf[0]);
+	if(b == 0) return 0;
+
+	if(b == 1) *output = inbuf[0];
+	if(b == 2) {
+		if(!utf8_later(inbuf[1])) return 0;
+		if((inbuf[0] & 0x1e) == 0) return 0;
+
+		*output = CAST_I32(inbuf[0] & 0x1f) << 6;
+		*output |= CAST_I32(inbuf[1] & 0x3f);
+	}
+	if(b == 3) {
+		if(!utf8_later(inbuf[1]) || !utf8_later(inbuf[2])) return 0;
+		if((inbuf[0] & 0x0f) == 0 && (inbuf[1] & 0x20) == 0) return 0;
+
+		*output = CAST_I32(inbuf[0] & 0x0f) << 12;
+		*output |= CAST_I32(inbuf[1] & 0x3f) << 6;
+		*output |= CAST_I32(inbuf[2] & 0x3f);
+	}
+	if(b == 4) {
+		if(!utf8_later(inbuf[1]) || !utf8_later(inbuf[2]) || !utf8_later(inbuf[3])) return 0;
+		if((inbuf[0] & 0x07) == 0 && (inbuf[1] & 0x30) == 0) return 0;
+
+		*output = CAST_I32(inbuf[0] & 0x07) << 18;
+		*output |= CAST_I32(inbuf[1] & 0x3f) << 12;
+		*output |= CAST_I32(inbuf[2] & 0x3f) << 6;
+		*output |= CAST_I32(inbuf[3] & 0x3f);
+	}
+
+	return b;
+}