From 63d787d9e963969ff42811c2f61dbd78dc1c3a9c Mon Sep 17 00:00:00 2001
From: Bill Currie <bill@taniwha.org>
Date: Thu, 5 May 2005 00:27:04 +0000
Subject: [PATCH] add MSG_WriteUTF8 and MSG_ReadUTF8 to handle 31 bit values
 nicely for future protocol extensions. Equivalent to MSG_*Byte or MSG_*Char.
 Use MSG_ReadString and MSG_WriteString with the appropriate libc functions
 for utf8 strings.

---
 include/QF/msg.h |  2 ++
 libs/util/msg.c  | 94 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+)

diff --git a/include/QF/msg.h b/include/QF/msg.h
index b43181334..167fd89c3 100644
--- a/include/QF/msg.h
+++ b/include/QF/msg.h
@@ -49,6 +49,7 @@ void MSG_WriteCoordAngleV (sizebuf_t *sb, const vec3_t coord,
 void MSG_WriteAngle (sizebuf_t *sb, float angle);
 void MSG_WriteAngleV (sizebuf_t *sb, const vec3_t angles);
 void MSG_WriteAngle16 (sizebuf_t *sb, float angle16);
+void MSG_WriteUTF8 (sizebuf_t *sb, unsigned utf8);
 
 typedef struct msg_s {
 	int readcount;
@@ -73,6 +74,7 @@ float MSG_ReadAngle (qmsg_t *msg);
 void MSG_ReadCoordAngleV (qmsg_t *msg, vec3_t coord, vec3_t angles);
 void MSG_ReadAngleV (qmsg_t *msg, vec3_t angles);
 float MSG_ReadAngle16 (qmsg_t *msg);
+int MSG_ReadUTF8 (qmsg_t *msg);
 
 //@}
 //@}
diff --git a/libs/util/msg.c b/libs/util/msg.c
index 83e0f1d81..5a7b4a1b9 100644
--- a/libs/util/msg.c
+++ b/libs/util/msg.c
@@ -165,6 +165,45 @@ MSG_WriteAngle16 (sizebuf_t *sb, float angle16)
 	MSG_WriteShort (sb, (int) (angle16 * (65536.0 / 360.0)) & 65535);
 }
 
+void
+MSG_WriteUTF8 (sizebuf_t *sb, unsigned utf8)
+{
+	byte	   *buf;
+	int			count;
+
+	if (utf8 & 0x80000000) {
+		return;					// invalid (FIXME die?)
+	} else if (utf8 & 0x7c000000) {
+		buf = SZ_GetSpace (sb, count = 6);
+		*buf = 0xfc | ((utf8 & 0x40000000) >> 30);	// 1 bit
+		utf8 <<= 2;
+	} else if (utf8 & 0x03e00000) {
+		buf = SZ_GetSpace (sb, count = 5);
+		*buf = 0xf8 | ((utf8 & 0x30000000) >> 28);	// 2 bits
+		utf8 <<= 4;
+	} else if (utf8 & 0x001f0000) {
+		buf = SZ_GetSpace (sb, count = 4);
+		*buf = 0xf0 | ((utf8 & 0x001c0000) >> 18);	// 3 bits
+		utf8 <<= 14;
+	} else if (utf8 & 0x0000f800) {
+		buf = SZ_GetSpace (sb, count = 3);
+		*buf = 0xe0 | ((utf8 & 0x0000f000) >> 12);	// 4 bits
+		utf8 <<= 20;
+	} else if (utf8 & 0x00000780) {
+		buf = SZ_GetSpace (sb, count = 2);
+		*buf = 0xc0 | ((utf8 & 0x00000780) >> 7);	// 5 bits
+		utf8 <<= 25;
+	} else {
+		buf = SZ_GetSpace (sb, count = 1);
+		*buf = utf8;
+		return;
+	}
+	while (--count) {
+		*buf++ = 0x80 | ((utf8 & 0xfc000000) >> 26);
+		utf8 <<= 6;
+	}
+}
+
 // reading functions ==========================================================
 
 void
@@ -343,3 +382,58 @@ MSG_ReadAngle16 (qmsg_t *msg)
 {
 	return MSG_ReadShort (msg) * (360.0 / 65536.0);
 }
+
+int
+MSG_ReadUTF8 (qmsg_t *msg)
+{
+	byte       *buf, *start, c;
+	int         val = 0, count;
+
+	if (msg->badread || msg->message->cursize == msg->readcount) {
+		msg->badread = true;
+		return -1;
+	}
+	buf = start = msg->message->data + msg->readcount;
+
+	c = *buf++;
+	if (c < 0x80) {
+		val = c;
+		count = 1;
+	} else if (c < 0xc0) {
+		msg->badread = true;
+		return -1;
+	} else if (c < 0xe0) {
+		count = 2;
+		val = c & 0x1f;
+	} else if (c < 0xf0) {
+		count = 3;
+		val = c & 0x0f;
+	} else if (c < 0xf8) {
+		count = 4;
+		val = c & 0x07;
+	} else if (c < 0xfc) {
+		count = 5;
+		val = c & 0x03;
+	} else if (c < 0xfe) {
+		count = 6;
+		val = c & 0x01;
+	} else {
+		msg->badread = true;
+		return -1;
+	}
+	if (count > (msg->message->cursize - msg->readcount)) {
+		msg->badread = true;
+		return -1;
+	}
+	while (--count) {
+		c = *buf++;
+		if ((c & 0xc0) != 0x80) {
+			msg->badread = true;
+			return -1;
+		}
+		val <<= 6;
+		val |= c & 0x3f;
+	}
+	msg->readcount += buf - start;
+	return val;
+}