From cc54db6e6df481e94d669f58fcbf58810a02504a Mon Sep 17 00:00:00 2001
From: Christoph Oelckers <coelckers@users.noreply.github.com>
Date: Sat, 2 Dec 2017 21:21:57 +0100
Subject: [PATCH] - let the 3 relevant text functions handle UTF-8 strings

These functions are: DCanvas::DrawTextCommon, V_BreakLines and FFont::StringWidth.
This will allow strings from UTF-8 encoded assets to display properly, but also handle the OpenAL device name on international systems, as this will be returned as an UTF-8 string.

Due to backwards compatibility needs the decoding function is rather lax to allow both UTF-8 and ISO 8859-1 to pass through correctly - and this also implies that it will allow mixed encodings which may happen if strings from different sources get concatenated.
---
 src/v_font.cpp | 10 +++----
 src/v_text.cpp | 76 ++++++++++++++++++++++++++++++++++++++++++++++++--
 src/v_text.h   |  2 ++
 3 files changed, 81 insertions(+), 7 deletions(-)

diff --git a/src/v_font.cpp b/src/v_font.cpp
index b475fb558..9716f2744 100644
--- a/src/v_font.cpp
+++ b/src/v_font.cpp
@@ -890,9 +890,10 @@ int FFont::StringWidth(const uint8_t *string) const
 
 	while (*string)
 	{
-		if (*string == TEXTCOLOR_ESCAPE)
+		auto chr = GetCharFromString(string);
+		if (chr == TEXTCOLOR_ESCAPE)
 		{
-			++string;
+			// We do not need to check for UTF-8 in here.
 			if (*string == '[')
 			{
 				while (*string != '\0' && *string != ']')
@@ -906,16 +907,15 @@ int FFont::StringWidth(const uint8_t *string) const
 			}
 			continue;
 		}
-		else if (*string == '\n')
+		else if (chr == '\n')
 		{
 			if (w > maxw)
 				maxw = w;
 			w = 0;
-			++string;
 		}
 		else
 		{
-			w += GetCharWidth(*string++) + GlobalKerning;
+			w += GetCharWidth(chr) + GlobalKerning;
 		}
 	}
 
diff --git a/src/v_text.cpp b/src/v_text.cpp
index cf8ce8cc7..f4829ec18 100644
--- a/src/v_text.cpp
+++ b/src/v_text.cpp
@@ -51,6 +51,78 @@
 
 int ListGetInt(VMVa_List &tags);
 
+//==========================================================================
+//
+// reads one character from the string.
+// This can handle both ISO 8859-1 and UTF-8, as well as mixed strings
+// between both encodings, which may happen if inconsistent encoding is 
+// used between different files in a mod.
+//
+//==========================================================================
+
+int GetCharFromString(const uint8_t *&string)
+{
+	int z, y, x;
+
+	z = *string++;
+
+	if (z < 192)
+	{
+		return z;
+	}
+	else if (z <= 223)
+	{
+		y = *string++;
+		if (y < 128 || y >= 192)
+		{
+			// not an UTF-8 sequence so return the first byte unchanged
+			string--;
+		}
+		else
+		{
+			z = (z - 192) * 64 + (y - 128);
+		}
+	}
+	else if (z >= 224 && z <= 239)
+	{
+		y = *string++;
+		if (y < 128 || y >= 192)
+		{
+			// not an UTF-8 sequence so return the first byte unchanged
+			string--;
+		}
+		else
+		{
+			x = *string++;
+			if (x < 128 || x >= 192)
+			{
+				// not an UTF-8 sequence so return the first byte unchanged
+				string -= 2;
+			}
+			else
+			{
+				z = (z - 224) * 4096 + (y - 128) * 64 + (x - 128);
+			}
+		}
+	}
+	else if (z >= 240)
+	{
+		y = *string++;
+		if (y < 128 || y >= 192)
+		{
+			// not an UTF-8 sequence so return the first byte unchanged
+			string--;
+		}
+		else
+		{
+			// we do not support 4-Byte UTF-8 here
+			string += 2;
+			return '?';
+		}
+	}
+	return z;
+}
+
 //==========================================================================
 //
 // DrawChar
@@ -170,7 +242,7 @@ void DCanvas::DrawTextCommon(FFont *font, int normalcolor, double x, double y, c
 
 	while ((const char *)ch - string < parms.maxstrlen)
 	{
-		c = *ch++;
+		c = GetCharFromString(ch);
 		if (!c)
 			break;
 
@@ -288,7 +360,7 @@ FBrokenLines *V_BreakLines (FFont *font, int maxwidth, const uint8_t *string, bo
 
 	w = 0;
 
-	while ( (c = *string++) )
+	while ( (c = GetCharFromString(string)) )
 	{
 		if (c == TEXTCOLOR_ESCAPE)
 		{
diff --git a/src/v_text.h b/src/v_text.h
index b76024fa5..cd7d3e6ad 100644
--- a/src/v_text.h
+++ b/src/v_text.h
@@ -86,4 +86,6 @@ inline FBrokenLines *V_BreakLines (FFont *font, int maxwidth, const char *str, b
 inline FBrokenLines *V_BreakLines (FFont *font, int maxwidth, const FString &str, bool preservecolor = false, unsigned int *count = nullptr)
  { return V_BreakLines (font, maxwidth, (const uint8_t *)str.GetChars(), preservecolor, count); }
 
+int GetCharFromString(const uint8_t *&string);
+
 #endif //__V_TEXT_H__