From 92d08d07de070c7b2194108a55ba8717d35210c8 Mon Sep 17 00:00:00 2001
From: TsFreddie <tsfreddiewang@gmail.com>
Date: Fri, 4 Sep 2020 17:34:27 +0100
Subject: [PATCH] add str_utf8_copy

---
 src/base/system.c                   | 44 +++++++++++++++++++++++++++++
 src/base/system.h                   | 17 +++++++++++
 src/engine/client/input.cpp         |  2 +-
 src/game/client/components/chat.cpp |  4 +--
 4 files changed, 64 insertions(+), 3 deletions(-)

diff --git a/src/base/system.c b/src/base/system.c
index 8f398b542..060924861 100644
--- a/src/base/system.c
+++ b/src/base/system.c
@@ -3182,6 +3182,50 @@ int str_utf8_check(const char *str)
 	return 1;
 }
 
+void str_utf8_copy(char *dst, const char *src, int dst_size)
+{
+	strncpy(dst, src, dst_size);
+	dst[dst_size-1] = 0; /* assure null termination */
+
+	// check whether we need to remove a broken utf8 character
+	char *end = dst + (dst_size-2);
+	char *ptr = end;
+	while(ptr > dst)
+	{
+		if((*ptr&0xC0) == 0x80)
+		{
+			ptr--;
+		}
+		else if((*ptr&0x80) == 0)
+		{
+			return;
+		}
+		else if((*ptr&0xE0) == 0xC0)
+		{
+			if(end-ptr != 1)
+				*ptr = 0;
+			return;
+		}
+		else if((*ptr&0xF0) == 0xE0)
+		{
+			if(end-ptr != 2)
+				*ptr = 0;
+			return;
+		}
+		else if((*ptr&0xF8) == 0xF0)
+		{
+			if(end-ptr != 3)
+				*ptr = 0;
+			return;
+		}
+		else
+		{
+			return;
+		}
+	}
+
+	*ptr = 0;
+}
 
 unsigned str_quickhash(const char *str)
 {
diff --git a/src/base/system.h b/src/base/system.h
index b936b33cc..a511e9ed3 100644
--- a/src/base/system.h
+++ b/src/base/system.h
@@ -1918,6 +1918,23 @@ int str_utf16le_encode(char *ptr, int chr);
 */
 int str_utf8_check(const char *str);
 
+/*
+	Function: str_utf8_copy
+		Copies a utf8 string to a buffer.
+
+	Parameters:
+		dst - Pointer to a buffer that shall receive the string.
+		src - utf8 string to be copied.
+		dst_size - Size of the buffer dst.
+
+	Remarks:
+		- The strings are treated as zero-terminated strings.
+		- Guarantees that dst string will contain zero-termination.
+		- Guarantees that dst always ends with a valid utf8 sequence.
+		- Does not guarantee the entire string is a valid utf8 string.
+*/
+void str_utf8_copy(char *dst, const char *src, int dst_size);
+
 /*
 	Function: str_next_token
 		Writes the next token after str into buf, returns the rest of the string.
diff --git a/src/engine/client/input.cpp b/src/engine/client/input.cpp
index 310c6e852..9171ecd3c 100644
--- a/src/engine/client/input.cpp
+++ b/src/engine/client/input.cpp
@@ -26,7 +26,7 @@ void CInput::AddEvent(char *pText, int Key, int Flags)
 		if(!pText)
 			m_aInputEvents[m_NumEvents].m_aText[0] = 0;
 		else
-			str_copy(m_aInputEvents[m_NumEvents].m_aText, pText, sizeof(m_aInputEvents[m_NumEvents].m_aText));
+			str_utf8_copy(m_aInputEvents[m_NumEvents].m_aText, pText, sizeof(m_aInputEvents[m_NumEvents].m_aText));
 		m_aInputEvents[m_NumEvents].m_InputCount = m_InputCounter;
 		m_NumEvents++;
 	}
diff --git a/src/game/client/components/chat.cpp b/src/game/client/components/chat.cpp
index d00428577..539c1d993 100644
--- a/src/game/client/components/chat.cpp
+++ b/src/game/client/components/chat.cpp
@@ -168,14 +168,14 @@ bool CChat::OnInput(IInput::CEvent Event)
 				if(Text[i] == '\n')
 				{
 					int max = minimum(i - Begin + 1, (int)sizeof(Line));
-					str_copy(Line, Text + Begin, max);
+					str_utf8_copy(Line, Text + Begin, max);
 					Begin = i+1;
 					SayChat(Line);
 					while(Text[i] == '\n') i++;
 				}
 			}
 			int max = minimum(i - Begin + 1, (int)sizeof(Line));
-			str_copy(Line, Text + Begin, max);
+			str_utf8_copy(Line, Text + Begin, max);
 			Begin = i+1;
 			m_Input.Add(Line);
 		}