add str_utf8_stats to base system

src/base/system.cpp

@@ -3454,6 +3454,23 @@ void str_utf8_copy(char *dst, const char *src, int dst_size)
 	str_utf8_truncate(dst, dst_size, src, dst_size);
 }
 
+void str_utf8_stats(const char *str, int max_size, int max_count, int *size, int *count)
+{
+	*size = 0;
+	*count = 0;
+	while(str[*size] && *size < max_size && *count < max_count)
+	{
+		int new_size = str_utf8_forward(str, *size);
+		if(new_size != *size)
+		{
+			if(new_size >= max_size || *count >= max_count)
+				break;
+			*size = new_size;
+			++(*count);
+		}
+	}
+}
+
 unsigned str_quickhash(const char *str)
 {
 	unsigned hash = 5381;

src/base/system.h

@@ -2089,6 +2089,23 @@ int str_utf8_check(const char *str);
 */
 void str_utf8_copy(char *dst, const char *src, int dst_size);
 
+/*
+	Function: str_utf8_stats
+		Determines the byte size and utf8 character count of a string.
+
+	Parameters:
+		str - Pointer to the string.
+		max_size - Maximum number of bytes to count.
+		max_count - Maximum number of utf8 characters to count.
+		size - Pointer to store size (number of non-zero bytes) of the string.
+		count - Pointer to store count of utf8 characters of the string.
+
+	Remarks:
+		- Assumes nothing about the encoding of the string.
+		  It's the users responsibility to make sure the bounds are aligned.
+*/
+void str_utf8_stats(const char *str, int max_size, int max_count, int *size, int *count);
+
 /*
 	Function: str_next_token
 		Writes the next token after str into buf, returns the rest of the string.

src/test/str.cpp

@@ -255,6 +255,43 @@ TEST(Str, StrCopyUtf8)
 	EXPECT_STREQ(aBuf, "DDNet最好了");
 }
 
+TEST(Str, Utf8Stats)
+{
+	int Size, Count;
+
+	str_utf8_stats("abc", 4, 3, &Size, &Count);
+	EXPECT_EQ(Size, 3);
+	EXPECT_EQ(Count, 3);
+
+	str_utf8_stats("abc", 2, 3, &Size, &Count);
+	EXPECT_EQ(Size, 1);
+	EXPECT_EQ(Count, 1);
+
+	str_utf8_stats("", 1, 0, &Size, &Count);
+	EXPECT_EQ(Size, 0);
+	EXPECT_EQ(Count, 0);
+
+	str_utf8_stats("abcde", 6, 5, &Size, &Count);
+	EXPECT_EQ(Size, 5);
+	EXPECT_EQ(Count, 5);
+
+	str_utf8_stats("любовь", 13, 6, &Size, &Count);
+	EXPECT_EQ(Size, 12);
+	EXPECT_EQ(Count, 6);
+
+	str_utf8_stats("abc愛", 7, 4, &Size, &Count);
+	EXPECT_EQ(Size, 6);
+	EXPECT_EQ(Count, 4);
+
+	str_utf8_stats("abc愛", 6, 4, &Size, &Count);
+	EXPECT_EQ(Size, 3);
+	EXPECT_EQ(Count, 3);
+
+	str_utf8_stats("любовь", 13, 3, &Size, &Count);
+	EXPECT_EQ(Size, 6);
+	EXPECT_EQ(Count, 3);
+}
+
 TEST(Str, StrTime)
 {
 	char aBuf[32] = "foobar";