public override int GetChars(byte [] bytes, int byteIndex, int byteCount, char [] chars, int charIndex)
{
CheckRange(bytes, byteIndex, byteCount, chars, charIndex);
int byteEnd = byteIndex + byteCount;
int charStart = charIndex;
while (byteIndex < byteEnd)
{
if (bytes [byteIndex] < 0x80)
{
chars [charIndex++] = (char)bytes [byteIndex++];
continue;
}
else if (bytes [byteIndex] == 0x80)
{
// Euro sign - actually it is obsolete,
// now it's just reserved but not used
chars [charIndex++] = '\u20AC';
byteIndex++;
continue;
}
else if (bytes [byteIndex] == 0xFF)
{
// invalid data - fill '?'
chars [charIndex++] = '?';
byteIndex++;
continue;
}
else if (byteIndex + 1 >= byteEnd)
{
//incomplete1 = bytes [byteIndex++];
//incomplete2 = -1;
//incomplete3 = -1;
break; // incomplete tail.
}
byte second = bytes [byteIndex + 1];
if (second == 0x7F || second == 0xFF)
{
// invalid data
chars [charIndex++] = '?';
byteIndex += 2;
}
else if (0x30 <= second && second <= 0x39)
{
// UCS mapping
if (byteIndex + 3 >= byteEnd)
{
// incomplete tail.
//incomplete1 = bytes [byteIndex];
//incomplete2 = bytes [byteIndex + 1];
//if (byteIndex + 3 == byteEnd)
// incomplete3 = bytes [byteIndex + 2];
break;
}
long value = GB18030Source.FromGBX(bytes, byteIndex);
if (value < 0)
{
// invalid data.
chars [charIndex++] = '?';
byteIndex -= (int)value;
}
else if (value >= 0x10000)
{
// UTF16 surrogate
value -= 0x10000;
chars [charIndex++] = (char)(value / 0x400 + 0xD800);
chars [charIndex++] = (char)(value % 0x400 + 0xDC00);
byteIndex += 4;
}
else
{
// UTF16 BMP
chars [charIndex++] = (char)value;
byteIndex += 4;
}
}
else
{
byte first = bytes [byteIndex];
int ord = ((first - 0x81) * 191 + second - 0x40) * 2;
char c1 = ord < 0 || ord >= gb2312.n2u.Length ?
'\0' : (char)(gb2312.n2u [ord] + gb2312.n2u [ord + 1] * 256);
if (c1 == 0)
{
chars [charIndex++] = '?';
}
else
{
chars [charIndex++] = c1;
}
byteIndex += 2;
}
}
return(charIndex - charStart);
}
public override int GetCharCount(byte [] bytes, int start, int len)
{
CheckRange(bytes, start, len);
int end = start + len;
int ret = 0;
while (start < end)
{
if (bytes [start] < 0x80)
{
ret++;
start++;
continue;
}
else if (bytes [start] == 0x80)
{
// Euro sign - actually it is obsolete,
// now it's just reserved but not used
ret++;
start++;
continue;
}
else if (bytes [start] == 0xFF)
{
// invalid data - fill '?'
ret++;
start++;
continue;
}
else if (start + 1 >= end)
{
// incomplete1 = bytes [start];
// incomplete2 = -1;
// incomplete3 = -1;
ret++;
break; // incomplete tail.
}
byte second = bytes [start + 1];
if (second == 0x7F || second == 0xFF)
{
// invalid data
ret++;
start += 2;
continue;
}
else if (0x30 <= second && second <= 0x39)
{
// UCS mapping
if (start + 3 >= end)
{
// incomplete tail.
// incomplete1 = bytes [start];
// incomplete2 = bytes [start + 1];
// if (start + 3 == end)
// incomplete3 = bytes [start + 2];
ret += start + 3 == end ? 3 : 2;
break;
}
long value = GB18030Source.FromGBX(bytes, start);
if (value < 0)
{
// invalid data.
ret++;
start -= (int)value;
}
else if (value >= 0x10000)
{
// UTF16 surrogate
ret += 2;
start += 4;
}
else
{
// UTF16 BMP
ret++;
start += 4;
}
}
else
{
// GB2312 mapping
start += 2;
ret++;
}
}
return(ret);
}