int Scan()
{
for (;;)
{
int next;
state = currentStart;
while ((next = NextState()) == goStart)
{
code = buffer.Read();
}
state = next;
code = buffer.Read();
while ((next = NextState()) > eofNum)
{
state = next;
code = buffer.Read();
}
if (state <= maxAccept)
{
#region ActionSwitch
#pragma warning disable 162
switch (state)
{
case eofNum:
switch (currentStart)
{
case 11:
if (utfX == 0 && uppr == 0)
{
return(-1); /* raw ascii */
}
else if (uppr * 10 > utfX)
{
return(0); /* default code page */
}
else
{
return(65001); /* UTF-8 encoding */
}
break;
}
return(EndToken);
case 1: // Recognized '{Upper128}', Shortest string "\xC0"
case 2: // Recognized '{Upper128}', Shortest string "\x80"
case 3: // Recognized '{Upper128}', Shortest string "\xE0"
case 4: // Recognized '{Upper128}', Shortest string "\xF0"
uppr++;
break;
case 5: // Recognized '{Utf8pfx4}{Utf8cont}', Shortest string "\xF0\x80"
uppr += 2;
break;
case 6: // Recognized '{Utf8pfx4}{Utf8cont}{2}', Shortest string "\xF0\x80\x80"
uppr += 3;
break;
case 7: // Recognized '{Utf8pfx4}{Utf8cont}{3}', Shortest string "\xF0\x80\x80\x80"
utfX += 3;
break;
case 8: // Recognized '{Utf8pfx3}{Utf8cont}', Shortest string "\xE0\x80"
uppr += 2;
break;
case 9: // Recognized '{Utf8pfx3}{Utf8cont}{2}', Shortest string "\xE0\x80\x80"
utfX += 2;
break;
case 10: // Recognized '{Utf8pfx2}{Utf8cont}', Shortest string "\xC0\x80"
utfX++;
break;
default:
break;
}
#pragma warning restore 162
#endregion
}
}
}
public void SetSource(Stream source)
{
buffer = new BuildBuffer(source);
code = buffer.Read();
}
public void SetSource(System.IO.Stream source)
{
this.buffer = new BuildBuffer(source);
code = buffer.Read();
}
public void SetSource(System.IO.Stream source)
{
this.buffer = new BuildBuffer(source);
code = buffer.Read();
}