public static String Decode(Utf8PointerLengthSlice text)
{
System.Text.StringBuilder builder = new System.Text.StringBuilder((int)text.length);
Utf8Pointer next = text.ptr;
Utf8Pointer limit = text.LimitPointer;
while (next < limit)
{
UInt32 c = Decode(ref next, limit);
builder.Append((Char)c);
}
return(builder.ToString());
}
public static UInt32 Decode(ref Utf8Pointer text, Utf8Pointer limit)
{
if (text >= limit)
{
throw new ArgumentException("Cannot pass an empty data segment to Utf8.Decode");
}
UInt32 c = text[0];
if (c <= 0x7F)
{
return(c);
}
if ((c & 0x40) == 0)
{
throw new Utf8Exception(Utf8ExceptionType.StartedInsideCodePoint);
}
text++;
if ((c & 0x20) == 0)
{
if (text >= limit)
{
throw new Utf8Exception(Utf8ExceptionType.MissingBytes);
}
return(((c << 6) & 0x7C0U) | (text[0] & 0x3FU));
}
if ((c & 0x10) == 0)
{
if (text + 1 >= limit)
{
throw new Utf8Exception(Utf8ExceptionType.MissingBytes);
}
return(((c << 12) & 0xF000U) | ((UInt32)(text[0] << 6) & 0xFC0U) | (text[1] & 0x3FU));
}
if ((c & 0x08) == 0)
{
if (text + 2 >= limit)
{
throw new Utf8Exception(Utf8ExceptionType.MissingBytes);
}
return(((c << 18) & 0x1C0000U) | ((UInt32)(text[0] << 12) & 0x3F000U) |
((UInt32)(text[1] << 6) & 0xFC0U) | (text[2] & 0x3FU));
}
throw new Utf8Exception(Utf8ExceptionType.OutOfRange);
}
unsafe void TestDecodeUtf8(Byte[] s, UInt32 start, UInt32 limit, params UInt32[] expectedChars)
{
foreach (var expected in expectedChars)
{
if (start >= limit)
{
Assert.Fail("Expected more decoded utf8 chars but input ended");
}
UInt32 encodedLength = limit - start;
{
var saveStart = start;
var decoded = Utf8.Decode(s, ref start, limit);
if (decoded != expected)
{
Assert.Fail("decodeUtf8: Expected '{0}' 0x{1} but decoded '{2}' 0x{3}",
expected, expected, decoded, decoded);
}
Console.WriteLine("decodeUtf8('{0}')", decoded);
}
fixed(byte *sPointer = s)
{
Utf8Pointer pointer = new Utf8Pointer(sPointer);
Utf8Pointer pointerLimit = new Utf8Pointer(sPointer + encodedLength);
var decoded = Utf8.Decode(ref pointer, pointerLimit);
if (decoded != expected)
{
Assert.Fail("decodeUtf8: Expected '{0}' 0x{1} but decoded '{2}' 0x{3}",
expected, expected, decoded, decoded);
}
}
}
if (start != limit)
{
Assert.Fail("Expected {0} characters but didn't get enough", expectedChars.Length);
}
}
public Utf8PointerLengthSlice(Utf8Pointer ptr, UInt32 length)
{
this.ptr = ptr;
this.length = length;
}
public Utf8PointerLimitSlice(Utf8Pointer ptr, Utf8Pointer limit)
{
this.ptr = ptr;
this.limit = limit;
}
