public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken token)
{
token = null;
if (inputBytes == null)
{
return(false);
}
if (currentByte != '(')
{
return(false);
}
var builder = stringBuilder;
var numberOfBrackets = 1;
var isEscapeActive = false;
var isLineBreaking = false;
var octalModeActive = false;
short[] octal = { 0, 0, 0 };
var octalsRead = 0;
while (inputBytes.MoveNext())
{
var b = inputBytes.CurrentByte;
var c = (char)b;
if (octalModeActive)
{
var nextCharacterOctal = c >= '0' && c <= '7';
if (nextCharacterOctal)
{
// left shift the octals.
LeftShiftOctal(c, octalsRead, octal);
octalsRead++;
}
if (octalsRead == 3 || !nextCharacterOctal)
{
var characterCode = OctalHelpers.FromOctalDigits(octal);
// For now :(
// TODO: I have a sneaking suspicion this is wrong, not sure what behaviour is for large octal numbers
builder.Append((char)characterCode);
octal[0] = 0;
octal[1] = 0;
octal[2] = 0;
octalsRead = 0;
octalModeActive = false;
}
if (nextCharacterOctal)
{
continue;
}
}
switch (c)
{
case ')':
isLineBreaking = false;
if (!isEscapeActive)
{
numberOfBrackets--;
}
isEscapeActive = false;
if (numberOfBrackets > 0)
{
builder.Append(c);
}
// TODO: Check for other ends of string where the string is improperly formatted. See commented method
numberOfBrackets = CheckForEndOfString(numberOfBrackets, inputBytes);
break;
case '(':
isLineBreaking = false;
if (!isEscapeActive)
{
numberOfBrackets++;
}
isEscapeActive = false;
builder.Append(c);
break;
// Escape
case '\\':
isLineBreaking = false;
// Escaped backslash
if (isEscapeActive)
{
builder.Append(c);
isEscapeActive = false;
}
else
{
isEscapeActive = true;
}
break;
default:
if (isLineBreaking)
{
if (ReadHelper.IsEndOfLine(c))
{
continue;
}
isLineBreaking = false;
builder.Append(c);
}
else if (isEscapeActive)
{
ProcessEscapedCharacter(c, builder, octal, ref octalModeActive, ref octalsRead, ref isLineBreaking);
isEscapeActive = false;
}
else
{
builder.Append(c);
}
break;
}
if (numberOfBrackets <= 0)
{
break;
}
}
StringToken.Encoding encodedWith;
string tokenStr;
if (builder.Length >= 2)
{
if (builder[0] == 0xFE && builder[1] == 0xFF)
{
var rawBytes = OtherEncodings.StringAsLatin1Bytes(builder.ToString());
tokenStr = Encoding.BigEndianUnicode.GetString(rawBytes).Substring(1);
encodedWith = StringToken.Encoding.Utf16BE;
}
else if (builder[0] == 0xFF && builder[1] == 0xFE)
{
var rawBytes = OtherEncodings.StringAsLatin1Bytes(builder.ToString());
tokenStr = Encoding.Unicode.GetString(rawBytes).Substring(1);
encodedWith = StringToken.Encoding.Utf16;
}
else
{
tokenStr = builder.ToString();
encodedWith = StringToken.Encoding.Iso88591;
}
}
else
{
tokenStr = builder.ToString();
encodedWith = StringToken.Encoding.Iso88591;
}
builder.Clear();
token = new StringToken(tokenStr, encodedWith);
return(true);
}
public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken token)
{
var builder = new StringBuilder();
token = null;
if (inputBytes == null)
{
return(false);
}
if (currentByte != '(')
{
return(false);
}
int numberOfBrackets = 1;
bool isEscapeActive = false;
bool isLineBreaking = false;
bool octalModeActive = false;
short[] octal = { 0, 0, 0 };
int octalsRead = 0;
while (inputBytes.MoveNext())
{
var b = inputBytes.CurrentByte;
var c = (char)b;
if (octalModeActive)
{
var nextCharacterOctal = c >= '0' && c <= '7';
if (nextCharacterOctal)
{
// left shift the octals.
LeftShiftOctal(c, octalsRead, octal);
octalsRead++;
}
if (octalsRead == 3 || !nextCharacterOctal)
{
var characterCode = OctalHelpers.FromOctalDigits(octal);
// For now :(
// TODO: I have a sneaking suspicion this is wrong, not sure what behaviour is for large octal numbers
builder.Append((char)characterCode);
octal[0] = 0;
octal[1] = 0;
octal[2] = 0;
octalsRead = 0;
octalModeActive = false;
}
if (nextCharacterOctal)
{
continue;
}
}
switch (c)
{
case ')':
isLineBreaking = false;
if (!isEscapeActive)
{
numberOfBrackets--;
}
isEscapeActive = false;
if (numberOfBrackets > 0)
{
builder.Append(c);
}
// TODO: Check for other ends of string where the string is improperly formatted. See commented method
// numberOfBrackets = CheckForEndOfString(inputBytes, numberOfBrackets);
break;
case '(':
isLineBreaking = false;
if (!isEscapeActive)
{
numberOfBrackets++;
}
isEscapeActive = false;
builder.Append(c);
break;
// Escape
case '\\':
isLineBreaking = false;
// Escaped backslash
if (isEscapeActive)
{
builder.Append(c);
}
else
{
isEscapeActive = true;
}
break;
default:
if (isLineBreaking)
{
if (ReadHelper.IsEndOfLine(c))
{
continue;
}
isLineBreaking = false;
builder.Append(c);
}
else if (isEscapeActive)
{
ProcessEscapedCharacter(c, builder, octal, ref octalModeActive, ref octalsRead, ref isLineBreaking);
isEscapeActive = false;
}
else
{
builder.Append(c);
}
break;
}
if (numberOfBrackets <= 0)
{
break;
}
}
token = new StringToken(builder.ToString());
return(true);
}