private Token ReadCommentRestOfLine(LexerIndex index)
{
var start = index.Index;
while (index.Next())
{
if (index.Current == EndOfLine[0])
{
if (index.Peek == EndOfLine[1])
{
index.Next();
}
break;
}
}
return(new Token(index.Text, start, index.Index + 1, TokenType.Comment));
}
private bool MoveToFirstNonWhite(LexerIndex index)
{
while (IsWhiteSpace(index.Current))
{
if (!index.Next())
{
return(false);
}
}
return(true);
}
private Token ReadNumber(LexerIndex index)
{
var start = index.Index;
while (IsDigit(index.Peek) || index.Current == DecimalPoint)
{
index.Next();
}
return(new Token(index.Text, start, index.Index + 1, TokenType.Litteral));
}
private Token ReadWord(LexerIndex index)
{
var start = index.Index;
while (IsLetterOrDigit(index.Peek))
{
index.Next();
}
return(new Token(index.Text, start, index.Index + 1, TokenType.Word));
}
private Token ReadComment(LexerIndex index)
{
var start = index.Index;
index.Next();
while (index.Next())
{
if (index.Current == CommentEnd[0])
{
if (index.Peek == CommentEnd[1])
{
index.Next();
break;
}
}
}
return(new Token(index.Text, start, index.Index + 1, TokenType.Comment));
}
private Token ReadVariable(LexerIndex index)
{
if (!index.PeekPeek.HasValue)
{
return(CreateSingelCharacterToken(index, TokenType.Unknown));
}
var start = index.Index;
while (IsLetterOrDigit(index.Peek) || IsStartVariable(index.Peek))
{
index.Next();
}
return(new Token(index.Text, start, index.Index + 1, TokenType.Variable));
}
private Token ReadStringLitteral(LexerIndex index)
{
var start = index.Index;
bool lastWasEscape = false;
while (index.Next())
{
if (lastWasEscape)
{
lastWasEscape = false;
}
else if (index.Current == StringLitteralEscape)
{
lastWasEscape = true;
}
else if (index.Current == StringLitteralEnd)
{
return(new Token(index.Text, start, index.Index - 1, TokenType.StringLitteral));
}
}
return(new Token(index.Text, start, index.Index - 1, TokenType.Unknown));
}
public Token[] Tokenize(char[] text)
{
var index = new LexerIndex(text);
var tokens = new List<Token>();
while (index.Next())
{
if (!MoveToFirstNonWhite(index))
break;
else if (IsSpecialCharacter(index.Current))
{
if (index.Current == CommentRestOfLine[0] && index.Current == CommentRestOfLine[1])
tokens.Add(ReadCommentRestOfLine(index));
else if (index.Current == CommentStart[0] && index.Current == CommentStart[1])
tokens.Add(ReadComment(index));
else if (IsEndOfQuery(index.Current))
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndOfQuery));
else
tokens.Add(CreateSingelCharacterToken(index, TokenType.SpecialCharacter));
}
else if (IsDigit(index.Current))
{
tokens.Add(ReadNumber(index));
}
else if (IsStartVariable(index.Current))
{
tokens.Add(ReadVariable(index));
}
else if (IsLetter(index.Current))
{
tokens.Add(ReadWord(index));
}
else if (index.Current == StartParenthesis)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.StartParenthesis));
}
else if (index.Current == EndParenthesis)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndParenthesis));
}
else if (IsStartIdentifierDelimiter(index.Current))
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.StartIdentifierDelimiter));
}
else if (IsEndIdentifierDelimiter(index.Current))
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndIdentifierDelimiter));
}
else if (index.Current == SepparatorPoint)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.SepparatorPoint));
}
else if (index.Current == Comma)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.Comma));
}
else if (index.Current == StringLitteralStart)
{
tokens.Add(ReadStringLitteral(index));
}
else
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.Unknown));
}
}
return tokens.ToArray();
}
private Token ReadWord(LexerIndex index)
{
var start = index.Index;
while (IsLetterOrDigit(index.Peek))
{
index.Next();
}
return new Token(index.Text, start, index.Index + 1, TokenType.Word);
}
private Token ReadVariable(LexerIndex index)
{
if (!index.PeekPeek.HasValue)
return CreateSingelCharacterToken(index, TokenType.Unknown);
var start = index.Index;
while (IsLetterOrDigit(index.Peek) || IsStartVariable(index.Peek))
{
index.Next();
}
return new Token(index.Text, start, index.Index + 1, TokenType.Variable);
}
private Token ReadStringLitteral(LexerIndex index)
{
var start = index.Index;
bool lastWasEscape = false;
while (index.Next())
{
if (lastWasEscape)
lastWasEscape = false;
else if (index.Current == StringLitteralEscape)
lastWasEscape = true;
else if (index.Current == StringLitteralEnd)
return new Token(index.Text, start, index.Index - 1, TokenType.StringLitteral);
}
return new Token(index.Text, start, index.Index - 1, TokenType.Unknown);
}
private Token ReadNumber(LexerIndex index)
{
var start = index.Index;
while (IsDigit(index.Peek) || index.Current == DecimalPoint)
{
index.Next();
}
return new Token(index.Text, start, index.Index + 1, TokenType.Litteral);
}
private Token ReadCommentRestOfLine(LexerIndex index)
{
var start = index.Index;
while (index.Next())
{
if (index.Current == EndOfLine[0])
{
if (index.Peek == EndOfLine[1])
index.Next();
break;
}
}
return new Token(index.Text, start, index.Index + 1, TokenType.Comment);
}
private Token ReadComment(LexerIndex index)
{
var start = index.Index;
index.Next();
while (index.Next())
{
if (index.Current == CommentEnd[0])
{
if (index.Peek == CommentEnd[1])
{
index.Next();
break;
}
}
}
return new Token(index.Text, start, index.Index + 1, TokenType.Comment);
}
private bool MoveToFirstNonWhite(LexerIndex index)
{
while (IsWhiteSpace(index.Current))
{
if (!index.Next())
return false;
}
return true;
}
public Token[] Tokenize(char[] text)
{
var index = new LexerIndex(text);
var tokens = new List <Token>();
while (index.Next())
{
if (!MoveToFirstNonWhite(index))
{
break;
}
else if (IsSpecialCharacter(index.Current))
{
if (index.Current == CommentRestOfLine[0] && index.Current == CommentRestOfLine[1])
{
tokens.Add(ReadCommentRestOfLine(index));
}
else if (index.Current == CommentStart[0] && index.Current == CommentStart[1])
{
tokens.Add(ReadComment(index));
}
else if (IsEndOfQuery(index.Current))
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndOfQuery));
}
else
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.SpecialCharacter));
}
}
else if (IsDigit(index.Current))
{
tokens.Add(ReadNumber(index));
}
else if (IsStartVariable(index.Current))
{
tokens.Add(ReadVariable(index));
}
else if (IsLetter(index.Current))
{
tokens.Add(ReadWord(index));
}
else if (index.Current == StartParenthesis)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.StartParenthesis));
}
else if (index.Current == EndParenthesis)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndParenthesis));
}
else if (IsStartIdentifierDelimiter(index.Current))
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.StartIdentifierDelimiter));
}
else if (IsEndIdentifierDelimiter(index.Current))
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.EndIdentifierDelimiter));
}
else if (index.Current == SepparatorPoint)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.SepparatorPoint));
}
else if (index.Current == Comma)
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.Comma));
}
else if (index.Current == StringLitteralStart)
{
tokens.Add(ReadStringLitteral(index));
}
else
{
tokens.Add(CreateSingelCharacterToken(index, TokenType.Unknown));
}
}
return(tokens.ToArray());
}