public void UpdateRange(Token token)
{
if (token.StartPos < startpos) startpos = token.StartPos;
if (token.EndPos > endpos) endpos = token.EndPos;
}
/// <summary>
/// executes a lookahead of the next token
/// and will advance the scan on the input string
/// </summary>
/// <returns></returns>
public Token Scan(params TokenType[] expectedtokens)
{
Token tok = LookAhead(expectedtokens); // temporarely retrieve the lookahead
LookAheadToken = null; // reset lookahead token, so scanning will continue
StartPos = tok.EndPos;
EndPos = tok.EndPos; // set the tokenizer to the new scan position
return tok;
}
/// <summary>
/// returns token with longest best match
/// </summary>
/// <returns></returns>
public Token LookAhead(params TokenType[] expectedtokens)
{
int i;
int startpos = StartPos;
Token tok = null;
List<TokenType> scantokens;
// this prevents double scanning and matching
// increased performance
if (LookAheadToken != null
&& LookAheadToken.Type != TokenType._UNDETERMINED_
&& LookAheadToken.Type != TokenType._NONE_) return LookAheadToken;
// if no scantokens specified, then scan for all of them (= backward compatible)
if (expectedtokens.Length == 0)
scantokens = Tokens;
else
{
scantokens = new List<TokenType>(expectedtokens);
scantokens.AddRange(SkipList);
}
do
{
int len = -1;
TokenType index = (TokenType)int.MaxValue;
string input = Input.Substring(startpos);
tok = new Token(startpos, EndPos);
for (i = 0; i < scantokens.Count; i++)
{
Regex r = Patterns[scantokens[i]];
Match m = r.Match(input);
if (m.Success && m.Index == 0 && ((m.Length > len) || (scantokens[i] < index && m.Length == len )))
{
len = m.Length;
index = scantokens[i];
}
}
if (index >= 0 && len >= 0)
{
tok.EndPos = startpos + len;
tok.Text = Input.Substring(tok.StartPos, len);
tok.Type = index;
}
else if (tok.StartPos < tok.EndPos - 1)
{
tok.Text = Input.Substring(tok.StartPos, 1);
}
if (SkipList.Contains(tok.Type))
{
startpos = tok.EndPos;
Skipped.Add(tok);
}
else
{
// only assign to non-skipped tokens
tok.Skipped = Skipped; // assign prior skips to this token
Skipped = new List<Token>(); //reset skips
}
}
while (SkipList.Contains(tok.Type));
LookAheadToken = tok;
return tok;
}
public void Init(string input)
{
this.Input = input;
StartPos = 0;
EndPos = 0;
CurrentLine = 0;
CurrentColumn = 0;
CurrentPosition = 0;
LookAheadToken = null;
}
public Token GetToken(TokenType type)
{
Token t = new Token(this.StartPos, this.EndPos);
t.Type = type;
return t;
}
public Scanner()
{
Regex regex;
Patterns = new Dictionary<TokenType, Regex>();
Tokens = new List<TokenType>();
LookAheadToken = null;
Skipped = new List<Token>();
SkipList = new List<TokenType>();
SkipList.Add(TokenType.WS);
regex = new Regex(@"\+", RegexOptions.Compiled);
Patterns.Add(TokenType.PLUS, regex);
Tokens.Add(TokenType.PLUS);
regex = new Regex(@"-", RegexOptions.Compiled);
Patterns.Add(TokenType.MINUS, regex);
Tokens.Add(TokenType.MINUS);
regex = new Regex(@"\*", RegexOptions.Compiled);
Patterns.Add(TokenType.STAR, regex);
Tokens.Add(TokenType.STAR);
regex = new Regex(@"\/", RegexOptions.Compiled);
Patterns.Add(TokenType.FSLASH, regex);
Tokens.Add(TokenType.FSLASH);
regex = new Regex(@"\^", RegexOptions.Compiled);
Patterns.Add(TokenType.EXPONENT, regex);
Tokens.Add(TokenType.EXPONENT);
regex = new Regex(@"\%", RegexOptions.Compiled);
Patterns.Add(TokenType.MODULO, regex);
Tokens.Add(TokenType.MODULO);
regex = new Regex(@"\(", RegexOptions.Compiled);
Patterns.Add(TokenType.LPAREN, regex);
Tokens.Add(TokenType.LPAREN);
regex = new Regex(@"\)", RegexOptions.Compiled);
Patterns.Add(TokenType.RPAREN, regex);
Tokens.Add(TokenType.RPAREN);
regex = new Regex(@"\[", RegexOptions.Compiled);
Patterns.Add(TokenType.LBRACKET, regex);
Tokens.Add(TokenType.LBRACKET);
regex = new Regex(@"\]", RegexOptions.Compiled);
Patterns.Add(TokenType.RBRACKET, regex);
Tokens.Add(TokenType.RBRACKET);
regex = new Regex(@"\{", RegexOptions.Compiled);
Patterns.Add(TokenType.LBRACE, regex);
Tokens.Add(TokenType.LBRACE);
regex = new Regex(@"\}", RegexOptions.Compiled);
Patterns.Add(TokenType.RBRACE, regex);
Tokens.Add(TokenType.RBRACE);
regex = new Regex(@"=", RegexOptions.Compiled);
Patterns.Add(TokenType.EQUAL, regex);
Tokens.Add(TokenType.EQUAL);
regex = new Regex(@"<", RegexOptions.Compiled);
Patterns.Add(TokenType.LT, regex);
Tokens.Add(TokenType.LT);
regex = new Regex(@">", RegexOptions.Compiled);
Patterns.Add(TokenType.GT, regex);
Tokens.Add(TokenType.GT);
regex = new Regex(@"<=", RegexOptions.Compiled);
Patterns.Add(TokenType.LTE, regex);
Tokens.Add(TokenType.LTE);
regex = new Regex(@">=", RegexOptions.Compiled);
Patterns.Add(TokenType.GTE, regex);
Tokens.Add(TokenType.GTE);
regex = new Regex(@"<>", RegexOptions.Compiled);
Patterns.Add(TokenType.NOTEQUAL, regex);
Tokens.Add(TokenType.NOTEQUAL);
regex = new Regex(@"[Aa][Nn][Dd]", RegexOptions.Compiled);
Patterns.Add(TokenType.AND, regex);
Tokens.Add(TokenType.AND);
regex = new Regex(@"[Oo][Rr]", RegexOptions.Compiled);
Patterns.Add(TokenType.OR, regex);
Tokens.Add(TokenType.OR);
regex = new Regex(@"[Nn][Oo][Tt]", RegexOptions.Compiled);
Patterns.Add(TokenType.NOT, regex);
Tokens.Add(TokenType.NOT);
regex = new Regex(@"\.", RegexOptions.Compiled);
Patterns.Add(TokenType.DOT, regex);
Tokens.Add(TokenType.DOT);
regex = new Regex(@"\.\.", RegexOptions.Compiled);
Patterns.Add(TokenType.DOTDOT, regex);
Tokens.Add(TokenType.DOTDOT);
regex = new Regex(@"\?", RegexOptions.Compiled);
Patterns.Add(TokenType.QUIZ, regex);
Tokens.Add(TokenType.QUIZ);
regex = new Regex(@"\?:", RegexOptions.Compiled);
Patterns.Add(TokenType.QUIZCOLON, regex);
Tokens.Add(TokenType.QUIZCOLON);
regex = new Regex(@",", RegexOptions.Compiled);
Patterns.Add(TokenType.COMMA, regex);
Tokens.Add(TokenType.COMMA);
regex = new Regex(@":", RegexOptions.Compiled);
Patterns.Add(TokenType.COLON, regex);
Tokens.Add(TokenType.COLON);
regex = new Regex(@"^$", RegexOptions.Compiled);
Patterns.Add(TokenType.EOF, regex);
Tokens.Add(TokenType.EOF);
regex = new Regex(@"[Tt][Rr][Uu][Ee]", RegexOptions.Compiled);
Patterns.Add(TokenType.TRUE, regex);
Tokens.Add(TokenType.TRUE);
regex = new Regex(@"[Ff][Aa][Ll][Ss][Ee]", RegexOptions.Compiled);
Patterns.Add(TokenType.FALSE, regex);
Tokens.Add(TokenType.FALSE);
regex = new Regex(@"[Nn][Uu][Ll][Ll]", RegexOptions.Compiled);
Patterns.Add(TokenType.NULL, regex);
Tokens.Add(TokenType.NULL);
regex = new Regex(@"(\+|-)?[0-9]+", RegexOptions.Compiled);
Patterns.Add(TokenType.INTEGER, regex);
Tokens.Add(TokenType.INTEGER);
regex = new Regex(@"(\+|-)?[0-9]*\.[0-9]+", RegexOptions.Compiled);
Patterns.Add(TokenType.DECIMAL, regex);
Tokens.Add(TokenType.DECIMAL);
regex = new Regex(@"[0-9]+\.\.[0-9]+", RegexOptions.Compiled);
Patterns.Add(TokenType.INTRANGE, regex);
Tokens.Add(TokenType.INTRANGE);
regex = new Regex(@"""([^""\\]*(\\""[^""\\]*)*)""", RegexOptions.Compiled);
Patterns.Add(TokenType.STRING, regex);
Tokens.Add(TokenType.STRING);
regex = new Regex(@"[a-zA-Z][a-zA-Z0-9_]*\(", RegexOptions.Compiled);
Patterns.Add(TokenType.FUNCTION, regex);
Tokens.Add(TokenType.FUNCTION);
regex = new Regex(@"[a-zA-Z_$][a-zA-Z0-9_$]*", RegexOptions.Compiled);
Patterns.Add(TokenType.IDENTIFIER, regex);
Tokens.Add(TokenType.IDENTIFIER);
regex = new Regex(@"\s+", RegexOptions.Compiled);
Patterns.Add(TokenType.WS, regex);
Tokens.Add(TokenType.WS);
}
