private Token IsString(string text, int lineNumber)
{
Token token = null;
//true when we find an invalid escape sequence \
bool isInvalid = false;
//if it starts with double quote, search for the ending quote
if (text.StartsWith("\"") && text.Length > 1)
{
for (int i = 1; i < text.Length; i++)
{
if (text[i] == '\\')
{
//isInvalid = true;
}
//l_num break
else if (text[i] == '\r' || text[i] == '\n')
{
//don't include the new l_num character in the value part
token = new Token(ClassPart.INVALID, text.Substring(0, i), lineNumber);
break;
}
//find the end of string and also deal the escape condition
if (text[i] == '\"' && text[i - 1] != '\\')
{
if (isInvalid)
{
return(new Token(ClassPart.INVALID, text.Substring(0, i + 1), lineNumber));
}
return(new Token(ClassPart.STRING_CONSTANT, text.Substring(0, i + 1), lineNumber));
}
//if text ends and we don't find the ending quote for string, means string is not closed
if (i == text.Length - 1)
{
token = new Token(ClassPart.INVALID, text, lineNumber);
}
}
}
//if it starts with single quote, search for the ending single
else if ((text.StartsWith("\'") && text.Length > 2))
{
for (int i = 0; i < 3; i++)
{
Console.WriteLine(text[i]);
if (text[2] == '\'')
{
token = new Token(ClassPart.CHAR_CONSTANT, text.Substring(0, i), lineNumber);
break;
}
// //l_num break
// if (text[i] == '\r' || text[i] == '\n')
// {
// //don't include the new l_num character in the value part
// token = new Token(ClassPart.CHAR_CONSTANT, text.Substring(0, i), lineNumber);
// break;
// }
// //find the end of string and also deal the escape condition
// if (text[i] == '\'' && text[i - 1] != '\\')
// {
// return new Token(ClassPart.CHAR_CONSTANT, text.Substring(0, i + 1), lineNumber);
// }
// //if text ends and we don't find the ending quote for string, means string is not closed
// if (i == text.Length - 1)
// token = new Token(ClassPart.INVALID, text, lineNumber);
}
}
//if string is too short and incomplete
else if ((text.StartsWith("\"") || text.StartsWith("'")) && text.Length < 2)
{
token = new Token(ClassPart.INVALID, text, lineNumber);
}
return(token);
}
/// <summary>
/// Tokenize the code according to the available grammar
/// </summary>
/// <param name="rawText"></param>
/// <returns></returns>
public List <Token> Analyze(string rawText)
{
var tokens = new List <Token>();
//main ind to track the current position
int postion = 0;
int noOflines = 1;
string text = rawText;
int length = rawText.Length;
while (postion < length)
{
// Remove spaces,tabs and new noOfLines from front of string and increase the current ind
var items = TrimStart(text);
noOflines += items.Item3;
var count = items.Item1 + items.Item2 + items.Item3;
postion += count;
text = text.Substring(count);
//if we are at the end of text
if (string.IsNullOrEmpty(text))
{
break;
}
Token token;
//check keyword
token = IsKeyword(text, noOflines);
//check punctuators
if (token == null)
{
token = IsPunctuator(text, noOflines);
}
//Check string
if (token == null)
{
token = IsString(text, noOflines);
}
//Check comment
if (token == null)
{
var val = IsComment(text, noOflines);
token = val.Item1;
noOflines += val.Item2;
}
//check operator
if (token == null)
{
token = IsOperator(text, noOflines);
}
if (token == null)
{
//now we have to break the word in order to validate it as arithmetic operators or identifier
var word = BreakWord(text);
token = IsDouble(word, noOflines);
if (token == null)
{
token = IsInt(word, noOflines);
}
if (token == null)
{
token = IsBool(word, noOflines);
}
if (token == null)
{
token = IsIdentifier(word, noOflines);
}
if (token == null)
{
token = new Token(ClassPart.INVALID, word, noOflines);
}
}
//increase current ind
postion += token.Value.Length;
//remove the token from text
text = text.Substring(token.Value.Length);
//add token
tokens.Add(token);
}
return(tokens);
}
