public static List<Lexeme> Analyze(String text)
{
List<Lexeme> lexemes = new List<Lexeme>();
State currentState = new DefaultState();
int line = 1;
for (int i = 0; i < text.Length; i++)
{
char ch = text[i];
StateInfo stateInfo = new StateInfo(line);
currentState = currentState.Execute(ch, stateInfo);
if (stateInfo.Lexemes.Any())
{
foreach (Lexeme addedLexeme in stateInfo.Lexemes)
{
if (addedLexeme is NewLine)
line++;
// Skip multiple consecutive newlines
if (!lexemes.Any() || !(lexemes.Last() is NewLine) || !(addedLexeme is NewLine))
lexemes.Add(addedLexeme);
}
}
if (stateInfo.ReReadCharacter)
i--;
}
lexemes.Add(new EndOfCode(line));
return lexemes;
}
private Lexeme getLexemeFromCurrentText(StateInfo stateInfo)
{
if (Keyword.Check(currentText))
{
return new Keyword(currentText, stateInfo.Line);
}
else
{
return new Identifier(currentText, stateInfo.Line);
}
}
internal override State Execute(char character, StateInfo stateInfo)
{
if (character == '>')
{
stateInfo.Lexemes.Add(new Arrow(stateInfo.Line));
return new DefaultState();
}
stateInfo.Lexemes.Add(new InvalidCharacter("-", stateInfo.Line));
stateInfo.ReReadCharacter = true;
return new DefaultState();
}
internal override State Execute(char character, StateInfo stateInfo)
{
if (char.IsLetter(character) || character == '_')
{
currentText += character;
return new WordState(currentText);
}
currentText = currentText.Substring(0, currentText.Length - 1);
stateInfo.Lexemes.Add(getLexemeFromCurrentText(stateInfo));
stateInfo.Lexemes.Add(new InvalidCharacter(".", stateInfo.Line));
stateInfo.ReReadCharacter = true;
return new DefaultState();
}
internal override State Execute(char character, StateInfo stateInfo)
{
if (char.IsLetter(character) || char.IsDigit(character) || character == '_')
{
currentText += character;
return this;
}
if (character == '.')
{
currentText += ".";
return new DotState(currentText);
}
stateInfo.Lexemes.Add(getLexemeFromCurrentText(stateInfo));
stateInfo.ReReadCharacter = true;
return new DefaultState();
}
internal override State Execute(char character, StateInfo stateInfo)
{
if (isComment)
{
if (character == '\n')
{
stateInfo.ReReadCharacter = true;
return new DefaultState();
}
return this;
}
if (character == '/')
{
isComment = true;
return this;
}
stateInfo.Lexemes.Add(new InvalidCharacter("/", stateInfo.Line));
stateInfo.ReReadCharacter = true;
return new DefaultState();
}
internal override State Execute(char character, StateInfo stateInfo)
{
if (character == '/')
{
return new CommentState();
}
if (specialCharacters.Contains(character))
{
stateInfo.Lexemes.Add(new Character(character.ToString(), stateInfo.Line));
return this;
}
if (character == '\n')
{
stateInfo.Lexemes.Add(new NewLine(stateInfo.Line));
return this;
}
if (char.IsLetter(character) || character == '_')
{
return new WordState(character.ToString());
}
if (skipCharacters.Contains(character))
{
return this;
}
if (character == '-')
{
return new ArrowState();
}
stateInfo.Lexemes.Add(new InvalidCharacter(character.ToString(), stateInfo.Line));
return this;
}
internal abstract State Execute(char character, StateInfo info);