package lipstone.joshua.parser;
import java.util.Stack;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import lipstone.joshua.parser.exceptions.SyntaxException;
import lipstone.joshua.parser.exceptions.UnbalancedParenthesesException;
import lipstone.joshua.parser.types.BigDec;
import lipstone.joshua.parser.util.ConsCell;
import lipstone.joshua.parser.util.ConsType;
public class Tokenizer {
private static final Pattern whitespace = Pattern.compile("[\t ]"),
numbers = Pattern.compile("(([0-9]+(\\\\.[0-9]*)?|\\\\.[0-9]+)(\\\\*?i)?|infinity)"),
identifiers = Pattern.compile("(['a-zA-Z_$%#&:;<=>?~\\|][a-zA-Z_\\.$%&:;<=>?~_\\|]*|(!)+)"),
strings = Pattern.compile("\"([^\"]|\\\\\")*\""),
separators = Pattern.compile("(,|\n)"), operators = Pattern.compile("[\\Q+-*/%^\\E\\\\]");
public static ConsCell tokenizeString(String input) throws SyntaxException, UnbalancedParenthesesException {
ConsCell head = new ConsCell(), result = head;
Stack<ConsCell> parenStack = new Stack<>();
Stack<Integer> parens = new Stack<>();
Matcher m;
int current = 0;
while (current < input.length()) {
boolean found = false;
//Parentheses
if (!found && input.charAt(current) == '(') {
parens.push(current);
ConsCell descent = new ConsCell();
parenStack.push(head.append(new ConsCell(descent, ConsType.CONS_CELL)));
head = descent;
current++;
found = true;
}
if (!found && input.charAt(current) == ')') {
if (parenStack.isEmpty())
throw new UnbalancedParenthesesException(current, input, null);
head = parenStack.pop();
parens.pop();
current++;
found = true;
}
//Object
if (!found && input.charAt(0) == '{') {
int end = Parser.getEndIndex(input, current, "{", "}") + 1;
head = head.append(new ConsCell(input.substring(current, end), ConsType.OBJECT));
current += end;
found = true;
}
//Operators
m = operators.matcher(input);
if (!found && m.find(current) && m.start() == current) { //An operator is a special identifier
head = head.append(new ConsCell((input.charAt(current) == '\\') ? '/' : input.charAt(current), ConsType.OPERATOR));
current++;
found = true;
}
//Separator
m = separators.matcher(input);
if (!found && m.find(current) && m.start() == current) {
head = head.append(new ConsCell(m.group(), ConsType.SEPARATOR));
current += m.group().length();
found = true;
}
//Whitespace
m = whitespace.matcher(input);
if (!found && m.find(current) && m.start() == current) {
current += m.group().length();
found = true;
}
//Numbers
m = numbers.matcher(input);
if (!found && m.find(current) && m.start() == current) {
head = head.append(new ConsCell(new BigDec(m.group()), ConsType.NUMBER));
current += m.group().length();
found = true;
}
//Identifiers
m = identifiers.matcher(input);
if (!found && m.find(current) && m.start() == current) {
head = head.append(new ConsCell(m.group(), ConsType.IDENTIFIER));
current += m.group().length();
found = true;
}
//Strings
m = strings.matcher(input);
if (!found && m.find(current) && m.start() == current) {
head = head.append(new ConsCell(m.group().substring(1, m.group().length() - 1), ConsType.STRING));
current += m.group().length();
found = true;
}
//If there aren't any matches
if (!found)
throw new SyntaxException(input, null);
}
if (!parenStack.isEmpty())
throw new UnbalancedParenthesesException(parens.peek(), input, null);
return result;
}
}