crafting-interpreters/cclox_src/scanner.cc

#include "scanner.hh"
#include "token.hh"
#include "error_handler.hh"
#include "object.hh"
#include <string>
#include <vector>

extern ErrorHandler error_handler;

Scanner::Scanner(const std::string &code) : code{code}, tokens{}, start{0}, current{0}, line{1} {
  keywords = {
    std::pair<std::string, TokenType>("and", TokenType::AND),
    std::pair<std::string, TokenType>("class", TokenType::CLASS),
    std::pair<std::string, TokenType>("else", TokenType::ELSE),
    std::pair<std::string, TokenType>("false", TokenType::FALSE),
    std::pair<std::string, TokenType>("for", TokenType::FOR),
    std::pair<std::string, TokenType>("fun", TokenType::FUN),
    std::pair<std::string, TokenType>("if", TokenType::IF),
    std::pair<std::string, TokenType>("nil", TokenType::NIL),
    std::pair<std::string, TokenType>("or", TokenType::OR),
    std::pair<std::string, TokenType>("print", TokenType::PRINT),
    std::pair<std::string, TokenType>("return", TokenType::RETURN),
    std::pair<std::string, TokenType>("super", TokenType::SUPER),
    std::pair<std::string, TokenType>("this", TokenType::THIS),
    std::pair<std::string, TokenType>("true", TokenType::TRUE),
    std::pair<std::string, TokenType>("var", TokenType::VAR),
    std::pair<std::string, TokenType>("while", TokenType::WHILE),
  };
}

std::vector<Token> Scanner::scan_tokens() {
  while (!is_at_end()) {
    start = current;
    scan_token();
  }

  tokens.push_back(Token(TokenType::EOFILE, "", Object{}, line));
  return tokens;
}

bool Scanner::is_at_end() {
  return current >= code.size();
}

bool Scanner::is_digit(char c) {
  return c >= '0' && c <= '9';
}

bool Scanner::is_alpha(char c) {
  return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_';
}

bool Scanner::is_alphanumeric(char c) {
  return is_alpha(c) || is_digit(c);
}

void Scanner::scan_token() {
  char c = advance();
  switch (c) {
    case '(': add_token(TokenType::LEFT_PAREN); break;
    case ')': add_token(TokenType::RIGHT_PAREN); break;
    case '{': add_token(TokenType::LEFT_BRACE); break;
    case '}': add_token(TokenType::RIGHT_BRACE); break;
    case ',': add_token(TokenType::COMMA); break;
    case '.': add_token(TokenType::DOT); break;
    case '-': add_token(TokenType::MINUS); break;
    case '+': add_token(TokenType::PLUS); break;
    case ';': add_token(TokenType::SEMICOLON); break;
    case '*': add_token(TokenType::STAR); break;
    case '!':
      add_token(match('=') ? TokenType::BANG_EQUAL : TokenType::BANG);
      break;
    case '=':
      add_token(match('=') ? TokenType::EQUAL_EQUAL : TokenType::EQUAL);
      break;
    case '<':
      add_token(match('=') ? TokenType::LESS_EQUAL : TokenType::LESS);
      break;
    case '>':
      add_token(match('=') ? TokenType::GREATER_EQUAL : TokenType::GREATER);
      break;
    case '/':
      if (match('/')) {
        while (peek() != '\n' && !is_at_end()) {
          advance();
        }
      } else {
        add_token(TokenType::SLASH);
      }

      break;
    case ' ':
    case '\r':
    case '\t':
      break;
    case '\n':
      ++line;
      break;
    case '"':
      string();
      break;
    default:
      if (is_digit(c)) {
        number();
      } else if(is_alpha(c)) {
        identifier();
      } else {
        error_handler.error(line, "Unexpected character.");
      }

      break;
  }
}

char Scanner::advance() {
  return code.at(current++);
}

bool Scanner::match(char expected) {
  if (is_at_end() || peek() != expected) {
    return false;
  }

  current++;
  return true;
}

char Scanner::peek() {
  if (is_at_end()) {
    return '\0';
  }

  return code.at(current);
}

char Scanner::peek_next() {
  if (current + 1 >= code.size()) {
    return '\0';
  }

  return code.at(current + 1);
}

void Scanner::add_token(TokenType type) {
  add_token(type, Object{});
}

void Scanner::add_token(TokenType type, Object literal) {
  std::string text = code.substr(start, current - start);
  tokens.push_back(Token(type, text, literal, line));
}

void Scanner::string() {
  while (peek() != '"' && !is_at_end()) {
    if (peek() == '\n') line++;
    advance();
  }

  if (is_at_end()) {
    error_handler.error(line, "Unterminated string.");
    return;
  }

  // The closing ".
  advance();

  // Trim the surrounding quotes.
  size_t no_quote_start = start + 1;
  size_t no_quote_end = current - 1;
  std::string value = code.substr(no_quote_start, no_quote_end - no_quote_start);
  add_token(TokenType::STRING, Object(StringObjectType::LITERAL, value));
}

void Scanner::number() {
  while (is_digit(peek())) {
    advance();
  }

  if (peek() == '.' && is_digit(peek_next())) {
    // Consume the "."
    advance();

    while (is_digit(peek())) {
      advance();
    }
  }

  std::string value = code.substr(start, current - start);
  add_token(TokenType::NUMBER, Object(std::stod(value)));
}

void Scanner::identifier() {
  while (is_alphanumeric(peek())) {
    advance();
  }

  std::string value = code.substr(start, current - start);
  auto it = keywords.find(value);
  if (it != keywords.end()) {
    TokenType type = it->second;
    add_token(type);
  } else {
    add_token(TokenType::IDENTIFIER, Object(StringObjectType::IDENTIFIER, value));
  }
}