joko/lib/Parser.cpp

#include "Parser.hpp"

namespace jk
{
  /*explicit*/ Parser::Parser(Logger& logger, std::shared_ptr<Lexer> lexer)
    : m_logger { logger }
    , m_lexer { lexer }
  {
  }

  /*virtual*/ Parser::~Parser()
  {
  }

  std::shared_ptr<Node> Parser::parse(std::string const& source)
  {
    std::shared_ptr<Node> tok;
    m_cursor = 0;
    m_lexer->scan(source);

    while ( (tok=m_lexer->next()) )
      {
        m_tokens.push_back(tok);
      }

    if (m_tokens.empty())
      {
        return std::make_shared<Node>(NODE_PROG, "",
                                      Loc {m_lexer->loc().path(), 1, 1});
      }

    return parse_prog();
  }

  std::shared_ptr<Node> Parser::current() const
  {
    assert(m_cursor < m_tokens.size());
    return m_tokens[m_cursor];
  }

  Loc Parser::loc() const
  {
    return current()->loc();
  }

  std::shared_ptr<Node> Parser::consume()
  {
    assert(m_cursor < m_tokens.size());

    auto tok = m_tokens[m_cursor];
    m_cursor++;
    return tok;
  }

  std::shared_ptr<Node> Parser::consume(NodeType type)
  {
    assert(m_cursor < m_tokens.size());

    if (m_tokens[m_cursor]->type() != type)
      {
        std::stringstream ss;
        ss << "expected '"
           << std::string(NodeTypeStr[type])
          .substr(std::string("NODE_").size())
           << "', got '"
           << std::string(NodeTypeStr[m_tokens[m_cursor]->type()])
          .substr(std::string("NODE_").size())
           << "'";

        m_logger.log<syntax_error>(LOG_ERROR,
                                   m_tokens[m_cursor]->loc(),
                                   ss.str());
      }

    return consume();
  }

  bool Parser::type_is(NodeType type, int lookahead) const
  {
    if (m_cursor + lookahead >= m_tokens.size())
      {
        return false;
      }

    return m_tokens[m_cursor + lookahead]->type() == type;
  }

  std::shared_ptr<Node> Parser::parse_prog()
  {
    auto root = std::make_shared<Node>(NODE_PROG, "", loc());

    while (m_cursor < m_tokens.size())
      {
        root->add_child(parse_expr());
      }

    return root;
  }

  std::shared_ptr<Node> Parser::parse_expr()
  {
    if (type_is(NODE_OPAR)
        && type_is(NODE_DECL, 1)
        && type_is(NODE_OPAR, 2))
      {
        return parse_fundecl();
      }

    if (type_is(NODE_OPAR)
        && type_is(NODE_IDENT, 1))
      {
        return parse_funcall();
      }

    if (type_is(NODE_OPAR)
        && type_is(NODE_DECL, 1))
      {
        return parse_vardecl();
      }

    if (type_is(NODE_OPAR)
        && type_is(NODE_RARROW, 1))
      {
        return parse_lambda();
      }

    return parse_literal();
  }

  std::shared_ptr<Node> Parser::parse_vardecl()
  {
    auto root = std::make_shared<Node>(NODE_VARDECL, "", loc());
    consume(NODE_OPAR);
    consume(NODE_DECL);
    root->add_child(consume(NODE_IDENT));
    root->add_child(parse_expr());
    consume(NODE_CPAR);

    return root;
  }

  std::shared_ptr<Node> Parser::parse_fundecl()
  {
    consume(NODE_OPAR);
    consume(NODE_DECL);

    consume(NODE_OPAR);
    auto ident = consume(NODE_IDENT);
    auto params = parse_params();

    consume(NODE_CPAR);

    auto body = parse_body();
    consume(NODE_CPAR);

    auto vardecl = std::make_shared<Node>(NODE_VARDECL, "", ident->loc());
    vardecl->add_child(ident);

    auto lambda = std::make_shared<Node>(NODE_LAMBDA, "", ident->loc());
    lambda->add_child(params);
    lambda->add_child(body);
    vardecl->add_child(lambda);

    return vardecl;
  }

  std::shared_ptr<Node> Parser::parse_lambda()
  {
    auto root = std::make_shared<Node>(NODE_LAMBDA, "", loc());
    consume(NODE_OPAR);
    consume(NODE_RARROW);
    consume(NODE_OPAR);
    root->add_child(parse_params());
    consume(NODE_CPAR);
    root->add_child(parse_body());
    consume(NODE_CPAR);

    return root;
  }

  std::shared_ptr<Node> Parser::parse_params()
  {
    auto root = std::make_shared<Node>(NODE_PARAMS, "", loc());

    while (!type_is(NODE_CPAR))
      {
        root->add_child(consume(NODE_IDENT));
      }

    return root;
  }

  std::shared_ptr<Node> Parser::parse_body()
  {
    auto root = std::make_shared<Node>(NODE_BODY, "", loc());

    while (!type_is(NODE_CPAR))
      {
        root->add_child(parse_expr());
      }

    return root;
  }

  std::shared_ptr<Node> Parser::parse_funcall()
  {
    auto root = std::make_shared<Node>(NODE_FUNCALL, "", loc());
    consume(NODE_OPAR);

    while (!type_is(NODE_CPAR))
      {
        root->add_child(parse_expr());
      }

    consume(NODE_CPAR);

    return root;
  }

  std::shared_ptr<Node> Parser::parse_literal()
  {
    if (type_is(NODE_INT)
        || type_is(NODE_BOOL)
        || type_is(NODE_IDENT))
      {
        return consume();
      }

    m_logger.log<syntax_error>(LOG_ERROR,
                               loc(),
                               std::string()
                               + "unexpected token '"
                               + NodeTypeStr[current()->type()]
                               + "' ('"
                               + current()->repr()
                               + "')");

    abort();
  }
}