elna/source/lexer.cpp

#include "elna/source/lexer.hpp"
#include <variant>
#include <sstream>

namespace elna::source
{
    using source_position = elna::source::position;
    using source_error = elna::source::error;

    std::pair<text_iterator, text_iterator> text_iterators(const std::string &buffer)
    {
        return std::make_pair<>(text_iterator(std::cbegin(buffer)),
                text_iterator(std::cend(buffer), position{0, 0}));
    }

    text_iterator::text_iterator(std::string::const_iterator buffer,
            const source_position start_position)
        : m_buffer(buffer), m_position(start_position)
    {
    }

    const source_position& text_iterator::position() const noexcept
    {
        return this->m_position;
    }

    text_iterator::reference text_iterator::operator*() const noexcept
    {
        return *m_buffer;
    }

    text_iterator::pointer text_iterator::operator->() const noexcept
    {
        return m_buffer.base();
    }

    text_iterator& text_iterator::operator++()
    {
        if (*this->m_buffer == '\n')
        {
            this->m_position.column = 1;
            ++this->m_position.line;
        }
        else
        {
            ++this->m_position.column;
        }
        std::advance(this->m_buffer, 1);

        return *this;
    }

    text_iterator& text_iterator::operator++(int)
    {
        auto tmp = *this;
        ++(*this);
        return *this;
    }

    text_iterator text_iterator::operator+(std::size_t step)
    {
        auto result = *this;
        return ++result;
    }

    bool text_iterator::operator==(const text_iterator& that) const noexcept
    {
        return this->m_buffer == that.m_buffer;
    }

    bool text_iterator::operator!=(const text_iterator& that) const noexcept
    {
        return !(*this == that);
    }

    token::value::value()
        : nil(nullptr)
    {
    }

    token::value::value(std::int32_t value)
        : number(value)
    {
    }

    token::value::value(const std::string& value)
        : identifier(value)
    {
    }

    token::value::~value()
    {
    }

    token::token(const type of, const std::string& value, const source_position position)
        : m_type(of), m_value(value), m_position(position)
    {
    }

    token::token(const type of, std::int32_t number, const source_position position)
        : m_type(of), m_value(number), m_position(position)
    {
    }

    token::token(const type of, source_position position)
        : m_type(of), m_position(position)
    {
    }

    token::token(const token& that)
    {
        *this = that;
    }

    token::token(token&& that)
    {
        *this = std::move(that);
    }

    token::~token()
    {
        if (has_identifier())
        {
            m_value.identifier.~basic_string();
        }
    }

    token& token::operator=(const token& that)
    {
        if (has_identifier())
        {
            m_value.identifier.~basic_string();
        }
        m_type = that.of();
        m_position = that.position();
        if (that.has_identifier())
        {
            new((void *) &m_value.identifier) std::string(that.identifier());
        }
        else if (that.is_numeric())
        {
            m_value.number = that.number();
        }
        else
        {
            m_value.nil = nullptr;
        }
        return *this;
    }

    token& token::operator=(token&& that)
    {
        if (has_identifier())
        {
            m_value.identifier.~basic_string();
        }
        m_type = that.of();
        m_position = that.position();
        if (that.has_identifier())
        {
            new((void *) &m_value.identifier) std::string(std::move(that.identifier()));
        }
        else if (that.is_numeric())
        {
            m_value.number = that.number();
        }
        else
        {
            m_value.nil = nullptr;
        }
        return *this;
    }

    token::type token::of() const noexcept
    {
        return m_type;
    }

    const std::string& token::identifier() const
    {
        if (!has_identifier())
        {
            throw std::bad_variant_access();
        }
        return m_value.identifier;
    }

    std::int32_t token::number() const
    {
        if (!is_numeric())
        {
            throw std::bad_variant_access();
        }
        return m_value.number;
    }

    const source_position& token::position() const noexcept
    {
        return m_position;
    }

    bool token::has_identifier() const noexcept
    {
        return of() == type::identifier
            || of() == type::term_operator
            || of() == type::factor_operator;
    }

    bool token::is_numeric() const noexcept
    {
        return of() == type::number
            || of() == type::boolean;
    }

    unexpected_character::unexpected_character(const std::string& character, const source::position position)
        : error(position), character(character)
    {
    }

    std::string unexpected_character::what() const
    {
        std::stringstream ss{ "Unexpected character '" };

        ss << character << "'";

        return ss.str();
    }

    unexpected_token::unexpected_token(const token& token)
        : error(token.position()), m_token(token)
    {
    }

    std::string unexpected_token::what() const
    {
        return "Unexpected token";
    }

    lexer::lexer(std::vector<token>&& tokens, const position last_position)
        : tokens(std::move(tokens)), iterator(this->tokens.cbegin()), eof(token(token::type::eof, last_position))
    {
    }

    lexer& lexer::operator++()
    {
        ++iterator;
        return *this;
    }

    const token& lexer::operator*() const
    {
        return *iterator;
    }

    const token *lexer::operator->() const
    {
        return iterator.base();
    }

    const token& lexer::current() const noexcept
    {
        if (iterator == tokens.cend())
        {
            return this->eof;
        }
        return *iterator;
    }

    bool lexer::current(const token::type token_type) const noexcept
    {
        return current().of() == token_type;
    }

    void lexer::add_error(const token& expected)
    {
        m_errors.push_back(std::make_unique<unexpected_token>(expected));
    }

    std::optional<std::reference_wrapper<const token>> lexer::advance(const token::type token_type)
    {
        if (iterator != tokens.cend() && iterator->of() == token_type)
        {
            return std::make_optional<>(std::cref(*iterator++));
        }
        add_error(current());
        return std::optional<std::reference_wrapper<const token>>();
    }

    const token& lexer::look_ahead() const
    {
        auto tmp = iterator;
        ++tmp;
        if (iterator == tokens.cend() || tmp == tokens.cend())
        {
            return eof;
        }
        return *tmp;
    }

    bool lexer::look_ahead(const token::type token_type) const
    {
        return look_ahead().of() == token_type;
    }

    bool lexer::skip(const token::type token_type)
    {
        return advance(token_type).has_value();
    }

    const std::list<std::unique_ptr<error>>& lexer::errors() const noexcept
    {
        return m_errors;
    }

    result<lexer> lex(const std::string& buffer)
    {
        std::vector<token> tokens;
        auto [iterator, text_end] = text_iterators(buffer);

        while (iterator != text_end)
        {
            if (*iterator == ' ' || *iterator == '\n')
            {
            }
            else if (std::isdigit(*iterator))
            {
                tokens.emplace_back(
                        token::type::number,
                        static_cast<std::int32_t>(*iterator - '0'),
                        iterator.position()
                );
            }
            else if (*iterator == '=')
            {
                tokens.emplace_back(token::type::equals, iterator.position());
            }
            else if (*iterator == '(')
            {
                tokens.emplace_back(token::type::left_paren, iterator.position());
            }
            else if (*iterator == ')')
            {
                tokens.emplace_back(token::type::right_paren, iterator.position());
            }
            else if (*iterator == ';')
            {
                tokens.emplace_back(token::type::semicolon, iterator.position());
            }
            else if (*iterator == ',')
            {
                tokens.emplace_back(token::type::comma, iterator.position());
            }
            else if (*iterator == '.')
            {
                tokens.emplace_back(token::type::dot, iterator.position());
            }
            else if (std::isalpha(*iterator))
            {
                std::string word;
                auto i = iterator;
                while (i != text_end && std::isalpha(*i))
                {
                    word.push_back(*i);
                    ++i;
                }
                if (word == "const")
                {
                    tokens.emplace_back(token::type::let, iterator.position());
                }
                else if (word == "var")
                {
                    tokens.emplace_back(token::type::var, iterator.position());
                }
                else if (word == "begin")
                {
                    tokens.emplace_back(token::type::begin, iterator.position());
                }
                else if (word == "end")
                {
                    tokens.emplace_back(token::type::end, iterator.position());
                }
                else if (word == "if")
                {
                    tokens.emplace_back(token::type::when, iterator.position());
                }
                else if (word == "then")
                {
                    tokens.emplace_back(token::type::then, iterator.position());
                }
                else if (word == "while")
                {
                    tokens.emplace_back(token::type::_while, iterator.position());
                }
                else if (word == "do")
                {
                    tokens.emplace_back(token::type::_do, iterator.position());
                }
                else if (word == "True")
                {
                    tokens.emplace_back(token::type::boolean, 1, iterator.position());
                }
                else if (word == "False")
                {
                    tokens.emplace_back(token::type::boolean, 0, iterator.position());
                }
                else if (word == "proc")
                {
                    tokens.emplace_back(token::type::procedure, 0, iterator.position());
                }
                else
                {
                    tokens.emplace_back(token::type::identifier, word.c_str(), iterator.position());
                }
                iterator = i;
                continue;
            }
            else if (*iterator == '+' || *iterator == '-')
            {
                std::string _operator{ *iterator };

                tokens.emplace_back(token::type::term_operator, _operator.c_str(), iterator.position());
            }
            else if (*iterator == '*' || *iterator == '/')
            {
                std::string _operator{ *iterator };

                tokens.emplace_back(token::type::factor_operator, _operator.c_str(), iterator.position());
            }
            else if (*iterator == ':' && iterator + 1 != text_end && *(iterator + 1) == '=')
            {
                tokens.emplace_back(token::type::assignment, iterator.position());
                ++iterator;
            }
            else if (*iterator == ':')
            {
                tokens.emplace_back(token::type::colon, iterator.position());
            }
            else
            {
                return result<lexer>(unexpected_character{ std::string{ *iterator }, iterator.position() });
            }
            ++iterator;
        }
        return result<lexer>(std::in_place, std::move(tokens), iterator.position());
    }
}