// rules.hpp
// Copyright (c) 2005-2018 Ben Hanson (http://www.benhanson.net/)
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file licence_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
#ifndef LEXERTL_RULES_HPP
#define LEXERTL_RULES_HPP

#include "enums.hpp"
#include <locale>
#include <map>
#include "narrow.hpp"
#include "observer_ptr.hpp"
#include "parser/tokeniser/re_tokeniser.hpp"
#include "runtime_error.hpp"
#include <set>
#include <sstream>
#include <string>
#include <vector>

namespace lexertl
{
template<typename r_ch_type, typename ch_type,
    typename id_ty = uint16_t>
class basic_rules
{
public:
    using bool_vector = std::vector<bool>;
    using bool_vector_vector = std::vector<bool_vector>;
    using char_type = ch_type;
    using rules_char_type = r_ch_type;
    using id_type = id_ty;
    using id_vector = std::vector<id_type>;
    using id_vector_vector = std::vector<id_vector>;
    using re_state = detail::basic_re_tokeniser_state<rules_char_type, id_type>;
    using string = std::basic_string<rules_char_type>;
    using string_token = basic_string_token<char_type>;
    using string_vector = std::vector<string>;
    using string_set = std::set<string>;
    using string_pair = std::pair<string, string>;
    using string_id_type_map = std::map<string, id_type>;
    using string_id_type_pair = std::pair<string, id_type>;
    using token = detail::basic_re_token<rules_char_type, char_type>;
    using token_vector = std::vector<token>;
    using token_vector_vector = std::vector<token_vector>;
    using token_vector_vector_vector = std::vector<token_vector_vector>;
    using macro_map = std::map<string, token_vector>;
    using macro_pair = std::pair<string, token_vector>;
    using tokeniser =
        detail::basic_re_tokeniser<rules_char_type, char_type, id_type>;

    // If you get a compile error here you have
    // failed to define an unsigned id type.
    static_assert(std::is_unsigned<id_type>::value, "Your id type is signed");

#ifdef _WIN32
    basic_rules(const std::size_t flags_ = dot_not_cr_lf) :
#else
    basic_rules(const std::size_t flags_ = dot_not_newline) :
#endif
        _statemap(),
        _macro_map(),
        _regexes(),
        _features(),
        _ids(),
        _user_ids(),
        _next_dfas(),
        _pushes(),
        _pops(),
        _flags(flags_),
        _locale(),
        _lexer_state_names()
    {
        push_state(initial());
    }

    void clear()
    {
        _statemap.clear();
        _macro_map.clear();
        _regexes.clear();
        _features.clear();
        _ids.clear();
        _user_ids.clear();
        _next_dfas.clear();
        _pushes.clear();
        _pops.clear();
#ifdef _WIN32
        _flags = dot_not_cr_lf;
#else
        _flags = dot_not_newline;
#endif
        _locale = std::locale();
        _lexer_state_names.clear();
        push_state(initial());
    }

    void clear(const id_type dfa_)
    {
        if (_regexes.size() > dfa_)
        {
            _regexes[dfa_].clear();
            _features[dfa_] = 0;
            _ids[dfa_].clear();
            _user_ids[dfa_].clear();
            _next_dfas[dfa_].clear();
            _pushes[dfa_].clear();
            _pops[dfa_].clear();
        }
    }

    void flags(const std::size_t flags_)
    {
        _flags = flags_;
    }

    std::size_t flags() const
    {
        return _flags;
    }

    static id_type skip()
    {
        return static_cast<id_type>(~1);
    }

    id_type eoi() const
    {
        return 0;
    }

    static id_type npos()
    {
        return static_cast<id_type>(~0);
    }

    std::locale imbue(const std::locale &locale_)
    {
        std::locale loc_ = _locale;

        _locale = locale_;
        return loc_;
    }

    const std::locale &locale() const
    {
        return _locale;
    }

    const rules_char_type *state(const id_type index_) const
    {
        if (index_ == 0)
        {
            return initial();
        }
        else
        {
            const id_type i_ = index_ - 1;

            if (_lexer_state_names.size() > i_)
            {
                return _lexer_state_names[i_].c_str();
            }
            else
            {
                return 0;
            }
        }
    }

    id_type state(const rules_char_type *name_) const
    {
        typename string_id_type_map::const_iterator iter_ =
            _statemap.find(name_);

        if (iter_ == _statemap.end())
        {
            return npos();
        }
        else
        {
            return iter_->second;
        }
    }

    id_type push_state(const rules_char_type *name_)
    {
        validate(name_);

        if (_statemap.insert(string_id_type_pair(name_,
            static_cast<id_type>(_statemap.size()))).second)
        {
            _regexes.push_back(token_vector_vector());
            _features.push_back(0);
            _ids.push_back(id_vector());
            _user_ids.push_back(id_vector());
            _next_dfas.push_back(id_vector());
            _pushes.push_back(id_vector());
            _pops.push_back(bool_vector());

            if (string(name_) != initial())
            {
                _lexer_state_names.push_back(name_);
            }
        }
        else
        {
            return _statemap.find(name_)->second;
        }

        if (_next_dfas.size() > npos())
        {
            // Overflow
            throw runtime_error("The data type you have chosen cannot hold "
                "this many lexer start states.");
        }

        // Initial is not stored, so no need to - 1.
        return static_cast<id_type>(_lexer_state_names.size());
    }

    void insert_macro(const rules_char_type *name_,
        const rules_char_type *regex_)
    {
        insert_macro(name_, string(regex_));
    }

    void insert_macro(const rules_char_type *name_,
        const rules_char_type *regex_start_,
        const rules_char_type *regex_end_)
    {
        insert_macro(name_, string(regex_start_, regex_end_));
    }

    void insert_macro(const rules_char_type *name_, const string &regex_)
    {
        validate(name_);

        typename macro_map::const_iterator iter_ = _macro_map.find(name_);

        if (iter_ == _macro_map.end())
        {
            auto pair_ = _macro_map.insert(macro_pair(name_, token_vector()));

            tokenise(regex_, pair_.first->second, npos(), name_);
        }
        else
        {
            std::ostringstream ss_;

            ss_ << "Attempt to redefine MACRO '";
            narrow(name_, ss_);
            ss_ << "'.";
            throw runtime_error(ss_.str());
        }
    }

    // Add rule to INITIAL
    void push(const rules_char_type *regex_, const id_type id_,
        const id_type user_id_ = npos())
    {
        push(string(regex_), id_, user_id_);
    }

    void push(const rules_char_type *regex_start_,
        const rules_char_type *regex_end_,
        const id_type id_, const id_type user_id_ = npos())
    {
        push(string(regex_start_, regex_end_), id_, user_id_);
    }

    void push(const string &regex_, const id_type id_,
        const id_type user_id_ = npos())
    {
        check_for_invalid_id(id_);
        _regexes.front().push_back(token_vector());
        tokenise(regex_, _regexes.front().back(), id_, 0);

        if (regex_[0] == '^')
        {
            _features.front() |= bol_bit;
        }

        if (regex_.size() > 0 && regex_[regex_.size() - 1] == '$')
        {
            _features.front() |= eol_bit;
        }

        if (id_ == skip())
        {
            _features.front() |= skip_bit;
        }
        else if (id_ == eoi())
        {
            _features.front() |= again_bit;
        }

        _ids.front().push_back(id_);
        _user_ids.front().push_back(user_id_);
        _next_dfas.front().push_back(0);
        _pushes.front().push_back(npos());
        _pops.front().push_back(false);
    }

    // Add rule with no id
    void push(const rules_char_type *curr_dfa_,
        const rules_char_type *regex_, const rules_char_type *new_dfa_)
    {
        push(curr_dfa_, string(regex_), new_dfa_);
    }

    void push(const rules_char_type *curr_dfa_,
        const rules_char_type *regex_start_, const rules_char_type *regex_end_,
        const rules_char_type *new_dfa_)
    {
        push(curr_dfa_, string(regex_start_, regex_end_), new_dfa_);
    }

    void push(const rules_char_type *curr_dfa_, const string &regex_,
        const rules_char_type *new_dfa_)
    {
        push(curr_dfa_, regex_, eoi(), new_dfa_, false);
    }

    // Add rule with id
    void push(const rules_char_type *curr_dfa_,
        const rules_char_type *regex_, const id_type id_,
        const rules_char_type *new_dfa_, const id_type user_id_ = npos())
    {
        push(curr_dfa_, string(regex_), id_, new_dfa_, user_id_);
    }

    void push(const rules_char_type *curr_dfa_,
        const rules_char_type *regex_start_,
        const rules_char_type *regex_end_, const id_type id_,
        const rules_char_type *new_dfa_, const id_type user_id_ = npos())
    {
        push(curr_dfa_, string(regex_start_, regex_end_),
            id_, new_dfa_, user_id_);
    }

    void push(const rules_char_type *curr_dfa_, const string &regex_,
        const id_type id_, const rules_char_type *new_dfa_,
        const id_type user_id_ = npos())
    {
        push(curr_dfa_, regex_, id_, new_dfa_, true, user_id_);
    }

    void reverse()
    {
        for (auto &state_ : _regexes)
        {
            for (auto &regex_ : state_)
            {
                reverse(regex_);
            }
        }

        for (auto &pair_ : _macro_map)
        {
            reverse(pair_.second);
        }
    }

    const string_id_type_map &statemap() const
    {
        return _statemap;
    }

    const token_vector_vector_vector &regexes() const
    {
        return _regexes;
    }

    const id_vector &features() const
    {
        return _features;
    }

    const id_vector_vector &ids() const
    {
        return _ids;
    }

    const id_vector_vector &user_ids() const
    {
        return _user_ids;
    }

    const id_vector_vector &next_dfas() const
    {
        return _next_dfas;
    }

    const id_vector_vector &pushes() const
    {
        return _pushes;
    }

    const bool_vector_vector &pops() const
    {
        return _pops;
    }

    bool empty() const
    {
        bool empty_ = true;

        for (const auto &regex_ : _regexes)
        {
            if (!regex_.empty())
            {
                empty_ = false;
                break;
            }
        }

        return empty_;
    }

    static const rules_char_type *initial()
    {
        static const rules_char_type initial_ [] =
        { 'I', 'N', 'I', 'T', 'I', 'A', 'L', 0 };

        return initial_;
    }

    static const rules_char_type *dot()
    {
        static const rules_char_type dot_ [] = { '.', 0 };

        return dot_;
    }

    static const rules_char_type *all_states()
    {
        static const rules_char_type star_ [] = { '*', 0 };

        return star_;
    }

private:
    string_id_type_map _statemap;
    macro_map _macro_map;
    token_vector_vector_vector _regexes;
    id_vector _features;
    id_vector_vector _ids;
    id_vector_vector _user_ids;
    id_vector_vector _next_dfas;
    id_vector_vector _pushes;
    bool_vector_vector _pops;
    std::size_t _flags;
    std::locale _locale;
    string_vector _lexer_state_names;

    void tokenise(const string &regex_, token_vector &tokens_,
        const id_type id_, const rules_char_type *name_)
    {
        re_state state_(regex_.c_str(), regex_.c_str() + regex_.size(), id_,
            _flags, _locale, name_);
        string macro_;
        rules_char_type diff_ = 0;

        tokens_.push_back(token());

        do
        {
            observer_ptr<token> lhs_ = &tokens_.back();
            token rhs_;

            tokeniser::next(*lhs_, state_, rhs_);

            if (rhs_._type != detail::DIFF &&
                lhs_->precedence(rhs_._type) == ' ')
            {
                std::ostringstream ss_;

                ss_ << "A syntax error occurred: '" <<
                    lhs_->precedence_string() <<
                    "' against '" << rhs_.precedence_string() <<
                    "' preceding index " << state_.index() <<
                    " in ";

                if (name_ != 0)
                {
                    ss_ << "macro ";
                    narrow(name_, ss_);
                }
                else
                {
                    ss_ << "rule id " << state_._id;
                }

                ss_ << '.';
                throw runtime_error(ss_.str());
            }

            if (rhs_._type == detail::MACRO)
            {
                typename macro_map::const_iterator iter_ =
                    _macro_map.find(rhs_._extra);

                macro_ = rhs_._extra;

                if (iter_ == _macro_map.end())
                {
                    const rules_char_type *rhs_name_ = rhs_._extra.c_str();
                    std::ostringstream ss_;

                    ss_ << "Unknown MACRO name '";
                    narrow(rhs_name_, ss_);
                    ss_ << "'.";
                    throw runtime_error(ss_.str());
                }
                else
                {
                    const bool multiple_ = iter_->second.size() > 3;

                    if (diff_)
                    {
                        if (multiple_)
                        {
                            std::ostringstream ss_;

                            ss_ << "Single CHARSET must follow {-} or {+} at "
                                "index " << state_.index() - 1 << " in ";

                            if (name_ != 0)
                            {
                                ss_ << "macro ";
                                narrow(name_, ss_);
                            }
                            else
                            {
                                ss_ << "rule id " << state_._id;
                            }

                            ss_ << '.';
                            throw runtime_error(ss_.str());
                        }
                        else
                        {
                            rhs_ = iter_->second[1];
                        }
                    }

                    // Any macro with more than one charset (or quantifiers)
                    // requires bracketing.
                    if (multiple_)
                    {
                        token open_;

                        open_._type = detail::OPENPAREN;
                        open_._str.insert('(');
                        tokens_.push_back(open_);
                    }

                    // Don't need to store token if it is diff.
                    if (!diff_)
                    {
                        // Don't insert BEGIN or END tokens
                        tokens_.insert(tokens_.end(), iter_->second.begin() + 1,
                            iter_->second.end() - 1);
                        lhs_ = &tokens_.back();
                    }

                    if (multiple_)
                    {
                        token close_;

                        close_._type = detail::CLOSEPAREN;
                        close_._str.insert(')');
                        tokens_.push_back(close_);
                    }
                }
            }
            else if (rhs_._type == detail::DIFF)
            {
                if (!macro_.empty())
                {
                    typename macro_map::const_iterator iter_ =
                        _macro_map.find(macro_);

                    if (iter_->second.size() > 3)
                    {
                        std::ostringstream ss_;

                        ss_ << "Single CHARSET must precede {-} or {+} at "
                            "index " << state_.index() - 1 << " in ";

                        if (name_ != 0)
                        {
                            ss_ << "macro ";
                            narrow(name_, ss_);
                        }
                        else
                        {
                            ss_ << "rule id " << state_._id;
                        }

                        ss_ << '.';
                        throw runtime_error(ss_.str());
                    }
                }

                diff_ = rhs_._extra[0];
                macro_.clear();
                continue;
            }
            else if (!diff_)
            {
                tokens_.push_back(rhs_);
                lhs_ = &tokens_.back();
                macro_.clear();
            }

            // diff_ may have been set by previous conditional.
            if (diff_)
            {
                if (rhs_._type != detail::CHARSET)
                {
                    std::ostringstream ss_;

                    ss_ << "CHARSET must follow {-} or {+} at index " <<
                        state_.index() - 1 << " in ";

                    if (name_ != 0)
                    {
                        ss_ << "macro ";
                        narrow(name_, ss_);
                    }
                    else
                    {
                        ss_ << "rule id " << state_._id;
                    }

                    ss_ << '.';
                    throw runtime_error(ss_.str());
                }

                switch (diff_)
                {
                case '-':
                    lhs_->_str.remove(rhs_._str);

                    if (lhs_->_str.empty())
                    {
                        std::ostringstream ss_;

                        ss_ << "Empty charset created by {-} at index " <<
                            state_.index() - 1 << " in ";

                        if (name_ != 0)
                        {
                            ss_ << "macro ";
                            narrow(name_, ss_);
                        }
                        else
                        {
                            ss_ << "rule id " << state_._id;
                        }

                        ss_ << '.';
                        throw runtime_error(ss_.str());
                    }

                    break;
                case '+':
                    lhs_->_str.insert(rhs_._str);
                    break;
                }

                diff_ = 0;
            }
        } while (tokens_.back()._type != detail::END);

        if (tokens_.size() == 2)
        {
            std::ostringstream ss_;

            ss_ << "Empty regex in ";

            if (name_ != 0)
            {
                ss_ << "macro ";
                narrow(name_, ss_);
            }
            else
            {
                ss_ << "rule id " << state_._id;
            }

            ss_ << " is not allowed.";
            throw runtime_error(ss_.str());
        }
    }

    void reverse(token_vector &vector_)
    {
        token_vector new_vector_(vector_.size(), token());
        auto iter_ = vector_.rbegin();
        auto end_ = vector_.rend();
        auto dest_ = new_vector_.begin();
        std::stack<typename token_vector::reverse_iterator> stack_;

        for (; iter_ != end_; ++iter_, ++dest_)
        {
            switch (iter_->_type)
            {
            case detail::BEGIN:
                iter_->swap(*dest_);
                dest_->_type = detail::END;
                break;
            case detail::BOL:
                iter_->swap(*dest_);
                dest_->_type = detail::EOL;
                break;
            case detail::EOL:
                iter_->swap(*dest_);
                dest_->_type = detail::BOL;
                break;
            case detail::OPENPAREN:
                iter_->swap(*dest_);
                dest_->_type = detail::CLOSEPAREN;

                if (stack_.top() != end_)
                {
                    ++dest_;
                    dest_->swap(*stack_.top());
                }

                stack_.pop();
                break;
            case detail::CLOSEPAREN:
                iter_->swap(*dest_);
                dest_->_type = detail::OPENPAREN;
                stack_.push(end_);
                break;
            case detail::OPT:
            case detail::AOPT:
            case detail::ZEROORMORE:
            case detail::AZEROORMORE:
            case detail::ONEORMORE:
            case detail::AONEORMORE:
            case detail::REPEATN:
            case detail::AREPEATN:
            {
                auto temp_ = iter_ + 1;

                if (temp_->_type == detail::CLOSEPAREN)
                {
                    stack_.push(iter_);
                    ++iter_;
                    iter_->swap(*dest_);
                    dest_->_type = detail::OPENPAREN;
                }
                else
                {
                    dest_->swap(*temp_);
                    ++dest_;
                    dest_->swap(*iter_);
                    ++iter_;
                }

                break;
            }
            case detail::END:
                iter_->swap(*dest_);
                dest_->_type = detail::BEGIN;
                break;
            default:
                // detail::OR
                // detail::CHARSET
                iter_->swap(*dest_);
                break;
            }
        }

        new_vector_.swap(vector_);
    }

    void push(const rules_char_type *curr_dfa_, const string &regex_,
        const id_type id_, const rules_char_type *new_dfa_,
        const bool check_, const id_type user_id_ = npos())
    {
        const bool star_ = *curr_dfa_ == '*' && *(curr_dfa_ + 1) == 0;
        const bool dot_ = *new_dfa_ == '.' && *(new_dfa_ + 1) == 0;
        const bool push_ = *new_dfa_ == '>';
        const rules_char_type *push_dfa_ = nullptr;
        const bool pop_ = *new_dfa_ == '<';

        if (push_ || pop_)
        {
            ++new_dfa_;
        }

        if (check_)
        {
            check_for_invalid_id(id_);
        }

        if (!dot_ && !pop_)
        {
            const rules_char_type *temp_ = new_dfa_;

            while (*temp_ && *temp_ != ':')
            {
                ++temp_;
            }

            if (*temp_) push_dfa_ = temp_ + 1;

            validate(new_dfa_, *temp_ ? temp_ : 0);

            if (push_dfa_)
            {
                validate(push_dfa_);
            }
        }

        // npos means pop here
        id_type new_dfa_id_ = npos();
        id_type push_dfa_id_ = npos();
        typename string_id_type_map::const_iterator iter_;
        auto end_ = _statemap.cend();
        id_vector next_dfas_;

        if (!dot_ && !pop_)
        {
            if (push_dfa_)
            {
                iter_ = _statemap.find(string(new_dfa_, push_dfa_ - 1));
            }
            else
            {
                iter_ = _statemap.find(new_dfa_);
            }

            if (iter_ == end_)
            {
                std::ostringstream ss_;

                ss_ << "Unknown state name '";
                narrow(new_dfa_, ss_);
                ss_ << "'.";
                throw runtime_error(ss_.str());
            }

            new_dfa_id_ = iter_->second;

            if (push_dfa_)
            {
                iter_ = _statemap.find(push_dfa_);

                if (iter_ == end_)
                {
                    std::ostringstream ss_;

                    ss_ << "Unknown state name '";
                    narrow(push_dfa_, ss_);
                    ss_ << "'.";
                    throw runtime_error(ss_.str());
                }

                push_dfa_id_ = iter_->second;
            }
        }

        if (star_)
        {
            const std::size_t size_ = _statemap.size();

            for (id_type i_ = 0; i_ < size_; ++i_)
            {
                next_dfas_.push_back(i_);
            }
        }
        else
        {
            const rules_char_type *start_ = curr_dfa_;
            string next_dfa_;

            while (*curr_dfa_)
            {
                while (*curr_dfa_ && *curr_dfa_ != ',')
                {
                    ++curr_dfa_;
                }

                next_dfa_.assign(start_, curr_dfa_);

                if (*curr_dfa_)
                {
                    ++curr_dfa_;
                    start_ = curr_dfa_;
                }

                validate(next_dfa_.c_str());
                iter_ = _statemap.find(next_dfa_.c_str());

                if (iter_ == end_)
                {
                    std::ostringstream ss_;

                    ss_ << "Unknown state name '";
                    curr_dfa_ = next_dfa_.c_str();
                    narrow(curr_dfa_, ss_);
                    ss_ << "'.";
                    throw runtime_error(ss_.str());
                }

                next_dfas_.push_back(iter_->second);
            }
        }

        for (std::size_t i_ = 0, size_ = next_dfas_.size();
            i_ < size_; ++i_)
        {
            const id_type curr_ = next_dfas_[i_];

            _regexes[curr_].push_back(token_vector());
            tokenise(regex_, _regexes[curr_].back(), id_, 0);

            if (regex_[0] == '^')
            {
                _features[curr_] |= bol_bit;
            }

            if (regex_[regex_.size() - 1] == '$')
            {
                _features[curr_] |= eol_bit;
            }

            if (id_ == skip())
            {
                _features[curr_] |= skip_bit;
            }
            else if (id_ == eoi())
            {
                _features[curr_] |= again_bit;
            }

            if (push_ || pop_)
            {
                _features[curr_] |= recursive_bit;
            }

            _ids[curr_].push_back(id_);
            _user_ids[curr_].push_back(user_id_);
            _next_dfas[curr_].push_back(dot_ ? curr_ : new_dfa_id_);
            _pushes[curr_].push_back(push_ ? (push_dfa_ ?
                push_dfa_id_ : curr_) : npos());
            _pops[curr_].push_back(pop_);
        }
    }

    void validate(const rules_char_type *name_,
        const rules_char_type *end_ = nullptr) const
    {
        const rules_char_type *start_ = name_;

        if (*name_ != '_' && !(*name_ >= 'A' && *name_ <= 'Z') &&
            !(*name_ >= 'a' && *name_ <= 'z'))
        {
            std::ostringstream ss_;

            ss_ << "Invalid name '";
            narrow(name_, ss_);
            ss_ << "'.";
            throw runtime_error(ss_.str());
        }
        else if (*name_)
        {
            ++name_;
        }

        while (*name_ && name_ != end_)
        {
            if (*name_ != '_' && *name_ != '-' &&
                !(*name_ >= 'A' && *name_ <= 'Z') &&
                !(*name_ >= 'a' && *name_ <= 'z') &&
                !(*name_ >= '0' && *name_ <= '9'))
            {
                std::ostringstream ss_;

                ss_ << "Invalid name '";
                name_ = start_;
                narrow(name_, ss_);
                ss_ << "'.";
                throw runtime_error(ss_.str());
            }

            ++name_;
        }
    }

    void check_for_invalid_id(const id_type id_) const
    {
        if (id_ == eoi())
        {
            throw runtime_error("Cannot resuse the id for eoi.");
        }

        if (id_ == npos())
        {
            throw runtime_error("The id npos is reserved for the "
                "UNKNOWN token.");
        }
    }
};

using rules = basic_rules<char, char>;
using wrules = basic_rules<wchar_t, wchar_t>;
using u32rules = basic_rules<char32_t, char32_t>;
}

#endif