structures.hxx

/* Copyright 2018-2024 Dimitrij Mijoski
 *
 * This file is part of Nuspell.
 *
 * Nuspell is free software: you can redistribute it and/or modify
 * it under the terms of the GNU Lesser General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * Nuspell is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public License
 * along with Nuspell.  If not, see <http://www.gnu.org/licenses/>.
 */
 
#ifndef NUSPELL_STRUCTURES_HXX
#define NUSPELL_STRUCTURES_HXX
 
#include "unicode.hxx"
 
#include <algorithm>
#include <cmath>
#include <forward_list>
#include <functional>
#include <iterator>
#include <stack>
#include <stdexcept>
#include <string>
#include <string_view>
#include <utility>
#include <vector>
 
namespace nuspell {
NUSPELL_BEGIN_INLINE_NAMESPACE
 
template <class It>
class Subrange {
    using Iter_Category =
        typename std::iterator_traits<It>::iterator_category;
    It a = {};
    It b = {};
 
      public:
    Subrange() = default;
    Subrange(std::pair<It, It> p) : a(p.first), b(p.second) {}
    Subrange(It first, It last) : a(first), b(last) {}
    template <class Range>
    Subrange(Range& r) : a(r.begin()), b(r.end())
    {
    }
    auto begin() const -> It { return a; }
    auto end() const -> It { return b; }
};
// CTAD
template <class Range>
Subrange(const Range& r) -> Subrange<typename Range::const_iterator>;
template <class Range>
Subrange(Range& r) -> Subrange<typename Range::iterator>;
 
template <class CharT>
class String_Set {
      private:
    std::basic_string<CharT> d;
    auto sort_uniq() -> void
    {
        auto first = begin();
        auto last = end();
        using t = traits_type;
        sort(first, last, t::lt);
        d.erase(unique(first, last, t::eq), last);
    }
    struct Char_Traits_Less_Than {
        auto operator()(CharT a, CharT b) const noexcept
        {
            return traits_type::lt(a, b);
        }
    };
 
      public:
    using Str = std::basic_string<CharT>;
    using traits_type = typename Str::traits_type;
 
    using key_type = typename Str::value_type;
    using key_compare = Char_Traits_Less_Than;
    using value_type = typename Str::value_type;
    using value_compare = key_compare;
    using allocator_type = typename Str::allocator_type;
    using pointer = typename Str::pointer;
    using const_pointer = typename Str::const_pointer;
    using reference = typename Str::reference;
    using const_reference = typename Str::const_reference;
    using size_type = typename Str::size_type;
    using difference_type = typename Str::difference_type;
    using iterator = typename Str::iterator;
    using const_iterator = typename Str::const_iterator;
    using reverse_iterator = typename Str::reverse_iterator;
    using const_reverse_iterator = typename Str::const_reverse_iterator;
 
    String_Set() = default;
    explicit String_Set(const Str& s) : d(s) { sort_uniq(); }
    explicit String_Set(Str&& s) : d(move(s)) { sort_uniq(); }
    explicit String_Set(const CharT* s) : d(s) { sort_uniq(); }
    template <class InputIterator>
    String_Set(InputIterator first, InputIterator last) : d(first, last)
    {
        sort_uniq();
    }
    String_Set(std::initializer_list<value_type> il) : d(il)
    {
        sort_uniq();
    }
 
    auto& operator=(const Str& s)
    {
        d = s;
        sort_uniq();
        return *this;
    }
    auto& operator=(Str&& s)
    {
        d = move(s);
        sort_uniq();
        return *this;
    }
    auto& operator=(std::initializer_list<value_type> il)
    {
        d = il;
        sort_uniq();
        return *this;
    }
    auto& operator=(const CharT* s)
    {
        d = s;
        sort_uniq();
        return *this;
    }
 
    // non standard underlying storage access:
    auto data() const noexcept -> const_pointer { return d.data(); }
    operator const Str&() const noexcept { return d; }
    auto& str() const noexcept { return d; }
 
    // iterators:
    iterator begin() noexcept { return d.begin(); }
    const_iterator begin() const noexcept { return d.begin(); }
    iterator end() noexcept { return d.end(); }
    const_iterator end() const noexcept { return d.end(); }
 
    reverse_iterator rbegin() noexcept { return d.rbegin(); }
    const_reverse_iterator rbegin() const noexcept { return d.rbegin(); }
    reverse_iterator rend() noexcept { return d.rend(); }
    const_reverse_iterator rend() const noexcept { return d.rend(); }
 
    const_iterator cbegin() const noexcept { return d.cbegin(); }
    const_iterator cend() const noexcept { return d.cend(); }
    const_reverse_iterator crbegin() const noexcept { return d.crbegin(); }
    const_reverse_iterator crend() const noexcept { return d.crend(); }
 
    // capacity:
    bool empty() const noexcept { return d.empty(); }
    size_type size() const noexcept { return d.size(); }
    size_type max_size() const noexcept { return d.max_size(); }
 
    // modifiers:
    std::pair<iterator, bool> insert(value_type x)
    {
        auto it = lower_bound(x);
        if (it != end() && *it == x) {
            return {it, false};
        }
        auto ret = d.insert(it, x);
        return {ret, true};
    }
    iterator insert(iterator hint, value_type x)
    {
        if (hint == end() || traits_type::lt(x, *hint)) {
            if (hint == begin() ||
                traits_type::lt(*(hint - 1), x)) {
                return d.insert(hint, x);
            }
        }
        return insert(x).first;
    }
 
    // iterator insert(const_iterator hint, value_type&& x);
    template <class InputIterator>
    void insert(InputIterator first, InputIterator last)
    {
        d.insert(end(), first, last);
        sort_uniq();
    }
    void insert(std::initializer_list<value_type> il)
    {
        d.insert(end(), il);
        sort_uniq();
    }
    template <class... Args>
    std::pair<iterator, bool> emplace(Args&&... args)
    {
        return insert(CharT(args...));
    }
 
    template <class... Args>
    iterator emplace_hint(iterator hint, Args&&... args)
    {
        return insert(hint, CharT(args...));
    }
 
    iterator erase(iterator position) { return d.erase(position); }
    size_type erase(key_type x)
    {
        auto i = d.find(x);
        if (i != d.npos) {
            d.erase(i, 1);
            return true;
        }
        return false;
    }
    iterator erase(iterator first, iterator last)
    {
        return d.erase(first, last);
    }
    void swap(String_Set& s) { d.swap(s.d); }
    void clear() noexcept { d.clear(); }
 
    // non standrd modifiers:
    auto insert(const Str& s) -> void
    {
        d += s;
        sort_uniq();
    }
    auto& operator+=(const Str& s)
    {
        insert(s);
        return *this;
    }
 
    // observers:
    key_compare key_comp() const { return Char_Traits_Less_Than(); }
    value_compare value_comp() const { return key_comp(); }
 
    // set operations:
      private:
    auto lookup(key_type x) const
    {
        auto i = d.find(x);
        if (i == d.npos)
            i = d.size();
        return i;
    }
 
      public:
    iterator find(key_type x) { return begin() + lookup(x); }
    const_iterator find(key_type x) const { return begin() + lookup(x); }
    size_type count(key_type x) const { return d.find(x) != d.npos; }
 
    iterator lower_bound(key_type x)
    {
        return std::lower_bound(begin(), end(), x, traits_type::lt);
    }
 
    const_iterator lower_bound(key_type x) const
    {
        return std::lower_bound(begin(), end(), x, traits_type::lt);
    }
    iterator upper_bound(key_type x)
    {
        return std::upper_bound(begin(), end(), x, traits_type::lt);
    }
 
    const_iterator upper_bound(key_type x) const
    {
        return std::upper_bound(begin(), end(), x, traits_type::lt);
    }
    std::pair<iterator, iterator> equal_range(key_type x)
    {
        return std::equal_range(begin(), end(), x, traits_type::lt);
    }
 
    std::pair<const_iterator, const_iterator> equal_range(key_type x) const
    {
        return std::equal_range(begin(), end(), x, traits_type::lt);
    }
 
    // non standard set operations:
    bool contains(key_type x) const { return count(x); }
 
    // compare
    bool operator<(const String_Set& rhs) const { return d < rhs.d; }
    bool operator<=(const String_Set& rhs) const { return d <= rhs.d; }
    bool operator==(const String_Set& rhs) const { return d == rhs.d; }
    bool operator!=(const String_Set& rhs) const { return d != rhs.d; }
    bool operator>=(const String_Set& rhs) const { return d >= rhs.d; }
    bool operator>(const String_Set& rhs) const { return d > rhs.d; }
};
 
template <class CharT>
auto swap(String_Set<CharT>& a, String_Set<CharT>& b)
{
    a.swap(b);
}
 
using Flag_Set = String_Set<char16_t>;
 
class Substr_Replacer {
      public:
    using Str = std::string;
    using Str_View = std::string_view;
    using Pair_Str = std::pair<Str, Str>;
    using Table_Pairs = std::vector<Pair_Str>;
 
      private:
    Table_Pairs table;
    auto sort_uniq() -> void;
    auto find_match(Str_View s) const;
 
      public:
    Substr_Replacer() = default;
    explicit Substr_Replacer(const Table_Pairs& v) : table(v)
    {
        sort_uniq();
    }
    explicit Substr_Replacer(Table_Pairs&& v) : table(std::move(v))
    {
        sort_uniq();
    }
 
    auto& operator=(const Table_Pairs& v)
    {
        table = v;
        sort_uniq();
        return *this;
    }
    auto& operator=(Table_Pairs&& v)
    {
        table = std::move(v);
        sort_uniq();
        return *this;
    }
 
    auto replace(Str& s) const -> Str&;
    auto replace_copy(Str s) const -> Str
    {
        replace(s);
        return s;
    }
};
 
auto inline Substr_Replacer::sort_uniq() -> void
{
    auto first = begin(table);
    auto last = end(table);
    sort(first, last, [](auto& a, auto& b) { return a.first < b.first; });
    auto it = unique(first, last,
                     [](auto& a, auto& b) { return a.first == b.first; });
    table.erase(it, last);
 
    // remove empty key ""
    if (!table.empty() && table.front().first.empty())
        table.erase(begin(table));
}
 
auto inline Substr_Replacer::find_match(Str_View s) const
{
    auto& t = table;
    struct Comparer_Str_Rep {
        auto static cmp_prefix_of(const Str& p, Str_View of)
        {
            return p.compare(0, p.npos, of.data(),
                             std::min(p.size(), of.size()));
        }
        auto operator()(const Pair_Str& a, Str_View b)
        {
            return cmp_prefix_of(a.first, b) < 0;
        }
        auto operator()(Str_View a, const Pair_Str& b)
        {
            return cmp_prefix_of(b.first, a) > 0;
        }
        auto static eq(const Pair_Str& a, Str_View b)
        {
            return cmp_prefix_of(a.first, b) == 0;
        }
    };
    Comparer_Str_Rep csr;
    auto it = begin(t);
    auto last_match = end(t);
    for (;;) {
        auto it2 = upper_bound(it, end(t), s, csr);
        if (it2 == it) {
            // not found, s is smaller that the range
            break;
        }
        --it2;
        if (csr.eq(*it2, s)) {
            // Match found. Try another search maybe for
            // longer.
            last_match = it2;
            it = ++it2;
        }
        else {
            // not found, s is greater that the range
            break;
        }
    }
    return last_match;
}
 
auto inline Substr_Replacer::replace(Str& s) const -> Str&
{
 
    if (table.empty())
        return s;
    for (size_t i = 0; i < s.size(); /*no increment here*/) {
        auto substr = Str_View(&s[i], s.size() - i);
        auto it = find_match(substr);
        if (it != end(table)) {
            auto& match = *it;
            // match found. match.first is the found string,
            // match.second is the replacement.
            s.replace(i, match.first.size(), match.second);
            i += match.second.size();
            continue;
        }
        ++i;
    }
    return s;
}
 
class Break_Table {
      public:
    using Str = std::string;
    using Table_Str = std::vector<Str>;
    using iterator = typename Table_Str::iterator;
    using const_iterator = typename Table_Str::const_iterator;
 
      private:
    Table_Str table;
    size_t start_word_breaks_last_idx = 0;
    size_t end_word_breaks_last_idx = 0;
 
    auto order_entries() -> void;
 
      public:
    Break_Table() = default;
    explicit Break_Table(const Table_Str& v) : table(v) { order_entries(); }
    explicit Break_Table(Table_Str&& v) : table(std::move(v))
    {
        order_entries();
    }
 
    auto& operator=(const Table_Str& v)
    {
        table = v;
        order_entries();
        return *this;
    }
 
    auto& operator=(Table_Str&& v)
    {
        table = std::move(v);
        order_entries();
        return *this;
    }
 
    auto start_word_breaks() const -> Subrange<const_iterator>
    {
        return {begin(table),
                begin(table) + start_word_breaks_last_idx};
    }
    auto end_word_breaks() const -> Subrange<const_iterator>
    {
        return {begin(table) + start_word_breaks_last_idx,
                begin(table) + end_word_breaks_last_idx};
    }
    auto middle_word_breaks() const -> Subrange<const_iterator>
    {
        return {begin(table) + end_word_breaks_last_idx, end(table)};
    }
};
auto inline Break_Table::order_entries() -> void
{
    auto it = remove_if(begin(table), end(table), [](auto& s) {
        return s.empty() ||
               (s.size() == 1 && (s[0] == '^' || s[0] == '$'));
    });
    table.erase(it, end(table));
 
    auto is_start_word_break = [=](auto& x) { return x[0] == '^'; };
    auto is_end_word_break = [=](auto& x) { return x.back() == '$'; };
    auto start_word_breaks_last =
        partition(begin(table), end(table), is_start_word_break);
    start_word_breaks_last_idx = start_word_breaks_last - begin(table);
 
    for_each(begin(table), start_word_breaks_last,
             [](auto& e) { e.erase(0, 1); });
 
    auto end_word_breaks_last =
        partition(start_word_breaks_last, end(table), is_end_word_break);
    end_word_breaks_last_idx = end_word_breaks_last - begin(table);
 
    for_each(start_word_breaks_last, end_word_breaks_last,
             [](auto& e) { e.pop_back(); });
}
 
struct identity {
    template <class T>
    constexpr auto operator()(T&& t) const noexcept -> T&&
    {
        return std::forward<T>(t);
    }
};
 
template <class Key, class T>
class Hash_Multimap {
    using bucket_type = std::forward_list<std::pair<Key, T>>;
    static constexpr float max_load_fact = 7.0 / 8.0;
    std::vector<bucket_type> data;
    size_t sz = 0;
    size_t max_load_factor_capacity = 0;
 
      public:
    using key_type = Key;
    using mapped_type = T;
    using value_type = std::pair<Key, T>;
    using size_type = std::size_t;
    using difference_type = std::ptrdiff_t;
    using hasher = std::hash<Key>;
    using reference = value_type&;
    using const_reference = const value_type&;
    using pointer = typename bucket_type::pointer;
    using const_pointer = typename bucket_type::const_pointer;
    using local_iterator = typename bucket_type::iterator;
    using local_const_iterator = typename bucket_type::const_iterator;
 
    Hash_Multimap() = default;
 
    auto size() const noexcept { return sz; }
    auto empty() const noexcept { return size() == 0; }
 
    auto rehash(size_t count)
    {
        if (empty()) {
            size_t capacity = 16;
            while (capacity <= count)
                capacity <<= 1;
            data.resize(capacity);
            max_load_factor_capacity =
                std::ceil(capacity * max_load_fact);
            return;
        }
        if (count < size() / max_load_fact)
            count = size() / max_load_fact;
        auto n = Hash_Multimap();
        n.rehash(count);
        for (auto& b : data) {
            for (auto& x : b) {
                n.insert(std::move(x));
            }
        }
        data.swap(n.data);
        sz = n.sz;
        max_load_factor_capacity = n.max_load_factor_capacity;
    }
 
    auto reserve(size_t count) -> void
    {
        rehash(std::ceil(count / max_load_fact));
    }
 
      private:
    auto prepare_for_insert(const key_type& key)
        -> std::pair<bucket_type&, local_iterator>
    {
        auto hash = hasher();
        if (sz == max_load_factor_capacity) {
            reserve(sz + 1);
        }
        auto h = hash(key);
        auto h_mod = h & (data.size() - 1);
        auto& bucket = data[h_mod];
        auto prev = bucket.before_begin();
        auto curr = begin(bucket);
        // find first entry with same key
        for (; curr != end(bucket); prev = curr++) {
            if (curr->first == key)
                break;
        }
        // find last entry with same key
        for (; curr != end(bucket); prev = curr++) {
            if (curr->first != key)
                break;
        }
        ++sz;
        return {bucket, prev};
    }
 
      public:
    auto insert(value_type&& value)
    {
        auto& key = value.first;
        auto [bucket, it] = prepare_for_insert(key);
        return bucket.insert_after(it, move(value));
    }
    template <class... Args>
    auto emplace(Args&&... a)
    {
        auto node = bucket_type();
        auto& val = node.emplace_front(std::forward<Args>(a)...);
        auto& key = val.first;
        auto [bucket, it] = prepare_for_insert(key);
        bucket.splice_after(it, node, node.before_begin());
        return next(it);
    }
 
    auto equal_range(const key_type& key) const
        -> std::pair<local_const_iterator, local_const_iterator>
    {
        auto hash = hasher();
        if (data.empty())
            return {};
        auto h = hash(key);
        auto h_mod = h & (data.size() - 1);
        auto& bucket = data[h_mod];
        auto eq_key = [&](auto& x) { return key == x.first; };
        auto first = std::find_if(begin(bucket), end(bucket), eq_key);
        if (first == end(bucket))
            return {first, first}; // ret empty
        auto last = std::find_if_not(next(first), end(bucket), eq_key);
        return {first, last};
    }
 
    auto bucket_count() const -> size_type { return data.size(); }
    auto bucket_data(size_type i) const { return Subrange(data[i]); }
};
 
struct Condition_Exception : public std::runtime_error {
    using std::runtime_error::runtime_error;
};
 
class Condition {
    using Str = std::string;
    using Str_View = std::string_view;
 
    Str cond;
    size_t num_cp = 0;
 
    auto construct() -> void;
 
      public:
    Condition() = default;
    explicit Condition(const Str& condition) : cond(condition)
    {
        construct();
    }
    explicit Condition(Str&& condition) : cond(std::move(condition))
    {
        construct();
    }
    explicit Condition(const char* condition) : cond(condition)
    {
        construct();
    }
    auto& operator=(const Str& condition)
    {
        cond = condition;
        num_cp = 0;
        construct();
        return *this;
    }
    auto& operator=(Str&& condition)
    {
        cond = std::move(condition);
        num_cp = 0;
        construct();
        return *this;
    }
    auto& operator=(const char* condition)
    {
        cond = condition;
        num_cp = 0;
        construct();
        return *this;
    }
    auto& str() const { return cond; }
    auto match_prefix(Str_View s) const -> bool;
    auto match_suffix(Str_View s) const
    {
        auto cp_cnt = size_t(0);
        auto i = size(s);
        for (; cp_cnt != num_cp && i != 0; ++cp_cnt) {
            valid_u8_reverse_index(s, i);
        }
        if (cp_cnt != num_cp)
            return false;
        return match_prefix(s.substr(i));
    }
};
auto inline Condition::construct() -> void
{
    for (size_t i = 0; i != size(cond);) {
        size_t j = cond.find_first_of("[].", i);
        if (j == cond.npos)
            j = size(cond);
        while (i != j) {
            valid_u8_advance_index(cond, i);
            ++num_cp;
        }
        if (i == size(cond))
            break;
        if (cond[i] == '.') {
            ++num_cp;
            ++i;
            continue;
        }
        else if (cond[i] == ']') {
            auto what =
                "closing bracket has no matching opening bracket";
            throw Condition_Exception(what);
        }
        else if (cond[i] == '[') {
            ++i;
            if (i == size(cond)) {
                auto what = "opening bracket has no matching "
                            "closing bracket";
                throw Condition_Exception(what);
            }
            if (cond[i] == '^')
                ++i;
            j = cond.find(']', i);
            if (j == i) {
                auto what = "empty bracket expression";
                throw Condition_Exception(what);
            }
            if (j == cond.npos) {
                auto what = "opening bracket has no matching "
                            "closing bracket";
                throw Condition_Exception(what);
            }
            ++num_cp;
            i = j + 1;
        }
    }
}
 
auto inline Condition::match_prefix(Str_View s) const -> bool
{
    if (size(s) < num_cp)
        return false;
 
    auto s_i = size_t(0);
    auto cond_i = size_t(0);
    for (; s_i != size(s) && cond_i != size(cond); ++cond_i) {
        auto s_cu = s[s_i];
        auto cond_cu = cond[cond_i];
        switch (cond_cu) {
        default:
            if (s_cu != cond_cu)
                return false;
            ++s_i;
            break;
        case '.':
            valid_u8_advance_index(s, s_i);
            break;
        case '[':
            ++cond_i;
            cond_cu = cond[cond_i];
            if (cond_cu == '^')
                ++cond_i;
            char32_t str_cp;
            valid_u8_advance_cp(s, s_i, str_cp);
            auto found = false;
            while (cond[cond_i] != ']') {
                char32_t cond_cp;
                valid_u8_advance_cp(cond, cond_i, cond_cp);
                if (str_cp == cond_cp)
                    found = true;
            }
            if (cond_cu != '^' && !found)
                return false;
            if (cond_cu == '^' && found)
                return false;
            break;
        }
    }
    return cond_i == size(cond);
}
 
struct Prefix {
    using Str = std::string;
 
    char16_t flag = 0;
    bool cross_product = false;
    Str stripping;
    Str appending;
    Flag_Set cont_flags;
    Condition condition;
 
    auto to_root(Str& word) const -> Str&
    {
        return word.replace(0, appending.size(), stripping);
    }
    auto to_root_copy(Str word) const -> Str
    {
        to_root(word);
        return word;
    }
 
    auto to_derived(Str& word) const -> Str&
    {
        return word.replace(0, stripping.size(), appending);
    }
    auto to_derived_copy(Str word) const -> Str
    {
        to_derived(word);
        return word;
    }
 
    auto check_condition(const Str& word) const -> bool
    {
        return condition.match_prefix(word);
    }
};
 
struct Suffix {
    using Str = std::string;
 
    char16_t flag = 0;
    bool cross_product = false;
    Str stripping;
    Str appending;
    Flag_Set cont_flags;
    Condition condition;
 
    auto to_root(Str& word) const -> Str&
    {
        return word.replace(word.size() - appending.size(),
                            appending.size(), stripping);
    }
    auto to_root_copy(Str word) const -> Str
    {
        to_root(word);
        return word;
    }
 
    auto to_derived(Str& word) const -> Str&
    {
        return word.replace(word.size() - stripping.size(),
                            stripping.size(), appending);
    }
    auto to_derived_copy(Str word) const -> Str
    {
        to_derived(word);
        return word;
    }
 
    auto check_condition(const Str& word) const -> bool
    {
        return condition.match_suffix(word);
    }
};
 
template <class T, class Key_Extr = identity, class Key_Transform = identity>
class Prefix_Multiset {
      public:
    using Value_Type = T;
    using Key_Type =
        std::remove_reference_t<decltype(std::declval<Key_Extr>()(
            std::declval<T>()))>;
    using Char_Type = typename Key_Type::value_type;
    using Traits = typename Key_Type::traits_type;
    using Vector_Type = std::vector<T>;
    using Iterator = typename Vector_Type::const_iterator;
 
      private:
    struct Ebo_Key_Extr : public Key_Extr {};
    struct Ebo_Key_Transf : public Key_Transform {};
    struct Ebo : public Ebo_Key_Extr, Ebo_Key_Transf {
        Vector_Type table;
    } ebo;
    std::basic_string<Char_Type> first_letter;
    std::vector<size_t> prefix_idx_with_first_letter;
 
    auto key_extractor() const -> const Ebo_Key_Extr& { return ebo; }
    auto key_transformator() const -> const Ebo_Key_Transf& { return ebo; }
    auto& get_table() { return ebo.table; }
    auto& get_table() const { return ebo.table; }
 
    auto sort()
    {
        auto& extract_key = key_extractor();
        auto& transform_key = key_transformator();
        auto& table = get_table();
 
        std::sort(begin(table), end(table), [&](T& a, T& b) {
            auto&& key_a = transform_key(extract_key(a));
            auto&& key_b = transform_key(extract_key(b));
            return key_a < key_b;
        });
 
        first_letter.clear();
        prefix_idx_with_first_letter.clear();
        auto first = begin(table);
        auto last = end(table);
        auto it = std::find_if_not(first, last, [=](const T& x) {
            auto&& k = transform_key(extract_key(x));
            return k.empty();
        });
        while (it != last) {
            auto&& k1 = transform_key(extract_key(*it));
            first_letter.push_back(k1[0]);
            prefix_idx_with_first_letter.push_back(it - first);
 
            it = std::upper_bound(
                it, last, k1[0],
                Comparator{0, extract_key, transform_key});
        }
        if (!prefix_idx_with_first_letter.empty())
            prefix_idx_with_first_letter.push_back(last - first);
    }
 
    struct Comparator {
        size_t i;
        Key_Extr extract_key;
        Key_Transform transform_key;
        auto operator()(const T& a, Char_Type b) const
        {
            auto&& key_a = transform_key(extract_key(a));
            return Traits::lt(key_a[i], b);
        }
        auto operator()(Char_Type a, const T& b) const
        {
            auto&& key_b = transform_key(extract_key(b));
            return Traits::lt(a, key_b[i]);
        }
    };
 
      public:
    Prefix_Multiset() = default;
    explicit Prefix_Multiset(Key_Extr ke, Key_Transform kt = {})
        : ebo{{ke}, {kt}, {}}
    {
    }
    explicit Prefix_Multiset(const Vector_Type& v, Key_Extr ke = {},
                             Key_Transform kt = {})
        : ebo{{ke}, {kt}, v}
    {
        sort();
    }
    explicit Prefix_Multiset(Vector_Type&& v, Key_Extr ke = {},
                             Key_Transform kt = {})
        : ebo{{ke}, {kt}, std::move(v)}
    {
        sort();
    }
    Prefix_Multiset(std::initializer_list<T> list, Key_Extr ke = {},
                    Key_Transform kt = {})
        : ebo{{ke}, {kt}, list}
    {
        sort();
    }
    auto& operator=(const Vector_Type& v)
    {
        get_table() = v;
        sort();
        return *this;
    }
    auto& operator=(Vector_Type&& v)
    {
        get_table() = std::move(v);
        sort();
        return *this;
    }
    auto& operator=(std::initializer_list<T> list)
    {
        get_table() = list;
        sort();
        return *this;
    }
    auto& data() const { return get_table(); }
 
    template <class Func>
    auto for_each_prefixes_of(const Key_Type& word, Func func) const;
 
    template <class OutIter>
    auto copy_all_prefixes_of(const Key_Type& word, OutIter o) const
    {
        for_each_prefixes_of(word, [&](const T& x) { *o++ = x; });
        return o;
    }
 
    class Iter_Prefixes_Of {
        const Prefix_Multiset* set = {};
        Iterator it = {};
        Iterator last = {};
        const Key_Type* search_key = {};
        size_t len = {};
        bool valid = false;
 
        auto advance() -> void;
 
          public:
        using iterator_category = std::input_iterator_tag;
        using value_type = T;
        using size_type = size_t;
        using difference_type = ptrdiff_t;
        using reference = const T&;
        using pointer = const T*;
 
        Iter_Prefixes_Of() = default;
        Iter_Prefixes_Of(const Prefix_Multiset& set,
                         const Key_Type& word)
            : set(&set), it(set.get_table().begin()),
              last(set.get_table().end()), search_key(&word),
              valid(true)
        {
            advance();
        }
        Iter_Prefixes_Of(const Prefix_Multiset&, Key_Type&&) = delete;
        Iter_Prefixes_Of(Prefix_Multiset&&, const Key_Type&) = delete;
        Iter_Prefixes_Of(Prefix_Multiset&&, Key_Type&&) = delete;
 
        auto& operator++()
        {
            ++it;
            advance();
            return *this;
        }
        auto& operator++(int)
        {
            auto old = *this;
            ++*this;
            return old;
        }
        auto& operator*() const { return *it; }
        auto operator->() const { return &*it; }
        auto operator==(const Iter_Prefixes_Of& other) const
        {
            return valid == other.valid;
        }
        auto operator!=(const Iter_Prefixes_Of& other) const
        {
            return valid != other.valid;
        }
 
        operator bool() const { return valid; }
 
        auto begin() const { return *this; }
        auto end() const { return Iter_Prefixes_Of(); }
    };
 
    auto iterate_prefixes_of(const Key_Type& word) const
    {
        return Iter_Prefixes_Of(*this, word);
    }
    auto iterate_prefixes_of(Key_Type&& word) const = delete;
};
template <class T, class Key_Extr, class Key_Transform>
auto Prefix_Multiset<T, Key_Extr, Key_Transform>::Iter_Prefixes_Of::advance()
    -> void
{
    auto& extract_key = set->key_extractor();
    auto& transform_key = set->key_transformator();
    auto& table = set->get_table();
 
    if (len == 0) {
        if (it != last) {
            if (transform_key(extract_key(*it)).empty())
                return;
            if (transform_key(*search_key).empty()) {
                valid = false;
                return;
            }
 
            auto& first_letter = set->first_letter;
            auto& prefix_idx_with_first_letter =
                set->prefix_idx_with_first_letter;
            auto idx =
                first_letter.find(transform_key(*search_key)[0]);
            if (idx == first_letter.npos) {
                valid = false;
                return;
            }
 
            auto first = table.begin();
            it = first + prefix_idx_with_first_letter[idx];
            last = first + prefix_idx_with_first_letter[idx + 1];
            ++len;
        }
        else {
            valid = false;
            return;
        }
    }
    for (;; ++len) {
        if (it != last) {
            if (transform_key(extract_key(*it)).size() == len)
                return;
            if (len == transform_key(*search_key).size()) {
                valid = false;
                break;
            }
        }
        else {
            valid = false;
            break;
        }
        tie(it, last) =
            equal_range(it, last, transform_key(*search_key)[len],
                        Comparator{len, extract_key, transform_key});
    }
}
 
template <class T, class Key_Extr, class Key_Transform>
template <class Func>
auto Prefix_Multiset<T, Key_Extr, Key_Transform>::for_each_prefixes_of(
    const Key_Type& word, Func func) const
{
    auto& extract_key = key_extractor();
    auto& transform_key = key_transformator();
    auto& table = get_table();
 
    auto first = begin(table);
    auto last = end(table);
    auto it = first;
    for (; it != last; ++it) {
        auto&& k = transform_key(extract_key(*it));
        if (k.empty())
            func(*it);
        else
            break;
    }
 
    if (it == last)
        return;
    if (transform_key(word).empty())
        return;
 
    auto idx = first_letter.find(transform_key(word)[0]);
    if (idx == first_letter.npos)
        return;
    first = begin(table) + prefix_idx_with_first_letter[idx];
    last = begin(table) + prefix_idx_with_first_letter[idx + 1];
 
    for (size_t len = 1;; ++len) {
        it = first;
        for (; it != last &&
               transform_key(extract_key(*it)).size() == len;
             ++it) {
            func(*it);
        }
        if (it == last)
            break;
        if (len == transform_key(word).size())
            break;
        tie(first, last) =
            equal_range(it, last, transform_key(word)[len],
                        Comparator{len, extract_key, transform_key});
    }
}
 
template <class CharT>
class Reversed_String_View {
      public:
    using Str = std::basic_string<CharT>;
    using traits_type = typename Str::traits_type;
    using value_type = typename Str::value_type;
    using size_type = typename Str::size_type;
    using const_iterator = typename Str::const_reverse_iterator;
 
      private:
    const_iterator first = {};
    size_type sz = {};
 
      public:
    Reversed_String_View() = default;
    explicit Reversed_String_View(const Str& s)
        : first(s.rbegin()), sz(s.size())
    {
    }
    Reversed_String_View(Str&& s) = delete;
    auto& operator[](size_type i) const { return first[i]; }
    auto size() const { return sz; }
    auto empty() const { return sz == 0; }
    auto begin() const { return first; }
    auto end() const { return first + sz; }
    auto operator<(Reversed_String_View other) const
    {
        return lexicographical_compare(begin(), end(), other.begin(),
                                       other.end(), traits_type::lt);
    }
};
 
template <class CharT>
struct String_Reverser {
    auto operator()(const std::basic_string<CharT>& x) const
    {
        return Reversed_String_View<CharT>(x);
    }
    // auto operator()(T&& x) const = delete;
};
 
template <class AffixT>
struct Extractor_Of_Appending_From_Affix {
    auto& operator()(const AffixT& a) const { return a.appending; }
};
 
template <class T, class Key_Extr = identity>
using Suffix_Multiset = Prefix_Multiset<
    T, Key_Extr,
    String_Reverser<typename Prefix_Multiset<T, Key_Extr>::Char_Type>>;
 
class Prefix_Table {
    using Prefix_Multiset_Type =
        Prefix_Multiset<Prefix, Extractor_Of_Appending_From_Affix<Prefix>>;
    using Key_Type = typename Prefix_Multiset_Type::Key_Type;
    using Vector_Type = typename Prefix_Multiset_Type::Vector_Type;
    Prefix_Multiset_Type table;
    Flag_Set all_cont_flags;
 
    auto populate()
    {
        for (auto& x : table.data())
            all_cont_flags += x.cont_flags;
    }
 
      public:
    Prefix_Table() = default;
    explicit Prefix_Table(const Vector_Type& t) : table(t) { populate(); }
    explicit Prefix_Table(Vector_Type&& t) : table(std::move(t))
    {
        populate();
    }
    auto& operator=(const Vector_Type& t)
    {
        table = t;
        populate();
        return *this;
    }
    auto& operator=(Vector_Type&& t)
    {
        table = std::move(t);
        populate();
        return *this;
    }
    auto begin() const { return table.data().begin(); }
    auto end() const { return table.data().end(); }
    auto size() const { return table.data().size(); }
 
    auto has_continuation_flags() const
    {
        return all_cont_flags.size() != 0;
    }
    auto has_continuation_flag(char16_t flag) const
    {
        return all_cont_flags.contains(flag);
    }
    auto iterate_prefixes_of(const Key_Type& word) const
    {
        return table.iterate_prefixes_of(word);
    }
    auto iterate_prefixes_of(Key_Type&& word) const = delete;
};
 
class Suffix_Table {
    using Suffix_Multiset_Type =
        Suffix_Multiset<Suffix, Extractor_Of_Appending_From_Affix<Suffix>>;
    using Key_Type = typename Suffix_Multiset_Type::Key_Type;
    using Vector_Type = typename Suffix_Multiset_Type::Vector_Type;
    Suffix_Multiset_Type table;
    Flag_Set all_cont_flags;
 
    auto populate()
    {
        for (auto& x : table.data())
            all_cont_flags += x.cont_flags;
    }
 
      public:
    Suffix_Table() = default;
    explicit Suffix_Table(const Vector_Type& t) : table(t) { populate(); }
    explicit Suffix_Table(Vector_Type&& t) : table(std::move(t))
    {
        populate();
    }
    auto& operator=(const Vector_Type& t)
    {
        table = t;
        populate();
        return *this;
    }
    auto& operator=(Vector_Type&& t)
    {
        table = std::move(t);
        populate();
        return *this;
    }
    auto begin() const { return table.data().begin(); }
    auto end() const { return table.data().end(); }
    auto size() const { return table.data().size(); }
 
    auto has_continuation_flags() const
    {
        return all_cont_flags.size() != 0;
    }
    auto has_continuation_flag(char16_t flag) const
    {
        return all_cont_flags.contains(flag);
    }
    auto iterate_suffixes_of(const Key_Type& word) const
    {
        return table.iterate_prefixes_of(word);
    }
    auto iterate_suffixes_of(Key_Type&& word) const = delete;
};
 
class String_Pair {
    using Str = std::string;
    using Str_View = std::string_view;
    size_t i = 0;
    Str s;
 
      public:
    String_Pair() = default;
    template <class Str1>
    explicit String_Pair(Str1&& str, size_t i)
        : i(i), s(std::forward<Str1>(str))
    {
        if (i > s.size())
            throw std::out_of_range("word split is too long");
    }
 
    template <class Str1, class Str2,
              class = std::enable_if_t<
                  std::is_same<std::remove_reference_t<Str1>, Str>::value &&
                  std::is_same<std::remove_reference_t<Str2>, Str>::value>>
    String_Pair(Str1&& first, Str2&& second)
        : i(first.size()) /* must be init before s, before we move
                             from first */
          ,
          s(std::forward<Str1>(first) + std::forward<Str2>(second))
 
    {
    }
    auto first() const { return Str_View(s).substr(0, i); }
    auto second() const { return Str_View(s).substr(i); }
    auto first(Str_View x)
    {
        s.replace(0, i, x);
        i = x.size();
    }
    auto second(Str_View x) { s.replace(i, s.npos, x); }
    auto& str() const { return s; }
    auto idx() const { return i; }
};
struct Compound_Pattern {
    String_Pair begin_end_chars;
    std::string replacement;
    char16_t first_word_flag = 0;
    char16_t second_word_flag = 0;
    bool match_first_only_unaffixed_or_zero_affixed = false;
};
 
class Compound_Rule_Table {
    std::vector<std::u16string> rules;
    Flag_Set all_flags;
 
    auto fill_all_flags() -> void;
 
      public:
    Compound_Rule_Table() = default;
    explicit Compound_Rule_Table(const std::vector<std::u16string>& tbl)
        : rules(tbl)
    {
        fill_all_flags();
    }
    explicit Compound_Rule_Table(std::vector<std::u16string>&& tbl)
        : rules(std::move(tbl))
    {
        fill_all_flags();
    }
    auto& operator=(const std::vector<std::u16string>& tbl)
    {
        rules = tbl;
        fill_all_flags();
        return *this;
    }
    auto& operator=(std::vector<std::u16string>&& tbl)
    {
        rules = std::move(tbl);
        fill_all_flags();
        return *this;
    }
    auto empty() const { return rules.empty(); }
    auto has_any_of_flags(const Flag_Set& f) const -> bool;
    auto match_any_rule(const std::vector<const Flag_Set*>& data) const
        -> bool;
};
auto inline Compound_Rule_Table::fill_all_flags() -> void
{
    for (auto& f : rules) {
        all_flags += f;
    }
    all_flags.erase(u'?');
    all_flags.erase(u'*');
}
 
auto inline Compound_Rule_Table::has_any_of_flags(const Flag_Set& f) const
    -> bool
{
    using std::begin;
    using std::end;
    struct Out_Iter_One_Bool {
        bool* value = nullptr;
        auto& operator++() { return *this; }
        auto& operator++(int) { return *this; }
        auto& operator*() { return *this; }
        auto& operator=(char16_t)
        {
            *value = true;
            return *this;
        }
    };
    auto has_intersection = false;
    auto out_it = Out_Iter_One_Bool{&has_intersection};
    std::set_intersection(begin(all_flags), end(all_flags), begin(f),
                          end(f), out_it);
    return has_intersection;
}
 
template <class DataIter, class PatternIter, class FuncEq = std::equal_to<>>
auto match_simple_regex(DataIter data_first, DataIter data_last,
                        PatternIter pat_first, PatternIter pat_last,
                        FuncEq eq = FuncEq())
{
    auto s = std::stack<std::pair<DataIter, PatternIter>>();
    s.emplace(data_first, pat_first);
    auto data_it = DataIter();
    auto pat_it = PatternIter();
    while (!s.empty()) {
        std::tie(data_it, pat_it) = s.top();
        s.pop();
        if (pat_it == pat_last) {
            if (data_it == data_last)
                return true;
            else
                return false;
        }
        auto node_type = *pat_it;
        if (pat_it + 1 == pat_last)
            node_type = 0;
        else
            node_type = *(pat_it + 1);
        switch (node_type) {
        case '?':
            s.emplace(data_it, pat_it + 2);
            if (data_it != data_last && eq(*data_it, *pat_it))
                s.emplace(data_it + 1, pat_it + 2);
            break;
        case '*':
            s.emplace(data_it, pat_it + 2);
            if (data_it != data_last && eq(*data_it, *pat_it))
                s.emplace(data_it + 1, pat_it);
 
            break;
        default:
            if (data_it != data_last && eq(*data_it, *pat_it))
                s.emplace(data_it + 1, pat_it + 1);
            break;
        }
    }
    return false;
}
 
template <class DataRange, class PatternRange, class FuncEq = std::equal_to<>>
auto match_simple_regex(const DataRange& data, const PatternRange& pattern,
                        FuncEq eq = FuncEq())
{
    using namespace std;
    return match_simple_regex(begin(data), end(data), begin(pattern),
                              end(pattern), eq);
}
 
auto inline match_compund_rule(const std::vector<const Flag_Set*>& words_data,
                               const std::u16string& pattern)
{
    return match_simple_regex(
        words_data, pattern,
        [](const Flag_Set* d, char16_t p) { return d->contains(p); });
}
 
auto inline Compound_Rule_Table::match_any_rule(
    const std::vector<const Flag_Set*>& data) const -> bool
{
    return any_of(begin(rules), end(rules), [&](const std::u16string& p) {
        return match_compund_rule(data, p);
    });
}
 
using List_Strings = std::vector<std::string>;
 
class Replacement_Table {
      public:
    using Str = std::string;
    using Table_Str = std::vector<std::pair<Str, Str>>;
    using iterator = typename Table_Str::iterator;
    using const_iterator = typename Table_Str::const_iterator;
 
      private:
    Table_Str table;
    size_t whole_word_reps_last_idx = 0;
    size_t start_word_reps_last_idx = 0;
    size_t end_word_reps_last_idx = 0;
 
    auto order_entries() -> void;
 
      public:
    Replacement_Table() = default;
    explicit Replacement_Table(const Table_Str& v) : table(v)
    {
        order_entries();
    }
    explicit Replacement_Table(Table_Str&& v) : table(std::move(v))
    {
        order_entries();
    }
 
    auto& operator=(const Table_Str& v)
    {
        table = v;
        order_entries();
        return *this;
    }
 
    auto& operator=(Table_Str&& v)
    {
        table = std::move(v);
        order_entries();
        return *this;
    }
 
    auto whole_word_replacements() const -> Subrange<const_iterator>
    {
        return {begin(table), begin(table) + whole_word_reps_last_idx};
    }
    auto start_word_replacements() const -> Subrange<const_iterator>
    {
        return {begin(table) + whole_word_reps_last_idx,
                begin(table) + start_word_reps_last_idx};
    }
    auto end_word_replacements() const -> Subrange<const_iterator>
    {
        return {begin(table) + start_word_reps_last_idx,
                begin(table) + end_word_reps_last_idx};
    }
    auto any_place_replacements() const -> Subrange<const_iterator>
    {
        return {begin(table) + end_word_reps_last_idx, end(table)};
    }
};
auto inline Replacement_Table::order_entries() -> void
{
    auto it = remove_if(begin(table), end(table), [](auto& p) {
        auto& s = p.first;
        return s.empty() ||
               (s.size() == 1 && (s[0] == '^' || s[0] == '$'));
    });
    table.erase(it, end(table));
 
    auto is_start_word_pat = [=](auto& x) { return x.first[0] == '^'; };
    auto is_end_word_pat = [=](auto& x) { return x.first.back() == '$'; };
 
    auto start_word_reps_last =
        partition(begin(table), end(table), is_start_word_pat);
    start_word_reps_last_idx = start_word_reps_last - begin(table);
    for_each(begin(table), start_word_reps_last,
             [](auto& e) { e.first.erase(0, 1); });
 
    auto whole_word_reps_last =
        partition(begin(table), start_word_reps_last, is_end_word_pat);
    whole_word_reps_last_idx = whole_word_reps_last - begin(table);
    for_each(begin(table), whole_word_reps_last,
             [](auto& e) { e.first.pop_back(); });
 
    auto end_word_reps_last =
        partition(start_word_reps_last, end(table), is_end_word_pat);
    end_word_reps_last_idx = end_word_reps_last - begin(table);
    for_each(start_word_reps_last, end_word_reps_last,
             [](auto& e) { e.first.pop_back(); });
}
 
struct Similarity_Group {
    using Str = std::string;
    Str chars;
    std::vector<Str> strings;
 
    auto parse(const Str& s) -> void;
    Similarity_Group() = default;
    explicit Similarity_Group(const Str& s) { parse(s); }
    auto& operator=(const Str& s)
    {
        parse(s);
        return *this;
    }
};
auto inline Similarity_Group::parse(const Str& s) -> void
{
    auto i = size_t(0);
    for (;;) {
        auto j = s.find('(', i);
        chars.append(s, i, j - i);
        if (j == s.npos)
            break;
        i = j + 1;
        j = s.find(')', i);
        if (j == s.npos)
            break;
        auto len = j - i;
        if (len == 1)
            chars += s[i];
        else if (len > 1)
            strings.push_back(s.substr(i, len));
        i = j + 1;
    }
}
 
class Phonetic_Table {
    using Char_Type = char; // not aware of unicode
    using Str = std::string;
    using Pair_Str = std::pair<Str, Str>;
 
    struct Phonet_Match_Result {
        size_t count_matched = 0;
        size_t go_back_before_replace = 0;
        size_t priority = 5;
        bool go_back_after_replace = false;
        bool treat_next_as_begin = false;
        operator bool() { return count_matched; }
    };
 
    std::vector<std::pair<Str, Str>> table;
    auto order() -> void;
    auto static match(const Str& data, size_t i, const Str& pattern,
                      bool at_begin) -> Phonet_Match_Result;
 
      public:
    Phonetic_Table() = default;
    explicit Phonetic_Table(const std::vector<Pair_Str>& v) : table(v)
    {
        order();
    }
    explicit Phonetic_Table(std::vector<Pair_Str>&& v) : table(std::move(v))
    {
        order();
    }
    auto& operator=(const std::vector<Pair_Str>& v)
    {
        table = v;
        order();
        return *this;
    }
    auto& operator=(std::vector<Pair_Str>&& v)
    {
        table = std::move(v);
        order();
        return *this;
    }
    auto replace(Str& word) const -> bool;
};
 
auto inline Phonetic_Table::order() -> void
{
    stable_sort(begin(table), end(table), [](auto& pair1, auto& pair2) {
        if (pair2.first.empty())
            return false;
        if (pair1.first.empty())
            return true;
        return pair1.first[0] < pair2.first[0];
    });
    auto it = find_if_not(begin(table), end(table),
                          [](auto& p) { return p.first.empty(); });
    table.erase(begin(table), it);
    for (auto& r : table) {
        if (r.second == "_")
            r.second.clear();
    }
}
 
auto inline Phonetic_Table::match(const Str& data, size_t i, const Str& pattern,
                                  bool at_begin) -> Phonet_Match_Result
{
    auto ret = Phonet_Match_Result();
    auto j = pattern.find_first_of("(<-0123456789^$");
    if (j == pattern.npos)
        j = pattern.size();
    if (data.compare(i, j, pattern, 0, j) == 0)
        ret.count_matched = j;
    else
        return {};
    if (j == pattern.size())
        return ret;
    if (pattern[j] == '(') {
        auto k = pattern.find(')', j);
        if (k == pattern.npos)
            return {}; // bad rule
        auto x = std::char_traits<Char_Type>::find(
            &pattern[j + 1], k - (j + 1), data[i + j]);
        if (!x)
            return {};
        j = k + 1;
        ret.count_matched += 1;
    }
    if (j == pattern.size())
        return ret;
    if (pattern[j] == '<') {
        ret.go_back_after_replace = true;
        ++j;
    }
    auto k = pattern.find_first_not_of('-', j);
    if (k == pattern.npos) {
        k = pattern.size();
        ret.go_back_before_replace = k - j;
        if (ret.go_back_before_replace >= ret.count_matched)
            return {}; // bad rule
        return ret;
    }
    else {
        ret.go_back_before_replace = k - j;
        if (ret.go_back_before_replace >= ret.count_matched)
            return {}; // bad rule
    }
    j = k;
    if (pattern[j] >= '0' && pattern[j] <= '9') {
        ret.priority = pattern[j] - '0';
        ++j;
    }
    if (j == pattern.size())
        return ret;
    if (pattern[j] == '^') {
        if (!at_begin)
            return {};
        ++j;
    }
    if (j == pattern.size())
        return ret;
    if (pattern[j] == '^') {
        ret.treat_next_as_begin = true;
        ++j;
    }
    if (j == pattern.size())
        return ret;
    if (pattern[j] != '$')
        return {}; // bad rule, no other char is allowed at this point
    if (i + ret.count_matched == data.size())
        return ret;
    return {};
}
 
auto inline Phonetic_Table::replace(Str& word) const -> bool
{
    struct Cmp {
        auto operator()(Char_Type c, const Pair_Str& s)
        {
            return c < s.first[0];
        }
        auto operator()(const Pair_Str& s, Char_Type c)
        {
            return s.first[0] < c;
        }
    };
    if (table.empty())
        return false;
    auto ret = false;
    auto treat_next_as_begin = true;
    size_t count_go_backs_after_replace = 0; // avoid infinite loop
    for (size_t i = 0; i != word.size(); ++i) {
        auto rules =
            equal_range(begin(table), end(table), word[i], Cmp());
        for (auto& r : Subrange(rules)) {
            auto rule = &r;
            auto m1 = match(word, i, r.first, treat_next_as_begin);
            if (!m1)
                continue;
            if (!m1.go_back_before_replace) {
                auto j = i + m1.count_matched - 1;
                auto rules2 = equal_range(
                    begin(table), end(table), word[j], Cmp());
                for (auto& r2 : Subrange(rules2)) {
                    auto m2 =
                        match(word, j, r2.first, false);
                    if (m2 && m2.priority >= m1.priority) {
                        i = j;
                        rule = &r2;
                        m1 = m2;
                        break;
                    }
                }
            }
            word.replace(
                i, m1.count_matched - m1.go_back_before_replace,
                rule->second);
            treat_next_as_begin = m1.treat_next_as_begin;
            if (m1.go_back_after_replace &&
                count_go_backs_after_replace < 100) {
                count_go_backs_after_replace++;
            }
            else {
                i += rule->second.size();
            }
            --i;
            ret = true;
            break;
        }
    }
    return ret;
}
NUSPELL_END_INLINE_NAMESPACE
} // namespace nuspell
#endif // NUSPELL_STRUCTURES_HXX