// libunicode

#pragma once

#include <algorithm>
#include <stdexcept>
#include <string>

#ifdef __cpp_char8_t
// char8_t available
 typedef char8_t utf8_t;
#else
 typedef char utf8_t;
#endif

namespace unicode {

 // usually, char32_t, uint32_t etc.
 template<typename T>
 static inline bool is_valid_unicode(const T& value)
 {
   return value <= 0x10FFFF && (value <= 0xD7FF || value >= 0xE000);
 }

}

namespace unicode::detail {

 using namespace std::string_literals;

 template<typename T>
 struct utf_iterator
 {
  typedef char32_t value_type;
  typedef char32_t& reference;
  typedef std::basic_string<T> string_type;

  utf_iterator(const typename string_type::const_iterator& cbegin, const typename string_type::const_iterator& cend):
   iterator(cbegin), end_iterator(cend)
  {
   calculate_value<T>();
  }

  utf_iterator<T>(const utf_iterator<T>& other) = default;
  utf_iterator<T>& operator=(const utf_iterator<T>& other) = default;

  size_t remaining_code_units()
  {
   return end_iterator - iterator;
  }

  template<size_t index>
  T get_code_unit()
  {
   return *(iterator + index);
  }

  // set value member
  // default: char32_t for UTF-32
  // specializations for UTF-8 and UTF-16 below
  template<typename T1>
  void calculate_value()
  {
   static_assert(sizeof(T1) == 4);

   size_t remaining{remaining_code_units()};

   if (!remaining)
    return;

   value = get_code_unit<0>();
   
   if (!unicode::is_valid_unicode(value))
    throw std::invalid_argument("Invalid Unicode character: "s + std::to_string(static_cast<uint32_t>(value)));

   sequence_length = 1;
  }

  inline static bool is_continuation_byte(T b)
  {
   return (b & 0b11000000) == 0b10000000;
  }

  template<typename... Targs>
  inline static bool is_continuation_byte(T b, Targs... Fargs)
  {
   return is_continuation_byte(b) && is_continuation_byte(Fargs...);
  }

  template<size_t n>
  inline static bool is_byte0_of(T b)
  {
   return (b & static_cast<T>(0xFF << (7 - n))) == static_cast<T>(0xFF << (8 - n));
  }

  inline static char32_t continuation_value(T b)
  {
   return static_cast<char32_t>(b & 0b00111111);
  }

  template<typename... Targs>
  inline static char32_t continuation_value(T b, Targs... Fargs)
  {
   return continuation_value(b) << (6 * sizeof...(Targs)) | continuation_value(Fargs...);
  }

  template<size_t n>
  inline static char32_t value_byte0_of(T b)
  {
   return static_cast<char32_t>(b & (0b1111111 >> n)) << ((n - 1) * 6);
  }

  // GCC Bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85282
  // specialization for UTF-8
  template<>
  void calculate_value<utf8_t>()
  {
   size_t remaining{remaining_code_units()};
   
   if (!remaining)
    return;

   utf8_t byte0 {get_code_unit<0>()};
   if (byte0 & 0x80) { // 2-4 bytes
    if (remaining >= 2) {
     utf8_t byte1 {get_code_unit<1>()};
     if (is_byte0_of<2>(byte0) && is_continuation_byte(byte1)) { // 2 bytes
      value = value_byte0_of<2>(byte0) | continuation_value(byte1);
      sequence_length = 2;
     } else if (remaining >= 3) {
      utf8_t byte2 {get_code_unit<2>()};
      if (is_byte0_of<3>(byte0) && is_continuation_byte(byte1, byte2)) { // 3 bytes
       value = value_byte0_of<3>(byte0) | continuation_value(byte1, byte2);
       sequence_length = 3;
      } else if (remaining >= 4) {
       utf8_t byte3 {get_code_unit<3>()};
       if (is_byte0_of<4>(byte0) && is_continuation_byte(byte1, byte2, byte3)) { // 4 bytes
        value = value_byte0_of<4>(byte0) | continuation_value(byte1, byte2, byte3);
        sequence_length = 4;
       } else
        throw std::invalid_argument("Bad input: Invalid 4 byte sequence");
      } else
       throw std::invalid_argument("Bad input: Invalid 3 byte sequence");
     } else
      throw std::invalid_argument("Bad input: Invalid 2 byte sequence");
    } else
     throw std::invalid_argument("Bad input: 2nd byte expected, none found");
  
    // check only for sequences >= 2 bytes (ASCII is always compliant)
    if (!unicode::is_valid_unicode(value))
     throw std::invalid_argument("Invalid Unicode character: "s + std::to_string(static_cast<uint32_t>(value)));

   } else { // 1 byte: 7 bit ASCII
    value = byte0;
    sequence_length = 1;
   }
  }

  // specialization for UTF-16
  template<>
  void calculate_value<char16_t>()
  {
   size_t remaining{remaining_code_units()};
   
   if (!remaining)
    return;

   char16_t unit0 {get_code_unit<0>()};

   if (unit0 <= 0xD7FF || unit0 >= 0xE000) { // 1 unit (BMP Basic Multilingual Plane)
    value = unit0;
    sequence_length = 1;
   } else {
    if (remaining < 2)
     throw std::invalid_argument("Bad input: Continuation of first UTF-16 unit missing");

    char16_t unit1 {get_code_unit<1>()};
    if ((unit0 & 0xFC00) != 0xD800 || (unit1 & 0xFC00) != 0xDC00)
     throw std::invalid_argument("Bad input: 2 malformed UTF-16 surrogates");

    value = (static_cast<char32_t>(unit0 & 0x03FF) << 10 | (unit1 & 0x03FF)) + 0x10000;
    sequence_length = 2;
   }
  }

  // pre-increment
  utf_iterator<T>& operator++()
  {
   iterator += sequence_length;
   calculate_value<T>();
   return *this;
  }

  bool operator!=(const utf_iterator<T>& other) const
  {
   return iterator != other.iterator;
  }

  reference operator*()
  {
   return value;
  }

  typename string_type::const_iterator iterator;
  typename string_type::const_iterator end_iterator;

  char32_t value{}; // always save complete unicode code point at this point
  size_t sequence_length{};
 };

 template<typename T>
 struct utf_back_insert_iterator
 {
  typedef std::basic_string<T> string_type;
  typedef utf_back_insert_iterator& reference;

  utf_back_insert_iterator(string_type& s): s(s) {}

  // no-op
  utf_back_insert_iterator& operator++()
  {
   return *this;
  }

  // support *x = value, together with operator=()
  reference operator*()
  {
   return *this;
  }

  // default: utf-32 code unit for UTF-32
  // specializations for UTF-8 and UTF-16 below
  template<typename T1=T>
  reference operator=(const char32_t& value)
  {
   // expect value to be already valid Unicode values
   s.push_back(value);
   return *this;
  }

  // n is number of UTF-8 bytes in sequence
  template<size_t n>
  inline static T byte0_of(char32_t value)
  {
   return (value >> 6 * (n - 1)) | (0xFF << (8 - n));
  }

  // n is index of 6-bit groups, counting from bit 0
  template<size_t n>
  inline static T trailing_byte(char32_t value)
  {
   return ((value >> n * 6) & 0b111111) | 0b10000000;
  }

  // calculate UTF-8 sequence byte for m >= 2 bytes sequences (i.e. non-ASCII)
  // assume value to be valid Unicode value for given byte position
  template<size_t n, size_t m>
  inline static T byte_n_of_m(char32_t value)
  {
   if constexpr (n == 0)
    return byte0_of<m>(value);
   else
    return trailing_byte<m - n - 1>(value);
  }

  // specialization for UTF-8
  // append utf-8 byte sequence
  template<>
  reference operator=<utf8_t>(const char32_t& value)
  {
   if (value < 0x80) { // 1 byte
    s.push_back(value);
   } else if (value < 0x800) { // 2 bytes
    s.push_back(byte_n_of_m<0,2>(value));
    s.push_back(byte_n_of_m<1,2>(value));
   } else if (value < 0x10000) { // 3 bytes
    s.push_back(byte_n_of_m<0,3>(value));
    s.push_back(byte_n_of_m<1,3>(value));
    s.push_back(byte_n_of_m<2,3>(value));
   } else if (value < 0x110000) { // 4 bytes
    s.push_back(byte_n_of_m<0,4>(value));
    s.push_back(byte_n_of_m<1,4>(value));
    s.push_back(byte_n_of_m<2,4>(value));
    s.push_back(byte_n_of_m<3,4>(value));
   } else
    throw std::runtime_error("Invalid internal Unicode value: "s + std::to_string(static_cast<uint32_t>(value)));
   return *this;
  }

  // specialization for UTF-16
  // append utf-16 word sequence
  template<>
  reference operator=<char16_t>(const char32_t& value)
  {
   if (value <= 0xFFFF) { // expect value to be already valid Unicode values
    s.push_back(value);
   } else {
    char32_t value_reduced{value - 0x10000};
    s.push_back((value_reduced >> 10) + 0xD800);
    s.push_back((value_reduced & 0x3FF) + 0xDC00);
   }
   return *this;
  }

  typename utf_back_insert_iterator::string_type& s;
 };

 template<typename T>
 utf_back_insert_iterator<T> utf_back_inserter(std::basic_string<T>& s)
 {
  return utf_back_insert_iterator<T>(s);
 }

 template<typename T>
 utf_iterator<T> utf_begin(const std::basic_string<T>& s)
 {
  return utf_iterator<T>{s.cbegin(), s.cend()};
 }

 template<typename T>
 utf_iterator<T> utf_end(const std::basic_string<T>& s)
 {
  return utf_iterator<T>{s.cend(), s.cend()};
 }

} // namespace

namespace unicode {

 using namespace detail;

 template<typename From, typename To>
 std::basic_string<To> utf_to_utf(const std::basic_string<From>& s)
 {
  std::basic_string<To> result;

  std::copy(utf_begin<From>(s), utf_end<From>(s), utf_back_inserter<To>(result));

  return result;
 }

 template<typename T>
 bool is_valid_utf(const std::basic_string<T>& s)
 {
  try {
   std::for_each(utf_begin<T>(s), utf_end<T>(s), [](const T& c){});
  } catch(...) {
   return false;
  }
  return true;
 }

} // namespace unicode