icuライブラリで書記素を扱うラッパーを作ってみた

#include <iostream>

// 自作ラッパー
#include "GraphemeText.hpp"

#pragma warning(disable:4996)

int main()
{
  // 文字列の定義
  GText gtext(u"aあغ山👨‍👧經经");

  gtext.insert(0, u"غ");

  FILE* fp = fopen("C:\\test\\test.txt", "wb");

  for (size_t i = 0; i < gtext.length(); i++) {

    wchar_t c[100]{ 0 };
    int len = gtext[i].size();
    
    // 一文字書き込み
    memcpy(c, gtext[i].ptr(), len);
    fwrite(c, 1, len, fp);
  }

  fclose(fp);

}

#pragma once

// 以下のdllを要求される
// icudt69.dll
// icuuc69.dll

//! @brief 書記素クラス
class Grapheme {
  void* _UnicodeString;
  int _index;
  int _length;
  enum UScriptCode _script;
  bool _emoji;
public:

  Grapheme(void* _UnicodeString_, int _index_, int _length_);
  ~Grapheme() {}

  const char16_t* ptr()const;

  //! @brief この書記素の元の文字列上のindexを取得。
  int index()const { return _index; }

  //! @brief この書記素のchar16_tとしての文字数。
  int length()const {return _length;}

  //! @brief この書記素のバイト数。
  int size()const { return _length * 2; }

  //! @brief この文字の用字の種類
  enum UScriptCode script()const { return _script; }

  //! @brief これが絵文字かどうか
  bool isEmoji()const { return _emoji; }
};

///////////////////////////////////
//! @brief 書記素の配列オブジェクト
class GText {
  struct GText_Impl* _impl;
public:
  GText(const char16_t* text);
  GText(const GText& src);
  ~GText();
  const Grapheme& operator[](const size_t index)const;

  //! @brief この文字列の書記素数
  size_t length()const;

  //! @brief 書記素を一つ置き換える
  //! @attention Graphemeリストを更新するのでこれまでのGraphemeは使えなくなる
  void replace(const size_t index, const char16_t c);
  void replace(const size_t index, const char16_t* c);

  void erase(const size_t index);

  void insert(const size_t index, const char16_t c);
  void insert(const size_t index, const char16_t* c);

  const char16_t* ptr()const;

  //! @brief この文字列のバイト数
  int size()const;

};

#include<vector>
#include <string>

#include <unicode/ucnv.h>
#include <unicode/brkiter.h>
#include <unicode/utypes.h>
#include <unicode/uscript.h>

#include "GraphemeText.hpp"


// 要リンク
#pragma comment(lib, "icuuc.lib")
#pragma comment(lib, "icudt.lib")

// 書記素クラス
Grapheme::Grapheme(void* _UnicodeString_, int _index_, int _length_) :
  _UnicodeString(_UnicodeString_),
  _index(_index_),
  _length(_length_)
{

  icu::UnicodeString* p = (icu::UnicodeString*)_UnicodeString;

  UErrorCode err;

  UChar32 c32 = p->char32At(_index);

  // スクリプト
  _script = uscript_getScript(c32, &err);

  _emoji = false;
  // 絵文字かどうかのフラグをセット

  _emoji |= (bool)u_getIntPropertyValue(c32, UCHAR_EMOJI);
  _emoji |= (bool)u_getIntPropertyValue(c32, UCHAR_EMOJI_PRESENTATION);
  _emoji |= (bool)u_getIntPropertyValue(c32, UCHAR_EMOJI_MODIFIER);
  _emoji |= (bool)u_getIntPropertyValue(c32, UCHAR_EMOJI_MODIFIER_BASE);
  _emoji |= (bool)u_getIntPropertyValue(c32, UCHAR_EMOJI_COMPONENT);


}


const char16_t* Grapheme::ptr()const {
  return static_cast<icu::UnicodeString*>(_UnicodeString)->getBuffer() + _index;
}

// Graphemeの配列を作成
void makeGraphemeList(std::vector< Grapheme >& glist, icu::UnicodeString& text) {
  UErrorCode err;
  icu::BreakIterator* bi = icu::BreakIterator::createCharacterInstance(
    icu::Locale::getDefault(), err);

  if (bi == nullptr)
    throw "bi is NULL";

  bi->setText(text);


  int32_t current = bi->first();
  while (current != icu::BreakIterator::DONE) {

    int32_t prev = current;
    current = bi->next();

    if (current == UBRK_DONE) {
      break;
    }

    int32_t count = current - prev;//文字の長さ

    //書記素保存
    glist.emplace_back(&text, prev, count);
  }

  delete bi;

}

struct GText_Impl {
  std::vector< Grapheme > glist;
  icu::UnicodeString utext;
};

GText::GText(const char16_t* text) {
  _impl = new GText_Impl;

  _impl->utext = text;

  makeGraphemeList(_impl->glist, _impl->utext);
}

const char16_t* GText::ptr()const {
  return _impl->utext.getBuffer();
}


GText::GText(const GText& src) {
  _impl->utext = src._impl->utext;

  _impl->glist.clear();

  makeGraphemeList(_impl->glist, _impl->utext);

}

GText::~GText() {
  delete _impl;
}

const Grapheme& GText::operator[](const size_t index)const {
  return _impl->glist[index];
}

size_t GText::length()const {
  return _impl->glist.size();
}
int GText::size()const {
  return _impl->utext.length() * 2;
}

void GText::replace(const size_t index, const char16_t c) {

  if (index >= _impl->glist.size())
    throw "GText replace index is too big";

  _impl->utext.replace(_impl->glist[index].index(), _impl->glist[index].length(), c);
  _impl->glist.clear();
  makeGraphemeList(_impl->glist, _impl->utext);


}

void GText::replace(const size_t index, const char16_t* c) {
  if (index >= _impl->glist.size())
    throw "GText replace index is too big";

  size_t len = std::char_traits<char16_t>::length(c);

  _impl->utext.replace(_impl->glist[index].index(), _impl->glist[index].length(),c, len);
  _impl->glist.clear();
  makeGraphemeList(_impl->glist, _impl->utext);

}


void GText::erase(const size_t index) {
  if (index >= _impl->glist.size())
    throw "GText erase index is too big";

  _impl->utext.remove(_impl->glist[index].index(), _impl->glist[index].length());

  _impl->glist.clear();
  makeGraphemeList(_impl->glist, _impl->utext);
}

void GText::insert(const size_t index, const char16_t c) {
  if (index >= _impl->glist.size())
    throw "GText insert index is too big";

  _impl->utext.insert(_impl->glist[index].index(), c);

  _impl->glist.clear();
  makeGraphemeList(_impl->glist, _impl->utext);

}

void GText::insert(const size_t index, const char16_t* c) {
  if (index >= _impl->glist.size())
    throw "GText insert index is too big";

  size_t len = std::char_traits<char16_t>::length(c);

  _impl->utext.insert(_impl->glist[index].index(), c, len);
  _impl->glist.clear();
  makeGraphemeList(_impl->glist, _impl->utext);

}