develop/UTF-8ひらがなカタカナ文字列考察C/C++ のバックアップ(No.1)

• バックアップ一覧
• 差分 を表示
• 現在との差分 を表示
• 現在との差分 - Visual を表示
• ソース を表示
• develop/UTF-8ひらがなカタカナ文字列考察C/C++ へ行く。
  • 1 (2012-02-15 (水) 01:50:09)
  • 2 (2012-02-17 (金) 02:25:47)
  • 3 (2012-02-18 (土) 12:39:28)
  • 4 (2012-02-20 (月) 08:15:52)
  • 5 (2012-05-31 (木) 00:07:12)

殆どというか、Cでしか書いていないのはご愛嬌でｗ 実行環境はCentOS5のコンパイル環境。文字コード設定はもちろんUTF-8

source

#include <string>
#include <iostream>

using namespace std;

int main()
{
  char kanaa[] = "あい";
  char hiragana[] = "ぁあぃいぅうぇえぉおかがきぎくぐけげこごさざしじすずせぜそぞただちぢっつづてでとどなにぬねのはばぱひびぴふぶぷへべぺほぼぽまみむめもゃやゅゆょよらりるれろゎわゐゑをん　　　ゝゞ";
  char katakana[] = "ァアィイゥウェエォオカガキギクグケゲコゴサザシジスズセゼソゾタダチヂッツヅテデトドナニヌネノハバパヒビピフブプヘベペホボポマミムメモャヤュユョヨラリルレロヮワヰヱヲンヴヵヶヽヾ";

  int i,j;

  printf("sizeof \"%s\": %d\n", &kanaa, sizeof(kanaa));
  for( i=0 ; i<=sizeof(kanaa) ; i++){
    printf(" , %x", kanaa[i]);
  }
  printf("\n\n");

  printf("%s\n", &hiragana);
  printf("%s\n", &katakana);

  if(sizeof(hiragana) == sizeof(katakana)){
    for( i=0 ; i<sizeof(hiragana) ; i+=3){
      printf("\"%c%c%c\" - \"%c%c%c\" = ",
             hiragana[i],
             hiragana[i+1],
             hiragana[i+2],
             katakana[i],
             katakana[i+1],
             katakana[i+2]);
      for( j=0 ; j<=2 ; j++){
        printf(" %x ,", katakana[i+j] - hiragana[i+j] );
      }
      printf("\n");
    }
  }else{
    printf("invaled compare string.\n");
  }
  return 0;
}

実行結果

$ ./a.out
sizeof "あい": 7
 , ffffffe3 , ffffff81 , ffffff82 , ffffffe3 , ffffff81 , ffffff84 , 0 , 0

ぁあぃいぅうぇえぉおかがきぎくぐけげこごさざしじすずせぜそぞただちぢっつづてでとどなにぬねのはばぱひびぴふぶぷへべぺほぼぽまみむめもゃやゅゆょよらりるれろゎわゐゑをん　　　ゝゞ
ァアィイゥウェエォオカガキギクグケゲコゴサザシジスズセゼソゾタダチヂッツヅテデトドナニヌネノハバパヒビピフブプヘベペホボポマミムメモャヤュユョヨラリルレロヮワヰヱヲンヴヵヶヽヾ
"ぁ" - "ァ" =  0 , 1 , 20 ,
"あ" - "ア" =  0 , 1 , 20 ,
"ぃ" - "ィ" =  0 , 1 , 20 ,
"い" - "イ" =  0 , 1 , 20 ,
"ぅ" - "ゥ" =  0 , 1 , 20 ,
"う" - "ウ" =  0 , 1 , 20 ,
"ぇ" - "ェ" =  0 , 1 , 20 ,
"え" - "エ" =  0 , 1 , 20 ,
"ぉ" - "ォ" =  0 , 1 , 20 ,
"お" - "オ" =  0 , 1 , 20 ,
"か" - "カ" =  0 , 1 , 20 ,
"が" - "ガ" =  0 , 1 , 20 ,
"き" - "キ" =  0 , 1 , 20 ,
"ぎ" - "ギ" =  0 , 1 , 20 ,
"く" - "ク" =  0 , 1 , 20 ,
"ぐ" - "グ" =  0 , 1 , 20 ,
"け" - "ケ" =  0 , 1 , 20 ,
"げ" - "ゲ" =  0 , 1 , 20 ,
"こ" - "コ" =  0 , 1 , 20 ,
"ご" - "ゴ" =  0 , 1 , 20 ,
"さ" - "サ" =  0 , 1 , 20 ,
"ざ" - "ザ" =  0 , 1 , 20 ,
"し" - "シ" =  0 , 1 , 20 ,
"じ" - "ジ" =  0 , 1 , 20 ,
"す" - "ス" =  0 , 1 , 20 ,
"ず" - "ズ" =  0 , 1 , 20 ,
"せ" - "セ" =  0 , 1 , 20 ,
"ぜ" - "ゼ" =  0 , 1 , 20 ,
"そ" - "ソ" =  0 , 1 , 20 ,
"ぞ" - "ゾ" =  0 , 1 , 20 ,
"た" - "タ" =  0 , 1 , 20 ,
"だ" - "ダ" =  0 , 2 , ffffffe0 ,
"ち" - "チ" =  0 , 2 , ffffffe0 ,
"ぢ" - "ヂ" =  0 , 2 , ffffffe0 ,
"っ" - "ッ" =  0 , 2 , ffffffe0 ,
"つ" - "ツ" =  0 , 2 , ffffffe0 ,
"づ" - "ヅ" =  0 , 2 , ffffffe0 ,
"て" - "テ" =  0 , 2 , ffffffe0 ,
"で" - "デ" =  0 , 2 , ffffffe0 ,
"と" - "ト" =  0 , 2 , ffffffe0 ,
"ど" - "ド" =  0 , 2 , ffffffe0 ,
"な" - "ナ" =  0 , 2 , ffffffe0 ,
"に" - "ニ" =  0 , 2 , ffffffe0 ,
"ぬ" - "ヌ" =  0 , 2 , ffffffe0 ,
"ね" - "ネ" =  0 , 2 , ffffffe0 ,
"の" - "ノ" =  0 , 2 , ffffffe0 ,
"は" - "ハ" =  0 , 2 , ffffffe0 ,
"ば" - "バ" =  0 , 2 , ffffffe0 ,
"ぱ" - "パ" =  0 , 2 , ffffffe0 ,
"ひ" - "ヒ" =  0 , 2 , ffffffe0 ,
"び" - "ビ" =  0 , 2 , ffffffe0 ,
"ぴ" - "ピ" =  0 , 2 , ffffffe0 ,
"ふ" - "フ" =  0 , 2 , ffffffe0 ,
"ぶ" - "ブ" =  0 , 2 , ffffffe0 ,
"ぷ" - "プ" =  0 , 2 , ffffffe0 ,
"へ" - "ヘ" =  0 , 2 , ffffffe0 ,
"べ" - "ベ" =  0 , 2 , ffffffe0 ,
"ぺ" - "ペ" =  0 , 2 , ffffffe0 ,
"ほ" - "ホ" =  0 , 2 , ffffffe0 ,
"ぼ" - "ボ" =  0 , 2 , ffffffe0 ,
"ぽ" - "ポ" =  0 , 2 , ffffffe0 ,
"ま" - "マ" =  0 , 2 , ffffffe0 ,
"み" - "ミ" =  0 , 2 , ffffffe0 ,
"む" - "ム" =  0 , 1 , 20 ,
"め" - "メ" =  0 , 1 , 20 ,
"も" - "モ" =  0 , 1 , 20 ,
"ゃ" - "ャ" =  0 , 1 , 20 ,
"や" - "ヤ" =  0 , 1 , 20 ,
"ゅ" - "ュ" =  0 , 1 , 20 ,
"ゆ" - "ユ" =  0 , 1 , 20 ,
"ょ" - "ョ" =  0 , 1 , 20 ,
"よ" - "ヨ" =  0 , 1 , 20 ,
"ら" - "ラ" =  0 , 1 , 20 ,
"り" - "リ" =  0 , 1 , 20 ,
"る" - "ル" =  0 , 1 , 20 ,
"れ" - "レ" =  0 , 1 , 20 ,
"ろ" - "ロ" =  0 , 1 , 20 ,
"ゎ" - "ヮ" =  0 , 1 , 20 ,
"わ" - "ワ" =  0 , 1 , 20 ,
"ゐ" - "ヰ" =  0 , 1 , 20 ,
"ゑ" - "ヱ" =  0 , 1 , 20 ,
"を" - "ヲ" =  0 , 1 , 20 ,
"ん" - "ン" =  0 , 1 , 20 ,
"　" - "ヴ" =  0 , 3 , 34 ,
"　" - "ヵ" =  0 , 3 , 35 ,
"　" - "ヶ" =  0 , 3 , 36 ,
"ゝ" - "ヽ" =  0 , 1 , 20 ,
"ゞ" - "ヾ" =  0 , 1 , 20 ,
"" - "?" =  0 , ffffffce , ffffff80 ,