cjkctype.cpp

This software performs code conversion of Chinese characters, including GB2312/GBK and BIG5. It a

//          strspn (), strcspn ()         //
////////////////////////////////////////////



// string2 偵懏偝側偄暥帤偑 string1 偱嵟弶偵弌尰偡傞埵抲傪曉偡乮MULTIBYTE斉乯
// 尒偮偐傜側偗傟偽枛旜傪曉偡
size_t mbsspn_cp ( const char *string1, const char *string2, int codepage ) {

   int c ;
   const char *current ;

   for ( current = string1 ; *current ; current ++ ) {

      int islead = ismbblead_cp ( (unsigned char) *current, codepage ) ;
      int issurrogate = ismbbsurrogate_cp ( (unsigned char) *(current + 1 ), codepage ) ;

      if ( islead ) {
         if ( issurrogate ) {
            int c1 = ( (unsigned char) *( current + 0 ) << 8 ) | (unsigned char) *( current + 1 ) ;
            int c2 = ( (unsigned char) *( current + 2 ) << 8 ) | (unsigned char) *( current + 3 ) ;
            c = ( c1 << 16 ) | c2 ;
         }
         else {
            c = ( (unsigned char) *current << 8 ) | (unsigned char) *( current + 1 ) ;
         }
      }
      else {
         c = *current ;
      }

      if ( ! mbschr_cp ( string2, c, codepage ) ) break ;

      if ( islead ) {
         if ( issurrogate ) {
            current ++ ;
            if ( *( current + 1 ) ) current ++ ;
            if ( *( current + 1 ) ) current ++ ;
         }
         else {
            if ( *( current + 1 ) ) current ++ ;
         }
      }
   }

   return current - string1 ;
}



// string2 偵懏偡暥帤偑 string1 偱嵟弶偵弌尰偡傞埵抲傪曉偡乮MULTIBYTE斉乯
// 尒偮偐傜側偗傟偽枛旜傪曉偡
size_t mbscspn_cp ( const char *string1, const char *string2, int codepage ) {

   int c ;
   const char *current ;

   for ( current = string1 ; *current ; current ++ ) {

      int islead = ismbblead_cp ( (unsigned char) *current, codepage ) ;
      int issurrogate = ismbbsurrogate_cp ( (unsigned char) *(current + 1 ), codepage ) ;

      if ( islead ) {
         if ( issurrogate ) {
            int c1 = ( (unsigned char) *( current + 0 ) << 8 ) | (unsigned char) *( current + 1 ) ;
            int c2 = ( (unsigned char) *( current + 2 ) << 8 ) | (unsigned char) *( current + 3 ) ;
            c = ( c1 << 16 ) | c2 ;
         }
         else {
            c = ( (unsigned char) *current << 8 ) | (unsigned char) *( current + 1 ) ;
         }
      }
      else {
         c = *current ;
      }

      if ( mbschr_cp ( string2, c, codepage ) ) break ;

      if ( islead ) {
         if ( issurrogate ) {
            current ++ ;
            if ( *( current + 1 ) ) current ++ ;
            if ( *( current + 1 ) ) current ++ ;
         }
         else {
            if ( *( current + 1 ) ) current ++ ;
         }
      }
   }

   return current - string1 ;
}



////////////////////////////////////////////
//         strspnp (), strpbrk ()         //
////////////////////////////////////////////



// string2 偵懏偝側偄暥帤偑 string1 偱嵟弶偵弌尰偡傞埵抲傪曉偡乮MULTIBYTE斉乯
// 尒偮偐傜側偗傟偽 NULL 傪曉偡
char *mbsspnp_cp ( const char *string1, const char *string2, int codepage ) {
   const char *result = string1 + mbsspn_cp ( string1, string2, codepage ) ;
   return *result ? (char*) result : NULL ;
}



// string2 偵懏偡暥帤偑 string1 偱嵟弶偵弌尰偡傞埵抲傪曉偡乮MULTIBYTE斉乯
// 尒偮偐傜側偗傟偽 NULL 傪曉偡
char *mbspbrk_cp ( const char *string1, const char *string2, int codepage ) {
   const char *result = string1 + mbscspn_cp ( string1, string2, codepage ) ;
   return *result ? (char*) result : NULL ;
}



////////////////////////////////////////////
//               strlwr ()                //
////////////////////////////////////////////



// 暥帤楍傪彫暥帤偵乮MULTIBYTE斉乯
char *mbslwr_cp ( char *string, int codepage ) {

   for ( char *current = string ; *current ; current ++ ) {
      if ( *current >= 'A' && *current <= 'Z' ) *current -= 'A' - 'a' ;
      if ( ismbblead_cp ( (unsigned char) *current, codepage ) && *( current + 1 ) ) current ++ ;
   }

   return string ;
}



////////////////////////////////////////////
//               strupr ()                //
////////////////////////////////////////////



// 暥帤楍傪戝暥帤偵乮MULTIBYTE斉乯
char *mbsupr_cp ( char *string, int codepage ) {

   for ( char *current = string ; *current ; current ++ ) {
      if ( *current >= 'a' && *current <= 'z' ) *current += 'A' - 'a' ;
      if ( ismbblead_cp ( (unsigned char) *current, codepage ) && *( current + 1 ) ) current ++ ;
   }

   return string ;
}



////////////////////////////////////////////
//               strrev ()                //
////////////////////////////////////////////



// 暥帤楍傪媡揮乮MULTIBYTE斉乯
char *mbsrev_cp ( char *string, int codepage ) {

   for ( char *current = string ; *current ; current ++ ) {
      if ( ismbblead_cp ( (unsigned char) *current, codepage ) ) {
         if ( ismbbsurrogate_cp ( (unsigned char) *( current + 1 ), codepage ) ) {
            if ( *( current + 2 ) && *( current + 3 ) ) {
               char c0 = *( current + 0 ) ;
               char c1 = *( current + 1 ) ;
               char c2 = *( current + 2 ) ;
               char c3 = *( current + 3 ) ;
               *( current + 3 ) = c0 ;
               *( current + 2 ) = c1 ;
               *( current + 1 ) = c2 ;
               *( current + 0 ) = c3 ;
               current ++ ;
               current ++ ;
               current ++ ;
               continue ;
            }
            else {
               *current = 0 ;
               break ;
            }
         }
         else {
            if ( *( current + 1 ) ) {
               char c0 = *( current + 0 ) ;
               char c1 = *( current + 1 ) ;
               *( current + 1 ) = c0 ;
               *( current + 0 ) = c1 ;
               current ++ ;
               continue ;
            }
            else {
               *current = 0 ;
               break ;
            }
         }
      }
   }

   return strrev ( string ) ;
}



////////////////////////////////////////////
//               strtrunc ()              //
////////////////////////////////////////////



// 暥帤楍傪抁弅乮MULTIBYTE斉乯
// length 偼嵟戝帤悢乮廔抂僰儖暥帤傪娷傑側偄乯
char *mbstrunc_cp ( char *string, size_t length, int codepage ) {

   char *end = string + length ;

   for ( char *current = string ; *current ; current ++ ) {
      if ( ismbblead_cp ( (unsigned char) *current, codepage ) ) {
         if ( ismbbsurrogate_cp ( (unsigned char) *( current + 1 ), codepage ) ) {
            if ( current >= end - 3 || ! *( current + 2 ) || ! *( current + 3 ) ) {
               *current = 0 ;
               break ;
            }
            current ++ ;
            current ++ ;
            current ++ ;
         }
         else {
            if ( current >= end - 1 || ! *( current + 1 ) ) {
               *current = 0 ;
               break ;
            }
            current ++ ;
         }
      }
      else {
         if ( current >= end ) {
            *current = 0 ;
            break ;
         }
      }
   }

   return string ;
}



////////////////////////////////////////////
//                strlen ()               //
////////////////////////////////////////////



// 暥帤悢傪曉偡乮MULTIBYTE斉乯
// 俀僶僀僩暥帤偼侾暥帤偲偟偰悢偊傞
size_t mbslen_cp ( const char *string, int codepage ) {

   size_t n ;

   for ( n = 0 ; *string ; n ++, string ++ ) {
      if ( ismbblead_cp ( (unsigned char) *string, codepage ) ) {
         if ( ismbbsurrogate_cp ( (unsigned char) *( string + 1 ), codepage ) ) {
            string ++ ;
            if ( *( string + 1 ) ) string ++ ;
            if ( *( string + 1 ) ) string ++ ;
         }
         else {
            if ( *( string + 1 ) ) string ++ ;
         }
      }
   }

   return n ;
}



////////////////////////////////////////////
//                a/wdup ()               //