[BACK]Return to utf8.h CVS log [TXT] [DIR] Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / libiconv / lib

Diff for /embedaddon/libiconv/lib/utf8.h between versions 1.1.1.1 and 1.1.1.2

version 1.1.1.1, 2012/02/21 22:57:48 version 1.1.1.2, 2021/03/17 13:38:46
Line 1 Line 1
 /*  /*
 * Copyright (C) 1999-2001, 2004 Free Software Foundation, Inc. * Copyright (C) 1999-2001, 2004, 2016 Free Software Foundation, Inc.
  * This file is part of the GNU LIBICONV Library.   * This file is part of the GNU LIBICONV Library.
  *   *
  * The GNU LIBICONV Library is free software; you can redistribute it   * The GNU LIBICONV Library is free software; you can redistribute it
Line 14 Line 14
  *   *
  * You should have received a copy of the GNU Library General Public   * You should have received a copy of the GNU Library General Public
  * License along with the GNU LIBICONV Library; see the file COPYING.LIB.   * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
 * If not, write to the Free Software Foundation, Inc., 51 Franklin Street, * If not, see <https://www.gnu.org/licenses/>.
 * Fifth Floor, Boston, MA 02110-1301, USA. 
  */   */
   
 /*  /*
Line 25 Line 24
 /* Specification: RFC 3629 */  /* Specification: RFC 3629 */
   
 static int  static int
utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n)
 {  {
   unsigned char c = s[0];    unsigned char c = s[0];
   
Line 46  utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned  Line 45  utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned 
     if (n < 3)      if (n < 3)
       return RET_TOOFEW(0);        return RET_TOOFEW(0);
     if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40      if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40
          && (c >= 0xe1 || s[1] >= 0xa0)))          && (c >= 0xe1 || s[1] >= 0xa0)
           && (c != 0xed || s[1] < 0xa0)))
       return RET_ILSEQ;        return RET_ILSEQ;
     *pwc = ((ucs4_t) (c & 0x0f) << 12)      *pwc = ((ucs4_t) (c & 0x0f) << 12)
            | ((ucs4_t) (s[1] ^ 0x80) << 6)             | ((ucs4_t) (s[1] ^ 0x80) << 6)
Line 57  utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned  Line 57  utf8_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned 
       return RET_TOOFEW(0);        return RET_TOOFEW(0);
     if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40      if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40
           && (s[3] ^ 0x80) < 0x40            && (s[3] ^ 0x80) < 0x40
          && (c >= 0xf1 || s[1] >= 0x90)))          && (c >= 0xf1 || s[1] >= 0x90)
           && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90))))
       return RET_ILSEQ;        return RET_ILSEQ;
     *pwc = ((ucs4_t) (c & 0x07) << 18)      *pwc = ((ucs4_t) (c & 0x07) << 18)
            | ((ucs4_t) (s[1] ^ 0x80) << 12)             | ((ucs4_t) (s[1] ^ 0x80) << 12)
            | ((ucs4_t) (s[2] ^ 0x80) << 6)             | ((ucs4_t) (s[2] ^ 0x80) << 6)
            | (ucs4_t) (s[3] ^ 0x80);             | (ucs4_t) (s[3] ^ 0x80);
     return 4;      return 4;
   } else if (c < 0xfc && sizeof(ucs4_t)*8 >= 32) {  
     if (n < 5)  
       return RET_TOOFEW(0);  
     if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40  
           && (s[3] ^ 0x80) < 0x40 && (s[4] ^ 0x80) < 0x40  
           && (c >= 0xf9 || s[1] >= 0x88)))  
       return RET_ILSEQ;  
     *pwc = ((ucs4_t) (c & 0x03) << 24)  
            | ((ucs4_t) (s[1] ^ 0x80) << 18)  
            | ((ucs4_t) (s[2] ^ 0x80) << 12)  
            | ((ucs4_t) (s[3] ^ 0x80) << 6)  
            | (ucs4_t) (s[4] ^ 0x80);  
     return 5;  
   } else if (c < 0xfe && sizeof(ucs4_t)*8 >= 32) {  
     if (n < 6)  
       return RET_TOOFEW(0);  
     if (!((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40  
           && (s[3] ^ 0x80) < 0x40 && (s[4] ^ 0x80) < 0x40  
           && (s[5] ^ 0x80) < 0x40  
           && (c >= 0xfd || s[1] >= 0x84)))  
       return RET_ILSEQ;  
     *pwc = ((ucs4_t) (c & 0x01) << 30)  
            | ((ucs4_t) (s[1] ^ 0x80) << 24)  
            | ((ucs4_t) (s[2] ^ 0x80) << 18)  
            | ((ucs4_t) (s[3] ^ 0x80) << 12)  
            | ((ucs4_t) (s[4] ^ 0x80) << 6)  
            | (ucs4_t) (s[5] ^ 0x80);  
     return 6;  
   } else    } else
     return RET_ILSEQ;      return RET_ILSEQ;
 }  }
   
 static int  static int
utf8_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n) /* n == 0 is acceptable */utf8_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n) /* n == 0 is acceptable */
 {  {
   int count;    int count;
   if (wc < 0x80)    if (wc < 0x80)
     count = 1;      count = 1;
   else if (wc < 0x800)    else if (wc < 0x800)
     count = 2;      count = 2;
  else if (wc < 0x10000)  else if (wc < 0x10000) {
    count = 3;    if (wc < 0xd800 || wc >= 0xe000)
  else if (wc < 0x200000)      count = 3;
     else
       return RET_ILUNI;
   } else if (wc < 0x110000)
     count = 4;      count = 4;
   else if (wc < 0x4000000)  
     count = 5;  
   else if (wc <= 0x7fffffff)  
     count = 6;  
   else    else
     return RET_ILUNI;      return RET_ILUNI;
   if (n < count)    if (n < count)
     return RET_TOOSMALL;      return RET_TOOSMALL;
   switch (count) { /* note: code falls through cases! */    switch (count) { /* note: code falls through cases! */
     case 6: r[5] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x4000000;  
     case 5: r[4] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x200000;  
     case 4: r[3] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x10000;      case 4: r[3] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x10000;
     case 3: r[2] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x800;      case 3: r[2] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x800;
     case 2: r[1] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0xc0;      case 2: r[1] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0xc0;
Removed from v.1.1.1.1  
changed lines
  Added in v.1.1.1.2

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>