src/src/utf8.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) Jeremy Harris 2015 */
   6 /* See the file NOTICE for conditions of use and distribution. */
   7
   8
   9 #include "exim.h"
  10
  11 #ifdef EXPERIMENTAL_INTERNATIONAL
  12
  13 #include <idna.h>
  14 #include <punycode.h>
  15 #include <stringprep.h>
  16
  17 BOOL
  18 string_is_utf8(const uschar * s)
  19 {
  20 uschar c;
  21 while ((c = *s++)) if (c & 0x80) return TRUE;
  22 return FALSE;
  23 }
  24
  25 /**************************************************/
  26 /* Domain conversions */
  27
  28 uschar *
  29 string_domain_utf8_to_alabel(const uschar * utf8, uschar ** err)
  30 {
  31 uschar * s1;
  32 uschar * s;
  33 int rc;
  34
  35 s = US stringprep_utf8_nfkc_normalize(CCS utf8, -1);
  36 if (  (rc = idna_to_ascii_8z(CCS s, CSS &s1, IDNA_USE_STD3_ASCII_RULES))
  37    != IDNA_SUCCESS)
  38   {
  39   free(s);
  40   if (err) *err = US idna_strerror(rc);
  41   return NULL;
  42   }
  43 free(s);
  44 s = string_copy(s1);
  45 free(s1);
  46 return s;
  47 }
  48
  49
  50
  51 uschar *
  52 string_domain_alabel_to_utf8(const uschar * alabel, uschar ** err)
  53 {
  54 uschar * s1;
  55 uschar * s;
  56 int rc;
  57 if (  (rc = idna_to_unicode_8z8z(CCS alabel, CSS &s1, IDNA_USE_STD3_ASCII_RULES))
  58    != IDNA_SUCCESS)
  59   {
  60   if (err) *err = US idna_strerror(rc);
  61   return NULL;
  62   }
  63 s = string_copy(s1);
  64 free(s1);
  65 return s;
  66 }
  67
  68 /**************************************************/
  69 /* localpart conversions */
  70
  71
  72 uschar *
  73 string_localpart_utf8_to_alabel(const uschar * utf8, uschar ** err)
  74 {
  75 size_t ucs4_len;
  76 punycode_uint * p = (punycode_uint *) stringprep_utf8_to_ucs4(CCS utf8, -1, &ucs4_len);
  77 size_t p_len = ucs4_len*4;      /* this multiplier is pure guesswork */
  78 uschar * res = store_get(p_len+5);
  79 int rc;
  80
  81 DEBUG(D_expand) debug_printf("l_u2a: ulen %d  plen %d\n", ucs4_len, p_len);
  82 DEBUG(D_expand) for (rc = 0; rc < ucs4_len; rc++) debug_printf("%08x ", p[rc]);
  83
  84 res[0] = 'x'; res[1] = 'n'; res[2] = res[3] = '-';
  85
  86 if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, res+4)) != PUNYCODE_SUCCESS)
  87   {
  88   DEBUG(D_expand) debug_printf("l_u2a: bad '%s'\n", punycode_strerror(rc));
  89   free(p);
  90   if (err) *err = US punycode_strerror(rc);
  91   return NULL;
  92   }
  93 DEBUG(D_expand) debug_printf("l_u2a: plen %d\n", p_len);
  94 p_len += 4;
  95 DEBUG(D_expand) for (rc = 0; rc < p_len; rc++) debug_printf("%02x ", res[rc]);
  96 DEBUG(D_expand) debug_printf("\n");
  97 free(p);
  98 res[p_len] = '\0';
  99 return res;
 100 }
 101
 102
 103 uschar *
 104 string_localpart_alabel_to_utf8(const uschar * alabel, uschar ** err)
 105 {
 106 size_t p_len = strlen(alabel);
 107 punycode_uint * p;
 108 uschar * s;
 109 uschar * res;
 110 int rc;
 111
 112 if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-')
 113   {
 114   if (err) *err = US"bad alabel prefix";
 115   return NULL;
 116   }
 117 p_len -= 4;
 118 DEBUG(D_expand) debug_printf("l_a2u: plen %d\n", p_len);
 119
 120 p = (punycode_uint *) store_get((p_len+1) * sizeof(*p));
 121
 122 if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUCCESS)
 123   {
 124   if (err) *err = US punycode_strerror(rc);
 125   return NULL;
 126   }
 127 DEBUG(D_expand) debug_printf("l_a2u: dlen %d\n", p_len);
 128
 129 s = stringprep_ucs4_to_utf8(p, p_len, NULL, &p_len);
 130 res = string_copyn(s, p_len);
 131 free(s);
 132 return res;
 133 }
 134
 135
 136 #endif  /* whole file */
 137
 138 /* vi: aw ai sw=2
 139 */
 140 /* End of utf8.c */