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 SUPPORT_I18N
  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 /* the *err string pointer should be null before the call */
  28
  29 uschar *
  30 string_domain_utf8_to_alabel(const uschar * utf8, uschar ** err)
  31 {
  32 uschar * s1;
  33 uschar * s;
  34 int rc;
  35
  36 s = US stringprep_utf8_nfkc_normalize(CCS utf8, -1);
  37 if (  (rc = idna_to_ascii_8z(CCS s, CSS &s1, IDNA_ALLOW_UNASSIGNED))
  38    != IDNA_SUCCESS)
  39   {
  40   free(s);
  41   if (err) *err = US idna_strerror(rc);
  42   return NULL;
  43   }
  44 free(s);
  45 s = string_copy(s1);
  46 free(s1);
  47 return s;
  48 }
  49
  50
  51
  52 uschar *
  53 string_domain_alabel_to_utf8(const uschar * alabel, uschar ** err)
  54 {
  55 uschar * s1;
  56 uschar * s;
  57 int rc;
  58
  59 if (  (rc = idna_to_unicode_8z8z(CCS alabel, CSS &s1, IDNA_USE_STD3_ASCII_RULES))
  60    != IDNA_SUCCESS)
  61   {
  62   if (err) *err = US idna_strerror(rc);
  63   return NULL;
  64   }
  65 s = string_copy(s1);
  66 free(s1);
  67 return s;
  68 }
  69
  70 /**************************************************/
  71 /* localpart conversions */
  72 /* the *err string pointer should be null before the call */
  73
  74
  75 uschar *
  76 string_localpart_utf8_to_alabel(const uschar * utf8, uschar ** err)
  77 {
  78 size_t ucs4_len;
  79 punycode_uint * p;
  80 size_t p_len;
  81 uschar * res;
  82 int rc;
  83
  84 if (!string_is_utf8(utf8)) return string_copy(utf8);
  85
  86 p = (punycode_uint *) stringprep_utf8_to_ucs4(CCS utf8, -1, &ucs4_len);
  87 p_len = ucs4_len*4;     /* this multiplier is pure guesswork */
  88 res = store_get(p_len+5);
  89
  90 res[0] = 'x'; res[1] = 'n'; res[2] = res[3] = '-';
  91
  92 if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, CS res+4)) != PUNYCODE_SUCCESS)
  93   {
  94   DEBUG(D_expand) debug_printf("l_u2a: bad '%s'\n", punycode_strerror(rc));
  95   free(p);
  96   if (err) *err = US punycode_strerror(rc);
  97   return NULL;
  98   }
  99 p_len += 4;
 100 free(p);
 101 res[p_len] = '\0';
 102 return res;
 103 }
 104
 105
 106 uschar *
 107 string_localpart_alabel_to_utf8(const uschar * alabel, uschar ** err)
 108 {
 109 size_t p_len = Ustrlen(alabel);
 110 punycode_uint * p;
 111 uschar * s;
 112 uschar * res;
 113 int rc;
 114
 115 if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-')
 116   {
 117   if (err) *err = US"bad alabel prefix";
 118   return NULL;
 119   }
 120
 121 p_len -= 4;
 122 p = (punycode_uint *) store_get((p_len+1) * sizeof(*p));
 123
 124 if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUCCESS)
 125   {
 126   if (err) *err = US punycode_strerror(rc);
 127   return NULL;
 128   }
 129
 130 s = US stringprep_ucs4_to_utf8(p, p_len, NULL, &p_len);
 131 res = string_copyn(s, p_len);
 132 free(s);
 133 return res;
 134 }
 135
 136
 137 /**************************************************/
 138 /* whole address conversion */
 139 /* the *err string pointer should be null before the call */
 140
 141 uschar *
 142 string_address_utf8_to_alabel(const uschar * utf8, uschar ** err)
 143 {
 144 const uschar * s;
 145 uschar * l;
 146 uschar * d;
 147
 148 if (!*utf8) return string_copy(utf8);
 149
 150 DEBUG(D_expand) debug_printf("addr from utf8 <%s>", utf8);
 151
 152 for (s = utf8; *s; s++)
 153   if (*s == '@')
 154     {
 155     l = string_copyn(utf8, s - utf8);
 156     if (  (l = string_localpart_utf8_to_alabel(l, err), err && *err)
 157        || (d = string_domain_utf8_to_alabel(++s, err),  err && *err)
 158        )
 159       return NULL;
 160     l = string_sprintf("%s@%s", l, d);
 161     DEBUG(D_expand) debug_printf(" -> <%s>\n", l);
 162     return l;
 163     }
 164
 165 l =  string_localpart_utf8_to_alabel(utf8, err);
 166 DEBUG(D_expand) debug_printf(" -> <%s>\n", l);
 167 return l;
 168 }
 169
 170
 171
 172 /*************************************************
 173 *         Report the library versions.           *
 174 *************************************************/
 175
 176 /* See a description in tls-openssl.c for an explanation of why this exists.
 177
 178 Arguments:   a FILE* to print the results to
 179 Returns:     nothing
 180 */
 181
 182 void
 183 utf8_version_report(FILE *f)
 184 {
 185 fprintf(f, "Library version: IDN: Compile: %s\n"
 186            "                      Runtime: %s\n",
 187         STRINGPREP_VERSION,
 188         stringprep_check_version(NULL));
 189 }
 190
 191 #endif  /* whole file */
 192
 193 /* vi: aw ai sw=2
 194 */
 195 /* End of utf8.c */