src/src/utf8.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) Jeremy Harris 2015 - 2018 */
   6 /* See the file NOTICE for conditions of use and distribution. */
   7
   8
   9 #include "exim.h"
  10
  11 #ifdef SUPPORT_I18N
  12
  13 #ifdef SUPPORT_I18N_2008
  14 # include <idn2.h>
  15 #else
  16 # include <idna.h>
  17 #endif
  18
  19 #include <punycode.h>
  20 #include <stringprep.h>
  21
  22 static uschar *
  23 string_localpart_alabel_to_utf8_(const uschar * alabel, uschar ** err);
  24
  25 /**************************************************/
  26
  27 BOOL
  28 string_is_utf8(const uschar * s)
  29 {
  30 uschar c;
  31 if (s) while ((c = *s++)) if (c & 0x80) return TRUE;
  32 return FALSE;
  33 }
  34
  35 static BOOL
  36 string_is_alabel(const uschar * s)
  37 {
  38 return s[0] == 'x' && s[1] == 'n' && s[2] == '-' && s[3] == '-';
  39 }
  40
  41 /**************************************************/
  42 /* Domain conversions.
  43 The *err string pointer should be null before the call
  44
  45 Return NULL for error, with optional errstr pointer filled in
  46 */
  47
  48 uschar *
  49 string_domain_utf8_to_alabel(const uschar * utf8, uschar ** err)
  50 {
  51 uschar * s1, * s;
  52 int rc;
  53
  54 #ifdef SUPPORT_I18N_2008
  55 /* Avoid lowercasing plain-ascii domains */
  56 if (!string_is_utf8(utf8))
  57   return string_copy(utf8);
  58
  59 /* Only lowercase is accepted by the library call.  A pity since we lose
  60 any mixed-case annotation.  This does not really matter for a domain. */
  61   {
  62   uschar c;
  63   for (s1 = s = US utf8; (c = *s1); s1++) if (!(c & 0x80) && isupper(c))
  64     {
  65     s = string_copy(utf8);
  66     for (s1 = s + (s1 - utf8); (c = *s1); s1++) if (!(c & 0x80) && isupper(c))
  67       *s1 = tolower(c);
  68     break;
  69     }
  70   }
  71 if ((rc = idn2_lookup_u8((const uint8_t *) s, &s1, IDN2_NFC_INPUT)) != IDN2_OK)
  72   {
  73   if (err) *err = US idn2_strerror(rc);
  74   return NULL;
  75   }
  76 #else
  77 s = US stringprep_utf8_nfkc_normalize(CCS utf8, -1);
  78 if (  (rc = idna_to_ascii_8z(CCS s, CSS &s1, IDNA_ALLOW_UNASSIGNED))
  79    != IDNA_SUCCESS)
  80   {
  81   free(s);
  82   if (err) *err = US idna_strerror(rc);
  83   return NULL;
  84   }
  85 free(s);
  86 #endif
  87 s = string_copy(s1);
  88 free(s1);
  89 return s;
  90 }
  91
  92
  93
  94 uschar *
  95 string_domain_alabel_to_utf8(const uschar * alabel, uschar ** err)
  96 {
  97 #ifdef SUPPORT_I18N_2008
  98 const uschar * label;
  99 int sep = '.';
 100 gstring * g = NULL;
 101
 102 while (label = string_nextinlist(&alabel, &sep, NULL, 0))
 103   if (  string_is_alabel(label)
 104      && !(label = string_localpart_alabel_to_utf8_(label, err))
 105      )
 106     return NULL;
 107   else
 108     g = string_append_listele(g, '.', label);
 109 return string_from_gstring(g);
 110
 111 #else
 112
 113 uschar * s1, * s;
 114 int rc;
 115
 116 if (  (rc = idna_to_unicode_8z8z(CCS alabel, CSS &s1, IDNA_USE_STD3_ASCII_RULES))
 117    != IDNA_SUCCESS)
 118   {
 119   if (err) *err = US idna_strerror(rc);
 120   return NULL;
 121   }
 122 s = string_copy(s1);
 123 free(s1);
 124 return s;
 125 #endif
 126 }
 127
 128 /**************************************************/
 129 /* localpart conversions */
 130 /* the *err string pointer should be null before the call */
 131
 132
 133 uschar *
 134 string_localpart_utf8_to_alabel(const uschar * utf8, uschar ** err)
 135 {
 136 size_t ucs4_len;
 137 punycode_uint * p;
 138 size_t p_len;
 139 uschar * res;
 140 int rc;
 141
 142 if (!string_is_utf8(utf8)) return string_copy(utf8);
 143
 144 p = (punycode_uint *) stringprep_utf8_to_ucs4(CCS utf8, -1, &ucs4_len);
 145 p_len = ucs4_len*4;     /* this multiplier is pure guesswork */
 146 res = store_get(p_len+5, is_tainted(utf8));
 147
 148 res[0] = 'x'; res[1] = 'n'; res[2] = res[3] = '-';
 149
 150 if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, CS res+4)) != PUNYCODE_SUCCESS)
 151   {
 152   DEBUG(D_expand) debug_printf("l_u2a: bad '%s'\n", punycode_strerror(rc));
 153   free(p);
 154   if (err) *err = US punycode_strerror(rc);
 155   return NULL;
 156   }
 157 p_len += 4;
 158 free(p);
 159 res[p_len] = '\0';
 160 return res;
 161 }
 162
 163
 164 static uschar *
 165 string_localpart_alabel_to_utf8_(const uschar * alabel, uschar ** err)
 166 {
 167 size_t p_len;
 168 punycode_uint * p;
 169 int rc;
 170 uschar * s, * res;
 171
 172 DEBUG(D_expand) debug_printf("l_a2u: '%s'\n", alabel);
 173 alabel += 4;
 174 p_len = Ustrlen(alabel);
 175 p = store_get((p_len+1) * sizeof(*p), is_tainted(alabel));
 176
 177 if ((rc = punycode_decode(p_len, CCS alabel, &p_len, p, NULL)) != PUNYCODE_SUCCESS)
 178   {
 179   if (err) *err = US punycode_strerror(rc);
 180   return NULL;
 181   }
 182
 183 s = US stringprep_ucs4_to_utf8(p, p_len, NULL, &p_len);
 184 res = string_copyn(s, p_len);
 185 free(s);
 186 return res;
 187 }
 188
 189
 190 uschar *
 191 string_localpart_alabel_to_utf8(const uschar * alabel, uschar ** err)
 192 {
 193 if (string_is_alabel(alabel))
 194   return string_localpart_alabel_to_utf8_(alabel, err);
 195
 196 if (err) *err = US"bad alabel prefix";
 197 return NULL;
 198 }
 199
 200
 201 /**************************************************/
 202 /* Whole address conversion.
 203 The *err string pointer should be null before the call.
 204
 205 Return NULL on error, with (optional) errstring pointer filled in
 206 */
 207
 208 uschar *
 209 string_address_utf8_to_alabel(const uschar * utf8, uschar ** err)
 210 {
 211 uschar * l, * d;
 212
 213 if (!*utf8) return string_copy(utf8);
 214
 215 DEBUG(D_expand) debug_printf("addr from utf8 <%s>", utf8);
 216
 217 for (const uschar * s = utf8; *s; s++)
 218   if (*s == '@')
 219     {
 220     l = string_copyn(utf8, s - utf8);
 221     if (  !(l = string_localpart_utf8_to_alabel(l, err))
 222        || !(d = string_domain_utf8_to_alabel(++s, err))
 223        )
 224       return NULL;
 225     l = string_sprintf("%s@%s", l, d);
 226     DEBUG(D_expand) debug_printf(" -> <%s>\n", l);
 227     return l;
 228     }
 229
 230 l =  string_localpart_utf8_to_alabel(utf8, err);
 231 DEBUG(D_expand) debug_printf(" -> <%s>\n", l);
 232 return l;
 233 }
 234
 235
 236
 237 /*************************************************
 238 *         Report the library versions.           *
 239 *************************************************/
 240
 241 /* See a description in tls-openssl.c for an explanation of why this exists.
 242
 243 Arguments:   a FILE* to print the results to
 244 Returns:     nothing
 245 */
 246
 247 void
 248 utf8_version_report(FILE *f)
 249 {
 250 #ifdef SUPPORT_I18N_2008
 251 fprintf(f, "Library version: IDN2: Compile: %s\n"
 252            "                       Runtime: %s\n",
 253         IDN2_VERSION,
 254         idn2_check_version(NULL));
 255 fprintf(f, "Library version: Stringprep: Compile: %s\n"
 256            "                             Runtime: %s\n",
 257         STRINGPREP_VERSION,
 258         stringprep_check_version(NULL));
 259 #else
 260 fprintf(f, "Library version: IDN: Compile: %s\n"
 261            "                      Runtime: %s\n",
 262         STRINGPREP_VERSION,
 263         stringprep_check_version(NULL));
 264 #endif
 265 }
 266
 267 #endif  /* whole file */
 268
 269 /* vi: aw ai sw=2
 270 */
 271 /* End of utf8.c */