smtp input
[exim.git] / 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 res[0] = 'x'; res[1] = 'n'; res[2] = res[3] = '-';
82
83 if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, res+4)) != PUNYCODE_SUCCESS)
84 {
85 DEBUG(D_expand) debug_printf("l_u2a: bad '%s'\n", punycode_strerror(rc));
86 free(p);
87 if (err) *err = US punycode_strerror(rc);
88 return NULL;
89 }
90 p_len += 4;
91 free(p);
92 res[p_len] = '\0';
93 return res;
94 }
95
96
97 uschar *
98 string_localpart_alabel_to_utf8(const uschar * alabel, uschar ** err)
99 {
100 size_t p_len = strlen(alabel);
101 punycode_uint * p;
102 uschar * s;
103 uschar * res;
104 int rc;
105
106 if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-')
107 {
108 if (err) *err = US"bad alabel prefix";
109 return NULL;
110 }
111
112 p_len -= 4;
113 p = (punycode_uint *) store_get((p_len+1) * sizeof(*p));
114
115 if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUCCESS)
116 {
117 if (err) *err = US punycode_strerror(rc);
118 return NULL;
119 }
120
121 s = stringprep_ucs4_to_utf8(p, p_len, NULL, &p_len);
122 res = string_copyn(s, p_len);
123 free(s);
124 return res;
125 }
126
127
128 #endif /* whole file */
129
130 /* vi: aw ai sw=2
131 */
132 /* End of utf8.c */