blob: 7e73d77596622debe26565169f0a4e4a23076c84 [file] [log] [blame]
vlmfa67ddc2004-06-03 03:38:44 +00001/*-
vlm8d3e6d52006-09-17 01:27:44 +00002 * Copyright (c) 2003, 2004, 2006 Lev Walkin <vlm@lionet.info>.
3 * All rights reserved.
vlmfa67ddc2004-06-03 03:38:44 +00004 * Redistribution and modifications are permitted subject to BSD license.
5 */
vlm39ba4c42004-09-22 16:06:28 +00006#include <asn_internal.h>
vlmfa67ddc2004-06-03 03:38:44 +00007#include <UTF8String.h>
8
9/*
10 * UTF8String basic type description.
11 */
vlmef6355b2004-09-29 13:26:15 +000012static ber_tlv_tag_t asn_DEF_UTF8String_tags[] = {
vlm72425de2004-09-13 08:31:01 +000013 (ASN_TAG_CLASS_UNIVERSAL | (12 << 2)), /* [UNIVERSAL 12] IMPLICIT ...*/
14 (ASN_TAG_CLASS_UNIVERSAL | (4 << 2)), /* ... OCTET STRING */
vlmfa67ddc2004-06-03 03:38:44 +000015};
vlmef6355b2004-09-29 13:26:15 +000016asn_TYPE_descriptor_t asn_DEF_UTF8String = {
vlmfa67ddc2004-06-03 03:38:44 +000017 "UTF8String",
vlm9de248e2004-10-20 15:50:55 +000018 "UTF8String",
vlm39ba4c42004-09-22 16:06:28 +000019 OCTET_STRING_free,
20 UTF8String_print,
vlmfa67ddc2004-06-03 03:38:44 +000021 UTF8String_constraint, /* Check for invalid codes, etc. */
22 OCTET_STRING_decode_ber, /* Implemented in terms of OCTET STRING */
vlm9de248e2004-10-20 15:50:55 +000023 OCTET_STRING_encode_der,
24 OCTET_STRING_decode_xer_utf8,
25 OCTET_STRING_encode_xer_utf8,
vlm86380d32006-10-09 12:07:58 +000026 OCTET_STRING_decode_uper,
27 OCTET_STRING_encode_uper,
vlmfa67ddc2004-06-03 03:38:44 +000028 0, /* Use generic outmost tag fetcher */
vlmef6355b2004-09-29 13:26:15 +000029 asn_DEF_UTF8String_tags,
30 sizeof(asn_DEF_UTF8String_tags)
31 / sizeof(asn_DEF_UTF8String_tags[0]) - 1,
32 asn_DEF_UTF8String_tags,
33 sizeof(asn_DEF_UTF8String_tags)
34 / sizeof(asn_DEF_UTF8String_tags[0]),
vlm337167e2005-11-26 11:25:14 +000035 0, /* No PER visible constraints */
vlme413c122004-08-20 13:23:42 +000036 0, 0, /* No members */
vlmb42843a2004-06-05 08:17:50 +000037 0 /* No specifics */
vlmfa67ddc2004-06-03 03:38:44 +000038};
39
vlm51283292004-10-02 11:37:38 +000040/*
41 * This is the table of length expectations.
vlm1a69ff32005-03-02 13:30:25 +000042 * The second half of this table is only applicable to the long sequences.
vlm51283292004-10-02 11:37:38 +000043 */
44static int UTF8String_ht[2][16] = {
45 { /* 0x0 ... 0x7 */
46 /* 0000..0111 */
47 1, 1, 1, 1, 1, 1, 1, 1,
48 /* 1000..1011(0), 1100..1101(2), 1110(3), 1111(-1) */
49 0, 0, 0, 0, 2, 2, 3, -1 },
50 { /* 0xF0 .. 0xF7 */
51 /* 11110000..11110111 */
52 4, 4, 4, 4, 4, 4, 4, 4,
53 5, 5, 5, 5, 6, 6, -1, -1 }
vlmfa67ddc2004-06-03 03:38:44 +000054};
vlm1e10a102004-10-02 15:55:07 +000055static int32_t UTF8String_mv[7] = { 0, 0,
56 0x00000080,
57 0x00000800,
58 0x00010000,
59 0x00200000,
60 0x04000000
61};
62
63/* Internal aliases for return codes */
64#define U8E_TRUNC -1 /* UTF-8 sequence truncated */
65#define U8E_ILLSTART -2 /* Illegal UTF-8 sequence start */
66#define U8E_NOTCONT -3 /* Continuation expectation failed */
67#define U8E_NOTMIN -4 /* Not minimal length encoding */
68#define U8E_EINVAL -5 /* Invalid arguments */
vlmfa67ddc2004-06-03 03:38:44 +000069
70int
vlmef6355b2004-09-29 13:26:15 +000071UTF8String_constraint(asn_TYPE_descriptor_t *td, const void *sptr,
vlmaf68ef52006-07-13 11:19:01 +000072 asn_app_constraint_failed_f *ctfailcb, void *app_key) {
vlm1e10a102004-10-02 15:55:07 +000073 ssize_t len = UTF8String_length((const UTF8String_t *)sptr);
74 switch(len) {
75 case U8E_EINVAL:
vlm18269b62006-10-16 12:18:41 +000076 _ASN_CTFAIL(app_key, td, sptr,
vlm1e10a102004-10-02 15:55:07 +000077 "%s: value not given", td->name);
78 break;
79 case U8E_TRUNC:
vlm18269b62006-10-16 12:18:41 +000080 _ASN_CTFAIL(app_key, td, sptr,
vlm1e10a102004-10-02 15:55:07 +000081 "%s: truncated UTF-8 sequence (%s:%d)",
82 td->name, __FILE__, __LINE__);
83 break;
84 case U8E_ILLSTART:
vlm18269b62006-10-16 12:18:41 +000085 _ASN_CTFAIL(app_key, td, sptr,
vlm1e10a102004-10-02 15:55:07 +000086 "%s: UTF-8 illegal start of encoding (%s:%d)",
87 td->name, __FILE__, __LINE__);
88 break;
89 case U8E_NOTCONT:
vlm18269b62006-10-16 12:18:41 +000090 _ASN_CTFAIL(app_key, td, sptr,
vlm1e10a102004-10-02 15:55:07 +000091 "%s: UTF-8 not continuation (%s:%d)",
92 td->name, __FILE__, __LINE__);
93 break;
94 case U8E_NOTMIN:
vlm18269b62006-10-16 12:18:41 +000095 _ASN_CTFAIL(app_key, td, sptr,
vlm1e10a102004-10-02 15:55:07 +000096 "%s: UTF-8 not minimal sequence (%s:%d)",
97 td->name, __FILE__, __LINE__);
98 break;
99 }
100 return (len < 0) ? -1 : 0;
vlmfa67ddc2004-06-03 03:38:44 +0000101}
102
vlma29c5bb2004-10-23 11:20:06 +0000103static ssize_t
104UTF8String__process(const UTF8String_t *st, uint32_t *dst, size_t dstlen) {
105 size_t length;
106 uint8_t *buf = st->buf;
107 uint8_t *end = buf + st->size;
vlm8d3e6d52006-09-17 01:27:44 +0000108 uint32_t *dstend = dst + dstlen;
vlmfa67ddc2004-06-03 03:38:44 +0000109
vlma29c5bb2004-10-23 11:20:06 +0000110 for(length = 0; buf < end; length++) {
111 int ch = *buf;
112 uint8_t *cend;
113 int32_t value;
114 int want;
vlmfa67ddc2004-06-03 03:38:44 +0000115
vlma29c5bb2004-10-23 11:20:06 +0000116 /* Compute the sequence length */
117 want = UTF8String_ht[0][ch >> 4];
118 switch(want) {
119 case -1:
120 /* Second half of the table, long sequence */
121 want = UTF8String_ht[1][ch & 0x0F];
122 if(want != -1) break;
123 /* Fall through */
124 case 0:
125 return U8E_ILLSTART;
vlmfa67ddc2004-06-03 03:38:44 +0000126 }
127
vlma29c5bb2004-10-23 11:20:06 +0000128 /* assert(want >= 1 && want <= 6) */
129
130 /* Check character sequence length */
131 if(buf + want > end) return U8E_TRUNC;
132
vlm8d3e6d52006-09-17 01:27:44 +0000133 value = ch & (0xff >> want);
vlma29c5bb2004-10-23 11:20:06 +0000134 cend = buf + want;
135 for(buf++; buf < cend; buf++) {
136 ch = *buf;
137 if(ch < 0x80 || ch > 0xbf) return U8E_NOTCONT;
138 value = (value << 6) | (ch & 0x3F);
139 }
140 if(value < UTF8String_mv[want])
141 return U8E_NOTMIN;
vlm8d3e6d52006-09-17 01:27:44 +0000142 if(dst < dstend)
143 *dst++ = value; /* Record value */
vlma29c5bb2004-10-23 11:20:06 +0000144 }
145
vlm8d3e6d52006-09-17 01:27:44 +0000146 if(dst < dstend) *dst = 0; /* zero-terminate */
vlma29c5bb2004-10-23 11:20:06 +0000147
148 return length;
149}
150
151
152ssize_t
153UTF8String_length(const UTF8String_t *st) {
154 if(st && st->buf) {
155 return UTF8String__process(st, 0, 0);
vlmfa67ddc2004-06-03 03:38:44 +0000156 } else {
vlm1e10a102004-10-02 15:55:07 +0000157 return U8E_EINVAL;
vlmfa67ddc2004-06-03 03:38:44 +0000158 }
159}
160
vlma29c5bb2004-10-23 11:20:06 +0000161size_t
162UTF8String_to_wcs(const UTF8String_t *st, uint32_t *dst, size_t dstlen) {
163 if(st && st->buf) {
164 ssize_t ret = UTF8String__process(st, dst, dstlen);
165 return (ret < 0) ? 0 : ret;
166 } else {
167 return 0;
168 }
169}
170
vlmfa67ddc2004-06-03 03:38:44 +0000171int
vlmef6355b2004-09-29 13:26:15 +0000172UTF8String_print(asn_TYPE_descriptor_t *td, const void *sptr, int ilevel,
vlmfa67ddc2004-06-03 03:38:44 +0000173 asn_app_consume_bytes_f *cb, void *app_key) {
vlmda674682004-08-11 09:07:36 +0000174 const UTF8String_t *st = (const UTF8String_t *)sptr;
vlmfa67ddc2004-06-03 03:38:44 +0000175
vlmb42843a2004-06-05 08:17:50 +0000176 (void)td; /* Unused argument */
177 (void)ilevel; /* Unused argument */
178
vlmfa67ddc2004-06-03 03:38:44 +0000179 if(st && st->buf) {
vlm6678cb12004-09-26 13:10:40 +0000180 return (cb(st->buf, st->size, app_key) < 0) ? -1 : 0;
vlmfa67ddc2004-06-03 03:38:44 +0000181 } else {
vlm6678cb12004-09-26 13:10:40 +0000182 return (cb("<absent>", 8, app_key) < 0) ? -1 : 0;
vlmfa67ddc2004-06-03 03:38:44 +0000183 }
184}