vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 1 | /*- |
| 2 | * Copyright (c) 2003, 2004 Lev Walkin <vlm@lionet.info>. All rights reserved. |
| 3 | * Redistribution and modifications are permitted subject to BSD license. |
| 4 | */ |
vlm | 39ba4c4 | 2004-09-22 16:06:28 +0000 | [diff] [blame] | 5 | #include <asn_internal.h> |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 6 | #include <UTF8String.h> |
| 7 | |
| 8 | /* |
| 9 | * UTF8String basic type description. |
| 10 | */ |
vlm | ef6355b | 2004-09-29 13:26:15 +0000 | [diff] [blame] | 11 | static ber_tlv_tag_t asn_DEF_UTF8String_tags[] = { |
vlm | 72425de | 2004-09-13 08:31:01 +0000 | [diff] [blame] | 12 | (ASN_TAG_CLASS_UNIVERSAL | (12 << 2)), /* [UNIVERSAL 12] IMPLICIT ...*/ |
| 13 | (ASN_TAG_CLASS_UNIVERSAL | (4 << 2)), /* ... OCTET STRING */ |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 14 | }; |
vlm | ef6355b | 2004-09-29 13:26:15 +0000 | [diff] [blame] | 15 | asn_TYPE_descriptor_t asn_DEF_UTF8String = { |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 16 | "UTF8String", |
vlm | 9de248e | 2004-10-20 15:50:55 +0000 | [diff] [blame] | 17 | "UTF8String", |
vlm | 39ba4c4 | 2004-09-22 16:06:28 +0000 | [diff] [blame] | 18 | OCTET_STRING_free, |
| 19 | UTF8String_print, |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 20 | UTF8String_constraint, /* Check for invalid codes, etc. */ |
| 21 | OCTET_STRING_decode_ber, /* Implemented in terms of OCTET STRING */ |
vlm | 9de248e | 2004-10-20 15:50:55 +0000 | [diff] [blame] | 22 | OCTET_STRING_encode_der, |
| 23 | OCTET_STRING_decode_xer_utf8, |
| 24 | OCTET_STRING_encode_xer_utf8, |
vlm | 337167e | 2005-11-26 11:25:14 +0000 | [diff] [blame] | 25 | 0, |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 26 | 0, /* Use generic outmost tag fetcher */ |
vlm | ef6355b | 2004-09-29 13:26:15 +0000 | [diff] [blame] | 27 | asn_DEF_UTF8String_tags, |
| 28 | sizeof(asn_DEF_UTF8String_tags) |
| 29 | / sizeof(asn_DEF_UTF8String_tags[0]) - 1, |
| 30 | asn_DEF_UTF8String_tags, |
| 31 | sizeof(asn_DEF_UTF8String_tags) |
| 32 | / sizeof(asn_DEF_UTF8String_tags[0]), |
vlm | 337167e | 2005-11-26 11:25:14 +0000 | [diff] [blame] | 33 | 0, /* No PER visible constraints */ |
vlm | e413c12 | 2004-08-20 13:23:42 +0000 | [diff] [blame] | 34 | 0, 0, /* No members */ |
vlm | b42843a | 2004-06-05 08:17:50 +0000 | [diff] [blame] | 35 | 0 /* No specifics */ |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 36 | }; |
| 37 | |
vlm | 5128329 | 2004-10-02 11:37:38 +0000 | [diff] [blame] | 38 | /* |
| 39 | * This is the table of length expectations. |
vlm | 1a69ff3 | 2005-03-02 13:30:25 +0000 | [diff] [blame] | 40 | * The second half of this table is only applicable to the long sequences. |
vlm | 5128329 | 2004-10-02 11:37:38 +0000 | [diff] [blame] | 41 | */ |
| 42 | static int UTF8String_ht[2][16] = { |
| 43 | { /* 0x0 ... 0x7 */ |
| 44 | /* 0000..0111 */ |
| 45 | 1, 1, 1, 1, 1, 1, 1, 1, |
| 46 | /* 1000..1011(0), 1100..1101(2), 1110(3), 1111(-1) */ |
| 47 | 0, 0, 0, 0, 2, 2, 3, -1 }, |
| 48 | { /* 0xF0 .. 0xF7 */ |
| 49 | /* 11110000..11110111 */ |
| 50 | 4, 4, 4, 4, 4, 4, 4, 4, |
| 51 | 5, 5, 5, 5, 6, 6, -1, -1 } |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 52 | }; |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 53 | static int32_t UTF8String_mv[7] = { 0, 0, |
| 54 | 0x00000080, |
| 55 | 0x00000800, |
| 56 | 0x00010000, |
| 57 | 0x00200000, |
| 58 | 0x04000000 |
| 59 | }; |
| 60 | |
| 61 | /* Internal aliases for return codes */ |
| 62 | #define U8E_TRUNC -1 /* UTF-8 sequence truncated */ |
| 63 | #define U8E_ILLSTART -2 /* Illegal UTF-8 sequence start */ |
| 64 | #define U8E_NOTCONT -3 /* Continuation expectation failed */ |
| 65 | #define U8E_NOTMIN -4 /* Not minimal length encoding */ |
| 66 | #define U8E_EINVAL -5 /* Invalid arguments */ |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 67 | |
| 68 | int |
vlm | ef6355b | 2004-09-29 13:26:15 +0000 | [diff] [blame] | 69 | UTF8String_constraint(asn_TYPE_descriptor_t *td, const void *sptr, |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 70 | asn_app_constraint_failed_f *ctfailcb, void *app_key) { |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 71 | ssize_t len = UTF8String_length((const UTF8String_t *)sptr); |
| 72 | switch(len) { |
| 73 | case U8E_EINVAL: |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 74 | _ASN_CTFAIL(app_key, td, |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 75 | "%s: value not given", td->name); |
| 76 | break; |
| 77 | case U8E_TRUNC: |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 78 | _ASN_CTFAIL(app_key, td, |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 79 | "%s: truncated UTF-8 sequence (%s:%d)", |
| 80 | td->name, __FILE__, __LINE__); |
| 81 | break; |
| 82 | case U8E_ILLSTART: |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 83 | _ASN_CTFAIL(app_key, td, |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 84 | "%s: UTF-8 illegal start of encoding (%s:%d)", |
| 85 | td->name, __FILE__, __LINE__); |
| 86 | break; |
| 87 | case U8E_NOTCONT: |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 88 | _ASN_CTFAIL(app_key, td, |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 89 | "%s: UTF-8 not continuation (%s:%d)", |
| 90 | td->name, __FILE__, __LINE__); |
| 91 | break; |
| 92 | case U8E_NOTMIN: |
vlm | af68ef5 | 2006-07-13 11:19:01 +0000 | [diff] [blame] | 93 | _ASN_CTFAIL(app_key, td, |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 94 | "%s: UTF-8 not minimal sequence (%s:%d)", |
| 95 | td->name, __FILE__, __LINE__); |
| 96 | break; |
| 97 | } |
| 98 | return (len < 0) ? -1 : 0; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 99 | } |
| 100 | |
vlm | a29c5bb | 2004-10-23 11:20:06 +0000 | [diff] [blame] | 101 | static ssize_t |
| 102 | UTF8String__process(const UTF8String_t *st, uint32_t *dst, size_t dstlen) { |
| 103 | size_t length; |
| 104 | uint8_t *buf = st->buf; |
| 105 | uint8_t *end = buf + st->size; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 106 | |
vlm | a29c5bb | 2004-10-23 11:20:06 +0000 | [diff] [blame] | 107 | for(length = 0; buf < end; length++) { |
| 108 | int ch = *buf; |
| 109 | uint8_t *cend; |
| 110 | int32_t value; |
| 111 | int want; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 112 | |
vlm | a29c5bb | 2004-10-23 11:20:06 +0000 | [diff] [blame] | 113 | /* Compute the sequence length */ |
| 114 | want = UTF8String_ht[0][ch >> 4]; |
| 115 | switch(want) { |
| 116 | case -1: |
| 117 | /* Second half of the table, long sequence */ |
| 118 | want = UTF8String_ht[1][ch & 0x0F]; |
| 119 | if(want != -1) break; |
| 120 | /* Fall through */ |
| 121 | case 0: |
| 122 | return U8E_ILLSTART; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 123 | } |
| 124 | |
vlm | a29c5bb | 2004-10-23 11:20:06 +0000 | [diff] [blame] | 125 | /* assert(want >= 1 && want <= 6) */ |
| 126 | |
| 127 | /* Check character sequence length */ |
| 128 | if(buf + want > end) return U8E_TRUNC; |
| 129 | |
| 130 | value = ch & (0xff >> (want + 1)); |
| 131 | cend = buf + want; |
| 132 | for(buf++; buf < cend; buf++) { |
| 133 | ch = *buf; |
| 134 | if(ch < 0x80 || ch > 0xbf) return U8E_NOTCONT; |
| 135 | value = (value << 6) | (ch & 0x3F); |
| 136 | } |
| 137 | if(value < UTF8String_mv[want]) |
| 138 | return U8E_NOTMIN; |
| 139 | if(dstlen) *dst++ = value; /* Record value */ |
| 140 | } |
| 141 | |
| 142 | if(dstlen) *dst = 0; /* zero-terminate */ |
| 143 | |
| 144 | return length; |
| 145 | } |
| 146 | |
| 147 | |
| 148 | ssize_t |
| 149 | UTF8String_length(const UTF8String_t *st) { |
| 150 | if(st && st->buf) { |
| 151 | return UTF8String__process(st, 0, 0); |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 152 | } else { |
vlm | 1e10a10 | 2004-10-02 15:55:07 +0000 | [diff] [blame] | 153 | return U8E_EINVAL; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 154 | } |
| 155 | } |
| 156 | |
vlm | a29c5bb | 2004-10-23 11:20:06 +0000 | [diff] [blame] | 157 | size_t |
| 158 | UTF8String_to_wcs(const UTF8String_t *st, uint32_t *dst, size_t dstlen) { |
| 159 | if(st && st->buf) { |
| 160 | ssize_t ret = UTF8String__process(st, dst, dstlen); |
| 161 | return (ret < 0) ? 0 : ret; |
| 162 | } else { |
| 163 | return 0; |
| 164 | } |
| 165 | } |
| 166 | |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 167 | int |
vlm | ef6355b | 2004-09-29 13:26:15 +0000 | [diff] [blame] | 168 | UTF8String_print(asn_TYPE_descriptor_t *td, const void *sptr, int ilevel, |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 169 | asn_app_consume_bytes_f *cb, void *app_key) { |
vlm | da67468 | 2004-08-11 09:07:36 +0000 | [diff] [blame] | 170 | const UTF8String_t *st = (const UTF8String_t *)sptr; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 171 | |
vlm | b42843a | 2004-06-05 08:17:50 +0000 | [diff] [blame] | 172 | (void)td; /* Unused argument */ |
| 173 | (void)ilevel; /* Unused argument */ |
| 174 | |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 175 | if(st && st->buf) { |
vlm | 6678cb1 | 2004-09-26 13:10:40 +0000 | [diff] [blame] | 176 | return (cb(st->buf, st->size, app_key) < 0) ? -1 : 0; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 177 | } else { |
vlm | 6678cb1 | 2004-09-26 13:10:40 +0000 | [diff] [blame] | 178 | return (cb("<absent>", 8, app_key) < 0) ? -1 : 0; |
vlm | fa67ddc | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 179 | } |
| 180 | } |