[457] | 1 | // Copyright (c) Microsoft. All rights reserved.
|
---|
| 2 | // Licensed under the MIT license. See LICENSE file in the project root for full license information.
|
---|
| 3 |
|
---|
| 4 | #ifdef __cplusplus
|
---|
| 5 | #include <cstdlib>
|
---|
| 6 | #include <cstdint>
|
---|
| 7 | #else
|
---|
| 8 | #include <stdlib.h>
|
---|
| 9 | #include <stdbool.h>
|
---|
| 10 | #include <stdint.h>
|
---|
| 11 | #endif
|
---|
| 12 |
|
---|
| 13 | #include "azure_c_shared_utility/utf8_checker.h"
|
---|
| 14 |
|
---|
| 15 | bool utf8_checker_is_valid_utf8(const unsigned char* utf8_str, size_t length)
|
---|
| 16 | {
|
---|
| 17 | bool result;
|
---|
| 18 |
|
---|
| 19 | if (utf8_str == NULL)
|
---|
| 20 | {
|
---|
| 21 | /* Codes_SRS_UTF8_CHECKER_01_002: [ If utf8_checker_is_valid_utf8 is called with NULL utf8_str it shall return false. ]*/
|
---|
| 22 | result = false;
|
---|
| 23 | }
|
---|
| 24 | else
|
---|
| 25 | {
|
---|
| 26 | size_t pos = 0;
|
---|
| 27 |
|
---|
| 28 | /* Codes_SRS_UTF8_CHECKER_01_003: [ If length is 0, utf8_checker_is_valid_utf8 shall consider utf8_str to be valid UTF-8 and return true. ]*/
|
---|
| 29 | result = true;
|
---|
| 30 |
|
---|
| 31 | while ((result == true) &&
|
---|
| 32 | (pos < length))
|
---|
| 33 | {
|
---|
| 34 | /* Codes_SRS_UTF8_CHECKER_01_001: [ utf8_checker_is_valid_utf8 shall verify that the sequence of chars pointed to by utf8_str represent UTF-8 encoded codepoints. ]*/
|
---|
| 35 | if ((utf8_str[pos] >> 3) == 0x1E)
|
---|
| 36 | {
|
---|
| 37 | /* 4 bytes */
|
---|
| 38 | /* Codes_SRS_UTF8_CHECKER_01_009: [ 000uuuuu zzzzyyyy yyxxxxxx 11110uuu 10uuzzzz 10yyyyyy 10xxxxxx ]*/
|
---|
| 39 | uint32_t code_point = (utf8_str[pos] & 0x07);
|
---|
| 40 |
|
---|
| 41 | pos++;
|
---|
| 42 | if ((pos < length) &&
|
---|
| 43 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 44 | {
|
---|
| 45 | code_point <<= 6;
|
---|
| 46 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 47 |
|
---|
| 48 | pos++;
|
---|
| 49 | if ((pos < length) &&
|
---|
| 50 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 51 | {
|
---|
| 52 | code_point <<= 6;
|
---|
| 53 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 54 |
|
---|
| 55 | pos++;
|
---|
| 56 | if ((pos < length) &&
|
---|
| 57 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 58 | {
|
---|
| 59 | code_point <<= 6;
|
---|
| 60 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 61 |
|
---|
| 62 | if (code_point <= 0xFFFF)
|
---|
| 63 | {
|
---|
| 64 | result = false;
|
---|
| 65 | }
|
---|
| 66 | else
|
---|
| 67 | {
|
---|
| 68 | /* Codes_SRS_UTF8_CHECKER_01_005: [ On success it shall return true. ]*/
|
---|
| 69 | result = true;
|
---|
| 70 | pos++;
|
---|
| 71 | }
|
---|
| 72 | }
|
---|
| 73 | else
|
---|
| 74 | {
|
---|
| 75 | result = false;
|
---|
| 76 | }
|
---|
| 77 | }
|
---|
| 78 | else
|
---|
| 79 | {
|
---|
| 80 | result = false;
|
---|
| 81 | }
|
---|
| 82 | }
|
---|
| 83 | else
|
---|
| 84 | {
|
---|
| 85 | result = false;
|
---|
| 86 | }
|
---|
| 87 | }
|
---|
| 88 | else if ((utf8_str[pos] >> 4) == 0x0E)
|
---|
| 89 | {
|
---|
| 90 | /* 3 bytes */
|
---|
| 91 | /* Codes_SRS_UTF8_CHECKER_01_008: [ zzzzyyyy yyxxxxxx 1110zzzz 10yyyyyy 10xxxxxx ]*/
|
---|
| 92 | uint32_t code_point = (utf8_str[pos] & 0x0F);
|
---|
| 93 |
|
---|
| 94 | pos++;
|
---|
| 95 | if ((pos < length) &&
|
---|
| 96 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 97 | {
|
---|
| 98 | code_point <<= 6;
|
---|
| 99 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 100 |
|
---|
| 101 | pos++;
|
---|
| 102 | if ((pos < length) &&
|
---|
| 103 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 104 | {
|
---|
| 105 | code_point <<= 6;
|
---|
| 106 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 107 |
|
---|
| 108 | if (code_point <= 0x7FF)
|
---|
| 109 | {
|
---|
| 110 | result = false;
|
---|
| 111 | }
|
---|
| 112 | else
|
---|
| 113 | {
|
---|
| 114 | /* Codes_SRS_UTF8_CHECKER_01_005: [ On success it shall return true. ]*/
|
---|
| 115 | result = true;
|
---|
| 116 | pos++;
|
---|
| 117 | }
|
---|
| 118 | }
|
---|
| 119 | else
|
---|
| 120 | {
|
---|
| 121 | result = false;
|
---|
| 122 | }
|
---|
| 123 | }
|
---|
| 124 | else
|
---|
| 125 | {
|
---|
| 126 | result = false;
|
---|
| 127 | }
|
---|
| 128 | }
|
---|
| 129 | else if ((utf8_str[pos] >> 5) == 0x06)
|
---|
| 130 | {
|
---|
| 131 | /* 2 bytes */
|
---|
| 132 | /* Codes_SRS_UTF8_CHECKER_01_007: [ 00000yyy yyxxxxxx 110yyyyy 10xxxxxx ]*/
|
---|
| 133 | uint32_t code_point = (utf8_str[pos] & 0x1F);
|
---|
| 134 |
|
---|
| 135 | pos++;
|
---|
| 136 | if ((pos < length) &&
|
---|
| 137 | ((utf8_str[pos] >> 6) == 0x02))
|
---|
| 138 | {
|
---|
| 139 | code_point <<= 6;
|
---|
| 140 | code_point += utf8_str[pos] & 0x3F;
|
---|
| 141 |
|
---|
| 142 | if (code_point <= 0x7F)
|
---|
| 143 | {
|
---|
| 144 | result = false;
|
---|
| 145 | }
|
---|
| 146 | else
|
---|
| 147 | {
|
---|
| 148 | /* Codes_SRS_UTF8_CHECKER_01_005: [ On success it shall return true. ]*/
|
---|
| 149 | result = true;
|
---|
| 150 | pos++;
|
---|
| 151 | }
|
---|
| 152 | }
|
---|
| 153 | else
|
---|
| 154 | {
|
---|
| 155 | result = false;
|
---|
| 156 | }
|
---|
| 157 | }
|
---|
| 158 | else if ((utf8_str[pos] >> 7) == 0x00)
|
---|
| 159 | {
|
---|
| 160 | /* 1 byte */
|
---|
| 161 | /* Codes_SRS_UTF8_CHECKER_01_006: [ 00000000 0xxxxxxx 0xxxxxxx ]*/
|
---|
| 162 | /* Codes_SRS_UTF8_CHECKER_01_005: [ On success it shall return true. ]*/
|
---|
| 163 | result = true;
|
---|
| 164 | pos++;
|
---|
| 165 | }
|
---|
| 166 | else
|
---|
| 167 | {
|
---|
| 168 | /* error */
|
---|
| 169 | result = false;
|
---|
| 170 | }
|
---|
| 171 | }
|
---|
| 172 | }
|
---|
| 173 |
|
---|
| 174 | return result;
|
---|
| 175 | }
|
---|