|
| 1 | +<?php |
| 2 | +namespace Aws\Api\Cbor; |
| 3 | + |
| 4 | +use Aws\Api\Cbor\Exception\CborException; |
| 5 | + |
| 6 | +/** |
| 7 | + * Encodes PHP values to Concise Binary Object Representation according to RFC 8949 |
| 8 | + * https://www.rfc-editor.org/rfc/rfc8949.html |
| 9 | + * |
| 10 | + * Supports Major types 0-7 including: |
| 11 | + * - Type 0: Unsigned integers |
| 12 | + * - Type 1: Negative integers |
| 13 | + * - Type 2: Byte strings (via ['__cbor_bytes' => $data] wrappers) |
| 14 | + * - Type 3: Text strings (UTF-8) |
| 15 | + * - Type 4: Arrays |
| 16 | + * - Type 5: Maps |
| 17 | + * - Type 6: Tagged values (timestamps) |
| 18 | + * - Type 7: Simple values (null, bool, float) |
| 19 | + * |
| 20 | + * @internal |
| 21 | + */ |
| 22 | +final class CborEncoder |
| 23 | +{ |
| 24 | + /** |
| 25 | + * Pre-encoded integers 0-23 (single byte) and common larger values |
| 26 | + * CBOR major type 0 (unsigned integer) |
| 27 | + */ |
| 28 | + private const INT_CACHE = [ |
| 29 | + 0 => "\x00", 1 => "\x01", 2 => "\x02", 3 => "\x03", |
| 30 | + 4 => "\x04", 5 => "\x05", 6 => "\x06", 7 => "\x07", |
| 31 | + 8 => "\x08", 9 => "\x09", 10 => "\x0A", 11 => "\x0B", |
| 32 | + 12 => "\x0C", 13 => "\x0D", 14 => "\x0E", 15 => "\x0F", |
| 33 | + 16 => "\x10", 17 => "\x11", 18 => "\x12", 19 => "\x13", |
| 34 | + 20 => "\x14", 21 => "\x15", 22 => "\x16", 23 => "\x17", |
| 35 | + 24 => "\x18\x18", 25 => "\x18\x19", 26 => "\x18\x1A", |
| 36 | + 32 => "\x18\x20", 50 => "\x18\x32", 64 => "\x18\x40", |
| 37 | + 100 => "\x18\x64", 128 => "\x18\x80", 200 => "\x18\xC8", |
| 38 | + 255 => "\x18\xFF", 256 => "\x19\x01\x00", 500 => "\x19\x01\xF4", |
| 39 | + 1000 => "\x19\x03\xE8", 1023 => "\x19\x03\xFF", |
| 40 | + ]; |
| 41 | + |
| 42 | + /** |
| 43 | + * Pre-encoded negative integers -1 to -24 and common larger values |
| 44 | + * CBOR major type 1 (negative integer) |
| 45 | + */ |
| 46 | + private const NEG_CACHE = [ |
| 47 | + -1 => "\x20", -2 => "\x21", -3 => "\x22", -4 => "\x23", |
| 48 | + -5 => "\x24", -10 => "\x29", -20 => "\x33", -24 => "\x37", |
| 49 | + -25 => "\x38\x18", -50 => "\x38\x31", -100 => "\x38\x63", |
| 50 | + ]; |
| 51 | + |
| 52 | + /** |
| 53 | + * Encode a PHP value to CBOR binary string |
| 54 | + * |
| 55 | + * @param mixed $value The value to encode |
| 56 | + * |
| 57 | + * @return string |
| 58 | + */ |
| 59 | + public function encode(mixed $value): string |
| 60 | + { |
| 61 | + return $this->encodeValue($value); |
| 62 | + } |
| 63 | + |
| 64 | + /** |
| 65 | + * Recursively encode a value to CBOR |
| 66 | + * |
| 67 | + * @param mixed $value Value to encode |
| 68 | + * @return string Encoded CBOR bytes |
| 69 | + */ |
| 70 | + private function encodeValue(mixed $value): string |
| 71 | + { |
| 72 | + switch (gettype($value)) { |
| 73 | + case 'string': |
| 74 | + $len = strlen($value); |
| 75 | + if ($len < 24) { |
| 76 | + return chr(0x60 | $len) . $value; |
| 77 | + } |
| 78 | + |
| 79 | + if ($len < 0x100) { |
| 80 | + return "\x78" . chr($len) . $value; |
| 81 | + } |
| 82 | + |
| 83 | + return $this->encodeTextString($value); |
| 84 | + |
| 85 | + case 'array': |
| 86 | + if (isset($value['__cbor_timestamp'])) { |
| 87 | + return "\xC1\xFB" . pack('E', $value['__cbor_timestamp']); |
| 88 | + } |
| 89 | + |
| 90 | + // Encode a byte string (major type 2) |
| 91 | + if (isset($value['__cbor_bytes'])) { |
| 92 | + $bytes = $value['__cbor_bytes']; |
| 93 | + $len = strlen($bytes); |
| 94 | + if ($len < 24) { |
| 95 | + return chr(0x40 | $len) . $bytes; |
| 96 | + } |
| 97 | + |
| 98 | + if ($len < 0x100) { |
| 99 | + return "\x58" . chr($len) . $bytes; |
| 100 | + } |
| 101 | + |
| 102 | + if ($len < 0x10000) { |
| 103 | + return "\x59" . pack('n', $len) . $bytes; |
| 104 | + } |
| 105 | + |
| 106 | + return "\x5A" . pack('N', $len) . $bytes; |
| 107 | + } |
| 108 | + |
| 109 | + if (array_is_list($value)) { |
| 110 | + return $this->encodeArray($value); |
| 111 | + } |
| 112 | + |
| 113 | + return $this->encodeMap($value); |
| 114 | + |
| 115 | + case 'integer': |
| 116 | + if (isset(self::INT_CACHE[$value])) { |
| 117 | + return self::INT_CACHE[$value]; |
| 118 | + } |
| 119 | + |
| 120 | + if (isset(self::NEG_CACHE[$value])) { |
| 121 | + return self::NEG_CACHE[$value]; |
| 122 | + } |
| 123 | + |
| 124 | + // Fast path for positive integers |
| 125 | + // Major type 0: unsigned integer |
| 126 | + if ($value >= 0) { |
| 127 | + if ($value < 24) { |
| 128 | + return chr($value); |
| 129 | + } |
| 130 | + |
| 131 | + if ($value < 0x100) { |
| 132 | + return "\x18" . chr($value); |
| 133 | + } |
| 134 | + |
| 135 | + if ($value < 0x10000) { |
| 136 | + return "\x19" . pack('n', $value); |
| 137 | + } |
| 138 | + |
| 139 | + if ($value < 0x100000000) { |
| 140 | + return "\x1A" . pack('N', $value); |
| 141 | + } |
| 142 | + |
| 143 | + return "\x1B" . pack('J', $value); |
| 144 | + } |
| 145 | + |
| 146 | + return $this->encodeInteger($value); |
| 147 | + |
| 148 | + case 'double': |
| 149 | + // Encode a float (major type 7, float 64) |
| 150 | + return "\xFB" . pack('E', $value); |
| 151 | + |
| 152 | + case 'boolean': |
| 153 | + // Encode a boolean (major type 7, simple) |
| 154 | + return $value ? "\xF5" : "\xF4"; |
| 155 | + |
| 156 | + case 'NULL': |
| 157 | + // Encode null (major type 7, simple) |
| 158 | + return "\xF6"; |
| 159 | + |
| 160 | + case 'object': |
| 161 | + throw new CborException("Cannot encode object of type: " . get_class($value)); |
| 162 | + |
| 163 | + default: |
| 164 | + throw new CborException("Cannot encode value of type: " . gettype($value)); |
| 165 | + } |
| 166 | + } |
| 167 | + |
| 168 | + /** |
| 169 | + * Encode an integer (major type 0 or 1) |
| 170 | + * |
| 171 | + * @param int $value |
| 172 | + * @return string |
| 173 | + */ |
| 174 | + private function encodeInteger(int $value): string |
| 175 | + { |
| 176 | + if (isset(self::INT_CACHE[$value])) { |
| 177 | + return self::INT_CACHE[$value]; |
| 178 | + } |
| 179 | + |
| 180 | + if (isset(self::NEG_CACHE[$value])) { |
| 181 | + return self::NEG_CACHE[$value]; |
| 182 | + } |
| 183 | + |
| 184 | + if ($value >= 0) { |
| 185 | + // Major type 0: unsigned integer |
| 186 | + if ($value < 24) { |
| 187 | + return chr($value); |
| 188 | + } |
| 189 | + |
| 190 | + if ($value < 0x100) { |
| 191 | + return "\x18" . chr($value); |
| 192 | + } |
| 193 | + |
| 194 | + if ($value < 0x10000) { |
| 195 | + return "\x19" . pack('n', $value); |
| 196 | + } |
| 197 | + |
| 198 | + if ($value < 0x100000000) { |
| 199 | + return "\x1A" . pack('N', $value); |
| 200 | + } |
| 201 | + |
| 202 | + return "\x1B" . pack('J', $value); |
| 203 | + } |
| 204 | + |
| 205 | + // Major type 1: negative integer (-1 - n) |
| 206 | + $value = -1 - $value; |
| 207 | + if ($value < 24) { |
| 208 | + return chr(0x20 | $value); |
| 209 | + } |
| 210 | + |
| 211 | + if ($value < 0x100) { |
| 212 | + return "\x38" . chr($value); |
| 213 | + } |
| 214 | + |
| 215 | + if ($value < 0x10000) { |
| 216 | + return "\x39" . pack('n', $value); |
| 217 | + } |
| 218 | + |
| 219 | + if ($value < 0x100000000) { |
| 220 | + return "\x3A" . pack('N', $value); |
| 221 | + } |
| 222 | + |
| 223 | + return "\x3B" . pack('J', $value); |
| 224 | + } |
| 225 | + |
| 226 | + /** |
| 227 | + * Encode a text string (major type 3) |
| 228 | + * |
| 229 | + * @param string $value |
| 230 | + * @return string |
| 231 | + */ |
| 232 | + private function encodeTextString(string $value): string |
| 233 | + { |
| 234 | + $len = strlen($value); |
| 235 | + |
| 236 | + if ($len < 24) { |
| 237 | + return chr(0x60 | $len) . $value; |
| 238 | + } |
| 239 | + |
| 240 | + if ($len < 0x100) { |
| 241 | + return "\x78" . chr($len) . $value; |
| 242 | + } |
| 243 | + |
| 244 | + if ($len < 0x10000) { |
| 245 | + return "\x79" . pack('n', $len) . $value; |
| 246 | + } |
| 247 | + |
| 248 | + if ($len < 0x100000000) { |
| 249 | + return "\x7A" . pack('N', $len) . $value; |
| 250 | + } |
| 251 | + |
| 252 | + return "\x7B" . pack('J', $len) . $value; |
| 253 | + } |
| 254 | + |
| 255 | + /** |
| 256 | + * Encode an array (major type 4) |
| 257 | + * |
| 258 | + * @param array $value |
| 259 | + * @return string |
| 260 | + */ |
| 261 | + private function encodeArray(array $value): string |
| 262 | + { |
| 263 | + $count = count($value); |
| 264 | + |
| 265 | + if ($count < 24) { |
| 266 | + $result = chr(0x80 | $count); |
| 267 | + } elseif ($count < 0x100) { |
| 268 | + $result = "\x98" . chr($count); |
| 269 | + } elseif ($count < 0x10000) { |
| 270 | + $result = "\x99" . pack('n', $count); |
| 271 | + } elseif ($count < 0x100000000) { |
| 272 | + $result = "\x9A" . pack('N', $count); |
| 273 | + } else { |
| 274 | + $result = "\x9B" . pack('J', $count); |
| 275 | + } |
| 276 | + |
| 277 | + foreach ($value as $item) { |
| 278 | + $result .= $this->encodeValue($item); |
| 279 | + } |
| 280 | + |
| 281 | + return $result; |
| 282 | + } |
| 283 | + |
| 284 | + /** |
| 285 | + * Encode a map (major type 5) |
| 286 | + * |
| 287 | + * @param array $value |
| 288 | + * @return string |
| 289 | + */ |
| 290 | + private function encodeMap(array $value): string |
| 291 | + { |
| 292 | + $count = count($value); |
| 293 | + |
| 294 | + if ($count < 24) { |
| 295 | + $result = chr(0xA0 | $count); |
| 296 | + } elseif ($count < 0x100) { |
| 297 | + $result = "\xB8" . chr($count); |
| 298 | + } elseif ($count < 0x10000) { |
| 299 | + $result = "\xB9" . pack('n', $count); |
| 300 | + } elseif ($count < 0x100000000) { |
| 301 | + $result = "\xBA" . pack('N', $count); |
| 302 | + } else { |
| 303 | + $result = "\xBB" . pack('J', $count); |
| 304 | + } |
| 305 | + |
| 306 | + foreach ($value as $k => $v) { |
| 307 | + if (is_int($k)) { |
| 308 | + $result .= $this->encodeInteger($k); |
| 309 | + } else { |
| 310 | + $len = strlen($k); |
| 311 | + if ($len < 24) { |
| 312 | + $result .= chr(0x60 | $len) . $k; |
| 313 | + } elseif ($len < 0x100) { |
| 314 | + $result .= "\x78" . chr($len) . $k; |
| 315 | + } else { |
| 316 | + $result .= "\x79" . pack('n', $len) . $k; |
| 317 | + } |
| 318 | + } |
| 319 | + |
| 320 | + $result .= $this->encodeValue($v); |
| 321 | + } |
| 322 | + |
| 323 | + return $result; |
| 324 | + } |
| 325 | + |
| 326 | + /** |
| 327 | + * Create an empty map (major type 5 with 0 elements) |
| 328 | + * |
| 329 | + * @return string |
| 330 | + */ |
| 331 | + public function encodeEmptyMap(): string |
| 332 | + { |
| 333 | + return "\xA0"; |
| 334 | + } |
| 335 | + |
| 336 | + /** |
| 337 | + * Create an empty indefinite map (major type 5 indefinite length) |
| 338 | + * |
| 339 | + * @return string |
| 340 | + */ |
| 341 | + public function encodeEmptyIndefiniteMap(): string |
| 342 | + { |
| 343 | + return "\xBF\xFF"; |
| 344 | + } |
| 345 | +} |
0 commit comments