| 1 |
<?php |
| 2 |
/** |
| 3 |
* WCPOS sync wire normalization. |
| 4 |
* |
| 5 |
* @package WCPOS\WooCommercePOS\Sync |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace WCPOS\WooCommercePOS\Sync; |
| 9 |
|
| 10 |
use WCPOS\WooCommercePOS\Logger; |
| 11 |
|
| 12 |
/** |
| 13 |
* Normalizes structured meta values before sync documents are hashed or emitted. |
| 14 |
*/ |
| 15 |
final class Meta_Normalizer { |
| 16 |
/** |
| 17 |
* Upper bound on array elements and object properties in ONE meta value, counted |
| 18 |
* across every nesting level. The POS's own structured meta (`_woocommerce_pos_data`, |
| 19 |
* attribute maps, line-item meta) is a few hundred nodes; anything past this is another |
| 20 |
* plugin's bulk data and is not served: one store carried a 16.7-million-element meta |
| 21 |
* value, and the JSON round trip below killed every request that touched the record |
| 22 |
* (Sentry WOOCOMMERCE-POS-2KQ, ~22 fatals an hour). |
| 23 |
*/ |
| 24 |
public const OVERSIZED_META_NODE_LIMIT = 20000; |
| 25 |
|
| 26 |
/** |
| 27 |
* Upper bound on string bytes in ONE meta value, summed across nesting and keys (8 MiB): |
| 28 |
* the same cut-off for values that are few entries but enormous strings. |
| 29 |
* |
| 30 |
* Deliberately well clear of legitimate use. `Test_Catalog_Proxy_Meta_Scaling` pins a |
| 31 |
* 1 MiB meta value as something that must round-trip, so the cut-off sits eight times |
| 32 |
* above the largest size this repo asserts is normal, and still three orders of |
| 33 |
* magnitude below the gigabyte that took a store's requests down. |
| 34 |
*/ |
| 35 |
public const OVERSIZED_META_BYTE_LIMIT = 8388608; |
| 36 |
|
| 37 |
/** |
| 38 |
* Meta keys already reported as oversized in this request (one warning per key). |
| 39 |
* |
| 40 |
* @var array<string, true> |
| 41 |
*/ |
| 42 |
private static array $oversized_logged = array(); |
| 43 |
|
| 44 |
/** |
| 45 |
* Clear the per-request set of already-reported keys. |
| 46 |
* |
| 47 |
* The dedupe is request-scoped in production, where the process ends with the |
| 48 |
* response. Long-lived processes and the test suite share one process across many |
| 49 |
* requests, so they reset at the boundary like the other request-scoped collectors. |
| 50 |
*/ |
| 51 |
public static function reset_request_state(): void { |
| 52 |
self::$oversized_logged = array(); |
| 53 |
} |
| 54 |
|
| 55 |
/** |
| 56 |
* Register the shared pre-stamping normalization seams. |
| 57 |
*/ |
| 58 |
public static function register_hooks(): void { |
| 59 |
add_filter( 'woocommerce_pos_sync_proxy_response', array( __CLASS__, 'normalize' ), 5 ); |
| 60 |
add_filter( 'woocommerce_pos_sync_serialized_product', array( __CLASS__, 'normalize' ), 5 ); |
| 61 |
add_filter( 'woocommerce_pos_sync_serialized_order', array( __CLASS__, 'normalize' ), 5 ); |
| 62 |
} |
| 63 |
|
| 64 |
/** |
| 65 |
* Unregister the shared pre-stamping normalization seams. |
| 66 |
*/ |
| 67 |
public static function unregister_hooks(): void { |
| 68 |
remove_filter( 'woocommerce_pos_sync_proxy_response', array( __CLASS__, 'normalize' ), 5 ); |
| 69 |
remove_filter( 'woocommerce_pos_sync_serialized_product', array( __CLASS__, 'normalize' ), 5 ); |
| 70 |
remove_filter( 'woocommerce_pos_sync_serialized_order', array( __CLASS__, 'normalize' ), 5 ); |
| 71 |
} |
| 72 |
|
| 73 |
/** |
| 74 |
* Recursively normalize every meta_data array in a document or payload. |
| 75 |
* |
| 76 |
* @param mixed $payload Document or payload being prepared for the wire. |
| 77 |
* |
| 78 |
* @return mixed |
| 79 |
*/ |
| 80 |
public static function normalize( $payload ) { |
| 81 |
if ( ! is_array( $payload ) ) { |
| 82 |
return $payload; |
| 83 |
} |
| 84 |
|
| 85 |
foreach ( $payload as $key => $value ) { |
| 86 |
if ( 'meta_data' === $key && is_array( $value ) ) { |
| 87 |
$value = self::normalize_meta_data( $value ); |
| 88 |
} |
| 89 |
|
| 90 |
$payload[ $key ] = is_array( $value ) ? self::normalize( $value ) : $value; |
| 91 |
} |
| 92 |
|
| 93 |
return $payload; |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Shape-tolerant reader for structured meta that may be stored in either |
| 98 |
* form: the historical JSON-encoded string, or (after a typed client push |
| 99 |
* lands through wc/v3) a native PHP array. Server-side consumers of |
| 100 |
* `_woocommerce_pos_data`-style meta must read through this — a bare |
| 101 |
* `json_decode( $raw )` fatals on PHP 8 the moment the storage holds an |
| 102 |
* array. |
| 103 |
* |
| 104 |
* @param mixed $raw Meta value as returned by get_meta(). |
| 105 |
* |
| 106 |
* @return array|null Decoded associative array, or null when the value is |
| 107 |
* neither a JSON object/array string nor an array. |
| 108 |
*/ |
| 109 |
public static function decode_to_array( $raw ): ?array { |
| 110 |
if ( is_array( $raw ) ) { |
| 111 |
return $raw; |
| 112 |
} |
| 113 |
if ( is_object( $raw ) ) { |
| 114 |
$raw = wp_json_encode( $raw ); |
| 115 |
} |
| 116 |
if ( ! is_string( $raw ) || '' === $raw ) { |
| 117 |
return null; |
| 118 |
} |
| 119 |
$decoded = json_decode( $raw, true ); |
| 120 |
if ( JSON_ERROR_NONE !== json_last_error() || ! is_array( $decoded ) ) { |
| 121 |
return null; |
| 122 |
} |
| 123 |
|
| 124 |
return $decoded; |
| 125 |
} |
| 126 |
|
| 127 |
/** |
| 128 |
* Normalize structured meta values and derived display fields. |
| 129 |
* |
| 130 |
* @param array $meta_data Serialized REST meta entries. |
| 131 |
* |
| 132 |
* @return array |
| 133 |
*/ |
| 134 |
private static function normalize_meta_data( array $meta_data ): array { |
| 135 |
$was_list = array_keys( $meta_data ) === array_keys( array_values( $meta_data ) ); |
| 136 |
$dropped = false; |
| 137 |
foreach ( $meta_data as $index => $entry ) { |
| 138 |
// Top-level entity meta reaches the filters as live WC_Meta_Data objects |
| 139 |
// (they only become arrays at JSON-encode time); convert a copy to the |
| 140 |
// exact shape it would serialize to, and only swap it in when normalization |
| 141 |
// actually happens — untouched entries keep their original form so |
| 142 |
// revision hashes of scalar-only records are unchanged. The budget check |
| 143 |
// runs on the live value BEFORE the round trip: encoding an oversized value |
| 144 |
// is the allocation that took the whole request down. |
| 145 |
$is_meta_object = $entry instanceof \WC_Meta_Data; |
| 146 |
if ( $is_meta_object ) { |
| 147 |
$data = $entry->get_data(); |
| 148 |
if ( self::exceeds_value_budget( $data['value'] ?? null ) ) { |
| 149 |
self::drop_oversized( $meta_data, $index, $data ); |
| 150 |
$dropped = true; |
| 151 |
continue; |
| 152 |
} |
| 153 |
$entry = json_decode( wp_json_encode( $data ), true ); |
| 154 |
} |
| 155 |
|
| 156 |
if ( ! is_array( $entry ) ) { |
| 157 |
continue; |
| 158 |
} |
| 159 |
|
| 160 |
if ( ! $is_meta_object && self::exceeds_value_budget( $entry['value'] ?? null ) ) { |
| 161 |
self::drop_oversized( $meta_data, $index, $entry ); |
| 162 |
$dropped = true; |
| 163 |
continue; |
| 164 |
} |
| 165 |
|
| 166 |
$entry_changed = false; |
| 167 |
if ( isset( $entry['value'] ) && is_string( $entry['value'] ) ) { |
| 168 |
$raw = $entry['value']; |
| 169 |
$trimmed = trim( $raw ); |
| 170 |
$opening = substr( $trimmed, 0, 1 ); |
| 171 |
if ( '{' === $opening || '[' === $opening ) { |
| 172 |
$decoded = json_decode( $raw ); |
| 173 |
if ( JSON_ERROR_NONE === json_last_error() && ( is_array( $decoded ) || $decoded instanceof \stdClass ) ) { |
| 174 |
$entry['value'] = self::preserve_json_object_shape( $decoded ); |
| 175 |
$entry_changed = true; |
| 176 |
} |
| 177 |
} |
| 178 |
} |
| 179 |
foreach ( array( 'display_key', 'display_value' ) as $display_field ) { |
| 180 |
if ( array_key_exists( $display_field, $entry ) && ! is_string( $entry[ $display_field ] ) ) { |
| 181 |
unset( $entry[ $display_field ] ); |
| 182 |
$entry_changed = true; |
| 183 |
} |
| 184 |
} |
| 185 |
|
| 186 |
if ( ! $is_meta_object || $entry_changed ) { |
| 187 |
$meta_data[ $index ] = $entry; |
| 188 |
} |
| 189 |
} |
| 190 |
|
| 191 |
if ( $dropped && $was_list ) { |
| 192 |
// A list with a hole JSON-encodes as an object; the wire expects a list. |
| 193 |
$meta_data = array_values( $meta_data ); |
| 194 |
} |
| 195 |
|
| 196 |
return $meta_data; |
| 197 |
} |
| 198 |
|
| 199 |
/** |
| 200 |
* Remove an oversized entry from the payload and say so once per key per request. |
| 201 |
* The value itself is never logged. |
| 202 |
* |
| 203 |
* @param array $meta_data Serialized REST meta entries (by reference). |
| 204 |
* @param int|string $index Index of the entry to drop. |
| 205 |
* @param array $entry Array form of the entry (`id`, `key`, `value`). |
| 206 |
*/ |
| 207 |
private static function drop_oversized( array &$meta_data, $index, array $entry ): void { |
| 208 |
unset( $meta_data[ $index ] ); |
| 209 |
self::note_oversized_meta( |
| 210 |
isset( $entry['key'] ) ? (string) $entry['key'] : '', |
| 211 |
(int) ( $entry['id'] ?? 0 ) |
| 212 |
); |
| 213 |
} |
| 214 |
|
| 215 |
/** |
| 216 |
* Record that an oversized meta entry was withheld, once per key per request. |
| 217 |
* The value is never logged. Shared with the v1 lane, which drops the same |
| 218 |
* entries at its own serializer rather than through this class. |
| 219 |
* |
| 220 |
* @param string $key Meta key that was withheld. |
| 221 |
* @param int $meta_id Meta row id, when known. |
| 222 |
*/ |
| 223 |
public static function note_oversized_meta( string $key, int $meta_id = 0 ): void { |
| 224 |
if ( isset( self::$oversized_logged[ $key ] ) ) { |
| 225 |
return; |
| 226 |
} |
| 227 |
self::$oversized_logged[ $key ] = true; |
| 228 |
Logger::warning( |
| 229 |
sprintf( |
| 230 |
'WCPOS sync: dropped oversized meta "%s" (meta id %d) from the POS payload; the stored value exceeds %d nodes or %d bytes and cannot be served to the POS.', |
| 231 |
$key, |
| 232 |
$meta_id, |
| 233 |
self::OVERSIZED_META_NODE_LIMIT, |
| 234 |
self::OVERSIZED_META_BYTE_LIMIT |
| 235 |
) |
| 236 |
); |
| 237 |
} |
| 238 |
|
| 239 |
/** |
| 240 |
* Whether a meta value is too large to serve. Walks iteratively and stops the moment a |
| 241 |
* limit is crossed, so the cost is bounded by the limits, not by the value: a |
| 242 |
* 16-million-element value costs the same as a 20,001-element one. |
| 243 |
* |
| 244 |
* EVERY object is expanded, not just stdClass. WordPress unserializes stored meta, so a |
| 245 |
* value can come back as an instance of some other plugin's class, and json_encode |
| 246 |
* serializes its public properties just the same — a custom object wrapping the |
| 247 |
* multi-million-element array would otherwise walk straight past this check into the |
| 248 |
* encode that killed the request. `get_object_vars()` is called from outside the value's |
| 249 |
* class, so it sees exactly the public properties the encoder will. Traversal is by |
| 250 |
* refcount, never a deep copy. |
| 251 |
* |
| 252 |
* Keys count toward both budgets: a value can be a few entries under enormous keys. |
| 253 |
* |
| 254 |
* A self-referencing object graph terminates on the node limit and reports oversized, |
| 255 |
* which is correct — json_encode cannot represent one either. |
| 256 |
* |
| 257 |
* @param mixed $value Meta value as stored. |
| 258 |
* |
| 259 |
* @return bool |
| 260 |
*/ |
| 261 |
public static function exceeds_value_budget( $value ): bool { |
| 262 |
$unencodable = false; |
| 263 |
$value = self::as_encoded( $value, $unencodable ); |
| 264 |
if ( $unencodable ) { |
| 265 |
return true; |
| 266 |
} |
| 267 |
if ( is_string( $value ) ) { |
| 268 |
return \strlen( $value ) > self::OVERSIZED_META_BYTE_LIMIT; |
| 269 |
} |
| 270 |
if ( ! is_array( $value ) && ! \is_object( $value ) ) { |
| 271 |
return false; |
| 272 |
} |
| 273 |
|
| 274 |
$nodes = 0; |
| 275 |
$bytes = 0; |
| 276 |
$stack = array( is_array( $value ) ? $value : get_object_vars( $value ) ); |
| 277 |
while ( array() !== $stack ) { |
| 278 |
$current = array_pop( $stack ); |
| 279 |
foreach ( $current as $child_key => $child ) { |
| 280 |
++$nodes; |
| 281 |
if ( is_string( $child_key ) ) { |
| 282 |
$bytes += \strlen( $child_key ); |
| 283 |
} |
| 284 |
$child = self::as_encoded( $child, $unencodable ); |
| 285 |
if ( $unencodable ) { |
| 286 |
return true; |
| 287 |
} |
| 288 |
if ( is_string( $child ) ) { |
| 289 |
$bytes += \strlen( $child ); |
| 290 |
} elseif ( is_array( $child ) ) { |
| 291 |
$stack[] = $child; |
| 292 |
} elseif ( \is_object( $child ) ) { |
| 293 |
$stack[] = get_object_vars( $child ); |
| 294 |
} |
| 295 |
if ( $nodes > self::OVERSIZED_META_NODE_LIMIT || $bytes > self::OVERSIZED_META_BYTE_LIMIT ) { |
| 296 |
return true; |
| 297 |
} |
| 298 |
} |
| 299 |
} |
| 300 |
|
| 301 |
return false; |
| 302 |
} |
| 303 |
|
| 304 |
/** |
| 305 |
* What `json_encode()` will actually serialize for a value. |
| 306 |
* |
| 307 |
* A `JsonSerializable` object is encoded from `jsonSerialize()`, NOT from its public |
| 308 |
* properties, so an object can expose nothing and still return a multi-million-element |
| 309 |
* array to the encoder. Budgeting `get_object_vars()` alone would wave exactly that |
| 310 |
* through. The encoder is going to call this method moments later anyway, so calling it |
| 311 |
* here adds no execution that was not already going to happen. |
| 312 |
* |
| 313 |
* A chain deeper than a handful of levels, or one that throws, is reported as |
| 314 |
* unencodable and the entry is withheld: `json_encode()` would fail on it too, and |
| 315 |
* failing there is the fatal this guard exists to prevent. |
| 316 |
* |
| 317 |
* @param mixed $value Value to resolve. |
| 318 |
* @param bool $unencodable Set to true when the value cannot be resolved safely. |
| 319 |
* |
| 320 |
* @return mixed |
| 321 |
*/ |
| 322 |
private static function as_encoded( $value, bool &$unencodable ) { |
| 323 |
$depth = 0; |
| 324 |
while ( $value instanceof \JsonSerializable ) { |
| 325 |
if ( ++$depth > 8 ) { |
| 326 |
$unencodable = true; |
| 327 |
|
| 328 |
return null; |
| 329 |
} |
| 330 |
try { |
| 331 |
$value = $value->jsonSerialize(); |
| 332 |
} catch ( \Throwable $error ) { |
| 333 |
$unencodable = true; |
| 334 |
|
| 335 |
return null; |
| 336 |
} |
| 337 |
} |
| 338 |
|
| 339 |
return $value; |
| 340 |
} |
| 341 |
|
| 342 |
/** |
| 343 |
* Convert JSON objects to arrays unless doing so would change their wire shape. |
| 344 |
* |
| 345 |
* @param mixed $value Decoded JSON value. |
| 346 |
* |
| 347 |
* @return mixed |
| 348 |
*/ |
| 349 |
private static function preserve_json_object_shape( $value ) { |
| 350 |
if ( is_array( $value ) ) { |
| 351 |
return array_map( array( __CLASS__, __FUNCTION__ ), $value ); |
| 352 |
} |
| 353 |
|
| 354 |
if ( ! $value instanceof \stdClass ) { |
| 355 |
return $value; |
| 356 |
} |
| 357 |
|
| 358 |
/** |
| 359 |
* Decoded object properties. |
| 360 |
* |
| 361 |
* @var array<int|string, mixed> $properties |
| 362 |
*/ |
| 363 |
$properties = array_map( array( __CLASS__, __FUNCTION__ ), get_object_vars( $value ) ); |
| 364 |
|
| 365 |
return array() === $properties || array_values( $properties ) === $properties |
| 366 |
? (object) $properties |
| 367 |
: $properties; |
| 368 |
} |
| 369 |
} |
| 370 |
|
