Completed
Push — sync/checksum-sort ( f97b1a...b30f1d )
by
unknown
15:47 queued 08:21
created

Module::get_metadata()   A

Complexity

Conditions 2
Paths 2

Size

Total Lines 21

Duplication

Lines 0
Ratio 0 %

Importance

Changes 0
Metric Value
cc 2
nc 2
nop 3
dl 0
loc 21
rs 9.584
c 0
b 0
f 0
1
<?php
2
/**
3
 * A base abstraction of a sync module.
4
 *
5
 * @package automattic/jetpack-sync
6
 */
7
8
namespace Automattic\Jetpack\Sync\Modules;
9
10
use Automattic\Jetpack\Sync\Listener;
11
use Automattic\Jetpack\Sync\Replicastore;
12
use Automattic\Jetpack\Sync\Sender;
13
use Automattic\Jetpack\Sync\Settings;
14
15
/**
16
 * Basic methods implemented by Jetpack Sync extensions.
17
 *
18
 * @abstract
19
 */
20
abstract class Module {
21
	/**
22
	 * Number of items per chunk when grouping objects for performance reasons.
23
	 *
24
	 * @access public
25
	 *
26
	 * @var int
27
	 */
28
	const ARRAY_CHUNK_SIZE = 10;
29
30
	/**
31
	 * Sync module name.
32
	 *
33
	 * @access public
34
	 *
35
	 * @return string
36
	 */
37
	abstract public function name();
38
39
	/**
40
	 * The id field in the database.
41
	 *
42
	 * @access public
43
	 *
44
	 * @return string
45
	 */
46
	public function id_field() {
47
		return 'ID';
48
	}
49
50
	/**
51
	 * The table in the database.
52
	 *
53
	 * @access public
54
	 *
55
	 * @return string|bool
56
	 */
57
	public function table_name() {
58
		return false;
59
	}
60
61
	// phpcs:disable VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
62
63
	/**
64
	 * Retrieve a sync object by its ID.
65
	 *
66
	 * @access public
67
	 *
68
	 * @param string $object_type Type of the sync object.
69
	 * @param int    $id          ID of the sync object.
70
	 * @return mixed Object, or false if the object is invalid.
71
	 */
72
	public function get_object_by_id( $object_type, $id ) {
73
		return false;
74
	}
75
76
	/**
77
	 * Initialize callables action listeners.
78
	 * Override these to set up listeners and set/reset data/defaults.
79
	 *
80
	 * @access public
81
	 *
82
	 * @param callable $callable Action handler callable.
83
	 */
84
	public function init_listeners( $callable ) {
85
	}
86
87
	/**
88
	 * Initialize module action listeners for full sync.
89
	 *
90
	 * @access public
91
	 *
92
	 * @param callable $callable Action handler callable.
93
	 */
94
	public function init_full_sync_listeners( $callable ) {
95
	}
96
97
	/**
98
	 * Initialize the module in the sender.
99
	 *
100
	 * @access public
101
	 */
102
	public function init_before_send() {
103
	}
104
105
	/**
106
	 * Set module defaults.
107
	 *
108
	 * @access public
109
	 */
110
	public function set_defaults() {
111
	}
112
113
	/**
114
	 * Perform module cleanup.
115
	 * Usually triggered when uninstalling the plugin.
116
	 *
117
	 * @access public
118
	 */
119
	public function reset_data() {
120
	}
121
122
	/**
123
	 * Enqueue the module actions for full sync.
124
	 *
125
	 * @access public
126
	 *
127
	 * @param array   $config               Full sync configuration for this sync module.
128
	 * @param int     $max_items_to_enqueue Maximum number of items to enqueue.
129
	 * @param boolean $state                True if full sync has finished enqueueing this module, false otherwise.
130
	 * @return array  Number of actions enqueued, and next module state.
131
	 */
132
	public function enqueue_full_sync_actions( $config, $max_items_to_enqueue, $state ) {
133
		// In subclasses, return the number of actions enqueued, and next module state (true == done).
134
		return array( null, true );
135
	}
136
137
	/**
138
	 * Retrieve an estimated number of actions that will be enqueued.
139
	 *
140
	 * @access public
141
	 *
142
	 * @param array $config Full sync configuration for this sync module.
143
	 * @return array Number of items yet to be enqueued.
144
	 */
145
	public function estimate_full_sync_actions( $config ) {
146
		// In subclasses, return the number of items yet to be enqueued.
147
		return null;
148
	}
149
150
	// phpcs:enable VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
151
152
	/**
153
	 * Retrieve the actions that will be sent for this module during a full sync.
154
	 *
155
	 * @access public
156
	 *
157
	 * @return array Full sync actions of this module.
158
	 */
159
	public function get_full_sync_actions() {
160
		return array();
161
	}
162
163
	/**
164
	 * Get the number of actions that we care about.
165
	 *
166
	 * @access protected
167
	 *
168
	 * @param array $action_names     Action names we're interested in.
169
	 * @param array $actions_to_count Unfiltered list of actions we want to count.
170
	 * @return array Number of actions that we're interested in.
171
	 */
172
	protected function count_actions( $action_names, $actions_to_count ) {
173
		return count( array_intersect( $action_names, $actions_to_count ) );
174
	}
175
176
	/**
177
	 * Calculate the checksum of one or more values.
178
	 *
179
	 * @access protected
180
	 *
181
	 * @param mixed $values Values to calculate checksum for.
182
	 * @param bool  $sort If $values should have ksort called on it.
183
	 * @return int The checksum.
184
	 */
185
	protected function get_check_sum( $values, $sort = true ) {
186
		// Associative array order changes the generated checksum value.
187
		if ( $sort && is_array( $values ) ) {
188
			$this->recursive_ksort( $values );
189
		}
190
		return crc32( wp_json_encode( jetpack_json_wrap( $values ) ) );
191
	}
192
193
	/**
194
	 * Recursively call ksort on an Array
195
	 *
196
	 * @param array $values Array.
197
	 */
198
	private function recursive_ksort( $values ) {
199
		ksort( $values );
200
		foreach ( $values as $value ) {
201
			if ( is_array( $value ) ) {
202
				$this->recursive_ksort( $value );
203
			}
204
		}
205
	}
206
207
	/**
208
	 * Whether a particular checksum in a set of checksums is valid.
209
	 *
210
	 * @access protected
211
	 *
212
	 * @param array  $sums_to_check Array of checksums.
213
	 * @param string $name          Name of the checksum.
214
	 * @param int    $new_sum       Checksum to compare against.
215
	 * @return boolean Whether the checksum is valid.
216
	 */
217
	protected function still_valid_checksum( $sums_to_check, $name, $new_sum ) {
218
		if ( isset( $sums_to_check[ $name ] ) && $sums_to_check[ $name ] === $new_sum ) {
219
			return true;
220
		}
221
222
		return false;
223
	}
224
225
	/**
226
	 * Enqueue all items of a sync type as an action.
227
	 *
228
	 * @access protected
229
	 *
230
	 * @param string  $action_name          Name of the action.
231
	 * @param string  $table_name           Name of the database table.
232
	 * @param string  $id_field             Name of the ID field in the database.
233
	 * @param string  $where_sql            The SQL WHERE clause to filter to the desired items.
234
	 * @param int     $max_items_to_enqueue Maximum number of items to enqueue in the same time.
235
	 * @param boolean $state                Whether enqueueing has finished.
236
	 * @return array Array, containing the number of chunks and TRUE, indicating enqueueing has finished.
237
	 */
238
	protected function enqueue_all_ids_as_action( $action_name, $table_name, $id_field, $where_sql, $max_items_to_enqueue, $state ) {
239
		global $wpdb;
240
241
		if ( ! $where_sql ) {
242
			$where_sql = '1 = 1';
243
		}
244
245
		$items_per_page        = 1000;
246
		$page                  = 1;
247
		$chunk_count           = 0;
248
		$previous_interval_end = $state ? $state : '~0';
249
		$listener              = Listener::get_instance();
250
251
		// Count down from max_id to min_id so we get newest posts/comments/etc first.
252
		// phpcs:ignore WordPress.CodeAnalysis.AssignmentInCondition.FoundInWhileCondition, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
253
		while ( $ids = $wpdb->get_col( "SELECT {$id_field} FROM {$table_name} WHERE {$where_sql} AND {$id_field} < {$previous_interval_end} ORDER BY {$id_field} DESC LIMIT {$items_per_page}" ) ) {
254
			// Request posts in groups of N for efficiency.
255
			$chunked_ids = array_chunk( $ids, self::ARRAY_CHUNK_SIZE );
256
257
			// If we hit our row limit, process and return.
258
			if ( $chunk_count + count( $chunked_ids ) >= $max_items_to_enqueue ) {
259
				$remaining_items_count                      = $max_items_to_enqueue - $chunk_count;
260
				$remaining_items                            = array_slice( $chunked_ids, 0, $remaining_items_count );
261
				$remaining_items_with_previous_interval_end = $this->get_chunks_with_preceding_end( $remaining_items, $previous_interval_end );
262
				$listener->bulk_enqueue_full_sync_actions( $action_name, $remaining_items_with_previous_interval_end );
263
264
				$last_chunk = end( $remaining_items );
265
				return array( $remaining_items_count + $chunk_count, end( $last_chunk ) );
266
			}
267
			$chunked_ids_with_previous_end = $this->get_chunks_with_preceding_end( $chunked_ids, $previous_interval_end );
268
269
			$listener->bulk_enqueue_full_sync_actions( $action_name, $chunked_ids_with_previous_end );
270
271
			$chunk_count += count( $chunked_ids );
272
			$page++;
273
			// The $ids are ordered in descending order.
274
			$previous_interval_end = end( $ids );
275
		}
276
277
		if ( $wpdb->last_error ) {
278
			// return the values that were passed in so all these chunks get retried.
279
			return array( $max_items_to_enqueue, $state );
280
		}
281
282
		return array( $chunk_count, true );
283
	}
284
285
	/**
286
	 * Given the Module Full Sync Configuration and Status return the next chunk of items to send.
287
	 *
288
	 * @param array $config This module Full Sync configuration.
289
	 * @param array $status This module Full Sync status.
290
	 * @param int   $chunk_size Chunk size.
291
	 *
292
	 * @return array|object|null
293
	 */
294
	public function get_next_chunk( $config, $status, $chunk_size ) {
295
		global $wpdb;
296
		return $wpdb->get_col(
297
			<<<SQL
298
SELECT {$this->id_field()}
299
FROM {$wpdb->{$this->table_name()}}
300
WHERE {$this->get_where_sql( $config )}
301
AND {$this->id_field()} < {$status['last_sent']}
302
ORDER BY {$this->id_field()}
303
DESC LIMIT {$chunk_size}
304
SQL
305
		);
306
	}
307
308
	/**
309
	 * Return the initial last sent object.
310
	 *
311
	 * @return string|array initial status.
312
	 */
313
	public function get_initial_last_sent() {
314
		return '~0';
315
	}
316
317
	/**
318
	 * Immediately send all items of a sync type as an action.
319
	 *
320
	 * @access protected
321
	 *
322
	 * @param string $config Full sync configuration for this module.
323
	 * @param array  $status the current module full sync status.
324
	 * @param float  $send_until timestamp until we want this request to send full sync events.
325
	 *
326
	 * @return array Status, the module full sync status updated.
327
	 */
328
	public function send_full_sync_actions( $config, $status, $send_until ) {
329
		global $wpdb;
330
331
		if ( empty( $status['last_sent'] ) ) {
332
			$status['last_sent'] = $this->get_initial_last_sent();
333
		}
334
335
		$limits = Settings::get_setting( 'full_sync_limits' )[ $this->name() ];
336
337
		$chunks_sent = 0;
338
		// phpcs:ignore WordPress.CodeAnalysis.AssignmentInCondition.FoundInWhileCondition
339
		while ( $objects = $this->get_next_chunk( $config, $status, $limits['chunk_size'] ) ) {
0 ignored issues
show
Documentation introduced by
$config is of type string, but the function expects a array.

It seems like the type of the argument is not accepted by the function/method which you are calling.

In some cases, in particular if PHP’s automatic type-juggling kicks in this might be fine. In other cases, however this might be a bug.

We suggest to add an explicit type cast like in the following example:

function acceptsInteger($int) { }

$x = '123'; // string "123"

// Instead of
acceptsInteger($x);

// we recommend to use
acceptsInteger((integer) $x);
Loading history...
340
			if ( $chunks_sent++ === $limits['max_chunks'] || microtime( true ) >= $send_until ) {
341
				return $status;
342
			}
343
344
			$result = $this->send_action( 'jetpack_full_sync_' . $this->name(), array( $objects, $status['last_sent'] ) );
345
346
			if ( is_wp_error( $result ) || $wpdb->last_error ) {
347
				return $status;
348
			}
349
			// The $ids are ordered in descending order.
350
			$status['last_sent'] = end( $objects );
351
			$status['sent']     += count( $objects );
352
		}
353
354
		if ( ! $wpdb->last_error ) {
355
			$status['finished'] = true;
356
		}
357
358
		return $status;
359
	}
360
361
362
	/**
363
	 * Immediately sends a single item without firing or enqueuing it
364
	 *
365
	 * @param string $action_name The action.
366
	 * @param array  $data The data associated with the action.
0 ignored issues
show
Documentation introduced by
Should the type for parameter $data not be array|null?

This check looks for @param annotations where the type inferred by our type inference engine differs from the declared type.

It makes a suggestion as to what type it considers more descriptive.

Most often this is a case of a parameter that can be null in addition to its declared types.

Loading history...
367
	 */
368
	public function send_action( $action_name, $data = null ) {
369
		$sender = Sender::get_instance();
370
		return $sender->send_action( $action_name, $data );
371
	}
372
373
	/**
374
	 * Retrieve chunk IDs with previous interval end.
375
	 *
376
	 * @access protected
377
	 *
378
	 * @param array $chunks                All remaining items.
379
	 * @param int   $previous_interval_end The last item from the previous interval.
380
	 * @return array Chunk IDs with the previous interval end.
381
	 */
382
	protected function get_chunks_with_preceding_end( $chunks, $previous_interval_end ) {
383
		$chunks_with_ends = array();
384
		foreach ( $chunks as $chunk ) {
385
			$chunks_with_ends[] = array(
386
				'ids'          => $chunk,
387
				'previous_end' => $previous_interval_end,
388
			);
389
			// Chunks are ordered in descending order.
390
			$previous_interval_end = end( $chunk );
391
		}
392
		return $chunks_with_ends;
393
	}
394
395
	/**
396
	 * Get metadata of a particular object type within the designated meta key whitelist.
397
	 *
398
	 * @access protected
399
	 *
400
	 * @todo Refactor to use $wpdb->prepare() on the SQL query.
401
	 *
402
	 * @param array  $ids                Object IDs.
403
	 * @param string $meta_type          Meta type.
404
	 * @param array  $meta_key_whitelist Meta key whitelist.
405
	 * @return array Unserialized meta values.
406
	 */
407
	protected function get_metadata( $ids, $meta_type, $meta_key_whitelist ) {
408
		global $wpdb;
409
		$table = _get_meta_table( $meta_type );
410
		$id    = $meta_type . '_id';
411
		if ( ! $table ) {
412
			return array();
413
		}
414
415
		$private_meta_whitelist_sql = "'" . implode( "','", array_map( 'esc_sql', $meta_key_whitelist ) ) . "'";
416
417
		return array_map(
418
			array( $this, 'unserialize_meta' ),
419
			$wpdb->get_results(
420
				// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQL.NotPrepared
421
				"SELECT $id, meta_key, meta_value, meta_id FROM $table WHERE $id IN ( " . implode( ',', wp_parse_id_list( $ids ) ) . ' )' .
422
				" AND meta_key IN ( $private_meta_whitelist_sql ) ",
423
				// phpcs:enable WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQL.NotPrepared
424
				OBJECT
425
			)
426
		);
427
	}
428
429
	/**
430
	 * Initialize listeners for the particular meta type.
431
	 *
432
	 * @access public
433
	 *
434
	 * @param string   $meta_type Meta type.
435
	 * @param callable $callable  Action handler callable.
436
	 */
437
	public function init_listeners_for_meta_type( $meta_type, $callable ) {
438
		add_action( "added_{$meta_type}_meta", $callable, 10, 4 );
439
		add_action( "updated_{$meta_type}_meta", $callable, 10, 4 );
440
		add_action( "deleted_{$meta_type}_meta", $callable, 10, 4 );
441
	}
442
443
	/**
444
	 * Initialize meta whitelist handler for the particular meta type.
445
	 *
446
	 * @access public
447
	 *
448
	 * @param string   $meta_type         Meta type.
449
	 * @param callable $whitelist_handler Action handler callable.
450
	 */
451
	public function init_meta_whitelist_handler( $meta_type, $whitelist_handler ) {
452
		add_filter( "jetpack_sync_before_enqueue_added_{$meta_type}_meta", $whitelist_handler );
453
		add_filter( "jetpack_sync_before_enqueue_updated_{$meta_type}_meta", $whitelist_handler );
454
		add_filter( "jetpack_sync_before_enqueue_deleted_{$meta_type}_meta", $whitelist_handler );
455
	}
456
457
	/**
458
	 * Retrieve the term relationships for the specified object IDs.
459
	 *
460
	 * @access protected
461
	 *
462
	 * @todo This feels too specific to be in the abstract sync Module class. Move it?
463
	 *
464
	 * @param array $ids Object IDs.
465
	 * @return array Term relationships - object ID and term taxonomy ID pairs.
466
	 */
467
	protected function get_term_relationships( $ids ) {
468
		global $wpdb;
469
470
		// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared
471
		return $wpdb->get_results( "SELECT object_id, term_taxonomy_id FROM $wpdb->term_relationships WHERE object_id IN ( " . implode( ',', wp_parse_id_list( $ids ) ) . ' )', OBJECT );
472
	}
473
474
	/**
475
	 * Unserialize the value of a meta object, if necessary.
476
	 *
477
	 * @access public
478
	 *
479
	 * @param object $meta Meta object.
480
	 * @return object Meta object with possibly unserialized value.
481
	 */
482
	public function unserialize_meta( $meta ) {
483
		$meta->meta_value = maybe_unserialize( $meta->meta_value );
484
		return $meta;
485
	}
486
487
	/**
488
	 * Retrieve a set of objects by their IDs.
489
	 *
490
	 * @access public
491
	 *
492
	 * @param string $object_type Object type.
493
	 * @param array  $ids         Object IDs.
494
	 * @return array Array of objects.
495
	 */
496
	public function get_objects_by_id( $object_type, $ids ) {
497
		if ( empty( $ids ) || empty( $object_type ) ) {
498
			return array();
499
		}
500
501
		$objects = array();
502
		foreach ( (array) $ids as $id ) {
503
			$object = $this->get_object_by_id( $object_type, $id );
504
505
			// Only add object if we have the object.
506
			if ( $object ) {
507
				$objects[ $id ] = $object;
508
			}
509
		}
510
511
		return $objects;
512
	}
513
514
	/**
515
	 * Gets a list of minimum and maximum object ids for each batch based on the given batch size.
516
	 *
517
	 * @access public
518
	 *
519
	 * @param int         $batch_size The batch size for objects.
520
	 * @param string|bool $where_sql  The sql where clause minus 'WHERE', or false if no where clause is needed.
521
	 *
522
	 * @return array|bool An array of min and max ids for each batch. FALSE if no table can be found.
523
	 */
524
	public function get_min_max_object_ids_for_batches( $batch_size, $where_sql = false ) {
525
		global $wpdb;
526
527
		if ( ! $this->table_name() ) {
528
			return false;
529
		}
530
531
		$results      = array();
532
		$table        = $wpdb->{$this->table_name()};
533
		$current_max  = 0;
534
		$current_min  = 1;
535
		$id_field     = $this->id_field();
536
		$replicastore = new Replicastore();
537
538
		$total = $replicastore->get_min_max_object_id(
539
			$id_field,
540
			$table,
541
			$where_sql,
0 ignored issues
show
Bug introduced by
It seems like $where_sql defined by parameter $where_sql on line 524 can also be of type boolean; however, Automattic\Jetpack\Sync\...get_min_max_object_id() does only seem to accept string, maybe add an additional type check?

This check looks at variables that have been passed in as parameters and are passed out again to other methods.

If the outgoing method call has stricter type requirements than the method itself, an issue is raised.

An additional type check may prevent trouble.

Loading history...
542
			false
0 ignored issues
show
Documentation introduced by
false is of type boolean, but the function expects a integer.

It seems like the type of the argument is not accepted by the function/method which you are calling.

In some cases, in particular if PHP’s automatic type-juggling kicks in this might be fine. In other cases, however this might be a bug.

We suggest to add an explicit type cast like in the following example:

function acceptsInteger($int) { }

$x = '123'; // string "123"

// Instead of
acceptsInteger($x);

// we recommend to use
acceptsInteger((integer) $x);
Loading history...
543
		);
544
545
		while ( $total->max > $current_max ) {
546
			$where  = $where_sql ?
547
				$where_sql . " AND $id_field > $current_max" :
548
				"$id_field > $current_max";
549
			$result = $replicastore->get_min_max_object_id(
550
				$id_field,
551
				$table,
552
				$where,
553
				$batch_size
554
			);
555
			if ( empty( $result->min ) && empty( $result->max ) ) {
556
				// Our query produced no min and max. We can assume the min from the previous query,
557
				// and the total max we found in the initial query.
558
				$current_max = (int) $total->max;
559
				$result      = (object) array(
560
					'min' => $current_min,
561
					'max' => $current_max,
562
				);
563
			} else {
564
				$current_min = (int) $result->min;
565
				$current_max = (int) $result->max;
566
			}
567
			$results[] = $result;
568
		}
569
570
		return $results;
571
	}
572
573
	/**
574
	 * Return Total number of objects.
575
	 *
576
	 * @param array $config Full Sync config.
577
	 *
578
	 * @return int total
579
	 */
580
	public function total( $config ) {
581
		global $wpdb;
582
		$table = $wpdb->{$this->table_name()};
583
		$where = $this->get_where_sql( $config );
584
585
		// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
586
		return $wpdb->get_var( "SELECT COUNT(*) FROM $table WHERE $where" );
587
	}
588
589
	/**
590
	 * Retrieve the WHERE SQL clause based on the module config.
591
	 *
592
	 * @access public
593
	 *
594
	 * @param array $config Full sync configuration for this sync module.
595
	 * @return string WHERE SQL clause, or `null` if no comments are specified in the module config.
596
	 */
597
	public function get_where_sql( $config ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
598
		return '1=1';
599
	}
600
601
}
602