ConfirmationEmailTemplate
1 month ago
ImportExport
2 weeks ago
RestApi
3 days ago
Statistics
1 month ago
BulkActionController.php
3 days ago
BulkActionException.php
1 month ago
BulkConfirmationEmailResender.php
2 months ago
ConfirmationEmailCustomizer.php
2 months ago
ConfirmationEmailMailer.php
2 months ago
ConfirmationEmailResolver.php
2 months ago
EngagementDataBackfiller.php
2 months ago
InactiveSubscribersController.php
4 days ago
LinkTokens.php
2 months ago
NewSubscriberNotificationMailer.php
2 months ago
RequiredCustomFieldValidator.php
2 months ago
SegmentsCountRecalculator.php
2 weeks ago
Source.php
2 months ago
SubscriberActions.php
2 months ago
SubscriberCustomFieldRepository.php
3 years ago
SubscriberIPsRepository.php
2 years ago
SubscriberLimitNotificationEvaluator.php
2 months ago
SubscriberLimitNotificationMailer.php
2 months ago
SubscriberLimitNotificationScheduler.php
2 months ago
SubscriberListingRepository.php
2 weeks ago
SubscriberPersonalDataEraser.php
2 months ago
SubscriberSaveController.php
3 days ago
SubscriberSegmentRepository.php
2 weeks ago
SubscriberSubscribeController.php
2 weeks ago
SubscriberTagRepository.php
4 years ago
SubscribersCountsController.php
2 weeks ago
SubscribersEmailCountsController.php
2 weeks ago
SubscribersRepository.php
3 days ago
TrackingConsentController.php
4 days ago
index.php
3 years ago
SegmentsCountRecalculator.php
276 lines
| 1 | <?php declare(strict_types = 1); |
| 2 | |
| 3 | namespace MailPoet\Subscribers; |
| 4 | |
| 5 | if (!defined('ABSPATH')) exit; |
| 6 | |
| 7 | |
| 8 | use MailPoet\Cron\CronWorkerScheduler; |
| 9 | use MailPoet\Cron\Workers\SubscribersSegmentsCountSync; |
| 10 | use MailPoet\DI\ContainerWrapper; |
| 11 | use MailPoet\Doctrine\WPDB\Connection; |
| 12 | use MailPoet\Entities\SegmentEntity; |
| 13 | use MailPoet\Entities\SubscriberEntity; |
| 14 | use MailPoet\Entities\SubscriberSegmentEntity; |
| 15 | use MailPoetVendor\Doctrine\DBAL\ArrayParameterType; |
| 16 | use MailPoetVendor\Doctrine\DBAL\ParameterType; |
| 17 | use MailPoetVendor\Doctrine\ORM\EntityManager; |
| 18 | |
| 19 | /** |
| 20 | * Keeps SubscriberEntity::$segmentsCount in sync. |
| 21 | * |
| 22 | * segments_count is the number of the subscriber's subscribed memberships in |
| 23 | * non-deleted segments. It mirrors the anti-join that used to power the |
| 24 | * "Subscribers without a list" count, so the read can become |
| 25 | * `WHERE segments_count = 0` instead of scanning every subscriber. |
| 26 | * |
| 27 | * Every recalculation re-derives the value from subscriber_segment + segments, |
| 28 | * so it is idempotent: it is safe to call from several write paths, to call |
| 29 | * twice, or to run concurrently with the backfill — the value always converges. |
| 30 | * The semantics intentionally match the previous query exactly: only |
| 31 | * status = 'subscribed' memberships in segments with deleted_at IS NULL are |
| 32 | * counted, with no filtering by segment type (WP/WooCommerce segments count too). |
| 33 | */ |
| 34 | class SegmentsCountRecalculator { |
| 35 | /** Subscribers touched per UPDATE when recalculating large/segment-wide sets. */ |
| 36 | public const BATCH_SIZE = 10000; |
| 37 | |
| 38 | /** |
| 39 | * When a single segment change would recompute at least this many memberships |
| 40 | * inline, defer to the background sweep instead. Read via static:: so tests |
| 41 | * can lower it without inserting hundreds of thousands of rows. |
| 42 | */ |
| 43 | protected const DEFER_THRESHOLD = 200000; |
| 44 | |
| 45 | /** @var EntityManager */ |
| 46 | private $entityManager; |
| 47 | |
| 48 | public function __construct( |
| 49 | EntityManager $entityManager |
| 50 | ) { |
| 51 | $this->entityManager = $entityManager; |
| 52 | } |
| 53 | |
| 54 | public function getDeferThreshold(): int { |
| 55 | return static::DEFER_THRESHOLD; |
| 56 | } |
| 57 | |
| 58 | /** |
| 59 | * Hand a recalculation off to the background sweep (scheduled to run as soon |
| 60 | * as possible) instead of doing it inline. Used when a single change touches |
| 61 | * too many subscribers to recompute within one request. The sweep re-derives |
| 62 | * every subscriber's count from source, so it converges regardless of what |
| 63 | * the deferred change was. |
| 64 | * |
| 65 | * CronWorkerScheduler is resolved lazily rather than injected: it transitively |
| 66 | * depends on SegmentsRepository, which depends on this class, so a constructor |
| 67 | * dependency would form a circular reference (the same reason SimpleWorker |
| 68 | * pulls it from the container). |
| 69 | */ |
| 70 | public function scheduleBackgroundRecalculation(): void { |
| 71 | // The sweep worker no-ops on SQLite and reads never trust the column there, |
| 72 | // so there is nothing to schedule. |
| 73 | if (Connection::isSQLite()) { |
| 74 | return; |
| 75 | } |
| 76 | ContainerWrapper::getInstance() |
| 77 | ->get(CronWorkerScheduler::class) |
| 78 | ->scheduleImmediatelyIfNotRunning(SubscribersSegmentsCountSync::TASK_TYPE); |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * Recalculate the count for an explicit set of subscribers. |
| 83 | * |
| 84 | * @param int[] $subscriberIds |
| 85 | */ |
| 86 | public function recalculateForSubscribers(array $subscriberIds): void { |
| 87 | // The UPDATE ... LEFT JOIN syntax below is not supported by the SQLite |
| 88 | // integration used in WordPress Playground. Reads stay on the anti-join |
| 89 | // there because the sync worker never flips the backfill flag (see |
| 90 | // SubscribersSegmentsCountSync::processTaskStrategy()). |
| 91 | if (Connection::isSQLite()) { |
| 92 | return; |
| 93 | } |
| 94 | |
| 95 | $subscriberIds = array_values(array_unique($subscriberIds)); |
| 96 | if ($subscriberIds === []) { |
| 97 | return; |
| 98 | } |
| 99 | |
| 100 | $subscribersTable = $this->getTableName(SubscriberEntity::class); |
| 101 | $membershipSelect = $this->membershipCountSubquery('ssg.subscriber_id IN (:ids)'); |
| 102 | $connection = $this->entityManager->getConnection(); |
| 103 | |
| 104 | foreach (array_chunk($subscriberIds, self::BATCH_SIZE) as $chunk) { |
| 105 | $connection->executeStatement( |
| 106 | "UPDATE {$subscribersTable} s |
| 107 | LEFT JOIN ({$membershipSelect}) m ON m.subscriber_id = s.id |
| 108 | SET s.segments_count = IFNULL(m.c, 0) |
| 109 | WHERE s.id IN (:ids)", |
| 110 | ['ids' => $chunk], |
| 111 | ['ids' => ArrayParameterType::INTEGER] |
| 112 | ); |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | /** |
| 117 | * Recalculate the count for an inclusive range of subscriber ids. |
| 118 | * Used by the backfill and reconcile workers. |
| 119 | */ |
| 120 | public function recalculateForIdRange(int $minId, int $maxId): void { |
| 121 | // See recalculateForSubscribers(): UPDATE ... LEFT JOIN is unsupported on SQLite. |
| 122 | if (Connection::isSQLite()) { |
| 123 | return; |
| 124 | } |
| 125 | |
| 126 | if ($minId > $maxId) { |
| 127 | return; |
| 128 | } |
| 129 | |
| 130 | $subscribersTable = $this->getTableName(SubscriberEntity::class); |
| 131 | $membershipSelect = $this->membershipCountSubquery('ssg.subscriber_id BETWEEN :minId AND :maxId'); |
| 132 | |
| 133 | $this->entityManager->getConnection()->executeStatement( |
| 134 | "UPDATE {$subscribersTable} s |
| 135 | LEFT JOIN ({$membershipSelect}) m ON m.subscriber_id = s.id |
| 136 | SET s.segments_count = IFNULL(m.c, 0) |
| 137 | WHERE s.id BETWEEN :minId AND :maxId", |
| 138 | ['minId' => $minId, 'maxId' => $maxId] |
| 139 | ); |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * Recalculate the count for every subscriber that has a membership in the |
| 144 | * given segment. Used when a segment is trashed, restored or deleted, which |
| 145 | * changes the count of all of its members at once. |
| 146 | */ |
| 147 | public function recalculateForSegment(int $segmentId, bool $subscribedOnly = true): void { |
| 148 | $this->recalculateForSegments([$segmentId], $subscribedOnly); |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Recalculate the count for every subscriber that has a membership in any of |
| 153 | * the given segments. Used when segments are trashed, restored or deleted, |
| 154 | * which changes the count of all of their members at once. |
| 155 | * |
| 156 | * Members are walked in keyset-paginated batches rather than materialized into |
| 157 | * one array, so this stays memory-safe even on multi-million-member segments. |
| 158 | * |
| 159 | * $subscribedOnly = true (default): only walk members whose current |
| 160 | * subscriber_segment.status = 'subscribed'. Safe when the segment's |
| 161 | * deleted_at changed but no membership statuses changed — non-subscribed |
| 162 | * members were never counted and their recomputation is a no-op. |
| 163 | * |
| 164 | * $subscribedOnly = false: walk all members regardless of status. Required |
| 165 | * when the caller performed raw-SQL writes that may have changed membership |
| 166 | * statuses (e.g. the WooCommerce sync), so subscribers transitioning away |
| 167 | * from subscribed must also be recomputed. |
| 168 | * |
| 169 | * @param int[] $segmentIds |
| 170 | */ |
| 171 | public function recalculateForSegments(array $segmentIds, bool $subscribedOnly = true): void { |
| 172 | // recalculateForSubscribers() is a no-op on SQLite, so skip the walk too. |
| 173 | if (Connection::isSQLite()) { |
| 174 | return; |
| 175 | } |
| 176 | |
| 177 | $segmentIds = array_values(array_unique(array_map('intval', $segmentIds))); |
| 178 | if ($segmentIds === []) { |
| 179 | return; |
| 180 | } |
| 181 | |
| 182 | // Recomputing a multi-million-member segment inline would blow the request |
| 183 | // budget, so hand the largest changes to the background sweep instead. |
| 184 | if ($this->countSegmentMembers($segmentIds, $subscribedOnly) >= $this->getDeferThreshold()) { |
| 185 | $this->scheduleBackgroundRecalculation(); |
| 186 | return; |
| 187 | } |
| 188 | |
| 189 | $subscriberSegmentTable = $this->getTableName(SubscriberSegmentEntity::class); |
| 190 | $connection = $this->entityManager->getConnection(); |
| 191 | |
| 192 | $lastId = 0; |
| 193 | do { |
| 194 | $batchSize = self::BATCH_SIZE; |
| 195 | $sql = "SELECT DISTINCT subscriber_id FROM {$subscriberSegmentTable} |
| 196 | WHERE segment_id IN (:segmentIds) AND subscriber_id > :lastId"; |
| 197 | $params = ['segmentIds' => $segmentIds, 'lastId' => $lastId]; |
| 198 | $types = ['segmentIds' => ArrayParameterType::INTEGER, 'lastId' => ParameterType::INTEGER]; |
| 199 | if ($subscribedOnly) { |
| 200 | $sql .= ' AND status = :status'; |
| 201 | $params['status'] = SubscriberEntity::STATUS_SUBSCRIBED; |
| 202 | $types['status'] = ParameterType::STRING; |
| 203 | } |
| 204 | $sql .= ' ORDER BY subscriber_id ASC LIMIT ' . $batchSize; |
| 205 | $ids = $connection->executeQuery($sql, $params, $types)->fetchFirstColumn(); |
| 206 | |
| 207 | if ($ids === []) { |
| 208 | break; |
| 209 | } |
| 210 | |
| 211 | $subscriberIds = array_map(function ($id): int { |
| 212 | return is_numeric($id) ? (int)$id : 0; |
| 213 | }, $ids); |
| 214 | $this->recalculateForSubscribers($subscriberIds); |
| 215 | $lastId = (int)end($subscriberIds); |
| 216 | } while (count($ids) === self::BATCH_SIZE); |
| 217 | } |
| 218 | |
| 219 | /** |
| 220 | * Count the memberships a segment change would touch. Uses COUNT(*) rather |
| 221 | * than COUNT(DISTINCT subscriber_id): a subscriber shared across several of |
| 222 | * the given segments is counted more than once, but an over-count only makes |
| 223 | * the deferral threshold trip slightly earlier, which is safe. |
| 224 | * |
| 225 | * When $type is null the query stays on the segment_id index with no join, |
| 226 | * which is what the recalculation path wants. Pass a $type to scope the count |
| 227 | * to segments of that type (joining the segments table), matching a |
| 228 | * type-scoped delete. |
| 229 | * |
| 230 | * @param int[] $segmentIds |
| 231 | */ |
| 232 | public function countSegmentMembers(array $segmentIds, bool $subscribedOnly, ?string $type = null): int { |
| 233 | if ($segmentIds === []) { |
| 234 | return 0; |
| 235 | } |
| 236 | |
| 237 | $subscriberSegmentTable = $this->getTableName(SubscriberSegmentEntity::class); |
| 238 | $sql = "SELECT COUNT(*) FROM {$subscriberSegmentTable} ss"; |
| 239 | $params = ['segmentIds' => $segmentIds]; |
| 240 | $types = ['segmentIds' => ArrayParameterType::INTEGER]; |
| 241 | if ($type !== null) { |
| 242 | $segmentsTable = $this->getTableName(SegmentEntity::class); |
| 243 | $sql .= " JOIN {$segmentsTable} s ON ss.segment_id = s.id AND s.type = :type"; |
| 244 | $params['type'] = $type; |
| 245 | $types['type'] = ParameterType::STRING; |
| 246 | } |
| 247 | $sql .= " WHERE ss.segment_id IN (:segmentIds)"; |
| 248 | if ($subscribedOnly) { |
| 249 | $sql .= ' AND ss.status = :status'; |
| 250 | $params['status'] = SubscriberEntity::STATUS_SUBSCRIBED; |
| 251 | $types['status'] = ParameterType::STRING; |
| 252 | } |
| 253 | $count = $this->entityManager->getConnection()->executeQuery($sql, $params, $types)->fetchOne(); |
| 254 | return is_numeric($count) ? (int)$count : 0; |
| 255 | } |
| 256 | |
| 257 | private function membershipCountSubquery(string $subscriberCondition): string { |
| 258 | $subscriberSegmentTable = $this->getTableName(SubscriberSegmentEntity::class); |
| 259 | $segmentsTable = $this->getTableName(SegmentEntity::class); |
| 260 | $subscribedStatus = SubscriberEntity::STATUS_SUBSCRIBED; |
| 261 | |
| 262 | return "SELECT ssg.subscriber_id, COUNT(*) AS c |
| 263 | FROM {$subscriberSegmentTable} ssg |
| 264 | JOIN {$segmentsTable} g ON g.id = ssg.segment_id AND g.deleted_at IS NULL |
| 265 | WHERE ssg.status = '{$subscribedStatus}' AND {$subscriberCondition} |
| 266 | GROUP BY ssg.subscriber_id"; |
| 267 | } |
| 268 | |
| 269 | /** |
| 270 | * @param class-string $entityClass |
| 271 | */ |
| 272 | private function getTableName(string $entityClass): string { |
| 273 | return $this->entityManager->getClassMetadata($entityClass)->getTableName(); |
| 274 | } |
| 275 | } |
| 276 |