2meet-data-optimizer/includes/migration/class-tmdo-post-group-reconciler.php

<?php
/**
 * TMDO_Post_Group_Reconciler — move post values written under legacy routing.
 *
 * Before post_type-scoped routing (tmdo_post_scoped_routing), a meta key
 * shared by several post groups was read and written through the last group
 * registered, e.g. a listing's hp_rating lived in wp_wpdo_post_hp_vendor_core.
 * Scoped lookups read it from the listing's own group instead, so those
 * values have to be copied there first.
 *
 * Rule: after the switch every read must return what it returned before.
 *   - A non-NULL value in the legacy owner's table is what reads served, so
 *     it overwrites the scoped group's column.
 *   - NULL / missing legacy values leave the scoped column alone (it holds
 *     the backfill copy, the only data there is).
 *   - Source rows are left in place; scoped lookups never read them again.
 *
 * Batched by entity id with a cursor option so an interrupted run resumes; a
 * final pass re-copies legacy rows updated since the first run started, then
 * scoped routing is switched on.
 *
 * @package WP_Data_Optimizer
 */

declare(strict_types=1);

defined( 'ABSPATH' ) || exit;

// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared,WordPress.DB.DirectDatabaseQuery,WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare -- placeholders are built for spread params; table/column names come from the entity registry via Schema_Manager and are sanitized there.

/**
 * Copies legacy-routed post values into their post type's group table.
 */
final class TMDO_Post_Group_Reconciler {

	/** Option: move id → last entity id copied. */
	public const CURSOR_OPTION = 'tmdo_post_reconcile_cursor';

	/** Option: DB time the first (possibly interrupted) run started. */
	public const STARTED_OPTION = 'tmdo_post_reconcile_started';

	/**
	 * Keys of scoped post groups whose legacy owner is another group.
	 *
	 * @return array<int, array{target:string, source:string, key:string, post_types:string[]}>
	 */
	public static function plan(): array {
		$moves = array();
		foreach ( TMDO_Entity_Registry::get_groups_for_type( 'post' ) as $group ) {
			$post_types = TMDO_Entity_Registry::get_group_post_types( 'post', $group );
			if ( empty( $post_types ) ) {
				continue;
			}
			foreach ( TMDO_Entity_Registry::get_group_keys( 'post', $group ) as $key ) {
				$legacy = TMDO_Entity_Registry::get_field( 'post', $key );
				if ( null === $legacy || $legacy['group'] === $group ) {
					continue;
				}
				$moves[] = array(
					'target'     => $group,
					'source'     => $legacy['group'],
					'key'        => $key,
					'post_types' => $post_types,
				);
			}
		}
		return $moves;
	}

	/**
	 * Copy the values (or, with $dry_run, count what would be copied).
	 *
	 * @param bool $dry_run Count only; write nothing.
	 * @param int  $batch   Rows per batch.
	 * @return array{moves: array<int, array<string, mixed>>, enabled: bool}
	 */
	public static function run( bool $dry_run = false, int $batch = 500 ): array {
		global $wpdb;

		$report  = array(
			'moves'   => array(),
			'enabled' => false,
		);
		$adapter = TMDO_Entity_Registry::get_adapter( 'post' );
		if ( ! $adapter ) {
			return $report;
		}
		$id_col = $adapter->get_entity_id_column();
		$batch  = max( 1, $batch );

		if ( ! $dry_run && false === get_option( self::STARTED_OPTION ) ) {
			update_option( self::STARTED_OPTION, (string) $wpdb->get_var( 'SELECT NOW()' ), false );
		}
		$cursor = get_option( self::CURSOR_OPTION, array() );
		$cursor = is_array( $cursor ) ? $cursor : array();

		$moves = self::plan();
		foreach ( $moves as $move ) {
			list( $from, $params ) = self::source_clause( $move, $id_col );
			$entry                 = array(
				'target'      => $move['target'],
				'source'      => $move['source'],
				'key'         => $move['key'],
				'copied'      => 0,
				'overwritten' => 0,
			);

			if ( $dry_run ) {
				$target = TMDO_Schema_Manager::get_table_name( 'post', $move['target'] );
				$col    = TMDO_Schema_Manager::sanitize_column_name( $move['key'] );

				$entry['copied']      = (int) $wpdb->get_var( $wpdb->prepare( "SELECT COUNT(*) {$from}", ...$params ) );
				$entry['overwritten'] = (int) $wpdb->get_var(
					$wpdb->prepare(
						"SELECT COUNT(*) {$from} AND EXISTS ( SELECT 1 FROM `{$target}` t WHERE t.`{$id_col}` = s.`{$id_col}` AND t.`{$col}` IS NOT NULL AND t.`{$col}` <> s.`{$col}` )",
						...$params
					)
				);
				$report['moves'][]    = $entry;
				continue;
			}

			$move_id = "{$move['source']}>{$move['target']}:{$move['key']}";
			$last    = (int) ( $cursor[ $move_id ] ?? 0 );
			do {
				$rows             = (array) $wpdb->get_results(
					$wpdb->prepare( "SELECT s.`{$id_col}` AS id, s.`" . TMDO_Schema_Manager::sanitize_column_name( $move['key'] ) . "` AS v {$from} AND s.`{$id_col}` > %d ORDER BY s.`{$id_col}` LIMIT %d", ...array_merge( $params, array( $last, $batch ) ) ),
					ARRAY_A
				);
				$entry['copied'] += self::copy_rows( $move, $id_col, $rows );
				$fetched          = count( $rows );
				if ( $fetched ) {
					$last               = (int) end( $rows )['id'];
					$cursor[ $move_id ] = $last;
					update_option( self::CURSOR_OPTION, $cursor, false );
				}
			} while ( $fetched === $batch );

			$report['moves'][] = $entry;
		}

		if ( $dry_run ) {
			return $report;
		}

		// Catch-up: legacy-routed writes that landed after a batch passed their row.
		$started = (string) get_option( self::STARTED_OPTION );
		foreach ( $moves as $move ) {
			list( $from, $params ) = self::source_clause( $move, $id_col );
			$rows                  = (array) $wpdb->get_results(
				$wpdb->prepare( "SELECT s.`{$id_col}` AS id, s.`" . TMDO_Schema_Manager::sanitize_column_name( $move['key'] ) . "` AS v {$from} AND s.updated_at >= %s", ...array_merge( $params, array( $started ) ) ),
				ARRAY_A
			);
			self::copy_rows( $move, $id_col, $rows );
		}

		update_option( TMDO_Entity_Registry::SCOPED_ROUTING_OPTION, '1', true );
		delete_option( self::CURSOR_OPTION );
		delete_option( self::STARTED_OPTION );
		$report['enabled'] = true;

		return $report;
	}

	/**
	 * FROM/WHERE for non-NULL legacy values of the move's post types.
	 *
	 * @param array<string, mixed> $move   One entry of plan().
	 * @param string               $id_col Entity id column of the flat tables.
	 * @return array{0:string, 1:array<int, mixed>}
	 */
	private static function source_clause( array $move, string $id_col ): array {
		global $wpdb;
		$source = TMDO_Schema_Manager::get_table_name( 'post', $move['source'] );
		$col    = TMDO_Schema_Manager::sanitize_column_name( $move['key'] );
		$types  = implode( ',', array_fill( 0, count( $move['post_types'] ), '%s' ) );

		return array(
			"FROM `{$source}` s INNER JOIN `{$wpdb->posts}` p ON p.ID = s.`{$id_col}` AND p.post_type IN ({$types}) WHERE s.`{$col}` IS NOT NULL",
			$move['post_types'],
		);
	}

	/**
	 * Upsert one column of the target group rows and drop their cached copies.
	 *
	 * @param array<string, mixed>             $move   One entry of plan().
	 * @param string                           $id_col Entity id column of the flat tables.
	 * @param array<int, array<string, mixed>> $rows   Source rows: list of { id, v }.
	 * @return int Rows written.
	 */
	private static function copy_rows( array $move, string $id_col, array $rows ): int {
		global $wpdb;
		if ( empty( $rows ) ) {
			return 0;
		}
		$target = TMDO_Schema_Manager::get_table_name( 'post', $move['target'] );
		$col    = TMDO_Schema_Manager::sanitize_column_name( $move['key'] );

		$values = array();
		$params = array();
		foreach ( $rows as $row ) {
			$values[] = '(%d, %s)';
			$params[] = (int) $row['id'];
			$params[] = (string) $row['v'];
		}
		$wpdb->query(
			$wpdb->prepare(
				"INSERT INTO `{$target}` (`{$id_col}`, `{$col}`) VALUES " . implode( ', ', $values ) . " ON DUPLICATE KEY UPDATE `{$col}` = VALUES(`{$col}`)",
				...$params
			)
		);

		if ( class_exists( 'TMDO_Cache_Orchestrator' ) ) {
			foreach ( $rows as $row ) {
				TMDO_Cache_Orchestrator::invalidate( 'post', (int) $row['id'], $move['target'] );
			}
		}
		return count( $rows );
	}
}