2meet-data-optimizer/includes/query/class-tmdo-post-query-router.php

<?php
/**
 * TMDO_Post_Query_Router — Post entity meta_query → flat table JOIN rewriter (v2.10.1).
 *
 * Companion to legacy TMDO_Query_Router (which targets zone hot tables).
 * Independent class — does NOT modify the user side path or the legacy zone
 * router. Both routers can coexist on the same WP_Query without conflict
 * because they extract clauses on disjoint criteria:
 *
 *   - TMDO_Query_Router         → only zone-registered keys with feature_flag query_active
 *   - TMDO_Post_Query_Router    → only entity-registered keys with mode=reads_from_flat
 *
 * Mode-gated: only rewrites when post mode is `shadow_read` or `aeav_only`.
 * In `disabled` / `dual_write` modes the router is a no-op.
 *
 * Hook order:
 *   1. pre_get_posts  — strip registered keys from meta_query, store in
 *                       $query->set( 'wpdo_post_clauses', $routed )
 *   2. posts_join     — LEFT JOIN wp_wpdo_post_<group> per routed post_type
 *   3. posts_where    — append WHERE conditions targeting flat columns
 *   4. posts_groupby  — ensure GROUP BY posts.ID to dedup join multiplication
 *
 * Status / date on the flat table (O-1): when the group's post_status /
 * post_date copies are ready (TMDO_Post_Status_Sync) and none is missing,
 * a single-post-type search with explicit statuses and date ordering
 * INNER JOINs the table, filters the copies, orders by the copied
 * post_date and skips GROUP BY (one row per post), so MySQL can walk
 * `idx_post_status_date` and stop at the LIMIT.
 *
 * 🔒 v2.9.x → v2.10.x frozen contract: never touches user entity. Routes
 * only entity_type='post'.
 *
 * @package WP_Data_Optimizer
 * @since   2.10.1
 */

declare(strict_types=1);

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared,WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Internal query rewriter: table names go through Schema_Manager::sanitize_column_name, column names go through esc_sql, user-controlled values use $wpdb->prepare() in build_condition().

/**
 * Rewrites WP_Query meta_query clauses to JOIN against post entity flat tables.
 */
class TMDO_Post_Query_Router {

	/** Query var key for capturing routed clauses (consumed by posts_join/where). */
	private const QUERY_VAR = 'wpdo_post_clauses';

	/** Query var holding the status / date copy route, when one applies. */
	private const COPY_VAR = 'wpdo_post_copy_route';

	/**
	 * Per-request answer to "are this group's copies usable?".
	 *
	 * @var array<string, bool>
	 */
	private static array $copies_usable = array();

	/**
	 * Register all WP_Query hooks.
	 *
	 * @return void
	 */
	public function register_hooks(): void {
		add_action( 'pre_get_posts', array( $this, 'pre_get_posts' ), 10, 1 );
		add_filter( 'posts_join', array( $this, 'posts_join' ), 10, 2 );
		add_filter( 'posts_where', array( $this, 'posts_where' ), 10, 2 );
		add_filter( 'posts_groupby', array( $this, 'posts_groupby' ), 10, 2 );
		add_filter( 'posts_orderby', array( $this, 'posts_orderby' ), 10, 2 );
	}

	/**
	 * Strip Entity-Registry-managed meta_query keys; store routed clauses.
	 *
	 * @param \WP_Query $query Current WP_Query.
	 * @return void
	 */
	public function pre_get_posts( \WP_Query $query ): void {
		// Mode gate: only run when post mode is reads_from_flat.
		if ( ! self::is_router_active() ) {
			return;
		}

		// is_admin guard (matches legacy TMDO_Query_Router behavior).
		if ( is_admin() && ! wp_doing_ajax() ) {
			return;
		}

		$post_types = (array) $query->get( 'post_type' );
		if ( empty( $post_types ) ) {
			return;
		}

		$raw_meta_query = (array) $query->get( 'meta_query' );
		if ( empty( $raw_meta_query ) ) {
			return;
		}

		$routed    = array(); // Keyed by post_type.
		$remaining = array();

		foreach ( $raw_meta_query as $k => $clause ) {
			if ( 'relation' === $k || ! is_array( $clause ) || ! isset( $clause['key'] ) ) {
				$remaining[ $k ] = $clause;
				continue;
			}

			$matched = false;
			foreach ( $post_types as $pt ) {
				$field = TMDO_Entity_Registry::get_field( 'post', $clause['key'], (string) $pt );
				if ( ! $field ) {
					continue;
				}
				// Confirm the field's group is appropriate for this post_type:
				// either cross-cutting wp_core (any) or the post_type-specific group.
				$group          = $field['group'] ?? '';
				$is_appropriate = ( 'wp_core' === $group ) || in_array( $group, TMDO_Post_Fields::groups_for_post_type( (string) $pt ), true );
				if ( ! $is_appropriate ) {
					continue;
				}

				$column = TMDO_Schema_Manager::sanitize_column_name( $clause['key'] );

				$routed[ $pt ][] = array(
					'group'   => $group,
					'column'  => $column,
					'value'   => $clause['value'] ?? '',
					'compare' => strtoupper( trim( (string) ( $clause['compare'] ?? '=' ) ) ),
					'type'    => strtoupper( trim( (string) ( $clause['type'] ?? 'CHAR' ) ) ),
				);
				$matched         = true;
				break;
			}

			if ( ! $matched ) {
				$remaining[ $k ] = $clause;
			}
		}

		if ( empty( $routed ) ) {
			return;
		}

		// Preserve relation if surviving clauses still have it.
		if ( isset( $raw_meta_query['relation'] ) && ! isset( $remaining['relation'] ) ) {
			$remaining['relation'] = $raw_meta_query['relation'];
		}

		$query->set( 'meta_query', $remaining );
		$query->set( self::QUERY_VAR, $routed );

		$copy = self::copy_route( $query, $routed );
		if ( $copy ) {
			$query->set( self::COPY_VAR, $copy );
		}
	}

	/**
	 * LEFT JOIN flat tables for each routed post_type.
	 *
	 * @param string|null $join  Current JOIN SQL.
	 * @param \WP_Query   $query Current WP_Query.
	 * @return string Modified JOIN SQL.
	 */
	public function posts_join( ?string $join, \WP_Query $query ): string {
		$join   = (string) ( $join ?? '' );
		$routed = $query->get( self::QUERY_VAR );
		if ( empty( $routed ) || ! is_array( $routed ) ) {
			return $join;
		}

		global $wpdb;

		foreach ( $routed as $post_type => $clauses ) {
			// Each routed clause carries its group; use the first clause's group
			// (all clauses for a given post_type currently target one group).
			$group = $clauses[0]['group'] ?? '';
			if ( '' === $group ) {
				continue;
			}
			$table = $wpdb->prefix . 'wpdo_post_' . sanitize_key( $group );
			$alias = self::table_alias( $post_type, $group );

			// Don't double-join.
			if ( false !== strpos( $join, "`{$alias}`" ) ) {
				continue;
			}

			$type  = self::copy_route_for( $query, $post_type ) ? 'INNER' : 'LEFT';
			$join .= " {$type} JOIN `{$table}` AS `{$alias}` "
				. "ON (`{$wpdb->posts}`.`ID` = `{$alias}`.`post_id`)";
		}

		return $join;
	}

	/**
	 * Append WHERE conditions for routed clauses.
	 *
	 * @param string|null $where Current WHERE SQL.
	 * @param \WP_Query   $query Current WP_Query.
	 * @return string Modified WHERE SQL.
	 */
	public function posts_where( ?string $where, \WP_Query $query ): string {
		$where  = (string) ( $where ?? '' );
		$routed = $query->get( self::QUERY_VAR );
		if ( empty( $routed ) || ! is_array( $routed ) ) {
			return $where;
		}

		global $wpdb;

		foreach ( $routed as $post_type => $clauses ) {
			foreach ( $clauses as $clause ) {
				$alias   = self::table_alias( $post_type, $clause['group'] );
				$col     = '`' . $alias . '`.`' . esc_sql( $clause['column'] ) . '`';
				$compare = self::sanitize_compare( $clause['compare'] );
				$value   = $clause['value'];

				$where .= self::build_condition( $col, $compare, $value );
			}
		}

		$copy = $query->get( self::COPY_VAR );
		if ( ! empty( $copy ) ) {
			$alias  = self::table_alias( $copy['post_type'], $copy['group'] );
			$where .= self::build_condition( "`{$alias}`.`post_status`", 'IN', $copy['statuses'] );
		}

		return $where;
	}

	/**
	 * Order by the flat table's post_date copy when the copy route applies.
	 *
	 * @param string|null $orderby Current ORDER BY SQL.
	 * @param \WP_Query   $query   Current WP_Query.
	 * @return string Modified ORDER BY SQL.
	 */
	public function posts_orderby( ?string $orderby, \WP_Query $query ): string {
		$orderby = (string) ( $orderby ?? '' );
		$copy    = $query->get( self::COPY_VAR );
		if ( empty( $copy ) ) {
			return $orderby;
		}
		return '`' . self::table_alias( $copy['post_type'], $copy['group'] ) . '`.`post_date` ' . $copy['order'];
	}

	/**
	 * Ensure GROUP BY posts.ID to deduplicate JOIN-multiplied rows.
	 *
	 * @param string|null $groupby Current GROUP BY SQL.
	 * @param \WP_Query   $query   Current WP_Query.
	 * @return string Modified GROUP BY SQL.
	 */
	public function posts_groupby( ?string $groupby, \WP_Query $query ): string {
		$groupby = (string) ( $groupby ?? '' );
		$routed  = $query->get( self::QUERY_VAR );
		if ( empty( $routed ) || ! is_array( $routed ) || ! empty( $query->get( self::COPY_VAR ) ) ) {
			return $groupby;
		}

		global $wpdb;
		if ( '' === trim( $groupby ) ) {
			$groupby = "`{$wpdb->posts}`.`ID`";
		}

		return $groupby;
	}

	// ─────────────────────────────────────────────────────────────────────────
	// Helpers
	// ─────────────────────────────────────────────────────────────────────────

	/**
	 * Whether the router is mode-active (post mode is shadow_read or aeav_only).
	 *
	 * @return bool
	 */
	public static function is_router_active(): bool {
		if ( ! class_exists( 'TMDO_Mode_Manager' ) ) {
			return false;
		}
		return TMDO_Mode_Manager::reads_from_flat( 'post' );
	}

	/**
	 * Forget per-request copy checks (tests, or after `post-status-sync`).
	 *
	 * @return void
	 */
	public static function flush_copy_cache(): void {
		self::$copies_usable = array();
	}

	/**
	 * The status / date copy route for a routed query, or [] when it does not apply.
	 *
	 * @param \WP_Query                        $query  Current WP_Query.
	 * @param array<string, array<int, array>> $routed Routed clauses keyed by post type.
	 * @return array{post_type:string, group:string, statuses:string[], order:string}|array{}
	 */
	private static function copy_route( \WP_Query $query, array $routed ): array {
		if ( 1 !== count( $routed ) ) {
			return array();
		}
		$post_type = (string) array_key_first( $routed );
		$groups    = array_values( array_unique( array_column( $routed[ $post_type ], 'group' ) ) );
		if ( 1 !== count( $groups ) || ! in_array( $query->get( 'orderby' ), array( '', 'date', 'post_date' ), true ) ) {
			return array();
		}
		$statuses = $query->get( 'post_status' );
		$statuses = array_values( array_filter( array_map( 'sanitize_key', is_array( $statuses ) ? $statuses : explode( ',', (string) $statuses ) ) ) );
		if ( empty( $statuses ) || in_array( 'any', $statuses, true ) || ! self::copies_usable( $groups[0] ) ) {
			return array();
		}
		return array(
			'post_type' => $post_type,
			'group'     => $groups[0],
			'statuses'  => $statuses,
			'order'     => 'ASC' === strtoupper( (string) $query->get( 'order' ) ) ? 'ASC' : 'DESC',
		);
	}

	/**
	 * Whether the copy route applies to this post type's join.
	 *
	 * @param \WP_Query $query     Current WP_Query.
	 * @param string    $post_type Post type slug.
	 * @return bool
	 */
	private static function copy_route_for( \WP_Query $query, string $post_type ): bool {
		$copy = $query->get( self::COPY_VAR );
		return ! empty( $copy ) && $copy['post_type'] === $post_type;
	}

	/**
	 * Copies are marked ready and no row lacks them (bulk writes that bypass
	 * the Hook Bus leave NULLs until `wp tmdo post-status-sync` runs again).
	 * One index probe per group per request.
	 *
	 * @param string $group Entity group name.
	 * @return bool
	 */
	private static function copies_usable( string $group ): bool {
		if ( ! isset( self::$copies_usable[ $group ] ) ) {
			global $wpdb;
			$table                         = TMDO_Schema_Manager::get_table_name( 'post', $group );
			self::$copies_usable[ $group ] = class_exists( 'TMDO_Post_Status_Sync' )
				&& 'ready' === TMDO_Post_Status_Sync::state( $group )
				&& TMDO_Schema_Manager::table_exists( $table )
				&& null === $wpdb->get_var( "SELECT 1 FROM `{$table}` WHERE `post_status` IS NULL LIMIT 1" );
		}
		return self::$copies_usable[ $group ];
	}

	/**
	 * Generate a deterministic JOIN alias for a (post_type, group) pair.
	 *
	 * @param string $post_type Post type slug.
	 * @param string $group     Entity group name.
	 * @return string SQL alias.
	 */
	private static function table_alias( string $post_type, string $group ): string {
		return 'wpdoflat_' . sanitize_key( $post_type ) . '_' . sanitize_key( $group );
	}

	/**
	 * Allow-list of comparison operators.
	 *
	 * @param string $compare Raw operator from clause.
	 * @return string Sanitized operator (defaults to '=').
	 */
	private static function sanitize_compare( string $compare ): string {
		$allowed = array( '=', '!=', '<>', '<', '<=', '>', '>=', 'IN', 'NOT IN', 'LIKE', 'NOT LIKE', 'BETWEEN', 'NOT BETWEEN', 'EXISTS', 'NOT EXISTS' );
		$compare = strtoupper( trim( $compare ) );
		return in_array( $compare, $allowed, true ) ? $compare : '=';
	}

	/**
	 * Build a single WHERE condition fragment (always prepended with " AND ").
	 *
	 * @param string $col_sql  Quoted column reference.
	 * @param string $compare  Sanitized comparison operator.
	 * @param mixed  $value    Raw value (scalar or array).
	 * @return string
	 */
	private static function build_condition( string $col_sql, string $compare, $value ): string {
		global $wpdb;

		if ( in_array( $compare, array( 'IN', 'NOT IN' ), true ) ) {
			$values = (array) $value;
			if ( empty( $values ) ) {
				return '';
			}
			$placeholders = implode( ',', array_fill( 0, count( $values ), '%s' ) );
			return ' AND ' . $col_sql . ' ' . $compare . ' (' . $wpdb->prepare( $placeholders, ...$values ) . ')';
		}

		if ( in_array( $compare, array( 'BETWEEN', 'NOT BETWEEN' ), true ) ) {
			$values = (array) $value;
			if ( count( $values ) < 2 ) {
				return '';
			}
			return ' AND ' . $col_sql . ' ' . $compare . ' '
				. $wpdb->prepare( '%s AND %s', $values[0], $values[1] );
		}

		if ( in_array( $compare, array( 'EXISTS', 'NOT EXISTS' ), true ) ) {
			$op = 'EXISTS' === $compare ? 'IS NOT NULL' : 'IS NULL';
			return ' AND ' . $col_sql . ' ' . $op;
		}

		return ' AND ' . $col_sql . ' ' . $compare . ' ' . $wpdb->prepare( '%s', (string) $value );
	}
}