<?php
/**
 * AI Search — dual-vector backfill (v1.48.0).
 *
 * The merchant enables dual embeddings on the Search Configuration tab, we
 * backfill the title/body pair for every already-published post in the
 * background so the hot path can start using it. New publishes dual-write
 * via AI_Search::cron_generate_embedding(); this class only exists for
 * posts that existed BEFORE the opt-in.
 *
 * The backfill runs under three triggers so it finishes even on sites
 * where WP cron is unreliable:
 *   1. A daily WP-Cron tick (CRON_HOOK), spawned by ensure_scheduled() on
 *      every plugins_loaded while the merchant is opted in and work is
 *      pending.
 *   2. A manual "Run next batch" button in the UI, via the AJAX endpoint
 *      ai_search_dual_backfill_batch.
 *   3. Nothing else — flipping the toggle off unschedules the cron.
 *
 * One batch = BATCH_SIZE posts (default 20). Each batch is ONE HTTP call to
 * the service (interleaved title+body texts in a single payload), so the
 * network cost is 1 round-trip per 20 posts, not 40.
 *
 * Reversible: flipping the dual toggle off leaves the postmeta in place —
 * zero data loss, re-enabling continues from where it stopped. `cleanup()`
 * is only called when the merchant explicitly asks to purge dual data.
 */

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

class AI_Search_Dual_Backfill {

	const CRON_HOOK            = 'ai_search_dual_backfill_cron';
	const CRON_LOCK_TRANSIENT  = 'ai_search_dual_backfill_lock';
	const DONE_NOTICE          = 'ai_search_dual_backfill_done';
	const CRON_BUDGET_SECONDS  = 10;
	const BATCH_SIZE           = 20;

	/* ====================================================================
	 * Introspection
	 * ================================================================== */

	/**
	 * Posts that still need a dual pair. Only counts posts that have a
	 * single-vector v2 embedding already (so we're never racing publish
	 * writes) and no dual pair yet.
	 */
	public static function count_pending() {
		global $wpdb;
		return (int) $wpdb->get_var( $wpdb->prepare(
			"SELECT COUNT(*) FROM {$wpdb->posts} p
			   JOIN {$wpdb->postmeta} ms ON ms.post_id = p.ID AND ms.meta_key = %s
			  LEFT JOIN {$wpdb->postmeta} mt ON mt.post_id = p.ID AND mt.meta_key = %s
			  WHERE p.post_status = 'publish'
			    AND mt.post_id IS NULL",
			'_ai_search_embedding_v2',
			AI_Search::META_KEY_TITLE_V2
		) );
	}

	/** Posts with a dual pair stored. */
	public static function count_done() {
		global $wpdb;
		return (int) $wpdb->get_var( $wpdb->prepare(
			"SELECT COUNT(DISTINCT post_id) FROM {$wpdb->postmeta}
			  WHERE meta_key = %s",
			AI_Search::META_KEY_TITLE_V2
		) );
	}

	/** Total posts with a single-vector embedding (eligible for dual). */
	public static function count_total() {
		global $wpdb;
		return (int) $wpdb->get_var( $wpdb->prepare(
			"SELECT COUNT(DISTINCT post_id) FROM {$wpdb->postmeta}
			  WHERE meta_key = %s",
			'_ai_search_embedding_v2'
		) );
	}

	public static function progress() {
		$total = self::count_total();
		$done  = self::count_done();
		return [
			'total'   => $total,
			'done'    => $done,
			'pending' => max( 0, $total - $done ),
			'percent' => $total > 0 ? (int) floor( $done / $total * 100 ) : 0,
		];
	}

	/* ====================================================================
	 * Scheduling
	 * ================================================================== */

	/**
	 * Keep the daily cron scheduled while dual is on AND there's work to do.
	 * Unschedule once the backfill is complete OR the merchant turns dual off.
	 * Called on every plugins_loaded via AI_Search::register_hooks.
	 */
	public static function ensure_scheduled() {
		if ( ! AI_Search::is_dual_enabled() ) {
			self::unschedule();
			return;
		}
		if ( self::count_pending() === 0 ) {
			self::unschedule();
			return;
		}
		if ( ! wp_next_scheduled( self::CRON_HOOK ) ) {
			wp_schedule_event( time() + 60, 'daily', self::CRON_HOOK );
		}
	}

	public static function unschedule() {
		$ts = wp_next_scheduled( self::CRON_HOOK );
		if ( $ts ) wp_unschedule_event( $ts, self::CRON_HOOK );
	}

	/* ====================================================================
	 * Workers
	 * ================================================================== */

	/**
	 * Daily cron handler. Takes one soft lock then runs batches until the
	 * CRON_BUDGET_SECONDS cap is hit, there's nothing left to do, or the
	 * provider returns an error that would fan out to every post in the
	 * batch (quota exceeded, auth failure, service down).
	 */
	public static function cron_tick() {
		if ( ! AI_Search::is_dual_enabled() ) return;

		if ( get_transient( self::CRON_LOCK_TRANSIENT ) ) return;
		set_transient( self::CRON_LOCK_TRANSIENT, 1, 60 );

		@set_time_limit( 60 );
		$started = microtime( true );

		while ( ( microtime( true ) - $started ) < self::CRON_BUDGET_SECONDS ) {
			$batch_results = self::run_one_batch();
			if ( ! $batch_results['processed'] ) break;        // nothing more
			if ( $batch_results['fatal'] )     break;          // quota / auth / 5xx
		}

		delete_transient( self::CRON_LOCK_TRANSIENT );

		if ( self::count_pending() === 0 ) {
			self::unschedule();
			set_transient( self::DONE_NOTICE, 1, 7 * DAY_IN_SECONDS );
		}
	}

	/**
	 * One batch of BATCH_SIZE posts through the dual-write path. Returns
	 * ['processed' => int, 'errors' => int, 'fatal' => bool]. 'fatal' is
	 * set when every post in the batch failed with the same provider error
	 * code — a signal the caller should stop retrying instead of hammering
	 * the service for the same reason.
	 */
	public static function run_one_batch() {
		$ids = self::next_batch_ids();
		if ( empty( $ids ) ) return [ 'processed' => 0, 'errors' => 0, 'fatal' => false ];

		$plugin  = AI_Search::get_instance();
		$results = $plugin->generate_embeddings_batch_dual( $ids );

		$errors = 0;
		$fatal_codes = [ 'quota_exceeded', 'invalid_token', 'no_token', 'service_unavailable', 'provider_error' ];
		$seen_fatal  = 0;
		foreach ( $results as $r ) {
			if ( is_wp_error( $r ) ) {
				$errors++;
				if ( in_array( $r->get_error_code(), $fatal_codes, true ) ) $seen_fatal++;
			}
		}
		return [
			'processed' => count( $results ),
			'errors'    => $errors,
			'fatal'     => $errors > 0 && $seen_fatal === count( $results ), // all-fail with fatal code
		];
	}

	/**
	 * Next BATCH_SIZE post ids that have single-vector v2 but no dual pair.
	 */
	private static function next_batch_ids() {
		global $wpdb;
		return array_map( 'intval', $wpdb->get_col( $wpdb->prepare(
			"SELECT p.ID FROM {$wpdb->posts} p
			   JOIN {$wpdb->postmeta} ms ON ms.post_id = p.ID AND ms.meta_key = %s
			  LEFT JOIN {$wpdb->postmeta} mt ON mt.post_id = p.ID AND mt.meta_key = %s
			  WHERE p.post_status = 'publish'
			    AND mt.post_id IS NULL
			  LIMIT %d",
			'_ai_search_embedding_v2',
			AI_Search::META_KEY_TITLE_V2,
			self::BATCH_SIZE
		) ) );
	}

	/* ====================================================================
	 * AJAX — Run Next Batch button on the settings UI
	 * ================================================================== */

	public static function ajax_batch() {
		check_ajax_referer( 'ai_search_dual_backfill' );
		if ( ! current_user_can( 'manage_options' ) ) {
			wp_send_json_error( [ 'message' => 'forbidden' ], 403 );
		}
		$res      = self::run_one_batch();
		$progress = self::progress();
		if ( $progress['pending'] === 0 ) {
			self::unschedule();
			set_transient( self::DONE_NOTICE, 1, 7 * DAY_IN_SECONDS );
		}
		wp_send_json_success( array_merge( $res, $progress ) );
	}

	/* ====================================================================
	 * Admin notice — one-shot "all done" message after backfill finishes
	 * ================================================================== */

	public static function maybe_render_done_notice() {
		if ( ! current_user_can( 'manage_options' ) ) return;
		if ( ! get_transient( self::DONE_NOTICE ) ) return;
		delete_transient( self::DONE_NOTICE );
		echo '<div class="notice notice-success is-dismissible"><p>'
			. esc_html__( 'AI Search: dual-vector backfill is complete. Semantic ranking now runs on both the title and body vectors.', 'ai-search' )
			. '</p></div>';
	}

	/* ====================================================================
	 * Clean-up (invoked only on explicit request — never automatically)
	 * ================================================================== */

	public static function purge_dual_postmeta() {
		global $wpdb;
		foreach ( [ AI_Search::META_KEY_TITLE_V2, AI_Search::META_KEY_BODY_V2 ] as $key ) {
			$wpdb->query( $wpdb->prepare(
				"DELETE FROM {$wpdb->postmeta} WHERE meta_key = %s", $key
			) );
		}
	}
}