<?php
/**
 * Schedules and throttles background scanning via Action Scheduler so a
 * full-site scan never runs inline on a real visitor's request.
 *
 * @package ABDS_SEO
 */

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

/**
 * Splits a full-site scan into throttled Action Scheduler batches and
 * tracks overall progress in wp_abds_seo_scans.
 */
class ABDS_SEO_Scan_Scheduler {

	/**
	 * Posts scanned per Action Scheduler batch action.
	 */
	const BATCH_SIZE = 20;

	/**
	 * Seconds between two consecutive batches, to avoid hammering the
	 * host's DB/CPU on shared hosting.
	 */
	const BATCH_THROTTLE_SECONDS = 15;

	/**
	 * Action Scheduler group all of this plugin's scheduled actions use.
	 */
	const GROUP = 'abds-seo';

	/**
	 * Registers this class's action hooks with the plugin's hook loader.
	 *
	 * @param ABDS_SEO_Loader $loader Hook loader.
	 */
	public function register_hooks( ABDS_SEO_Loader $loader ): void {
		$loader->add_action( 'abds_seo_run_site_scan', $this, 'run_site_scan' );
		$loader->add_action( 'abds_seo_scan_batch', $this, 'scan_batch', 10, 3 );
	}

	/**
	 * Entry point for the recurring scheduled scan (and for a manually
	 * triggered "Rescan site" action from the admin). Creates a scan record
	 * and schedules one throttled Action Scheduler batch per BATCH_SIZE
	 * posts.
	 */
	public function run_site_scan(): void {
		if ( ! function_exists( 'as_schedule_single_action' ) ) {
			return; // Action Scheduler not available; caller falls back to no-op rather than blocking a real request.
		}

		global $wpdb;

		$post_ids = get_posts(
			array(
				'post_type'      => 'any',
				'post_status'    => 'publish',
				'posts_per_page' => -1, // phpcs:ignore WordPress.WP.PostsPerPage.posts_per_page_posts_per_page -- internal background batch: collects every published post's ID once, then queues a separate per-post scheduled action below; no cap on how many pages are audited.
				'fields'         => 'ids',
			)
		);

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

		$scans_table = $wpdb->prefix . 'abds_seo_scans';

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- $scans_table is built only from $wpdb->prefix plus a hardcoded literal, never user input; $wpdb->insert() safely escapes its values via the format array. Runs once per scheduled/manually-triggered scan, not a cacheable read path.
		$wpdb->insert(
			$scans_table,
			array(
				'started_at'    => current_time( 'mysql' ),
				'status'        => 'running',
				'pages_scanned' => 0,
			),
			array( '%s', '%s', '%d' )
		);
		$scan_id     = (int) $wpdb->insert_id;
		$pages_total = count( $post_ids );

		$batches = array_chunk( $post_ids, self::BATCH_SIZE );

		foreach ( $batches as $index => $batch_ids ) {
			as_schedule_single_action(
				time() + ( $index * self::BATCH_THROTTLE_SECONDS ),
				'abds_seo_scan_batch',
				array( $batch_ids, $scan_id, $pages_total ),
				self::GROUP
			);
		}
	}

	/**
	 * Scans one batch of posts and updates the scan's running total. Marks
	 * the scan as completed once the cumulative pages_scanned count reaches
	 * $pages_total.
	 *
	 * $pages_total is passed through from run_site_scan() rather than
	 * re-derived from Action Scheduler's own queue state (e.g.
	 * as_has_scheduled_action()): during a batch's own execution, Action
	 * Scheduler still reports that action itself as in-progress, so a
	 * "any batches still pending/running?" check can never reliably tell
	 * the last batch apart from an earlier one — comparing against a
	 * known total sidesteps that entirely.
	 *
	 * @param int[] $post_ids    Post IDs in this batch.
	 * @param int   $scan_id     wp_abds_seo_scans row ID this batch belongs to.
	 * @param int   $pages_total Total pages across every batch of this scan.
	 */
	public function scan_batch( array $post_ids, int $scan_id, int $pages_total ): void {
		global $wpdb;

		$scanner  = new ABDS_SEO_Site_Scanner();
		$analyzer = new ABDS_SEO_Content_Analyzer();

		foreach ( $post_ids as $post_id ) {
			$post_id = (int) $post_id;
			$scanner->scan_post( $post_id );

			// Backfills/refreshes content_score for every post a full site
			// scan touches, not just ones someone happens to open in the
			// editor (see ABDS_SEO_Content_Score_Cache, which handles the
			// save_post case) — analyze() is a cheap no-op when no focus
			// keyword is set (see its own empty_result() early return), so
			// this doesn't meaningfully add to the batch's cost for posts
			// that haven't been given one yet.
			$keyword = (string) get_post_meta( $post_id, '_abds_seo_focus_keyword', true );
			$analyzer->analyze( $post_id, $keyword );
		}

		$scans_table = $wpdb->prefix . 'abds_seo_scans';

		// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- $scans_table is built only from $wpdb->prefix plus a hardcoded literal, never user input; values are passed through $wpdb->prepare(). Runs once per background batch, not on a cacheable read path.
		$wpdb->query(
			$wpdb->prepare(
				"UPDATE {$scans_table} SET pages_scanned = pages_scanned + %d WHERE id = %d",
				count( $post_ids ),
				$scan_id
			)
		);

		$pages_scanned = (int) $wpdb->get_var(
			$wpdb->prepare( "SELECT pages_scanned FROM {$scans_table} WHERE id = %d", $scan_id )
		);
		// phpcs:enable

		if ( $pages_scanned >= $pages_total ) {
			$this->finish_scan( $scan_id );
		}
	}

	/**
	 * Marks a scan as completed and records the site's average technical
	 * score at that point.
	 *
	 * @param int $scan_id wp_abds_seo_scans row ID.
	 */
	protected function finish_scan( int $scan_id ): void {
		global $wpdb;
		$scans_table       = $wpdb->prefix . 'abds_seo_scans';
		$page_scores_table = $wpdb->prefix . 'abds_seo_page_scores';

		// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- $page_scores_table is built only from $wpdb->prefix plus a hardcoded literal, never user input. Runs once per completed scan, not on a cacheable read path.
		$avg_score = $wpdb->get_var( "SELECT AVG(technical_score) FROM {$page_scores_table}" );

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- $scans_table is built only from $wpdb->prefix plus a hardcoded literal, never user input; $wpdb->update() safely escapes its values via the format array. Runs once per completed scan, not a cacheable read path.
		$wpdb->update(
			$scans_table,
			array(
				'finished_at' => current_time( 'mysql' ),
				'status'      => 'completed',
				'avg_score'   => $avg_score,
			),
			array( 'id' => $scan_id ),
			array( '%s', '%s', '%f' ),
			array( '%d' )
		);
	}
}