<?php
/**
 * Takes snapshots of the site's SEO signals as a background job.
 *
 * @package OrbitstemSEO
 */

namespace Orbitstem\SEO\Monitor;

use Orbitstem\SEO\Content\TemplateEngine;
use Orbitstem\SEO\Content\TypeRegistry;
use Orbitstem\SEO\Core\Jobs;
use Orbitstem\SEO\Core\Settings;
use Orbitstem\SEO\Diagnostics\HeadInspector;
use Orbitstem\SEO\Diagnostics\Tables;
use Orbitstem\SEO\Frontend\Resolver;
use Orbitstem\SEO\Links\Url;

defined( 'ABSPATH' ) || exit;

/**
 * A snapshot records, without changing anything:
 *
 * - site switches (search engine visibility, permalink structure, theme,
 *   other SEO plugins, Orbitstem output and settings, robots.txt);
 * - for every published item: address, title, description, noindex,
 *   canonical, sitemap eligibility and inbound links (from the database);
 * - for a sample of pages: HTTP status, robots, canonical and title as
 *   actually rendered (loopback requests to this site only).
 *
 * Stages: site (one step), posts (100 per step), rendered (2 per step), finish.
 */
class Snapshotter {

	const JOB           = 'snapshot';
	const SOURCE_OPTION = 'orbitstem_seo_snapshot_source';

	/**
	 * Settings service.
	 *
	 * @var Settings
	 */
	private $settings;

	/**
	 * Content type registry.
	 *
	 * @var TypeRegistry
	 */
	private $types;

	/**
	 * SEO value resolver.
	 *
	 * @var Resolver
	 */
	private $resolver;

	/**
	 * Head inspector.
	 *
	 * @var HeadInspector
	 */
	private $head;

	/**
	 * Background job runner.
	 *
	 * @var Jobs
	 */
	private $jobs;

	/**
	 * Returns site switches that live outside this class (output state, other SEO plugins).
	 *
	 * @var callable
	 */
	private $extra_site;

	/**
	 * Constructor.
	 *
	 * @param Settings      $settings   Settings.
	 * @param TypeRegistry  $types      Type registry.
	 * @param Resolver      $resolver   Resolver.
	 * @param HeadInspector $head       Head inspector.
	 * @param Jobs          $jobs       Jobs.
	 * @param callable      $extra_site Returns array of extra site signals.
	 */
	public function __construct( Settings $settings, TypeRegistry $types, Resolver $resolver, HeadInspector $head, Jobs $jobs, callable $extra_site ) {
		$this->settings   = $settings;
		$this->types      = $types;
		$this->resolver   = $resolver;
		$this->head       = $head;
		$this->jobs       = $jobs;
		$this->extra_site = $extra_site;
	}

	/**
	 * Registers the job handler.
	 */
	public function register() {
		$this->jobs->register( self::JOB, array( $this, 'step' ), 1 );
	}

	/**
	 * Starts a snapshot unless one is running.
	 *
	 * @param string $source manual|scheduled.
	 * @return bool
	 */
	public function start( $source = 'manual' ) {
		if ( ! Tables::v4_ready() ) {
			return false;
		}
		$state = $this->jobs->status( self::JOB );
		if ( $state && 'running' === $state['status'] ) {
			return false;
		}
		update_option( self::SOURCE_OPTION, 'scheduled' === $source ? 'scheduled' : 'manual', false );
		return $this->jobs->start( self::JOB );
	}

	/**
	 * Job handler.
	 *
	 * @param mixed $cursor Cursor.
	 * @return array{cursor:array,done:bool}
	 */
	public function step( $cursor ) {
		global $wpdb;
		if ( ! is_array( $cursor ) ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery
			$wpdb->insert(
				Tables::snapshots(),
				array(
					'created_at' => current_time( 'mysql', true ),
					'source'     => (string) get_option( self::SOURCE_OPTION, 'manual' ),
					'status'     => 'running',
					'site'       => wp_json_encode( $this->site_signals() ),
				)
			);
			return array(
				'cursor' => array(
					'id'    => (int) $wpdb->insert_id,
					'stage' => 'posts',
					'last'  => 0,
				),
				'done'   => false,
			);
		}

		if ( 'posts' === $cursor['stage'] ) {
			$r              = $this->posts_batch( $cursor['id'], (int) $cursor['last'], 100 );
			$cursor['last'] = $r['last'];
			if ( $r['done'] ) {
				$cursor['stage'] = 'rendered';
				$cursor['urls']  = $this->head->sample_urls();
				$cursor['pos']   = 0;
			}
		} elseif ( 'rendered' === $cursor['stage'] ) {
			$total = count( $cursor['urls'] );
			for ( $i = 0; $i < 2 && $cursor['pos'] < $total; $i++ ) {
				$this->rendered_item( $cursor['id'], $cursor['urls'][ $cursor['pos'] ] );
				++$cursor['pos'];
			}
			if ( $cursor['pos'] >= $total ) {
				$this->finish( $cursor['id'] );
				return array(
					'cursor' => $cursor,
					'done'   => true,
				);
			}
		}
		return array(
			'cursor' => $cursor,
			'done'   => false,
		);
	}

	/**
	 * Site-wide signals.
	 *
	 * @return array
	 */
	public function site_signals() {
		$settings = $this->settings->all();
		$robots   = wp_remote_get(
			home_url( '/robots.txt' ),
			array(
				'timeout'   => 10,
				'sslverify' => apply_filters( 'https_local_ssl_verify', false ), // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- core filter.
				'headers'   => array( \Orbitstem\SEO\Diagnostics\UrlChecker::CHECK_HEADER => '1' ),
			)
		);
		return array(
			'blog_public'         => (int) get_option( 'blog_public' ),
			'permalink_structure' => (string) get_option( 'permalink_structure' ),
			'theme'               => get_stylesheet(),
			'sitemap_enabled'     => (bool) $settings['sitemap_enabled'],
			'settings_hash'       => md5( wp_json_encode( $settings ) ),
			'robots_txt_hash'     => is_wp_error( $robots ) ? '' : md5( (string) wp_remote_retrieve_body( $robots ) ),
		) + (array) call_user_func( $this->extra_site );
	}

	/**
	 * Records one batch of published items.
	 *
	 * @param int $snapshot_id Snapshot.
	 * @param int $after       Last post ID.
	 * @param int $limit       Batch size.
	 * @return array{last:int,done:bool}
	 */
	private function posts_batch( $snapshot_id, $after, $limit ) {
		global $wpdb;
		$types = array_keys( $this->types->post_types() );
		if ( ! $types ) {
			return array(
				'last' => $after,
				'done' => true,
			);
		}
		$in   = implode( ',', array_fill( 0, count( $types ), '%s' ) );
		$args = array_merge( $types, array( (int) $after, (int) $limit ) );
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber, PluginCheck.Security.DirectDB.UnescapedDBParameter -- one %s per type, built above.
		$ids = array_map( 'intval', $wpdb->get_col( $wpdb->prepare( "SELECT ID FROM {$wpdb->posts} WHERE post_status = 'publish' AND post_type IN ({$in}) AND ID > %d ORDER BY ID ASC LIMIT %d", ...$args ) ) );
		if ( ! $ids ) {
			return array(
				'last' => $after,
				'done' => true,
			);
		}
		_prime_post_caches( $ids, false, true );
		$inbound = $this->inbound_counts( $ids );
		$home    = home_url( '/' );
		$rows    = array();
		foreach ( $ids as $id ) {
			$post = get_post( $id );
			$data = $post ? $this->resolver->for_post( $post ) : null;
			if ( ! $data ) {
				continue;
			}
			$url       = Url::normalize( (string) get_permalink( $post ), $home );
			$canonical = '' !== $data['canonical'] ? Url::normalize( $data['canonical'], $home ) : $url;
			$sitemap   = $this->settings->get( 'sitemap_enabled' ) && $this->settings->post_type( $post->post_type, 'sitemap' ) && ! $data['noindex'] && $canonical === $url && '' === $post->post_password;
			$rows[]    = $wpdb->prepare(
				'(%d, 1, %s, %s, %d, 0, %d, %s, %s, %s, %d, %d, 0)',
				$snapshot_id,
				Url::hash( $url ),
				mb_substr( $url, 0, 500 ),
				$id,
				$data['noindex'] ? 1 : 0,
				mb_substr( $canonical, 0, 500 ),
				mb_substr( TemplateEngine::plain( $data['title'] ), 0, 255 ),
				md5( $data['description'] ),
				$sitemap ? 1 : 0,
				isset( $inbound[ $id ] ) ? $inbound[ $id ] : 0
			);
		}
		if ( $rows ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.NotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- each row prepared above.
			$wpdb->query( $wpdb->prepare( 'INSERT INTO %i (snapshot_id, kind, url_hash, url, object_id, status, noindex, canonical, title, desc_hash, sitemap, inbound, duplicates) VALUES ', Tables::snapshot_items() ) . implode( ',', $rows ) );
		}
		return array(
			'last' => end( $ids ),
			'done' => count( $ids ) < $limit,
		);
	}

	/**
	 * Number of other posts linking to each post (from the link index).
	 *
	 * @param int[] $ids Post IDs.
	 * @return array<int,int>
	 */
	private function inbound_counts( array $ids ) {
		global $wpdb;
		if ( ! Tables::exist_cached() || ! $ids ) {
			return array();
		}
		$in   = implode( ',', array_fill( 0, count( $ids ), '%d' ) );
		$args = array_merge( array( Tables::links() ), $ids );
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber, PluginCheck.Security.DirectDB.UnescapedDBParameter -- one %d per ID, built above.
		$rows   = $wpdb->get_results( $wpdb->prepare( "SELECT target_id, COUNT(DISTINCT source_id) AS n FROM %i WHERE kind = 'link' AND source_id <> target_id AND target_id IN ({$in}) GROUP BY target_id", ...$args ) );
		$counts = array();
		foreach ( $rows as $row ) {
			$counts[ (int) $row->target_id ] = (int) $row->n;
		}
		return $counts;
	}

	/**
	 * Records one rendered page.
	 *
	 * @param int    $snapshot_id Snapshot.
	 * @param string $url         Page URL.
	 */
	private function rendered_item( $snapshot_id, $url ) {
		global $wpdb;
		$page       = $this->head->inspect( $url );
		$normalized = Url::normalize( $url, home_url( '/' ) );
		$noindex    = isset( $page['robots_values'] ) && preg_grep( '/noindex/', $page['robots_values'] );
		$noindex    = $noindex || ( isset( $page['x_robots'] ) && false !== stripos( $page['x_robots'], 'noindex' ) );
		$duplicates = isset( $page['problems'] ) && array_filter(
			$page['problems'],
			function ( $p ) {
				return 0 === strpos( $p[1], 'duplicate_' );
			}
		);
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$wpdb->insert(
			Tables::snapshot_items(),
			array(
				'snapshot_id' => $snapshot_id,
				'kind'        => 2,
				'url_hash'    => Url::hash( $normalized ),
				'url'         => mb_substr( $normalized, 0, 500 ),
				'object_id'   => (int) url_to_postid( $normalized ),
				'status'      => isset( $page['status'] ) ? (int) $page['status'] : -1,
				'noindex'     => $noindex ? 1 : 0,
				'canonical'   => isset( $page['canonicals'][0] ) ? mb_substr( Url::normalize( $page['canonicals'][0], home_url( '/' ) ), 0, 500 ) : '',
				'title'       => isset( $page['titles'][0] ) ? mb_substr( $page['titles'][0], 0, 255 ) : '',
				'duplicates'  => $duplicates ? 1 : 0,
			)
		);
	}

	/**
	 * Completes a snapshot and lets the monitor compare it.
	 *
	 * @param int $snapshot_id Snapshot.
	 */
	private function finish( $snapshot_id ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$wpdb->update(
			Tables::snapshots(),
			array(
				'status'       => 'complete',
				'completed_at' => current_time( 'mysql', true ),
			),
			array( 'id' => $snapshot_id )
		);
		$this->prune( (int) $this->settings->get( 'snapshot_retention' ) );

		/**
		 * Fires when a snapshot is complete.
		 *
		 * @param int $snapshot_id Snapshot ID.
		 */
		do_action( 'orbitstem_seo_snapshot_complete', $snapshot_id );
	}

	/**
	 * Keeps the newest $keep complete snapshots.
	 *
	 * @param int $keep Snapshots to keep.
	 */
	public function prune( $keep ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$old = $wpdb->get_col( $wpdb->prepare( "SELECT id FROM %i WHERE status = 'complete' ORDER BY id DESC LIMIT 1000 OFFSET %d", Tables::snapshots(), max( 2, (int) $keep ) ) );
		foreach ( $old as $id ) {
			$this->delete( (int) $id );
		}
	}

	/**
	 * Deletes one snapshot and its rows.
	 *
	 * @param int $id Snapshot.
	 */
	public function delete( $id ) {
		global $wpdb;
		// phpcs:disable WordPress.DB.DirectDatabaseQuery
		$wpdb->delete( Tables::snapshot_items(), array( 'snapshot_id' => (int) $id ), array( '%d' ) );
		$wpdb->delete( Tables::snapshots(), array( 'id' => (int) $id ), array( '%d' ) );
		// phpcs:enable
	}

	/**
	 * Recent snapshots, newest first.
	 *
	 * @param int $limit Rows.
	 * @return array<int,array>
	 */
	public function all( $limit = 50 ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$rows = $wpdb->get_results( $wpdb->prepare( 'SELECT s.*, (SELECT COUNT(*) FROM %i i WHERE i.snapshot_id = s.id) AS items FROM %i s ORDER BY s.id DESC LIMIT %d', Tables::snapshot_items(), Tables::snapshots(), (int) $limit ), ARRAY_A );
		foreach ( $rows as &$row ) {
			$row['site']    = json_decode( (string) $row['site'], true );
			$row['summary'] = json_decode( (string) $row['summary'], true );
		}
		return $rows;
	}

	/**
	 * One snapshot header.
	 *
	 * @param int $id Snapshot.
	 * @return array|null
	 */
	public function get( $id ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$row = $wpdb->get_row( $wpdb->prepare( 'SELECT * FROM %i WHERE id = %d', Tables::snapshots(), (int) $id ), ARRAY_A );
		if ( $row ) {
			$row['site']    = json_decode( (string) $row['site'], true );
			$row['summary'] = json_decode( (string) $row['summary'], true );
		}
		return $row ? $row : null;
	}

	/**
	 * The complete snapshot before a given one.
	 *
	 * @param int $id Snapshot.
	 * @return int 0 when none.
	 */
	public function previous( $id ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		return (int) $wpdb->get_var( $wpdb->prepare( "SELECT id FROM %i WHERE status = 'complete' AND id < %d ORDER BY id DESC LIMIT 1", Tables::snapshots(), (int) $id ) );
	}

	/**
	 * Stores a diff summary on a snapshot.
	 *
	 * @param int   $id      Snapshot.
	 * @param array $summary Summary.
	 */
	public function set_summary( $id, array $summary ) {
		global $wpdb;
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$wpdb->update( Tables::snapshots(), array( 'summary' => wp_json_encode( $summary ) ), array( 'id' => (int) $id ) );
	}
}