<?php
/**
 * Scan lifecycle: start, step, finish.
 *
 * @package A11yMate
 */

namespace A11yMate\Scanner;

use A11yMate\Rules\Registry;
use A11yMate\Settings;
use A11yMate\Store\Issues;
use A11yMate\Store\Scans;
use A11yMate\Store\Score;
use A11yMate\Plan;

defined( 'ABSPATH' ) || exit;

/**
 * Drives a scan. The same code backs the admin, WP-Cron and WP-CLI.
 */
class Runner {

	/**
	 * Start a scan.
	 *
	 * @param string $source        Source: admin, cron or cli.
	 * @param bool   $with_rendered Whether the browser will run the rendered pass.
	 * @param array  $overrides     Config overrides (max_pages, post_types...).
	 * @return int|\WP_Error
	 */
	public static function start( $source, $with_rendered, array $overrides = array() ) {
		$active = Scans::active();
		if ( $active ) {
			// Abandon stale scans (nothing happened for an hour) rather than blocking forever.
			$started = strtotime( $active['started_at'] . ' UTC' );
			if ( 'admin' === $source && $started && ( time() - $started ) < DAY_IN_SECONDS ) {
				return new \WP_Error( 'a11ymate_busy', __( 'A scan is already in progress. Resume it or cancel it first.', 'a11ymate' ), array( 'status' => 409 ) );
			}
			Scans::update(
				$active['id'],
				array(
					'status'      => 'cancelled',
					'finished_at' => gmdate( 'Y-m-d H:i:s' ),
				)
			);
		}
		$config = array_merge( Settings::all(), $overrides );
		$urls   = Discovery::urls( $config );
		if ( ! $urls ) {
			return new \WP_Error( 'a11ymate_no_urls', __( 'No pages were found to scan.', 'a11ymate' ), array( 'status' => 400 ) );
		}
		$snapshot = array(
			'max_pages'  => (int) $config['max_pages'],
			'post_types' => $config['post_types'],
			'target'     => $config['wcag_target'],
			'categories' => $config['categories'],
		);
		$id       = Scans::create( $source, $urls, $with_rendered, $snapshot );
		return $id;
	}

	/**
	 * Run the server pass over the next few pages.
	 *
	 * @param int $scan_id Scan.
	 * @param int $batch   Pages per call.
	 * @param int $budget  Seconds to spend at most.
	 * @return array Progress.
	 */
	public static function step( $scan_id, $batch = 3, $budget = 20 ) {
		$scan = Scans::get( $scan_id );
		if ( ! $scan || 'running' !== $scan['status'] ) {
			return self::progress( $scan_id );
		}
		$started = microtime( true );
		$done    = 0;
		foreach ( Scans::pending_pages( $scan_id, 'server', $batch ) as $page ) {
			self::scan_page( $scan, $page );
			++$done;
			if ( ( microtime( true ) - $started ) > $budget ) {
				break;
			}
		}
		Issues::refresh_counts( $scan_id );
		$progress = self::progress( $scan_id );
		if ( $progress['server_pending'] < 1 && ! $scan['with_rendered'] ) {
			self::finish( $scan_id );
			$progress = self::progress( $scan_id );
		}
		return $progress;
	}

	/**
	 * Fetch and analyse one page.
	 *
	 * @param array $scan Scan.
	 * @param array $page Page row.
	 * @return void
	 */
	private static function scan_page( array $scan, array $page ) {
		$res  = Fetcher::fetch( $page['url'] );
		$data = array(
			'server_done' => 1,
			'http_code'   => $res['code'],
		);
		$ok   = $res['code'] >= 200 && $res['code'] < 400;
		if ( '404' === $page['kind'] && 404 === $res['code'] && '' !== $res['html'] ) {
			$ok = true;
		}
		if ( ! $ok || '' === $res['html'] ) {
			$data['error'] = '' !== $res['error'] ? $res['error'] : sprintf(
				/* translators: %d: HTTP status */
				__( 'The page returned HTTP %d.', 'a11ymate' ),
				$res['code']
			);
			$data['rendered_done'] = 1;
			Scans::update_page( $page['id'], $data );
			Scans::recount( $scan['id'] );
			return;
		}
		$out           = ServerScan::analyse( $res['html'], $page['url'] );
		$data['title'] = $out['title'];
		Issues::record( $scan['id'], $page['id'], $page['url'], $out['findings'], 'all' );
		$data['issue_count'] = count( $out['findings'] );
		Scans::update_page( $page['id'], $data );
		Scans::recount( $scan['id'] );
	}

	/**
	 * Save what the browser found on one page.
	 *
	 * @param int   $scan_id  Scan.
	 * @param int   $page_id  Page.
	 * @param array $findings Findings, each with a viewport key.
	 * @param array $meta     Studio samples.
	 * @param bool  $failed   Whether the browser could not examine the page.
	 * @return array|\WP_Error
	 */
	public static function save_rendered( $scan_id, $page_id, array $findings, array $meta, $failed = false ) {
		$page = Scans::page( $page_id );
		if ( ! $page || (int) $page['scan_id'] !== (int) $scan_id ) {
			return new \WP_Error( 'a11ymate_bad_page', __( 'That page is not part of this scan.', 'a11ymate' ), array( 'status' => 404 ) );
		}
		if ( $failed ) {
			Scans::update_page(
				$page_id,
				array(
					'rendered_done' => 1,
					'error'         => __( 'The browser could not examine this page (it may stop being shown in a frame).', 'a11ymate' ),
				)
			);
			Scans::recount( $scan_id );
			return self::progress( $scan_id );
		}
		$enabled = array_flip( Registry::enabled_ids( 'rendered' ) );
		$by_vp   = array();
		foreach ( $findings as $f ) {
			if ( empty( $f['rule'] ) || ! isset( $enabled[ $f['rule'] ] ) ) {
				continue;
			}
			$vp             = isset( $f['viewport'] ) && in_array( $f['viewport'], array( 'desktop', 'mobile' ), true ) ? $f['viewport'] : 'all';
			$by_vp[ $vp ][] = array(
				'rule'     => sanitize_key( $f['rule'] ),
				'selector' => isset( $f['selector'] ) ? self::clean_code( $f['selector'] ) : '',
				'snippet'  => isset( $f['snippet'] ) ? self::clean_code( $f['snippet'] ) : '',
				'detail'   => isset( $f['detail'] ) && is_array( $f['detail'] ) ? $f['detail'] : array(),
			);
		}
		$stored = 0;
		foreach ( $by_vp as $vp => $list ) {
			$stored += Issues::record( $scan_id, $page_id, $page['url'], $list, $vp );
		}
		Scans::update_page(
			$page_id,
			array(
				'rendered_done' => 1,
				'issue_count'   => (int) $page['issue_count'] + $stored,
			)
		);
		Scans::recount( $scan_id );
		Issues::refresh_counts( $scan_id );
		return self::progress( $scan_id );
	}

	/**
	 * Keep markup-looking text intact (it is data, shown with textContent), but strip control characters.
	 *
	 * @param mixed $value Value.
	 * @return string
	 */
	private static function clean_code( $value ) {
		$value = wp_check_invalid_utf8( (string) $value );
		return substr( preg_replace( '/[\x00-\x08\x0B\x0C\x0E-\x1F]/', '', $value ), 0, 600 );
	}

	/**
	 * Progress for the UI.
	 *
	 * @param int $scan_id Scan.
	 * @return array
	 */
	public static function progress( $scan_id ) {
		$scan = Scans::get( $scan_id );
		if ( ! $scan ) {
			return array();
		}
		$counts = Scans::recount( $scan_id );
		$scan   = Scans::get( $scan_id );
		$total  = max( 1, $counts['total'] );
		$steps  = $scan['with_rendered'] ? 2 : 1;
		$done   = $counts['server_done'] + ( $scan['with_rendered'] ? $counts['rendered_done'] : 0 );
		$open   = Issues::counts();
		return array(
			'id'               => $scan['id'],
			'status'           => $scan['status'],
			'source'           => $scan['source'],
			'with_rendered'    => $scan['with_rendered'],
			'total'            => $counts['total'],
			'server_done'      => $counts['server_done'],
			'server_pending'   => $counts['total'] - $counts['server_done'],
			'rendered_done'    => $counts['rendered_done'],
			'rendered_pending' => $scan['with_rendered'] ? $counts['total'] - $counts['rendered_done'] : 0,
			'errors'           => $counts['errors'],
			'percent'          => (int) round( 100 * $done / ( $total * $steps ) ),
			'issues'           => $open,
			'score'            => $scan['score'],
		);
	}

	/**
	 * Finish a scan: scores, fixed issues, summary, hooks.
	 *
	 * @param int $scan_id Scan.
	 * @return array|null Summary.
	 */
	public static function finish( $scan_id ) {
		global $wpdb;
		$scan = Scans::get( $scan_id );
		if ( ! $scan || 'complete' === $scan['status'] ) {
			return $scan ? $scan['summary'] : null;
		}
		$prev   = Scans::latest_complete();
		$passes = array( 'server' );
		if ( $scan['with_rendered'] ) {
			$passes[] = 'rendered';
		}
		self::duplicate_titles( $scan_id );
		Issues::refresh_counts( $scan_id );
		$max_pages = isset( $scan['config']['max_pages'] ) ? (int) $scan['config']['max_pages'] : 0;
		$full      = $max_pages > 0 && (int) $scan['total_pages'] < $max_pages;
		$fixed     = Issues::mark_fixed( $scan_id, $passes, Registry::enabled_ids(), $full );

		// Page scores.
		$issues   = \A11yMate\Schema::table( 'issues' );
		$occ      = \A11yMate\Schema::table( 'occurrences' );
		$rows     = (array) $wpdb->get_results( // phpcs:ignore WordPress.DB.DirectDatabaseQuery
			$wpdb->prepare(
				"SELECT o.page_id, i.severity, COUNT(*) AS n FROM %i o JOIN %i i ON i.id = o.issue_id WHERE o.scan_id = %d AND i.mode = 'automated' AND i.state = 'open' GROUP BY o.page_id, i.rule_id, i.severity",
				$occ,
				$issues,
				$scan_id
			),
			ARRAY_A
		);
		$per_page = array();
		foreach ( $rows as $r ) {
			$per_page[ $r['page_id'] ][] = array(
				'severity' => $r['severity'],
				'n'        => (int) $r['n'],
			);
		}
		foreach ( Scans::pages( $scan_id ) as $p ) {
			if ( ! empty( $p['error'] ) ) {
				continue;
			}
			$entries = isset( $per_page[ $p['id'] ] ) ? $per_page[ $p['id'] ] : array();
			Scans::update_page( $p['id'], array( 'score' => Score::page( $entries ) ) );
		}

		$site     = Score::site();
		$ok_pages = 0;
		foreach ( Scans::pages( $scan_id ) as $p ) {
			if ( empty( $p['error'] ) ) {
				++$ok_pages;
			}
		}
		if ( 0 === $ok_pages ) {
			Scans::update(
				$scan_id,
				array(
					'status'      => 'failed',
					'finished_at' => gmdate( 'Y-m-d H:i:s' ),
					'summary'     => wp_json_encode( array( 'error' => __( 'None of the pages could be fetched. A11yMate loads your pages from the server itself; check the site can reach its own address, or add credentials with the a11ymate_fetch_args filter if the site is behind a login.', 'a11ymate' ) ) ),
				)
			);
			return null;
		}
		$new    = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM %i WHERE first_scan = %d', $issues, $scan_id ) ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$counts = Issues::counts();
		$pages  = Scans::pages( $scan_id );
		$errors = 0;
		foreach ( $pages as $p ) {
			if ( ! empty( $p['error'] ) ) {
				++$errors;
			}
		}
		$summary = array(
			'score'          => $site['score'],
			'previous_score' => $prev ? $prev['score'] : null,
			'categories'     => $site['categories'],
			'severity'       => $counts['severity'],
			'review'         => $counts['review'],
			'open'           => $counts['open'],
			'new'            => $new,
			'fixed'          => $fixed,
			'pages'          => count( $pages ),
			'page_errors'    => $errors,
			'rendered'       => (bool) $scan['with_rendered'],
		);
		Scans::update(
			$scan_id,
			array(
				'status'      => 'complete',
				'finished_at' => gmdate( 'Y-m-d H:i:s' ),
				'score'       => $site['score'],
				'summary'     => wp_json_encode( $summary ),
			)
		);
		/**
		 * Fires when a scan has finished.
		 *
		 * @param int   $scan_id Scan id.
		 * @param array $summary Summary: score, severity counts, new, fixed...
		 */
		do_action( 'a11ymate_scan_complete', $scan_id, $summary );
		return $summary;
	}

	/**
	 * Cross-page check: pages that share a title.
	 *
	 * @param int $scan_id Scan.
	 * @return void
	 */
	private static function duplicate_titles( $scan_id ) {
		global $wpdb;
		if ( ! in_array( 'title-duplicate', Registry::enabled_ids( 'server' ), true ) ) {
			return;
		}
		$t     = \A11yMate\Schema::table( 'pages' );
		$dupes = $wpdb->get_col( $wpdb->prepare( "SELECT title FROM %i WHERE scan_id = %d AND title IS NOT NULL AND title <> '' AND error IS NULL AND kind <> '404' GROUP BY title HAVING COUNT(*) > 1", $t, $scan_id ) ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery
		foreach ( (array) $dupes as $title ) {
			$pages = $wpdb->get_results( $wpdb->prepare( "SELECT id, url FROM %i WHERE scan_id = %d AND title = %s AND kind <> '404'", $t, $scan_id, $title ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery
			foreach ( $pages as $p ) {
				Issues::record(
					$scan_id,
					$p['id'],
					$p['url'],
					array(
						array(
							'rule'     => 'title-duplicate',
							'selector' => 'html > head > title',
							'snippet'  => '<title>' . $title . '</title>',
							'detail'   => array(
								'title' => $title,
								'pages' => count( $pages ),
							),
						),
					)
				);
			}
		}
	}

	/**
	 * Pause, resume or cancel.
	 *
	 * @param int    $scan_id Scan.
	 * @param string $status  running|paused|cancelled.
	 * @return array
	 */
	public static function set_status( $scan_id, $status ) {
		if ( in_array( $status, array( 'running', 'paused', 'cancelled' ), true ) ) {
			$data = array( 'status' => $status );
			if ( 'cancelled' === $status ) {
				$data['finished_at'] = gmdate( 'Y-m-d H:i:s' );
			}
			Scans::update( $scan_id, $data );
		}
		return self::progress( $scan_id );
	}

	/**
	 * Run a whole server-pass scan in one go (CLI) with an optional progress callback.
	 *
	 * @param string        $source   cli|cron.
	 * @param array         $overrides Overrides.
	 * @param callable|null $tick     Called with the progress array after each batch.
	 * @param int           $budget   Seconds before returning (0 for unlimited).
	 * @return int|\WP_Error Scan id.
	 */
	public static function run_all( $source, array $overrides = array(), $tick = null, $budget = 0 ) {
		$id = self::start( $source, false, $overrides );
		if ( is_wp_error( $id ) ) {
			return $id;
		}
		self::continue_scan( $id, $tick, $budget );
		return $id;
	}

	/**
	 * Keep stepping a scan until it completes or the time budget runs out.
	 *
	 * @param int           $scan_id Scan.
	 * @param callable|null $tick    Callback.
	 * @param int           $budget  Seconds (0 for unlimited).
	 * @return array Progress.
	 */
	public static function continue_scan( $scan_id, $tick = null, $budget = 0 ) {
		$t0 = time();
		do {
			$p = self::step( $scan_id, 3, 20 );
			if ( $tick ) {
				call_user_func( $tick, $p );
			}
			if ( empty( $p ) || 'running' !== $p['status'] ) {
				break;
			}
		} while ( $p['server_pending'] > 0 && ( 0 === $budget || ( time() - $t0 ) < $budget ) );
		return $p;
	}
}