<?php
namespace Clusterlane\Api;
if ( ! defined( 'ABSPATH' ) ) exit;

class RestScanner {
    public function __construct( private \Clusterlane\Core\Plugin $plugin ) {}

    public function register_routes(): void {
        // Get all post IDs to scan.
        register_rest_route( 'clusterlane/v1', '/scan/posts', [
            'methods'             => 'GET',
            'callback'            => [ $this, 'get_post_ids' ],
            'permission_callback' => fn() => \Clusterlane\Core\Capabilities::can_use(),
        ] );

        // Scan a BATCH of posts in one request (fast).
        register_rest_route( 'clusterlane/v1', '/scan/batch', [
            'methods'             => 'POST',
            'callback'            => [ $this, 'scan_batch' ],
            'permission_callback' => fn() => \Clusterlane\Core\Capabilities::can_use(),
        ] );

        // Published pages waiting to be read after a scan (0.98.178): where it
        // stands (GET), or read the next ~20 s of them (POST).
        register_rest_route( 'clusterlane/v1', '/scan/pages', [
            [
                'methods'             => 'GET',
                'callback'            => fn() => rest_ensure_response( \Clusterlane\Core\PageReader::status() ),
                'permission_callback' => fn() => \Clusterlane\Core\Capabilities::can_use(),
            ],
            [
                'methods'             => 'POST',
                'callback'            => fn() => rest_ensure_response( \Clusterlane\Core\PageReader::run_queue() ),
                'permission_callback' => fn() => \Clusterlane\Core\Capabilities::can_use(),
            ],
        ] );

        // Scan a single post (kept for compatibility).
        register_rest_route( 'clusterlane/v1', '/scan/post/(?P<id>\d+)', [
            'methods'             => 'POST',
            'callback'            => [ $this, 'scan_post' ],
            'permission_callback' => fn() => \Clusterlane\Core\Capabilities::can_use(),
            'args'                => [
                'id' => [
                    'validate_callback' => fn( $v ) => is_numeric( $v ),
                    'sanitize_callback' => 'absint',
                ],
            ],
        ] );
    }

    public function get_post_ids(): \WP_REST_Response {
        $settings = $this->plugin->get('settings');
        // A full scan starts here: drop data for posts that left the scope
        // (unpublished, deleted or password-protected since the last scan).
        \Clusterlane\Core\Scope::purge_out_of_scope();
        \Clusterlane\Core\PageReader::reset_log();   // this scan's page-reading report starts empty
        \Clusterlane\Core\PageReader::reset_queue(); // and so does the queue of pages to read
        $ids      = $this->plugin->get('scanner')->get_all_post_ids( $settings->post_types() );
        return rest_ensure_response( [ 'ids' => $ids, 'total' => count( $ids ), 'reads_pages' => \Clusterlane\Core\PageReader::any_enabled() ] );
    }

    /**
     * Scan a batch of posts in one request.
     * Expects JSON body: { "ids": [1, 2, 3, ...] }
     * Returns: { "scanned": N, "results": [{post_id, links_found}, ...] }
     */
    /** Maximum post IDs accepted per batch request. */
    const MAX_BATCH_SIZE = 100;

    /** Seconds a batch may run before it hands the rest back (0.98.177). */
    const TIME_BUDGET = 20;

    public function scan_batch( \WP_REST_Request $request ): \WP_REST_Response {
        $raw = $request->get_json_params()['ids'] ?? [];

        if ( empty( $raw ) || ! is_array( $raw ) ) {
            return new \WP_REST_Response( [ 'error' => 'Missing ids array' ], 400 );
        }

        // Cast every element to int and enforce an upper limit to prevent
        // a single request from scanning thousands of posts and timing out.
        $ids = array_slice( array_map( 'intval', $raw ), 0, self::MAX_BATCH_SIZE );

        // A batch of 25 posts can exceed a strict host max_execution_time
        // (often 30s on shared hosting) because link extraction resolves
        // every URL. Without this guard the request dies mid-batch and
        // the whole batch is lost, which shows up as "only N of M scanned".
        if ( function_exists( 'set_time_limit' ) ) @set_time_limit( 300 ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged, Squiz.PHP.DiscouragedFunctions.Discouraged -- one scan batch on an admin's request; harmless where disabled
        ignore_user_abort( true );

        $scanner = $this->plugin->get('scanner');
        $results = [];
        $start   = microtime( true );
        $left    = [];

        // 'later' (the Scan button, 0.98.178): read the text now, queue the
        // published pages. 'now' (onboarding, older screens): read them here.
        $later = ( $request->get_json_params()['pages'] ?? 'now' ) === 'later';
        \Clusterlane\Core\PageReader::defer( $later );

        foreach ( $ids as $i => $post_id ) {
            $post_id = (int) $post_id;
            if ( ! $post_id ) continue;

            // Reading published pages takes a second or so each (0.98.177): stop after
            // ~20 s and hand the rest back, well inside hosts' and proxies' time limits.
            if ( $results && microtime( true ) - $start > self::TIME_BUDGET ) {
                $left = array_values( array_slice( $ids, $i ) );
                break;
            }
            $links_found = $scanner->scan_post( $post_id );
            $results[]   = [ 'post_id' => $post_id, 'links_found' => $links_found ];
        }
        \Clusterlane\Core\PageReader::defer( false );

        return rest_ensure_response( [
            'scanned'   => count( $results ),
            'results'   => $results,
            'remaining' => $left,                          // not scanned yet: the client sends them again
            'pages'     => \Clusterlane\Core\PageReader::flush_log(), // published pages read / failed in this batch
            'queued'    => \Clusterlane\Core\PageReader::flush_queue(), // published pages left for later (pages: 'later')
        ] );
    }

    public function scan_post( \WP_REST_Request $request ): \WP_REST_Response {
        $post_id = (int) $request['id'];
        $count   = $this->plugin->get('scanner')->scan_post( $post_id );
        return rest_ensure_response( [ 'post_id' => $post_id, 'links_found' => $count ] );
    }
}