<?php

declare(strict_types=1);

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

/**
 * Builds bounded, read-only Content Health snapshots from eligible public
 * WordPress pages and posts. This adapter never fetches a URL, renders a
 * theme, persists a result, or changes content or metadata.
 */
final class TheDotDevAgenticBridge_Content_Health_Snapshot_Adapter
{
    public const MAX_POSTS = 50;

    /** @return list<array<string, mixed>> */
    public static function snapshots_for_current_site(int $site_id): array
    {
        if ($site_id <= 0 || $site_id !== (int) get_current_blog_id()) {
            return [];
        }

        $snapshots = [];
        TheDotDevAgenticBridge_Content_Policy::each_outbound_candidate(
            static function (WP_Post $post) use (&$snapshots): bool {
                if (count($snapshots) >= self::MAX_POSTS) {
                    return false;
                }
                if (!TheDotDevAgenticBridge_Content_Policy::is_eligible_post($post)) {
                    return true;
                }

                $snapshot = self::snapshot_from_post($post);
                if ($snapshot !== null) {
                    $snapshots[] = $snapshot;
                }
                return true;
            }
        );

        return $snapshots;
    }

    /** @return array<string, mixed>|null */
    private static function snapshot_from_post(WP_Post $post): ?array
    {
        $permalink = (string) get_permalink($post);
        $path = self::path_from_permalink($permalink);
        $origin = rtrim((string) home_url('/'), '/');
        if ($path === '' || $origin === '') {
            return null;
        }

        $content = (string) $post->post_content;
        return [
            'siteId' => (int) get_current_blog_id(),
            'pageId' => (int) $post->ID,
            'path' => $path,
            'pageType' => $post->post_type === 'page' ? 'page' : 'post',
            'status' => 'publish',
            'title' => sanitize_text_field((string) get_the_title($post)),
            'metaDescription' => self::meta_description($post),
            // A post query cannot establish the rendered canonical tag. Keep
            // this unknown rather than making a claim about theme output.
            'canonical' => '',
            'canonicalKnown' => false,
            'robots' => TheDotDevAgenticBridge_Content_Policy::has_post_noindex($post) ? 'noindex' : '',
            'headings' => self::headings($content),
            'imageMissingAlt' => self::images_missing_alt($content),
            'decorativeImages' => 0,
            'contentWords' => self::word_count($content),
            'snapshotSource' => 'local_post_snapshot',
            'snapshotTime' => TheDotDevAgenticBridge_Readiness_Clock::format_mysql_gmt_as_utc_z(
                get_post_modified_time('Y-m-d H:i:s', true, $post)
            ),
            'scope' => 'indexable',
            'exclusions' => [],
            'siteOrigin' => $origin,
        ];
    }

    private static function path_from_permalink(string $permalink): string
    {
        $parts = wp_parse_url($permalink);
        if (!is_array($parts) || isset($parts['query'], $parts['fragment'])) {
            return '';
        }
        $path = (string) ($parts['path'] ?? '/');
        return str_starts_with($path, '/') ? $path : '';
    }

    private static function meta_description(WP_Post $post): string
    {
        foreach (['_yoast_wpseo_metadesc', 'rank_math_description', '_aioseo_description'] as $key) {
            $value = get_post_meta((int) $post->ID, $key, true);
            if (is_string($value) && trim($value) !== '') {
                return sanitize_text_field($value);
            }
        }
        return '';
    }

    /** @return list<array{level:int}> */
    private static function headings(string $content): array
    {
        preg_match_all('/<h([1-6])\\b[^>]*>/i', $content, $matches);
        return array_map(static fn(string $level): array => ['level' => (int) $level], $matches[1] ?? []);
    }

    private static function images_missing_alt(string $content): int
    {
        preg_match_all('/<img\\b[^>]*>/i', $content, $images);
        $missing = 0;
        foreach ($images[0] ?? [] as $image) {
            if (preg_match('/\\balt\\s*=\\s*(["\\\']).*?\\1/i', $image) !== 1) {
                ++$missing;
            }
        }
        return $missing;
    }

    private static function word_count(string $content): int
    {
        $plain = trim(wp_strip_all_tags($content));
        if ($plain === '') {
            return 0;
        }
        preg_match_all('/[^\\s]+/u', $plain, $words);
        return min(100000, count($words[0] ?? []));
    }
}