<?php
if (!defined('ABSPATH')) exit;

/** Automatic SEO and GEO baseline for every AI-managed content lifecycle. */
class AIWEST_Search_Optimization {
    const REPORT_META = '_aiwest_search_optimization_report';
    protected static $finalizing = false;

    public static function init() {
        add_action('wp_head', [__CLASS__, 'output_search_metadata'], 2);
        add_action('save_post', [__CLASS__, 'audit_saved_content'], 60, 3);
    }

    public static function audit_saved_content($post_id, $post, $update) {
        if (self::$finalizing || !$post || !in_array($post->post_type, ['page','post'], true)) return;
        if (wp_is_post_revision($post_id) || wp_is_post_autosave($post_id)) return;
        self::$finalizing = true;
        self::finalize($post_id, (string)$post->post_content, ['post_type'=>$post->post_type,'post_title'=>$post->post_title]);
        self::$finalizing = false;
    }

    public static function prompt_contract($context = []) {
        $host = strtolower((string) wp_parse_url(home_url('/'), PHP_URL_HOST));
        return "SEO AND GEO CONTRACT (mandatory):\n"
            . "- Satisfy the page's real user intent with accurate, useful, non-commodity content.\n"
            . "- Write a concise introduction, then use a logical H2/H3 hierarchy. Never add an H1 inside page content.\n"
            . "- Produce a truthful SEO title and meta description that match the visible content; avoid keyword stuffing.\n"
            . "- Make entities, product names, relationships, dates and definitions explicit and internally consistent.\n"
            . "- Make important answers easy to identify through descriptive headings, steps, lists, tables or FAQ when genuinely useful.\n"
            . "- Never invent statistics, quotations, customer claims, citations or external sources.\n"
            . "- Internal links must use verified existing URLs on {$host}; never use chatgpt.com, localhost or example.com as internal destinations.\n"
            . "- Structured information must agree with the visible content and the administrator's project data.\n"
            . "- Optimize for people first; do not create repetitive query variations or search-engine-only text.";
    }

    public static function normalize($post_id, $content, $context = []) {
        $title = (string)($context['post_title'] ?? ($post_id ? get_the_title($post_id) : ''));
        $post_type = (string)($context['post_type'] ?? ($post_id ? get_post_type($post_id) : 'page'));
        $normalized = ['content' => (string)$content, 'report' => []];
        if (class_exists('AIWEST_Page_Title_Heading_Normalizer')) {
            $normalized = AIWEST_Page_Title_Heading_Normalizer::normalize($post_id, (string)$content, [
                'post_title' => $title, 'post_type' => $post_type,
                'mode' => !empty($context['custom_hero']) ? 'custom_hero' : 'standard',
                'theme_title_visible' => empty($context['theme_title_hidden']),
            ]);
        }
        $normalized['report'] = self::analyze($title, $normalized['content'], $context, $normalized['report']);
        return $normalized;
    }

    public static function finalize($post_id, $content = null, $context = [], $provided = []) {
        $post = get_post($post_id);
        if (!$post) return [];
        $content = $content === null ? (string)$post->post_content : (string)$content;
        $title = (string)($context['post_title'] ?? $post->post_title);
        $report = self::analyze($title, $content, $context);
        $meta_title = sanitize_text_field($provided['meta_title'] ?? get_post_meta($post_id, '_aiwest_meta_title', true));
        if ($meta_title === '') $meta_title = $title;
        $description = sanitize_textarea_field($provided['meta_description'] ?? get_post_meta($post_id, '_aiwest_meta_description', true));
        if (preg_match('/\[(?:\/)?aiwest_[a-z0-9_-]*/i', $description)) $description = '';
        if ($description === '') $description = self::description($post->post_excerpt ?: $content);
        update_post_meta($post_id, '_aiwest_meta_title', $meta_title);
        update_post_meta($post_id, '_aiwest_meta_description', $description);
        update_post_meta($post_id, self::REPORT_META, $report);
        update_post_meta($post_id, '_aiwest_search_optimized_at', current_time('mysql'));
        return $report;
    }

    public static function analyze($title, $content, $context = [], $heading_report = []) {
        $plain = trim(preg_replace('/\s+/u', ' ', html_entity_decode(wp_strip_all_tags((string)$content, true), ENT_QUOTES, 'UTF-8')));
        preg_match_all('/<h([1-6])\b[^>]*>(.*?)<\/h\1>/is', $content, $headings, PREG_SET_ORDER);
        preg_match_all('/<a\b[^>]*\bhref\s*=\s*(["\'])(.*?)\1/is', $content, $links, PREG_SET_ORDER);
        $internal = 0; $external = 0; $invalid = [];
        $home_host = strtolower((string)wp_parse_url(home_url('/'), PHP_URL_HOST));
        foreach ($links as $link) {
            $url = html_entity_decode(trim($link[2]), ENT_QUOTES, 'UTF-8');
            if ($url === '' || $url[0] === '#' || stripos($url, 'mailto:') === 0 || stripos($url, 'tel:') === 0) continue;
            $host = strtolower((string)wp_parse_url($url, PHP_URL_HOST));
            if ($host === '' || $host === $home_host) $internal++; else $external++;
            if ($home_host === 'aiwebmaster.cloud' && in_array($host, ['chatgpt.com','www.chatgpt.com','localhost','example.com','www.example.com'], true)) $invalid[] = $url;
        }
        $h1_count = preg_match_all('/<h1\b/i', $content);
        $h2_count = preg_match_all('/<h2\b/i', $content);
        $question_count = preg_match_all('/<h[2-4]\b[^>]*>[^<]*(?:\?|faq|how|what|why|when|where|who)\b.*?<\/h[2-4]>/iu', $content);
        $structured = preg_match('/<(?:ul|ol|table|details)\b/i', $content) === 1;
        $word_count = self::word_count($plain);
        $warnings = [];
        if ($word_count < 120) $warnings[] = 'Content may be too short to satisfy the topic completely.';
        if ($h1_count > 0) $warnings[] = 'Page content contains an H1; the WordPress title should normally remain the only H1.';
        if ($h2_count < 1 && $word_count > 250) $warnings[] = 'Long content has no H2 sections.';
        if ($internal < 1 && $word_count > 250) $warnings[] = 'No internal contextual link was detected.';
        if ($invalid) $warnings[] = 'A prohibited internal-link destination was detected.';
        $seo = 35 + min(20, intval($word_count / 50)) + ($h2_count ? 15 : 0) + ($internal ? 15 : 0) + ($h1_count === 0 ? 15 : 0);
        $geo = 30 + min(20, intval($word_count / 60)) + ($structured ? 15 : 0) + ($question_count ? 10 : 0) + ($external ? 10 : 0) + ($h2_count ? 15 : 0);
        if ($invalid) { $seo -= 20; $geo -= 20; }
        return [
            'contract' => 'ADL-028', 'checked_at' => current_time('mysql'),
            'seo_score' => max(0, min(100, $seo)), 'geo_readiness_score' => max(0, min(100, $geo)),
            'word_count' => $word_count, 'h1_count' => $h1_count, 'h2_count' => $h2_count,
            'internal_links' => $internal, 'external_sources' => $external, 'question_headings' => $question_count,
            'structured_content' => $structured, 'schema_type' => self::schema_type($context, $content),
            'indexability' => 'review WordPress and SEO-plugin settings',
            'prohibited_link_targets' => array_values(array_unique($invalid)),
            'duplicate_title_heading' => $heading_report,
            'warnings' => $warnings, 'passed' => empty($warnings),
        ];
    }

    public static function output_search_metadata() {
        if (!is_singular() || self::seo_plugin_active()) return;
        $post_id = get_queried_object_id();
        if (!$post_id) return;
        $description = trim((string)get_post_meta($post_id, '_aiwest_meta_description', true));
        if ($description !== '') echo '<meta name="description" content="' . esc_attr($description) . '">' . "\n";
        $schema = self::schema($post_id, $description);
        if ($schema) {
            $json = wp_json_encode(
                $schema,
                JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT
            );
            if (is_string($json)) {
                wp_print_inline_script_tag($json, ['type' => 'application/ld+json']);
            }
        }
    }

    protected static function schema($post_id, $description) {
        $post = get_post($post_id);
        if (!$post) return [];
        $report = get_post_meta($post_id, self::REPORT_META, true);
        $type = is_array($report) ? (string)($report['schema_type'] ?? 'WebPage') : 'WebPage';
        $schema = [
            '@context' => 'https://schema.org',
            '@type' => in_array($type, ['WebPage','Article','FAQPage'], true) ? $type : 'WebPage',
            'name' => (string)$post->post_title,
            'url' => get_permalink($post_id),
        ];
        if ($description !== '') $schema['description'] = $description;
        if ($schema['@type'] === 'Article') {
            $schema['datePublished'] = get_post_time('c', true, $post);
            $schema['dateModified'] = get_post_modified_time('c', true, $post);
        } elseif ($schema['@type'] === 'FAQPage') {
            $entities = self::faq_entities((string)$post->post_content);
            if ($entities) $schema['mainEntity'] = $entities; else $schema['@type'] = 'WebPage';
        }
        return $schema;
    }

    protected static function faq_entities($content) {
        preg_match_all('/<details\b[^>]*>\s*<summary\b[^>]*>(.*?)<\/summary>(.*?)<\/details>/is', $content, $matches, PREG_SET_ORDER);
        $entities = [];
        foreach ($matches as $match) {
            $question = trim(wp_strip_all_tags($match[1]));
            $answer = trim(preg_replace('/\s+/u', ' ', wp_strip_all_tags($match[2])));
            if ($question === '' || $answer === '') continue;
            $entities[] = ['@type'=>'Question','name'=>$question,'acceptedAnswer'=>['@type'=>'Answer','text'=>$answer]];
        }
        return array_slice($entities, 0, 50);
    }

    protected static function schema_type($context, $content) {
        if (preg_match('/<details\b/i', $content)) return 'FAQPage';
        return (($context['post_type'] ?? '') === 'post') ? 'Article' : 'WebPage';
    }
    protected static function description($content) {
        $raw = (string) $content;
        $without_shortcodes = function_exists('strip_shortcodes') ? strip_shortcodes($raw) : preg_replace('/\[[^\]]+\]/', '', $raw);
        $text = trim(preg_replace('/\s+/u', ' ', wp_strip_all_tags((string) $without_shortcodes)));
        if ($text === '' && (has_shortcode($raw, 'aiwest_directory_map') || has_shortcode($raw, 'aiwest_company_map'))) {
            $text = sanitize_textarea_field((string) get_option(
                'aiwest_directory_map_description',
                __('Find our business location, branches or approved partners on the map.', 'ai-webmaster-studio')
            ));
        }
        return function_exists('mb_substr') ? mb_substr($text, 0, 160) : substr($text, 0, 160);
    }
    protected static function word_count($text) { preg_match_all('/[\p{L}\p{N}][\p{L}\p{N}\p{M}\x{2019}\'\-]*/u', (string)$text, $m); return count($m[0]); }
    protected static function seo_plugin_active() { return defined('WPSEO_VERSION') || defined('RANK_MATH_VERSION') || defined('AIOSEO_VERSION'); }
}
AIWEST_Search_Optimization::init();