<?php
/**
 * AEO BOB - AEO Score Analyzer
 * Calculates Answer Engine Optimization scores for posts
 */

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

class AEO_Bob_AEO_Analyzer {

    private static $instance = null;

    // AEO Score weights (adjustable constants)
    const WEIGHT_QA_BLOCKS = 15;
    const WEIGHT_STRUCTURED_ANSWERS = 10;
    const WEIGHT_SCHEMA_JSON_LD = 15;
    const WEIGHT_CITATIONS = 10;
    const WEIGHT_INTERNAL_LINKS = 10;
    const WEIGHT_READABILITY = 10;
    const WEIGHT_ENTITY_COVERAGE = 10;
    const WEIGHT_DISCOVERABILITY = 10;
    const WEIGHT_SAFETY_BRAND = 10;

    public static function get_instance() {
        if (null === self::$instance) {
            self::$instance = new self();
        }
        return self::$instance;
    }

    private function __construct() {
        // Initialize hooks if needed
    }

    /**
     * Calculate AEO score for a post
     */
    public function calculate_aeo_score($post_id) {
        $post = get_post($post_id);
        if (!$post) {
            return 0;
        }

        $score = 0;
        $details = array();

        // Get post content and meta
        $content = $post->post_content;
        $title = $post->post_title;
        $meta_description = get_post_meta($post_id, '_yoast_wpseo_metadesc', true);

        // 1. Q&A/AEO blocks present (+15)
        $qa_score = $this->check_qa_blocks($content);
        $score += $qa_score;
        $details['qa_blocks'] = $qa_score;

        // 2. Structured answers & clarity (+10)
        $structured_score = $this->check_structured_answers($content, $title);
        $score += $structured_score;
        $details['structured_answers'] = $structured_score;

        // 3. Schema JSON-LD valid (+15)
        $schema_score = $this->check_schema_json_ld($post_id);
        $score += $schema_score;
        $details['schema_json_ld'] = $schema_score;

        // 4. Citations & support (+10)
        $citations_score = $this->check_citations($content);
        $score += $citations_score;
        $details['citations'] = $citations_score;

        // 5. Internal linking (+10)
        $internal_links_score = $this->check_internal_links($content);
        $score += $internal_links_score;
        $details['internal_links'] = $internal_links_score;

        // 6. Readability (+10)
        $readability_score = $this->check_readability($content);
        $score += $readability_score;
        $details['readability'] = $readability_score;

        // 7. Entity coverage (+10)
        $entity_score = $this->check_entity_coverage($content, $title);
        $score += $entity_score;
        $details['entity_coverage'] = $entity_score;

        // 8. llms.txt + engine prompt discoverable (+10)
        $discoverability_score = $this->check_discoverability($post_id);
        $score += $discoverability_score;
        $details['discoverability'] = $discoverability_score;

        // 9. Safety/brand constraints met (+10)
        $safety_score = $this->check_safety_brand($content);
        $score += $safety_score;
        $details['safety_brand'] = $safety_score;

        // Ensure score is between 0-100
        $score = max(0, min(100, $score));

        // Save score and details
        update_post_meta($post_id, '_aeo_bob_aeo_score', $score);
        update_post_meta($post_id, '_aeo_bob_aeo_details', $details);

        return $score;
    }

    /**
     * Check for Q&A/AEO blocks
     */
    private function check_qa_blocks($content) {
        $qa_patterns = array(
            '/<h[2-6][^>]*>.*?(?:faq|question|q&a|q and a|frequently asked).*?<\/h[2-6]>/i',
            '/<div[^>]*class[^>]*faq[^>]*>/i',
            '/<section[^>]*class[^>]*faq[^>]*>/i',
            '/\b(?:what|how|why|when|where|who)\b.*?\?/i',
            '/<dl[^>]*>.*?<dt[^>]*>.*?<\/dt>.*?<dd[^>]*>.*?<\/dd>.*?<\/dl>/s'
        );

        foreach ($qa_patterns as $pattern) {
            if (preg_match($pattern, $content)) {
                return self::WEIGHT_QA_BLOCKS;
            }
        }

        return 0;
    }

    /**
     * Check for structured answers within first 180 words
     */
    private function check_structured_answers($content, $title) {
        $clean_content = wp_strip_all_tags($content);
        $first_180_words = implode(' ', array_slice(explode(' ', $clean_content), 0, 180));

        // Check if there's a clear answer structure
        $answer_indicators = array(
            'answer is', 'solution is', 'result is', 'conclusion is',
            'the answer', 'the solution', 'the result', 'the conclusion',
            'here\'s how', 'here is how', 'this means', 'this indicates'
        );

        foreach ($answer_indicators as $indicator) {
            if (stripos($first_180_words, $indicator) !== false) {
                return self::WEIGHT_STRUCTURED_ANSWERS;
            }
        }

        return 0;
    }

    /**
     * Check for valid Schema JSON-LD
     */
    private function check_schema_json_ld($post_id) {
        // Check if we have stored schema
        $schema = get_post_meta($post_id, '_aeobob_engine_prompt', true);
        if ($schema) {
            $schema_data = json_decode($schema, true);
            if ($schema_data && isset($schema_data['schema_ld'])) {
                return self::WEIGHT_SCHEMA_JSON_LD;
            }
        }

        // Check for existing schema in content
        $content = get_post_field('post_content', $post_id);
        if (preg_match('/<script[^>]*type[^>]*application\/ld\+json[^>]*>(.*?)<\/script>/s', $content, $matches)) {
            $json_ld = json_decode($matches[1], true);
            if ($json_ld && (isset($json_ld['@type']) && in_array($json_ld['@type'], array('Article', 'FAQPage')))) {
                return self::WEIGHT_SCHEMA_JSON_LD;
            }
        }

        return 0;
    }

    /**
     * Check for citations and external links
     */
    private function check_citations($content) {
        // Count external links
        preg_match_all('/<a[^>]*href[^>]*>/i', $content, $links);
        $external_links = 0;

        if (isset($links[0])) {
            foreach ($links[0] as $link) {
                if (preg_match('/href[^>]*https?:\/\/(?!' . preg_quote(wp_parse_url(home_url(), PHP_URL_HOST)) . ')/i', $link)) {
                    $external_links++;
                }
            }
        }

        return $external_links > 0 ? self::WEIGHT_CITATIONS : 0;
    }

    /**
     * Check for internal links
     */
    private function check_internal_links($content) {
        preg_match_all('/<a[^>]*href[^>]*>/i', $content, $links);
        $internal_links = 0;

        if (isset($links[0])) {
            foreach ($links[0] as $link) {
                if (preg_match('/href[^>]*https?:\/\/' . preg_quote(wp_parse_url(home_url(), PHP_URL_HOST)) . '/i', $link)) {
                    $internal_links++;
                }
            }
        }

        return $internal_links >= 2 ? self::WEIGHT_INTERNAL_LINKS : 0;
    }

    /**
     * Check readability (target grade 8-10)
     */
    private function check_readability($content) {
        $clean_content = wp_strip_all_tags($content);
        $sentences = preg_split('/[.!?]+/', $clean_content);
        $words = str_word_count($clean_content);
        $syllables = $this->count_syllables($clean_content);

        if (count($sentences) > 0 && $words > 0) {
            $avg_sentence_length = $words / count($sentences);
            $avg_syllables_per_word = $syllables / $words;

            // Flesch Reading Ease calculation
            $flesch_score = 206.835 - (1.015 * $avg_sentence_length) - (84.6 * $avg_syllables_per_word);

            // Grade 8-10 corresponds to Flesch scores of 60-70
            if ($flesch_score >= 60 && $flesch_score <= 70) {
                return self::WEIGHT_READABILITY;
            } elseif ($flesch_score >= 50 && $flesch_score <= 80) {
                return self::WEIGHT_READABILITY / 2; // Partial credit
            }
        }

        return 0;
    }

    /**
     * Count syllables in text
     */
    private function count_syllables($text) {
        $text = strtolower($text);
        $text = preg_replace('/[^a-z]/', '', $text);
        $syllables = 0;
        $previous_was_vowel = false;

        for ($i = 0; $i < strlen($text); $i++) {
            $is_vowel = in_array($text[$i], array('a', 'e', 'i', 'o', 'u', 'y'));
            if ($is_vowel && !$previous_was_vowel) {
                $syllables++;
            }
            $previous_was_vowel = $is_vowel;
        }

        return max(1, $syllables);
    }

    /**
     * Check entity coverage
     */
    private function check_entity_coverage($content, $title) {
        // Extract primary keyword
        $keyword = get_post_meta(get_the_ID(), '_aeo_bob_target_keyword', true);
        if (!$keyword) {
            return 0;
        }

        // Check if keyword appears in headings and body
        $headings = array();
        preg_match_all('/<h[1-6][^>]*>(.*?)<\/h[1-6]>/i', $content, $headings);

        $keyword_in_headings = false;
        $keyword_in_body = false;

        if (isset($headings[1])) {
            foreach ($headings[1] as $heading) {
                if (stripos($heading, $keyword) !== false) {
                    $keyword_in_headings = true;
                    break;
                }
            }
        }

        if (stripos($content, $keyword) !== false) {
            $keyword_in_body = true;
        }

        return ($keyword_in_headings && $keyword_in_body) ? self::WEIGHT_ENTITY_COVERAGE : 0;
    }

    /**
     * Check discoverability (llms.txt + engine prompt)
     */
    private function check_discoverability($post_id) {
        // Check if engine prompt exists
        $engine_prompt = get_post_meta($post_id, '_aeobob_engine_prompt', true);
        if (!$engine_prompt) {
            return 0;
        }

        // Check if llms.txt is accessible
        $upload_dir = wp_upload_dir();
        $llms_file = $upload_dir['basedir'] . '/llms.txt';

        if (!file_exists($llms_file)) {
            return self::WEIGHT_DISCOVERABILITY / 2; // Partial credit for engine prompt only
        }

        return self::WEIGHT_DISCOVERABILITY;
    }

    /**
     * Check safety and brand constraints
     */
    private function check_safety_brand($content) {
        $forbidden_phrases = array(
            'guaranteed results', '100% success', 'overnight success',
            'get rich quick', 'make money fast', 'secret formula',
            'limited time offer', 'act now', 'don\'t miss out'
        );

        $overclaiming_phrases = array(
            'best in the world', 'number one', 'revolutionary',
            'breakthrough', 'miracle', 'cure all'
        );

        $content_lower = strtolower($content);

        foreach ($forbidden_phrases as $phrase) {
            if (stripos($content_lower, $phrase) !== false) {
                return 0; // Fail if forbidden phrases found
            }
        }

        $overclaiming_count = 0;
        foreach ($overclaiming_phrases as $phrase) {
            if (stripos($content_lower, $phrase) !== false) {
                $overclaiming_count++;
            }
        }

        // Penalize for overclaiming but don't fail completely
        if ($overclaiming_count > 2) {
            return self::WEIGHT_SAFETY_BRAND / 2;
        }

        return self::WEIGHT_SAFETY_BRAND;
    }

    /**
     * Get AEO score for a post
     */
    public function get_aeo_score($post_id) {
        $score = get_post_meta($post_id, '_aeo_bob_aeo_score', true);
        return $score ? intval($score) : 0;
    }

    /**
     * Get AEO details for a post
     */
    public function get_aeo_details($post_id) {
        $details = get_post_meta($post_id, '_aeo_bob_aeo_details', true);
        return $details ? $details : array();
    }
}