<?php
/**
 * Internal link suggestion service.
 *
 * @package 7sky-internal-post-linker
 */

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

/**
 * Builds deterministic internal link suggestions from existing content.
 */
final class skyinpo_Link_Suggestions {
    /**
     * Content search helper.
     *
     * @var skyinpo_Content_Search
     */
    private $content_search;

    /**
     * Keyword matching helper.
     *
     * @var skyinpo_Keyword_Matcher
     */
    private $keyword_matcher;

    /**
     * Constructor.
     *
     * @param skyinpo_Content_Search  $content_search Content search helper.
     * @param skyinpo_Keyword_Matcher $keyword_matcher Keyword matching helper.
     */
    public function __construct(skyinpo_Content_Search $content_search, skyinpo_Keyword_Matcher $keyword_matcher) {
        $this->content_search = $content_search;
        $this->keyword_matcher = $keyword_matcher;
    }

    /**
     * Find deterministic internal-link suggestions from existing content titles and slugs.
     *
     * @param string $content Current editor content.
     * @param int    $current_post_id Current editor post ID.
     * @return array<int, array<string, mixed>>
     */
    public function get_link_suggestions($content, $current_post_id) {
        $scan_text = $this->get_suggestion_scan_text($content);
        if (strlen($scan_text) < 20) {
            return array();
        }

        $post_types = $this->content_search->get_searchable_post_types();
        if (empty($post_types)) {
            return array();
        }

        $post_statuses = array('publish');
        if (current_user_can('read_private_posts')) {
            $post_statuses[] = 'private';
        }

        $query = new WP_Query(
            array(
                'post_type' => $post_types,
                'post_status' => $post_statuses,
                'posts_per_page' => 200,
                'orderby' => 'modified',
                'order' => 'DESC',
                'no_found_rows' => true,
                'ignore_sticky_posts' => true,
            )
        );

        $suggestions = array();
        foreach ($query->posts as $post) {
            $best_suggestion = null;
            $phrases = $this->get_suggestion_phrases($post);

            foreach ($phrases as $phrase) {
                $count = $this->keyword_matcher->count_phrase_matches($scan_text, $phrase);
                if ($count < 1) {
                    continue;
                }

                $rating = $this->get_suggestion_rating($post, $phrase, $count);
                $suggestion = array(
                    'keyword' => $phrase,
                    'count' => $count,
                    'rating' => $rating['label'],
                    'ratingScore' => $rating['score'],
                    'specificityScore' => $this->get_phrase_specificity_score($phrase),
                    'targetContentCount' => $rating['targetContentCount'],
                    'id' => (int) $post->ID,
                    'title' => html_entity_decode(get_the_title($post), ENT_QUOTES, get_bloginfo('charset')),
                    'type' => $this->content_search->get_post_type_label($post->post_type),
                    'postType' => $post->post_type,
                    'url' => get_permalink($post),
                    'isCurrent' => absint($current_post_id) === (int) $post->ID,
                    'date' => get_the_date('', $post),
                    'excerpt' => $this->content_search->get_result_excerpt($post),
                );

                if (!$best_suggestion || $this->compare_suggestions($suggestion, $best_suggestion) < 0) {
                    $best_suggestion = $suggestion;
                }
            }

            if ($best_suggestion) {
                $suggestions[] = $best_suggestion;
            }
        }

        wp_reset_postdata();

        usort(
            $suggestions,
            function ($a, $b) {
                return $this->compare_suggestions($a, $b);
            }
        );

        return array_slice($suggestions, 0, 20);
    }

    /**
     * Compare suggestions with phrase specificity ahead of raw frequency.
     *
     * @param array<string, mixed> $a First suggestion.
     * @param array<string, mixed> $b Second suggestion.
     * @return int
     */
    private function compare_suggestions($a, $b) {
        if ($a['specificityScore'] !== $b['specificityScore']) {
            return $b['specificityScore'] <=> $a['specificityScore'];
        }

        if ($a['ratingScore'] !== $b['ratingScore']) {
            return $b['ratingScore'] <=> $a['ratingScore'];
        }

        if ($a['count'] !== $b['count']) {
            return $b['count'] <=> $a['count'];
        }

        return strlen($b['keyword']) <=> strlen($a['keyword']);
    }

    /**
     * Score phrase specificity so long-tail matches outrank generic single words.
     *
     * @param string $phrase Suggested keyword phrase.
     * @return int
     */
    private function get_phrase_specificity_score($phrase) {
        $words = preg_split('/\s+/', trim((string) $phrase));
        $word_count = is_array($words) ? count(array_filter($words)) : 0;

        if ($word_count >= 4) {
            return 400 + strlen($phrase);
        }

        if (3 === $word_count) {
            return 300 + strlen($phrase);
        }

        if (2 === $word_count) {
            return 200 + strlen($phrase);
        }

        return strlen($phrase);
    }

    /**
     * Rate a suggestion using deterministic match strength signals.
     *
     * @param WP_Post $post Target post.
     * @param string  $phrase Suggested keyword.
     * @param int     $count Keyword occurrence count in current article.
     * @return array{label: string, score: int, targetContentCount: int}
     */
    private function get_suggestion_rating(WP_Post $post, $phrase, $count) {
        $title = strtolower(html_entity_decode(get_the_title($post), ENT_QUOTES, get_bloginfo('charset')));
        $slug = strtolower(str_replace(array('-', '_'), ' ', $post->post_name));
        $excerpt = strtolower($this->content_search->get_result_excerpt($post));
        $keyword = strtolower((string) $phrase);
        $word_count = str_word_count($keyword);
        $target_content_count = $this->keyword_matcher->count_phrase_matches($this->get_target_content_scan_text($post), $phrase);
        $score = 0;

        if ($keyword && false !== strpos($title, $keyword)) {
            $score += 35;
        }

        if ($keyword && false !== strpos($slug, $keyword)) {
            $score += 20;
        }

        if ($keyword && false !== strpos($excerpt, $keyword)) {
            $score += 15;
        }

        if ($target_content_count > 0) {
            $score += 10;
            $score += min(20, $target_content_count * 4);
        }

        $score += min(20, max(0, (int) $count) * 5);
        $score += min(20, max(0, $word_count - 1) * 5);

        if ($keyword === $title || $keyword === $slug) {
            $score += 15;
        }

        if ($score >= 85) {
            return array('label' => __('Ideal', '7sky-internal-post-linker'), 'score' => $score, 'targetContentCount' => $target_content_count);
        }

        if ($score >= 65) {
            return array('label' => __('Best', '7sky-internal-post-linker'), 'score' => $score, 'targetContentCount' => $target_content_count);
        }

        if ($score >= 45) {
            return array('label' => __('Good', '7sky-internal-post-linker'), 'score' => $score, 'targetContentCount' => $target_content_count);
        }

        return array('label' => __('Normal', '7sky-internal-post-linker'), 'score' => $score, 'targetContentCount' => $target_content_count);
    }

    /**
     * Build plain target article text used only for server-side relevance scoring.
     *
     * @param WP_Post $post Target post.
     * @return string
     */
    private function get_target_content_scan_text(WP_Post $post) {
        $content = wp_strip_all_tags(strip_shortcodes((string) $post->post_content));
        $content = html_entity_decode($content, ENT_QUOTES, get_bloginfo('charset'));

        return preg_replace('/\s+/', ' ', trim($content));
    }

    /**
     * Strip linked text and markup so suggestions do not target existing links.
     *
     * @param string $content Current editor content.
     * @return string
     */
    private function get_suggestion_scan_text($content) {
        $content = preg_replace('#<a\b[^>]*>.*?</a>#is', ' ', $content);
        $content = wp_strip_all_tags(strip_shortcodes((string) $content));
        $content = html_entity_decode($content, ENT_QUOTES, get_bloginfo('charset'));

        return preg_replace('/\s+/', ' ', trim($content));
    }

    /**
     * Build useful non-AI match phrases from a target post.
     *
     * @param WP_Post $post Target post.
     * @return string[]
     */
    private function get_suggestion_phrases(WP_Post $post) {
        $phrases = array();
        $title = html_entity_decode(get_the_title($post), ENT_QUOTES, get_bloginfo('charset'));
        $slug = str_replace(array('-', '_'), ' ', $post->post_name);

        foreach (array($title, $slug) as $phrase) {
            $phrase = preg_replace('/\s+/', ' ', trim(wp_strip_all_tags((string) $phrase)));

            if ($this->keyword_matcher->is_linkworthy_phrase($phrase)) {
                $phrases[strtolower($phrase)] = $phrase;
            }

            foreach ($this->keyword_matcher->get_title_phrase_candidates($phrase) as $candidate) {
                $phrases[strtolower($candidate)] = $candidate;
            }
        }

        return array_values($phrases);
    }

}