<?php
/**
 * Keyword matching helper.
 *
 * @package 7sky-internal-post-linker
 */

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

/**
 * Filters generic phrases and counts whole-phrase matches.
 */
final class skyinpo_Keyword_Matcher {
    /**
     * Cached stopwords.
     *
     * @var string[]|null
     */
    private $stopwords = null;

    /**
     * Build useful phrase candidates from a title or slug.
     *
     * @param string $phrase Full title or slug phrase.
     * @return string[]
     */
    public function get_title_phrase_candidates($phrase) {
        $words = preg_split('/\s+/', $phrase);
        $candidates = array();
        $count = count($words);

        for ($length = $count; $length >= 1; $length--) {
            for ($offset = 0; $offset <= $count - $length; $offset++) {
                $slice = array_slice($words, $offset, $length);
                $candidate = preg_replace('/[^\p{L}\p{N}\s&+-]/u', '', implode(' ', $slice));
                $candidate = preg_replace('/\s+/', ' ', trim($candidate));

                if (!$this->is_linkworthy_phrase($candidate)) {
                    continue;
                }

                $candidates[strtolower($candidate)] = $candidate;
            }
        }

        return array_values($candidates);
    }

    /**
     * Determine whether a generated keyword phrase is specific enough for linking.
     *
     * @param string $phrase Candidate phrase.
     * @return bool
     */
    public function is_linkworthy_phrase($phrase) {
        $phrase = preg_replace('/\s+/', ' ', trim(wp_strip_all_tags((string) $phrase)));
        $words = $this->get_normalized_phrase_words($phrase);
        $word_count = count($words);
        $meaningful_words = array();

        if (strlen($phrase) < 4 || 0 === $word_count) {
            return false;
        }

        if (in_array($words[0], $this->get_stopwords(), true) || in_array($words[$word_count - 1], $this->get_stopwords(), true)) {
            return false;
        }

        foreach ($words as $word) {
            if (!in_array($word, $this->get_stopwords(), true) && strlen($word) > 2) {
                $meaningful_words[] = $word;
            }
        }

        if (empty($meaningful_words)) {
            return false;
        }

        if (1 === $word_count) {
            return strlen($meaningful_words[0]) >= 5 || preg_match('/\d/', $meaningful_words[0]);
        }

        return true;
    }

    /**
     * Count whole-phrase matches in plain editor text.
     *
     * @param string $text Plain text.
     * @param string $phrase Candidate phrase.
     * @return int
     */
    public function count_phrase_matches($text, $phrase) {
        $pattern = '/(?<![A-Za-z0-9])' . preg_quote($phrase, '/') . '(?![A-Za-z0-9])/i';
        preg_match_all($pattern, $text, $matches);

        return count($matches[0]);
    }

    /**
     * Normalize a phrase into lowercase words.
     *
     * @param string $phrase Candidate phrase.
     * @return string[]
     */
    private function get_normalized_phrase_words($phrase) {
        $words = preg_split('/\s+/', strtolower((string) $phrase));
        $normalized = array();

        foreach ($words as $word) {
            $word = preg_replace('/[^\p{L}\p{N}&+-]/u', '', $word);
            if ('' !== $word) {
                $normalized[] = $word;
            }
        }

        return $normalized;
    }

    /**
     * Return common filler words that should not drive linking suggestions.
     *
     * @return string[]
     */
    private function get_stopwords() {
        if (null === $this->stopwords) {
            $this->stopwords = require SKYINPO_PLUGIN_PATH . 'includes/data/skyinpo-stopwords.php';
        }

        return $this->stopwords;
    }
}