<?php
declare(strict_types=1);
namespace TFGraphics\SmartImport\Search;

/** Deterministic editorial cleanup: no generated facts or external services. */
final class ProductDescriptionCleaner
{
    public function clean(?string $value, string $title = ''): ?string
    {
        if ($value === null) { return null; }
        $text = preg_replace('~<(script|style|nav|footer)\b[^>]*>.*?</\1>~is', '', $value) ?? $value;
        // Preserve word boundaries before stripping markup.
        $text = preg_replace('~<(?:br\b[^>]*|/p|/div|/li|/h[1-6]|/tr)>~i', "\n", $text) ?? $text;
        // phpcs:ignore WordPress.WP.AlternativeFunctions.strip_tags_strip_tags -- Standalone-test fallback is used only when WordPress is unavailable.
        $stripped = function_exists('wp_strip_all_tags') ? wp_strip_all_tags($text) : \strip_tags($text);
        $text = html_entity_decode($stripped, ENT_QUOTES | ENT_HTML5, 'UTF-8');
        $text = preg_replace('~(?:https?://|www\.)[^\s<>]+|[\w.+-]+@[\w.-]+\.[a-z]{2,}~iu', '', $text) ?? $text;
        $parts = preg_split('/\R+|(?<=[.!?])\s+/u', $text) ?: [];
        $kept = [];
        foreach ($parts as $part) {
            $part = trim((string) preg_replace('/[ \t]+/u', ' ', $part));
            // Cut vendor notices even when appended to product prose.
            $part = preg_replace('/(?:\s*[-–|]\s*)?(?:Prestalia\s+e-commerce\s+solutions|powered by\b|venduto da\b|sold by\b|fonte\s*:|source\s*:|descrizione (?:fornita|tratta) da\b|description (?:provided|taken) (?:by|from)\b|il prodotto verrà spedito\b|spedizione gratuita\b|free shipping\b|buyer protection\b|fast shipping\b|visita il nostro\b|visit our\b|contattaci\b|contact us\b|acquista (?:ora|sul nostro)\b|buy now\b).*$/iu', '', $part) ?? $part;
            $part = trim($part, " \t\n\r\0\x0B-–|");
            if ($part === '') { continue; }
            if ($title !== '' && str_starts_with($part, $title . ' ')) {
                $part = trim(substr($part, strlen($title)));
            }
            if ($part !== '' && !in_array($part, $kept, true)) { $kept[] = $part; }
        }
        return trim(implode("\n", $kept));
    }
}