<?php

namespace App\Services\Crawl;

/**
 * 投资界正文 HTML：补全图片地址，便于后台编辑器与前台展示。
 */
class PedailyContentNormalizer
{
    public static function normalize(?string $html): ?string
    {
        if ($html === null || trim($html) === '') {
            return null;
        }

        $html = html_entity_decode($html, ENT_QUOTES | ENT_HTML5, 'UTF-8');

        $html = preg_replace_callback(
            '/<img\b([^>]*)>/iu',
            static function (array $m): string {
                $attrs = $m[1];
                if (preg_match('/\sdata-src=(["\'])([^"\']+)\1/i', $attrs, $data)) {
                    $realSrc = $data[2];
                    if (preg_match('/\ssrc=(["\'])([^"\']*)\1/i', $attrs)) {
                        $attrs = preg_replace('/\ssrc=(["\'])[^"\']*\1/i', ' src="'.$realSrc.'"', $attrs);
                    } else {
                        $attrs .= ' src="'.$realSrc.'"';
                    }
                }

                if (! preg_match('/\sreferrerpolicy=/i', $attrs)) {
                    $attrs .= ' referrerpolicy="no-referrer"';
                }

                return '<img'.$attrs.'>';
            },
            $html
        ) ?? $html;

        $html = preg_replace('#\ssrc=(["\'])/(?!/)#', ' src=$1https://www.pedaily.cn/', $html) ?? $html;
        $html = preg_replace('#\ssrc=(["\'])//#', ' src=$1https://', $html) ?? $html;

        return trim($html);
    }
}