dotnet/molilian.core/Plus/pdd/PddUnionPlus.cs

using dodohold.core;
using Sayaka.Common;
using System.Diagnostics;
using System.Text.RegularExpressions;
using System.Web;

namespace molilian.core
{
    public partial class PddUnionPlus
    {
        public static string _url_pattern = @"https?://(?:[\w-]+\.)*(?:yangkeduo\.com|pinduoduo\.com)(?:/[^\s]*)?";
        public static string GetLink(string content)
        {
            if (string.IsNullOrEmpty(content)) return content;
            string result = content;
            //
            Regex regex = new(_url_pattern);
            MatchCollection matches = regex.Matches(content);

            foreach (Match m in matches.Cast<Match>())
            {
                string url = m.Value;
                return url;
            }
            return string.Empty;
        }


        public static string GetDeeplink(string url)
        {
            string result = $"pinduoduo://com.xunmeng.pinduoduo/{url}";
            return result;
        }



        private string ExtractGoodsIdUrl(string url)
        {
            var uri = new Uri(url);
            var queryParams = HttpUtility.ParseQueryString(uri.Query);
            string goodsId = queryParams["goods_id"];

            if (!string.IsNullOrEmpty(goodsId) && uri.AbsolutePath.StartsWith("/goods"))
            {
                // 保持域名和路径不变，仅保留 goods_id 参数
                var baseUrl = $"{uri.Scheme}://{uri.Host}{uri.AbsolutePath}";
                return $"{baseUrl}?goods_id={goodsId}";
            }
            return url; // 如果没有找到 goods_id 参数，则返回原始 URL
        }



        private bool IsTrackUrl(string sourceUrl)
        {
            string pattern = @"^https?://mobile\.yangkeduo\.com/goods(\d*)\.html\?ps=.*$";
            Regex regex = new Regex(pattern);
            return regex.IsMatch(sourceUrl);

        }

        /// <summary>
        /// bool:是否ps链接，string 处理后的链接
        /// </summary>
        /// <param name="sourceUrl"></param>
        /// <param name="cancellationToken"></param>
        /// <returns></returns>
        public async Task<string> GetRedirectedUrlAsync(string sourceUrl, CancellationToken cancellationToken = default)
        {
            var isTrackUrl = IsTrackUrl(sourceUrl);
            if (!isTrackUrl) return ExtractGoodsIdUrl(sourceUrl);
            string redirectedUrl = sourceUrl;

            try
            {
                string useragent = ProviderFakeUserAgent.RandomMobile;
                WebClientUtility client = new()
                {
                    AllowAutoRedirect = false,
                    UserAgent = useragent,
                };
                client.Proxy = _proxy;
#if DEBUG
                client.Proxy = null;
#endif
                var response = await client.RequestAsync(sourceUrl, "GET", cancellationToken);
                if (!response.Successed) return null;

                if (response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.RedirectKeepVerb ||
                  response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.TemporaryRedirect)
                {
                    redirectedUrl = response.ResponseMessage.Headers.Location.ToString();
                    redirectedUrl = ExtractGoodsIdUrl(redirectedUrl);
                    return redirectedUrl;
                }
            }
            catch (Exception ex)
            {
            }
            return sourceUrl;
        }



        public static PddDataDTO GetFormattedObject(string content, string ip = "", string oaid = "")
        {
            string shortLinkurl = GetLink(content);
            string deeplink_url = GetDeeplink(shortLinkurl);
            var link_type = LinkTypeEnum.unknown;
            if (!string.IsNullOrEmpty(shortLinkurl) && shortLinkurl.StartsWith("https://p.pinduoduo.com/"))
            {
                link_type = LinkTypeEnum.other_aff;
            }

            return new PddDataDTO()
            {
                message = string.Empty,
                link_type = link_type,
                channel = TkChannelEnum.pdd,
                accountName = string.Empty,
                success = false,
                content = content,
                ip = ip,
                oaid = oaid,
                elapsedTime = 0,
                itemName = "点击打开拼多多APP",
                shortLinkurl = string.Empty,
                deeplink_url = string.Empty,
                create_time = DateTime.Now,
                end_point = _end_point,
            };
        }

        public async Task<PddDataDTO> PddParseAsync(string content, PddDataDTO result, CancellationToken cancellationToken = default)
        {
            string message = "OK";
            string url = GetLink(content);
            if (string.IsNullOrEmpty(url))
            {
                result.success = false;
                result.link_type = LinkTypeEnum.unknown;
                result.channel_type = ChannelTypeEnum.pdd;
                result.message = "放弃转链";
                result.reason = "无效链接";
                result.accountId = 0;
                result.accountName = string.Empty;
                result.content = content;
                return result;
            }
            string shortLinkurl = url;

            if (result.link_type == LinkTypeEnum.other_aff)
            {
                result.success = false;
                result.link_type = LinkTypeEnum.unknown;
                result.channel_type = ChannelTypeEnum.pdd;
                result.message = "放弃转链";
                result.reason = "其他推广链接";
                result.accountId = 0;
                result.accountName = string.Empty;
                result.content = content;
                return result;
            }

            if (!result.ip.StartsWith("127.0.0") && FlowControlIgnoreRequest(_config, out string reason))
            {
                result.success = false;
                result.link_type = LinkTypeEnum.unknown;
                result.channel_type = ChannelTypeEnum.pdd;
                result.message = "放弃转链";
                result.reason = reason;
                result.accountId = 0;
                result.accountName = string.Empty;
                result.content = content;
                return result;
            }


            var isTrackUrl = IsTrackUrl(url);
            //不是跟踪链接 ps=*** 就返回
            if (!isTrackUrl)
            {
                result.success = false;
                result.link_type = LinkTypeEnum.unknown;
                result.channel_type = ChannelTypeEnum.pdd;
                result.message = "放弃转链";
                result.reason = "数字id链接";
                result.accountId = 0;
                result.accountName = string.Empty;
                result.content = content;
                return result;
            }

            //去追踪
            Stopwatch stopwatch = Stopwatch.StartNew();
            stopwatch.Start();
            var redirectedUrl = await GetRedirectedUrlAsync(url, cancellationToken);
            stopwatch.Stop();


            if (redirectedUrl != url)
            {
                result.rawContent2 = redirectedUrl;
                url = redirectedUrl;
                result.elapsedTime2 = (int)stopwatch.ElapsedMilliseconds;
            }



            Stopwatch stopwatch2 = Stopwatch.StartNew();
            stopwatch2.Start();

            switch (_account.work_mode)
            {
                case PddUnionWorkMode.SiteApi:
                    result = await UnionGeneratelink(result, url, cancellationToken);
                    break;
                //case PddUnionWorkMode.AffApi:
                //    result = await GetPromotionByAffAsync(result, url, cancellationToken);
                //    break;
                case PddUnionWorkMode.Crawler:
                default:
                    result = await transferUrl(result, url, cancellationToken);
                    break;
            }
            stopwatch2.Stop();
            result.elapsedTime3 = (int)stopwatch2.ElapsedMilliseconds;
            return result;
        }
    }
}