dotnet/molilian.core/Plus/ks/KsUnionPlus.cs

using COSXML.Network;
using dodohold.core;
using Spire.Pdf.Graphics;
using System.Diagnostics;
using System.Net;
using System.Text.RegularExpressions;
using TencentCloud.Apm.V20210622.Models;
using static dodohold.core.ZTOExpress.CreateOrderArgs;

namespace molilian.core
{
    public partial class KsUnionPlus
    {
        public static string _url_pattern = @"https?://(?:[\w-]+\.)*(?:chenzhongtech\.com|kuaishou\.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, LinkTypeEnum type)
        {
            string shareObjectId = string.Empty;
            string url_pattern = @"shareObjectId=([^\s&]+)";
            var match = Regex.Match(url, url_pattern);
            if (match.Success) shareObjectId = match.Groups[1].Value;

            string result = type switch
            {
                LinkTypeEnum.video => $"kwai://work/{shareObjectId}",
                LinkTypeEnum.profile => $"kwai://profile/{shareObjectId}",
                LinkTypeEnum.live => $"kwai://live/play/{shareObjectId}",
                LinkTypeEnum.baseDomain or LinkTypeEnum.other_aff => "kwai://home",
                LinkTypeEnum.goods or LinkTypeEnum.note => "kwai://home",
                _ => string.Empty,
            };
            return result;
        }

        public static KsDataDTO GetFormattedObject(string content, string ip = "", string oaid = "")
        {
            string shortLinkurl = GetLink(content);
            LinkTypeEnum link_type = string.IsNullOrEmpty(shortLinkurl) ? LinkTypeEnum.unknown : LinkTypeEnum.baseDomain;

            string deeplink_url = GetDeeplink(shortLinkurl, link_type);
            return new KsDataDTO()
            {
                message = string.Empty,
                link_type = link_type,
                channel = TkChannelEnum.ks,
                accountName = string.Empty,
                success = false,
                content = content,
                ip = ip,
                oaid = oaid,
                elapsedTime = 0,
                itemName = "点击打开快手APP",
                shortLinkurl = shortLinkurl,
                deeplink_url = deeplink_url,
                create_time = DateTime.Now,
                end_point = _end_point,
            };
        }


        private LinkTypeEnum GetLinkType(string url)
        {
            if (string.IsNullOrEmpty(url)) return LinkTypeEnum.unknown;

            string url_pattern = @"https?://(?:[\w-]+\.)*(?:chenzhongtech\.com|kuaishou\.com)/fw/(?:photo|long-video)/[^\s]*\bshareObjectId=[^\s&]+";
            var match = Regex.Match(url, url_pattern);
            if (match.Success) return LinkTypeEnum.video;

            url_pattern = @"https?://(?:[\w-]+\.)*(?:chenzhongtech\.com|kuaishou\.com)/fw/user/[^\s]*\bshareObjectId=[^\s&]+";
            match = Regex.Match(url, url_pattern);
            if (match.Success) return LinkTypeEnum.profile;

            url_pattern = @"https?://(?:[\w-]+\.)*(?:chenzhongtech\.com|kuaishou\.com)/fw/live/[^\s]*\bshareObjectId=[^\s&]+";
            match = Regex.Match(url, url_pattern);
            if (match.Success) return LinkTypeEnum.live;

            if (url.Contains("chenzhongtech.com") || url.Contains("kuaishou.com")) return LinkTypeEnum.baseDomain;

            return LinkTypeEnum.unknown;
        }




        private bool IsAffCode(string content)
        {
            string url_pattern = @"復制整段消息，打開快手|复制整段消息，打开快手";
            var match = Regex.Match(content, url_pattern);
            if (match.Success) return true;
            return false;
        }

        private async Task<(bool, string)> GetDesiredUrlAsync(string url)
        {
            string result = url;
            try
            {
                string pattern = @"https:\/\/v\.kuaishou\.com\/[A-Za-z0-9]+";
                Match match = Regex.Match(url, pattern);

                //不是短网址，不用展开
                if (!match.Success) return (true, url);

                WebClientUtility client = new()
                {
                    Proxy = _proxy ?? ProxyNodesCore.RandomOne(),
                    UserAgent = Sayaka.Common.ProviderFakeUserAgent.RandomComputer,
                    AllowAutoRedirect = false
                };
#if DEBUG
                client.Proxy = null;
#endif
                client.Timeout = GetRequestTimeout(0);
                var response = await client.RequestAsync(url);

                if (response.ResponseMessage != null)
                {
                    result = response.ResponseMessage.Headers.Location.ToString();
                    result = result.UrlDecode();
                    return (true, result);
                }
            }
            catch (Exception ex)
            {
                result = $"{ex.Message}\n{ex.StackTrace}";
            }
            return (false, result);
        }

        public TimeSpan GetRequestTimeout(int deduction = 0)
        {
#if DEBUG
            return TimeSpan.FromMilliseconds(10 * 1000);
#endif
            int min = (int)(_config.rt_max * 0.3);
            if (_config.rt_max == 0) return TimeSpan.FromMilliseconds(1000);

            int timeout = _config.rt_max;
            if (deduction == 0)
            {
                timeout -= min;
                return TimeSpan.FromMilliseconds(timeout);
            }

            timeout -= deduction;
            if (timeout <= min) return TimeSpan.FromMilliseconds(1);
            return TimeSpan.FromMilliseconds(min);
        }

        public async Task<KsDataDTO> KsParseAsync(string content, KsDataDTO result, CancellationToken cancellationToken = default)
        {
            //正文里面不包含url
            if (string.IsNullOrEmpty(result.shortLinkurl) && IsAffCode(content))
            {
                result.success = false;
                result.link_type = LinkTypeEnum.other_aff;
                result.message = "放弃转链";
                result.reason = "其他推广链接";
                result.accountId = 0;
                result.accountName = string.Empty;
                result.content = content;
                result.deeplink_url = GetDeeplink(content, LinkTypeEnum.goods);
                return result;
            }

            string url = string.Empty;
            if (!string.IsNullOrEmpty(result.shortLinkurl))
            {

                //从http获取跳转的url
                Stopwatch stopwatch = Stopwatch.StartNew();
                stopwatch.Start();
                (bool success, url) = await GetDesiredUrlAsync(result.shortLinkurl);
                stopwatch.Stop();
                // 获取执行时间
                result.elapsedTime2 = (int)stopwatch.ElapsedMilliseconds;


                result.link_type = GetLinkType(url);

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

                //没有链接、 不是链接
                if (!success)
                {
                    result.success = false;
                    result.link_type = LinkTypeEnum.unknown;
                    result.message = "放弃转链";
                    result.reason = url;
                    result.accountId = 0;
                    result.accountName = string.Empty;
                    result.content = content;
                    return result;
                }
                result.deeplink_url = GetDeeplink(url, result.link_type);
                return result;
            }


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



            switch (_account.work_mode)
            {
                case KsUnionWorkMode.SiteApi:
                    result = await UnionGeneratelink(result, content, cancellationToken);
                    break;
                case KsUnionWorkMode.Crawler:
                default:
                    //result = await transferUrl(result, url, cancellationToken);
                    break;
            }
            return result;
        }
    }
}