dotnet/molilian.core/Plus/Alimama/parse.cs

using System.Text;
using dodohold.core;
using System.Text.RegularExpressions;
using System.Net;
using System.Security.Cryptography;
using TencentCloud.Ecm.V20190719.Models;
using System.Diagnostics;

namespace molilian.core
{
    public partial class AlimamaPlus
    {

        private string _accountName;
        private int _accountId;
        private string _company;
        private string _tb_token;
        private string _floorId;
        private string _refpid;
        private string _user_agent;
        private string _cookies;
        private WebProxy? _proxy = null;

        private string _api;
        private int _api_id = 0;
        public TkConfigDTO _config;
        private static string _end_point;
        static Random _random = new Random();

        static AlimamaPlus()
        {
            _end_point = Environment.GetEnvironmentVariable("EndPoint");
        }

        public AlimamaPlus(TkPoolDTO account)
        {
            //    string accountName, string tb_token, string floorId, string refpid,
            //    string cookies, string user_agent, string nodeName
            _config = TkConfigCore.Get();


            _accountId = account.id;
            _accountName = account.name;
            _company = account.company;
            _tb_token = account.tb_token;
            _floorId = account.floorId;
            _refpid = account.refpid;
            _cookies = account.cookies;
            _user_agent = account.user_agent;
            if (!string.IsNullOrEmpty(account.nodeName))
            {
                _proxy = ProxyNodesCore.GetOne(account.nodeName);
            }
            _api = account.api;
            _api_id = account.api_id;
        }

        public bool ShouldIgnoreOtherAff(string content)
        {
            TkDataDTO result = new();

            //(_, _, var link_type, _) = InternalTextProcessing(content, ref result);
            InternalTextProcessing(content, ref result);
            return result.link_type == LinkTypeEnum.other_aff;
        }

        public TkDataDTO UnionParse(string content, ref TkDataDTO result)
        {
            InternalTextProcessing(content, ref result);
            //(result.success, result.content, result.link_type, result.shortLinkurl) = InternalTextProcessing(content,ref result);
            if (result.link_type == LinkTypeEnum.other_aff)
            {
                result.channel = TkChannelEnum.tb;
                result.channel_type = ChannelTypeEnum.tb;
                result.link_type = LinkTypeEnum.unknown;
                result.success = false;
                result.message = "其他推广链接";
                result.accountName = string.Empty;
                result.content = content;
                result.deeplink_url = GetDeeplink(null);
                result.itemName = GetTitle(content);
            }
            else
            {
                //处理过后的正文 用于转链
                if (result.success) content = result.content;
                result = alimamaParse(content, result);
            }

            result.content = content;
            result.link_type = result.link_type;
            return result;
        }

        public async Task<TkDataDTO> UnionParseAsync(string content, TkDataDTO result,
            CancellationToken cancellationToken = default,
            Dictionary<string, long> swData = null)
        {
            Stopwatch stopwatch = Stopwatch.StartNew();
            result = await InternalTextProcessingAsync(content, result, cancellationToken, swData);
            if (swData != null)
            {
                stopwatch.Stop();
                swData.Add("InternalTextProcessingAsync", stopwatch.ElapsedMilliseconds);
            }


            //(result.success, result.content, result.link_type, result.shortLinkurl) = InternalTextProcessing(content);
            if (result.link_type == LinkTypeEnum.other_aff)
            {
                result.channel = TkChannelEnum.tb;
                result.channel_type = ChannelTypeEnum.tb;
                result.link_type = LinkTypeEnum.unknown;
                result.success = false;
                result.message = "其他推广链接";
                result.accountName = string.Empty;
                result.content = content;
                result.deeplink_url = GetDeeplink(null);
                //result.itemName = GetTitle(content);
                result.itemName = "点击打开淘宝APP";
            }
            else
            {
                //处理过后的正文 用于转链
                if (!result.success)
                {
                    result.content = content;
                    result.itemName = "点击打开淘宝APP";
                    return result;
                }

                if (swData != null)
                {
                    stopwatch.Reset();
                    stopwatch.Start();
                }
                result = await alimamaParseAsync(result.content, result, cancellationToken);

                if (swData != null)
                {
                    stopwatch.Stop();
                    swData.Add("alimamaParseAsync", stopwatch.ElapsedMilliseconds);
                }

                if (!result.success)
                {
                    result.itemName = "点击打开淘宝APP";
                }
                //if (string.IsNullOrEmpty(result.itemName)) result.itemName = GetTitle(content);
            }
            if (string.IsNullOrEmpty(result.itemName) || result.link_type == LinkTypeEnum.profile)
            {
                result.itemName = "点击打开淘宝APP";
            }
            result.link_type = result.link_type;
            return result;
        }

        public static string ReplaceUrls(string content, string shortLink)
        {
            if (string.IsNullOrEmpty(content) || string.IsNullOrEmpty(shortLink)) return content;
            string result = content;
            string pattern = @"https?:\/\/?([\da-z\.-]+)\.([a-z]{2,6})(:\d{1,5})?([\/\w\.-]*)*\/?(#[\S]+)?(\?[\S]+)?";

            Regex regex = new(pattern);
            MatchCollection matches = regex.Matches(content);

            foreach (Match m in matches.Cast<Match>())
            {
                string url = m.Value;
                if (url.Contains("https://m.tb.cn/") ||
                    url.Contains("http://m.tb.cn/"))
                {
                    result = result.Replace(url, shortLink);
                }
            }
            return result;
        }

        public static string GetLink(string content)
        {
            if (string.IsNullOrEmpty(content)) return content;
            string result = content;
            string pattern = @"https?:\/\/?([\da-z\.-]+)\.([a-z]{2,6})(:\d{1,5})?([\/\w\.-]*)*\/?(#[\S]+)?(\?[\S]+)?";
            //string pattern = @"https?:\/\/[\da-z\.-]+\.[a-z]{2,6}(:\d{1,5})?(\/[\w\.-]*)*";
            Regex regex = new(pattern);
            MatchCollection matches = regex.Matches(content);

            if (matches.Count > 0)
            {
                return matches[^1].Value; // 返回第一个匹配到的 URL
            }
            return string.Empty;
        }


        public static string GetTaobaoLink(string content)
        {
            if (string.IsNullOrEmpty(content)) return content;
            string result = content;
            string pattern = @"(https?://(?:[\w-]+\.)*(?:tb\.cn|taobao\.com|juhuasuan.com|tmall\.com|tmall\.hk)(?:/[^\s]*)?)";
            Regex regex = new(pattern);
            MatchCollection matches = regex.Matches(content);

            if (matches.Count > 0)
            {
                return matches[^1].Value; // 返回第一个匹配到的 URL
            }
            return string.Empty;
        }

        public static string GetDeeplink(string url)
        {
            if (string.IsNullOrEmpty(url)) return "tbopen://m.taobao.com/tbopen/index.html";
            string result = $"tbopen://m.taobao.com/tbopen/index.html?action=ali.open.nav&module=h5&h5Url={url.UrlEncode()}&backURL=__back_url__";
            return result;
        }


        public static string GetTitle(string content)
        {
            if (string.IsNullOrEmpty(content)) return string.Empty;
            string url = GetLink(content);
            if (!string.IsNullOrEmpty(url)) content = content.Replace(url, string.Empty);

            content = content.Replace("点击链接直接打开 或者 淘宝搜索直接打开", string.Empty).Trim();
            string[] words = Regex.Split(content, @"\s+");
            string word = words[^1];
            word = word.TrimStart('「').TrimEnd('」');
            return word;
        }




        static bool IgnoreRegionIncluded(string _ignoreRegions, string? ipInfo, out string regionInfo)
        {
            regionInfo = string.Empty;
            if (string.IsNullOrEmpty(_ignoreRegions) || string.IsNullOrEmpty(ipInfo)) return false;

            string[] parts = ipInfo.Split('|');
            string countryInfo = null;
            if (parts.Length >= 4)
            {
                countryInfo = parts[0];
                regionInfo = parts[3];
            }
            if (string.IsNullOrEmpty(regionInfo)) return false;
            if (string.IsNullOrEmpty(countryInfo)) return false;

            string[] ignoreRegions = _ignoreRegions.Split('|');

            if (ignoreRegions.Contains(regionInfo)) return true;
            if (ignoreRegions.Contains(countryInfo))
            {
                regionInfo = countryInfo;
                return true;
            }
            foreach (var region in ignoreRegions)
            {
                if ("海外".Equals(region) && !"中国".Equals(countryInfo) && !"0".Equals(countryInfo))
                {
                    regionInfo = "海外";
                    return true;
                }
            }
            return false;
        }
        public static async Task<bool> OtherPlatformAsync(string content)
        {
            try
            {
                //腾讯会议
                string wemeetId = await ToolParsePlus.GetWemeetIdAsync(content);
                if (!string.IsNullOrEmpty(wemeetId)) return true;

                //百度网盘
                string surl = string.Empty, pwd = string.Empty;
                _ = ToolParsePlus.GetPanLink(content, ref surl, ref pwd);
                if (!string.IsNullOrEmpty(surl)) return true;

                //JD
                string url = JdUnionPlus.GetLink(content);
                var urlType = JdUnionPlus.GetLinkType(url, out _, out _);
                if (urlType != LinkTypeEnum.unknown) return true;
            }
            catch { }
            return false;
        }

        public static bool MatchRegexes(string text)
        {
            string regex1 = @"(.*?\d{2}\s[a-zA-Z]{2}\d{4}\s.*?[$￥🗝₴€₤£₳✔《💰🔑🎵✔💲🔐📲]+[a-zA-Z0-9\s]+[$￥🗝₴€₤£₳✔《💰🔑🎵✔💲🔐📲]).* [$￥🗝₴€₤£₳✔《💰🔑🎵✔💲🔐📲]+[a-zA-Z0-9\s]+[$￥🗝₴€₤£₳✔《💰🔑🎵✔💲🔐📲](\s*(https?://[^\s]+)?)?\s*([a-zA-Z]{2}\d{4})?\s*(.*)";
            Regex r1 = new(regex1);
            bool match1 = r1.IsMatch(text);
            return match1;
        }
        public static bool FirstFilterIgnoreRequest(string content, out string reason)
        {
            reason = string.Empty;
            try
            {
                // 正则过滤
                var matchResult = MatchRegexes(content);
                if (matchResult) return false;
                reason = "初步筛选";
                return true;
            }
            catch (Exception ex)
            {
                reason = "配置异常";
                return true;
            }

        }

        public static bool ShouldIgnoreRequest(string ip, string oaid, out string reason)
        {
            reason = string.Empty;
            try
            {
                // IP和流量控制
                var config = TkConfigCore.Get();
                string ignorePercentageCity = config.ignorePercentageCity;
                string? ipInfo = IP2RegionPlus.Search(ip);
                if (IgnoreRegionIncluded(ignorePercentageCity, ipInfo, out string regionInfo))
                {
                    reason = $"地区控制：{regionInfo}";
                    return true;
                }

                int ignorePercentage = config.ignorePercentage;
                if (ignorePercentage == 0) return false;
                var randomValue = (decimal)_random.Next(0, 100);
                bool result = randomValue <= ignorePercentage; // 如果生成的随机数小于忽略百分比，则返回 true，表示需要忽略请求
                if (result) reason = "流量控制";
                return result;

            }
            catch (Exception ex)
            {
                reason = $"配置异常";
                return true;
            }
        }

        public static TkDataDTO GetFormattedObject(string content, string ip, string oaid)
        {
            string shortLinkurl = GetTaobaoLink(content);
            string deeplink_url = GetDeeplink(shortLinkurl);
            return new TkDataDTO()
            {
                message = string.Empty,
                channel = TkChannelEnum.tb,
                accountName = string.Empty,
                success = false,
                content = content,
                link_type = LinkTypeEnum.unknown,
                ip = ip,
                oaid = oaid,
                couponAmount = 0,
                taoToken = string.Empty,
                elapsedTime = 0,
                itemName = "点击打开淘宝APP",
                shortLinkurl = shortLinkurl,
                deeplink_url = deeplink_url,
                create_time = DateTime.Now,
                end_point = _end_point,
            };
        }


        public static string OppoDecode(string content)
        {
            try
            {
                string secretKey = "BwFeIvOEDZy9MFGi0JLZBO4yEhR44DGV";
                byte[] keyBytes = Encoding.UTF8.GetBytes(secretKey);


                if (!content.Contains("%IV%")) return content;
                string cipherText = content.GetContentPart("", "%IV%");
                string ivText = content.GetContentPart("%IV%", "");
                byte[] cipherBytes = Convert.FromBase64String(cipherText);
                byte[] ivBytes = Convert.FromBase64String(ivText);

                using RijndaelManaged aes = new();
                aes.Key = keyBytes;
                aes.IV = ivBytes;
                aes.Mode = CipherMode.CFB;
                aes.Padding = PaddingMode.PKCS7;

                ICryptoTransform decryptor = aes.CreateDecryptor(aes.Key, aes.IV);

                var outBytes = decryptor.TransformFinalBlock(cipherBytes, 0, cipherBytes.Length);
                content = outBytes.ConvertToString();

            }
            catch (Exception e) { }
            return content;
        }

    }
}