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

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

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;
        private static TkPoolDTO _account;
        static Random _random = new Random();

        public static string _url_pattern = @"https?://[-A-Za-z0-9+&@#/%?=~_|!:,.;]+[-A-Za-z0-9+&@#/%=~_|]";

        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();

            _account = account;
            _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 = "放弃转链";
                result.reason = "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)
        {
            result.itemName = "点击打开淘宝APP";
            result = await InternalTextProcessingAsync(content, result, cancellationToken, swData);

            if (result.link_type == LinkTypeEnum.other_aff)
            {
                //============================== 其他推广链接 ==============================
                result.channel = TkChannelEnum.tb;
                result.channel_type = ChannelTypeEnum.tb;
                result.link_type = LinkTypeEnum.other_aff;
                result.success = false;
                result.message = "放弃转链";
                result.reason = "其他推广链接";
                result.accountName = string.Empty;
                result.content = content;
                result.deeplink_url = GetDeeplink(null);
                //result.itemName = GetTitle(content);
                return result;
            }

            //处理过后的正文 用于转链
            if (!result.success)
            {
                switch (result.link_type)
                {
                    case LinkTypeEnum.profile:
                    case LinkTypeEnum.live:
                    case LinkTypeEnum.video:
                    case LinkTypeEnum.note:
                    case LinkTypeEnum.goods:
                        break;
                    default:
                        return result;
                }
            }

            //============================== 放弃转链-地区过滤 ==============================
            bool is_ignore2 = FlowControlIgnoreRequest(out string reason);
            if (is_ignore2)
            {
                result.success = false;
                result.message = "放弃转链";
                result.reason = reason;
                return result;
            }

            result = await alimamaParseAsync(result.content, result, cancellationToken);

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

        public static string ReplaceUrls(string content, string shortLink)
        {
            if (string.IsNullOrEmpty(content) || string.IsNullOrEmpty(shortLink)) 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;
                if (url.Contains("https://s.tb.cn/") ||
                    url.Contains("http://s.tb.cn/") ||
                    url.Contains("http://m.tb.cn/") ||
                    url.Contains("http://m.tb.cn/"))
                {
                    result = result.Replace(url, shortLink);
                }
            }
            return result;
        }



        public static string RemoveUrls(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;
                result = result.Replace(url, string.Empty);

            }
            return result;
        }

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

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


        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)(?:/[a-zA-Z0-9\-\._~:/?#[\]@!$&'()*+,;=%]*)?)";
            Regex regex = new(pattern, RegexOptions.IgnoreCase);
            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()}";
            //string result = $"tbopen://m.taobao.com/tbopen/index.html?action=ali.open.nav&module=h5&bootImage=0&afcPromotionOpen=false&bc_fl_src=h5_huanduan&source=slk_dp&h5Url={url.UrlEncode()}";
            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;
        }




        public 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;


                //DY
                url = DyUnionPlus.GetLink(content);
                urlType = DyUnionPlus.GetLinkType(content, url);
                if (urlType != LinkTypeEnum.unknown) return true;
            }
            catch { }
            return false;
        }

        public static bool MatchRegexes(string text, List<string> extended)
        {
            text = RemoveUrls(text);
            string special_symbols = $"\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲";

            string[] base_patterns =
            [
                $@"(.*?\d{{2}}\s[a-zA-Z]{{2}}\d{{4}}\s.*?[{special_symbols}]+[a-zA-Z0-9\s]+[{special_symbols}]).*",
                $@"[{special_symbols}]+\s*([a-zA-Z]{{1,2}}\d{{2,4}}\s[a-zA-Z0-9]{{11}}|[a-zA-Z0-9]{{11}}[a-zA-Z]{{1,2}}\d{{2,4}}\s|[a-zA-Z0-9]{{11}})\s*[{special_symbols}]+",
                @"🍑♭ɑ◑下单|Tao宝"
            ];

            List<string> mergedArray = extended.Union(base_patterns).ToList();

            /*
            (.*?\\d{2}\\s[a-zA-Z]{2}\\d{4}\\s.*?[\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+[a-zA-Z0-9\\s]+[\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲])).*\n[\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+\\s*([a-zA-Z]{1,2}\\d{2,4}\\s[a-zA-Z0-9]{11}|[a-zA-Z0-9]{11}[a-zA-Z]{1,2}\\d{2,4}\\s|[a-zA-Z0-9]{11})\\s*[\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+\n🍑♭ɑ◑下单|Tao宝
            
             
            【3瓶】恩威万蓝莓叶黄素酯软糖儿童成人上班族如酸涩肿胀可服用，<z5TYWAblvTw >:// HU7709,打開/
置顶小助手，方便下次查询\r\n1★【劵】无\r\n2★【返.佣】 0.29圆 \r\n3★【付款.价】24.9 圆\r\n \r\n 1 ￠HZv5WAboi0D￠(/:/ HU8866\r\
            n长按覆制本内容，打开TaoBao/\r\n-\r\n返佣是指商家的推广佣.金\r\n通过我回复的口令下単收货后，我
            会微信红.苞返给你！！\r\n\r\n有疑问可回复“人工”咨询客服*/

            //🍑🔗上 水和肌精华水& CZ0 VWdCWC2gj97&,
            //6฿04ZDWzrwxjN)/ CA97
            //1😘  腹制这条(CZ00 iHvtWzrXa7O(:/
            //￥EC8DG4Zj0dhd8DOw￥MF6563 oIo0WzOF1Cp##X-PyVpHiPlyj1OP##
            //6.6₵vU2jWzIW3Qx₲/ CZ5224
            //＄F56nWzIDSFZ＄
            //0復製这条口令( CZ00 D0M4WzI0jFJ¢，📱咑開[Tao宝]就可下单:/
            //8fuzhi本条消息：¤ CZ3458 JSR1WzrPCBU¤，打开🍑♭ɑ◑下单/
            //5€Lz6SWzrkLDw¥:// CZ13
            //3₲Igq3Wzkc5np₳:// CZ15
            //9.9🎈【康师傅易拉罐6罐】口伶：6₰TnxKWzr0wff₪:// CZ69
            //4₲99Y0WzJdK6S₳:// CZ46
            //4fuzhi本条消息：¤ CZ0002 WVCSWzrk6Ix¤，打开🍑♭ɑ◑下单/
            bool anyMatch = false;
            foreach (string pattern in mergedArray)
            {
                if (Regex.IsMatch(text, pattern, RegexOptions.IgnoreCase))
                {
                    anyMatch = true;
                }
            }
            return anyMatch;
        }
        public static bool MatchOtherInfo(string text, List<string> extended)
        {
            string[] base_patterns =
            [
                "[ďȌƯƔɪƝɖƟɏƖԂȎȗϓɫɧơȲǏđǫƲƳɨǹƊȮǚƴɬɲƿȖȳɳɗʮϔӢǙɦɒɎƥʘưΊƞǪȈʠΌǓϒίȠƠƱƗǸþŌŬÿĩŅÞÓũŸÎŎûįŇĎÖÙŷÏňÜÝÍñĎÒüŶľńĎ0ųŷÏŇdǒǔyí℡抖yí℡ÞŐûŷÌñ]{6}",
                @"(?:.+)极速版(?:口令|搜索)",
                @"##[a-zA-Z0-9]+##\[抖音(?:极速版)?(?:口令|搜索)\]",
                @"岽|亰|菄|京东|婛",
                "打开【剪映】",
                "weishi://|baiduhaokan://",
                "复制打开「UC浏览器」|UC瀏覽器",
                "番茄免费小说",
                @"\?chanTag=[^&\s]*口令",
                @"(https?://pan\.baidu\.com/s/[A-Za-z0-9?\-_=]+)",
                @"https?:\/\/meeting\.tencent\.com[^\s]*",
                @"#腾讯会议：(\d{3}-\d{3}-\d{3})"
            ];

            List<string> mergedArray = extended.Union(base_patterns).ToList();

            bool anyMatch = false;
            foreach (string pattern in mergedArray)
            {
                if (Regex.IsMatch(text, pattern, RegexOptions.IgnoreCase))
                {
                    anyMatch = true;
                }
            }
            return anyMatch;
        }

        public static bool MatchMultiToken(string text, List<string> extended)
        {
            text = RemoveUrls(text);
            string special_symbols = $"\\$\\(\\)\\/\\#\\&\\<\\>฿￠￥¥＄🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲";
            string[] base_patterns =
            [
                @"[！$￥%]+([a-zA-Z0-9]{16})[！$￥%]+",
                @"\^v(\d+#[a-zA-Z0-9]+)(?:\?.*)?\^",
                "##([a-zA-Z0-9\\-]+)##",
                $@"(.*?\d{{2}}\s[a-zA-Z]{{2}}\d{{4}}\s.*?[{special_symbols}]+[a-zA-Z0-9\s]+[{special_symbols}]).*",
                $@"[{special_symbols}]+\s*([a-zA-Z]{{1,2}}\d{{2,4}}\s[a-zA-Z0-9]{{11}}|[a-zA-Z0-9]{{11}}[a-zA-Z]{{1,2}}\d{{2,4}}\s|[a-zA-Z0-9]{{11}})\s*[{special_symbols}]+",
            ];

            // PAGE6:/^v130091#P2J6EDxN8p8?chanTag=d9dc522bdea27437514052ea50f6a787^  


            List<string> mergedArray = extended.Union(base_patterns).ToList();
            int matchCount = 0;

            foreach (string pattern in mergedArray)
            {
                Match match = Regex.Match(text, pattern, RegexOptions.IgnoreCase);
                if (match.Success)
                {
                    matchCount++;
                    text = text.Replace(match.Groups[^1].Value, ""); // 移除匹配的部分文本
                }
            }
            return matchCount >= 2;
        }
        public static bool FilterRiskLink(string ip, string oaid, string content, out string reason)
        {
            reason = string.Empty;
            string itemId = ExtractItemId(content);
            if (string.IsNullOrEmpty(itemId)) return false;

            string cacheKey = $":cache:parse:{ip}_{oaid}_{itemId}";
            int flag = RedisHelper.Get<int>(cacheKey);
            if (flag > 0)
            {
                reason = $"口令用户";
                return true;
            }
            return false;
        }

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

        public static bool FlowControlIgnoreRequest(out string reason)
        {
            reason = string.Empty;
            try
            {
                // IP和流量控制
                var config = TkConfigCore.Get();
                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 true;
                }
            }
            catch (Exception ex)
            {
                reason = $"配置异常";
            }
            return false;
        }

        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;
        }

    }
}