dotnet/molilian.core/Plus/Alimama/parse_2.cs

using dodohold.core;
using System.Text.RegularExpressions;
using System.Diagnostics;
using System.Text.Json;
using TencentCloud.Fmu.V20191213.Models;

namespace molilian.core
{
    public partial class AlimamaPlus
    {
        public void UnionParse2(string content, ref TkDataDTO result)
        {
            if ("veapi".Equals(_api) && _api_id > 0)
            {
                var account = VeapiPoolCore.Get(_api_id);
                if (account != null)
                {
                    var core = new VeapiPlus(account.name, account.key, account.sessionKey);
                    core.UnionParse(content, ref result);
                    return;
                }
            }
            UnionParse(content, ref result);
        }

        public async Task<TkDataDTO> UnionParse2Async(string content, TkDataDTO result)
        {
            if ("veapi".Equals(_api) && _api_id > 0)
            {
                var account = VeapiPoolCore.Get(_api_id);
                if (account != null)
                {
                    var core = new VeapiPlus(account.name, account.key, account.sessionKey);
                    result = await core.UnionParseAsync(content, result);
                    return result;
                }
            }
            result = await UnionParseAsync(content, result);
            return result;
        }

        /// <summary>
        /// 提取url里面的id参数
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        static string? ExtractItemId(string url)
        {
            Match match = Regex.Match(url, @"(?:\?|&)id=(\d+)");
            if (match.Success)
            {
                return match.Groups[1].Value;
            }
            return null;
        }

        /// <summary>
        /// 是否淘宝链接
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        private static (bool, string) IsTaobaoUrl(string url)
        {
            if (url.Contains("//a.m.taobao.com/i"))
            {
                //第一种提取方法
                string itemId = url.GetContentPart("//a.m.taobao.com/i", ".htm");
                if (!string.IsNullOrEmpty(itemId))
                {
                    string resultUrl = $"https://item.taobao.com/item.htm?id={itemId}";
                    return (true, resultUrl);
                }
            }

            if (url.Contains("//item.taobao.com/item.htm?") ||
                url.Contains("//h5.m.taobao.com/awp/core/detail.htm?") ||
                url.Contains("//new.m.taobao.com/detail.htm?") ||
                url.Contains("//outfliggys.m.taobao.com/app/trip/rx-travel-detail") ||
                url.Contains("//detail.m.tmall.com/item.htm?") ||
                url.Contains("//tmallx.tmall.com/app/tmallx-src/goods-detail") ||
                url.Contains("//detail.tmall.com/item.htm?") ||
                url.Contains("//detail.tmall.hk/hk/item.htm?"))
            {
                string? itemId = ExtractItemId(url);
                if (!string.IsNullOrEmpty(itemId))
                {
                    string resultUrl = $"https://item.taobao.com/item.htm?id={itemId}";
                    return (true, resultUrl);
                }
            }
            return (false, url);
        }

        /// <summary>
        /// 排除已知的淘客链接
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        static bool IsAffLink(string url)
        {
            if (!url.StartsWith("https://") || !url.StartsWith("https://")) url = "https://" + url;
            Uri uri = new(url);
            if (uri.Host.Equals("s.click.taobao.com", StringComparison.OrdinalIgnoreCase) ||
                uri.Host.Equals("uland.taobao.com", StringComparison.OrdinalIgnoreCase))
            {
                return true;
            }

            if (uri.Host.Equals("m.tb.cn", StringComparison.OrdinalIgnoreCase) &&
                uri.AbsolutePath.StartsWith("/h.") &&
                string.IsNullOrEmpty(uri.Query) && // No query parameters
                !url.Contains("?tk=")) // No "tk" parameter
            {
                return true;
            }
            return false;
        }

        /// <summary>
        /// 判断淘客的￥羊角符号
        /// </summary>
        /// <param name="content"></param>
        /// <returns></returns>
        static bool ContainsSpecialFormat(string content)
        {
            // 使用正则表达式匹配内容中是否含有特定格式的字符串
            //string pattern = @"￥[a-zA-Z0-9\s]+￥";
            string pattern = @"[$￥🗝₴€₤£₳✔《💰🔑🎵✔️💲🔐📲]+[a-zA-Z0-9]+[$￥🗝₴€₤£₳✔《💰🔑🎵✔️💲🔐📲]*";
            return Regex.IsMatch(content, pattern);
        }

        /// <summary>
        /// 从http获取跳转的url
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        (bool, string) GetDesiredUrl(string url)
        {
            string result;
            try
            {
                string desiredUrlPattern = "var url = '(.*?)'";
                WebClientUtility client = new()
                {
                    Proxy = _proxy,
                    UserAgent = Sayaka.Common.ProviderFakeUserAgent.RandomComputer
                };
#if DEBUG
                client.Proxy = null;
#else
                client.Timeout = TimeSpan.FromMilliseconds(3000);
#endif
                var response = client.Request(url);

                var responseBody = response.Body();
                result = responseBody;
                if (response.Successed)
                {

                    if (response.ResponseMessage.IsSuccessStatusCode)
                    {
                        Match match = Regex.Match(responseBody, desiredUrlPattern);
                        if (match.Success)
                        {
                            result = match.Groups[1].Value; // 返回匹配的 URL

                            if (string.IsNullOrEmpty(result)) return (false, "没有匹配的URL");
                            return (true, result);
                        }
                    }
                    else
                    {
                        result = $"{response.ResponseMessage.StatusCode}\n{responseBody}";
                    }
                }
                else
                {
                    throw response.ResponseException;

                }
            }
            catch (Exception ex)
            {
                result = $"{ex.Message}\n{ex.StackTrace}";
            }
            return (false, result);
        }


        private (bool, string, LinkTypeEnum, string) InternalTextProcessing(string content)
        {
            bool success;
            string result;
            LinkTypeEnum link_type = LinkTypeEnum.unknown;

            string url = GetLink(content);
            if (!string.IsNullOrEmpty(url))
            {
                //排除已知的淘客链接
                if (IsAffLink(url)) return (false, "排除淘客链接", LinkTypeEnum.other_aff, url);

                //直接提取
                (success, result) = IsTaobaoUrl(url);
                if (success) return (true, result, LinkTypeEnum.goods, result);


                //从http获取跳转的url
                (success, result) = GetDesiredUrl(url);
                if (success)
                {
                    //排除已知的淘客链接
                    if (IsAffLink(result)) return (false, "排除淘客链接", LinkTypeEnum.other_aff, result);

                    (success, result) = IsTaobaoUrl(result);
                    if (success) return (true, result, LinkTypeEnum.goods, result);
                }

                link_type = GetLinkType(url);
                return (false, result, link_type, url);
            }

            //判断淘客的￥羊角符号
            if (ContainsSpecialFormat(content)) return (false, "排除淘口令", LinkTypeEnum.other_aff, null);

            return (false, "纯口令 没链接", link_type, null);
        }

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

            if (url.StartsWith("https://huodong.m.taobao.com/act/talent/live.html")) return LinkTypeEnum.live;
            if (url.StartsWith("https://web.m.taobao.com/app/tnode/web/index")) return LinkTypeEnum.video;

            string pattern = @"^https://shop\d+\.m\.taobao\.com";
            if (Regex.IsMatch(url, pattern))
            {
                return LinkTypeEnum.profile;
            }
            return LinkTypeEnum.unknown;
        }

        private void alimamaParse(string content, ref TkDataDTO result)
        {
            string t = $"{DateTime.Now.Convert2UnixTimestamp(true)}";

            Random random = new();
            double randomNumber = random.NextDouble();
            string randomString = randomNumber.ToString()[2..];

            string cna = _cookies.GetContentPart("cna=", ";");
            string firstFiveCharacters = cna[..Math.Min(5, cna.Length)];

            var variableMap = new
            {
                url = content,
                union_lens = $"b_pvid:a219t._portal_v2_tool_links_page_home_index_htm_{t}_{randomString}_{firstFiveCharacters}",
                lensScene = "PUB",
                spmB = "_portal_v2_tool_links_page_home_index_htm"
            }.Convert2Json(true).UrlEncode();

            variableMap = variableMap.Replace(" ", "%20");

            string url = "https://pub.alimama.com/openapi/param2/1/gateway.unionpub/xt.entry.json?" +
                $"t={t}&_tb_token_={_tb_token}&floorId={_floorId}&refpid={_refpid}&variableMap={variableMap}";


            var client = new WebClientUtility().SetContentType("application/json;charset=utf-8")
                        .AddHeaders("X-Requested-With", "XMLHttpRequest")
                        .AddHeaders("Cookie", _cookies);
            client.Proxy = _proxy;
#if DEBUG
            client.Proxy = null;
#endif
            if (!string.IsNullOrEmpty(_user_agent)) client.UserAgent = _user_agent;

            client.Timeout = TimeSpan.FromMilliseconds(1000);
            var response = client.Request(url);

            var body = string.Empty;
            try
            {
                if (!response.Successed)
                {
                    result.channel = TkChannelEnum.tb;
                    result.accountName = _accountName;
                    result.link_type = LinkTypeEnum.unknown;
                    result.success = false;
                    result.message = "fail";

                    if (response.ResponseException != null)
                    {
                        new LoggerLibrary("api_error", "fail")
                            .Info(response.ResponseException.Message, response.ResponseException.StackTrace)
                            .Save();
                        throw response.ResponseException;
                    }

                    return;
                }


                body = response.Body();
                JsonElement root = body.Convert2JsonElement();

                bool success = root.Read<bool>("success", false);
                string message = root.Read("message", string.Empty);
                string info_message = root.PathRead("info.message", string.Empty);
                if (!string.IsNullOrEmpty(info_message)) message = info_message;

                string taoToken = root.PathRead("data.taoToken", string.Empty);
                string shortLinkurl = root.PathRead("data.shortLinkurl", string.Empty);

                string couponLinkTaoToken = root.PathRead("data.couponLinkTaoToken", string.Empty);
                string couponShortLinkUrl = root.PathRead("data.couponShortLinkUrl", string.Empty);
                decimal couponAmount = root.PathRead<decimal>("data.couponAmount", 0);

                string itemId = root.PathRead("data.itemId", string.Empty);
                string itemName = root.PathRead("data.itemName", string.Empty);
                decimal promotionPrice = root.PathRead<decimal>("data.promotionPrice", 0);
                string sellerNickName = root.PathRead("data.sellerNickName", string.Empty);
                string shopTitle = root.PathRead("data.shopTitle", string.Empty);
                string pic = root.PathRead("data.pic", string.Empty);
                string qrCodeUrl = root.PathRead("data.qrCodeUrl", string.Empty);


                if (response.ResponseMessage != null)
                {

                    switch (response.ResponseMessage.StatusCode)
                    {
                        case System.Net.HttpStatusCode.OK:
                            {
                                if (couponAmount > 0 && !string.IsNullOrEmpty(couponLinkTaoToken) && !string.IsNullOrEmpty(couponShortLinkUrl))
                                {
                                    taoToken = couponLinkTaoToken;
                                    shortLinkurl = couponShortLinkUrl;
                                }
                                string shortLink = GetLink(taoToken);
                                content = ReplaceUrls(content, shortLink);
                            }
                            break;
                        case System.Net.HttpStatusCode.Found:
                            string location = response.ResponseMessage.Headers.Location.OriginalString;
                            if (!string.IsNullOrEmpty(location) && location.Contains("www.alimama.com/member/login.htm"))
                            {
                                success = false;
                                message = "nologin";
                            }
                            break;
                        default:
                            {
                                success = false;
                                message = "other";
                            }
                            break;
                    }
                }
                if (!success)
                {
                    shortLinkurl = GetLink(content);
                    switch (message)
                    {
                        case "该链接不支持转化，请更换链接尝试":
                        case "网络错误":
                            break;
                        default:
                            var _headers = response.ResponseMessage.Headers;
                            var headers = "";
                            foreach (var h in _headers)
                            {
                                foreach (var Value in h.Value)
                                {
                                    headers += $"{h.Key}: {Value}\n";
                                }
                            }
                            new LoggerLibrary("api_error", "fail")
                                .Info(message, content)
                                .Info(headers, body)
                                .Save();
                            break;
                    }
                }
                string deeplink_url = GetDeeplink(shortLinkurl);

                result.channel = TkChannelEnum.tb;
                result.accountName = _accountName;
                result.link_type = success ? LinkTypeEnum.goods : result.link_type;
                result.success = success;
                result.message = message;
                result.content = content;
                result.couponAmount = couponAmount;
                result.taoToken = taoToken;
                result.shortLinkurl = shortLinkurl;
                result.deeplink_url = deeplink_url;

                result.itemId = itemId;
                result.itemName = itemName;
                result.promotionPrice = promotionPrice;
                result.sellerNickName = sellerNickName;
                result.shopTitle = shopTitle;
                result.pic = pic;
                result.qrCodeUrl = qrCodeUrl;
            }
            catch
            {
                if (response.ResponseMessage?.StatusCode == System.Net.HttpStatusCode.OK)
                {
                    if (body.Contains("<p>抱歉！页面无法访问……</p>"))
                    {
                        throw new APIException("抱歉！页面无法访问……");
                    }
                }
                throw;
            }
        }



        private async Task<TkDataDTO> alimamaParseAsync(string content, TkDataDTO result)
        {
            string t = $"{DateTime.Now.Convert2UnixTimestamp(true)}";

            Random random = new();
            double randomNumber = random.NextDouble();
            string randomString = randomNumber.ToString()[2..];

            string cna = _cookies.GetContentPart("cna=", ";");
            string firstFiveCharacters = cna[..Math.Min(5, cna.Length)];

            var variableMap = new
            {
                url = content,
                union_lens = $"b_pvid:a219t._portal_v2_tool_links_page_home_index_htm_{t}_{randomString}_{firstFiveCharacters}",
                lensScene = "PUB",
                spmB = "_portal_v2_tool_links_page_home_index_htm"
            }.Convert2Json(true).UrlEncode();

            variableMap = variableMap.Replace(" ", "%20");

            string url = "https://pub.alimama.com/openapi/param2/1/gateway.unionpub/xt.entry.json?" +
                $"t={t}&_tb_token_={_tb_token}&floorId={_floorId}&refpid={_refpid}&variableMap={variableMap}";


            var client = new WebClientUtility().SetContentType("application/json;charset=utf-8")
                        .AddHeaders("X-Requested-With", "XMLHttpRequest")
                        .AddHeaders("Cookie", _cookies);
            client.Proxy = _proxy;
            if (!string.IsNullOrEmpty(_user_agent)) client.UserAgent = _user_agent;
            client.Timeout = TimeSpan.FromMilliseconds(1000);
#if DEBUG
            client.Proxy = null;
            client.Timeout = TimeSpan.FromMilliseconds(10000);
#endif
            var response = client.Request(url);

            var body = string.Empty;
            try
            {
                if (!response.Successed)
                {
                    result.channel = TkChannelEnum.tb;
                    result.accountName = _accountName;
                    result.link_type = LinkTypeEnum.unknown;
                    result.success = false;
                    result.message = "fail";

                    if (response.ResponseException != null)
                    {
                        new LoggerLibrary("api_error", "fail")
                            .Info(response.ResponseException.Message, response.ResponseException.StackTrace)
                            .Save();
                        throw response.ResponseException;
                    }
                    return result;
                }


                body = response.Body();
                //{"code":601,"info":{"ok":false,"message":"nologin"}}


                JsonElement root = body.Convert2JsonElement();

                bool success = root.Read<bool>("success", false);
                string message = root.Read("message", string.Empty);
                string info_message = root.PathRead("info.message", string.Empty);
                if (!string.IsNullOrEmpty(info_message)) message = info_message;

                string taoToken = root.PathRead("data.taoToken", string.Empty);
                string shortLinkurl = root.PathRead("data.shortLinkurl", string.Empty);

                string couponLinkTaoToken = root.PathRead("data.couponLinkTaoToken", string.Empty);
                string couponShortLinkUrl = root.PathRead("data.couponShortLinkUrl", string.Empty);
                decimal couponAmount = root.PathRead<decimal>("data.couponAmount", 0);
                string couponEffectiveEndTime = root.PathRead("data.couponEffectiveEndTime", string.Empty);
                string couponEffectiveStartTime = root.PathRead("data.couponEffectiveStartTime", string.Empty);

                string itemId = root.PathRead("data.itemId", string.Empty);
                string itemName = root.PathRead("data.itemName", string.Empty);
                decimal promotionPrice = root.PathRead<decimal>("data.promotionPrice", 0);
                string sellerNickName = root.PathRead("data.sellerNickName", string.Empty);
                string shopTitle = root.PathRead("data.shopTitle", string.Empty);
                string pic = root.PathRead("data.pic", string.Empty);
                string qrCodeUrl = root.PathRead("data.qrCodeUrl", string.Empty);


                if (response.ResponseMessage != null)
                {

                    switch (response.ResponseMessage.StatusCode)
                    {
                        case System.Net.HttpStatusCode.OK:
                            {
                                if (couponAmount > 0 && !string.IsNullOrEmpty(couponLinkTaoToken) && !string.IsNullOrEmpty(couponShortLinkUrl))
                                {
                                    taoToken = couponLinkTaoToken;
                                    shortLinkurl = couponShortLinkUrl;
                                }
                                string shortLink = GetLink(taoToken);
                                content = ReplaceUrls(content, shortLink);
                            }
                            break;
                        case System.Net.HttpStatusCode.Found:
                            string location = response.ResponseMessage.Headers.Location.OriginalString;
                            if (!string.IsNullOrEmpty(location) && location.Contains("www.alimama.com/member/login.htm"))
                            {
                                success = false;
                                message = "nologin";
                            }
                            break;
                        default:
                            {
                                success = false;
                                message = "other";
                            }
                            break;
                    }
                }
                if (!success)
                {
                    shortLinkurl = GetLink(content);
                    switch (message)
                    {
                        case "该链接不支持转化，请更换链接尝试":
                        case "网络错误":
                            break;
                        default:
                            var _headers = response.ResponseMessage.Headers;
                            var headers = "";
                            foreach (var h in _headers)
                            {
                                foreach (var Value in h.Value)
                                {
                                    headers += $"{h.Key}: {Value}\n";
                                }
                            }
                            new LoggerLibrary("api_error", "fail")
                                .Info(message, content)
                                .Info(headers, body)
                                .Save();
                            break;
                    }
                }
                string deeplink_url = GetDeeplink(shortLinkurl);

                result.channel = TkChannelEnum.tb;
                result.accountName = _accountName;
                result.link_type = success ? LinkTypeEnum.goods : result.link_type;
                result.success = success;
                result.message = message;
                result.content = content;
                result.couponAmount = couponAmount;
                result.couponEffectiveEndTime = couponEffectiveEndTime;
                result.couponEffectiveStartTime = couponEffectiveStartTime;
                result.taoToken = taoToken;
                result.shortLinkurl = shortLinkurl;
                result.deeplink_url = deeplink_url;

                result.itemId = itemId;
                result.itemName = itemName;
                result.promotionPrice = promotionPrice;
                result.sellerNickName = sellerNickName;
                result.shopTitle = shopTitle;
                result.pic = pic;
                result.qrCodeUrl = qrCodeUrl;
            }
            catch
            {
                if (response.ResponseMessage?.StatusCode == System.Net.HttpStatusCode.OK)
                {
                    if (body.Contains("<p>抱歉！页面无法访问……</p>"))
                    {
                        throw new APIException("抱歉！页面无法访问……");
                    }
                }
                throw;
            }
            return result;
        }

    }
}