parse.cs 36 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860
  1. using System.Text;
  2. using dodohold.core;
  3. using System.Text.RegularExpressions;
  4. using System.Net;
  5. using System.Linq;
  6. using System.Security.Cryptography;
  7. using TencentCloud.Ecm.V20190719.Models;
  8. using System.Diagnostics;
  9. using static dodohold.core.ZTOExpress.CreateOrderArgs;
  10. using System.Text.Json;
  11. using TencentCloud.Tcm.V20210413.Models;
  12. using TencentCloud.Tione.V20211111.Models;
  13. using System.Runtime.InteropServices.Marshalling;
  14. namespace molilian.core
  15. {
  16. public partial class AlimamaPlus
  17. {
  18. private string _accountName;
  19. private int _accountId;
  20. private string _company;
  21. private string _tb_token;
  22. private string _floorId;
  23. private string _refpid;
  24. private string _user_agent;
  25. private string _cookies;
  26. public string _proxyName;
  27. public WebProxy? _proxy = null;
  28. private string _api;
  29. private int _api_id = 0;
  30. public TkConfigDTO _config;
  31. private static string _end_point;
  32. private TkPoolDTO _account;
  33. static Random _random = new Random();
  34. public static string _url_pattern = @"https?://[-A-Za-z0-9+&@#/%?=~_|!:,.;]+[-A-Za-z0-9+&@#/%=~_|]";
  35. static AlimamaPlus()
  36. {
  37. _end_point = Environment.GetEnvironmentVariable("EndPoint");
  38. }
  39. public AlimamaPlus(TkPoolDTO account)
  40. {
  41. // string accountName, string tb_token, string floorId, string refpid,
  42. // string cookies, string user_agent, string nodeName
  43. _config = TkConfigCore.Get();
  44. _account = account;
  45. _accountId = account.id;
  46. _accountName = account.name;
  47. _company = account.company;
  48. _tb_token = account.tb_token;
  49. _floorId = account.floorId;
  50. _refpid = account.refpid;
  51. _cookies = account.cookies;
  52. _user_agent = account.user_agent?.Trim();
  53. if (!string.IsNullOrEmpty(account.nodeName))
  54. {
  55. _proxyName = ProxyNodesCore.GetSelectOne(account.nodeName);
  56. _proxy = ProxyNodesCore.GetOne(_proxyName);
  57. }
  58. _api = account.api;
  59. _api_id = account.api_id;
  60. }
  61. public bool ShouldIgnoreOtherAff(string content)
  62. {
  63. TkDataDTO result = new();
  64. //(_, _, var link_type, _) = InternalTextProcessing(content, ref result);
  65. InternalTextProcessing(content, ref result);
  66. return result.link_type == LinkTypeEnum.other_aff;
  67. }
  68. public async Task<TkDataDTO> UnionParseAsync(UnionParseRequest request, string content, TkDataDTO result, bool isTaobaoUrl, CancellationToken cancellationToken = default,
  69. Dictionary<string, long> swData = null)
  70. {
  71. result.itemName = "点击打开淘宝APP";
  72. result = await InternalTextProcessingAsync(content, result, cancellationToken, swData);
  73. if (result.link_type == LinkTypeEnum.other_aff)
  74. {
  75. //============================== 其他推广链接 ==============================
  76. result.channel = TkChannelEnum.tb;
  77. result.channel_type = ChannelTypeEnum.tb;
  78. result.link_type = LinkTypeEnum.other_aff;
  79. result.success = false;
  80. result.message = "放弃转链";
  81. result.reason = "其他推广链接";
  82. result.subCode = TkSubCodeEnum.OtherPromo;
  83. result.accountName = string.Empty;
  84. result.content = content;
  85. //result.deeplink_url = GetDeeplink(null);
  86. //result.itemName = GetTitle(content);
  87. //if (!content.Contains("助我领红包") && !content.Contains("双11超级红包主会场"))
  88. //{
  89. // //20241108 特例
  90. // string ruleKey = $"tmp_rules:{DateTime.Now:yyyyMMdd}_2";
  91. // var count = RedisHelper.IncrBy(ruleKey);
  92. // if (DateTime.Now < DateTime.Parse("2024-11-11") && count < 1000)
  93. // {
  94. // string hb_url = "https://s.click.taobao.com/t?union_lens=lensId%3APUB%401730991020%400bbb0a39_0e75_193071ad936_c566%4001%40eyJmbG9vcklkIjozODg1Miiwiic3BtQiiI6Il9wb3J0YWxfdjJfcGFnZXNfYWN0aXZpdHlfb2ZmaWNpYWxfaW5kZXhfaHRtIn0ie%3BeventPageId%3A20150318020007201&e=m%3D2%26s%3DipuVKjYImbNw4vFB6t2Z2iperVdZeJviU%2F9%2F0taeK29yINtkUhsv0KWvzu%2FHMi3%2BFPhC4MirEWStw6LBB4w3HYXT%2BqQ2P5OIQVNSryTmvs7YddWYbxmvGTMNfihJN%2FGnAGIx0oe2X2hZfJ7ZQxC1%2FU3rpAhWgptfEBnEBk3xaGylLmcSHKfaX1CIFRJhZoJ2keMqUwSQcLRC0TzsMy8kp7Hfv3%2BBYcPW3eHn14rW1PB5SunYCHQHRbw6LwQYf%2F1JuyhJem%2BM5OoR2cbDLktPLiIOenYkFIXa2HzNl%2Bm1VXoil%2F2L3b1aA6a%2FoVHlY%2BqZ2kyLuBX5NHCAQVYCmAmuozC39JfnbJNrE%2FSR3F9BSjOYe76q6w0hyfybhwAASyyp5XMWEyz25ulHMwkL2n%2F7nNBiFmadXyjFDPu12TgSnbZtfjDbrVTLHb6MowVEq7z9LgOFRSuRLesvlLykGd2CaRk8ezhE61zXa8yR1eex6ip3KqbOh7CEmgCLaFxcnT%2BrGuviRFaBil8Mlu5kMKse3g%3D%3D";
  95. // result.deeplink_url = GetDeeplink(hb_url);
  96. // RedisHelper.Expire(ruleKey, 86400 * 10);
  97. // }
  98. //}
  99. return result;
  100. }
  101. //处理过后的正文 用于转链
  102. if (!result.success)
  103. {
  104. switch (result.link_type)
  105. {
  106. case LinkTypeEnum.profile:
  107. case LinkTypeEnum.live:
  108. case LinkTypeEnum.video:
  109. case LinkTypeEnum.note:
  110. case LinkTypeEnum.goods:
  111. case LinkTypeEnum.baseDomain:
  112. result.reason = string.Empty;
  113. break;
  114. default:
  115. result.reason = "非标准链接";
  116. result.subCode = TkSubCodeEnum.NonStdLink;
  117. return result;
  118. }
  119. }
  120. //============================== 放弃转链-地区过滤 ==============================
  121. bool is_ignore2 = FlowControlIgnoreRequest(request, out string reason);
  122. string message = "放弃转链";
  123. if (is_ignore2)
  124. {
  125. if ("7".Equals(_account.parseEndpoint))
  126. {
  127. message = "转链失败";
  128. result.reason = "fail2";
  129. }
  130. result.success = false;
  131. result.message = message;
  132. result.reason = reason;
  133. result.subCode = TkSubCodeEnum.TrafficCtrl;
  134. return result;
  135. }
  136. switch (request.CommerceType)
  137. {
  138. case 1:
  139. case 2:
  140. result.success = false;
  141. result.message = "放弃转链";
  142. result.reason = $"厂商放弃{request.CommerceType}";
  143. result.subCode = TkSubCodeEnum.PartnersDrop;
  144. return result;
  145. }
  146. //2025-02-06 增加重试机制
  147. int try_count = 0;
  148. while (try_count == 0 || try_count <= _config.tk_parse_retry_count)
  149. {
  150. if (try_count >= 3) break;
  151. (int ep_id, string tkEndpoint) = await TkEndpointManager.GetConvertApiAsync(_accountId, isTaobaoUrl, _account.parseEndpoint);
  152. result.tkEndpoint = ep_id;
  153. switch (tkEndpoint)
  154. {
  155. case "moose_share":
  156. result = await endpoint_moose_share(result.content, result, cancellationToken);
  157. break;
  158. case "promo_goods":
  159. result = await endpoint_promo_goods(result.content, result, cancellationToken);
  160. break;
  161. case "tool_links":
  162. result = await endpoint_tool_links(result.content, result, cancellationToken);
  163. break;
  164. case "aliyun":
  165. result = await endpoint_aliyun(result.content, result, cancellationToken);
  166. break;
  167. case "tbk_dg":
  168. result = await endpoint_tbk_dg(result.content, result, cancellationToken);
  169. break;
  170. case "veapi":
  171. case "ddx":
  172. case "other_api":
  173. if (!string.IsNullOrEmpty(_api) && _api_id > 0)
  174. {
  175. var account = OtherApiPoolCore.Get(_api_id);
  176. if (account != null)
  177. {
  178. var core = new OtherApiPlus(account.name, account.key, account.sessionKey, tkEndpoint, account.api_url);
  179. string refpid = string.IsNullOrEmpty(account.refpid) ? _refpid : account.refpid;
  180. result = await core.UnionParseAsync(result.content, refpid, result, cancellationToken);
  181. return result;
  182. }
  183. }
  184. else
  185. {
  186. result.link_type = LinkTypeEnum.unknown;
  187. result.subCode = TkSubCodeEnum.Other;
  188. result.success = false;
  189. result.message = "放弃转链";
  190. result.reason = "接口配置失败";
  191. return result;
  192. }
  193. break;
  194. //case "ALL":
  195. // //没有可以用的接口了
  196. // result.link_type = LinkTypeEnum.unknown;
  197. // result.subCode = TkSubCodeEnum.ParseDeny;
  198. // result.success = false;
  199. // result.message = "放弃转链";
  200. // result.reason = "接口风控ALL";
  201. // return result;
  202. default:
  203. result.link_type = LinkTypeEnum.unknown;
  204. result.subCode = TkSubCodeEnum.ParseDeny;
  205. result.success = false;
  206. result.message = "放弃转链";
  207. result.reason = $"接口风控{tkEndpoint}";
  208. break;
  209. }
  210. if (!"该链接不支持转化,请更换链接尝试".Equals(result.message)) break;
  211. try_count++;
  212. }
  213. result.retry_count = try_count;
  214. result.commercial = result.success;
  215. if (result.success && "brwsimilar".Equals(request.RiskStrategy))
  216. {
  217. // 需要搜同款才进入这个逻辑
  218. result.PromotionImg = await InternalCallPromotionQuery(result.pic, request.Oaid, request.Ip, _account.useCouponLinkFirst);
  219. // "pic": "https://img.alicdn.com/i4/6000000004750/O1CN01w4wgqa1kxYS1BadZp_!!6000000004750-0-sm.jpg",
  220. //result.pic
  221. //result.PromotionImg[*].pic
  222. int i = 0;
  223. foreach (var item in result.PromotionImg)
  224. {
  225. i++;
  226. item.clickTracks = TracksCore.BuildBrwSimilarPath(TrackType.Click, result, item, i);
  227. item.exposeTracks = TracksCore.BuildBrwSimilarPath(TrackType.Expose, result, item, i);
  228. item.pic = NormalizeAlicdnImageUrl(item.pic);
  229. }
  230. }
  231. result.pic = NormalizeAlicdnImageUrl(result.pic);
  232. if (!result.success)
  233. {
  234. result.itemName = "点击打开淘宝APP";
  235. }
  236. if (string.IsNullOrEmpty(result.itemName) || result.link_type == LinkTypeEnum.profile)
  237. {
  238. result.itemName = "点击打开淘宝APP";
  239. }
  240. return result;
  241. }
  242. private async Task<List<PromotionQueryItemDTO>> InternalCallPromotionQuery(string pic, string oaid, string ip, bool useCouponLinkFirst)
  243. {
  244. if (string.IsNullOrWhiteSpace(pic))
  245. {
  246. return [];
  247. }
  248. static bool TryParseItems(JsonElement element, out List<PromotionQueryItemDTO> items)
  249. {
  250. items = [];
  251. if (element.ValueKind == JsonValueKind.Array)
  252. {
  253. items = JsonSerializer.Deserialize<List<PromotionQueryItemDTO>>(element.GetRawText()) ?? [];
  254. return true;
  255. }
  256. if (element.ValueKind != JsonValueKind.Object)
  257. {
  258. return false;
  259. }
  260. if (element.TryGetProperty("promotionImg", out var promotionImg) && promotionImg.ValueKind == JsonValueKind.Array)
  261. {
  262. items = JsonSerializer.Deserialize<List<PromotionQueryItemDTO>>(promotionImg.GetRawText()) ?? [];
  263. return true;
  264. }
  265. if (element.TryGetProperty("PromotionImg", out promotionImg) && promotionImg.ValueKind == JsonValueKind.Array)
  266. {
  267. items = JsonSerializer.Deserialize<List<PromotionQueryItemDTO>>(promotionImg.GetRawText()) ?? [];
  268. return true;
  269. }
  270. return false;
  271. }
  272. static List<PromotionQueryItemDTO> RemoveFirstItem(List<PromotionQueryItemDTO> items)
  273. {
  274. return items.Count > 0 ? items.Skip(1).ToList() : [];
  275. }
  276. string requestUrl = "https://similar.molilian.com/api/unsafePromotionQuery";
  277. string targetPic = pic.StartsWith("//") ? $"https:{pic}" : pic;
  278. object postData = new
  279. {
  280. url = targetPic,
  281. oaid,
  282. ip,
  283. recommand = false,
  284. useCouponLinkFirst
  285. };
  286. string post = postData.Convert2Json();
  287. try
  288. {
  289. var client = new WebClientUtility().SetContentType("application/json")
  290. .Post(post);
  291. var response = await client.RequestAsync(requestUrl, "POST");
  292. var body = response.Body();
  293. if (string.IsNullOrWhiteSpace(body))
  294. {
  295. return [];
  296. }
  297. var root = body.Convert2JsonElement();
  298. if (TryParseItems(root, out var result))
  299. {
  300. return RemoveFirstItem(result);
  301. }
  302. if (root.ValueKind == JsonValueKind.Object)
  303. {
  304. if (root.TryGetProperty("data", out var data) && TryParseItems(data, out result))
  305. {
  306. return RemoveFirstItem(result);
  307. }
  308. if (root.TryGetProperty("result", out var resultNode) && TryParseItems(resultNode, out result))
  309. {
  310. return RemoveFirstItem(result);
  311. }
  312. }
  313. _ = new LoggerLibrary("AlimamaPlus", "InternalCallPromotionQuery")
  314. .Info(post, body)
  315. .SaveAsync();
  316. }
  317. catch (Exception ex)
  318. {
  319. _ = new LoggerLibrary("AlimamaPlus", "InternalCallPromotionQuery")
  320. .Info(requestUrl, post)
  321. .Info(ex.Message, ex.StackTrace)
  322. .SaveAsync();
  323. }
  324. return [];
  325. }
  326. private static string NormalizeAlicdnImageUrl(string pic)
  327. {
  328. const string formatSuffix = "_400x400q90.jpg_.webp";
  329. if (string.IsNullOrWhiteSpace(pic))
  330. {
  331. return pic;
  332. }
  333. if (!pic.Contains("img.alicdn.com", StringComparison.OrdinalIgnoreCase))
  334. {
  335. return pic;
  336. }
  337. if (pic.Contains(formatSuffix, StringComparison.OrdinalIgnoreCase))
  338. {
  339. return pic;
  340. }
  341. return $"{pic}{formatSuffix}";
  342. }
  343. public static bool ContainsTaobaoId(string content, string rawContent)
  344. {
  345. var match = Regex.Match(content, @"(?<=[?&]id=)\d+(?=&|$)");
  346. if (!match.Success) return false;
  347. string id = match.Value;
  348. return Regex.IsMatch(rawContent, $@"(^|\D){id}($|\D)");
  349. }
  350. public static string ReplaceUrls(string content, string shortLink)
  351. {
  352. if (string.IsNullOrEmpty(content) || string.IsNullOrEmpty(shortLink)) return content;
  353. string result = content;
  354. Regex regex = new(_url_pattern);
  355. MatchCollection matches = regex.Matches(content);
  356. foreach (Match m in matches.Cast<Match>())
  357. {
  358. string url = m.Value;
  359. if (url.Contains("https://s.tb.cn/") ||
  360. url.Contains("http://s.tb.cn/") ||
  361. url.Contains("http://m.tb.cn/") ||
  362. url.Contains("http://m.tb.cn/") ||
  363. url.Contains("http://e.tb.cn/") ||
  364. url.Contains("https://e.tb.cn/"))
  365. {
  366. result = result.Replace(url, shortLink);
  367. }
  368. }
  369. return result;
  370. }
  371. public static string RemoveUrls(string content)
  372. {
  373. if (string.IsNullOrEmpty(content)) return content;
  374. string result = content;
  375. Regex regex = new(_url_pattern);
  376. MatchCollection matches = regex.Matches(content);
  377. foreach (Match m in matches.Cast<Match>())
  378. {
  379. string url = m.Value;
  380. result = result.Replace(url, string.Empty);
  381. }
  382. return result;
  383. }
  384. public static string GetLink(string content)
  385. {
  386. if (string.IsNullOrEmpty(content)) return content;
  387. //string pattern = @"https?:\/\/?([\da-z\.-]+)\.([a-z]{2,6})(:\d{1,5})?([\/\w\.-]*)*\/?(#[\S]+)?(\?[\S]+)?";
  388. Regex regex = new(_url_pattern);
  389. MatchCollection matches = regex.Matches(content);
  390. string result = string.Empty;
  391. if (matches.Count > 0)
  392. {
  393. result = Regex.Replace(matches[^1].Value, @"\++$", "");
  394. }
  395. result = result.Replace("&nbsp;", "");
  396. if (result.StartsWith("http://e.tb.cn/")) result = result.Replace("http://", "https://");
  397. return result;
  398. }
  399. public static string GetTaobaoLink(string content)
  400. {
  401. if (string.IsNullOrEmpty(content)) return content;
  402. string pattern = @"(https?://(?:[\w-]+\.)*(?:tb\.cn|taobao\.com|juhuasuan\.com|tmall\.com|tmall\.hk)(?:/[a-zA-Z0-9\-\._~:/?#[\]@!$&'()*+,;=%]*)?)";
  403. Regex regex = new(pattern, RegexOptions.IgnoreCase);
  404. MatchCollection matches = regex.Matches(content);
  405. if (matches.Count > 0)
  406. {
  407. string result = Regex.Replace(matches[^1].Value, @"\++$", "");
  408. return result;
  409. }
  410. return string.Empty;
  411. }
  412. public static string GetDeeplink(string url)
  413. {
  414. if (string.IsNullOrEmpty(url)) return "tbopen://m.taobao.com/tbopen/index.html";
  415. string result = $"tbopen://m.taobao.com/tbopen/index.html?action=ali.open.nav&module=h5&h5Url={url.UrlEncode()}";
  416. //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()}";
  417. return result;
  418. }
  419. public static string GetTitle(string content)
  420. {
  421. if (string.IsNullOrEmpty(content)) return string.Empty;
  422. string url = GetLink(content);
  423. if (!string.IsNullOrEmpty(url)) content = content.Replace(url, string.Empty);
  424. content = content.Replace("点击链接直接打开 或者 淘宝搜索直接打开", string.Empty).Trim();
  425. string[] words = Regex.Split(content, @"\s+");
  426. string word = words[^1];
  427. word = word.TrimStart('「').TrimEnd('」');
  428. return word;
  429. }
  430. public static bool IgnoreRegionIncluded(string _ignoreRegions, string? ipInfo, out string regionInfo)
  431. {
  432. regionInfo = string.Empty;
  433. if (string.IsNullOrEmpty(_ignoreRegions) || string.IsNullOrEmpty(ipInfo)) return false;
  434. string[] parts = ipInfo.Split('|');
  435. string countryInfo = null;
  436. if (parts.Length >= 4)
  437. {
  438. countryInfo = parts[0];
  439. regionInfo = parts[3];
  440. }
  441. if (string.IsNullOrEmpty(regionInfo)) return false;
  442. if (string.IsNullOrEmpty(countryInfo)) return false;
  443. string[] ignoreRegions = _ignoreRegions.Split('|');
  444. if (ignoreRegions.Contains(regionInfo)) return true;
  445. if (ignoreRegions.Contains(countryInfo))
  446. {
  447. regionInfo = countryInfo;
  448. return true;
  449. }
  450. foreach (var region in ignoreRegions)
  451. {
  452. if ("海外".Equals(region) && !"中国".Equals(countryInfo) && !"0".Equals(countryInfo))
  453. {
  454. regionInfo = "海外";
  455. return true;
  456. }
  457. }
  458. return false;
  459. }
  460. public static async Task<bool> OtherPlatformAsync(string content)
  461. {
  462. try
  463. {
  464. //腾讯会议
  465. string wemeetId = await ToolParsePlus.GetWemeetIdAsync(content);
  466. if (!string.IsNullOrEmpty(wemeetId)) return true;
  467. //百度网盘
  468. string surl = string.Empty, pwd = string.Empty;
  469. _ = ToolParsePlus.GetPanLink(content, ref surl, ref pwd);
  470. if (!string.IsNullOrEmpty(surl)) return true;
  471. //JD
  472. string url = JdUnionPlus.GetLink(content);
  473. var urlType = JdUnionPlus.GetLinkType(url, out _, out _);
  474. if (urlType != LinkTypeEnum.unknown) return true;
  475. //DY
  476. url = DyUnionPlus.GetLink(content);
  477. urlType = DyUnionPlus.GetLinkType(content, url);
  478. if (urlType != LinkTypeEnum.unknown) return true;
  479. }
  480. catch { }
  481. return false;
  482. }
  483. public static bool MatchRegexes(string text, List<string> extended)
  484. {
  485. text = RemoveUrls(text);
  486. string special_symbols = $"\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲";
  487. string[] base_patterns =
  488. [
  489. $@"(.*?\d{{2}}\s[a-zA-Z]{{2}}\d{{4}}\s.*?[{special_symbols}]+[a-zA-Z0-9\s]+[{special_symbols}]).*",
  490. $@"[{special_symbols}]+\s*([a-zA-Z]{{1,2}}\d{{2,4}}\s[a-zA-Z0-9]{{11}}|[a-zA-Z0-9]{{11}}[a-zA-Z]{{1,2}}\d{{2,4}}\s|[a-zA-Z0-9]{{11}})\s*[{special_symbols}]+",
  491. @"🍑♭ɑ◑下单|Tao宝"
  492. ];
  493. List<string> mergedArray = extended.Union(base_patterns).ToList();
  494. /*
  495. (.*?\\d{2}\\s[a-zA-Z]{2}\\d{4}\\s.*?[\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+[a-zA-Z0-9\\s]+[\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲])).*\n[\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+\\s*([a-zA-Z]{1,2}\\d{2,4}\\s[a-zA-Z0-9]{11}|[a-zA-Z0-9]{11}[a-zA-Z]{1,2}\\d{2,4}\\s|[a-zA-Z0-9]{11})\\s*[\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲]+\n🍑♭ɑ◑下单|Tao宝
  496. 【3瓶】恩威万蓝莓叶黄素酯软糖儿童成人上班族如酸涩肿胀可服用,<z5TYWAblvTw >:// HU7709,打開/
  497. 置顶小助手,方便下次查询\r\n1★【劵】无\r\n2★【返.佣】 0.29圆 \r\n3★【付款.价】24.9 圆\r\n \r\n 1 ¢HZv5WAboi0D¢(/:/ HU8866\r\
  498. n长按覆制本内容,打开TaoBao/\r\n-\r\n返佣是指商家的推广佣.金\r\n通过我回复的口令下単收货后,我
  499. 会微信红.苞返给你!!\r\n\r\n有疑问可回复“人工”咨询客服*/
  500. //🍑🔗上 水和肌精华水& CZ0 VWdCWC2gj97&,
  501. //6฿04ZDWzrwxjN)/ CA97
  502. //1😘 腹制这条(CZ00 iHvtWzrXa7O(:/
  503. //¥EC8DG4Zj0dhd8DOw¥MF6563 oIo0WzOF1Cp##X-PyVpHiPlyj1OP##
  504. //6.6₵vU2jWzIW3Qx₲/ CZ5224
  505. //$F56nWzIDSFZ$
  506. //0復製这条口令( CZ00 D0M4WzI0jFJ¢,📱咑開[Tao宝]就可下单:/
  507. //8fuzhi本条消息:¤ CZ3458 JSR1WzrPCBU¤,打开🍑♭ɑ◑下单/
  508. //5€Lz6SWzrkLDw¥:// CZ13
  509. //3₲Igq3Wzkc5np₳:// CZ15
  510. //9.9🎈【康师傅易拉罐6罐】口伶:6₰TnxKWzr0wff₪:// CZ69
  511. //4₲99Y0WzJdK6S₳:// CZ46
  512. //4fuzhi本条消息:¤ CZ0002 WVCSWzrk6Ix¤,打开🍑♭ɑ◑下单/
  513. bool anyMatch = false;
  514. foreach (string pattern in mergedArray)
  515. {
  516. if (Regex.IsMatch(text, pattern, RegexOptions.IgnoreCase))
  517. {
  518. anyMatch = true;
  519. }
  520. }
  521. return anyMatch;
  522. }
  523. public static bool MatchOtherInfo(string text, List<string> extended)
  524. {
  525. string[] base_patterns =
  526. [
  527. "[ďȌƯƔɪƝɖƟɏƖԂȎȗϓɫɧơȲǏđǫƲƳɨǹƊȮǚƴɬɲƿȖȳɳɗʮϔӢǙɦɒɎƥʘưΊƞǪȈʠΌǓϒίȠƠƱƗǸþŌŬÿĩŅÞÓũŸÎŎûįŇĎÖÙŷÏňÜÝÍñĎÒüŶľńĎ0ųŷÏŇdǒǔyí℡抖yí℡ÞŐûŷÌñ]{6}",
  528. @"(?:.+)极速版(?:口令|搜索)",
  529. @"##[a-zA-Z0-9]+##\[抖音(?:极速版)?(?:口令|搜索)\]",
  530. @"岽|亰|菄|京东|婛",
  531. "打开【剪映】",
  532. "weishi://|baiduhaokan://",
  533. "复制打开「UC浏览器」|UC瀏覽器",
  534. "番茄免费小说",
  535. @"\?chanTag=[^&\s]*口令",
  536. @"(https?://pan\.baidu\.com/s/[A-Za-z0-9?\-_=]+)",
  537. @"https?:\/\/meeting\.tencent\.com[^\s]*",
  538. @"#腾讯会议:(\d{3}-\d{3}-\d{3})"
  539. ];
  540. List<string> mergedArray = extended.Union(base_patterns).ToList();
  541. bool anyMatch = false;
  542. foreach (string pattern in mergedArray)
  543. {
  544. if (Regex.IsMatch(text, pattern, RegexOptions.IgnoreCase))
  545. {
  546. anyMatch = true;
  547. }
  548. }
  549. return anyMatch;
  550. }
  551. public static bool MatchMultiToken(string text, List<string> extended)
  552. {
  553. text = RemoveUrls(text);
  554. string special_symbols = $"\\$\\(\\)\\/\\#\\&\\<\\>฿¢¥¥$🗝₰₵₲¤₪₴€₤£₳《¢💰🔑🎵✔💲🔐📲";
  555. string[] base_patterns =
  556. [
  557. @"[!$¥%]+([a-zA-Z0-9]{16})[!$¥%]+",
  558. @"\^v(\d+#[a-zA-Z0-9]+)(?:\?.*)?\^",
  559. "##([a-zA-Z0-9\\-]+)##",
  560. $@"(.*?\d{{2}}\s[a-zA-Z]{{2}}\d{{4}}\s.*?[{special_symbols}]+[a-zA-Z0-9\s]+[{special_symbols}]).*",
  561. $@"[{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}]+",
  562. ];
  563. // PAGE6:/^v130091#P2J6EDxN8p8?chanTag=d9dc522bdea27437514052ea50f6a787^
  564. List<string> mergedArray = extended.Union(base_patterns).ToList();
  565. int matchCount = 0;
  566. foreach (string pattern in mergedArray)
  567. {
  568. Match match = Regex.Match(text, pattern, RegexOptions.IgnoreCase);
  569. if (match.Success)
  570. {
  571. matchCount++;
  572. text = text.Replace(match.Groups[^1].Value, ""); // 移除匹配的部分文本
  573. }
  574. }
  575. return matchCount >= 2;
  576. }
  577. public static bool FilterRiskLink(string ip, string oaid, string content, out string reason)
  578. {
  579. reason = string.Empty;
  580. string itemId = ExtractItemId(content);
  581. if (string.IsNullOrEmpty(itemId)) return false;
  582. string cacheKey = $":cache:parse:{ip}_{oaid}_{itemId}";
  583. int flag = RedisHelper.Get<int>(cacheKey);
  584. if (flag > 0)
  585. {
  586. reason = $"口令用户";
  587. return true;
  588. }
  589. return false;
  590. }
  591. public static bool ShouldIgnoreRequest(UnionParseRequest request, out string reason)
  592. {
  593. reason = string.Empty;
  594. ////2025-05-11 brw不做风控
  595. //if (!string.IsNullOrEmpty(riskStrategy) && !"os".Equals(riskStrategy) && launchScene != -1) return false;
  596. switch (request.RiskStrategy)
  597. {
  598. case "tbpush":
  599. case "brwsimilar":
  600. case "icon":
  601. return false;
  602. }
  603. try
  604. {
  605. // IP和流量控制
  606. var config = TkConfigCore.Get();
  607. string ignorePercentageCity = config.ignorePercentageCity;
  608. string? ipInfo = IP2RegionPlus.Search(request.Ip);
  609. if (!TestParseCore.InWhitelist(request.Ip, request.Oaid) && IgnoreRegionIncluded(ignorePercentageCity, ipInfo, out string regionInfo))
  610. {
  611. reason = $"地区控制:{regionInfo}";
  612. return true;
  613. }
  614. if (TkLogCore.InBlacklist(config.blacklist_oaid, request.Oaid))
  615. {
  616. reason = "OAID黑名单";
  617. return true;
  618. }
  619. int limit_num = config.tk_limit_per_ip_24h;
  620. if (!TestParseCore.InWhitelist(request.Ip, request.Oaid) && limit_num > 0)
  621. {
  622. int num = TkLogCore.getClientRequestTotalByIp(TkChannelEnum.tb, request.Ip);
  623. if (num >= limit_num)
  624. {
  625. reason = "IP控制";
  626. return true;
  627. }
  628. }
  629. limit_num = config.tk_limit_per_oaid_24h;
  630. if (!TestParseCore.InWhitelist(request.Ip, request.Oaid) && limit_num > 0)
  631. {
  632. int num = TkLogCore.getClientRequestTotalByOAID(TkChannelEnum.tb, request.Oaid);
  633. if (num >= limit_num)
  634. {
  635. reason = "OAID控制";
  636. return true;
  637. }
  638. }
  639. }
  640. catch (Exception ex)
  641. {
  642. reason = "配置异常";
  643. }
  644. return false;
  645. }
  646. public static bool FlowControlIgnoreRequest(UnionParseRequest request, out string reason)
  647. {
  648. reason = string.Empty;
  649. ////2025-05-11 brw不做风控
  650. //if (!string.IsNullOrEmpty(riskStrategy) && !"os".Equals(riskStrategy) && launchScene != -1) return false;
  651. switch (request.RiskStrategy)
  652. {
  653. case "tbpush":
  654. case "brwsimilar":
  655. case "icon":
  656. return false;
  657. }
  658. if (TestParseCore.InWhitelist(request.Ip, request.Oaid)) return false;
  659. try
  660. {
  661. // IP和流量控制
  662. var config = TkConfigCore.Get();
  663. int ignorePercentage = config.ignorePercentage;
  664. if (ignorePercentage == 0) return false;
  665. var randomValue = (decimal)_random.Next(0, 100);
  666. 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
  667. if (result)
  668. {
  669. reason = "流量控制";
  670. return true;
  671. }
  672. }
  673. catch (Exception ex)
  674. {
  675. reason = $"配置异常";
  676. }
  677. return false;
  678. }
  679. public static bool IsDigitsOnly(string? s)
  680. {
  681. return !string.IsNullOrEmpty(s)
  682. && s.AsSpan().IndexOfAnyExceptInRange('0', '9') < 0;
  683. }
  684. public static TkDataDTO GetFormattedObject(UnionParseRequest request)
  685. {
  686. string content = request.Content;
  687. switch (request.RiskStrategy)
  688. {
  689. case "tbpush":
  690. case "brwsimilar":
  691. case "icon":
  692. if (IsDigitsOnly(content))
  693. {
  694. content = $"https://item.taobao.com/item.htm?id={content}";
  695. }
  696. break;
  697. }
  698. string shortLinkurl = GetTaobaoLink(content);
  699. string deeplink_url = GetDeeplink(shortLinkurl);
  700. return new TkDataDTO()
  701. {
  702. message = string.Empty,
  703. subCode = TkSubCodeEnum.Other,
  704. channel = TkChannelEnum.tb,
  705. accountName = string.Empty,
  706. success = false,
  707. content = content,
  708. rawContent = request.Content,
  709. link_type = LinkTypeEnum.unknown,
  710. ip = request.Ip,
  711. oaid = request.Oaid,
  712. couponAmount = 0,
  713. taoToken = string.Empty,
  714. elapsedTime = 0,
  715. itemName = "点击打开淘宝APP",
  716. shortLinkurl = shortLinkurl,
  717. deeplink_url = deeplink_url,
  718. create_time = DateTime.Now,
  719. end_point = _end_point,
  720. parse_type = request.Type,
  721. riskStrategy = request.RiskStrategy,
  722. launchScene = request.LaunchScene,
  723. };
  724. }
  725. public static string OppoDecode(string content)
  726. {
  727. try
  728. {
  729. string secretKey = "BwFeIvOEDZy9MFGi0JLZBO4yEhR44DGV";
  730. byte[] keyBytes = Encoding.UTF8.GetBytes(secretKey);
  731. if (!content.Contains("%IV%")) return content;
  732. string cipherText = content.GetContentPart("", "%IV%");
  733. string ivText = content.GetContentPart("%IV%", "");
  734. byte[] cipherBytes = Convert.FromBase64String(cipherText);
  735. byte[] ivBytes = Convert.FromBase64String(ivText);
  736. using RijndaelManaged aes = new();
  737. aes.Key = keyBytes;
  738. aes.IV = ivBytes;
  739. aes.Mode = CipherMode.CFB;
  740. aes.Padding = PaddingMode.PKCS7;
  741. ICryptoTransform decryptor = aes.CreateDecryptor(aes.Key, aes.IV);
  742. var outBytes = decryptor.TransformFinalBlock(cipherBytes, 0, cipherBytes.Length);
  743. content = outBytes.ConvertToString();
  744. }
  745. catch (Exception e) { }
  746. return content;
  747. }
  748. }
  749. }