PddUnionPlus.cs 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528
  1. using dodohold.core;
  2. using Sayaka.Common;
  3. using System.Diagnostics;
  4. using System.Text.RegularExpressions;
  5. using System.Threading.Channels;
  6. using System.Web;
  7. namespace molilian.core
  8. {
  9. public enum RrecheckGoodsIdResult
  10. {
  11. Success,
  12. Empty,
  13. Exception,
  14. InvalidLink,
  15. OtherAff,
  16. }
  17. public partial class PddUnionPlus
  18. {
  19. private static readonly object _precheckCircuitLock = new();
  20. private static bool _precheckDisabled = false;
  21. private static int _precheckConsecutiveExceptionCount = 0;
  22. private static int _precheckDisableErrorThreshold = 10;
  23. public static string _url_pattern = @"https?://(?:[\w-]+\.)*(?:yangkeduo\.com|pinduoduo\.com)(?:/[^\s]*)?";
  24. public static string GetLink(string content)
  25. {
  26. if (string.IsNullOrEmpty(content)) return content;
  27. string result = content;
  28. //
  29. Regex regex = new(_url_pattern);
  30. MatchCollection matches = regex.Matches(content);
  31. foreach (Match m in matches.Cast<Match>())
  32. {
  33. string url = m.Value;
  34. return url;
  35. }
  36. return string.Empty;
  37. }
  38. public static string GetDeeplink(string url)
  39. {
  40. string result = $"pinduoduo://com.xunmeng.pinduoduo/{url}";
  41. return result;
  42. }
  43. private static bool IsGoodsDetailPath(string path)
  44. {
  45. if (string.IsNullOrWhiteSpace(path)) return false;
  46. return Regex.IsMatch(path, @"^/goods\d*\.html$", RegexOptions.IgnoreCase);
  47. }
  48. private static bool IsPddShortLinkHost(string host)
  49. {
  50. return "p.pinduoduo.com".Equals(host, StringComparison.OrdinalIgnoreCase);
  51. }
  52. private static bool TryGetPrecheckPsGoodsUrl(string url, out string normalizedUrl)
  53. {
  54. normalizedUrl = string.Empty;
  55. if (string.IsNullOrWhiteSpace(url)) return false;
  56. if (!Uri.TryCreate(url, UriKind.Absolute, out var uri)) return false;
  57. if (IsPddShortLinkHost(uri.Host) && !string.IsNullOrWhiteSpace(uri.AbsolutePath) && uri.AbsolutePath != "/")
  58. {
  59. normalizedUrl = $"{uri.Scheme}://{uri.Host}{uri.PathAndQuery}";
  60. return true;
  61. }
  62. if (!IsGoodsDetailPath(uri.AbsolutePath)) return false;
  63. var queryParams = HttpUtility.ParseQueryString(uri.Query);
  64. string goodsId = queryParams["goods_id"];
  65. if (!string.IsNullOrWhiteSpace(goodsId) && long.TryParse(goodsId, out _))
  66. {
  67. return false;
  68. }
  69. string ps = queryParams["ps"];
  70. if (string.IsNullOrWhiteSpace(ps)) return false;
  71. normalizedUrl = $"{uri.Scheme}://{uri.Host}{uri.AbsolutePath}?ps={HttpUtility.UrlEncode(ps)}";
  72. return true;
  73. }
  74. private static bool IsPrecheckDisabled()
  75. {
  76. lock (_precheckCircuitLock)
  77. {
  78. return _precheckDisabled;
  79. }
  80. }
  81. private static void ResetPrecheckExceptionState()
  82. {
  83. lock (_precheckCircuitLock)
  84. {
  85. _precheckConsecutiveExceptionCount = 0;
  86. }
  87. }
  88. public static void ResetPrecheckCircuitState()
  89. {
  90. lock (_precheckCircuitLock)
  91. {
  92. _precheckDisabled = false;
  93. _precheckConsecutiveExceptionCount = 0;
  94. }
  95. }
  96. private static (int count, bool disabledNow) RegisterPrecheckException()
  97. {
  98. lock (_precheckCircuitLock)
  99. {
  100. _precheckConsecutiveExceptionCount++;
  101. bool disabledNow = false;
  102. if (!_precheckDisabled && _precheckConsecutiveExceptionCount >= _precheckDisableErrorThreshold)
  103. {
  104. _precheckDisabled = true;
  105. disabledNow = true;
  106. }
  107. return (_precheckConsecutiveExceptionCount, disabledNow);
  108. }
  109. }
  110. private static void NotifyPrecheckException(string goodsUrl, string stage, string detail, int count, bool disabledNow)
  111. {
  112. string message = $"【PDD商品Precheck异常】{_end_point}\n" +
  113. $"url:{goodsUrl}\n" +
  114. $"stage:{stage}\n" +
  115. $"连续异常:{count}/{_precheckDisableErrorThreshold}\n" +
  116. $"{detail}";
  117. if (disabledNow)
  118. {
  119. message += "\n状态: 已临时禁用Precheck,后续默认返回Success";
  120. }
  121. NotifyCore.Notify(new NifyMessage
  122. {
  123. message = message,
  124. priority = NifyMessagePriority.high,
  125. tags = ["red_circle"]
  126. });
  127. }
  128. private string ExtractGoodsIdUrl(string url)
  129. {
  130. var uri = new Uri(url);
  131. var queryParams = HttpUtility.ParseQueryString(uri.Query);
  132. string goodsId = queryParams["goods_id"];
  133. if (!string.IsNullOrEmpty(goodsId) && uri.AbsolutePath.StartsWith("/goods"))
  134. {
  135. // 保持域名和路径不变,仅保留 goods_id 参数
  136. var baseUrl = $"{uri.Scheme}://{uri.Host}{uri.AbsolutePath}";
  137. return $"{baseUrl}?goods_id={goodsId}";
  138. }
  139. return url; // 如果没有找到 goods_id 参数,则返回原始 URL
  140. }
  141. private bool IsTrackUrl(string sourceUrl)
  142. {
  143. string pattern = @"^https?://mobile\.yangkeduo\.com/goods(\d*)\.html\?ps=.*$";
  144. Regex regex = new Regex(pattern);
  145. return regex.IsMatch(sourceUrl);
  146. }
  147. /// <summary>
  148. /// bool:是否ps链接,string 处理后的链接
  149. /// </summary>
  150. /// <param name="sourceUrl"></param>
  151. /// <param name="cancellationToken"></param>
  152. /// <returns></returns>
  153. public async Task<string> GetRedirectedUrlAsync(string sourceUrl, CancellationToken cancellationToken = default)
  154. {
  155. var isTrackUrl = IsTrackUrl(sourceUrl);
  156. if (!isTrackUrl) return ExtractGoodsIdUrl(sourceUrl);
  157. string redirectedUrl = sourceUrl;
  158. try
  159. {
  160. string useragent = ProviderFakeUserAgent.RandomMobile;
  161. WebClientUtility client = new()
  162. {
  163. AllowAutoRedirect = false,
  164. UserAgent = useragent,
  165. };
  166. client.Proxy = _proxy;
  167. #if DEBUG
  168. client.Proxy = null;
  169. #endif
  170. var response = await client.RequestAsync(sourceUrl, "GET", cancellationToken);
  171. if (!response.Successed) return null;
  172. if (response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.RedirectKeepVerb ||
  173. response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.TemporaryRedirect)
  174. {
  175. redirectedUrl = response.ResponseMessage.Headers.Location.ToString();
  176. redirectedUrl = ExtractGoodsIdUrl(redirectedUrl);
  177. return redirectedUrl;
  178. }
  179. }
  180. catch (Exception ex)
  181. {
  182. }
  183. return sourceUrl;
  184. }
  185. public static PddDataDTO GetFormattedObject(UnionParseRequest request)
  186. {
  187. switch (request.RiskStrategy)
  188. {
  189. case "tbpush":
  190. if (AlimamaPlus.IsDigitsOnly(request.Content))
  191. {
  192. request.Content = $"https://mobile.yangkeduo.com/goods.html?goods_id={request.Content}";
  193. }
  194. break;
  195. }
  196. string shortLinkurl = GetLink(request.Content);
  197. string deeplink_url = GetDeeplink(shortLinkurl);
  198. var link_type = LinkTypeEnum.unknown;
  199. if (!string.IsNullOrEmpty(shortLinkurl) && shortLinkurl.StartsWith("https://p.pinduoduo.com/"))
  200. {
  201. link_type = LinkTypeEnum.other_aff;
  202. }
  203. return new PddDataDTO()
  204. {
  205. message = string.Empty,
  206. link_type = link_type,
  207. channel = TkChannelEnum.pdd,
  208. accountName = string.Empty,
  209. success = false,
  210. content = request.Content,
  211. ip = request.Ip,
  212. oaid = request.Oaid,
  213. elapsedTime = 0,
  214. itemName = "点击打开拼多多APP",
  215. shortLinkurl = shortLinkurl,
  216. deeplink_url = string.Empty,
  217. create_time = DateTime.Now,
  218. end_point = _end_point,
  219. parse_type = request.Type,
  220. riskStrategy = request.RiskStrategy,
  221. launchScene = request.LaunchScene,
  222. };
  223. }
  224. public static async Task<RrecheckGoodsIdResult> PrecheckGoodsIdAsync(string content, CancellationToken cancellationToken = default)
  225. {
  226. var config = TkConfigCore.Get();
  227. if (config != null && !config.pdd_rrecheck_enabled)
  228. {
  229. return RrecheckGoodsIdResult.Success;
  230. }
  231. string url = GetLink(content);
  232. if (string.IsNullOrEmpty(url))
  233. {
  234. return RrecheckGoodsIdResult.InvalidLink;
  235. }
  236. if (!TryGetPrecheckPsGoodsUrl(url, out string goods_url))
  237. {
  238. return RrecheckGoodsIdResult.Success;
  239. }
  240. if (IsPrecheckDisabled()) return RrecheckGoodsIdResult.Success;
  241. if (string.IsNullOrWhiteSpace(goods_url)) return RrecheckGoodsIdResult.Empty;
  242. try
  243. {
  244. string precheckClientId = config?.pdd_precheck_client_id;
  245. string precheckClientSecret = config?.pdd_precheck_client_secret;
  246. string precheckPid = config?.pdd_precheck_pid;
  247. if (string.IsNullOrWhiteSpace(precheckClientId) || string.IsNullOrWhiteSpace(precheckClientSecret) || string.IsNullOrWhiteSpace(precheckPid))
  248. {
  249. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync.Config")
  250. .Info(goods_url, "precheck config missing")
  251. .SaveAsync();
  252. var status = RegisterPrecheckException();
  253. NotifyPrecheckException(goods_url, "Config", "precheck config missing", status.count, status.disabledNow);
  254. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  255. }
  256. var precheckAccount = new PddPoolDTO
  257. {
  258. app_key = precheckClientId,
  259. app_secret = precheckClientSecret
  260. };
  261. var ts = DateTime.Now.Convert2UnixTimestamp(true);
  262. var args = new Dictionary<string, string>
  263. {
  264. { "type", "pdd.ddk.goods.search" },
  265. { "data_type", "JSON" },
  266. { "client_id", precheckClientId },
  267. { "keyword", goods_url.UrlDecode() },
  268. { "pid", precheckPid },
  269. { "timestamp", ts.ToString() }
  270. };
  271. args = GenerateSignature(precheckAccount, args);
  272. var query = HttpUtility.ParseQueryString(string.Empty);
  273. foreach (var kvp in args) query[kvp.Key] = kvp.Value;
  274. string urlWithQuery = $"{_baseUrl}?{query}";
  275. WebClientUtility client = new();
  276. client.SetContentType("application/json");
  277. var response = await client.RequestAsync(urlWithQuery, "GET", cancellationToken);
  278. string body = response.Body();
  279. if (!response.Successed)
  280. {
  281. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  282. var status = RegisterPrecheckException();
  283. NotifyPrecheckException(goods_url, "RequestAsync", $"response.Successed=false\nbody:{body}", status.count, status.disabledNow);
  284. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  285. }
  286. var root = body.Convert2Object<GoodsSearchDTO>();
  287. if (root?.error_response != null)
  288. {
  289. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  290. var status = RegisterPrecheckException();
  291. var error = root.error_response;
  292. NotifyPrecheckException(
  293. goods_url,
  294. "ErrorResponse",
  295. $"error_code:{error.error_code}\nsub_code:{error.sub_code}\nerror_msg:{error.error_msg}\nsub_msg:{error.sub_msg}\nbody:{body}",
  296. status.count,
  297. status.disabledNow);
  298. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  299. }
  300. var goodsList = root?.goods_search_response?.goods_list;
  301. if (goodsList == null)
  302. {
  303. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  304. var status = RegisterPrecheckException();
  305. NotifyPrecheckException(goods_url, "GoodsSearchResponse", $"goods_list is null\nbody:{body}", status.count, status.disabledNow);
  306. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  307. }
  308. bool isValidGoods = goodsList.Any(item => item.goods_id > 0 && !string.IsNullOrWhiteSpace(item.goods_name));
  309. if (!isValidGoods && goodsList.Count > 0)
  310. {
  311. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync.Empty").Info(urlWithQuery, body).SaveAsync();
  312. }
  313. ResetPrecheckExceptionState();
  314. return isValidGoods ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Empty;
  315. }
  316. catch (Exception ex)
  317. {
  318. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(goods_url, ex.ToString()).SaveAsync();
  319. var status = RegisterPrecheckException();
  320. NotifyPrecheckException(goods_url, "Exception", ex.ToString(), status.count, status.disabledNow);
  321. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  322. }
  323. // client_id:c8823690b47842649e4fee054317e0c1
  324. //client_secret:0b5eeab761a6b61df59f11296387b95523a26b09
  325. //https://gw-api.pinduoduo.com/api/router
  326. //{
  327. // "type": "pdd.ddk.goods.search",
  328. // "data_type": "JSON",
  329. // "client_id": "c8823690b47842649e4fee054317e0c1",
  330. // "keyword": "https://mobile.yangkeduo.com/goods.html?ps=ROyjHRQJ7d",
  331. // "pid": "13585632_187155840",
  332. // "timestamp": 1779396871,
  333. // "sign": "3DE407843BECA8BA480E8BA4C51E6049"
  334. //}
  335. // 这样可以用字符串链接去查询是否多多商品。返回goods_list里面有内容就是可以去转链的。
  336. //下面这个就是不能转链的
  337. //{
  338. // "goods_search_response": {
  339. // "goods_list": [{
  340. // "subsidy_goods_type": 0,
  341. // "subsidy_list": [],
  342. // "platform_discount_list": [],
  343. // "goods_id": 922492775451,
  344. // "goods_sign": "E9L2dKKUUo9LeKDhwePAPXLcFjD6hB3MKw_JGaDyX4DX"
  345. // }],
  346. // "total_count": 1,
  347. // "request_id": "17793969978830062"
  348. // }
  349. // }
  350. }
  351. public async Task<PddDataDTO> PddParseAsync(string content, int commerceType, PddDataDTO result, CancellationToken cancellationToken = default)
  352. {
  353. string message = "OK";
  354. string url = GetLink(content);
  355. if (string.IsNullOrEmpty(url))
  356. {
  357. result.success = false;
  358. result.link_type = LinkTypeEnum.unknown;
  359. result.channel_type = ChannelTypeEnum.pdd;
  360. result.message = "放弃转链";
  361. result.reason = "无效链接";
  362. result.accountId = 0;
  363. result.accountName = string.Empty;
  364. result.content = content;
  365. return result;
  366. }
  367. string shortLinkurl = url;
  368. if (result.link_type == LinkTypeEnum.other_aff)
  369. {
  370. result.success = false;
  371. result.link_type = LinkTypeEnum.unknown;
  372. result.channel_type = ChannelTypeEnum.pdd;
  373. result.deeplink_url = GetDeeplink(url);
  374. result.message = "放弃转链";
  375. result.reason = "其他推广链接";
  376. result.accountId = 0;
  377. result.accountName = string.Empty;
  378. result.content = content;
  379. return result;
  380. }
  381. // 移到外围了
  382. //if (!TestParseCore.InWhitelist(result.ip, result.oaid) && FlowControlIgnoreRequest(_config, out string reason))
  383. //{
  384. // result.success = false;
  385. // result.link_type = LinkTypeEnum.unknown;
  386. // result.channel_type = ChannelTypeEnum.pdd;
  387. // result.message = "放弃转链";
  388. // result.reason = reason;
  389. // result.accountId = 0;
  390. // result.accountName = string.Empty;
  391. // result.content = content;
  392. // result.deeplink_url = GetDeeplink(result.shortLinkurl);
  393. // return result;
  394. //}
  395. var isTrackUrl = IsTrackUrl(url);
  396. //不是跟踪链接 ps=*** 就返回
  397. //2024-09-26 放开数字id转链
  398. //if (!isTrackUrl)
  399. //{
  400. // result.success = false;
  401. // result.link_type = LinkTypeEnum.unknown;
  402. // result.channel_type = ChannelTypeEnum.pdd;
  403. // result.message = "放弃转链";
  404. // result.reason = "数字id链接";
  405. // result.accountId = 0;
  406. // result.accountName = string.Empty;
  407. // result.content = content;
  408. // return result;
  409. //}
  410. //去追踪
  411. Stopwatch stopwatch = Stopwatch.StartNew();
  412. stopwatch.Start();
  413. var redirectedUrl = await GetRedirectedUrlAsync(url, cancellationToken);
  414. stopwatch.Stop();
  415. bool redirected = !string.IsNullOrWhiteSpace(redirectedUrl) &&
  416. !string.Equals(redirectedUrl, url, StringComparison.OrdinalIgnoreCase);
  417. if (redirected)
  418. {
  419. // 字符串ID转数字ID
  420. result.rawContent2 = redirectedUrl;
  421. url = redirectedUrl;
  422. result.elapsedTime2 = (int)stopwatch.ElapsedMilliseconds;
  423. }
  424. switch (commerceType)
  425. {
  426. case 1:
  427. case 2:
  428. result.success = false;
  429. result.message = "放弃转链";
  430. result.reason = $"厂商放弃{commerceType}";
  431. return result;
  432. }
  433. Stopwatch stopwatch2 = Stopwatch.StartNew();
  434. stopwatch2.Start();
  435. result = _account.work_mode switch
  436. {
  437. PddUnionWorkMode.SiteApi => await UnionGeneratelink(result, url, cancellationToken),
  438. //case PddUnionWorkMode.AffApi:
  439. // result = await GetPromotionByAffAsync(result, url, cancellationToken);
  440. // break;
  441. _ => await transferUrl(result, url, cancellationToken),
  442. };
  443. if (!result.success)
  444. {
  445. result.shortLinkurl = shortLinkurl;
  446. result.deeplink_url = GetDeeplink(shortLinkurl);
  447. }
  448. stopwatch2.Stop();
  449. result.elapsedTime3 = (int)stopwatch2.ElapsedMilliseconds;
  450. return result;
  451. }
  452. }
  453. }