PddUnionPlus.cs 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502
  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. private static (int count, bool disabledNow) RegisterPrecheckException()
  89. {
  90. lock (_precheckCircuitLock)
  91. {
  92. _precheckConsecutiveExceptionCount++;
  93. bool disabledNow = false;
  94. if (!_precheckDisabled && _precheckConsecutiveExceptionCount >= _precheckDisableErrorThreshold)
  95. {
  96. _precheckDisabled = true;
  97. disabledNow = true;
  98. }
  99. return (_precheckConsecutiveExceptionCount, disabledNow);
  100. }
  101. }
  102. private static void NotifyPrecheckException(string goodsUrl, string stage, string detail, int count, bool disabledNow)
  103. {
  104. string message = $"【PDD商品Precheck异常】{_end_point}\n" +
  105. $"url:{goodsUrl}\n" +
  106. $"stage:{stage}\n" +
  107. $"连续异常:{count}/{_precheckDisableErrorThreshold}\n" +
  108. $"{detail}";
  109. if (disabledNow)
  110. {
  111. message += "\n状态: 已临时禁用Precheck,后续默认返回Success";
  112. }
  113. NotifyCore.Notify(new NifyMessage
  114. {
  115. message = message,
  116. priority = NifyMessagePriority.high,
  117. tags = ["red_circle"]
  118. });
  119. }
  120. private string ExtractGoodsIdUrl(string url)
  121. {
  122. var uri = new Uri(url);
  123. var queryParams = HttpUtility.ParseQueryString(uri.Query);
  124. string goodsId = queryParams["goods_id"];
  125. if (!string.IsNullOrEmpty(goodsId) && uri.AbsolutePath.StartsWith("/goods"))
  126. {
  127. // 保持域名和路径不变,仅保留 goods_id 参数
  128. var baseUrl = $"{uri.Scheme}://{uri.Host}{uri.AbsolutePath}";
  129. return $"{baseUrl}?goods_id={goodsId}";
  130. }
  131. return url; // 如果没有找到 goods_id 参数,则返回原始 URL
  132. }
  133. private bool IsTrackUrl(string sourceUrl)
  134. {
  135. string pattern = @"^https?://mobile\.yangkeduo\.com/goods(\d*)\.html\?ps=.*$";
  136. Regex regex = new Regex(pattern);
  137. return regex.IsMatch(sourceUrl);
  138. }
  139. /// <summary>
  140. /// bool:是否ps链接,string 处理后的链接
  141. /// </summary>
  142. /// <param name="sourceUrl"></param>
  143. /// <param name="cancellationToken"></param>
  144. /// <returns></returns>
  145. public async Task<string> GetRedirectedUrlAsync(string sourceUrl, CancellationToken cancellationToken = default)
  146. {
  147. var isTrackUrl = IsTrackUrl(sourceUrl);
  148. if (!isTrackUrl) return ExtractGoodsIdUrl(sourceUrl);
  149. string redirectedUrl = sourceUrl;
  150. try
  151. {
  152. string useragent = ProviderFakeUserAgent.RandomMobile;
  153. WebClientUtility client = new()
  154. {
  155. AllowAutoRedirect = false,
  156. UserAgent = useragent,
  157. };
  158. client.Proxy = _proxy;
  159. #if DEBUG
  160. client.Proxy = null;
  161. #endif
  162. var response = await client.RequestAsync(sourceUrl, "GET", cancellationToken);
  163. if (!response.Successed) return null;
  164. if (response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.RedirectKeepVerb ||
  165. response.ResponseMessage.StatusCode == System.Net.HttpStatusCode.TemporaryRedirect)
  166. {
  167. redirectedUrl = response.ResponseMessage.Headers.Location.ToString();
  168. redirectedUrl = ExtractGoodsIdUrl(redirectedUrl);
  169. return redirectedUrl;
  170. }
  171. }
  172. catch (Exception ex)
  173. {
  174. }
  175. return sourceUrl;
  176. }
  177. public static PddDataDTO GetFormattedObject(UnionParseRequest request)
  178. {
  179. switch (request.RiskStrategy)
  180. {
  181. case "tbpush":
  182. if (AlimamaPlus.IsDigitsOnly(request.Content))
  183. {
  184. request.Content = $"https://mobile.yangkeduo.com/goods.html?goods_id={request.Content}";
  185. }
  186. break;
  187. }
  188. string shortLinkurl = GetLink(request.Content);
  189. string deeplink_url = GetDeeplink(shortLinkurl);
  190. var link_type = LinkTypeEnum.unknown;
  191. if (!string.IsNullOrEmpty(shortLinkurl) && shortLinkurl.StartsWith("https://p.pinduoduo.com/"))
  192. {
  193. link_type = LinkTypeEnum.other_aff;
  194. }
  195. return new PddDataDTO()
  196. {
  197. message = string.Empty,
  198. link_type = link_type,
  199. channel = TkChannelEnum.pdd,
  200. accountName = string.Empty,
  201. success = false,
  202. content = request.Content,
  203. ip = request.Ip,
  204. oaid = request.Oaid,
  205. elapsedTime = 0,
  206. itemName = "点击打开拼多多APP",
  207. shortLinkurl = shortLinkurl,
  208. deeplink_url = string.Empty,
  209. create_time = DateTime.Now,
  210. end_point = _end_point,
  211. parse_type = request.Type,
  212. riskStrategy = request.RiskStrategy,
  213. launchScene = request.LaunchScene,
  214. };
  215. }
  216. public static async Task<RrecheckGoodsIdResult> PrecheckGoodsIdAsync(string content, CancellationToken cancellationToken = default)
  217. {
  218. string url = GetLink(content);
  219. if (string.IsNullOrEmpty(url))
  220. {
  221. return RrecheckGoodsIdResult.InvalidLink;
  222. }
  223. if (!TryGetPrecheckPsGoodsUrl(url, out string goods_url))
  224. {
  225. return RrecheckGoodsIdResult.Success;
  226. }
  227. if (IsPrecheckDisabled()) return RrecheckGoodsIdResult.Success;
  228. if (string.IsNullOrWhiteSpace(goods_url)) return RrecheckGoodsIdResult.Empty;
  229. try
  230. {
  231. const string precheckClientId = "c8823690b47842649e4fee054317e0c1";
  232. const string precheckClientSecret = "0b5eeab761a6b61df59f11296387b95523a26b09";
  233. const string precheckPid = "13585632_187155840";
  234. var precheckAccount = new PddPoolDTO
  235. {
  236. app_key = precheckClientId,
  237. app_secret = precheckClientSecret
  238. };
  239. var ts = DateTime.Now.Convert2UnixTimestamp(true);
  240. var args = new Dictionary<string, string>
  241. {
  242. { "type", "pdd.ddk.goods.search" },
  243. { "data_type", "JSON" },
  244. { "client_id", precheckClientId },
  245. { "keyword", goods_url.UrlDecode() },
  246. { "pid", precheckPid },
  247. { "timestamp", ts.ToString() }
  248. };
  249. args = GenerateSignature(precheckAccount, args);
  250. var query = HttpUtility.ParseQueryString(string.Empty);
  251. foreach (var kvp in args) query[kvp.Key] = kvp.Value;
  252. string urlWithQuery = $"{_baseUrl}?{query}";
  253. WebClientUtility client = new();
  254. client.SetContentType("application/json");
  255. var response = await client.RequestAsync(urlWithQuery, "GET", cancellationToken);
  256. string body = response.Body();
  257. if (!response.Successed)
  258. {
  259. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  260. var status = RegisterPrecheckException();
  261. NotifyPrecheckException(goods_url, "RequestAsync", $"response.Successed=false\nbody:{body}", status.count, status.disabledNow);
  262. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  263. }
  264. var root = body.Convert2Object<GoodsSearchDTO>();
  265. if (root?.error_response != null)
  266. {
  267. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  268. var status = RegisterPrecheckException();
  269. var error = root.error_response;
  270. NotifyPrecheckException(
  271. goods_url,
  272. "ErrorResponse",
  273. $"error_code:{error.error_code}\nsub_code:{error.sub_code}\nerror_msg:{error.error_msg}\nsub_msg:{error.sub_msg}\nbody:{body}",
  274. status.count,
  275. status.disabledNow);
  276. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  277. }
  278. var goodsList = root?.goods_search_response?.goods_list;
  279. if (goodsList == null)
  280. {
  281. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(urlWithQuery, body).SaveAsync();
  282. var status = RegisterPrecheckException();
  283. NotifyPrecheckException(goods_url, "GoodsSearchResponse", $"goods_list is null\nbody:{body}", status.count, status.disabledNow);
  284. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  285. }
  286. bool isValidGoods = goodsList.Any(item => item.goods_id > 0 && !string.IsNullOrWhiteSpace(item.goods_name));
  287. if (!isValidGoods && goodsList.Count > 0)
  288. {
  289. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync.Empty").Info(urlWithQuery, body).SaveAsync();
  290. }
  291. ResetPrecheckExceptionState();
  292. return isValidGoods ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Empty;
  293. }
  294. catch (Exception ex)
  295. {
  296. _ = new LoggerLibrary("PddUnion", "PrecheckGoodsIdAsync").Info(goods_url, ex.ToString()).SaveAsync();
  297. var status = RegisterPrecheckException();
  298. NotifyPrecheckException(goods_url, "Exception", ex.ToString(), status.count, status.disabledNow);
  299. return status.disabledNow ? RrecheckGoodsIdResult.Success : RrecheckGoodsIdResult.Exception;
  300. }
  301. // client_id:c8823690b47842649e4fee054317e0c1
  302. //client_secret:0b5eeab761a6b61df59f11296387b95523a26b09
  303. //https://gw-api.pinduoduo.com/api/router
  304. //{
  305. // "type": "pdd.ddk.goods.search",
  306. // "data_type": "JSON",
  307. // "client_id": "c8823690b47842649e4fee054317e0c1",
  308. // "keyword": "https://mobile.yangkeduo.com/goods.html?ps=ROyjHRQJ7d",
  309. // "pid": "13585632_187155840",
  310. // "timestamp": 1779396871,
  311. // "sign": "3DE407843BECA8BA480E8BA4C51E6049"
  312. //}
  313. // 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。
  314. //下面这个就是不能转链的
  315. //{
  316. // "goods_search_response": {
  317. // "goods_list": [{
  318. // "subsidy_goods_type": 0,
  319. // "subsidy_list": [],
  320. // "platform_discount_list": [],
  321. // "goods_id": 922492775451,
  322. // "goods_sign": "E9L2dKKUUo9LeKDhwePAPXLcFjD6hB3MKw_JGaDyX4DX"
  323. // }],
  324. // "total_count": 1,
  325. // "request_id": "17793969978830062"
  326. // }
  327. // }
  328. }
  329. public async Task<PddDataDTO> PddParseAsync(string content, int commerceType, PddDataDTO result, CancellationToken cancellationToken = default)
  330. {
  331. string message = "OK";
  332. string url = GetLink(content);
  333. if (string.IsNullOrEmpty(url))
  334. {
  335. result.success = false;
  336. result.link_type = LinkTypeEnum.unknown;
  337. result.channel_type = ChannelTypeEnum.pdd;
  338. result.message = "放弃转链";
  339. result.reason = "无效链接";
  340. result.accountId = 0;
  341. result.accountName = string.Empty;
  342. result.content = content;
  343. return result;
  344. }
  345. string shortLinkurl = url;
  346. if (result.link_type == LinkTypeEnum.other_aff)
  347. {
  348. result.success = false;
  349. result.link_type = LinkTypeEnum.unknown;
  350. result.channel_type = ChannelTypeEnum.pdd;
  351. result.deeplink_url = GetDeeplink(url);
  352. result.message = "放弃转链";
  353. result.reason = "其他推广链接";
  354. result.accountId = 0;
  355. result.accountName = string.Empty;
  356. result.content = content;
  357. return result;
  358. }
  359. // 移到外围了
  360. //if (!TestParseCore.InWhitelist(result.ip, result.oaid) && FlowControlIgnoreRequest(_config, out string reason))
  361. //{
  362. // result.success = false;
  363. // result.link_type = LinkTypeEnum.unknown;
  364. // result.channel_type = ChannelTypeEnum.pdd;
  365. // result.message = "放弃转链";
  366. // result.reason = reason;
  367. // result.accountId = 0;
  368. // result.accountName = string.Empty;
  369. // result.content = content;
  370. // result.deeplink_url = GetDeeplink(result.shortLinkurl);
  371. // return result;
  372. //}
  373. var isTrackUrl = IsTrackUrl(url);
  374. //不是跟踪链接 ps=*** 就返回
  375. //2024-09-26 放开数字id转链
  376. //if (!isTrackUrl)
  377. //{
  378. // result.success = false;
  379. // result.link_type = LinkTypeEnum.unknown;
  380. // result.channel_type = ChannelTypeEnum.pdd;
  381. // result.message = "放弃转链";
  382. // result.reason = "数字id链接";
  383. // result.accountId = 0;
  384. // result.accountName = string.Empty;
  385. // result.content = content;
  386. // return result;
  387. //}
  388. //去追踪
  389. Stopwatch stopwatch = Stopwatch.StartNew();
  390. stopwatch.Start();
  391. var redirectedUrl = await GetRedirectedUrlAsync(url, cancellationToken);
  392. stopwatch.Stop();
  393. bool redirected = !string.IsNullOrWhiteSpace(redirectedUrl) &&
  394. !string.Equals(redirectedUrl, url, StringComparison.OrdinalIgnoreCase);
  395. if (redirected)
  396. {
  397. // 字符串ID转数字ID
  398. result.rawContent2 = redirectedUrl;
  399. url = redirectedUrl;
  400. result.elapsedTime2 = (int)stopwatch.ElapsedMilliseconds;
  401. }
  402. switch (commerceType)
  403. {
  404. case 1:
  405. case 2:
  406. result.success = false;
  407. result.message = "放弃转链";
  408. result.reason = $"厂商放弃{commerceType}";
  409. return result;
  410. }
  411. Stopwatch stopwatch2 = Stopwatch.StartNew();
  412. stopwatch2.Start();
  413. result = _account.work_mode switch
  414. {
  415. PddUnionWorkMode.SiteApi => await UnionGeneratelink(result, url, cancellationToken),
  416. //case PddUnionWorkMode.AffApi:
  417. // result = await GetPromotionByAffAsync(result, url, cancellationToken);
  418. // break;
  419. _ => await transferUrl(result, url, cancellationToken),
  420. };
  421. if (!result.success)
  422. {
  423. result.shortLinkurl = shortLinkurl;
  424. result.deeplink_url = GetDeeplink(shortLinkurl);
  425. }
  426. stopwatch2.Stop();
  427. result.elapsedTime3 = (int)stopwatch2.ElapsedMilliseconds;
  428. return result;
  429. }
  430. }
  431. }