Arzon.cs 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404
  1. using System;
  2. using System.Collections.Generic;
  3. using System.Linq;
  4. using System.Text;
  5. using AVSORTER;
  6. using System.Text.RegularExpressions;
  7. using HtmlAgilityPack;
  8. using System.IO;
  9. using System.Runtime.Serialization.Formatters.Binary;
  10. namespace Gets
  11. {
  12. public class Arzon : AVSORTER.IGetable, ICloneable
  13. {
  14. Uri uri;
  15. MyWebClient wc;
  16. /// <summary>
  17. /// 构造函数
  18. /// </summary>
  19. /// <param name="IsInit">决定是否初始化Cookie,初始化Cookie需要耗费时间并且需要联网,或者根据需要手动调用CookiesInit()方法</param>
  20. public Arzon(bool IsInit)
  21. {
  22. wc = new MyWebClient();
  23. if (IsInit)
  24. {
  25. CookiesInit();
  26. }
  27. }
  28. public void CookiesInit()
  29. {
  30. const string url = @"http://www.arzon.jp/index.php?action=adult_customer_agecheck&agecheck=1";//&redirect=http%3A%2F%2Fwww.arzon.jp%2F";
  31. uri = new Uri(url);
  32. wc.DownloadDataCompleted += wc_DownloadDataCompleted;
  33. wc.DownloadDataAsync(uri, "init");
  34. }
  35. public event EventHandler InitCompleted;
  36. void wc_DownloadDataCompleted(object sender, System.Net.DownloadDataCompletedEventArgs e)
  37. {
  38. if ((e.UserState as string) == "init")
  39. {
  40. this.IsInitCompleted = true;
  41. if (this.InitCompleted!=null)
  42. {
  43. this.InitCompleted(this, new EventArgs());
  44. }
  45. }
  46. }
  47. public bool IsInitCompleted { get; set; }
  48. public List<MovieBasic> Query(string fcode)
  49. {
  50. Uri u = urlCombine(fcode);
  51. return PageParse(u);
  52. }
  53. public List<MovieBasic> PageParse(Uri u)
  54. {
  55. HtmlAgilityPack.HtmlDocument doc = new HtmlDocument();
  56. doc.LoadHtml(wc.GetHTML(u));
  57. //doc.DocumentNode.SelectNodes("//li[@class='saledate']/span")[3].InnerText
  58. var listitems = doc.DocumentNode.SelectNodes("//div[@id='listitem']/table/tr/td/div[@class='data']/ul[1]/li[2]");
  59. var htmnode = doc.DocumentNode.SelectSingleNode("//div[@class='autopagerize_page_element']");
  60. if (htmnode == null)
  61. {
  62. //查无此片
  63. return new List<MovieBasic>();
  64. }
  65. string htm = htmnode.InnerHtml;
  66. List<MovieBasic> l = new List<MovieBasic>();
  67. //<div class="autopagerize_page_element">
  68. int cou = doc.DocumentNode.SelectNodes("//div[@id='itemd']").Count;
  69. for (int i = 0; i < cou; i++)
  70. {
  71. string n_title = doc.DocumentNode.SelectNodes("//div[@id='itemd']")[i].ChildNodes["ul"].ChildNodes["li"].ChildNodes["h3"].InnerText;
  72. n_title = Tools.RemoveInvalidChars(n_title);
  73. string n_itemURL = "http://" + u.Host + doc.DocumentNode.SelectNodes("//div[@id='itemd']")[i].ChildNodes["ul"].ChildNodes["li"].ChildNodes["h3"].ChildNodes["a"].Attributes["href"].Value;
  74. string n_date = doc.DocumentNode.SelectSingleNode("//div[@id='itemd']/ul/li/span[@class='date']").InnerText;
  75. var datanode = doc.DocumentNode.SelectNodes("//div[@class='data']")[i];
  76. HtmlDocument docData = new HtmlDocument();
  77. docData.LoadHtml(datanode.InnerHtml);
  78. int k = docData.DocumentNode.SelectNodes("//ul[1]/li").Count;
  79. string n_actor = string.Empty;
  80. List<string> actors = new List<string>();
  81. if (k > 1)
  82. {
  83. for (int j = k; j > 1; j--)
  84. {
  85. n_actor = docData.DocumentNode.SelectNodes("//ul[1]/li[" + j + "]")[0].InnerText.Trim();
  86. actors.Add(n_actor);
  87. }
  88. }
  89. string n_maker = docData.DocumentNode.SelectSingleNode("//ul[2]/li[2]").InnerText.Trim();
  90. string n_company = docData.DocumentNode.SelectSingleNode("//ul[3]/li[2]").InnerText.Trim();
  91. var n_ticai_node = docData.DocumentNode.SelectSingleNode("//ul[4]/li[2]");
  92. string n_ticai;
  93. if (n_ticai_node == null)
  94. {
  95. n_ticai = string.Empty;
  96. }
  97. else
  98. {
  99. n_ticai = n_ticai_node.InnerText.Trim();
  100. }
  101. string s_img = doc.DocumentNode.SelectNodes("//div[@id='listitem']/table/tr/td/div/a/img")[i].Attributes["src"].Value;
  102. MovieBasic mb = new MovieBasic()
  103. {
  104. Title = n_title,
  105. ItemURL = n_itemURL,
  106. Img_s = s_img,
  107. Actor = actors,
  108. Maker = n_company,
  109. Label = n_ticai
  110. };
  111. l.Add(mb);
  112. }
  113. return l;
  114. }
  115. /// <summary>
  116. /// this is the old version.
  117. /// </summary>
  118. /// <param name="basic"></param>
  119. /// <returns></returns>
  120. //public Movie GetMovie(MovieBasic basic)
  121. //{
  122. // try
  123. // {
  124. // string html = wc.GetHTML(new Uri(basic.ItemURL));
  125. // var docc = new HtmlDocument();
  126. // docc.LoadHtml(html);
  127. // //web DOM Changed 2014/12/7
  128. // string Title = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/div[@class='detail_title']/h1").InnerText.Trim();
  129. // //string Title = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/div[@class='detail_title_new']/h1").InnerHtml;
  130. // Title = Tools.RemoveInvalidChars(Title);
  131. // while (Title.IndexOf("&nbsp;") != -1)
  132. // {
  133. // Title= Title.Replace("&nbsp;", "");
  134. // }
  135. // while (Title.IndexOf("廃盤") != -1)
  136. // {
  137. // Title = Title.Replace("廃盤", "");
  138. // }
  139. // //var ddd ="[MIDD-983]Baby Entertainment×MOODYZコラボ作品 淫神の女泥棒 哀しき痙攣の追憶 Dear.F 1 恥辱的、屈辱的なイカせの拷問! 反反复复反反复复方法";
  140. // if (Title.Length>82)
  141. // {
  142. // Title = Title.Substring(0, 81);
  143. // }
  144. // //string Title = basic.Title;
  145. // string label = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[3]/td/a").InnerText.Trim();
  146. // string changjia = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[2]/td/a").InnerText.Trim();
  147. // string jiandu = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[5]/td[2]").InnerText.Trim();
  148. // string date = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[6]/td[2]").InnerText.Trim();
  149. // //2008/01/25 (DVD レンタル版)
  150. // System.Text.RegularExpressions.Regex r = new System.Text.RegularExpressions.Regex(@"\d{4}/\d{2}/\d{2}");
  151. // if (r.IsMatch(date))
  152. // {
  153. // date = r.Match(date).Value;
  154. // }
  155. // else
  156. // {
  157. // date = "1900/01/01";
  158. // }
  159. // DateTime dtime = DateTime.Parse(date);
  160. // string minutes = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[7]/td[2]").InnerText.Trim();
  161. // string f_code = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[8]/td[2]").InnerText.Trim();
  162. // while (f_code.IndexOf("&nbsp;")!=-1)
  163. // {
  164. // f_code = f_code.Replace("&nbsp;", "");
  165. // }
  166. // while (f_code.IndexOf("廃盤")!=-1)
  167. // {
  168. // f_code = f_code.Replace("廃盤", "");
  169. // }
  170. // f_code = Tools.Fcode(f_code);
  171. // string xilie = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/table/tr/td/table[@class='item_detail']/tr/td[@class='caption']/table[@class='item']/tr[4]/td[2]").InnerText.Trim();
  172. // //f_code = Tools.Fcode(f_code);
  173. // string intro = docc.DocumentNode.SelectSingleNode("//table[@class='item_detail']/tr/td[@class='text']").InnerText.Trim();
  174. // string coverImg = docc.DocumentNode.SelectSingleNode("//table[@class='item_detail']/tr/td/div/a").Attributes["href"].Value.Trim();
  175. // Movie m = new Movie()
  176. // {
  177. // Actor = basic.Actor,
  178. // Title = Title,
  179. // Lable = label,
  180. // Maker = changjia,
  181. // ReleaseDate = dtime,
  182. // Minutes = minutes,
  183. // AVCode = f_code,
  184. // Introduction = intro,
  185. // CoverURL = coverImg,
  186. // ItemURL = basic.ItemURL,
  187. // Series = xilie,
  188. // Producer = jiandu
  189. // };
  190. // //Console.WriteLine(string.Format("Actor:{0}\r\nTitle:{1}\r\nLabel:{2}\r\nMaker:{3}\r\nReleaseDate:{4}\r\n番号:{5}\r\n ", m.Actor[0], m.Title, m.Lable, m.Maker, m.ReleaseDate.ToShortDateString(), m.AVCode));
  191. // return m;
  192. // }
  193. // catch (Exception)
  194. // {
  195. // throw new Exception("获取影片信息失败!");
  196. // }
  197. //}
  198. public Movie GetMovie(MovieBasic basic)
  199. {
  200. try
  201. {
  202. string html = wc.GetHTML(new Uri(basic.ItemURL));
  203. var docc = new HtmlDocument();
  204. docc.LoadHtml(html);
  205. //web DOM Changed 2014/12/7
  206. string Title = docc.DocumentNode.SelectSingleNode("//div[@id='detail_new']/div[@class='detail_title_new']/h1").InnerText.Trim();
  207. string coverImg = docc.DocumentNode.SelectSingleNode("//*[@id='detail_new']//table//table//a").Attributes["href"].Value.ToString().Trim();
  208. //AV女優:
  209. var item1 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[1]/td[2]").InnerText.Trim();
  210. //AVメーカー 制造厂
  211. var item2 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[2]/td[2]").InnerText.Trim();
  212. //AVレーベル 唱片公司
  213. var item3 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[3]/td[2]").InnerText.Trim();
  214. //シリーズ 系列
  215. var item4 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[4]/td[2]").InnerText.Trim();
  216. //監督 导演
  217. var item5 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[5]/td[2]").InnerText.Trim();
  218. //発売日:
  219. var item6 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[6]/td[2]").InnerText.Trim();
  220. //収録時間
  221. var item7 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[7]/td[2]").InnerText.Trim();
  222. //品番
  223. var item8 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[8]/td[2]").InnerText.Trim();
  224. //タグ: 标签
  225. var item9 = docc.DocumentNode.SelectSingleNode("//table[@class='item']/tr[9]/td[2]").InnerText.Trim();
  226. //作品紹介
  227. string descrition;
  228. try
  229. {
  230. descrition = docc.DocumentNode.SelectSingleNode("//div[@class='item_text']").InnerText.Trim();
  231. }
  232. catch (Exception)
  233. {
  234. descrition = "";
  235. }
  236. Movie m = new Movie()
  237. {
  238. Actor = basic.Actor,
  239. Title = titleCleaner(Title),
  240. Lable = item9,
  241. Maker = item2,//ideapocket
  242. ReleaseDate = dateCleaner(item6),
  243. Minutes = item7,
  244. AVCode = f_codeCleaner(item8),
  245. Introduction = descrition,
  246. CoverURL = coverImg,
  247. ItemURL = basic.ItemURL,
  248. Series = item4,
  249. //导演
  250. Producer = item5
  251. };
  252. //Console.WriteLine(string.Format("Actor:{0}\r\nTitle:{1}\r\nLabel:{2}\r\nMaker:{3}\r\nReleaseDate:{4}\r\n番号:{5}\r\n ", m.Actor[0], m.Title, m.Lable, m.Maker, m.ReleaseDate.ToShortDateString(), m.AVCode));
  253. return m;
  254. }
  255. catch (Exception)
  256. {
  257. throw new Exception("获取影片信息失败!");
  258. }
  259. }
  260. //标题除杂
  261. private string titleCleaner(string title)
  262. {
  263. title = Tools.RemoveInvalidChars(title);
  264. while (title.IndexOf("&nbsp;") != -1)
  265. {
  266. title = title.Replace("&nbsp;", "");
  267. }
  268. while (title.IndexOf("廃盤") != -1)
  269. {
  270. title = title.Replace("廃盤", "");
  271. }
  272. //var ddd ="[MIDD-983]Baby Entertainment×MOODYZコラボ作品 淫神の女泥棒 哀しき痙攣の追憶 Dear.F 1 恥辱的、屈辱的なイカせの拷問! 反反复复反反复复方法";
  273. if (title.Length > 82)
  274. {
  275. title = title.Substring(0, 81);
  276. }
  277. return title;
  278. }
  279. //日期
  280. private DateTime dateCleaner(string date)
  281. {
  282. System.Text.RegularExpressions.Regex r = new System.Text.RegularExpressions.Regex(@"\d{4}/\d{2}/\d{2}");
  283. if (r.IsMatch(date))
  284. {
  285. date = r.Match(date).Value;
  286. }
  287. else
  288. {
  289. date = "1900/01/01";
  290. }
  291. return DateTime.Parse(date);
  292. }
  293. //番号
  294. private string f_codeCleaner(string f_code)
  295. {
  296. while (f_code.IndexOf("&nbsp;") != -1)
  297. {
  298. f_code = f_code.Replace("&nbsp;", "");
  299. }
  300. while (f_code.IndexOf("廃盤") != -1)
  301. {
  302. f_code = f_code.Replace("廃盤", "");
  303. }
  304. f_code = Tools.Fcode(f_code);
  305. return f_code;
  306. }
  307. public bool GetCover(Movie mo)
  308. {
  309. wc.ReferURL = mo.ItemURL;
  310. FileInfo f = new FileInfo(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "Cover", mo.AVCode + ".jpg"));
  311. try
  312. {
  313. if (!Directory.Exists(Path.GetDirectoryName(f.FullName)))
  314. {
  315. Directory.CreateDirectory(Path.GetDirectoryName(f.FullName));
  316. }
  317. if (!File.Exists(f.FullName))
  318. {
  319. wc.DownloadFile(mo.CoverURL, f.FullName);
  320. }
  321. else
  322. {
  323. Console.WriteLine("已有封面 " + mo.Title);
  324. }
  325. }
  326. catch (Exception err)
  327. {
  328. return false;
  329. }
  330. mo.CoverFile = f.FullName;
  331. return true;
  332. }
  333. Uri urlCombine(string fcode)
  334. {
  335. string ur = "http://www.arzon.jp/itemlist.html?t=&m=all&s=&mkt=all&disp=30&sort=-saledate&list=list&q=" + fcode;
  336. return new Uri(ur);
  337. }
  338. public object Clone()
  339. {
  340. //MemoryStream ms = new MemoryStream();
  341. //BinaryFormatter bf = new BinaryFormatter();
  342. //bf.Serialize(ms, this);
  343. //ms.Seek(0, 0);
  344. //object value = bf.Deserialize(ms);
  345. //ms.Close();
  346. //return value;//this.MemberwiseClone();
  347. MemoryStream ms = new MemoryStream();
  348. BinaryFormatter bf = new BinaryFormatter();
  349. bf.Serialize(ms, this.wc.m_container);
  350. ms.Seek(0, 0);
  351. object value = bf.Deserialize(ms);
  352. Arzon ar = new Arzon(false);
  353. ar.wc.m_container = value as System.Net.CookieContainer;
  354. return ar;
  355. }
  356. }
  357. }