public static void GetForumsList(BufferBlock <string> imageTargetBlock) { var block = new BufferBlock <string>(); var task = GetNewsDetail.GenerateForumDetail(block, imageTargetBlock); int errorTime = 0; var firstPage = "http://www.ihchina.cn/luntan/p/1.html"; var lastPageNumber = DebugHelperTools.IsDebugMode() ? 2 : WebpageHelper.GetPageLastIndex(firstPage); for (int i = 1; i < lastPageNumber && errorTime < 10; i++) { var listUrl = string.Format("http://www.ihchina.cn/luntan/p/{0}.html", i); Console.WriteLine("starting process page:{0}", listUrl); var doc = WebpageHelper.GetHttpRequestDocument(listUrl); var listNodes = doc.DocumentNode.SelectNodes("//div[@id='datalist']/div[@class='list-item']"); if (listNodes == null) { errorTime++; continue; } List <BsonDocument> result = new List <BsonDocument>(); foreach (var node in listNodes) { if (errorTime == 10) { break; } var bson = WebpageHelper.AnalizeGeneralListInformation(node, MongodbChecker.CheckForumsListExist, imageTargetBlock); if (bson == null) { errorTime++; Console.WriteLine("duplicated url: page {0}", i); continue; } if (bson != null) { var link = bson.GetElement("link").Value.ToString(); block.Post(link); result.Add(bson); } //每10条进行一次数据库插入,减少内存负担 if (result.Count == 10) { MongodbSaver.SaveForumsList(result); result.Clear(); } } if (result.Count > 0) { MongodbSaver.SaveForumsList(result); } } block.Complete(); task.Wait(); }
private static async Task GetPeoplePageList(BufferBlock <string> imageTargetBlock) { var block = new BufferBlock <string>(); var task = GetNewsDetail.GeneratePeopleDetail(block, imageTargetBlock); int errorTime = 0; var firstPage = "http://www.ihchina.cn/character/p/1.html"; var lastPageNumber = DebugHelperTools.IsDebugMode() ? 1 : WebpageHelper.GetPageLastIndex(firstPage); for (int i = 1; i <= lastPageNumber && errorTime < 10; i++) { var listUrl = string.Format("http://www.ihchina.cn/character/p/{0}.html", i); Console.WriteLine("starting process people page: {0}", listUrl); var doc = WebpageHelper.GetHttpRequestDocument(listUrl); var listNodes = doc.DocumentNode.SelectNodes("//div[@class='list-item']"); if (listNodes == null) { errorTime++; continue; } List <BsonDocument> result = new List <BsonDocument>(); foreach (var node in listNodes) { if (errorTime == 10) { break; } var bson = WebpageHelper.AnalizeGeneralListInformation(node, MongodbChecker.CheckPeoplePageListExist, imageTargetBlock); if (bson == null) { errorTime++; Console.WriteLine("duplicated people url page{0}", i); continue; } if (bson != null) { var link = bson.GetElement("link").Value.ToString(); block.Post(link); result.Add(bson); } } if (result.Count > 0) { MongodbSaver.SavePeopleListInformation(result); } result.Clear(); } block.Complete(); await task; }
public static void Main(string[] args) { if (!DebugHelperTools.IsDebugMode()) { Task.Run(async() => { await GetIhChina.StartReptile(); }); System.Timers.Timer timer = new System.Timers.Timer(); timer.Enabled = true; timer.Interval = TimeSpan.FromDays(1).TotalMilliseconds; timer.Start(); timer.Elapsed += (o, e) => { Task.Run(async() => { await GetIhChina.StartReptile(); }); }; } CreateWebHostBuilder(args).Build().Run(); }
public static async Task <int> GetNewsList(Task <int> imageSaverTask, BufferBlock <string> imageTargetBlock) { BufferBlock <string> newsDetailTargetBlock = new BufferBlock <string>(); int errorTime = 0; var newsDetailPageGenerate = GetNewsDetail.GenerateNewsDetail(newsDetailTargetBlock, imageTargetBlock); int pageNumber = DebugHelperTools.IsDebugMode() ? 2 : 255; for (int page = 1; page < pageNumber; page++) { if (errorTime > 10) { Console.WriteLine("reach the limitation of error time"); break; } Console.WriteLine("starting process {0} page", page); string pageURL = String.Format("{0}?category_id=9&page={1}&limit=0", GetIhChina.NewsListUrl, page); var result = WebpageHelper.GetRequest(pageURL); var jsonObject = JsonConvert.DeserializeObject <NewsListResponse>(result); if (jsonObject.more != 1 && String.IsNullOrEmpty(jsonObject.data)) { Console.WriteLine("GetNewsList: There is no more here, break from the loop"); break; } var doc = new HtmlDocument(); doc.LoadHtml(jsonObject.data); var newsListNodes = from links in doc.DocumentNode.Descendants() where links.Name == "div" && links.Attributes["class"] != null && links.Attributes["class"].Value == "list-item" select links; var newslistBsons = new List <BsonDocument>(); foreach (var node in newsListNodes) { var newsBson = new BsonDocument(); var titleNode = node.SelectSingleNode(".//div[@class='h16']/a"); if (titleNode != null) { var link = titleNode.Attributes["href"].Value; if (MongodbChecker.CheckNewsExist(link)) { Console.WriteLine("duplicated url: page {0}", link); errorTime++; continue; } newsDetailTargetBlock.Post(link); newsBson.Add("link", link); newsBson.Add("title", titleNode.Attributes["title"].Value); } var imgNode = node.SelectSingleNode(".//img"); if (imgNode != null) { var imgUrl = WebpageHelper.GetSubUrl(imgNode.Attributes["src"].Value); newsBson.Add("img", imgUrl); newsBson.Add("compressImg", WebImageSaver.Instance.GetComressImageName(imgUrl)); imageTargetBlock.Post(GetIhChina.MainPage + imgNode.Attributes["src"].Value); } var dataNode = node.SelectSingleNode(".//div[@class='date']/div"); if (dataNode != null) { newsBson.Add("date", dataNode.InnerText); } var contentNode = node.SelectSingleNode(".//div[@class='p']"); if (contentNode != null) { newsBson.Add("content", contentNode.InnerText); } if (newsBson.Count() != 0) { newslistBsons.Add(newsBson); } //每10条进行一次数据库插入,减少内存负担 if (newslistBsons.Count == 10) { MongodbSaver.SaveNewsList(newslistBsons); for (int i = 0; i < 10; i++) { newslistBsons[i] = null; } newslistBsons.Clear(); } } if (newslistBsons.Count != 0) { MongodbSaver.SaveNewsList(newslistBsons); } } newsDetailTargetBlock.Complete(); return(await newsDetailPageGenerate); }