Ejemplo n.º 1
0
        static void Main(string[] args)
        {
            for (char c = 'a'; c <= 'z'; ++c)
            {
                alph.Add(c);
            }
            alph.AddRange(Enumerable.Range(1, 9).Select(x => x.ToString()[0]).ToArray());

            HtmlReaderManager hrm = new HtmlReaderManager();
            LimitedConcurrencyLevelTaskScheduler lcts =
               new LimitedConcurrencyLevelTaskScheduler(200);
            TaskFactory factory = new TaskFactory(lcts);
            var tasks = new List<Task>();

            int counter = 1;

            foreach (string url in GetNextUrl())
            {

                Options opt = new Options();
                opt.Url = url;
                opt.counter = counter;

                tasks.Add(factory.StartNew((opts) =>
                                               {
                                                   bool isDownloaded = false;
                                                   int tryCount = 0;
                                                   do
                                                   {
                                                       Options options = (Options) opts;
                                                       string targetUrl = "http://prntscr.com/1" + options.Url;

                                                       try
                                                       {
                                                           hrm.Get(targetUrl);
                                                           string html = hrm.Html;
                                                           HtmlAgilityPack.HtmlDocument doc = new HtmlDocument();
                                                           doc.LoadHtml(html);
                                                           string imgUrl = doc.DocumentNode.SelectSingleNode("//meta[@name='twitter:image:src']").Attributes["content"].Value;
                                                           if (!string.IsNullOrEmpty(imgUrl))
                                                           {
                                                               ImageDownloader.DownloadRemoteImageFile(imgUrl, "images/" + Guid.NewGuid() + ".jpeg");
                                                               Console.WriteLine(options.counter + ". Скачиваем " + imgUrl);

                                                           }
                                                           else
                                                               Console.WriteLine(options.counter + ". " + imgUrl + " пуст");
                                                           break;
                                                       }
                                                       catch (Exception ex)
                                                       {
                                                           tryCount++;
                                                           if (ex.ToString().StartsWith("System.Net.WebException: The remote server returned an error: (503) Server Unavailable.") && tryCount<20)
                                                           {
                                                               Random s = new Random();
                                                               int randomVal = s.Next(1000, 10000);
                                                               Thread.Sleep(randomVal);
                                                               Console.WriteLine(options.counter + ". ждём " + targetUrl +" ("+tryCount+" раз)");
                                                           }
                                                           else
                                                           {
                                                               Console.WriteLine(options.counter + ". ошибка " + targetUrl);
                                                               isDownloaded = true;
                                                           }

                                                       }
                                                   } while (!isDownloaded);
                                               }, opt, TaskCreationOptions.LongRunning));

                if (counter%200 == 0)
                    Task.WaitAll(tasks.ToArray());
                tasks.Clear();
                counter++;
            }

            Console.ReadLine();
        }
Ejemplo n.º 2
0
        static void Main(string[] args)
        {
            for (char c = 'a'; c <= 'z'; ++c)
            {
                alph.Add(c);
            }
            alph.AddRange(Enumerable.Range(1, 9).Select(x => x.ToString()[0]).ToArray());

            HtmlReaderManager hrm = new HtmlReaderManager();
            LimitedConcurrencyLevelTaskScheduler lcts =
                new LimitedConcurrencyLevelTaskScheduler(200);
            TaskFactory factory = new TaskFactory(lcts);
            var         tasks   = new List <Task>();

            int counter = 1;


            foreach (string url in GetNextUrl())
            {
                Options opt = new Options();
                opt.Url     = url;
                opt.counter = counter;

                tasks.Add(factory.StartNew((opts) =>
                {
                    bool isDownloaded = false;
                    int tryCount      = 0;
                    do
                    {
                        Options options  = (Options)opts;
                        string targetUrl = "http://prntscr.com/1" + options.Url;

                        try
                        {
                            hrm.Get(targetUrl);
                            string html = hrm.Html;
                            HtmlAgilityPack.HtmlDocument doc = new HtmlDocument();
                            doc.LoadHtml(html);
                            string imgUrl = doc.DocumentNode.SelectSingleNode("//meta[@name='twitter:image:src']").Attributes["content"].Value;
                            if (!string.IsNullOrEmpty(imgUrl))
                            {
                                ImageDownloader.DownloadRemoteImageFile(imgUrl, "images/" + Guid.NewGuid() + ".jpeg");
                                Console.WriteLine(options.counter + ". Скачиваем " + imgUrl);
                            }
                            else
                            {
                                Console.WriteLine(options.counter + ". " + imgUrl + " пуст");
                            }
                            break;
                        }
                        catch (Exception ex)
                        {
                            tryCount++;
                            if (ex.ToString().StartsWith("System.Net.WebException: The remote server returned an error: (503) Server Unavailable.") && tryCount < 20)
                            {
                                Random s      = new Random();
                                int randomVal = s.Next(1000, 10000);
                                Thread.Sleep(randomVal);
                                Console.WriteLine(options.counter + ". ждём " + targetUrl + " (" + tryCount + " раз)");
                            }
                            else
                            {
                                Console.WriteLine(options.counter + ". ошибка " + targetUrl);
                                isDownloaded = true;
                            }
                        }
                    } while (!isDownloaded);
                }, opt, TaskCreationOptions.LongRunning));

                if (counter % 200 == 0)
                {
                    Task.WaitAll(tasks.ToArray());
                }
                tasks.Clear();
                counter++;
            }

            Console.ReadLine();
        }