Пример #1
0
        public static void StartCrawlEbuyer(string url)
        {
            try
            {
                PoliteWebCrawler crawler = new PoliteWebCrawler();
                crawler.PageCrawlStartingAsync        += crawler_ProcessPageCrawlStarting;
                crawler.PageCrawlCompletedAsync       += crawler_ProcessPageCrawlCompleted;
                crawler.PageCrawlDisallowedAsync      += crawler_PageCrawlDisallowed;
                crawler.PageLinksCrawlDisallowedAsync += crawler_PageLinksCrawlDisallowed;


                TimeSpan ts = new TimeSpan(0, 0, 5);
                CancellationTokenSource cancellationTokenSource = new CancellationTokenSource(ts);
                CrawlResult             result = crawler.Crawl(new Uri(url), cancellationTokenSource);

                if (result.ErrorOccurred)
                {
                    Console.WriteLine("Crawl of {0} completed with error: {1}", result.RootUri.AbsoluteUri, result.ErrorException.Message);
                }
                else
                {
                    Console.WriteLine("Crawl of {0} completed without error.", result.RootUri.AbsoluteUri);
                }
            }catch (Exception)
            {
            }
            ExtractingHtml.ExtractDetailsEbuyer();
        }
Пример #2
0
//Crawling code for GSM
        public static void StartCrawlGSM(string url)
        {
            PoliteWebCrawler crawler = new PoliteWebCrawler();

            crawler.PageCrawlStartingAsync        += crawler_ProcessPageCrawlStartingGSM;
            crawler.PageCrawlCompletedAsync       += crawler_ProcessPageCrawlCompletedGSM;
            crawler.PageCrawlDisallowedAsync      += crawler_PageCrawlDisallowedGSM;
            crawler.PageLinksCrawlDisallowedAsync += crawler_PageLinksCrawlDisallowedGSM;


            TimeSpan ts = new TimeSpan(0, 0, 0);
            CancellationTokenSource cancellationTokenSource = new CancellationTokenSource(ts);
            CrawlResult             result = crawler.Crawl(new Uri(url), cancellationTokenSource);

            if (result.ErrorOccurred)
            {
                Console.WriteLine("Crawl of {0} completed with error: {1}", result.RootUri.AbsoluteUri, result.ErrorException.Message);
            }
            else
            {
                Console.WriteLine("Crawl of {0} completed without error.", result.RootUri.AbsoluteUri);
            }


            //FileStream fs = new FileStream("url.txt", FileMode.Open);
            //StreamReader sr = new StreamReader(fs);
            //string str = "";
            //while ((str = sr.ReadLine()) != null)
            //{
            //    StartCrawl(str);
            //}

            ExtractingHtml.ExtractingDetailsGSM();
        }