Exemple #1
0
        public List<Link> GetLinks(string urlHead,int pageNeededNumber)
        {
            List<Link> links = new List<Link>();
            string url ;
            for (int i = 0; i < pageNeededNumber; i++)
            {
                url = urlHead + 50 * i;
                WebPage page = new WebPage(url);
                links.AddRange(page.getSpecialLinksByUrl("/p/", 50));
            }

            StreamWriter sw = new StreamWriter("C:\\Users\\LENOVO\\Desktop\\urls3.0.txt",false,Encoding.UTF8);
            for (int i = 0; i < links.Count; i++)
            {
                sw.Write(links[i].NavigateUrl + "\r\n");
            }
            sw.Close();

            return links;
        }
Exemple #2
0
        private void bt_go_Click(object sender, EventArgs e)
        {
            startUrl = "http://tieba.baidu.com/f?kw=%E7%BD%91%E7%BB%9C%E6%B5%81%E8%A1%8C%E8%AF%AD&ie=utf-8&pn=";
            List<Link> links =GetLinks(startUrl, 50);

            if (File.Exists("C:\\Users\\LENOVO\\Desktop\\回复提取3.0.txt"))
                File.Delete("C:\\Users\\LENOVO\\Desktop\\回复提取3.0.txt");

            for (int i = 0; i < links.Count; i++)
            {
                string fstUrl = links[i].NavigateUrl;
                WebPage page = new WebPage(fstUrl);
                int pageNumber = GetPageNumber(page.Context);
                if (pageNumber == 0)
                    continue;
                for (int j = 0; j < pageNumber; j++)
                {
                    string currentUrl = fstUrl + "?pn=" + (j + 1).ToString();
                    WebPage currentPage = new WebPage(currentUrl);
                    GetAndWrite(currentPage.Context,"回复提取3.0.txt",true);
                }
            }
        }