Пример #1
0
        /// <summary>
        /// 解析网站
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        public HtmlDocument AnalysisHtml(string url)
        {
            EnCodHelp    enCodHelp = new EnCodHelp();
            var          EncodeUrl = enCodHelp.HttpGet(url);
            HtmlDocument document  = new HtmlDocument();

            document.LoadHtml(EncodeUrl);
            return(document);
        }
Пример #2
0
        public List <life_Comme.ParagraphAddModel> PostPackHtmlData(string url)
        {
            EnCodHelp enCodHelp = new EnCodHelp();
            var       EncodeUrl = enCodHelp.HttpGet(url);
            List <life_Comme.ParagraphAddModel> paragraphAddModel = new List <life_Comme.ParagraphAddModel>();
            HtmlDocument document = new HtmlDocument();

            document.LoadHtml(EncodeUrl);
            HtmlNodeCollection htmlNodes;

            if (url == "http://ent.163.com/")
            {
                htmlNodes = document.DocumentNode.SelectNodes("//div[@class='top_news']/ul/li/h3/a[@href]");
            }
            else if (url == "http://www.9game.cn/wzry/gonglue-0-1/")
            {
                htmlNodes = document.DocumentNode.SelectNodes("//div[@class='box-text']/ul[@class='icon-word-list statistics-show']/li[@class='no-pic statistics-click']/div[@class='right-text']/p[@class='tit']/a[@href]");
            }
            else
            {
                htmlNodes = document.DocumentNode.SelectNodes("//div[@class='news-box news-box-thr news-box-pic clear']/h4/a[@href]");
            }

            foreach (var item in htmlNodes)
            {
                life_Comme.ParagraphAddModel paragraphAdd = new life_Comme.ParagraphAddModel
                {
                    Title = item.InnerHtml.Trim()
                };
                var hrefString = item.Attributes["href"].Value.ToLower().Trim();;
                var lasthref   = "";
                if (hrefString.Contains("http:"))
                {
                    lasthref = hrefString.Replace("http:", "");
                }
                if (lasthref != "")
                {
                    paragraphAdd.VidUrl = lasthref;
                }
                else
                {
                    paragraphAdd.VidUrl = hrefString;
                }
                paragraphAddModel.Add(paragraphAdd);
            }
            return(paragraphAddModel);
        }