58.com Qiyi

usingAnflecrawler.common;usingSystem;usingSystem.Collections.Generic;usingSystem.Linq;usingSystem.Text;usingSystem.Threading.Tasks;namespaceanflecrawler.dataanalyzer{Internal classQy58:analyzerbase { Public Override voidInit (Pagecrawler crawler) {Base.            Init (crawler); varURL =NewUri ("http://qy.58.com/caohejing/pn1/?PGTID=14177711280840.45006677554920316&ClickID=1"); //http://qy.58.com/19583455460359/?PGTID=14177659184690.5166369006238447&clickid=4Crawler. Pushurl (URL,0); }        protected Override voidanalyzeinternal (pagelandentity current) {varLander =Crawler.lander; varPhandler =Createcontenthandler (current); Switch(current. Depth) { Case 0: {pHandler.AjaxBlocks.Add (HACK); varDom =Lander.                        GetDocument (Phandler); Doperpaging (current, Dom. Documentnode,". Next"); foreach(varNodeinchQuerynodes (DOM. Documentnode,". Complist a"))                        {                            varURL =gethref (node, current.                            URL); Crawler.pushurl (URL,1); }                    }                     Break;  Case 1:                    {                        varDom =Lander.                        GetDocument (Phandler); varattr =NewAttributefiller (); attr. Append ("name:{0}", Querytexts (DOM. Documentnode,". Compt").                        First ()); foreach(varThinchQuerynodes (DOM. Documentnode,". basicmsg table th"). Skip (1))                        {                            stringSTh =th.                            InnerText, STd; Switch(sTh) { Case "Contact Phone":                                 Case "Email":                                    varClient =NewSystem.Net.WebClient (); varINode = querynode (th. NextSibling,"img"); byte[] Imgraw = client. Downloaddata (Gethref (INode, current. URL, Attrname:"src")); varIMG =NewSystem.Drawing.Bitmap (NewSystem.IO.MemoryStream (Imgraw)); STd=OCR (IMG);  Break;  Case "Company Address": STd= querytexts (th. NextSibling,"span").                                    First ();  Break; default: STd=th.                                    NextSibling.InnerText.HtmlTrim ();  Break; } attr. Append ("{0}:{1}", STh, sTd); }                        varBo =Newcompanyentity (); Bo. City="Shanghai"; Bo. GroupName="caohejing Enterprise"; Bo. Pageurl=Current .                        url.originalstring; Bo. Updatedate=DateTime.Now; attr. Fillentity (Bo,Newdictionary<string,string>()                         {                            {"Company Nature","Nature"},                            {"Company Industry","Industry"},                            {"Company Size"," Scale"},                            {"Contact Person","ContactPerson"},                            {"Corporate website","Website"},                            {"Contact Phone","Tel"},                            {"Email","Email"},                            {"Company Address","Address"},                        });                        Repository.savecompany (BO); Crawler.outwrite ("save Enterprise {0}", Bo.                    Name); }                     Break; }        }    }}

58.com Qiyi