private void SaveTabir(string WordName, string TabirUrl) { ReqUtils Reqs = new ReqUtils(); TabirUrl = TabirUrl.Replace("\" target=\"_blank", ""); string Result = Reqs.GetHTML(TabirUrl, System.Text.Encoding.UTF8); string _pattern = @" class=""contentpaneopen_text"">\s*(.*?)</div>"; Regex r = new Regex(_pattern, RegexOptions.IgnoreCase | RegexOptions.Singleline); Match m = r.Match(Result); if (m.Success) { string Meaning = m.Groups[1].Captures[0].ToString(); Meaning = Reqs.RemoveTags(Meaning); TabirDataContext dc = new TabirDataContext(); Parsetv91._1.Code.DAL.Tabir NewRecord = new Code.DAL.Tabir(); dc.Tabirs.InsertOnSubmit(NewRecord); NewRecord.WordName = WordName; NewRecord.Meaning = Meaning; NewRecord.ID = MainCounter.ToString(); MainCounter++; dc.SubmitChanges(); } }
protected void btnakairan_Click(object sender, EventArgs e) { int Counter = 0; ReqUtils Reqs = new ReqUtils(); string Params = "-a.html,-b.html,-p.html,-t.html,-se.html,-j.html,-ch.html,-h.html,-kh.html,-d.html-zal.html,-r.html,-z.html,24263.html,-sin.html,-shin.html,-sad.html,-zad.html,-ta.html,-za.html,-eyn.html,-ghein.html,-f.html-gh.html,-k.html,-g.html,-l.html,-m.html,-n.html,-v.html,-he.html,-ye.html"; string[] ParamsArray = Params.Split(','); for (int i = 0; i < ParamsArray.Length; i++) { string CurUrl = ""; if (ParamsArray[i] == "") { CurUrl = "http://tabirkhab.akairan.com/sleep/sleep/"; } else { CurUrl = "http://tabirkhab.akairan.com/sleep/sleep/jadval" + ParamsArray[i]; } string Result = Reqs.GetHTML(CurUrl, System.Text.Encoding.UTF8); string _pattern = @"<td class=""buttmtbl""><a href=""(.*?)"">(.*?)</a></td>"; Regex r = new Regex(_pattern, RegexOptions.IgnoreCase | RegexOptions.Singleline); Match m = r.Match(Result); while (m.Success) { Title = m.Groups[2].Captures[0].ToString(); Title = Tools.PersianTextCorrection(Title); if (GetTabirCount(Title) == 0) { string TabirUrl = m.Groups[1].Captures[0].ToString(); SaveTabir(Title, TabirUrl); Counter++; } m = m.NextMatch(); } lblMessage.Text = Counter + " Tabirs added"; } }
void ListNews(int privateSiteCode, string privateNewsUrl, string privateNewsTitle, string privateREDetail, string privateREImage, string privateREVideo, int Count, string LinkDomainName, int EncodingTypeCode) { ReqUtils gn = new ReqUtils(); string NewsContentHtml = ""; string FullStory = ""; string ImageSource = ""; string VideoSource = ""; string TextTitle = ""; string NewsCode = ""; System.Text.Encoding enc = System.Text.Encoding.UTF8; if (EncodingTypeCode != 1) { IBaseBOL <DataTable> BolHardCode = new BOLHardCode(); BolHardCode.QueryObjName = "HCEncodingTypes"; DataTable dt = BolHardCode.GetDetails(EncodingTypeCode); enc = System.Text.Encoding.GetEncoding(dt.Rows[0]["Description"].ToString()); } privateNewsUrl = privateNewsUrl.Replace("//", "/"); privateNewsUrl = privateNewsUrl.Replace("http:/", "http://"); privateNewsUrl = privateNewsUrl.Replace("https:/", "https://"); if (privateREDetail.Length > 0) { NewsContentHtml = gn.GetHTML(privateNewsUrl, enc); if (privateREDetail != null && privateREDetail != "") { FullStory = gn.GetREGroup(NewsContentHtml, privateREDetail, "CONTENT"); } if (privateREImage != null && privateREImage != "") { ImageSource = gn.GetREGroup(NewsContentHtml, privateREImage, "IMAGE"); } if (privateREVideo != null && privateREVideo != "") { VideoSource = gn.GetREGroup(NewsContentHtml, privateREVideo, "VIDEO"); } if (!ImageSource.StartsWith("http://") && !ImageSource.StartsWith("https://") && ImageSource != "") { ImageSource = LinkDomainName + ImageSource; } if (!VideoSource.StartsWith("http://") && VideoSource != "") { VideoSource = LinkDomainName + VideoSource; } FullStory = gn.RemoveTags(FullStory, "br"); FullStory = FullStory.Replace("'", ""); FullStory = FullStory.Replace("\"", ""); FullStory = FullStory.Replace("\r", ""); FullStory = FullStory.Replace("\t", ""); FullStory = FullStory.Replace(" ", ""); ImageSource = ImageSource.Replace("[", "%5B"); ImageSource = ImageSource.Replace("]", "%5D"); VideoSource = VideoSource.Replace("[", "%5B"); VideoSource = VideoSource.Replace("]", "%5D"); if (ImageSource != "") //Save News File { try { int SlashPos = ImageSource.LastIndexOf("/"); string FileName = ImageSource.Substring(SlashPos + 1, ImageSource.Length - SlashPos - 1); WebClient WebCl = new WebClient(); string FName = Server.MapPath("~/Files/News/" + Tools.GetRandomFileName(FileName)); //WebCl.DownloadFile(ImageSource, FName); } catch { } } } TextTitle = gn.RemoveTags(privateNewsTitle); string ImageTag = ""; if (ImageSource != "") { ImageSource = "<img src=\"" + ImageSource + "\">"; } string outStr = string.Format("<table border=\"1\" class=\"cNews\" dir=\"rtl\" width=\"100%\" ><tr><td>{0}</td><td >{1}</td><td>{2}</td><td>{3}</td><td>{4}</td><td>{5}</td></tr></table>", Count, TextTitle, FullStory.Length, ImageSource, VideoSource, privateNewsUrl); Response.Write(outStr); Response.Flush(); gn.Dispose(); }
public void GetSingleSite(int privateSiteCode, int privateLimitCount) { int Count = 1; string NewsTitle = ""; string HtmlContent = ""; string EditedNewsTitle = ""; string NewsTextTitle = ""; ReqUtils gn; string SelectStatement = ""; BOLResourseSiteCats ResourceSiteCatsBOL = new BOLResourseSiteCats(1); vResourseSiteCats SingleSite = ResourceSiteCatsBOL.GetSingleSite(privateSiteCode); int SiteCode = SingleSite.Code; string SiteName = SingleSite.Name; string SiteUrl = SingleSite.Url; int? EncodingTypeCode = SingleSite.HCEncodingTypeCode; string BaseURL = SingleSite.BaseURL; string RELink = SingleSite.RELink; string REDetail = SingleSite.REDetail; string REImage = SingleSite.REImage; string REVideo = SingleSite.REVideo; ArrayList NewList = new ArrayList(); gn = new ReqUtils(); System.Text.Encoding enc = System.Text.Encoding.UTF8; if (EncodingTypeCode != 1) { IBaseBOL <DataTable> BolHardCode = new BOLHardCode(); BolHardCode.QueryObjName = "HCEncodingTypes"; DataTable dt = BolHardCode.GetDetails((int)EncodingTypeCode); enc = System.Text.Encoding.GetEncoding(dt.Rows[0]["Description"].ToString()); } int LastSlash = SiteUrl.LastIndexOf("/"); string LinkDomainName; if (BaseURL != null && BaseURL != "") { LinkDomainName = BaseURL; } else { LinkDomainName = SiteUrl.Substring(0, LastSlash + 1); } if (!(bool)SingleSite.RssIsActive) { HtmlContent = gn.GetHTML(SiteUrl, enc); NewList = gn.ExtractNewsLinks(HtmlContent, RELink, LinkDomainName); } else { NewList = GetNewsList(SingleSite.RssUrl); } IEnumerator NewENum = NewList.GetEnumerator(); gn = new ReqUtils(); while (NewENum.MoveNext()) { NewsTitle = NewENum.Current.ToString(); NewsTextTitle = gn.RemoveTags(NewsTitle); BOLNews NewsBOl = new BOLNews(); if (!NewsBOl.CheckNewsExists(NewsTextTitle, SiteCode)) { string RealLink = gn.ExtractLink(NewsTitle); string outStr = ""; //outStr = string.Format("<table width=100% ><tr><td class=\"cNews\">{1}</td><td>{0}</td></tr></table>", Count, NewsTitle); //Response.Write(outStr); //Response.Flush(); ListNews(SiteCode, RealLink, NewsTitle, REDetail, REImage, REVideo, Count, LinkDomainName, (int)EncodingTypeCode); } Count++; if (privateLimitCount != 0) { if (privateLimitCount == Count) { break; } } } }