override public void __Processor(BotCycle bc) { for (int i = 0; i < 3; i++) { bc.Add(new Product(url: i + Url)); } //list next page bc.Add(new Category(url: "qqqqq2", t: T)); }
/// <summary> /// InputItem can define a method to process it. /// When it is not defined within InputItem class, it must be defined in CustomBot as PROCESSOR_[InputItem class name] /// </summary> /// <param name="bc">BotCycle that keeps the current thread</param> override public void __Processor(BotCycle bc) { bc.Add(new Category(url: Url + "?q=1", t: new Category.Tag(name: "fff", description: "ttttt"))); //it is possible to get the current CustomBot when access to common members is needed ((CustomSession)bc.Session).counter++; }
static public void get_links(int depth2, HttpRoutine hr, BotCycle bc) { if (depth2 > Bot.Properties.Spider.Default.MaxDownloadLinkDepth) { return; } string domain = Spider.GetDomain(hr.ResponseUrl); int page_count = domain2page_count[domain]; if (Bot.Properties.Spider.Default.MaxPageCountPerSite > -1 && page_count >= Bot.Properties.Spider.Default.MaxPageCountPerSite) { return; } AgileSpider ags = new AgileSpider(hr.ResponseUrl, hr.HtmlResult); List <WebLink> wls = ags.GetWebLinks(WebLinkType.Anchor | WebLinkType.Area | WebLinkType.Form | WebLinkType.MetaTag | WebLinkType.Frame | WebLinkType.Image | WebLinkType.Javascript); List <WebLink> beyond_domain_web_links; wls = Spider.GetSpiderableLinks(ags.BaseUri, wls, out beyond_domain_web_links); bool download = true; if (depth2 >= Bot.Properties.Spider.Default.MaxDownloadLinkDepth) { download = false; } foreach (WebLink wl in wls) { bc.Add(new Link(url: wl.Url, depth: depth2, download: download)); page_count++; if (Bot.Properties.Spider.Default.MaxPageCountPerSite > -1 && Bot.Properties.Spider.Default.MaxPageCountPerSite <= page_count) { Log.Warning(domain + " reached MaxPageCountPerSite: " + Bot.Properties.Spider.Default.MaxPageCountPerSite.ToString()); break; } } domain2page_count[domain] = page_count; foreach (WebLink wl in beyond_domain_web_links) { bc.Add(new Link(url: wl.Url, depth: depth2, download: false)); } }
public void get_links(int depth2, string queue) { if (depth2 > Bot.Properties.Spider.Default.MaxDownloadLinkDepth) { return; } string domain = Spider.GetDomain(hr.ResponseUrl); int page_count = domain2page_count[domain]; if (Bot.Properties.Spider.Default.MaxPageCountPerSite > -1 && page_count >= Bot.Properties.Spider.Default.MaxPageCountPerSite) { return; } string queue2 = domain + "-" + depth2.ToString(); if (depth2 > 1) { Session.SetInputItemQueuePositionAfterQueue(queue2, queue);//by default queue name is item type name but it can be different if needed } AgileSpider ags = new AgileSpider(hr.ResponseUrl, hr.HtmlResult); List <WebLink> wls = ags.GetWebLinks(WebLinkType.Anchor | WebLinkType.Area | WebLinkType.Form | WebLinkType.MetaTag | WebLinkType.Frame | WebLinkType.Image | WebLinkType.Javascript); List <WebLink> beyond_domain_web_links; wls = Spider.GetSpiderableLinks(ags.BaseUri, wls, out beyond_domain_web_links); bool download = true; if (depth2 >= Bot.Properties.Spider.Default.MaxDownloadLinkDepth) { download = false; } foreach (WebLink wl in wls) { BotCycle.Add(queue2, new Link(url: wl.Url, depth: depth2, download: download)); page_count++; if (Bot.Properties.Spider.Default.MaxPageCountPerSite > -1 && Bot.Properties.Spider.Default.MaxPageCountPerSite <= page_count) { Log.Warning(domain + " reached MaxPageCountPerSite: " + Bot.Properties.Spider.Default.MaxPageCountPerSite.ToString()); break; } } domain2page_count[domain] = page_count; foreach (WebLink wl in beyond_domain_web_links) { BotCycle.Add(queue2, new Link(url: wl.Url, depth: depth2, download: false));//by default queue name is item type name but it can be deifferent if needed } }
override public void PROCESSOR(BotCycle bc) { CustomBot cb = (CustomBot)bc.Bot; string url = "http://www.rent.com/" + Regex.Replace(State, @"\s", "-"); if (!cb.HR.GetPage(url)) { throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + url); } DataSifter.Capture c = cities.Parse(cb.HR.HtmlResult); string[] us = c.ValuesOf("Url"); for (int i = 0; i < us.Length; i++) { bc.Add(new SearchItem("http://www.rent.com" + us[i])); } }
void search_processor(string url) { if (!HR.GetPage(url)) { throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + url); } DataSifter.Capture c0 = search.Parse(HR.HtmlResult); string npu = c0.ValueOf("NextPageUrl"); if (npu != null) { BotCycle.Add(new SearchNextPageItem(npu)); } foreach (DataSifter.Capture c in c0["Product"]) { BotCycle.Add(new CompanyItem(Spider.GetAbsoluteUrl(c.ValueOf("Url"), url), c.ValueOf("Name"), c.ValueOf("City"), c.ValueOf("State"), c.ValueOf("Phone"))); } }
override public void __Processor(BotCycle bc) { CustomSession session = (CustomSession)bc.Session; string address = new System.Globalization.CultureInfo("en-US", false).TextInfo.ToTitleCase(Address.ToLower()); string file_dir = PathRoutines.CreateDirectory(session.output_dir + "\\" + address, true); string output_pdf = file_dir + "\\" + Regex.Replace(address + " RPA.pdf", @"\s+", " "); { //lock (template_pdf) //{ // File.Copy(template_pdf, pdf); //} PdfReader.unethicalreading = true; PdfReader pr = new PdfReader(template_pdf); //pr.RemoveUsageRights(); //pr.SelectPages("7,8"); PdfStamper ps = new PdfStamper(pr, new FileStream(output_pdf, FileMode.Create, FileAccess.Write, FileShare.None)); //string fs = ""; //foreach (KeyValuePair<string, AcroFields.Item> kvp in ps.AcroFields.Fields) // fs += "\n{\"" + kvp.Key + "\", \"\"},"; set_field(ps.AcroFields, "Todays Date", DateTime.Today.ToShortDateString()); set_field(ps.AcroFields, "Buyer Name", Settings.Parties.BuyerProfile.Name); set_field(ps.AcroFields, "Address and Unit Number", address + " " + UnitNumber); set_field(ps.AcroFields, "City/Town", City); //set_field(ps.AcroFields, "CLARK", ); set_field(ps.AcroFields, "Zip", ZipCode); set_field(ps.AcroFields, "PARCEL NUMBER", ParcelNumber); set_field(ps.AcroFields, "OfferAmt", OfferAmt); string offer_amt_ = Regex.Replace(OfferAmt, @"[^\d]", ""); if (offer_amt_.Length > 0) { set_field(ps.AcroFields, "OfferAmt in words", ConvertionRoutines.NumberToWords(int.Parse(offer_amt_)).ToUpper()); } set_field(ps.AcroFields, "EMD", Settings.Offer.Emd); //set_field(ps.AcroFields, "Check Box1", ); //set_field(ps.AcroFields, "Balance", ); set_field(ps.AcroFields, "Co Buyer Name", Settings.Parties.BuyerProfile.CoBuyerName); set_field(ps.AcroFields, "<address> <UnitNumber> <City/town> NV <ZIP Code>", address + " " + UnitNumber + ", " + City + " NV " + ZipCode); set_field(ps.AcroFields, "ML#", "ALL PER ML# " + ML_Id); set_field(ps.AcroFields, "Title Company", Settings.Parties.EscrowProfile.TitleCompany); set_field(ps.AcroFields, "Escrow Officer", Settings.Parties.EscrowProfile.Officer); set_field(ps.AcroFields, "Close of Escrow", Settings.Offer.CloseOfEscrow.ToShortDateString()); //String[] values = ps.AcroFields.GetAppearanceStates("Licensee Yes"); //String[] values2 = ps.AcroFields.GetAppearanceStates("Licensee No"); if (Settings.Parties.BuyerProfile.UseLicensee) { set_field(ps.AcroFields, "Licensee Yes", "Yes"); //set_field(ps.AcroFields, "Licensee No", "Off"); //set_field(ps.AcroFields, "Licensee relationship", DateTime.Today.ToShortDateString()); set_field(ps.AcroFields, "Licensee relationship", Settings.Parties.BuyerProfile.LicenseeRelationship); switch (Settings.Parties.BuyerProfile.RelationshipType) { case "Family Firm": set_field(ps.AcroFields, "Licensee Family Firm", "Yes"); break; case "Principal": set_field(ps.AcroFields, "Licensee Principal", "Yes"); break; } } else { set_field(ps.AcroFields, "Licensee No", "Yes"); //set_field(ps.AcroFields, "Licensee Yes", "Off"); } DateTime date = DateTime.Now.AddDays(7); set_field(ps.AcroFields, "Response Month", date.ToString("MMMM")); set_field(ps.AcroFields, "Response day", date.ToString("%d")); set_field(ps.AcroFields, "year", date.ToString("yyyy"));; string emd_ = Regex.Replace(Settings.Offer.Emd, @"[^\d]", ""); if (offer_amt_.Length > 0 && emd_.Length > 0) { set_field(ps.AcroFields, "Balance", (int.Parse(offer_amt_) - int.Parse(emd_)).ToString()); } // string AdditionalTerms = @"This form is available for use by the real estate industry. It is not intended to identify the user as a REALTOR®. //8 REALTOR® is a registered collective membership mark which may be used only by members of the NATIONAL //9 ASSOCIATION OF REALTORS® who subscribe to its Code."; string s = AdditionalTerms; s = fill_field_by_words(ps.AcroFields, "AdditionalTerms1", s); s = fill_field_by_words(ps.AcroFields, "AdditionalTerms2", s); s = fill_field_by_words(ps.AcroFields, "AdditionalTerms3", s); if (s.Length > 0) { s = AdditionalTerms; s = fill_field_by_chars(ps.AcroFields, "AdditionalTerms1", s); s = fill_field_by_chars(ps.AcroFields, "AdditionalTerms2", s); set_field(ps.AcroFields, "AdditionalTerms3", s); } set_field(ps.AcroFields, "Buyer Broker", Settings.Parties.BrokerProfile.Name); set_field(ps.AcroFields, "Agent Name", Settings.Parties.AgentProfile.Name); set_field(ps.AcroFields, "ListAgentFullName", ListAgentFullName); set_field(ps.AcroFields, "Company Name", Settings.Parties.BrokerProfile.Company); set_field(ps.AcroFields, "ListOfficeName", ListOfficeName); set_field(ps.AcroFields, "Agents License", Settings.Parties.AgentProfile.LicenseNo); set_field(ps.AcroFields, "Brokers License", Settings.Parties.BrokerProfile.LicenseNo); set_field(ps.AcroFields, "Office Address", Settings.Parties.BrokerProfile.Address); set_field(ps.AcroFields, "Office Phone", Settings.Parties.BrokerProfile.Phone); set_field(ps.AcroFields, "City State Zip", Settings.Parties.BrokerProfile.City + " " + Settings.Parties.BrokerProfile.State + " " + Settings.Parties.BrokerProfile.Zip); set_field(ps.AcroFields, "Agent Email", Settings.Parties.AgentProfile.Email); //set_field(ps.AcroFields, "Response Month", ); //set_field(ps.AcroFields, "Response day", ); //set_field(ps.AcroFields, "year", ); //set_field(ps.AcroFields, "Date_2", ); //set_field(ps.AcroFields, "Date_3", ); //set_field(ps.AcroFields, "Sellers Broker", ); set_field(ps.AcroFields, "Agents Name_2", Settings.Parties.AgentProfile.Name); set_field(ps.AcroFields, "Company Name_2", Settings.Parties.BrokerProfile.Company); set_field(ps.AcroFields, "Agents License Number_2", Settings.Parties.AgentProfile.LicenseNo); set_field(ps.AcroFields, "Brokers License Number_2", Settings.Parties.BrokerProfile.LicenseNo); set_field(ps.AcroFields, "Office Address_2", Settings.Parties.BrokerProfile.Address); set_field(ps.AcroFields, "Phone_2", Settings.Parties.BrokerProfile.Phone); set_field(ps.AcroFields, "City State Zip_2", Settings.Parties.BrokerProfile.Zip); set_field(ps.AcroFields, "Email_2", Settings.Parties.AgentProfile.Email); set_field(ps.AcroFields, "ListAgentEmail", ListAgentEmail); ps.FormFlattening = true; for (int i = 1; i <= pr.NumberOfPages; i++) { var pcb = ps.GetOverContent(i); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.InitialFile), new System.Drawing.Point(497, 67)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerInitialFile), new System.Drawing.Point(536, 67)); } } { var pcb = ps.GetOverContent(3); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.InitialFile), new System.Drawing.Point(140, 103)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerInitialFile), new System.Drawing.Point(280, 103)); } } { var pcb = ps.GetOverContent(9); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.SignatureFile), new System.Drawing.Point(60, 190)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerSignatureFile), new System.Drawing.Point(60, 155)); } } ps.Close(); pr.Close(); } string output_addendum_pdf = null; if (Settings.Offer.ShortSaleAddendum) { output_addendum_pdf = file_dir + "\\" + Regex.Replace(address + " SS Addendum.pdf", @"\s+", " "); PdfReader.unethicalreading = true; PdfReader pr = new PdfReader(template_addendum_pdf); PdfStamper ps = new PdfStamper(pr, new FileStream(output_addendum_pdf, FileMode.Create, FileAccess.Write, FileShare.None)); string fs = ""; foreach (KeyValuePair <string, AcroFields.Item> kvp in ps.AcroFields.Fields) { fs += "\n{\"" + kvp.Key + "\", \"\"},"; } if (Settings.Parties.BuyerProfile.UseCoBuyer) { set_field(ps.AcroFields, "<Buyer Name> and <Co Buyer Name>", Settings.Parties.BuyerProfile.Name + " and " + Settings.Parties.BuyerProfile.CoBuyerName); } else { set_field(ps.AcroFields, "<Buyer Name> and <Co Buyer Name>", Settings.Parties.BuyerProfile.Name); } set_field(ps.AcroFields, "Todays Date", DateTime.Today.ToShortDateString()); set_field(ps.AcroFields, "<Address> <UnitNumber> <City/Town> NV <Zip Code>", address + " " + UnitNumber + ", " + City + " NV " + ZipCode); set_field(ps.AcroFields, "Agent Name", Settings.Parties.AgentProfile.Name); //set_field(ps.AcroFields, "Agent Phone", ); ps.FormFlattening = true; for (int i = 1; i <= pr.NumberOfPages; i++) { var pcb = ps.GetOverContent(i); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.InitialFile), new System.Drawing.Point(120, 70)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerInitialFile), new System.Drawing.Point(157, 70)); } } { var pcb = ps.GetOverContent(1); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.InitialFile), new System.Drawing.Point(140, 400)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerInitialFile), new System.Drawing.Point(197, 400)); } } { var pcb = ps.GetOverContent(3); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.SignatureFile), new System.Drawing.Point(150, 547)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerSignatureFile), new System.Drawing.Point(150, 504)); } } ps.Close(); pr.Close(); } string output_addendum1_pdf = null; if (Settings.Offer.OtherAddendum1) { output_addendum1_pdf = file_dir + "\\" + Regex.Replace("duites " + address + ".pdf", @"\s+", " "); PdfReader.unethicalreading = true; PdfReader pr = new PdfReader(template_addendum1_pdf); PdfStamper ps = new PdfStamper(pr, new FileStream(output_addendum1_pdf, FileMode.Create, FileAccess.Write, FileShare.None)); string fs = ""; foreach (KeyValuePair <string, AcroFields.Item> kvp in ps.AcroFields.Fields) { fs += "\n{\"" + kvp.Key + "\", \"\"},"; } set_field(ps.AcroFields, "AgentName", Settings.Parties.AgentProfile.Name); set_field(ps.AcroFields, "LicenseNumber", Settings.Parties.AgentProfile.LicenseNo); set_field(ps.AcroFields, "Buyer Name", Settings.Parties.BuyerProfile.Name); set_field(ps.AcroFields, "Co-Buyer Name", Settings.Parties.BuyerProfile.CoBuyerName); set_field(ps.AcroFields, "BrokerName", Settings.Parties.BrokerProfile.Name); set_field(ps.AcroFields, "Company", Settings.Parties.BrokerProfile.Company); set_field(ps.AcroFields, "Date_3", DateTime.Now.ToShortDateString()); set_field(ps.AcroFields, "Time_3", DateTime.Now.ToShortTimeString()); set_field(ps.AcroFields, "Date_4", DateTime.Now.ToShortDateString()); set_field(ps.AcroFields, "Time_4", DateTime.Now.ToShortTimeString()); ps.FormFlattening = true; { var pcb = ps.GetOverContent(1); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.InitialFile), new System.Drawing.Point(70, 190)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerInitialFile), new System.Drawing.Point(140, 190)); } } { var pcb = ps.GetOverContent(1); add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.SignatureFile), new System.Drawing.Point(110, 85)); if (Settings.Parties.BuyerProfile.UseCoBuyer) { add_image(pcb, System.Drawing.Image.FromFile(Settings.Parties.BuyerProfile.CoBuyerSignatureFile), new System.Drawing.Point(110, 85)); } } ps.Close(); pr.Close(); } bc.Add(new EmailItem(output_pdf, output_addendum_pdf, output_addendum1_pdf)); }