private static string GetConsultantInsertText(string firmID, Contact c)
{
string insert = string.Format(@"
INSERT INTO Consultant (Name, Title, Industry, Country, City, Email, Phone, BioPage, FirmID)
values ('{0}','{1}','{2}','{3}','{4}','{5}','{6}','{7}','{8}')
", c.Name, c.Title, c.Industry, c.Country, c.City, c.Email, c.Phone, c.BioPage, firmID);
return insert;
}
public static bool ParseConsultants(MainViewModel vm, bool persistToDatabase)
{
var watch = new Stopwatch(); watch.Start();
vm.LoadDoc();
// page expertease
vm.PageIndustry = string.Format("{0}", LookupQueryStringValue(vm, vm.Industries, "Expertise"));
// page country
vm.PageCountry = string.Format("{0}", LookupQueryStringValue(vm, vm.Countries, "Location"));
// page city
vm.PageCity = string.Format("{0}", LookupQueryStringValue(vm, vm.Cities, "Office"));
// page number
vm.PageNumber = string.IsNullOrEmpty(vm.QueryPairs["page"]) ? null : (int?)Int32.Parse(vm.QueryPairs["page"]);
// pages
// page consultants
var consultantElements = DomUtil.GetElementsByClass(vm.Doc, "td", "submain");
if (consultantElements == null || consultantElements.Count() < 1)
return false;
var sb = new StringBuilder();
var contacts = new List<Contact>();
foreach (var consutlantElement in consultantElements) {
var c = new Contact();
c.Industry = vm.PageIndustry;
c.Country = vm.PageCountry;
c.City = vm.PageCity;
c.Source = vm.GetHost();
foreach (var e in consutlantElement.ChildNodes) {
var val = e.InnerText;
switch (e.Name) {
case "#text":
if (string.IsNullOrEmpty(c.Title))
c.Title = val;
break;
case "a":
if (val.Contains('@'))
c.Email = val;
else
c.Name = val;
if (e.Attributes["href"] != null && e.Attributes["href"].Value.IndexOf("/Bios") > -1)
c.BioPage = string.Format("http://{0}{1}", c.Source, e.Attributes["href"].Value.Substring(e.Attributes["href"].Value.IndexOf("/Bios")));
break;
}
}
sb.Append(c.ToString());
sb.Append("\n");
contacts.Add(c);
}
if (persistToDatabase)
new ConsultantDataAccess().SaveConsultant(contacts);
Task.Factory.StartNew(() => {
vm.RetrieveStatus = string.Format("{0} retrieved in {1}:{2}", vm.Url, watch.Elapsed.Minutes, watch.Elapsed.Seconds);
vm.PageConsultants = sb.ToString();
});
return true;
}