public IList<IVehicle> Scrape(IExtractionArguments args, HtmlDocument page)
{
IList<IVehicle> vehicles = new List<IVehicle>();
var nodes = page.DocumentNode.SelectNodes("//div[@class='box_item_detail section ']");
var gooNodes = page.DocumentNode.SelectNodes("//div[@class='box_item_detail section no_goo_area']");
if (nodes == null)
nodes = gooNodes;
else if (gooNodes != null)
{
foreach (var selectNode in gooNodes)
{
nodes.Add(selectNode);
}
}
foreach (var node in nodes)
{
Vehicle vehicle = new Vehicle();
vehicle.Title =
node.SelectNodes("div/div[@class='heading_inner']")
.First()
.InnerText.Trim()
.Replace(" ", " ")
.Replace("\t", string.Empty)
.Replace("\n", string.Empty);
vehicle.Make = args.Make;
vehicle.Model = args.Model;
vehicle.Milage = node.SelectNodes("div/div/table/tr/td[@class='w63']").First().InnerText;
double price = 0;
double.TryParse(node.SelectNodes("div/div/table/tr/td/div[@class='priceInfo']/p/em").First().InnerText,
out price);
vehicle.Price = price;
var dirtyYear = node.SelectNodes("div/div/table/tr/td[@class='w66']")[0].InnerText;
dirtyYear = dirtyYear.Substring(0, dirtyYear.IndexOf("("));
vehicle.Year = int.Parse(dirtyYear);
vehicles.Add(vehicle);
}
return vehicles;
}
public IList<IVehicle> Scrape(IExtractionArguments args, HtmlDocument page)
{
var vehicleList = new List<IVehicle>();
var searchResults = page.DocumentNode.SelectNodes("//div[@class='searchResult ']");
if (searchResults == null)
return vehicleList;
foreach (var searchResult in searchResults)
{
IVehicle vehicle = new Vehicle();
vehicle.Make = args.Make;
vehicle.Model = args.Model;
vehicle.Title = searchResult.SelectSingleNode(".//h2[@class='serpTitle']").InnerText.Trim();
vehicle.Price = GetVehiclePrice(searchResult);
vehicle.Milage = GetMilage(searchResult);
vehicle.Year = GetAge(searchResult);
vehicleList.Add(vehicle);
_log.DebugFormat("Added {0}", vehicle);
}
return vehicleList;
}