override public void PROCESSOR(BotCycle bc)
{
CustomBot cb = (CustomBot)bc.Bot;
if (!cb.HR.Get(Url))
{
throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + Url);
}
DataSifter.Capture gc = cb.list.Parse(cb.HR.HtmlResult);
{
string url = gc.ValueOf("NextPageUrl");
if (url != null)
{
cb.BotCycle.Add(new ListItem(Spider.GetAbsoluteUrl(url, cb.HR.ResponseUrl)));
}
}
string[] urls = Spider.GetAbsoluteUrls(gc.ValuesOf("ProductUrl"), cb.HR.ResponseUrl, cb.HR.HtmlResult);
foreach (string url in urls)
{
cb.BotCycle.Add(new ProductItem(url));
}
}
override public void PROCESSOR(BotCycle bc)
{
CustomBot cb = (CustomBot)bc.Bot;
string name = FieldPreparation.Html.GetCsvField(Name);
if (!cb.HR.GetPage(Url))
{
throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + Url);
}
DataSifter.Capture c = product.Parse(cb.HR.HtmlResult);
string zip_code = Regex.Replace(c.ValueOf("ZipCode"), @"[^\d]", "", RegexOptions.Singleline);
string url2 = "http://www.yellowpages.com/search?search_terms=" + name + "&geo_location_terms=" + zip_code;
string email = null;
string url3 = url2;
if (cb.HR.GetPage(url2))
{
DataSifter.Capture c2 = yp.Parse(cb.HR.HtmlResult);
string regex_name = get_stripped_name(name);
regex_name = Regex.Escape((regex_name.Length > 10 ? regex_name.Substring(0, 10) : regex_name).Trim());
foreach (DataSifter.Capture cc in c2["Company"])
{
if (cc.ValueOf("ZipCode") != null &&
Regex.Replace(cc.ValueOf("ZipCode"), @"[^\d]", "", RegexOptions.Singleline) == zip_code &&
Regex.IsMatch(get_stripped_name(cc.ValueOf("Name")), regex_name, RegexOptions.IgnoreCase)
)
{
url3 = Spider.GetAbsoluteUrl(cc.ValueOf("Url"), url2);
if (!cb.HR.GetPage(url3))
{
throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + url3);
}
DataSifter.Capture c3 = yp2.Parse(cb.HR.HtmlResult);
email = c3.ValueOf("Email");
break;
}
}
}
else if (cb.HR.HWResponse.StatusCode != HttpStatusCode.NotFound)
{
throw new ProcessorException(ProcessorExceptionType.RESTORE_AS_NEW, "Could not get: " + url2);
}
FileWriter.This.PrepareAndWriteHtmlLineWithHeader(
"Name", Name,
"City", City,
"ZipCode", zip_code,
"State", State,
"Phone", Phone,
"Email", email,
"Url", Url,
"Url2", url3
);
}
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")));
}
}