protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//*[@id='consultant-header']/h1");
profile.OfficeAddress = this.GetTextByXPath("//*[@id='consultant-tabs']/div[2]/table/tbody/tr[1]/td[1]/p");
profile.EmailAddress = this.GetTextByXPath("//*[@id='consultant-tabs']//table//tr//p[starts-with(text(), 'Email:')]//a");
profile.WebsiteAddress = this.GetTextByXPath("//*[@id='consultant-tabs']//table//tr//p[starts-with(text(), 'Website:')]//a");
profile.AboutUs = this.GroupAllTextByXPath("//*[@id='consultant-tabs']/following-sibling::p", "\r\n\r\n");
string otherDetails = this.GetTextByXPath("//*[@id='consultant-tabs']//table/tbody/tr[1]/td[2]/p");
// Take line from paragraph by word, then replace the text
profile.PhoneNumber = this.GetLineContaining(otherDetails, "Telephone: ").Replace("Telephone: ", "");
profile.FaxNumber = this.GetLineContaining(otherDetails, "Fax: ").Replace("Fax: ", "");
profile.MobileNumber = this.GetLineContaining(otherDetails, "Mobile: ").Replace("Mobile: ", "");
profile.ContactName = this.GetLineContaining(otherDetails, "Contact: ").Replace("Contact: ", "");
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//*[@id='mainContainer']/h2");
profile.OfficeAddress = this.GetTextByXPath("//*[@id='companyDetails']//span[text()='Address:']/../following-sibling::td");
profile.PhoneNumber = this.GetTextByXPath("//*[@id='companyDetails']//span[text()='Tel:']/../following-sibling::td");
profile.FaxNumber = this.GetTextByXPath("//*[@id='companyDetails']//span[text()='Fax:']/../following-sibling::td");
profile.EmailAddress = this.GetTextByXPath("//*[@id='companyDetails']//span[text()='Email:']/../following-sibling::td");
profile.WebsiteAddress = this.GetTextByXPath("//*[@id='companyDetails']//span[text()='Web:']/../following-sibling::td");
profile.Category = this.GroupAllTextByXPath("//*[@id='companyDetails']//span[text()='See also:']/../following-sibling::td/a", ", ");
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
string nameOfCompanyXpath = string.Format("//*[@id='member-listing']/li[{0}]/div[1]/h3", this.CurrentResultOnPage + 1);
string emailAddressXpath = string.Format("//*[@id='member-listing']/li[{0}]/div[3]/ul/li/a[contains(.,'Email us')]", this.CurrentResultOnPage + 1);
string phoneNumberXpath = string.Format("//*[@id='member-listing']/li[{0}]/div[3]/p/span/strong", this.CurrentResultOnPage + 1);
string officeAddressXpath = string.Format("//*[@id='member-listing']/li[{0}]/div[2]/p[@class='location']", this.CurrentResultOnPage + 1);
//string aboutUsAddressXpath = string.Format("", this.CurrentResultOnPage + 1);
string servicesProvidedXPath = string.Format("//*[@id='member-listing']/li[{0}]/div[1]/p[2]/strong", this.CurrentResultOnPage + 1);
string websiteAddressXpath = string.Format("//*[@id='member-listing']/li[{0}]/div[3]/ul/li/a[contains(.,'Visit Website')]", this.CurrentResultOnPage + 1);
profile.NameOfCompany = this.GetTextByXPath(nameOfCompanyXpath);
profile.EmailAddress = this.GetAttributeValueByXPath(emailAddressXpath, "href")?.Replace("mailto:", "");
profile.PhoneNumber = this.GetTextByXPath(phoneNumberXpath);
profile.OfficeAddress = this.GetTextByXPath(officeAddressXpath);
profile.ServicesProvided = this.GetTextByXPath(servicesProvidedXPath);
profile.WebsiteAddress = this.GetAttributeValueByXPath(websiteAddressXpath, "href");
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//*[@id='njh_container']/div[3]/div/div/div[1]/h2");
List <string> address = new List <string>();
string addressLine = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Address: ']/following-sibling::td");
string county = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='County: ']/following-sibling::td");
string postcode = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Postcode: ']/following-sibling::td");
string country = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Country: ']/following-sibling::td");
if (!string.IsNullOrEmpty(addressLine))
{
address.Add(addressLine.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(county))
{
address.Add(county.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(postcode))
{
address.Add(postcode.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(country))
{
address.Add(addressLine.Trim().Trim(','));
}
profile.OfficeAddress = string.Join(", ", address);
profile.PhoneNumber = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Telephone: ']/following-sibling::td");
profile.BusinessType = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Type: ']/following-sibling::td");
profile.AreasServed = this.GetTextByXPath("//*[@id='OrderingForm']/table//tr//td//table//td[text()='Search District: ']/following-sibling::td");
profile.EmailAddress = this.GetTextByXPath("//*[@id='OrderingForm']/table//span[@class='email_l']");
profile.WebsiteAddress = this.GetTextByXPath("//*[@id='OrderingForm']/table//span[@class='website_l']");
profile.AboutUs = this.GroupAllTextByXPath("//*[@id='OrderingForm']/table/tbody//div[@class='description']/following-sibling::*");
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected void ExportTable(List <string> columns, string fileName, bool updateExportedStatus = true, List <string> headings = null)
{
using (var ctx = new ScrapedItem())
{
Trace.TraceInformation("Exporting results to CSV file " + fileName);
// Loop through each row in database which isn't deleted and hasn't got a don't export flag
var exportResults = ctx.BuildUppProfiles.Where(p => p.IsDeletedFlag != true && p.DoNotExportFlag != true && p.WebsiteScraped == WebsiteName);
// Write it to the CSV file specified for only specific columns
using (StreamWriter outfile = new StreamWriter(fileName))
{
headings = headings ?? columns;
outfile.WriteLine(
string.Join(",", headings)
);
foreach (var profile in exportResults)
{
List <string> rowValues = new List <string>();
// Loop over each colum name and get a list of the values
foreach (var column in columns)
{
rowValues.Add(
"\"" + this.ReplaceSafe(Convert.ToString(this.GetPropertyValue(profile, column))) + "\""
);
}
// Print the row of values
outfile.WriteLine(string.Join(",", rowValues));
if (updateExportedStatus)
{
profile.HasBeenExportedFlag = true;
}
}
}
// Update field to say they've exported
if (updateExportedStatus)
{
ctx.SaveChanges();
}
// Log the number of results exported
Trace.TraceInformation("CSV file exported with " + exportResults.Count() + " results to " + fileName);
}
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("/html/body/div/section[1]/div/div/div/div/h2");
profile.EmailAddress = this.GetTextByXPath("//*[@id='member-info']/tbody/tr/th[text()='Email']/following-sibling::td/a");
profile.PhoneNumber = this.GetTextByXPath("//*[@id='member-info']/tbody/tr/th[text()='Telephone']/following-sibling::td");
profile.WebsiteAddress = this.GetTextByXPath("//*[@id='member-info']/tbody/tr/th[text()='Website']/following-sibling::td/a");
var addressLines = new List <string>();
var addressLinesCorrected = new List <string>();
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[5]/td"));
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[7]/td"));
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[9]/td"));
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[11]/td"));
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[13]/td"));
addressLines.Add(this.GroupAllTextByXPath("//*[@id='member-info']/tbody/tr[15]/td"));
// Loop over each addressLine only, copying non blanks into corrected version
foreach (string addressLine in addressLines)
{
if (!string.IsNullOrEmpty(addressLine?.Trim()))
{
addressLinesCorrected.Add(addressLine);
}
}
profile.OfficeAddress = string.Join(", ", addressLinesCorrected);
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//h1");
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//*[@id='page_container']/header/div[3]/div/h1/span");
profile.WebsiteAddress = this.GetAttributeValueByXPath("//*[@id='page_container']/section/div/div/div[2]/div[2]/div/div/p[1]/a", "href");
profile.OfficeAddress = this.GetTextByXPath("//*[@id='page_container']/section/div/div/div[2]/div[2]/div/div/p[2]");
profile.AreasServed = this.GroupAllTextByXPath("//*[@id='page_container']/section/div/div/div[3]/div[2]/ul/li", ", ");
profile.AboutUs = this.GroupAllTextByXPath("//*[@id='page_container']/section/div/div/div[3]/div[1]/p");
profile.ContactName = this.GetTextByXPath("//*[@id='page_container']/header/div[3]/div/h1");
if (!string.IsNullOrEmpty(profile.NameOfCompany))
{
profile.ContactName = profile.ContactName.Replace(profile.NameOfCompany, "").Trim();
}
string infoBox = this.GetTextByXPath("//*[@id='page_container']/section/div/div/div[2]/div[2]/div/div/p[1]");
profile.PhoneNumber = this.GetLineContaining(infoBox, "Tel:").Replace("Tel: ", "");
try
{
profile.PageURL = driver.Url;
}
catch (Exception e) { };
try
{
profile.WebsiteScraped = WebsiteName;
}
catch (Exception e) { };
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
/// <summary>
/// Loop over all profiles in the database and crawl for email addresses
/// </summary>
public void LoopOverResults()
{
IQueryable <BuildUppProfile> profiles;
List <BuildUppProfile> profilesList;
using (var ctx = new ScrapedItem())
{
profiles = ctx.BuildUppProfiles.Where(m => m.Id >= this.idToStart && m.WebsiteAddress != null && m.EmailAddress == null && m.HasEmailBeenChecked != true).Take(100);
profilesList = profiles.ToList();
}
foreach (var profile in profilesList)
{
Trace.TraceInformation("About to attempt to get URL for {0}", profile.NameOfCompany);
string email = "";
try
{
email = this.GetEmailFromWebsite(profile.WebsiteAddress);
}
catch (Exception e) {
Trace.TraceInformation("Unexpected Exception thrown: {0}", e.ToString());
}
string nameOfCompany = profile.NameOfCompany;
using (var ctx = new ScrapedItem())
{
//update all entries with that name of company to email address
foreach (var profileToUpdate in ctx.BuildUppProfiles.Where(m => m.NameOfCompany == nameOfCompany))
{
profileToUpdate.EmailAddress = email;
// if not found then update the EmailCheckedFlag
profileToUpdate.HasEmailBeenChecked = true;
}
ctx.SaveChanges();
}
}
}
protected override void ExtractItemFromPage()
{
Trace.TraceInformation("Extracting infomation from " + driver.Url);
using (var ctx = new ScrapedItem())
{
var profile = new BuildUppProfile();
profile.NameOfCompany = this.GetTextByXPath("//*[@id='block-system-main']/div/div/table/tbody/tr/td/div[1]/h1");
profile.ContactName = this.GetTextByXPath("//div[contains(@class,'views-field')]/strong[text()='Contact Name: ']/following-sibling::span");
profile.PhoneNumber = this.GetTextByXPath("//div[contains(@class,'views-field')]/strong[text()='Telephone: ']/following-sibling::span");
profile.EmailAddress = this.GetTextByXPath("//div[contains(@class,'views-field')]/strong[text()='Email Address: ']/following-sibling::span");
profile.WebsiteAddress = this.GetTextByXPath("//div[contains(@class,'views-field')]/strong[text()='Website: ']/following-sibling::span");
profile.OfficeAddress = this.GetTextByXPath("//span[contains(@class,'views-field')]/strong[text()='Contact Name: ']/following-sibling::span");
profile.AboutUs = this.GetTextByXPath("//span[contains(@class, 'field-content')]/strong[text()='Company Information:']/../..");
profile.AboutUs = string.IsNullOrEmpty(profile.AboutUs) ? null : profile.AboutUs.Replace("Company Information:", "").Trim();
profile.ServicesProvided = this.GroupAllTextByXPath("//*[@id='buyersGuideCompany_productsAndServices']/li");
List <string> address = new List <string>();
string streetAddress = this.GetTextByXPath("//span[contains(@class, 'views-field-street-address')]/span");
string addressLine1 = this.GetTextByXPath("//span[contains(@class, 'views-field-supplemental-address-1')]/span");
string city = this.GetTextByXPath("//span[contains(@class, 'views-field-city')]/span");
string province = this.GetTextByXPath("//span[contains(@class, 'views-field-state-province')]/span");
string postcode = this.GetTextByXPath("//span[contains(@class, 'views-field-postal-code')]/span");
string country = this.GetTextByXPath("//span[contains(@class, 'views-field-country')]/span");
if (!string.IsNullOrEmpty(streetAddress))
{
address.Add(streetAddress.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(addressLine1))
{
address.Add(addressLine1.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(city))
{
address.Add(city.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(province))
{
address.Add(province.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(postcode))
{
address.Add(postcode.Trim().Trim(','));
}
if (!string.IsNullOrEmpty(country))
{
address.Add(country.Trim().Trim(','));
}
profile.OfficeAddress = string.Join(", ", address);
profile.PageURL = driver.Url;
profile.WebsiteScraped = WebsiteName;
ctx.BuildUppProfiles.Add(profile);
ctx.SaveChanges();
Trace.TraceInformation("Completed scraping infomation from " + driver.Url);
};
}
protected void StandardClean(bool RemoveExactDuplicates)
{
Trace.TraceInformation("Cleaning results data");
Trace.TraceInformation("Remove duplicate profiles");
using (var ctx = new ScrapedItem())
{
// Loop over each row and delete duplicates by adding isDeleted flag
var duplicateProfiles = from r in ctx.BuildUppProfiles
group r by new
{
NameOfCompany = r.NameOfCompany,
OfficeAddress = r.OfficeAddress,
PhoneNumber = r.PhoneNumber,
EmailAddress = r.EmailAddress,
WebsiteScraped = this.WebsiteName
}
into g
where g.Count() > 1
select g;
foreach (var g in duplicateProfiles)
{
var removeProfiles = g.Skip(1);
foreach (var record in removeProfiles)
{
record.IsDeletedFlag = true;
}
}
// Add do not expoert for profiles that do not meet the basic requirment
var invalidProfiles = ctx.BuildUppProfiles.Where(p =>
String.IsNullOrEmpty(p.NameOfCompany) ||
String.IsNullOrEmpty(p.EmailAddress) ||
String.IsNullOrEmpty(p.OfficeAddress)
);
foreach (var profile in invalidProfiles)
{
profile.DoNotExportFlag = true;
}
Trace.TraceInformation("do not export flag set for multiple company name listings");
// Add the do not export flag for duplicate names
var multipleAddressProfiles = from r in ctx.BuildUppProfiles
where r.IsDeletedFlag != true
group r by new
{
NameOfCompany = r.NameOfCompany,
}
into g
where g.Count() > 1
select g;
foreach (var g in multipleAddressProfiles)
{
var removeProfiles = g.Skip(1);
foreach (var record in removeProfiles)
{
if (record.IsDeletedFlag != true)
{
record.DoNotExportFlag = true;
}
}
}
ctx.SaveChanges();
}
}
