public async Task GetProspectsForTrade()
{
var trade = new Trade()
{
Id = "-12039",
Name = "Handyman Services"
};
var ha = new HomeAdvisor();
var prospects = await ha.GetProspectsForTrade(trade, "10707");
Assert.NotNull(prospects);
Assert.NotEqual(0, prospects.Count);
}
public void GetProspectsForTradesWithZipCodes()
{
var tradeHandyman = new Trade()
{
Id = "-12039",
Name = "Handyman Services"
};
var tradePlumbing = new Trade()
{
Id = "-12058",
Name = "Plumbing"
};
var trades = new List<Trade>();
trades.Add(tradeHandyman);
trades.Add(tradePlumbing);
var zipCodes = new List<string>();
zipCodes.Add("10707");
zipCodes.Add("11782");
var ha = new HomeAdvisor();
var results = ha.GetProspectsForTradesWithZipCodes(trades, zipCodes);
using (var file = new System.IO.StreamWriter("prospects.txt"))
{
foreach (var result in results)
{
file.WriteLine($"{result.Name} - {result.Service.Name} - {result.PostalCode} - {result.StreetAddress}");
}
}
Assert.NotNull(results);
Assert.NotEqual(0, results.Count);
}
public async Task<List<Prospect>> GetProspectsForTrade(Trade trade, string postalCode)
{
List<Prospect> prospects = null;
// URL to POST to, requiring following form data:
// - findContractor: searchByZip
// - catOid: [trade ID here]
// - zip: [zip code here]
var values = new List<KeyValuePair<string, string>>();
values.Add(new KeyValuePair<string,string>("findContractor", "searchByZip"));
values.Add(new KeyValuePair<string,string>("catOid", trade.Id));
values.Add(new KeyValuePair<string,string>("zip", postalCode));
using (var postContent = new FormUrlEncodedContent(values))
using (var client = new HttpClient())
using (var response = await client.PostAsync(_proReviewsUrl, postContent))
using (var content = response.Content)
{
// Now have first page with the review content
var xhtml = await content.ReadAsStringAsync();
// Parse the XHTML result
var doc = new HtmlDocument();
doc.LoadHtml(xhtml);
// Look for the review pane containing prospects
var reviewPane = doc.DocumentNode.SelectSingleNode("//div[@class='xmd-content-main']");
var prospectDivClass = string.Format(".//div[@data-srzip='{0}']", postalCode);
foreach (var child in reviewPane.SelectNodes(prospectDivClass))
{
var prospect = new Prospect()
{
Service = trade,
RatingAvailable = false,
Id = child.Attributes["data-spid"].Value
};
// Parse the itemprops
foreach (var span in child.SelectNodes(".//span[@itemprop]"))
{
var attrValue = span.GetAttributeValue("itemprop", "").ToLower();
switch (attrValue)
{
case "name":
prospect.Name = span.InnerText;
break;
case "telephone":
prospect.Phone = span.InnerText;
break;
case "streetaddress":
prospect.StreetAddress = span.InnerText;
break;
case "addresslocality":
prospect.AddressLocality = span.InnerText;
break;
case "addressregion":
prospect.AddressRegion = span.InnerText;
break;
case "postalcode":
prospect.PostalCode = span.InnerText;
break;
default:
break;
}
}
// Get the URL for the detail page of the prospect
var urlNode = child.SelectNodes(".//a[@itemprop='url']").FirstOrDefault();
if (urlNode != null)
{
prospect.Url = new Uri(UrlCombine(_rootUrl, urlNode.Attributes["href"].Value));
}
var ratingRefNodes = child.SelectNodes(".//div[@class='l-column ratings-reference']/div");
if (ratingRefNodes != null && ratingRefNodes.First().Attributes["class"].Value.Equals("t-stars-small t-stars-rating"))
{
prospect.RatingAvailable = true;
// Parse star percentage
var reviewRatingNode = ratingRefNodes[0].ChildNodes.Where(x => x.Name.Equals("div")).FirstOrDefault();
if (reviewRatingNode != null)
{
// TODO: Move cleaning double to utility class
// Replace non digit and non decimal characters with nothing
var ratingCleaned = Regex.Replace(reviewRatingNode.Attributes["style"].Value, "[^0-9.]", "");
var starRating = 0.0;
if (double.TryParse(ratingCleaned, out starRating))
{
prospect.StarRating = starRating;
}
}
// Parse review count
var reviewCountNode = ratingRefNodes.Where(x => x.Attributes["class"].Value.Contains("verified-reviews")).FirstOrDefault();
if (reviewCountNode != null)
{
// TODO: Move cleaning integer to utility class
// Replace non digit characters with nothing
var reviewCountCleaned = Regex.Replace(reviewCountNode.InnerText, "[^0-9]", "");
var reviewCount = 0;
if (int.TryParse(reviewCountCleaned, out reviewCount))
{
prospect.ReviewCount = reviewCount;
}
}
}
// Finished parsing itemprops, add the object to the list
if (prospects == null) prospects = new List<Prospect>();
prospects.Add(prospect);
}
}
return prospects;
}
