public void EmptyContentTest()
{
StringBuilder htmlString = new StringBuilder();
ArticleFinder af = new ArticleFinder();
//10 pages = 150 results
var records = af.GetArticles("arrested", pageCount);
List<string> htmlList = new List<string>();
using (WebClient client = new WebClient())
{
foreach (var item in records)
{
try
{
htmlString.Append(client.DownloadString(item.URL));
htmlList.Add(htmlString.ToString());
}
catch (WebException)
{
string url = item.URL;
}
finally
{
htmlString.Clear();
}
}
}
foreach (var item in htmlList)
{
HtmlDocument htmlDoc = new HtmlDocument();
htmlDoc.Load(new StringReader(htmlString.ToString()));
IEnumerable<HtmlNode> nodes;
try
{
nodes = from HtmlNode node in
htmlDoc.DocumentNode.SelectNodes("//div/article/p") //get the everything in the article
select node;
}
catch (ArgumentNullException)
{
try
{
nodes = from HtmlNode node in htmlDoc.DocumentNode.SelectNodes("//div/p") // get every p tag
select node;
}
catch (ArgumentNullException)
{
nodes = from HtmlNode node in htmlDoc.DocumentNode.DescendantsAndSelf()// get everything
select node;
}
}
Assert.IsNotEmpty(nodes);
}
}
public List<Article> GetArticleData(string searchTerm, int pages)
{
//get article finder results
ArticleFinder af = new ArticleFinder();
var finderResults = af.GetArticles(searchTerm, pages); //get the articles from the article finder
List<Article> articleList = new List<Article>();
foreach (var item in finderResults)
{
articleList.Add(GenerateArticle(item));
}
return articleList;
}
public void NoDuplicates()
{
ArticleFinder af = new ArticleFinder();
//10 pages = 150 results
var records = af.GetArticles("arrested", pageCount);
var dupliactes = from a in records
group a by a.BingArticleID into grouped
where grouped.Count() > 1
select grouped.Key;
Assert.Less(dupliactes.Count(), 1);
}
