public void EmptyArticleContentTest()
{
string[] urls = {"http://www.news.com.au/breaking-news/world/indon-says-2-arrested-for-myanmar-plot/story-e6frfkui-1226634619523",
"http://www.elpasotimes.com/newupdated/ci_23175234/el-paso-man-arrested-charged-cocaine-possession",
"http://www.twincities.com/crime/ci_23160767/st-paul-man-arrested-after-mother-shot-is",
"http://www.news-record.com/news/1174493-91/three-arrested-in-home-invasion",
"http://www.mercurynews.com/news/ci_23159399/teacher-arrested-allegedly-molesting-girls",
"http://www.menafn.com/menafn/9c990483-5c0b-4bb9-87ff-8ade7ac1a9c2/BRIEF-Tulsa-man-arrested-on-drug-gun-complaints"};
List<Bing.NewsResult> bnrList = new List<Bing.NewsResult>();
foreach (var item in urls)
{
bnrList.Add(new Bing.NewsResult { Url = item });
}
ArticleFinder af = new ArticleFinder();
var temp = af.GetResults(bnrList);
int emptyCounter = 0;
foreach (var item in temp)
{
if (string.IsNullOrEmpty(item.Content) || string.IsNullOrWhiteSpace(item.Content))
{
emptyCounter++;
}
}
Assert.Less(emptyCounter, 1);
}
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);
}