public IDictionary <string, int> FindKeywords(string pageContent)
{
if (string.IsNullOrEmpty(pageContent))
{
return(new Dictionary <string, int>());
}
string keywordsContent = _htmlWeb.SelectNodeAttribute(pageContent, MetaXPath, ContentAttribute);
if (string.IsNullOrEmpty(keywordsContent))
{
return(new Dictionary <string, int>());
}
IEnumerable <string> keywords = keywordsContent.Split(',')
.Select(keyword => keyword[0] == ' ' ? keyword.Remove(0, 1) : keyword)
.Distinct();
string body = _htmlWeb.SelectNode(pageContent, BodyXPath);
if (string.IsNullOrEmpty(body))
{
return(new Dictionary <string, int>());
}
return(keywords
.Select(key => new { Key = key, Regex.Matches(body, key, RegexOptions.IgnoreCase | RegexOptions.Compiled).Count })
.OrderByDescending(key => key.Count)
.ToDictionary(k => k.Key, k => k.Count));
}
public void GetPageContent_KeywordsNotExist_ReturnEmptyDictionary(string keywords)
{
IHtmlWebWrapper htmlWeb = Substitute.For <IHtmlWebWrapper>();
htmlWeb.SelectNodeAttribute(Arg.Any <string>(), Arg.Any <string>(), Arg.Any <string>()).Returns(keywords);
KeywordsFinder keyworsdFinder = new KeywordsFinder(htmlWeb);
var result = keyworsdFinder.FindKeywords("Test");
result.Should().BeEmpty();
}
public void GetPageContent_KeywordsAndBodyExist_ReturnProperDictionary()
{
IHtmlWebWrapper htmlWeb = Substitute.For <IHtmlWebWrapper>();
htmlWeb.SelectNode(Arg.Any <string>(), Arg.Any <string>()).Returns("keywrod keyword1 keyword2 keyword keyword2");
htmlWeb.SelectNodeAttribute(Arg.Any <string>(), Arg.Any <string>(), Arg.Any <string>()).Returns("keyword1, keyword2, keyword1");
KeywordsFinder keyworsdFinder = new KeywordsFinder(htmlWeb);
var result = keyworsdFinder.FindKeywords("Test");
result.Should().BeEquivalentTo(new Dictionary <string, int>
{
{ "keyword2", 2 },
{ "keyword1", 1 },
});
}