/// <summary>
/// Calculates number of occurrences of each word listed in keywords Meta Tag.
/// </summary>
/// <returns>Dictionary of each word listed in keywords meta tags.</returns>
public override Dictionary <string, int> CalculateOccuranceInMetaTag()
{
if (IsFilterStopWords && stopwordDictionary == null)
{
stopwordDictionary = CoreUtil.ProcessStopWords(StopWords);
}
if (htmlPage == null)
{
var htmlWeb = new HtmlWeb();
var lastStatusCode = HttpStatusCode.OK;
htmlWeb.PostResponse = (request, response) =>
{
if (response != null)
{
lastStatusCode = response.StatusCode;
}
};
htmlPage = htmlWeb.Load(Input);
if (lastStatusCode != HttpStatusCode.OK)
{
throw new Exception($"{Constant.WEB_REQUEST_ERROR_MESSAGE} {lastStatusCode}");
}
}
var metaTagCollection = htmlPage.DocumentNode.SelectNodes("//meta");
return(CoreUtil.ProcessMetaTag(metaTagCollection, stopwordDictionary, allWordDictionary));
}
/// <summary>
/// Calculates number of occurrences of each word in a webpage.
/// </summary>
/// <returns>Dictionary of each word.</returns>
public override Dictionary <string, int> CalculateOccuranceInText()
{
if (IsFilterStopWords && stopwordDictionary == null)
{
stopwordDictionary = CoreUtil.ProcessStopWords(StopWords);
}
if (htmlPage == null)
{
var htmlWeb = new HtmlWeb();
var lastStatusCode = HttpStatusCode.OK;
htmlWeb.PostResponse = (request, response) =>
{
if (response != null)
{
lastStatusCode = response.StatusCode;
}
};
htmlPage = htmlWeb.Load(Input);
if (lastStatusCode != HttpStatusCode.OK)
{
throw new Exception($"{Constant.WEB_REQUEST_ERROR_MESSAGE} {lastStatusCode}");
}
}
var bodyText = htmlPage.DocumentNode.SelectSingleNode("//body").InnerText;
allWordDictionary = CoreUtil.ProcessInput(bodyText, stopwordDictionary);
return(allWordDictionary);
}
/// <summary>
/// Calculates number of occurrences of each word in a english text.
/// </summary>
/// <returns>Dictionary of each word.</returns>
public override Dictionary <string, int> CalculateOccuranceInText()
{
if (IsFilterStopWords && stopwordDictionary == null)
{
stopwordDictionary = CoreUtil.ProcessStopWords(StopWords);
}
return(CoreUtil.ProcessInput(Input, stopwordDictionary));
}
public void ProcessStopWord_TextWithNonLetterOrWhiteSpace_ReturnEmptyDictionary(string input)
{
//arrange
var expected = new Dictionary <string, int>(StringComparer.InvariantCultureIgnoreCase);
//act
var actual = CoreUtil.ProcessStopWords(input);
//assert
Assert.AreEqual(actual, expected);
Assert.IsEmpty(actual);
}
public void ProcessStopWord_WithEnglishText_ReturnDictionaryOfSixElements(string input)
{
//arrange
var expected = new Dictionary <string, int>(StringComparer.InvariantCultureIgnoreCase)
{
{ "don't", 1 },
{ "or", 1 },
{ "and", 1 },
{ "are", 1 },
{ "the", 1 },
{ "I'm", 1 }
};
//act
var actual = CoreUtil.ProcessStopWords(input);
//assert
Assert.AreEqual(actual, expected);
}