/// <summary>
/// Fetch words from a website page, count the frequencies,
/// and then put them into a list of word objects.
/// </summary>
/// <param name="url"></param>
/// <returns></returns>
public async Task <IEnumerable <HtmlWordDto> > GetWordCloudData(string url)
{
List <HtmlWordDto> wordCloudData = new List <HtmlWordDto>();
string htmlContent = await ParseHtml(url);
if (htmlContent == String.Empty)
{
return(null);
}
HtmlDocument htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(htmlContent);
HtmlNode rootNode = htmlDocument.DocumentNode;
GetWords(rootNode);
IEnumerable <KeyValuePair <string, int> > wordsCount = CountWords();
//Save words count data into database
var x = _wordsDbContext.HtmlWord.OrderByDescending(i => i.count);
foreach (var item in wordsCount)
{
wordCloudData.Add(
new HtmlWordDto()
{
Text = item.Key,
Weight = item.Value
});
var wordInDb = x.Where(w => w.Word == item.Key).FirstOrDefault();
if (wordInDb == null)
{
var hashedWord = _saltHashHelper.GenerateSaltedHash(64, item.Key);
_wordsDbContext.Add(
new HtmlWord
{
SaltedHash = hashedWord.Hash,
Word = item.Key,
count = item.Value,
Salt = hashedWord.Salt
}
);
}
else
{
wordInDb.count += item.Value;
}
}
;
_wordsDbContext.SaveChanges();
return(wordCloudData);
}
public void CreateSaltedHashShould()
{
HashSalt res = _saltHashHelper.GenerateSaltedHash(64, "Jason");
Assert.True(_saltHashHelper.VerifyStr("Jason", res));
}