public void GetWordDictRemovesCapitalization()
{
var req = new GetPageResponse {
Content = "<html><body><p>TEsT</p></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Equal(1, res["test"]);
}
public void GetWordDictPicksUpWordsInMultipleTags()
{
var req = new GetPageResponse {
Content = "<html><body><p>test</p><a>test</a><span>test</span></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Equal(3, res["test"]);
}
public void GetWordDictReturnsDictWithSingleEntry()
{
var req = new GetPageResponse {
Content = "<html><body><p>testword</p></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Equal(1, res["testword"]);
}
public void GetWordDictReturnsEmptyDictGivenNoInput()
{
var req = new GetPageResponse {
Content = ""
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Empty(res);
}
public void GetWordDictResultIsSorted()
{
var req = new GetPageResponse
{
Content = "<html><body><p>1 2 2 3 3 3</p></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Equal("3", res.First().Key);
}
public void GetWordDictRemovesSpecialChars()
{
var req = new GetPageResponse
{
Content = "<html><body><p>\n | ( ) - +</p></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Empty(res);
}
public void GetWordDictBreaksDownSentance()
{
var req = new GetPageResponse {
Content = "<html><body><p>this is a test</p></body></html>"
};
var mfw = new MostFrequentWords();
var res = mfw.GetWordDict(req);
Assert.Equal(4, res.Count);
Assert.Equal(1, res["test"]);
}
