public override void Describe(DescribeContext <Article> context)
{
var tokenizer = new WhiteSpaceTokenizer();
var filter = new StopWordFilter();
context.For <ArticleByWord, string>()
.Map(article => filter
.Filter(tokenizer.Tokenize(article.Content))
.Select(x => new ArticleByWord {
Word = x, Count = 1
})
)
.Group(article => article.Word)
.Reduce(group => new ArticleByWord
{
Word = group.Key,
Count = group.Sum(y => y.Count)
})
.Delete((index, map) =>
{
index.Count -= map.Sum(x => x.Count);
// if Count == 0 then delete the index
return(index.Count > 0 ? index : null);
});
}
public void RemovesStopWordsFromArray()
{
string[] stopWords = { "fum", "fee" };
string[] filtered = new StopWordFilter(stopWords).Filter(new string[] { "foo", "bar", "fum", "fee" });
Assert.That(filtered, Is.EqualTo(new string[] { "foo", "bar" }));
}
