public void AndMergeTest_SomeOverlap_ReturnsMatchingPostings()
{
//Test with two
IList <Posting> first = UnitTest.GeneratePostings("(1,[0,5]), (2,[10,99])");
IList <Posting> second = UnitTest.GeneratePostings("(1,[2,9]), (3,[77])");
IList <Posting> result = Merge.AndMerge(first, second);
result.Should().HaveCount(1, "because only 1 posting overlap");
//Test with list
List <IList <Posting> > list = new List <IList <Posting> > {
UnitTest.GeneratePostings("(1,[0,5]), (2,[8])"),
UnitTest.GeneratePostings("(2,[1,9])"),
UnitTest.GeneratePostings("(1,[3]), (2,[4]), (3,[9])"),
UnitTest.GeneratePostings("(2,[7]), (3,[7])")
};
result = Merge.AndMerge(list);
result.Should().HaveCount(1, "because only 1 posting overlap");
}
public void AndMergeTest_AllOverlap_ReturnsOneOfThePostingLists()
{
//Test with two
IList <Posting> first = UnitTest.GeneratePostings("(1,[0,5]), (2,[10,99])");
IList <Posting> second = UnitTest.GeneratePostings("(1,[2,9]), (2,[77])");
IList <Posting> result = Merge.AndMerge(first, second);
result.Should().HaveCount(first.Count, "because all overlaps");
//Test with list
List <IList <Posting> > list = new List <IList <Posting> > {
UnitTest.GeneratePostings("(1,[0,5]), (2,[1])"),
UnitTest.GeneratePostings("(1,[3,6]), (2,[7])"),
UnitTest.GeneratePostings("(1,[4]), (2,[3,6])"),
UnitTest.GeneratePostings("(1,[2,8]), (2,[9])")
};
result = Merge.AndMerge(list);
result.Should().HaveCount(2, "because all in the list overlaps");
}
public void AndMergeTest_NoOverlap_ReturnsEmpty()
{
//Test with two
IList <Posting> first = UnitTest.GeneratePostings("(1,[0,5]), (3,[10])");
IList <Posting> second = UnitTest.GeneratePostings("(2,[2,9]), (4,[77])");
IList <Posting> result = Merge.AndMerge(first, second);
result.Should().BeEmpty("because nothing overlaps between the two");
result.Should().HaveCount(0, "because nothing overlaps between the two");
//Test with list
List <IList <Posting> > list = new List <IList <Posting> > {
UnitTest.GeneratePostings("(1,[0,5]), (2,[4])"),
UnitTest.GeneratePostings("(2,[6]), (3,[7])"),
UnitTest.GeneratePostings("(1,[2]), (3,[1,9])"),
UnitTest.GeneratePostings("(3,[8]), (4,[6])")
};
result = Merge.AndMerge(list);
result.Should().BeEmpty("because nothing in the list overlaps");
}
/// <summary>
/// Gets Postings of documents which contain the similar sounding author names.
/// </summary>
/// <param name="nameQuery">author name to find documents</param>
/// <returns>postings with document and empty positions</returns>
public IList <Posting> GetPostings(string nameQuery)
{
IList <Posting> result = new List <Posting>();
//Check author name can contains multiple terms
string[] terms = nameQuery.Split(' ');
//remove the empry string in the string array
terms = terms.Where(x => !string.IsNullOrEmpty(x)).ToArray();
List <IList <Posting> > list = new List <IList <Posting> >();
foreach (string term in terms)
{
string soundCode = ParseToSoundex(term);
List <int> docIDs;
try
{
List <int> temp = this.map.Get(soundCode);
docIDs = temp == default(List <int>) ? new List <int>() : temp;
}
catch (KeyNotFoundException)
{
continue;
}
IList <Posting> postings = new List <Posting>();
foreach (int id in docIDs)
{
postings.Add(new Posting(id, new List <int>()));
}
list.Add(postings);
}
result = Merge.AndMerge(list);
return(result);
}