public PostingArray Union(PostingArray other, bool max)
{
int min = Math.Max(this.list.Length, other.list.Length);
if (this.list.Length == 0)
{
return(other);
}
if (other.list.Length == 0)
{
return(this);
}
List <Posting> result = new List <Posting>(min);
int i0 = 0;
int i1 = 0;
Posting item0 = this.list[i0];
Posting item1 = other.list[i1];
while (i0 < this.list.Length && i1 < other.list.Length)
{
if (i0 == this.list.Length)
{
result.Add(item1);
i1 = other.Next(i1, out item1);
}
else if (i1 == other.list.Length)
{
result.Add(item0);
i0 = this.Next(i0, out item0);
}
else if (item0.Document == item1.Document)
{
if (max)
{
result.Add(new Posting(item0.Document, Math.Max(item0.Bm25F, item1.Bm25F)));
}
else
{
result.Add(new Posting(item0.Document, item0.Bm25F + item1.Bm25F));
}
i0 = this.Next(i0, out item0);
i1 = other.Next(i1, out item1);
}
else if (item0.Document.hash < item1.Document.hash)
{
result.Add(item0);
i0 = this.Next(i0, out item0);
}
else
{
result.Add(item1);
i1 = other.Next(i1, out item1);
}
}
return(new PostingArray(result.ToArray()));
}
void Resolve()
{
if (arrays.Count() == 1)
{
return;
}
arrays.Sort((x, y) => x.Count().CompareTo(y.Count()));
PostingArray array = arrays[0];
for (int i = 1; i < arrays.Count(); i++)
{
array = array.Union(arrays[i], max) as PostingArray;
}
arrays.Clear();
arrays.Add(array);
}
public PostingArray Intersection(PostingArray other, bool max)
{
int min = Math.Min(this.list.Length, other.list.Length);
if (min == 0)
{
return(new PostingArray(new Posting[0]));
}
int span0 = this.list.Length / other.list.Length;
int span1 = other.list.Length / this.list.Length;
List <Posting> result = new List <Posting>(min / 10 + 1);
int i0 = 0;
int i1 = 0;
Posting item0 = this.list[i0];
Posting item1 = other.list[i1];
while (i0 < this.list.Length && i1 < other.list.Length)
{
if (item0.Document == item1.Document)
{
if (max)
{
result.Add(new Posting(item0.Document, Math.Max(item0.Bm25F, item1.Bm25F)));
}
else
{
result.Add(new Posting(item0.Document, item0.Bm25F + item1.Bm25F));
}
i0 = this.Next(i0, out item0);
i1 = other.Next(i1, out item1);
}
else if (item0.Document.hash < item1.Document.hash)
{
i0 = this.SkipAhead(i0, span0, item1.Document, out item0);
}
else
{
i1 = other.SkipAhead(i1, span1, item0.Document, out item1);
}
}
return(new PostingArray(result.ToArray()));
}
static void Process(string src, string dst)
{
string fn = Path.GetFileName(src);
if (fn.StartsWith("_"))
{
PostingArray plist = new PostingArray(src);
plist.Sort();
plist = plist.RemoveRepeats();
plist.Write(dst);
}
else
{
File.Copy(src, dst, true);
}
sofar += new FileInfo(src).Length;
double remain = (DateTime.Now - start).TotalHours * (tot - sofar) / sofar;
Console.WriteLine($"{sofar / 1e6:0,000} MBs processed {remain:0.00} hours remaining");
}
public PostingUnion(PostingArray a)
{
this.arrays.Add(a);
}
