public IMatch NextMatch()
{
if (consumed)
{
return(null);
}
var target = Occurrence.Empty;
var i = 0;
matches[i] = null;
while (i < queries.Length)
{
var newTarget = Occurrence.Empty;
while (matches[i] == null || matches[i].Right < target)
{
var m = queries[i].NextMatch();
matches[i] = m;
if (m == null)
{
consumed = true;
return(null);
}
newTarget = Occurrence.O(
Math.Max(target.DocumentId, m.Right.DocumentId),
Math.Max(target.FieldId, m.Right.FieldId),
0);
}
if (target < newTarget)
{
target = newTarget;
i = 0;
}
else if (i == 0)
{
i++;
}
else if (matches[i - 1].Right.TokenId + 1 == matches[i].Left.TokenId)
{
i++;
}
else
{
matches[i] = null;
}
}
return(new SequenceMatch(matches));
}
private void AddTerms(ulong id, IEnumerable <string> terms)
{
ulong tokenId = 0;
foreach (var term in terms)
{
List <Occurrence> postingList;
if (!temp.TryGetValue(term, out postingList))
{
postingList = new List <Occurrence>();
temp.Add(term, postingList);
}
postingList.Add(Occurrence.O(id, 1, ++tokenId));
}
}
private void AddTerms(ulong id, ulong fieldId, IEnumerable <ScopedToken> terms)
{
var positions = new List <TextPosition>();
ulong tokenId = 0;
foreach (var token in terms)
{
var term = token.AsString();
if (!temp.TryGetValue(term, out var postingList))
{
postingList = new List <Occurrence>();
temp.Add(term, postingList);
}
positions.Add(TextPosition.P(token.CharOffset, token.Length));
postingList.Add(Occurrence.O(id, fieldId, ++tokenId));
}
AddDocVector(id, fieldId, positions);
}
private static void ParseBufferTo(byte[] buffer, IList <Occurrence> occurrences)
{
var numbers = GroupVarint.Decode(buffer);
var o = new Occurrence((ulong)numbers[0],
(ulong)numbers[1],
(ulong)numbers[2]);
occurrences.Add(o);
int i = 3;
while (i < numbers.Count)
{
uint deltaSelector = (uint)numbers[i];
++i;
while (deltaSelector > 0)
{
int delta = (int)(deltaSelector & 0b00000011);
deltaSelector >>= 2;
if (i + delta > numbers.Count)
{
throw new Exception("Attempt to read above data");
}
switch (delta)
{
case 0:
{
throw new Exception("Zero delta is not used, see comments in DeltaWriter");
}
case 1:
{
o = Occurrence.O(o.DocumentId, o.FieldId, o.TokenId + (ulong)numbers[i]);
i += 1;
occurrences.Add(o);
break;
}
case 2:
{
o = Occurrence.O(o.DocumentId, o.FieldId + (ulong)numbers[i], (ulong)numbers[i + 1]);
i += 2;
occurrences.Add(o);
break;
}
case 3:
{
o = Occurrence.O(o.DocumentId + (ulong)numbers[i], (ulong)numbers[i + 1], (ulong)numbers[i + 2]);
i += 3;
occurrences.Add(o);
break;
}
default:
{
throw new Exception("Something wrong");
}
}
}
}
}