public static IList <DocumentScore> Sum(this IList <DocumentScore> scores)
{
var sum = new List <DocumentScore>();
DocumentScore tmp = null;
foreach (var score in scores)
{
if (tmp == null)
{
tmp = score;
continue;
}
if (score.DocumentId == tmp.DocumentId)
{
tmp.Add(score);
}
else
{
sum.Add(tmp);
}
tmp = score;
}
if (tmp != null)
{
sum.Add(tmp);
}
return(sum);
}
private ScoredDocument GetDoc(DocumentScore score)
{
using (var readSession = (IFullTextReadSession)_sessionFactory.OpenReadSession(score.Ix))
{
return(readSession.ReadDocument(score));
}
}
private Document GetDoc(DocumentScore score)
{
var doc = _docReader.Get(score.DocumentId);
doc.Fields["__score"] = score.Score.ToString(CultureInfo.InvariantCulture);
return(doc);
}
public void Add(DocumentScore score)
{
if (!score.DocumentId.Equals(DocumentId))
{
throw new ArgumentException("Document IDs differ. Cannot combine.", "score");
}
Score = (Score + score.Score);
}
public ScoredDocument ReadDocument(DocumentScore score)
{
var address = AddressReader.Read(new BlockInfo(score.DocumentId * BlockSize, BlockSize));
var documentData = _documentsReader.ReadOverNetwork(address);
var documentStream = new MemoryStream(documentData);
var document = DocumentSerializer.DeserializeDocument(documentStream, address.Length, Version.Compression, KeyIndex);
document.TableId = score.DocumentId;
return(new ScoredDocument(document, score.Score));
}
public static DocumentScore TakeLatestVersion(DocumentScore first, DocumentScore second)
{
if (!first.DocHash.Equals(second.DocHash))
{
throw new ArgumentException("Document hashes differ. Cannot take latest version.", "score");
}
if (first.Ix.VersionId > second.Ix.VersionId)
{
return(first);
}
return(second);
}
public Document GetDocFromDisk(DocumentScore doc)
{
var fileId = _docIdToFileIndex[doc.DocId];
Dictionary <int, Dictionary <string, List <string> > > dics;
if (!_docFiles.TryGetValue(fileId, out dics))
{
dics = ReadDocFile(Path.Combine(_scanner.Dir, fileId + ".d"));
_docFiles.Add(fileId, dics);
}
var dic = dics[doc.DocId];
var d = Document.FromDictionary(doc.DocId, dic);
return(d);
}
public static IList <DocumentScore> Sum(this IList <DocumentScore>[] scores)
{
if (scores.Length == 0)
{
return(new DocumentScore[0]);
}
if (scores.Length == 1)
{
return(scores[0].Sum());
}
var first = scores[0];
for (int i = 1; i < scores.Length; i++)
{
first = DocumentScore.CombineAnd(first, scores[i]);
}
return(first);
}
private IEnumerable <DocumentScore> GetScoredResult(Term term, IScoringScheme scoringScheme)
{
var trie = GetTrie(term.Field);
if (_ix == null)
{
yield break;
}
var totalNumOfDocs = _ix.DocCount[term.Field];
if (trie.ContainsToken(term.Value))
{
var termData = GetPostingsFile(term.Field, term.Value);
var scorer = scoringScheme.CreateScorer(totalNumOfDocs, termData.Postings.Count);
foreach (var posting in termData.Postings)
{
var hit = new DocumentScore(posting.Key, posting.Value, totalNumOfDocs);
scorer.Score(hit);
yield return(hit);
}
}
}
public ScoredDocument ReadDocument(DocumentScore score)
{
var document = ReadDocument(score.DocumentId);
return(new ScoredDocument(document, score.Score));
}
public void Score(DocumentScore doc)
{
var tf = Math.Sqrt((int)doc.PostingData);
doc.Score = tf * _idf;
}
