public IDictionary <long, SortedList <long, byte> > Parse(
string collectionName, HttpRequest request, ReadSession readSession, SessionFactory sessionFactory)
{
string[] fields;
var docs = new Dictionary <long, SortedList <long, byte> >();
if (request.Query.ContainsKey("fields"))
{
fields = request.Query["fields"].ToArray();
}
else
{
fields = new[] { "title", "body" };
}
var phrase = request.Query["q"];
foreach (var field in fields)
{
var keyId = sessionFactory.GetKeyId(collectionName.ToHash(), field.ToLower().ToHash());
var vector = BOWWriteSession.CreateDocumentVector(phrase, readSession.CreateIndexReader(keyId), _tokenizer);
docs.Add(keyId, vector);
}
return(docs);
}
public Query Parse(
ulong collectionId,
HttpRequest request,
ReadSession mappingSession)
{
var query = _httpQueryParser.Parse(collectionId, request);
mappingSession.Map(query);
return(query);
}
public Query Parse(
ulong collectionId,
HttpRequest request,
ReadSession readSession,
IStringModel tokenizer)
{
var query = _httpQueryParser.Parse(collectionId, tokenizer, request);
readSession.Map(query);
return(query);
}
public OptimizeSession(
string collectionName,
ulong collectionId,
SessionFactory sessionFactory,
IConfigurationProvider config,
ConcurrentDictionary <long, NodeReader> indexReaders) : base(collectionName, collectionId, sessionFactory)
{
_config = config;
_readSession = new ReadSession(collectionName, collectionId, sessionFactory, config, indexReaders);
_postingsWriter = new RemotePostingsWriter(config, collectionName);
}
public ReadSession CreateReadSession(ulong collectionId)
{
var sess = new ReadSession(_dir, collectionId, this)
{
ValueStream = _valueStream,
KeyStream = CreateReadStream(string.Format("{0}.key", collectionId)),
DocStream = CreateReadStream(string.Format("{0}.docs", collectionId)),
ValueIndexStream = _valueIndexStream,
KeyIndexStream = CreateReadStream(string.Format("{0}.kix", collectionId)),
DocIndexStream = CreateReadStream(string.Format("{0}.dix", collectionId)),
PostingsStream = CreateReadStream(string.Format("{0}.pos", collectionId)),
VectorStream = CreateReadStream(string.Format("{0}.vec", collectionId)),
Index = GetIndex(collectionId)
};
return(sess);
}
private ReadResult Reduce(IDictionary <long, SortedList <long, byte> > query, ReadSession readSession, int skip, int take)
{
IDictionary <long, BOCHit> scored = new Dictionary <long, BOCHit>();
foreach (var term in query)
{
var hit = Scan(term.Key, term.Value);
BOCHit score;
if (scored.TryGetValue(hit.PostingsOffsets.PostingsOffset, out score))
{
scored[hit.Key].Score = score.Score + hit.Value.Score;
}
else
{
scored.Add(hit.Key, hit.Value);
}
}
var sortedHits = scored.Values.OrderByDescending(h => h.Score);
var offsets = sortedHits.Select(h => h.PostingsOffset).ToArray();
var docIds = _postingsReader.Read(skip, take, offsets);
var window = docIds.GroupBy(x => x).Select(x => (x.Key, x.Count()))
.OrderByDescending(x => x.Item2)
.Skip(skip)
.Take(take)
.Select(x => x.Key).ToList();
var docs = readSession.ReadDocs(window);
return(new ReadResult {
Docs = docs, Total = docIds.Count
});
}
public ReadResult Read(IDictionary <long, SortedList <long, byte> > query, ReadSession readSession, int skip, int take)
{
return(Reduce(query, readSession, skip, take));
}
