public WriteSession(
ulong collectionId,
LocalStorageSessionFactory sessionFactory,
ITokenizer tokenizer) : base(collectionId, sessionFactory)
{
_tokenizer = tokenizer;
_log = Logging.CreateLogWriter("writesession");
_indexQueue = new ProducerConsumerQueue <IndexJob>(Write);
_buildQueue = new ProducerConsumerQueue <BuildJob>(Write);
ValueStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.val", collectionId)));
KeyStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.key", collectionId)));
DocStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.docs", collectionId)));
ValueIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.vix", collectionId)));
KeyIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.kix", collectionId)));
DocIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.dix", collectionId)));
PostingsStream = sessionFactory.CreateReadWriteStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.pos", collectionId)));
VectorStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.vec", collectionId)));
Index = sessionFactory.GetCollectionIndex(collectionId);
_vals = new ValueWriter(ValueStream);
_keys = new ValueWriter(KeyStream);
_docs = new DocWriter(DocStream);
_valIx = new ValueIndexWriter(ValueIndexStream);
_keyIx = new ValueIndexWriter(KeyIndexStream);
_docIx = new DocIndexWriter(DocIndexStream);
_postingsReader = new PagedPostingsReader(PostingsStream);
_dirty = new Dictionary <long, VectorNode>();
}
public WriteSession(
string collectionId,
SessionFactory sessionFactory) : base(collectionId, sessionFactory)
{
_log = Logging.CreateWriter("writesession");
var collection = collectionId.ToHash();
ValueStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.val", collection)));
KeyStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.key", collection)));
DocStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.docs", collection)));
ValueIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.vix", collection)));
KeyIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.kix", collection)));
DocIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.dix", collection)));
_vals = new ValueWriter(ValueStream);
_keys = new ValueWriter(KeyStream);
_docs = new DocWriter(DocStream);
_valIx = new ValueIndexWriter(ValueIndexStream);
_keyIx = new ValueIndexWriter(KeyIndexStream);
_docIx = new DocIndexWriter(DocIndexStream);
}
public WriteSession(ulong collectionId, LocalStorageSessionFactory sessionFactory)
: base(collectionId, sessionFactory)
{
_dirty = new Dictionary <long, VectorNode>();
ValueStream = sessionFactory.WritableValueStream;
KeyStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.key", collectionId)));
DocStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.docs", collectionId)));
ValueIndexStream = sessionFactory.WritableValueIndexStream;
KeyIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.kix", collectionId)));
DocIndexStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.dix", collectionId)));
PostingsStream = sessionFactory.CreateReadWriteStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.pos", collectionId)));
VectorStream = sessionFactory.CreateAppendStream(Path.Combine(sessionFactory.Dir, string.Format("{0}.vec", collectionId)));
Index = sessionFactory.GetIndex(collectionId);
_vals = new ValueWriter(ValueStream);
_keys = new ValueWriter(KeyStream);
_docs = new DocWriter(DocStream);
_valIx = new ValueIndexWriter(ValueIndexStream);
_keyIx = new ValueIndexWriter(KeyIndexStream);
_docIx = new DocIndexWriter(DocIndexStream);
_postingsReader = new PagedPostingsReader(PostingsStream);
}
public void Write(IEnumerable <IDictionary> data, ITokenizer tokenizer)
{
var vals = new ValueWriter(ValueStream);
var keys = new ValueWriter(KeyStream);
var docs = new DocWriter(DocStream);
var valIx = new ValueIndexWriter(ValueIndexStream);
var keyIx = new ValueIndexWriter(KeyIndexStream);
var docIx = new DocIndexWriter(DocIndexStream);
foreach (var model in data)
{
var docId = docIx.GetNextDocId();
var docMap = new List <(uint keyId, uint valId)>();
foreach (var key in model.Keys)
{
var keyStr = key.ToString();
var keyHash = keyStr.ToHash();
var fieldIndex = GetIndex(keyHash);
var val = (IComparable)model[key];
var str = val as string;
var indexTokens = new List <Term>();
uint keyId, valId;
if (str != null) //TODO: implement numeric index
{
foreach (var token in tokenizer.Tokenize(str))
{
indexTokens.Add(new Term(keyStr, token));
}
}
if (fieldIndex == null)
{
// We have a new key!
// store key
var keyInfo = keys.Append(keyStr);
keyId = keyIx.Append(keyInfo.offset, keyInfo.len, keyInfo.dataType);
SessionFactory.AddKey(keyHash, keyId);
// add new index to global in-memory tree
fieldIndex = new VectorNode();
Index.Add(keyId, fieldIndex);
}
else
{
keyId = SessionFactory.GetKey(keyHash);
}
foreach (var token in indexTokens)
{
var match = fieldIndex.ClosestMatch((string)token.Value);
if (match.Highscore < VectorNode.IdenticalAngle)
{
// We have a new unique value!
// store value
var valInfo = vals.Append(val);
valId = valIx.Append(valInfo.offset, valInfo.len, valInfo.dataType);
}
else
{
valId = match.ValueId;
}
// add posting to index
fieldIndex.Add((string)token.Value, valId, docId);
// store refs to keys and values
docMap.Add((keyId, valId));
}
var indexName = string.Format("{0}.{1}", CollectionId, keyId);
if (!_dirty.ContainsKey(indexName))
{
_dirty.Add(indexName, fieldIndex);
}
}
var docMeta = docs.Append(docMap);
docIx.Append(docMeta.offset, docMeta.length);
}
}
