/**
* Returns the NGrams of the given word sequence
*
* @param wordSequence the word sequence from which to get the buffer
* @return the NGramBuffer of the word sequence
*/
private NGramBuffer GetNGramBuffer(WordSequence wordSequence)
{
NGramBuffer nGramBuffer = null;
var order = wordSequence.Size;
if (order > 1)
{
nGramBuffer = _loadedNGramBuffers[order - 1].Get(wordSequence); // better
}
// when
// using
// containsKey
if (nGramBuffer == null)
{
nGramBuffer = LoadNGramBuffer(wordSequence);
if (nGramBuffer != null)
{
Java.Put(_loadedNGramBuffers[order - 1], wordSequence, nGramBuffer); // optimizable
}
// by
// adding
// an
// 'empty'
// nGramBuffer
}
return(nGramBuffer);
}
/**
* Loads into a buffer all the NGram followers of the given N-1Gram.
*
* @param ws the N-1Gram to find followers
*
* @return a NGramBuffer of all the NGram followers of the given sequence
*/
private NGramBuffer LoadNGramBuffer(WordSequence ws)
{
var firstWordID = GetWordID(ws.GetWord(0));
int firstCurrentNGramEntry;
var numberNGrams = 0;
int size;
long position;
var orderBuffer = ws.Size + 1;
NGramBuffer currentBuffer = null;
NGramBuffer nMinus1Buffer = null;
firstCurrentNGramEntry = _unigrams[firstWordID].FirstBigramEntry;
numberNGrams = GetNumberBigramFollowers(firstWordID) + 1;
if (numberNGrams == 1) // 1 means that there is no bigram starting with
// firstWordID
{
return(null);
}
if (orderBuffer == 2)
{
size =
numberNGrams *
((_loader.MaxDepth == orderBuffer) ? BytesPerNmaxgram
: BytesPerNgram) * _loader.BytesPerField;
position = _loader.GetNGramOffset(orderBuffer)
+ firstCurrentNGramEntry
* (long)((_loader.MaxDepth == orderBuffer) ? BytesPerNmaxgram
: BytesPerNgram) * _loader.BytesPerField;
}
else // only for ws.size() >= 2
{
var lastWordId = GetWordID(ws.GetWord(ws.Size - 1));
nMinus1Buffer = GetNGramBuffer(ws.GetOldest());
var index = nMinus1Buffer.FindNGramIndex(lastWordId);
if (index == -1)
{
return(null);
}
var firstNMinus1GramEntry = nMinus1Buffer.FirstNGramEntry;
firstCurrentNGramEntry =
GetFirstNGramEntry(nMinus1Buffer.GetNGramProbability(index),
firstNMinus1GramEntry,
orderBuffer);
var firstNextNGramEntry =
GetFirstNGramEntry(nMinus1Buffer.GetNGramProbability(index + 1),
firstNMinus1GramEntry,
orderBuffer);
numberNGrams = firstNextNGramEntry - firstCurrentNGramEntry;
if (numberNGrams == 0)
{
return(null);
}
if (_loader.MaxDepth != orderBuffer)
{
numberNGrams++;
}
size = numberNGrams *
((_loader.MaxDepth == orderBuffer) ? BytesPerNmaxgram
: BytesPerNgram) * _loader.BytesPerField;
position =
_loader.GetNGramOffset(orderBuffer) +
firstCurrentNGramEntry *
(long)((_loader.MaxDepth == orderBuffer) ? BytesPerNmaxgram
: BytesPerNgram) * _loader.BytesPerField;
}
try {
var buffer = _loader.LoadBuffer(position, size);
if (_loader.MaxDepth == orderBuffer)
{
currentBuffer = new NMaxGramBuffer(buffer, numberNGrams,
_loader.GetBigEndian(), Is32Bits(), orderBuffer,
firstCurrentNGramEntry);
}
else
{
currentBuffer = new NGramBuffer(buffer, numberNGrams,
_loader.GetBigEndian(), Is32Bits(), orderBuffer,
firstCurrentNGramEntry);
}
}
catch (IOException ioe)
{
Trace.TraceError(ioe.ToString());
throw new Exception("Error loading " + orderBuffer + "-Grams.");
}
return(currentBuffer);
}
