/// <summary>
/// Split source index into multiple parts. </summary>
/// <param name="version">lucene compatibility version</param>
/// <param name="in"> source index, can have deletions, can have
/// multiple segments (or multiple readers). </param>
/// <param name="outputs"> list of directories where the output parts will be stored. </param>
/// <param name="seq"> if true, then the source index will be split into equal
/// increasing ranges of document id-s. If false, source document id-s will be
/// assigned in a deterministic round-robin fashion to one of the output splits. </param>
/// <exception cref="IOException"> If there is a low-level I/O error </exception>
public virtual void Split(LuceneVersion version, IndexReader @in, Store.Directory[] outputs, bool seq)
{
if (outputs == null || outputs.Length < 2)
{
throw new IOException("Invalid number of outputs.");
}
if (@in == null || @in.NumDocs < 2)
{
throw new IOException("Not enough documents for splitting");
}
int numParts = outputs.Length;
// wrap a potentially read-only input
// this way we don't have to preserve original deletions because neither
// deleteDocument(int) or undeleteAll() is applied to the wrapped input index.
FakeDeleteIndexReader input = new FakeDeleteIndexReader(@in);
int maxDoc = input.MaxDoc;
int partLen = maxDoc / numParts;
for (int i = 0; i < numParts; i++)
{
input.UndeleteAll();
if (seq) // sequential range
{
int lo = partLen * i;
int hi = lo + partLen;
// below range
for (int j = 0; j < lo; j++)
{
input.DeleteDocument(j);
}
// above range - last part collects all id-s that remained due to
// integer rounding errors
if (i < numParts - 1)
{
for (int j = hi; j < maxDoc; j++)
{
input.DeleteDocument(j);
}
}
}
else
{
// round-robin
for (int j = 0; j < maxDoc; j++)
{
if ((j + numParts - i) % numParts != 0)
{
input.DeleteDocument(j);
}
}
}
using (IndexWriter w = new IndexWriter(outputs[i],
new IndexWriterConfig(version, null)
{
OpenMode = OpenMode.CREATE
}))
{
Console.Error.WriteLine("Writing part " + (i + 1) + " ...");
// pass the subreaders directly, as our wrapper's numDocs/hasDeletetions are not up-to-date
IList <IndexReader> sr = input.GetSequentialSubReaders();
w.AddIndexes(sr.ToArray()); // TODO: maybe take List<IR> here?
}
}
Console.Error.WriteLine("Done.");
}
/// <summary>
/// Split source index into multiple parts. </summary>
/// <param name="in"> source index, can have deletions, can have
/// multiple segments (or multiple readers). </param>
/// <param name="outputs"> list of directories where the output parts will be stored. </param>
/// <param name="seq"> if true, then the source index will be split into equal
/// increasing ranges of document id-s. If false, source document id-s will be
/// assigned in a deterministic round-robin fashion to one of the output splits. </param>
/// <exception cref="IOException"> If there is a low-level I/O error </exception>
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void split(org.apache.lucene.util.Version version, IndexReader in, org.apache.lucene.store.Directory[] outputs, boolean seq) throws java.io.IOException
public virtual void Split(Version version, IndexReader @in, Directory[] outputs, bool seq)
{
if (outputs == null || outputs.Length < 2)
{
throw new IOException("Invalid number of outputs.");
}
if (@in == null || @in.numDocs() < 2)
{
throw new IOException("Not enough documents for splitting");
}
int numParts = outputs.Length;
// wrap a potentially read-only input
// this way we don't have to preserve original deletions because neither
// deleteDocument(int) or undeleteAll() is applied to the wrapped input index.
FakeDeleteIndexReader input = new FakeDeleteIndexReader(@in);
int maxDoc = input.maxDoc();
int partLen = maxDoc / numParts;
for (int i = 0; i < numParts; i++)
{
input.undeleteAll();
if (seq) // sequential range
{
int lo = partLen * i;
int hi = lo + partLen;
// below range
for (int j = 0; j < lo; j++)
{
input.deleteDocument(j);
}
// above range - last part collects all id-s that remained due to
// integer rounding errors
if (i < numParts - 1)
{
for (int j = hi; j < maxDoc; j++)
{
input.deleteDocument(j);
}
}
}
else
{
// round-robin
for (int j = 0; j < maxDoc; j++)
{
if ((j + numParts - i) % numParts != 0)
{
input.deleteDocument(j);
}
}
}
IndexWriter w = new IndexWriter(outputs[i], new IndexWriterConfig(version, null)
.setOpenMode(OpenMode.CREATE));
Console.Error.WriteLine("Writing part " + (i + 1) + " ...");
// pass the subreaders directly, as our wrapper's numDocs/hasDeletetions are not up-to-date
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.List<? extends FakeDeleteAtomicIndexReader> sr = input.getSequentialSubReaders();
//JAVA TO C# CONVERTER TODO TASK: Java wildcard generics are not converted to .NET:
IList <?> sr = input.SequentialSubReaders;
w.addIndexes(sr.ToArray()); // TODO: maybe take List<IR> here?
w.close();
}
Console.Error.WriteLine("Done.");
}