/*
Copyright (C) SYSTAP, LLC DBA Blazegraph 2006-2016. All rights reserved.
Contact:
SYSTAP, LLC DBA Blazegraph
2501 Calvert ST NW #106
Washington, DC 20008
licenses@blazegraph.com
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
package com.bigdata.search;
import java.io.IOException;
import java.io.ObjectInput;
import java.io.ObjectOutput;
import org.apache.log4j.Logger;
import com.bigdata.btree.IIndex;
import com.bigdata.btree.proc.AbstractKeyArrayIndexProcedure;
import com.bigdata.btree.proc.AbstractKeyArrayIndexProcedureConstructor;
import com.bigdata.btree.proc.IParallelizableIndexProcedure;
import com.bigdata.btree.proc.IResultHandler;
import com.bigdata.btree.proc.LongAggregator;
import com.bigdata.btree.raba.IRaba;
import com.bigdata.btree.raba.codec.IRabaCoder;
import com.bigdata.relation.IMutableRelationIndexWriteProcedure;
/**
* Writes on the text index.
*
* @author <a href="mailto:thompsonbry@users.sourceforge.net">Bryan Thompson</a>
*/
public class TextIndexWriteProc extends AbstractKeyArrayIndexProcedure<Long>
implements IParallelizableIndexProcedure<Long>,
IMutableRelationIndexWriteProcedure<Long> {
/**
*
*/
private static final long serialVersionUID = 9013449121306914750L;
private static transient final Logger log = Logger.getLogger(TextIndexWriteProc.class);
public static class IndexWriteProcConstructor extends
AbstractKeyArrayIndexProcedureConstructor<TextIndexWriteProc> {
/**
* Variant which always overwrites any existing entry. Note that you
* must still delete all entries for a document before re-indexing that
* document.
*/
public static TextIndexWriteProc.IndexWriteProcConstructor OVERWRITE = new IndexWriteProcConstructor(
true);
/**
* Variant that will not overwrite an existing entry for a
* {term,doc,field}. This is useful when you have a corpus which (a)
* only grows in size; and (b) the content of each document is
* unchanging.
*/
public static TextIndexWriteProc.IndexWriteProcConstructor NO_OVERWRITE = new IndexWriteProcConstructor(
false);
private final boolean overwrite;
/**
* Values are required.
*/
@Override
public final boolean sendValues() {
return true;
}
/**
*
* @param overwrite
*/
private IndexWriteProcConstructor(final boolean overwrite) {
this.overwrite = overwrite;
}
@Override
public TextIndexWriteProc newInstance(final IRabaCoder keySer,
final IRabaCoder valSer, final int fromIndex,
final int toIndex, final byte[][] keys, final byte[][] vals) {
return new TextIndexWriteProc(keySer, valSer, fromIndex, toIndex,
keys, vals, overwrite);
}
}
/**
* De-serialization constructor.
*/
public TextIndexWriteProc() {
}
private boolean overwrite;
protected TextIndexWriteProc(final IRabaCoder keySer,
final IRabaCoder valSer, final int fromIndex, final int toIndex,
final byte[][] keys, final byte[][] vals, final boolean overwrite) {
super(keySer, valSer, fromIndex, toIndex, keys, vals);
assert vals != null;
}
@Override
public final boolean isReadOnly() {
return false;
}
/**
* @return The #of pre-existing tuples that were updated as an
* {@link Integer}.
*/
@Override
public Long applyOnce(final IIndex ndx, final IRaba keys, final IRaba vals) {
long updateCount = 0;
final int n = keys.size();
for (int i = 0; i < n; i++) {
final byte[] key = keys.get(i);
assert key != null;
assert key.length > 0;
/*
* Note: The value MAY be used to encoded information. While it is
* no longer used to encode information in the com.bigdata.search
* package, the RDF specific full text indices still use the value.
* Therefore it now MAY be null and these asserts have been removed.
*/
final byte[] val = vals.get(i);
// assert val != null;
// assert val.length > 0;
/*
* Write on the index if (a) overwrite was specified; or (b) the
* index does not contain an entry for the key.
*
* Note: This is an optimization which avoids mutation of the btree
* when there would be no change in the data.
*/
if(overwrite) {
// overwrite.
if (ndx.insert(key, val) != null) {
updateCount++;
}
} else {
// conditional mutation.
if (ndx.putIfAbsent(key, val) != null) {
updateCount++;
}
}
// final boolean write = overwrite || !ndx.contains(key);
//
// if (write && ndx.insert(key, val) != null) {
//
// updateCount++;
//
// }
}
if (log.isInfoEnabled())
log.info("wrote " + n + " tuples of which " + updateCount
+ " were updated rows");
return updateCount;
}
@Override
protected void readMetadata(final ObjectInput in) throws IOException, ClassNotFoundException {
super.readMetadata(in);
overwrite = in.readBoolean();
}
/**
* Extended to write the {@link #overwrite} flag.
*/
@Override
protected void writeMetadata(final ObjectOutput out) throws IOException {
super.writeMetadata(out);
out.writeBoolean(overwrite);
}
/**
* Uses {@link LongAggregator} to combine the mutation counts.
*/
@Override
protected IResultHandler<Long, Long> newAggregator() {
return new LongAggregator();
}
}