| 1 | |
|
| 2 | |
|
| 3 | |
package org.galagosearch.core.index; |
| 4 | |
|
| 5 | |
import java.io.FileNotFoundException; |
| 6 | |
import java.io.IOException; |
| 7 | |
import java.io.OutputStream; |
| 8 | |
import org.galagosearch.tupleflow.Utility; |
| 9 | |
import org.galagosearch.tupleflow.InputClass; |
| 10 | |
import org.galagosearch.tupleflow.Parameters; |
| 11 | |
import org.galagosearch.tupleflow.TupleFlowParameters; |
| 12 | |
import org.galagosearch.tupleflow.execution.Verified; |
| 13 | |
import org.galagosearch.core.types.NumberedValuedExtent; |
| 14 | |
|
| 15 | |
|
| 16 | |
|
| 17 | |
|
| 18 | |
|
| 19 | |
@InputClass(className = "org.galagosearch.core.types.NumberedValuedExtent", order = {"+extentName", "+number", "+begin"}) |
| 20 | |
@Verified |
| 21 | 4 | public class ExtentValueIndexWriter implements NumberedValuedExtent.ExtentNameNumberBeginOrder.ShreddedProcessor { |
| 22 | 0 | long minimumSkipListLength = 2048; |
| 23 | 0 | int skipByteLength = 128; |
| 24 | |
byte[] lastWord; |
| 25 | 0 | long lastPosition = 0; |
| 26 | 0 | long lastDocument = 0; |
| 27 | |
ExtentListBuffer invertedList; |
| 28 | |
OutputStream output; |
| 29 | |
long filePosition; |
| 30 | |
IndexWriter writer; |
| 31 | 0 | long documentCount = 0; |
| 32 | 0 | long collectionLength = 0; |
| 33 | |
Parameters header; |
| 34 | |
|
| 35 | |
|
| 36 | |
|
| 37 | |
|
| 38 | 0 | public ExtentValueIndexWriter(TupleFlowParameters parameters) throws FileNotFoundException, IOException { |
| 39 | 0 | writer = new IndexWriter(parameters); |
| 40 | 0 | writer.getManifest().add("readerClass", ExtentIndexReader.class.getName()); |
| 41 | 0 | writer.getManifest().add("writerClass", getClass().toString()); |
| 42 | 0 | header = parameters.getXML(); |
| 43 | 0 | } |
| 44 | |
|
| 45 | |
public void processExtentName(byte[] wordBytes) throws IOException { |
| 46 | 0 | if (invertedList != null) { |
| 47 | 0 | invertedList.close(); |
| 48 | 0 | writer.add(invertedList); |
| 49 | 0 | invertedList = null; |
| 50 | |
} |
| 51 | |
|
| 52 | 0 | invertedList = new ExtentListBuffer(); |
| 53 | 0 | invertedList.setWord(wordBytes); |
| 54 | |
|
| 55 | 0 | assert lastWord == null || 0 != Utility.compare(lastWord, wordBytes) : "Duplicate word"; |
| 56 | 0 | lastWord = wordBytes; |
| 57 | 0 | } |
| 58 | |
|
| 59 | |
public void processNumber(long document) { |
| 60 | 0 | invertedList.addDocument(document); |
| 61 | 0 | } |
| 62 | |
|
| 63 | |
public void processBegin(int begin) throws IOException { |
| 64 | 0 | invertedList.addBegin(begin); |
| 65 | 0 | } |
| 66 | |
|
| 67 | |
public void processTuple(int end, long value) throws IOException { |
| 68 | 0 | invertedList.setValue(value); |
| 69 | 0 | invertedList.addEnd(end); |
| 70 | 0 | } |
| 71 | |
|
| 72 | |
public void close() throws IOException { |
| 73 | 0 | if (invertedList != null) { |
| 74 | 0 | invertedList.close(); |
| 75 | 0 | writer.add(invertedList); |
| 76 | |
} |
| 77 | |
|
| 78 | 0 | writer.close(); |
| 79 | 0 | } |
| 80 | |
} |