| 1 | |
|
| 2 | |
|
| 3 | |
package org.galagosearch.core.parse; |
| 4 | |
|
| 5 | |
import java.io.FileNotFoundException; |
| 6 | |
import java.io.IOException; |
| 7 | |
import java.util.HashMap; |
| 8 | |
import org.galagosearch.core.index.IndexReader; |
| 9 | |
import org.galagosearch.tupleflow.DataStream; |
| 10 | |
import org.galagosearch.tupleflow.VByteInput; |
| 11 | |
|
| 12 | |
|
| 13 | |
|
| 14 | |
|
| 15 | |
|
| 16 | |
public class DocumentIndexReader { |
| 17 | |
IndexReader reader; |
| 18 | |
|
| 19 | 12 | public DocumentIndexReader(String fileName) throws FileNotFoundException, IOException { |
| 20 | 12 | reader = new IndexReader(fileName); |
| 21 | 12 | } |
| 22 | |
|
| 23 | 4 | public DocumentIndexReader(IndexReader reader) { |
| 24 | 4 | this.reader = reader; |
| 25 | 4 | } |
| 26 | |
|
| 27 | |
public void close() throws IOException { |
| 28 | 0 | reader.close(); |
| 29 | 0 | } |
| 30 | |
|
| 31 | |
public Iterator getIterator() throws IOException { |
| 32 | 12 | return new Iterator(reader.getIterator()); |
| 33 | |
} |
| 34 | |
|
| 35 | |
public Document getDocument(String key) throws IOException { |
| 36 | 0 | IndexReader.Iterator iterator = reader.getIterator(key); |
| 37 | 0 | if (iterator == null) return null; |
| 38 | 0 | return new Iterator(iterator).getDocument(); |
| 39 | |
} |
| 40 | |
|
| 41 | |
public class Iterator { |
| 42 | |
IndexReader.Iterator iterator; |
| 43 | |
|
| 44 | 12 | Iterator(IndexReader.Iterator iterator) throws IOException { |
| 45 | 12 | this.iterator = iterator; |
| 46 | 12 | } |
| 47 | |
|
| 48 | |
public void skipTo(byte[] key) throws IOException { |
| 49 | 12 | iterator.skipTo(key); |
| 50 | 12 | } |
| 51 | |
|
| 52 | |
public String getKey() { |
| 53 | 8 | return iterator.getKey(); |
| 54 | |
} |
| 55 | |
|
| 56 | |
public boolean isDone() { |
| 57 | 16 | return iterator.isDone(); |
| 58 | |
} |
| 59 | |
|
| 60 | |
public Document getDocument() throws IOException { |
| 61 | 4 | String key = iterator.getKey(); |
| 62 | 4 | DataStream stream = iterator.getValueStream(); |
| 63 | 4 | return decodeDocument(key, stream); |
| 64 | |
} |
| 65 | |
|
| 66 | |
public boolean nextDocument() throws IOException { |
| 67 | 4 | return iterator.nextKey(); |
| 68 | |
} |
| 69 | |
|
| 70 | |
Document decodeDocument(String key, DataStream stream) throws IOException { |
| 71 | 4 | VByteInput input = new VByteInput(stream); |
| 72 | 4 | Document document = new Document(); |
| 73 | |
|
| 74 | |
|
| 75 | |
|
| 76 | 4 | document.identifier = key; |
| 77 | 4 | document.text = input.readString(); |
| 78 | 4 | document.metadata = new HashMap<String, String>(); |
| 79 | |
|
| 80 | 12 | while (!stream.isDone()) { |
| 81 | 8 | String mapKey = input.readString(); |
| 82 | 8 | String mapValue = input.readString(); |
| 83 | |
|
| 84 | 8 | document.metadata.put(mapKey, mapValue); |
| 85 | 8 | } |
| 86 | |
|
| 87 | 4 | return document; |
| 88 | |
} |
| 89 | |
} |
| 90 | |
} |