| 1 | |
|
| 2 | |
|
| 3 | |
package org.galagosearch.core.retrieval.structured; |
| 4 | |
|
| 5 | |
import java.io.FileNotFoundException; |
| 6 | |
import java.io.IOException; |
| 7 | |
import java.util.ArrayList; |
| 8 | |
import java.util.PriorityQueue; |
| 9 | |
import org.galagosearch.core.index.StructuredIndex; |
| 10 | |
import org.galagosearch.core.retrieval.query.Node; |
| 11 | |
import org.galagosearch.core.retrieval.query.StructuredQuery; |
| 12 | |
import org.galagosearch.core.retrieval.Retrieval; |
| 13 | |
import org.galagosearch.core.retrieval.ScoredDocument; |
| 14 | |
import org.galagosearch.core.retrieval.query.NodeType; |
| 15 | |
import org.galagosearch.core.retrieval.traversal.AddCombineTraversal; |
| 16 | |
import org.galagosearch.core.retrieval.traversal.ImplicitFeatureCastTraversal; |
| 17 | |
import org.galagosearch.core.retrieval.traversal.IndriWindowCompatibilityTraversal; |
| 18 | |
import org.galagosearch.core.retrieval.traversal.TextFieldRewriteTraversal; |
| 19 | |
import org.galagosearch.core.retrieval.traversal.WeightConversionTraversal; |
| 20 | |
import org.galagosearch.tupleflow.Parameters; |
| 21 | |
|
| 22 | |
|
| 23 | |
|
| 24 | |
|
| 25 | |
|
| 26 | |
public class StructuredRetrieval extends Retrieval { |
| 27 | |
StructuredIndex index; |
| 28 | |
FeatureFactory featureFactory; |
| 29 | |
|
| 30 | 12 | public StructuredRetrieval(StructuredIndex index) { |
| 31 | 12 | this.index = index; |
| 32 | 12 | Parameters featureParameters = new Parameters(); |
| 33 | 12 | featureParameters.add("collectionLength", Long.toString(index.getCollectionLength())); |
| 34 | 12 | featureParameters.add("documentCount", Long.toString(index.getDocumentCount())); |
| 35 | 12 | featureFactory = new FeatureFactory(featureParameters); |
| 36 | 12 | } |
| 37 | |
|
| 38 | |
public StructuredRetrieval(String filename) throws FileNotFoundException, IOException { |
| 39 | 8 | this(new StructuredIndex(filename)); |
| 40 | 8 | } |
| 41 | |
|
| 42 | |
public ScoredDocument[] getArrayResults(PriorityQueue<ScoredDocument> scores) { |
| 43 | 4 | ScoredDocument[] results = new ScoredDocument[scores.size()]; |
| 44 | |
|
| 45 | 24 | for (int i = scores.size() - 1; i >= 0; i--) { |
| 46 | 20 | results[i] = scores.poll(); |
| 47 | |
} |
| 48 | |
|
| 49 | 4 | return results; |
| 50 | |
} |
| 51 | |
|
| 52 | |
public NodeType getNodeType(Node node) throws Exception { |
| 53 | 0 | NodeType nodeType = index.getNodeType(node); |
| 54 | 0 | if (nodeType == null) { |
| 55 | 0 | nodeType = featureFactory.getNodeType(node); |
| 56 | |
} |
| 57 | 0 | return nodeType; |
| 58 | |
} |
| 59 | |
|
| 60 | |
public StructuredIterator createIterator(Node node) throws Exception { |
| 61 | 20 | ArrayList<StructuredIterator> internalIterators = new ArrayList<StructuredIterator>(); |
| 62 | |
|
| 63 | 20 | for (Node internalNode : node.getInternalNodes()) { |
| 64 | 16 | StructuredIterator internalIterator = createIterator(internalNode); |
| 65 | 16 | internalIterators.add(internalIterator); |
| 66 | 16 | } |
| 67 | |
|
| 68 | 20 | StructuredIterator iterator = index.getIterator(node); |
| 69 | 20 | if (iterator == null) { |
| 70 | 12 | iterator = featureFactory.getIterator(node, internalIterators); |
| 71 | |
} |
| 72 | |
|
| 73 | 20 | return iterator; |
| 74 | |
} |
| 75 | |
|
| 76 | |
public Node transformQuery(Node queryTree) throws Exception { |
| 77 | 0 | queryTree = StructuredQuery.copy(new AddCombineTraversal(), queryTree); |
| 78 | 0 | queryTree = StructuredQuery.copy(new WeightConversionTraversal(), queryTree); |
| 79 | 0 | queryTree = StructuredQuery.copy(new IndriWindowCompatibilityTraversal(), queryTree); |
| 80 | 0 | queryTree = StructuredQuery.copy(new TextFieldRewriteTraversal(index), queryTree); |
| 81 | 0 | queryTree = StructuredQuery.copy(new ImplicitFeatureCastTraversal(this), queryTree); |
| 82 | 0 | return queryTree; |
| 83 | |
} |
| 84 | |
|
| 85 | |
|
| 86 | |
|
| 87 | |
|
| 88 | |
|
| 89 | |
|
| 90 | |
|
| 91 | |
|
| 92 | |
|
| 93 | |
public ScoredDocument[] runQuery(Node queryTree, int requested) throws Exception { |
| 94 | |
|
| 95 | 4 | ScoreIterator iterator = (ScoreIterator) createIterator(queryTree); |
| 96 | |
|
| 97 | |
|
| 98 | 4 | PriorityQueue<ScoredDocument> queue = new PriorityQueue<ScoredDocument>(); |
| 99 | |
|
| 100 | 24 | while (!iterator.isDone()) { |
| 101 | 20 | int document = iterator.nextCandidate(); |
| 102 | 20 | int length = index.getLength(document); |
| 103 | 20 | double score = iterator.score(document, length); |
| 104 | |
|
| 105 | 20 | if (queue.size() <= requested || queue.peek().score < score) { |
| 106 | 20 | ScoredDocument scoredDocument = new ScoredDocument(document, score); |
| 107 | 20 | queue.add(scoredDocument); |
| 108 | |
|
| 109 | 20 | if (queue.size() > requested) { |
| 110 | 0 | queue.poll(); |
| 111 | |
} |
| 112 | |
} |
| 113 | |
|
| 114 | 20 | iterator.movePast(document); |
| 115 | 20 | } |
| 116 | |
|
| 117 | 4 | return getArrayResults(queue); |
| 118 | |
} |
| 119 | |
|
| 120 | |
public String getDocumentName(int document) { |
| 121 | 20 | return index.getDocumentName(document); |
| 122 | |
} |
| 123 | |
|
| 124 | |
public void close() throws IOException { |
| 125 | 0 | index.close(); |
| 126 | 0 | } |
| 127 | |
} |