| 1 | |
|
| 2 | |
package org.galagosearch.core.index; |
| 3 | |
|
| 4 | |
import java.io.DataInput; |
| 5 | |
import java.io.FileNotFoundException; |
| 6 | |
import java.io.IOException; |
| 7 | |
import java.util.HashMap; |
| 8 | |
import org.galagosearch.core.retrieval.*; |
| 9 | |
import org.galagosearch.core.retrieval.query.Node; |
| 10 | |
import org.galagosearch.core.retrieval.query.NodeType; |
| 11 | |
import org.galagosearch.core.retrieval.structured.*; |
| 12 | |
import org.galagosearch.core.util.ExtentArray; |
| 13 | |
import org.galagosearch.tupleflow.DataStream; |
| 14 | |
import org.galagosearch.tupleflow.VByteInput; |
| 15 | |
|
| 16 | |
|
| 17 | |
|
| 18 | |
|
| 19 | |
|
| 20 | 12 | public class ExtentIndexReader implements StructuredIndexPartReader { |
| 21 | |
public class Iterator extends ExtentIterator implements IndexIterator { |
| 22 | |
IndexReader.Iterator iterator; |
| 23 | |
DataInput stream; |
| 24 | |
int documentCount; |
| 25 | |
int options; |
| 26 | |
boolean done; |
| 27 | |
int document; |
| 28 | |
ExtentArray extents; |
| 29 | |
int termDocs; |
| 30 | |
|
| 31 | 12 | public Iterator(IndexReader.Iterator iterator) throws IOException { |
| 32 | 12 | this.iterator = iterator; |
| 33 | 12 | this.extents = new ExtentArray(); |
| 34 | 12 | this.done = false; |
| 35 | 12 | loadIndex(); |
| 36 | 12 | } |
| 37 | |
|
| 38 | |
public void loadIndex() throws IOException { |
| 39 | 12 | readHeader(iterator.getValueStream()); |
| 40 | 12 | } |
| 41 | |
|
| 42 | |
public void reset() throws IOException { |
| 43 | 0 | done = false; |
| 44 | |
|
| 45 | 0 | document = 0; |
| 46 | 0 | extents.reset(); |
| 47 | 0 | documentCount = 0; |
| 48 | 0 | termDocs = 0; |
| 49 | |
|
| 50 | 0 | loadIndex(); |
| 51 | 0 | } |
| 52 | |
|
| 53 | |
public String getRecordString() { |
| 54 | 0 | StringBuilder builder = new StringBuilder(); |
| 55 | 0 | builder.append(iterator.getKey()); |
| 56 | 0 | builder.append(","); |
| 57 | 0 | builder.append(document); |
| 58 | 0 | for (int i = 0; i < extents.getPosition(); ++i) { |
| 59 | 0 | builder.append(",("); |
| 60 | 0 | builder.append(extents.getBuffer()[i].begin); |
| 61 | 0 | builder.append(","); |
| 62 | 0 | builder.append(extents.getBuffer()[i].end); |
| 63 | 0 | builder.append(")"); |
| 64 | |
} |
| 65 | 0 | return builder.toString(); |
| 66 | |
} |
| 67 | |
|
| 68 | |
public boolean nextRecord() throws IOException { |
| 69 | 0 | nextDocument(); |
| 70 | 0 | if (!isDone()) return true; |
| 71 | 0 | if (iterator.nextKey()) { |
| 72 | 0 | loadIndex(); |
| 73 | 0 | return true; |
| 74 | |
} else { |
| 75 | 0 | return false; |
| 76 | |
} |
| 77 | |
} |
| 78 | |
|
| 79 | |
private void readHeader(DataStream compressedStream) throws IOException { |
| 80 | 12 | stream = new VByteInput(compressedStream); |
| 81 | |
|
| 82 | 12 | options = stream.readInt(); |
| 83 | 12 | documentCount = stream.readInt(); |
| 84 | 12 | termDocs = 0; |
| 85 | 12 | done = false; |
| 86 | |
|
| 87 | 12 | nextDocument(); |
| 88 | 12 | } |
| 89 | |
|
| 90 | |
public void nextDocument() throws IOException { |
| 91 | 32 | extents.reset(); |
| 92 | |
|
| 93 | 32 | if (termDocs >= documentCount) { |
| 94 | 12 | done = true; |
| 95 | 12 | return; |
| 96 | |
} |
| 97 | |
|
| 98 | 20 | termDocs++; |
| 99 | 20 | int deltaDocument = stream.readInt(); |
| 100 | 20 | document += deltaDocument; |
| 101 | |
|
| 102 | 20 | int extentCount = stream.readInt(); |
| 103 | 20 | int begin = 0; |
| 104 | |
|
| 105 | 48 | for (int i = 0; i < extentCount; i++) { |
| 106 | 28 | int deltaBegin = stream.readInt(); |
| 107 | 28 | int extentLength = stream.readInt(); |
| 108 | 28 | long value = stream.readLong(); |
| 109 | |
|
| 110 | 28 | begin = deltaBegin + begin; |
| 111 | 28 | int end = begin + extentLength; |
| 112 | |
|
| 113 | 28 | extents.add(document, begin, end, value); |
| 114 | |
} |
| 115 | 20 | } |
| 116 | |
|
| 117 | |
public int document() { |
| 118 | 20 | return document; |
| 119 | |
} |
| 120 | |
|
| 121 | |
public int count() { |
| 122 | 0 | return extents.getPosition(); |
| 123 | |
} |
| 124 | |
|
| 125 | |
public ExtentArray extents() { |
| 126 | 12 | return extents; |
| 127 | |
} |
| 128 | |
|
| 129 | |
public boolean isDone() { |
| 130 | 48 | return done; |
| 131 | |
} |
| 132 | |
} |
| 133 | |
IndexReader reader; |
| 134 | |
|
| 135 | 24 | public ExtentIndexReader(IndexReader reader) throws FileNotFoundException, IOException { |
| 136 | 24 | this.reader = reader; |
| 137 | 24 | } |
| 138 | |
|
| 139 | |
public Iterator getIterator() throws IOException { |
| 140 | 0 | return new Iterator(reader.getIterator()); |
| 141 | |
} |
| 142 | |
|
| 143 | |
public Iterator getExtents(String term) throws IOException { |
| 144 | 12 | IndexReader.Iterator iterator = reader.getIterator(term); |
| 145 | |
|
| 146 | 12 | if (iterator != null) { |
| 147 | 12 | return new Iterator(iterator); |
| 148 | |
} |
| 149 | 0 | return null; |
| 150 | |
} |
| 151 | |
|
| 152 | |
public CountIterator getCounts(String term) throws IOException { |
| 153 | 0 | IndexReader.Iterator iterator = reader.getIterator(term); |
| 154 | |
|
| 155 | 0 | if (iterator != null) { |
| 156 | 0 | return new Iterator(iterator); |
| 157 | |
} |
| 158 | 0 | return null; |
| 159 | |
} |
| 160 | |
|
| 161 | |
public void close() throws IOException { |
| 162 | 12 | reader.close(); |
| 163 | 12 | } |
| 164 | |
|
| 165 | |
public HashMap<String, NodeType> getNodeTypes() { |
| 166 | 12 | HashMap<String, NodeType> nodeTypes = new HashMap<String, NodeType>(); |
| 167 | 12 | nodeTypes.put("extents", new NodeType(Iterator.class)); |
| 168 | 12 | return nodeTypes; |
| 169 | |
} |
| 170 | |
|
| 171 | |
public IndexIterator getIterator(Node node) throws IOException { |
| 172 | 0 | if (node.getOperator().equals("extents")) { |
| 173 | 0 | return getExtents(node.getDefaultParameter()); |
| 174 | |
} else { |
| 175 | 0 | throw new UnsupportedOperationException("Node type " + node.getOperator() + |
| 176 | |
" isn't supported."); |
| 177 | |
} |
| 178 | |
} |
| 179 | |
} |