Use identifiers index position as hash key to cache the mtabix block values instead of the identifiers hash. Fixes #156

jordeu · jordeu · commit 31f7315c9cd2 · 2016-02-02T10:31:08.000+01:00
diff --git a/org.gitools.matrix/src/main/java/org/gitools/matrix/model/mtabixmatrix/MTabixBlockValues.java b/org.gitools.matrix/src/main/java/org/gitools/matrix/model/mtabixmatrix/MTabixBlockValues.java
@@ -21,16 +21,35 @@
  */
 package org.gitools.matrix.model.mtabixmatrix;
 
+import edu.upf.bg.mtabix.MTabixConfig;
+import edu.upf.bg.mtabix.parse.IKeyParser;
 import gnu.trove.map.TLongDoubleMap;
 import gnu.trove.map.hash.TLongDoubleHashMap;
 
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
 public class MTabixBlockValues {
 
+    private IKeyParser keyParser;
+    private Map<Integer, Map<String, Integer>> identifierToPosition;
     private TLongDoubleMap map;
 
-    public MTabixBlockValues() {
+    public MTabixBlockValues(IKeyParser keyParser, Map<Integer, List<String>> identifiers) {
         super();
+
         this.map = new TLongDoubleHashMap();
+        this.keyParser = keyParser;
+        this.identifierToPosition = new HashMap<>();
+        for (Map.Entry<Integer, List<String>> e : identifiers.entrySet()) {
+            Map<String, Integer> pos = new HashMap<>();
+            List<String> ids = e.getValue();
+            for (int i=0; i < ids.size(); i++) {
+                pos.put(ids.get(i), i);
+            }
+            this.identifierToPosition.put(e.getKey(), pos);
+        }
     }
 
     public Double get(String... identifiers) {
@@ -48,9 +67,9 @@ public void put(long key, double value) {
         map.put(key, value);
     }
 
-    public static long hash(String... identifiers) {
-        int a = identifiers[0].hashCode();
-        int b = identifiers[1].hashCode();
+    public long hash(String... identifiers) {
+        int a = identifierToPosition.get(keyParser.getKeys()[0]).get(identifiers[0]);
+        int b = identifierToPosition.get(keyParser.getKeys()[1]).get(identifiers[1]);
         return (long) a << 32 | b & 0xFFFFFFFFL;
     }
 }
diff --git a/org.gitools.matrix/src/main/java/org/gitools/matrix/model/mtabixmatrix/MTabixMatrix.java b/org.gitools.matrix/src/main/java/org/gitools/matrix/model/mtabixmatrix/MTabixMatrix.java
@@ -130,19 +130,18 @@ private synchronized MTabixBlockValues loadBlock(Long filePointer, int column) t
 
         long block = getBlockAddress(filePointer);
         LOGGER.info("Loading block '" + Long.toHexString(block) + "' column " + column);
-        MTabixBlockValues matrix = new MTabixBlockValues();
+        MTabixBlockValues matrix = new MTabixBlockValues(this.index.getConfig().getKeyParser(), this.index.getIndexIdentifiers());
         dataStream.seek(filePointer);
 
         while ((getBlockAddress(dataStream.getFilePointer()) == block) && ((lineLength = dataStream.readLine(line)) != -1)) {
 
             final Double value = DoubleTranslator.get().stringToValue(parseLine(line, column, lineLength));
             if (value!=null) {
-                int a = parseHash(line, 1, lineLength);
-                int b = parseHash(line, 0, lineLength);
-                long key = (long) a << 32 | b & 0xFFFFFFFFL;
+                String id1 = parseLine(line, 1, lineLength);
+                String id0 = parseLine(line, 0, lineLength);
+                long key = matrix.hash(id1, id0);
                 matrix.put(key, value);
             }
-
         }
 
         return matrix;
@@ -199,22 +198,6 @@ public String parseLine(byte[] str, int pos, int length) {
         return new String(str, start, size);
     }
 
-    public int parseHash(byte[] str, int pos, int length) {
-
-        int size = parseTab(str, pos, length);
-
-        if (size == -1) {
-            return 0;
-        }
-
-        int hash = 0;
-        for (int i=start; i < end; i++) {
-            hash = 31 * hash + str[i];
-        }
-
-        return hash;
-    }
-
     private class BlockLoader extends CacheLoader<Long, MTabixBlockValues> {
 
         private int column;
