less memory usage during collectionIndex-rebuild

should also speed up that process a little bit

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3524 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/kelondro/kelondroCollectionIndex.java

@@ -174,11 +174,10 @@ public class kelondroCollectionIndex {
                     key = aentry.getColBytes(0);
                     assert (key != null);
                     if (key == null) continue; // skip deleted entries
-                    kelondroRowSet indexrows = new kelondroRowSet(this.payloadrow, aentry.getColBytes(1));
                     ientry = irow.newEntry();
                     ientry.setCol(idx_col_key,        key);
                     ientry.setCol(idx_col_chunksize,  chunksize);
-                    ientry.setCol(idx_col_chunkcount, indexrows.size());
+                    ientry.setCol(idx_col_chunkcount, kelondroRowCollection.sizeOfExportedCollectionRows(this.payloadrow, aentry.getColBytes(1)));
                     ientry.setCol(idx_col_clusteridx, (byte) partitionNumber);
                     ientry.setCol(idx_col_flags,      (byte) 0);
                     ientry.setCol(idx_col_indexpos,   aentry.index());

source/de/anomic/kelondro/kelondroRow.java

@@ -138,6 +138,35 @@ public class kelondroRow {
         return new String(s);
     }
     
+    public long getColLong(byte[] rowinstance, int column) {
+        // uses the column definition to choose the right encoding
+        return getColLong(rowinstance, row[column].encoder(), colstart[column], row[column].cellwidth());
+    }
+    
+    protected static final long getColLong(byte[] rowinstance, int encoder, int offset, int length) {
+        // start - fix for badly stored parameters
+        if ((length >= 3) && (rowinstance[offset] == '[') && (rowinstance[offset + 1] == 'B') && (rowinstance[offset + 2] == '@')) return 0;
+        if ((length == 2) && (rowinstance[offset] == '[') && (rowinstance[offset + 1] == 'B')) return 0;
+        if ((length == 1) && (rowinstance[offset] == '[')) return 0;
+        // stop - fix for badly stored parameters
+        switch (encoder) {
+        case kelondroColumn.encoder_none:
+            throw new kelondroException("ROW", "getColLong has celltype none, no encoder given");
+        case kelondroColumn.encoder_b64e:
+            // start - fix for badly stored parameters
+            boolean maxvalue = true;
+            for (int i = 0; i < length; i++) if (rowinstance[offset + i] != '_') {maxvalue = false; break;}
+            if (maxvalue) return 0;
+            // stop - fix for badly stored parameters
+            return kelondroBase64Order.enhancedCoder.decodeLong(rowinstance, offset, length);
+        case kelondroColumn.encoder_b256:
+            return kelondroNaturalOrder.decodeLong(rowinstance, offset, length);
+        case kelondroColumn.encoder_bytes:
+            throw new kelondroException("ROW", "getColLong of celltype bytes not applicable");
+        }
+        throw new kelondroException("ROW", "getColLong did not find appropriate encoding");
+    }
+    
     public Entry newEntry() {
         return new Entry();
     }
@@ -405,36 +434,12 @@ public class kelondroRow {
             if (ref == null) return dflt;
             kelondroColumn col = (kelondroColumn) ref[0];
             int colstart = ((Integer) ref[1]).intValue();
-            return getColLong(col.encoder(), colstart, col.cellwidth());
+            return kelondroRow.getColLong(rowinstance, col.encoder(), colstart, col.cellwidth());
         }
         
         public long getColLong(int column) {
             // uses the column definition to choose the right encoding
-            return getColLong(row[column].encoder(), colstart[column], row[column].cellwidth());
+            return kelondroRow.getColLong(rowinstance, row[column].encoder(), colstart[column], row[column].cellwidth());
-        }
-
-        private long getColLong(int encoder, int offset, int length) {
-            // start - fix for badly stored parameters
-            if ((length >= 3) && (rowinstance[offset] == '[') && (rowinstance[offset + 1] == 'B') && (rowinstance[offset + 2] == '@')) return 0;
-            if ((length == 2) && (rowinstance[offset] == '[') && (rowinstance[offset + 1] == 'B')) return 0;
-            if ((length == 1) && (rowinstance[offset] == '[')) return 0;
-            // stop - fix for badly stored parameters
-            switch (encoder) {
-            case kelondroColumn.encoder_none:
-                throw new kelondroException("ROW", "getColLong has celltype none, no encoder given");
-            case kelondroColumn.encoder_b64e:
-                // start - fix for badly stored parameters
-                boolean maxvalue = true;
-                for (int i = 0; i < length; i++) if (rowinstance[offset + i] != '_') {maxvalue = false; break;}
-                if (maxvalue) return 0;
-                // stop - fix for badly stored parameters
-                return kelondroBase64Order.enhancedCoder.decodeLong(rowinstance, offset, length);
-            case kelondroColumn.encoder_b256:
-                return kelondroNaturalOrder.decodeLong(rowinstance, offset, length);
-            case kelondroColumn.encoder_bytes:
-                throw new kelondroException("ROW", "getColLong of celltype bytes not applicable");
-            }
-            throw new kelondroException("ROW", "getColLong did not find appropriate encoding");
         }
 
         public byte getColByte(String nickname, byte dflt) {

source/de/anomic/kelondro/kelondroRowCollection.java

@@ -113,13 +113,20 @@ public class kelondroRowCollection {
         this.chunkcache = exportedCollection.getColBytes(exp_collection);        
     }
     
+    private static final kelondroRow exportMeasureRow = exportRow(0 /* no relevance */);
+    
+    protected static final int sizeOfExportedCollectionRows(kelondroRow rowdef, byte[] exportedCollectionRowinstance) {
+        int chunkcount = (int) exportMeasureRow.getColLong(exportedCollectionRowinstance, exp_chunkcount);
+        return chunkcount;
+    }
+    
     private static final long day = 1000 * 60 * 60 * 24;
     
     public static int daysSince2000(long time) {
         return (int) (time / day) - 10957;
     }
     
-    private kelondroRow exportRow(int chunkcachelength) {
+    private static kelondroRow exportRow(int chunkcachelength) {
         // find out the size of this collection
         return new kelondroRow(
                 "int size-4 {b256}," +