%PDF- %PDF-
| Direktori : /home/waritko/yacy/source/net/yacy/cora/federate/yacy/ |
| Current File : //home/waritko/yacy/source/net/yacy/cora/federate/yacy/Distribution.java |
/**
* VerticalPartition
* Copyright 2009 by Michael Peter Christen
* First released 28.01.2009 at http://yacy.net
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with this program in the file lgpl21.txt
* If not, see <http://www.gnu.org/licenses/>.
*/
package net.yacy.cora.federate.yacy;
import net.yacy.cora.document.encoding.ASCII;
import net.yacy.cora.document.encoding.UTF8;
import net.yacy.cora.order.Base64Order;
/**
* calculate the DHT position for horizontal and vertical performance scaling:
* horizontal: scale with number of words
* vertical: scale with number of references for every word
* The vertical scaling is selected using the corresponding reference hash, the url hash
* This has the effect that every vertical position accumulates references for the same url
* and the urls are not spread over all positions of the DHT.
*/
public class Distribution {
private final int verticalPartitionExponent;
private final int shiftLength;
private final int partitionCount;
private final long partitionSize;
private final long partitionMask;
/**
*
* @param verticalPartitionExponent, the number of partitions should be computed with partitions = 2**n, n = scaling factor
*/
public Distribution(int verticalPartitionExponent) {
// the partition exponent is the number of bits that we use for the partition
this.verticalPartitionExponent = verticalPartitionExponent;
// number of partitions that is possible for the given number of partition exponent bits
this.partitionCount = 1 << this.verticalPartitionExponent;
// we use Long.SIZE - 1 as bitlength since we use only the 63 bits of 0..Long.MAX_VALUE
this.shiftLength = Long.SIZE - 1 - this.verticalPartitionExponent;
// the partition size is the cardinal number of possible hash positions for each segment of the DHT
this.partitionSize = 1L << this.shiftLength;
// the partition mask is a bitmask for each partition
this.partitionMask = this.partitionSize - 1L;
}
public int verticalPartitions() {
return this.partitionCount;
}
/**
* the horizontal DHT position uses simply the ordering on hashes, the base 64 order to assign a cardinal
* in the range of 0..Long.MAX_VALUE to the word.
* @param wordHash
* @return
*/
public final static long horizontalDHTPosition(byte[] wordHash) {
assert wordHash != null;
assert wordHash[2] != '@';
return Base64Order.enhancedCoder.cardinal(wordHash);
}
/**
* the horizontal DHT distance is the cardinal number between the cardinal position of the hashes of two objects in the DHT
* Since the DHT is closed at the end, a cardinal at the high-end of 0..Long.MAX_VALUE can be very close to a low cardinal number.
* @param from the start DHT position as word hash
* @param to the end DHT position as word hash
* @return the distance of two positions. The maximal distance is Long.MAX_VALUE / 2
*/
public final static long horizontalDHTDistance(final byte[] from, final byte[] to) {
// the dht distance is a positive value between 0 and 1
// if the distance is small, the word more probably belongs to the peer
final long toPos = horizontalDHTPosition(to);
final long fromPos = horizontalDHTPosition(from);
return horizontalDHTDistance(fromPos, toPos);
}
/**
* the horizontalDHTDistance computes the closed-at-the-end ordering of two cardinal DHT positions
* @param fromPos the start DHT position as cardinal of the word hash
* @param toPos the end DHT position as cardinal of the word hash
* @return the distance of two positions. The maximal distance is Long.MAX_VALUE / 2
*/
public final static long horizontalDHTDistance(final long fromPos, final long toPos) {
return (toPos >= fromPos) ? toPos - fromPos : (Long.MAX_VALUE - fromPos) + toPos + 1;
}
/**
* the reverse function to horizontalDHTPosition
* This is a bit fuzzy since the horizontalDHTPosition cannot represent all 72 bits of the word hash (Yes, its a HASH!)
* @param l the cardinal position in the DHT
* @return the abstract/computed word of the cardinal.
*/
public final static byte[] positionToHash(final long l) {
// transform the position of a peer position into a close peer hash
byte[] h = Base64Order.enhancedCoder.uncardinal(l);
assert h.length == 12;
return h;
}
/**
* the partition size is (Long.MAX + 1) / 2 ** e == 2 ** (63 - e)
* compute the position using a specific fragment of the word hash and the url hash:
* - from the word hash take the 63 - <partitionExponent> lower bits
* - from the url hash take the <partitionExponent> higher bits
* in case that the partitionExpoent is 1, only one bit is taken from the urlHash,
* which means that the partition is in two parts.
* With partitionExponent = 2 it is divided in four parts and so on.
* @param wordHash
* @param urlHash
* @return
*/
public final long verticalDHTPosition(final byte[] wordHash, final String urlHash) {
// this creates 1^^e different positions for the same word hash (according to url hash)
return (Distribution.horizontalDHTPosition(wordHash) & partitionMask) | (Distribution.horizontalDHTPosition(ASCII.getBytes(urlHash)) & ~partitionMask);
}
/**
* compute a vertical DHT position for a given word
* This is used when a word is searched and the peers holding the word must be computed
* @param wordHash, the hash of the word
* @param verticalPosition (0 <= verticalPosition < verticalPartitions())
* @return a number that can represents a position and can be computed to a word hash again
*/
public final long verticalDHTPosition(final byte[] wordHash, final int verticalPosition) {
assert verticalPosition >= 0 && verticalPosition < verticalPartitions();
long verticalMask = ((long) verticalPosition) << this.shiftLength; // don't remove the cast! it will become an integer result which is wrong.
return (Distribution.horizontalDHTPosition(wordHash) & partitionMask) | verticalMask;
}
/**
* compute the vertical position of a url hash. Thats the same value as second parameter in verticalDHTPosition/2
* @param urlHash
* @return a number from 0..verticalPartitions()
*/
public final int verticalDHTPosition(final byte[] urlHash) {
int vdp = (int) (Distribution.horizontalDHTPosition(urlHash) >> this.shiftLength); // take only the top-<partitionExponent> bits
assert vdp >= 0;
assert vdp < this.partitionCount;
return vdp;
}
public static void main(String[] args) {
// java -classpath classes de.anomic.yacy.yacySeed hHJBztzcFn76
// java -classpath classes de.anomic.yacy.yacySeed hHJBztzcFG76 M8hgtrHG6g12 3
// test the DHT position calculation
byte[] wordHash = UTF8.getBytes("hHJBztzcFn76");
long dhtl;
int partitionExponent = 4;
Distribution partition = new Distribution(partitionExponent);
dhtl = Distribution.horizontalDHTPosition(wordHash);
System.out.println("DHT Long = " + dhtl);
System.out.println("DHT as b64 from Long = " + ASCII.String(Distribution.positionToHash(dhtl)));
System.out.println("all " + partition.verticalPartitions() + " DHT positions from long : ");
for (int i = 0; i < partition.verticalPartitions(); i++) {
long l = partition.verticalDHTPosition(wordHash, i);
System.out.println(ASCII.String(Distribution.positionToHash(l)));
}
System.out.println();
long c1 = Base64Order.enhancedCoder.cardinal("AAAAAAAAAAAA".getBytes());
System.out.println(ASCII.String(Base64Order.enhancedCoder.uncardinal(c1)));
long c2 = Base64Order.enhancedCoder.cardinal("____________".getBytes());
System.out.println(ASCII.String(Base64Order.enhancedCoder.uncardinal(c2)));
}
}