// ReferenceContainer.java // (C) 2006 by Michael Peter Christen; mc@yacy.net, Frankfurt a. M., Germany // first published 04.07.2006 on http://yacy.net // // This is a part of YaCy, a peer-to-peer based web search engine // // $LastChangedDate$ // $LastChangedRevision$ // $LastChangedBy$ // // LICENSE // // This program is free software; you can redistribute it and/or modify // it under the terms of the GNU General Public License as published by // the Free Software Foundation; either version 2 of the License, or // (at your option) any later version. // // This program 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 General Public License for more details. // // You should have received a copy of the GNU General Public License // along with this program; if not, write to the Free Software // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA package net.yacy.kelondro.rwi; import java.lang.reflect.Array; import java.lang.reflect.Method; import java.util.ArrayList; import java.util.Arrays; import java.util.Collection; import java.util.ConcurrentModificationException; import java.util.Iterator; import java.util.List; import java.util.TreeMap; import net.yacy.cora.document.encoding.ASCII; import net.yacy.cora.order.Base64Order; import net.yacy.cora.order.ByteOrder; import net.yacy.cora.storage.HandleSet; import net.yacy.cora.util.ConcurrentLog; import net.yacy.cora.util.SpaceExceededException; import net.yacy.kelondro.index.Row; import net.yacy.kelondro.index.RowSet; /** * A ReferenceContainer is a set of ReferenceRows entries for a specific term. * Since ReferenceRow entries are special Row entries, a collection of ReferenceRows * can be contained in a RowSet. * This class extends the RowSet with methods for the handling of * special ReferenceRow Row entry objects. */ public class ReferenceContainer extends RowSet { private static final long serialVersionUID=-540567425172727979L; private byte[] termHash; protected ReferenceFactory factory; public static int maxReferences = 0; // overwrite this to enable automatic index shrinking. 0 means no shrinking public ReferenceContainer(final ReferenceFactory factory, final byte[] termHash, final RowSet collection) { super(collection); assert termHash == null || (termHash[2] != '@' && termHash.length == this.rowdef.primaryKeyLength); this.factory = factory; this.termHash = termHash; } public ReferenceContainer(final ReferenceFactory factory, final byte[] termHash) { super(factory.getRow()); assert termHash == null || (termHash[2] != '@' && termHash.length == this.rowdef.primaryKeyLength); this.termHash = termHash; this.factory = factory; this.lastTimeWrote = 0; } public ReferenceContainer(final ReferenceFactory factory, final byte[] termHash, final int objectCount) throws SpaceExceededException { super(factory.getRow(), objectCount); assert termHash == null || (termHash[2] != '@' && termHash.length == this.rowdef.primaryKeyLength); this.termHash = termHash; this.factory = factory; this.lastTimeWrote = 0; } public ReferenceContainer topLevelClone() throws SpaceExceededException { final ReferenceContainer newContainer = new ReferenceContainer(this.factory, this.termHash, size()); newContainer.addAllUnique(this); return newContainer; } public static ReferenceContainer emptyContainer(final ReferenceFactory factory, final byte[] termHash) { assert termHash == null || (termHash[2] != '@' && termHash.length == factory.getRow().primaryKeyLength); return new ReferenceContainer(factory, termHash); } public static ReferenceContainer emptyContainer(final ReferenceFactory factory, final byte[] termHash, final int elementCount) throws SpaceExceededException { assert termHash == null || (termHash[2] != '@' && termHash.length == factory.getRow().primaryKeyLength); return new ReferenceContainer(factory, termHash, elementCount); } public void setWordHash(final byte[] newTermHash) { assert this.termHash == null || (this.termHash[2] != '@' && this.termHash.length == this.rowdef.primaryKeyLength); this.termHash = newTermHash; } public long updated() { return super.lastWrote(); } public byte[] getTermHash() { return this.termHash; } public void add(final Reference entry) throws SpaceExceededException { // add without double-occurrence test assert entry.toKelondroEntry().objectsize() == super.rowdef.objectsize; this.addUnique(entry.toKelondroEntry()); } public ReferenceContainer merge(final ReferenceContainer c) throws SpaceExceededException { return new ReferenceContainer(this.factory, this.termHash, super.merge(c)); } public Reference replace(final Reference entry) throws SpaceExceededException { assert entry.toKelondroEntry().objectsize() == super.rowdef.objectsize; final Row.Entry r = super.replace(entry.toKelondroEntry()); if (r == null) return null; return this.factory.produceSlow(r); } public void put(final Reference entry) throws SpaceExceededException { assert entry.toKelondroEntry().objectsize() == super.rowdef.objectsize; super.put(entry.toKelondroEntry()); } public boolean putRecent(final Reference entry) throws SpaceExceededException { assert entry.toKelondroEntry().objectsize() == super.rowdef.objectsize; // returns true if the new entry was added, false if it already existed final Row.Entry oldEntryRow = this.replace(entry.toKelondroEntry()); if (oldEntryRow == null) { return true; } final Reference oldEntry = this.factory.produceSlow(oldEntryRow); if (entry.isOlder(oldEntry)) { // A more recent Entry is already in this container this.replace(oldEntry.toKelondroEntry()); // put it back return false; } return true; } public int putAllRecent(final ReferenceContainer c) throws SpaceExceededException { // adds all entries in c and checks every entry for double-occurrence // returns the number of new elements if (c == null) return 0; int x = 0; synchronized (c) { final Iterator i = c.entries(); while (i.hasNext()) { try { if (putRecent(i.next())) x++; } catch (final ConcurrentModificationException e) { ConcurrentLog.logException(e); } } } this.lastTimeWrote = java.lang.Math.max(this.lastTimeWrote, c.updated()); return x; } public ReferenceType getReference(final byte[] urlHash) { final Row.Entry entry = super.get(urlHash, false); if (entry == null) return null; return this.factory.produceSlow(entry); } /** * remove a url reference from the container. * if the url hash was found, return the entry, but delete the entry from the container * if the entry was not found, return null. */ public ReferenceType removeReference(final byte[] urlHash) { final Row.Entry entry = super.remove(urlHash); if (entry == null) return null; return this.factory.produceSlow(entry); } public int removeEntries(final HandleSet urlHashes) { int count = 0; final Iterator i = urlHashes.iterator(); while (i.hasNext()) count += (delete(i.next())) ? 1 : 0; return count; } /** * Shrink the reference size in such a way that it does not exceed maxReferences * In case that the index is too large, old entries are deleted * @return the number of deleted entries */ public int shrinkReferences() { final int oldsize = size(); final int diff = oldsize - maxReferences; if (maxReferences <= 0 || diff <= 0) return 0; synchronized (this) { final int[] indexes = oldPostions(diff); Arrays.sort(indexes); for (int i = indexes.length - 1; i >= 0; i--) { if (indexes[i] < 0) break; removeRow(indexes[i], false); } sort(); } trim(); return oldsize - size(); } private int[] oldPostions(final int count) { final int[] indexes = new int[count]; int i = 0; for (final List positions : positionsByLastMod()) { for (final Integer pos : positions) { indexes[i++] = pos; if (i >= count) return indexes; } } return indexes; } private Collection> positionsByLastMod() { long mod; List positions; ReferenceType r; final TreeMap> tm = new TreeMap>(); final Iterator i = this.entries(); int pos = 0; while (i.hasNext()) { r = i.next(); if (r == null) continue; mod = r.lastModified(); positions = tm.get(mod); if (positions == null) positions = new ArrayList(); positions.add(pos++); tm.put(mod, positions); } return tm.values(); } public Iterator entries() { // returns an iterator of indexRWIEntry objects return new entryIterator(); } public class entryIterator implements Iterator { Iterator rowEntryIterator; public entryIterator() { this.rowEntryIterator = iterator(); } @Override public boolean hasNext() { return this.rowEntryIterator.hasNext(); } @Override public ReferenceType next() { final Row.Entry rentry = this.rowEntryIterator.next(); if (rentry == null) return null; return ReferenceContainer.this.factory.produceSlow(rentry); } @Override public void remove() { this.rowEntryIterator.remove(); } } public static Object mergeUnique(final Object a, final Object b) throws SpaceExceededException { if (a instanceof ReferenceContainer) { final ReferenceContainer c = (ReferenceContainer) a; c.addAllUnique((ReferenceContainer) b); return c; } throw new UnsupportedOperationException("Objects have wrong type: " + a.getClass().getName()); } public static final Method containerMergeMethod; static { Method meth = null; try { final Class c = net.yacy.kelondro.rwi.ReferenceContainer.class; final Class[] args = (Class[]) Array.newInstance(Class.class, 2); args[0] = Object.class; args[1] = Object.class; meth = c.getMethod("mergeUnique", args); } catch (final SecurityException e) { System.out.println("Error while initializing containerMerge.SecurityException: " + e.getMessage()); meth = null; } catch (final NoSuchMethodException e) { System.out.println("Error while initializing containerMerge.NoSuchMethodException: " + e.getMessage()); meth = null; } assert meth != null; containerMergeMethod = meth; } public static ReferenceContainer joinExcludeContainers( final ReferenceFactory factory, final Collection> includeContainers, final Collection> excludeContainers, final int maxDistance) throws SpaceExceededException { // join a search result and return the joincount (number of pages after join) // since this is a conjunction we return an empty entity if any word is not known if (includeContainers == null) return ReferenceContainer.emptyContainer(factory, null, 0); // join the result final ReferenceContainer rcLocal = ReferenceContainer.joinContainers(factory, includeContainers, maxDistance); if (rcLocal == null) return ReferenceContainer.emptyContainer(factory, null, 0); excludeContainers(factory, rcLocal, excludeContainers); return rcLocal; } public static ReferenceContainer joinContainers( final ReferenceFactory factory, final Collection> containers, final int maxDistance) throws SpaceExceededException { // order entities by their size final TreeMap> map = new TreeMap>(); ReferenceContainer singleContainer; final Iterator> i = containers.iterator(); int count = 0; while (i.hasNext()) { // get next entity: singleContainer = i.next(); // check result if (singleContainer == null || singleContainer.isEmpty()) return null; // as this is a cunjunction of searches, we have no result if any word is not known // store result in order of result size map.put(Long.valueOf(singleContainer.size() * 1000 + count), singleContainer); count++; } // check if there is any result if (map.isEmpty()) return null; // no result, nothing found // the map now holds the search results in order of number of hits per word // we now must pairwise build up a conjunction of these sets Long k = map.firstKey(); // the smallest, which means, the one with the least entries ReferenceContainer searchA, searchB, searchResult = map.remove(k); while (!map.isEmpty() && !searchResult.isEmpty()) { // take the first element of map which is a result and combine it with result k = map.firstKey(); // the next smallest... searchA = searchResult; searchB = map.remove(k); searchResult = ReferenceContainer.joinConstructive(factory, searchA, searchB, maxDistance); // free resources searchA = null; searchB = null; } // in 'searchResult' is now the combined search result if (searchResult.isEmpty()) return null; return searchResult; } public static ReferenceContainer excludeContainers( final ReferenceFactory factory, ReferenceContainer pivot, final Collection> containers) { // check if there is any result if (containers == null || containers.isEmpty()) return pivot; // no result, nothing found final Iterator> i = containers.iterator(); while (i.hasNext()) { pivot = excludeDestructive(factory, pivot, i.next()); if (pivot == null || pivot.isEmpty()) return null; } return pivot; } // join methods private static int log2(int x) { int l = 0; while (x > 0) {x = x >> 1; l++;} return l; } public static ReferenceContainer joinConstructive( final ReferenceFactory factory, final ReferenceContainer i1, final ReferenceContainer i2, final int maxDistance) throws SpaceExceededException { if ((i1 == null) || (i2 == null)) return null; if (i1.isEmpty() || i2.isEmpty()) return null; // decide which method to use final int high = ((i1.size() > i2.size()) ? i1.size() : i2.size()); final int low = ((i1.size() > i2.size()) ? i2.size() : i1.size()); final int stepsEnum = 10 * (high + low - 1); final int stepsTest = 12 * log2(high) * low; // start most efficient method if (stepsEnum > stepsTest) { if (i1.size() < i2.size()) return joinConstructiveByTest(factory, i1, i2, maxDistance); return joinConstructiveByTest(factory, i2, i1, maxDistance); } return joinConstructiveByEnumeration(factory, i1, i2, maxDistance); } private static ReferenceContainer joinConstructiveByTest( final ReferenceFactory factory, final ReferenceContainer small, final ReferenceContainer large, final int maxDistance) throws SpaceExceededException { //System.out.println("DEBUG: JOIN METHOD BY TEST, maxdistance = " + maxDistance); assert small.rowdef.equals(large.rowdef) : "small = " + small.rowdef.toString() + "; large = " + large.rowdef.toString(); final int keylength = small.rowdef.width(0); assert (keylength == large.rowdef.width(0)); final ReferenceContainer conj = new ReferenceContainer(factory, null, 0); // start with empty search result final Iterator se = small.entries(); ReferenceType ie1; ReferenceType ie2; while (se.hasNext()) { ie1 = se.next(); ie2 = large.getReference(ie1.urlhash()); if ((ie1 != null) && (ie2 != null)) { assert (ie1.urlhash().length == keylength) : "ie0.urlHash() = " + ASCII.String(ie1.urlhash()); assert (ie2.urlhash().length == keylength) : "ie1.urlHash() = " + ASCII.String(ie2.urlhash()); // this is a hit. Calculate word distance: ie1 = factory.produceFast(ie2, true); ie1.join(ie2); if (ie1.distance() <= maxDistance) conj.add(ie1); } } return conj; } private static ReferenceContainer joinConstructiveByEnumeration( final ReferenceFactory factory, final ReferenceContainer i1, final ReferenceContainer i2, final int maxDistance) throws SpaceExceededException { //System.out.println("DEBUG: JOIN METHOD BY ENUMERATION, maxdistance = " + maxDistance); assert i1.rowdef.equals(i2.rowdef) : "i1 = " + i1.rowdef.toString() + "; i2 = " + i2.rowdef.toString(); final int keylength = i1.rowdef.width(0); assert (keylength == i2.rowdef.width(0)); final ReferenceContainer conj = new ReferenceContainer(factory, null, 0); // start with empty search result if (!((i1.rowdef.getOrdering().signature().equals(i2.rowdef.getOrdering().signature())))) return conj; // ordering must be equal final ByteOrder ordering = i1.rowdef.getOrdering(); final Iterator e1 = i1.entries(); final Iterator e2 = i2.entries(); int c; if ((e1.hasNext()) && (e2.hasNext())) { ReferenceType ie1; ReferenceType ie2; ie1 = e1.next(); ie2 = e2.next(); while (true) { assert (ie1.urlhash().length == keylength) : "ie1.urlHash() = " + ASCII.String(ie1.urlhash()); assert (ie2.urlhash().length == keylength) : "ie2.urlHash() = " + ASCII.String(ie2.urlhash()); c = ordering.compare(ie1.urlhash(), ie2.urlhash()); //System.out.println("** '" + ie1.getUrlHash() + "'.compareTo('" + ie2.getUrlHash() + "')="+c); if (c < 0) { if (e1.hasNext()) ie1 = e1.next(); else break; } else if (c > 0) { if (e2.hasNext()) ie2 = e2.next(); else break; } else { // we have found the same urls in different searches! ie1 = factory.produceFast(ie1, true); ie1.join(ie2); if (ie1.distance() <= maxDistance) conj.add(ie1); if (e1.hasNext()) ie1 = e1.next(); else break; if (e2.hasNext()) ie2 = e2.next(); else break; } } } return conj; } public static ReferenceContainer excludeDestructive( final ReferenceFactory factory, final ReferenceContainer pivot, final ReferenceContainer excl) { if (pivot == null) return null; if (excl == null) return pivot; if (pivot.isEmpty()) return null; if (excl.isEmpty()) return pivot; // decide which method to use final int high = ((pivot.size() > excl.size()) ? pivot.size() : excl.size()); final int low = ((pivot.size() > excl.size()) ? excl.size() : pivot.size()); final int stepsEnum = 10 * (high + low - 1); final int stepsTest = 12 * log2(high) * low; // start most efficient method if (stepsEnum > stepsTest) { return excludeDestructiveByTest(pivot, excl); } return excludeDestructiveByEnumeration(factory, pivot, excl); } private static ReferenceContainer excludeDestructiveByTest( final ReferenceContainer pivot, final ReferenceContainer excl) { assert pivot.rowdef.equals(excl.rowdef) : "small = " + pivot.rowdef.toString() + "; large = " + excl.rowdef.toString(); final int keylength = pivot.rowdef.width(0); assert (keylength == excl.rowdef.width(0)); final boolean iterate_pivot = pivot.size() < excl.size(); final Iterator se = (iterate_pivot) ? pivot.entries() : excl.entries(); Reference ie0, ie1; while (se.hasNext()) { ie0 = se.next(); ie1 = excl.getReference(ie0.urlhash()); if ((ie0 != null) && (ie1 != null)) { assert (ie0.urlhash().length == keylength) : "ie0.urlHash() = " + ASCII.String(ie0.urlhash()); assert (ie1.urlhash().length == keylength) : "ie1.urlHash() = " + ASCII.String(ie1.urlhash()); if (iterate_pivot) se.remove(); pivot.delete(ie0.urlhash()); } } return pivot; } private static ReferenceContainer excludeDestructiveByEnumeration( final ReferenceFactory factory, final ReferenceContainer pivot, final ReferenceContainer excl) { assert pivot.rowdef.equals(excl.rowdef) : "i1 = " + pivot.rowdef.toString() + "; i2 = " + excl.rowdef.toString(); final int keylength = pivot.rowdef.width(0); assert (keylength == excl.rowdef.width(0)); if (!((pivot.rowdef.getOrdering().signature().equals(excl.rowdef.getOrdering().signature())))) return pivot; // ordering must be equal final Iterator e1 = pivot.entries(); final Iterator e2 = excl.entries(); int c; if ((e1.hasNext()) && (e2.hasNext())) { ReferenceType ie1; ReferenceType ie2; ie1 = e1.next(); ie2 = e2.next(); while (true) { assert (ie1.urlhash().length == keylength) : "ie1.urlHash() = " + ASCII.String(ie1.urlhash()); assert (ie2.urlhash().length == keylength) : "ie2.urlHash() = " + ASCII.String(ie2.urlhash()); c = pivot.rowdef.getOrdering().compare(ie1.urlhash(), ie2.urlhash()); //System.out.println("** '" + ie1.getUrlHash() + "'.compareTo('" + ie2.getUrlHash() + "')="+c); if (c < 0) { if (e1.hasNext()) ie1 = e1.next(); else break; } else if (c > 0) { if (e2.hasNext()) ie2 = e2.next(); else break; } else { // we have found the same urls in different searches! ie1 = factory.produceFast(ie1, true); ie1.join(ie2); e1.remove(); if (e1.hasNext()) ie1 = e1.next(); else break; if (e2.hasNext()) ie2 = e2.next(); else break; } } } return pivot; } @Override public synchronized String toString() { return "C[" + ASCII.String(this.termHash) + "] has " + size() + " entries"; } @Override public int hashCode() { return (int) Base64Order.enhancedCoder.decodeLong(this.termHash, 0, 4); } }