added property index.storeCommons to switch commons storage on or off

with index.storeCommons=false all currently stored commons are deleted!
Default is now 'true', but in future full releases it will be switched to 'false'

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5315 6c8d7289-2bf4-0310-a012-ef5d649a1542

defaults/yacy.init

@@ -260,6 +260,13 @@ dbPath=DATA/PLASMADB
 indexPrimaryPath=DATA/INDEX
 indexSecondaryPath=
 
+# the commons are words that appear in the index more than 64k times in references.
+# Since indexes with such size cannot be handled efficiently, they are sorted in such a way that references with high ranking
+# are stored back into the index, and references with bad ranking are sorted out. Such sorted-out references can be stored
+# for later use (but there is no at this time). If the sorted-out references should be stored, the following property should be
+# set to true. If set to false, they are abandoned (deleted), and previously stored commons are removed.
+index.storeCommons=true
+
 # the path to the LISTS files. Most lists are used to filter web content
 listsPath=DATA/LISTS
 

source/de/anomic/index/indexCollectionRI.java

@@ -44,7 +44,7 @@ public class indexCollectionRI implements indexRI {
 
     kelondroCollectionIndex collectionIndex;
     
-    public indexCollectionRI(final File path, final String filenameStub, final int maxpartition, final kelondroRow payloadrow) {
+    public indexCollectionRI(final File path, final String filenameStub, final int maxpartition, final kelondroRow payloadrow, boolean useCommons) {
         try {
             collectionIndex = new kelondroCollectionIndex(
                     path,
@@ -53,7 +53,8 @@ public class indexCollectionRI implements indexRI {
                     kelondroBase64Order.enhancedCoder,
                     4 /*loadfactor*/,
                     maxpartition,
-                    payloadrow);
+                    payloadrow,
+                    useCommons);
         } catch (final IOException e) {
             serverLog.logSevere("PLASMA", "unable to open collection index at " + path.toString() + ":" + e.getMessage());
         }

source/de/anomic/kelondro/kelondroCollectionIndex.java

@@ -59,7 +59,7 @@ public class kelondroCollectionIndex {
     private final int           keylength;
     private final File          path;
     private final String        filenameStub;
-    private final File          commonsPath;
+    private final File          commonsPath1;
     private final int           loadfactor;
     private Map<String, kelondroFixedWidthArray> arrays; // Map of (partitionNumber"-"chunksize)/kelondroFixedWidthArray - Objects
     private final kelondroRow   payloadrow; // definition of the payload (chunks inside the collections)
@@ -113,7 +113,7 @@ public class kelondroCollectionIndex {
     }
     
     public kelondroCollectionIndex(final File path, final String filenameStub, final int keyLength, final kelondroByteOrder indexOrder,
-                                   final int loadfactor, final int maxpartitions, final kelondroRow rowdef) throws IOException {
+                                   final int loadfactor, final int maxpartitions, final kelondroRow rowdef, boolean useCommons) throws IOException {
         // the buffersize is number of bytes that are only used if the kelondroFlexTable is backed up with a kelondroTree
         indexErrors = 0;
         this.path = path;
@@ -122,8 +122,13 @@ public class kelondroCollectionIndex {
         this.payloadrow = rowdef;
         this.loadfactor = loadfactor;
         this.maxPartitions = maxpartitions;
-        this.commonsPath = new File(path, filenameStub + "." + fillZ(Integer.toHexString(rowdef.objectsize).toUpperCase(), 4) + ".commons");
-        this.commonsPath.mkdirs();
+        File cop = new File(path, filenameStub + "." + fillZ(Integer.toHexString(rowdef.objectsize).toUpperCase(), 4) + ".commons");
+        this.commonsPath1 = (useCommons) ? cop : null;
+        if (this.commonsPath1 == null) {
+            serverFileUtils.deleteDirectory(cop);
+        } else {
+            this.commonsPath1.mkdirs();
+        }
         final File f = new File(path, filenameStub + ".index");
         
         if (f.exists()) {
@@ -640,11 +645,12 @@ public class kelondroCollectionIndex {
         serverLog.logInfo("kelondroCollectionIndex", "shrinked common word " + new String(key) + "; old size = " + oldsize + ", new size = " + collection.size() + ", maximum size = " + targetSize + ", newcommon size = " + newcommon.size() + ", first newcommon = " + firstnewcommon);
         
         // finally dump the removed entries to a file
+        if (commonsPath1 != null) {
             newcommon.sort();
             final SimpleDateFormat formatter = new SimpleDateFormat("yyyyMMddHHmmss");
             formatter.setTimeZone(TimeZone.getTimeZone("GMT"));
             final String filename = serverCodings.encodeHex(kelondroBase64Order.enhancedCoder.decode(new String(key), "de.anomic.kelondro.kelondroCollectionIndex.shrinkCollection(...)")) + "_" + formatter.format(new Date()) + ".collection";
-        final File storagePath = new File(commonsPath, filename.substring(0, 2)); // make a subpath
+            final File storagePath = new File(commonsPath1, filename.substring(0, 2)); // make a subpath
             storagePath.mkdirs();
             final File file = new File(storagePath, filename);
             try {
@@ -654,7 +660,7 @@ public class kelondroCollectionIndex {
                 e.printStackTrace();
                 serverLog.logWarning("kelondroCollectionIndex", "failed to dump common word " + new String(key) + " to " + file.toString() + "; size = " + newcommon.size());
             }
-        
+        }        
     }
     
     public synchronized int remove(final byte[] key, final Set<String> removekeys) throws IOException, kelondroOutOfLimitsException {
@@ -885,7 +891,7 @@ public class kelondroCollectionIndex {
             final kelondroCollectionIndex collectionIndex  = new kelondroCollectionIndex(
                         path, filenameStub, 9 /*keyLength*/,
                         kelondroNaturalOrder.naturalOrder,
-                        4 /*loadfactor*/, 7, rowdef);
+                        4 /*loadfactor*/, 7, rowdef, false);
             
             // fill index with values
             kelondroRowSet collection = new kelondroRowSet(rowdef, 0);

source/de/anomic/plasma/plasmaHTCache.java

@@ -49,6 +49,7 @@ import de.anomic.kelondro.kelondroBLOBBuffer;
 import de.anomic.kelondro.kelondroBLOBHeap;
 import de.anomic.kelondro.kelondroBase64Order;
 import de.anomic.kelondro.kelondroMap;
+import de.anomic.server.serverFileUtils;
 import de.anomic.server.logging.serverLog;
 import de.anomic.yacy.yacySeedDB;
 import de.anomic.yacy.yacyURL;
@@ -95,7 +96,7 @@ public final class plasmaHTCache {
                     object.getName().equals("yacy") ||
                     object.getName().equals("https") ||
                     object.getName().equals("ftp")) {
-                    deleteOldHTCache(cachePath);
+                    serverFileUtils.deleteDirectory(cachePath);
                 }
             }
         }
@@ -141,22 +142,6 @@ public final class plasmaHTCache {
         }
     }
 
-    private static void deleteOldHTCache(final File directory) {
-        final String[] list = directory.list();
-        if (list != null) {
-            File object;
-            for (int i = list.length - 1; i >= 0; i--) {
-                object = new File(directory, list[i]);
-                if (object.isFile()) {
-                    object.delete();
-                } else {
-                    deleteOldHTCache(object);
-                }
-            }
-        }
-        directory.delete();
-    }
-
     public static int responseHeaderDBSize() {
         return responseHeaderDB.size();
     }

source/de/anomic/plasma/plasmaRankingCRProcess.java

@@ -243,7 +243,7 @@ public class plasmaRankingCRProcess {
         if (newdb) {
             final File path = to_file.getParentFile(); // path to storage place
             newacc = new kelondroFlexTable(path, CRG_accname, CRG_accrow, 0, false);
-            newseq = new kelondroCollectionIndex(path, CRG_seqname, 12, kelondroBase64Order.enhancedCoder, 2, 9, CRG_colrow);
+            newseq = new kelondroCollectionIndex(path, CRG_seqname, 12, kelondroBase64Order.enhancedCoder, 2, 9, CRG_colrow, false);
         } else {
             if (!(to_file.exists())) {
                 acc = new kelondroAttrSeq("Global Ranking Accumulator File",
@@ -372,8 +372,8 @@ public class plasmaRankingCRProcess {
     
     public static int genrcix(final File cr_path_in, final File rci_path_out) throws IOException {
         //kelondroFlexTable       acc = new kelondroFlexTable(cr_path_in, CRG_accname, kelondroBase64Order.enhancedCoder, 128 * 1024 * 1024, -1, CRG_accrow, true);
-        final kelondroCollectionIndex seq = new kelondroCollectionIndex(cr_path_in, CRG_seqname, 12, kelondroBase64Order.enhancedCoder, 2, 9, CRG_colrow);
-        final kelondroCollectionIndex rci = new kelondroCollectionIndex(rci_path_out, RCI_colname, 6, kelondroBase64Order.enhancedCoder, 2, 9, RCI_coli);
+        final kelondroCollectionIndex seq = new kelondroCollectionIndex(cr_path_in, CRG_seqname, 12, kelondroBase64Order.enhancedCoder, 2, 9, CRG_colrow, false);
+        final kelondroCollectionIndex rci = new kelondroCollectionIndex(rci_path_out, RCI_colname, 6, kelondroBase64Order.enhancedCoder, 2, 9, RCI_coli, false);
         
         // loop over all referees
         int count = 0;

source/de/anomic/plasma/plasmaSwitchboard.java

@@ -304,7 +304,8 @@ public final class plasmaSwitchboard extends serverAbstractSwitch<IndexingStack.
         // start indexing management
         log.logConfig("Starting Indexing Management");
         final String networkName = getConfig(plasmaSwitchboardConstants.NETWORK_NAME, "");
-        webIndex = new plasmaWordIndex(networkName, log, indexPrimaryPath, indexSecondaryPath, wordCacheMaxCount);
+        final boolean useCommons = getConfigBool("index.storeCommons", false);
+        webIndex = new plasmaWordIndex(networkName, log, indexPrimaryPath, indexSecondaryPath, wordCacheMaxCount, useCommons);
         crawlResults = new ResultURLs();
         
         // start yacy core
@@ -738,7 +739,8 @@ public final class plasmaSwitchboard extends serverAbstractSwitch<IndexingStack.
             final File indexPrimaryPath = getConfigPath(plasmaSwitchboardConstants.INDEX_PRIMARY_PATH, plasmaSwitchboardConstants.INDEX_PATH_DEFAULT);
             final File indexSecondaryPath = (getConfig(plasmaSwitchboardConstants.INDEX_SECONDARY_PATH, "").length() == 0) ? indexPrimaryPath : new File(getConfig(plasmaSwitchboardConstants.INDEX_SECONDARY_PATH, ""));
             final int wordCacheMaxCount = (int) getConfigLong(plasmaSwitchboardConstants.WORDCACHE_MAX_COUNT, 20000);
-            this.webIndex = new plasmaWordIndex(getConfig(plasmaSwitchboardConstants.NETWORK_NAME, ""), getLog(), indexPrimaryPath, indexSecondaryPath, wordCacheMaxCount);
+            final boolean useCommons = getConfigBool("index.storeCommons", false);
+            this.webIndex = new plasmaWordIndex(getConfig(plasmaSwitchboardConstants.NETWORK_NAME, ""), getLog(), indexPrimaryPath, indexSecondaryPath, wordCacheMaxCount, useCommons);
         }
         // start up crawl jobs
         continueCrawlJob(plasmaSwitchboardConstants.CRAWLJOB_LOCAL_CRAWL);

source/de/anomic/plasma/plasmaWordIndex.java

@@ -115,7 +115,7 @@ public final class plasmaWordIndex implements indexRI {
     private final File                     queuesRoot;
     public  yacyPeerActions                peerActions;
 
-    public plasmaWordIndex(final String networkName, final serverLog log, final File indexPrimaryRoot, final File indexSecondaryRoot, final int entityCacheMaxSize) {
+    public plasmaWordIndex(final String networkName, final serverLog log, final File indexPrimaryRoot, final File indexSecondaryRoot, final int entityCacheMaxSize, boolean useCommons) {
         if (networkName == null || networkName.length() == 0) {
             log.logSevere("no network name given - shutting down");
             System.exit(0);
@@ -148,7 +148,7 @@ public final class plasmaWordIndex implements indexRI {
         // create collections storage path
         final File textindexcollections = new File(indexPrimaryTextLocation, "RICOLLECTION");
         if (!(textindexcollections.exists())) textindexcollections.mkdirs();
-        this.collections = new indexCollectionRI(textindexcollections, "collection", maxCollectionPartition, indexRWIRowEntry.urlEntryRow);
+        this.collections = new indexCollectionRI(textindexcollections, "collection", maxCollectionPartition, indexRWIRowEntry.urlEntryRow, useCommons);
 
         // create LURL-db
         referenceURL = new indexRepositoryReference(this.secondaryRoot);

source/de/anomic/server/serverFileUtils.java

@@ -620,4 +620,25 @@ public final class serverFileUtils {
         writer1.flush();
         return count;
     }
+    
+    /**
+     * delete a directory
+     * if the directory is not empty, delete also everything inside
+     * @param directory
+     */
+    public static void deleteDirectory(final File directory) {
+        final String[] list = directory.list();
+        if (list != null) {
+            File object;
+            for (int i = list.length - 1; i >= 0; i--) {
+                object = new File(directory, list[i]);
+                if (object.isFile()) {
+                    object.delete();
+                } else {
+                    deleteDirectory(object);
+                }
+            }
+        }
+        directory.delete();
+    }
 }

source/yacy.java

@@ -674,7 +674,7 @@ public final class yacy {
             final int cacheMem = (int)(serverMemory.max() - serverMemory.total());
             if (cacheMem < 2048000) throw new OutOfMemoryError("Not enough memory available to start clean up.");
                 
-            final plasmaWordIndex wordIndex = new plasmaWordIndex(networkName, log, indexPrimaryRoot, indexSecondaryRoot, 10000);
+            final plasmaWordIndex wordIndex = new plasmaWordIndex(networkName, log, indexPrimaryRoot, indexSecondaryRoot, 10000, false);
             final Iterator<indexContainer> indexContainerIterator = wordIndex.wordContainers("AAAAAAAAAAAA", false, false);
             
             long urlCounter = 0, wordCounter = 0;
@@ -865,7 +865,7 @@ public final class yacy {
         try {
             Iterator<indexContainer> indexContainerIterator = null;
             if (resource.equals("all")) {
-                WordIndex = new plasmaWordIndex("freeworld", log, indexPrimaryRoot, indexSecondaryRoot, 10000);
+                WordIndex = new plasmaWordIndex("freeworld", log, indexPrimaryRoot, indexSecondaryRoot, 10000, false);
                 indexContainerIterator = WordIndex.wordContainers(wordChunkStartHash, false, false);
             }
             int counter = 0;