hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From st...@apache.org
Subject svn commit: r656868 [6/10] - in /hadoop/hbase/trunk: ./ src/java/org/apache/hadoop/hbase/ src/java/org/apache/hadoop/hbase/client/ src/java/org/apache/hadoop/hbase/filter/ src/java/org/apache/hadoop/hbase/hql/ src/java/org/apache/hadoop/hbase/io/ src/j...
Date Thu, 15 May 2008 22:10:50 GMT
Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStore.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStore.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStore.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStore.java Thu May 15 15:10:47 2008
@@ -52,11 +52,10 @@
 import org.apache.hadoop.hbase.filter.RowFilterInterface;
 import org.apache.hadoop.hbase.io.Cell;
 import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
-import org.apache.hadoop.hbase.io.TextSequence;
+import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.hbase.util.FSUtils;
 import org.apache.hadoop.io.MapFile;
 import org.apache.hadoop.io.SequenceFile;
-import org.apache.hadoop.io.Text;
 import org.apache.hadoop.io.Writable;
 import org.apache.hadoop.util.Progressable;
 import org.apache.hadoop.util.StringUtils;
@@ -104,7 +103,8 @@
 
   private final ReentrantReadWriteLock lock = new ReentrantReadWriteLock();
 
-  final Text storeName;
+  final byte [] storeName;
+  private final String storeNameStr;
 
   /*
    * Sorted Map of readers keyed by sequence id (Most recent should be last in
@@ -180,8 +180,9 @@
       this.ttl *= 1000;
     this.memcache = new Memcache(this.ttl);
     this.compactionDir = HRegion.getCompactionDir(basedir);
-    this.storeName =
-      new Text(this.info.getEncodedName() + "/" + this.family.getFamilyName());
+    this.storeName = Bytes.toBytes(this.info.getEncodedName() + "/" +
+      Bytes.toString(this.family.getName()));
+    this.storeNameStr = Bytes.toString(this.storeName);
     
     // By default, we compact if an HStore has more than
     // MIN_COMMITS_FOR_COMPACTION map files
@@ -203,12 +204,12 @@
     }
     
     Path mapdir = HStoreFile.getMapDir(basedir, info.getEncodedName(),
-        family.getFamilyName());
+        family.getName());
     if (!fs.exists(mapdir)) {
       fs.mkdirs(mapdir);
     }
     Path infodir = HStoreFile.getInfoDir(basedir, info.getEncodedName(),
-        family.getFamilyName());
+        family.getName());
     if (!fs.exists(infodir)) {
       fs.mkdirs(infodir);
     }
@@ -218,7 +219,7 @@
       this.bloomFilter = null;
     } else {
       this.filterDir = HStoreFile.getFilterDir(basedir, info.getEncodedName(),
-          family.getFamilyName());
+          family.getName());
       if (!fs.exists(filterDir)) {
         fs.mkdirs(filterDir);
       }
@@ -232,9 +233,9 @@
     // loadHStoreFiles also computes the max sequence id internally.
     this.maxSeqId = -1L;
     this.storefiles.putAll(loadHStoreFiles(infodir, mapdir));
-    if (LOG.isDebugEnabled()) {
+    if (LOG.isDebugEnabled() && this.storefiles.size() > 0) {
       LOG.debug("Loaded " + this.storefiles.size() + " file(s) in hstore " +
-        this.storeName + ", max sequence id " + this.maxSeqId);
+        Bytes.toString(this.storeName) + ", max sequence id " + this.maxSeqId);
     }
     
     try {
@@ -244,7 +245,7 @@
       // HADOOP-1700; for now keep going but this is probably not what we want
       // long term.  If we got here there has been data-loss
       LOG.warn("Exception processing reconstruction log " + reconstructionLog +
-        " opening " + this.storeName +
+        " opening " + Bytes.toString(this.storeName) +
         " -- continuing.  Probably DATA LOSS!", e);
     }
 
@@ -317,10 +318,10 @@
         }
         // Check this edit is for me. Also, guard against writing
         // METACOLUMN info such as HBASE::CACHEFLUSH entries
-        Text column = val.getColumn();
-        if (column.equals(HLog.METACOLUMN)
-            || !key.getRegionName().equals(info.getRegionName())
-            || !HStoreKey.extractFamily(column).equals(family.getFamilyName())) {
+        byte [] column = val.getColumn();
+        if (Bytes.equals(column, HLog.METACOLUMN)
+            || !Bytes.equals(key.getRegionName(), info.getRegionName())
+            || !HStoreKey.matchingFamily(family.getName(), column)) {
           continue;
         }
         HStoreKey k = new HStoreKey(key.getRow(), column, val.getTimestamp());
@@ -386,7 +387,7 @@
         reference = readSplitInfo(p, fs);
       }
       curfile = new HStoreFile(conf, fs, basedir, info.getEncodedName(),
-        family.getFamilyName(), fid, reference);
+        family.getName(), fid, reference);
       storeSize += curfile.length();
       long storeSeqId = -1;
       try {
@@ -450,7 +451,7 @@
     Filter bloomFilter = null;
     if(fs.exists(filterFile)) {
       if (LOG.isDebugEnabled()) {
-        LOG.debug("loading bloom filter for " + this.storeName);
+        LOG.debug("loading bloom filter for " + this.storeNameStr);
       }
       
       BloomFilterDescriptor.BloomFilterType type =
@@ -482,7 +483,7 @@
       }
     } else {
       if (LOG.isDebugEnabled()) {
-        LOG.debug("creating bloom filter for " + this.storeName);
+        LOG.debug("creating bloom filter for " + this.storeNameStr);
       }
 
       BloomFilterDescriptor.BloomFilterType type =
@@ -517,7 +518,7 @@
    */
   private void flushBloomFilter() throws IOException {
     if (LOG.isDebugEnabled()) {
-      LOG.debug("flushing bloom filter for " + this.storeName);
+      LOG.debug("flushing bloom filter for " + this.storeNameStr);
     }
     FSDataOutputStream out =
       fs.create(new Path(filterDir, BLOOMFILTER_FILE_NAME));
@@ -527,7 +528,7 @@
       out.close();
     }
     if (LOG.isDebugEnabled()) {
-      LOG.debug("flushed bloom filter for " + this.storeName);
+      LOG.debug("flushed bloom filter for " + this.storeNameStr);
     }
   }
   
@@ -568,7 +569,7 @@
       synchronized (this.storefiles) {
         result = new ArrayList<HStoreFile>(storefiles.values());
       }
-      LOG.debug("closed " + this.storeName);
+      LOG.debug("closed " + this.storeNameStr);
       return result;
     } finally {
       this.lock.writeLock().unlock();
@@ -621,7 +622,7 @@
       long now = System.currentTimeMillis();
       // A. Write the Maps out to the disk
       HStoreFile flushedFile = new HStoreFile(conf, fs, basedir,
-        info.getEncodedName(),  family.getFamilyName(), -1L, null);
+        info.getEncodedName(),  family.getName(), -1L, null);
       MapFile.Writer out = flushedFile.getWriter(this.fs, this.compression,
         this.bloomFilter);
       
@@ -641,8 +642,7 @@
         for (Map.Entry<HStoreKey, byte []> es: cache.entrySet()) {
           HStoreKey curkey = es.getKey();
           byte[] bytes = es.getValue();
-          TextSequence f = HStoreKey.extractFamily(curkey.getColumn());
-          if (f.equals(this.family.getFamilyName())) {
+          if (HStoreKey.matchingFamily(this.family.getName(), curkey.getColumn())) {
             if (ttl == HConstants.FOREVER ||
                   now < curkey.getTimestamp() + ttl) {
               entries++;
@@ -775,11 +775,14 @@
    * @return mid key if a split is needed, null otherwise
    * @throws IOException
    */
-  Text compact(final boolean force) throws IOException {
+  byte [] compact(final boolean force) throws IOException {
     synchronized (compactLock) {
       long maxId = -1;
       List<HStoreFile> filesToCompact = null;
       synchronized (storefiles) {
+        if (this.storefiles.size() <= 0) {
+          return null;
+        }
         filesToCompact = new ArrayList<HStoreFile>(this.storefiles.values());
         if (!force && !hasReferences(filesToCompact) &&
              filesToCompact.size() < compactionThreshold) {
@@ -800,7 +803,7 @@
 
       // Step through them, writing to the brand-new MapFile
       HStoreFile compactedOutputFile = new HStoreFile(conf, fs, 
-          this.compactionDir, info.getEncodedName(), family.getFamilyName(),
+          this.compactionDir, info.getEncodedName(), family.getName(),
           -1L, null);
       if (LOG.isDebugEnabled()) {
         LOG.debug("started compaction of " + filesToCompact.size() +
@@ -822,7 +825,7 @@
       completeCompaction(filesToCompact, compactedOutputFile);
 
       if (LOG.isDebugEnabled()) {
-        LOG.debug("Completed compaction of " + this.storeName +
+        LOG.debug("Completed compaction of " + this.storeNameStr +
             " store size is " + StringUtils.humanReadableInt(storeSize));
       }
     }
@@ -883,10 +886,10 @@
 
       long now = System.currentTimeMillis();
       int timesSeen = 0;
-      Text lastRow = new Text();
-      Text lastColumn = new Text();
+      byte [] lastRow = null;
+      byte [] lastColumn = null;
       // Map of a row deletes keyed by column with a list of timestamps for value
-      Map<Text, List<Long>> deletes = null;
+      Map<byte [], List<Long>> deletes = null;
       while (numDone < done.length) {
         // Find the reader with the smallest key.  If two files have same key
         // but different values -- i.e. one is delete and other is non-delete
@@ -909,13 +912,13 @@
 
         // Reflect the current key/val in the output
         HStoreKey sk = keys[smallestKey];
-        if(lastRow.equals(sk.getRow())
-            && lastColumn.equals(sk.getColumn())) {
+        if (Bytes.equals(lastRow, sk.getRow())
+            && Bytes.equals(lastColumn, sk.getColumn())) {
           timesSeen++;
         } else {
-          timesSeen = 1;
+          timesSeen = 0;
           // We are on to a new row.  Create a new deletes list.
-          deletes = new HashMap<Text, List<Long>>();
+          deletes = new TreeMap<byte [], List<Long>>(Bytes.BYTES_COMPARATOR);
         }
 
         byte [] value = (vals[smallestKey] == null)?
@@ -924,7 +927,7 @@
             timesSeen <= family.getMaxVersions()) {
           // Keep old versions until we have maxVersions worth.
           // Then just skip them.
-          if (sk.getRow().getLength() != 0 && sk.getColumn().getLength() != 0) {
+          if (sk.getRow().length != 0 && sk.getColumn().length != 0) {
             // Only write out objects which have a non-zero length key and
             // value
             if (ttl == HConstants.FOREVER || now < sk.getTimestamp() + ttl) {
@@ -938,8 +941,8 @@
         }
 
         // Update last-seen items
-        lastRow.set(sk.getRow());
-        lastColumn.set(sk.getColumn());
+        lastRow = sk.getRow();
+        lastColumn = sk.getColumn();
 
         // Advance the smallest key.  If that reader's all finished, then 
         // mark it as done.
@@ -962,7 +965,7 @@
         try {
           rdrs[i].close();
         } catch (IOException e) {
-          LOG.warn("Exception closing reader for " + this.storeName, e);
+          LOG.warn("Exception closing reader for " + this.storeNameStr, e);
         }
       }
     }
@@ -984,7 +987,7 @@
    * passed value is HGlobals.deleteBytes.
   */
   private boolean isDeleted(final HStoreKey hsk, final byte [] value,
-      final boolean checkMemcache, final Map<Text, List<Long>> deletes) {
+      final boolean checkMemcache, final Map<byte [], List<Long>> deletes) {
     if (checkMemcache && memcache.isDeleted(hsk)) {
       return true;
     }
@@ -1042,7 +1045,7 @@
     try {
       // 1. Moving the new MapFile into place.
       HStoreFile finalCompactedFile = new HStoreFile(conf, fs, basedir,
-        info.getEncodedName(), family.getFamilyName(), -1, null);
+        info.getEncodedName(), family.getName(), -1, null);
       if (LOG.isDebugEnabled()) {
         LOG.debug("moving " + FSUtils.getPath(compactedFile.getMapFilePath()) +
           " to " + FSUtils.getPath(finalCompactedFile.getMapFilePath()));
@@ -1093,7 +1096,8 @@
           }
         } catch (IOException e) {
           e = RemoteExceptionHandler.checkIOException(e);
-          LOG.error("Failed replacing compacted files for " + this.storeName +
+          LOG.error("Failed replacing compacted files for " +
+            this.storeNameStr +
             ". Compacted file is " + finalCompactedFile.toString() +
             ".  Files replaced are " + compactedFiles.toString() +
             " some of which may have been already removed", e);
@@ -1120,10 +1124,12 @@
    *
    * The returned object should map column names to Cells.
    */
-  void getFull(HStoreKey key, final Set<Text> columns, Map<Text, Cell> results)
+  void getFull(HStoreKey key, final Set<byte []> columns,
+      Map<byte [], Cell> results)
   throws IOException {
-    Map<Text, Long> deletes = new HashMap<Text, Long>();
-    
+    Map<byte [], Long> deletes =
+      new TreeMap<byte [], Long>(Bytes.BYTES_COMPARATOR);
+
     // if the key is null, we're not even looking for anything. return.
     if (key == null) {
       return;
@@ -1152,7 +1158,7 @@
   }
   
   private void getFullFromMapFile(MapFile.Reader map, HStoreKey key, 
-    Set<Text> columns, Map<Text, Long> deletes, Map<Text, Cell> results) 
+    Set<byte []> columns, Map<byte [], Long> deletes, Map<byte [], Cell> results) 
   throws IOException {
     synchronized(map) {
       long now = System.currentTimeMillis();
@@ -1167,7 +1173,7 @@
         return;
       }
       do {
-        Text readcol = readkey.getColumn();
+        byte [] readcol = readkey.getColumn();
         
         // if we're looking for this column (or all of them), and there isn't 
         // already a value for this column in the results map, and the key we 
@@ -1182,18 +1188,18 @@
             // recent delete timestamp, record it for later
             if (!deletes.containsKey(readcol) 
               || deletes.get(readcol).longValue() < readkey.getTimestamp()) {
-              deletes.put(new Text(readcol), readkey.getTimestamp());              
+              deletes.put(readcol, readkey.getTimestamp());              
             }
           } else if (!(deletes.containsKey(readcol) 
             && deletes.get(readcol).longValue() >= readkey.getTimestamp()) ) {
             // So the cell itself isn't a delete, but there may be a delete 
             // pending from earlier in our search. Only record this result if
             // there aren't any pending deletes.
-            if (!(deletes.containsKey(readcol) 
-              && deletes.get(readcol).longValue() >= readkey.getTimestamp())) {
+            if (!(deletes.containsKey(readcol) &&
+                deletes.get(readcol).longValue() >= readkey.getTimestamp())) {
               if (ttl == HConstants.FOREVER ||
                       now < readkey.getTimestamp() + ttl) {
-                results.put(new Text(readcol), 
+                results.put(readcol, 
                   new Cell(readval.get(), readkey.getTimestamp()));
                 // need to reinstantiate the readval so we can reuse it, 
                 // otherwise next iteration will destroy our result
@@ -1205,7 +1211,7 @@
               }
             }
           }
-        } else if(key.getRow().compareTo(readkey.getRow()) < 0) {
+        } else if (Bytes.compareTo(key.getRow(), readkey.getRow()) < 0) {
           // if we've crossed into the next row, then we can just stop 
           // iterating
           break;
@@ -1253,7 +1259,8 @@
       // This List of deletes should not large since we are only keeping rows
       // and columns that match those set on the scanner and which have delete
       // values.  If memory usage becomes an issue, could redo as bloom filter.
-      Map<Text, List<Long>> deletes = new HashMap<Text, List<Long>>();
+      Map<byte [], List<Long>> deletes =
+        new TreeMap<byte [], List<Long>>(Bytes.BYTES_COMPARATOR);
       // This code below is very close to the body of the getKeys method.
       MapFile.Reader[] maparray = getReaders();
       for(int i = maparray.length - 1; i >= 0; i--) {
@@ -1337,8 +1344,7 @@
    * @throws IOException
    */
   List<HStoreKey> getKeys(final HStoreKey origin, final int versions)
-    throws IOException {
-    
+  throws IOException {
     List<HStoreKey> keys = this.memcache.getKeys(origin, versions);
     if (versions != ALL_VERSIONS && keys.size() >= versions) {
       return keys;
@@ -1414,13 +1420,13 @@
    * with stricly increasing timestamps. This method assumes this pattern of 
    * writes in order to make it reasonably performant. 
    */
-  Text getRowKeyAtOrBefore(final Text row)
+  byte [] getRowKeyAtOrBefore(final byte [] row)
   throws IOException{
     // Map of HStoreKeys that are candidates for holding the row key that
     // most closely matches what we're looking for. We'll have to update it 
     // deletes found all over the place as we go along before finally reading
     // the best key out of it at the end.
-    SortedMap<HStoreKey, Long> candidateKeys = new TreeMap<HStoreKey, Long>();    
+    SortedMap<HStoreKey, Long> candidateKeys = new TreeMap<HStoreKey, Long>();
     
     // Obtain read lock
     this.lock.readLock().lock();
@@ -1446,14 +1452,13 @@
    * Check an individual MapFile for the row at or before a given key 
    * and timestamp
    */
-  private void rowAtOrBeforeFromMapFile(MapFile.Reader map, Text row, 
+  private void rowAtOrBeforeFromMapFile(MapFile.Reader map, final byte [] row, 
     SortedMap<HStoreKey, Long> candidateKeys)
   throws IOException {
     HStoreKey searchKey = null;
     ImmutableBytesWritable readval = new ImmutableBytesWritable();
     HStoreKey readkey = new HStoreKey();
     HStoreKey strippedKey = null;
-    
     synchronized(map) {
       // don't bother with the rest of this if the file is empty
       map.reset();
@@ -1462,7 +1467,6 @@
       }
 
       long now = System.currentTimeMillis();
-
       // if there aren't any candidate keys yet, we'll do some things slightly
       // different 
       if (candidateKeys.isEmpty()) {
@@ -1472,7 +1476,7 @@
         // save time and add the last key to the candidates.
         HStoreKey finalKey = new HStoreKey(); 
         map.finalKey(finalKey);
-        if (finalKey.getRow().compareTo(row) < 0) {
+        if (Bytes.compareTo(finalKey.getRow(), row) < 0) {
           candidateKeys.put(stripTimestamp(finalKey), 
             new Long(finalKey.getTimestamp()));
           return;
@@ -1489,7 +1493,7 @@
         do {
           // if we have an exact match on row, and it's not a delete, save this
           // as a candidate key
-          if (readkey.getRow().equals(row)) {
+          if (Bytes.equals(readkey.getRow(), row)) {
             if (!HLogEdit.isDeleted(readval.get())) {
               if (ttl == HConstants.FOREVER || 
                     now < readkey.getTimestamp() + ttl) {
@@ -1502,7 +1506,7 @@
                 }
               }
             }
-          } else if (readkey.getRow().compareTo(row) > 0 ) {
+          } else if (Bytes.compareTo(readkey.getRow(), row) > 0 ) {
             // if the row key we just read is beyond the key we're searching for,
             // then we're done. return.
             return;
@@ -1538,7 +1542,7 @@
         // save time and add the last key to the candidates.
         HStoreKey finalKey = new HStoreKey(); 
         map.finalKey(finalKey);
-        if (finalKey.getRow().compareTo(searchKey.getRow()) < 0) {
+        if (Bytes.compareTo(finalKey.getRow(), searchKey.getRow()) < 0) {
           strippedKey = stripTimestamp(finalKey);
           
           // if the candidate keys has a cell like this one already,
@@ -1568,7 +1572,7 @@
       do {
         // if we have an exact match on row, and it's not a delete, save this
         // as a candidate key
-        if (readkey.getRow().equals(row)) {
+        if (Bytes.equals(readkey.getRow(), row)) {
           strippedKey = stripTimestamp(readkey);
           if (!HLogEdit.isDeleted(readval.get())) {
             if (ttl == HConstants.FOREVER || 
@@ -1593,7 +1597,7 @@
               } 
             }
           }
-        } else if (readkey.getRow().compareTo(row) > 0 ) {
+        } else if (Bytes.compareTo(readkey.getRow(), row) > 0 ) {
           // if the row key we just read is beyond the key we're searching for,
           // then we're done. return.
           return;
@@ -1644,9 +1648,9 @@
    */
   private boolean cellMatches(HStoreKey origin, HStoreKey target){
     // if the origin's column is empty, then we're matching any column
-    if (origin.getColumn().equals(new Text())){
+    if (Bytes.equals(origin.getColumn(), HConstants.EMPTY_BYTE_ARRAY)){
       // if the row matches, then...
-      if (target.getRow().equals(origin.getRow())) {
+      if (Bytes.equals(target.getRow(), origin.getRow())) {
         // check the timestamp
         return target.getTimestamp() <= origin.getTimestamp();
       }
@@ -1665,9 +1669,9 @@
    */
   private boolean rowMatches(HStoreKey origin, HStoreKey target){
     // if the origin's column is empty, then we're matching any column
-    if (origin.getColumn().equals(new Text())){
+    if (Bytes.equals(origin.getColumn(), HConstants.EMPTY_BYTE_ARRAY)) {
       // if the row matches, then...
-      return target.getRow().equals(origin.getRow());
+      return Bytes.equals(target.getRow(), origin.getRow());
     }
     // otherwise, we want to match on row and column
     return target.matchesRowCol(origin);
@@ -1678,7 +1682,7 @@
    * 
    * @return midKey if store can be split, null otherwise
    */
-  Text checkSplit() {
+  byte [] checkSplit() {
     if (this.storefiles.size() <= 0) {
       return null;
     }
@@ -1726,14 +1730,14 @@
       if (mk != null) {
         // if the midkey is the same as the first and last keys, then we cannot
         // (ever) split this region. 
-        if (mk.getRow().equals(firstKey.getRow()) && 
-            mk.getRow().equals(lastKey.getRow())) {
+        if (Bytes.equals(mk.getRow(), firstKey.getRow()) && 
+            Bytes.equals(mk.getRow(), lastKey.getRow())) {
           return null;
         }
         return mk.getRow();
       }
     } catch(IOException e) {
-      LOG.warn("Failed getting store size for " + this.storeName, e);
+      LOG.warn("Failed getting store size for " + this.storeNameStr, e);
     } finally {
       this.lock.readLock().unlock();
     }
@@ -1752,8 +1756,8 @@
   /**
    * Return a scanner for both the memcache and the HStore files
    */
-  InternalScanner getScanner(long timestamp, Text targetCols[],
-      Text firstRow, RowFilterInterface filter)
+  InternalScanner getScanner(long timestamp, byte [][] targetCols,
+      byte [] firstRow, RowFilterInterface filter)
   throws IOException {
     lock.readLock().lock();
     try {
@@ -1766,7 +1770,7 @@
   /** {@inheritDoc} */
   @Override
   public String toString() {
-    return this.storeName.toString();
+    return this.storeNameStr;
   }
 
   /*

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreFile.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreFile.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreFile.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreFile.java Thu May 15 15:10:47 2008
@@ -39,10 +39,10 @@
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.hbase.io.BlockFSInputStream;
 import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
+import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.hbase.util.Writables;
 import org.apache.hadoop.io.MapFile;
 import org.apache.hadoop.io.SequenceFile;
-import org.apache.hadoop.io.Text;
 import org.apache.hadoop.io.Writable;
 import org.apache.hadoop.io.WritableComparable;
 import org.onelab.filter.Filter;
@@ -50,6 +50,7 @@
 
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HBaseConfiguration;
+import org.apache.hadoop.hbase.HRegionInfo;
 import org.apache.hadoop.hbase.HStoreKey;
 
 /**
@@ -113,8 +114,8 @@
   private final static Random rand = new Random();
 
   private final Path basedir;
-  private final String encodedRegionName;
-  private final Text colFamily;
+  private final int encodedRegionName;
+  private final byte [] colFamily;
   private final long fileId;
   private final HBaseConfiguration conf;
   private final FileSystem fs;
@@ -131,13 +132,13 @@
    * @throws IOException
    */
   HStoreFile(HBaseConfiguration conf, FileSystem fs, Path basedir,
-      String encodedRegionName, Text colFamily, long fileId,
+      int encodedRegionName, byte [] colFamily, long fileId,
       final Reference ref) throws IOException {
     this.conf = conf;
     this.fs = fs;
     this.basedir = basedir;
     this.encodedRegionName = encodedRegionName;
-    this.colFamily = new Text(colFamily);
+    this.colFamily = colFamily;
     
     long id = fileId;
     if (id == -1) {
@@ -145,7 +146,7 @@
       Path testpath = null;
       do {
         id = Math.abs(rand.nextLong());
-        testpath = new Path(mapdir, createHStoreFilename(id, null));
+        testpath = new Path(mapdir, createHStoreFilename(id, -1));
       } while(fs.exists(testpath));
     }
     this.fileId = id;
@@ -165,12 +166,12 @@
     return reference;
   }
 
-  String getEncodedRegionName() {
-    return encodedRegionName;
+  int getEncodedRegionName() {
+    return this.encodedRegionName;
   }
 
   /** @return the column family */
-  Text getColFamily() {
+  byte [] getColFamily() {
     return colFamily;
   }
 
@@ -187,18 +188,22 @@
       return getMapFilePath(encodedRegionName, fileId,
           reference.getEncodedRegionName());
     }
-    return getMapFilePath(encodedRegionName, fileId, null);
+    return getMapFilePath(this.encodedRegionName, fileId);
   }
 
   private Path getMapFilePath(final Reference r) {
     if (r == null) {
       return getMapFilePath();
     }
-    return getMapFilePath(r.getEncodedRegionName(), r.getFileId(), null);
+    return getMapFilePath(r.getEncodedRegionName(), r.getFileId());
   }
 
-  private Path getMapFilePath(final String encodedName, final long fid,
-      final String ern) {
+  private Path getMapFilePath(final int encodedName, final long fid) {
+    return getMapFilePath(encodedName, fid, HRegionInfo.NO_HASH);
+  }
+  
+  private Path getMapFilePath(final int encodedName, final long fid,
+      final int ern) {
     return new Path(HStoreFile.getMapDir(basedir, encodedName, colFamily), 
       createHStoreFilename(fid, ern));
   }
@@ -210,11 +215,15 @@
           reference.getEncodedRegionName());
  
     }
-    return getInfoFilePath(encodedRegionName, fileId, null);
+    return getInfoFilePath(encodedRegionName, fileId);
+  }
+  
+  private Path getInfoFilePath(final int encodedName, final long fid) {
+    return getInfoFilePath(encodedName, fid, HRegionInfo.NO_HASH);
   }
   
-  private Path getInfoFilePath(final String encodedName, final long fid,
-      final String ern) {
+  private Path getInfoFilePath(final int encodedName, final long fid,
+      final int ern) {
     return new Path(HStoreFile.getInfoDir(basedir, encodedName, colFamily), 
       createHStoreFilename(fid, ern));
   }
@@ -293,8 +302,7 @@
   long loadInfo(FileSystem fs) throws IOException {
     Path p = null;
     if (isReference()) {
-      p = getInfoFilePath(reference.getEncodedRegionName(),
-          reference.getFileId(), null);
+      p = getInfoFilePath(reference.getEncodedRegionName(), reference.getFileId());
     } else {
       p = getInfoFilePath();
     }
@@ -400,7 +408,6 @@
   public synchronized MapFile.Reader getReader(final FileSystem fs,
       final Filter bloomFilter, final boolean blockCacheEnabled)
   throws IOException {
-    
     if (isReference()) {
       return new HStoreFile.HalfMapFileReader(fs,
           getMapFilePath(reference).toString(), conf, 
@@ -475,55 +482,72 @@
     return r.equals(Range.top);
   }
 
+  private static String createHStoreFilename(final long fid) {
+    return createHStoreFilename(fid, HRegionInfo.NO_HASH);
+  }
+
   private static String createHStoreFilename(final long fid,
-      final String encodedRegionName) {
-    return Long.toString(fid) +
-      ((encodedRegionName != null) ? "." + encodedRegionName : "");
+      final int encodedRegionName) {
+    return Long.toString(fid) + 
+      ((encodedRegionName != HRegionInfo.NO_HASH)?
+        "." + encodedRegionName : "");
   }
-  
+
   /**
-   * @param dir
-   * @param encodedRegionName
-   * @param colFamily
+   * @param dir Base directory
+   * @param encodedRegionName Encoding of region name.
+   * @param f Column family.
    * @return path for map file directory
    */
-  public static Path getMapDir(Path dir, String encodedRegionName, Text colFamily) {
-    return new Path(dir, new Path(encodedRegionName, 
-        new Path(colFamily.toString(), HSTORE_DATFILE_DIR)));
+  public static Path getMapDir(Path dir, int encodedRegionName,
+      final byte [] f) {
+    return getFamilySubDir(dir, encodedRegionName, f, HSTORE_DATFILE_DIR);
   }
 
   /**
-   * @param dir
-   * @param encodedRegionName
-   * @param colFamily
+   * @param dir Base directory
+   * @param encodedRegionName Encoding of region name.
+   * @param f Column family.
    * @return the info directory path
    */
-  public static Path getInfoDir(Path dir, String encodedRegionName, Text colFamily) {
-    return new Path(dir, new Path(encodedRegionName, 
-        new Path(colFamily.toString(), HSTORE_INFO_DIR)));
+  public static Path getInfoDir(Path dir, int encodedRegionName, byte [] f) {
+    return getFamilySubDir(dir, encodedRegionName, f, HSTORE_INFO_DIR);
   }
 
   /**
-   * @param dir
-   * @param encodedRegionName
-   * @param colFamily
+   * @param dir Base directory
+   * @param encodedRegionName Encoding of region name.
+   * @param f Column family.
    * @return the bloom filter directory path
    */
-  public static Path getFilterDir(Path dir, String encodedRegionName, Text colFamily) {
-    return new Path(dir, new Path(encodedRegionName,
-        new Path(colFamily.toString(), HSTORE_FILTER_DIR)));
+  public static Path getFilterDir(Path dir, int encodedRegionName,
+      final byte [] f) {
+    return getFamilySubDir(dir, encodedRegionName, f, HSTORE_FILTER_DIR);
+  }
+  
+  /*
+   * @param base Base directory
+   * @param encodedRegionName Encoding of region name.
+   * @param f Column family.
+   * @param subdir Subdirectory to create under column family/store directory.
+   * @return
+   */
+  private static Path getFamilySubDir(final Path base,
+      final int encodedRegionName, final byte [] f, final String subdir) {
+    return new Path(base, new Path(Integer.toString(encodedRegionName),
+        new Path(Bytes.toString(f), subdir)));
   }
 
   /*
    * Data structure to hold reference to a store file over in another region.
    */
   static class Reference implements Writable {
-    private String encodedRegionName;
+    private int encodedRegionName;
     private long fileid;
     private Range region;
     private HStoreKey midkey;
     
-    Reference(final String ern, final long fid, final HStoreKey m,
+    Reference(final int ern, final long fid, final HStoreKey m,
         final Range fr) {
       this.encodedRegionName = ern;
       this.fileid = fid;
@@ -532,7 +556,7 @@
     }
     
     Reference() {
-      this(null, -1, null, Range.bottom);
+      this(-1, -1, null, Range.bottom);
     }
 
     long getFileId() {
@@ -547,8 +571,8 @@
       return midkey;
     }
     
-    String getEncodedRegionName() {
-      return encodedRegionName;
+    int getEncodedRegionName() {
+      return this.encodedRegionName;
     }
    
     /** {@inheritDoc} */
@@ -561,7 +585,7 @@
 
     /** {@inheritDoc} */
     public void write(DataOutput out) throws IOException {
-      out.writeUTF(encodedRegionName);
+      out.writeInt(this.encodedRegionName);
       out.writeLong(fileid);
       // Write true if we're doing top of the file.
       out.writeBoolean(isTopFileRegion(region));
@@ -570,7 +594,7 @@
 
     /** {@inheritDoc} */
     public void readFields(DataInput in) throws IOException {
-      encodedRegionName = in.readUTF();
+      this.encodedRegionName = in.readInt();
       fileid = in.readLong();
       boolean tmp = in.readBoolean();
       // If true, set region to top.

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreScanner.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreScanner.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreScanner.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/HStoreScanner.java Thu May 15 15:10:47 2008
@@ -21,19 +21,19 @@
 package org.apache.hadoop.hbase.regionserver;
 
 import java.io.IOException;
-import java.util.TreeMap;
-import java.util.SortedMap;
+import java.util.ArrayList;
 import java.util.Iterator;
-import java.util.Map;
 import java.util.List;
-import java.util.ArrayList;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
 
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
-import org.apache.hadoop.io.Text;
+import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HStoreKey;
 import org.apache.hadoop.hbase.filter.RowFilterInterface;
-import org.apache.hadoop.hbase.HConstants;
+import org.apache.hadoop.hbase.util.Bytes;
 
 /**
  * Scanner scans both the memcache and the HStore
@@ -42,7 +42,7 @@
   static final Log LOG = LogFactory.getLog(HStoreScanner.class);
 
   private InternalScanner[] scanners;
-  private TreeMap<Text, byte []>[] resultSets;
+  private TreeMap<byte [], byte []>[] resultSets;
   private HStoreKey[] keys;
   private boolean wildcardMatch = false;
   private boolean multipleMatchers = false;
@@ -51,8 +51,8 @@
   
   /** Create an Scanner with a handle on the memcache and HStore files. */
   @SuppressWarnings("unchecked")
-  HStoreScanner(HStore store, Text[] targetCols, Text firstRow, long timestamp,
-    RowFilterInterface filter) 
+  HStoreScanner(HStore store, byte [][] targetCols, byte [] firstRow,
+    long timestamp, RowFilterInterface filter) 
   throws IOException {
     this.store = store;
     this.dataFilter = filter;
@@ -87,7 +87,7 @@
     // All results will match the required column-set and scanTime.
     for (int i = 0; i < scanners.length; i++) {
       keys[i] = new HStoreKey();
-      resultSets[i] = new TreeMap<Text, byte []>();
+      resultSets[i] = new TreeMap<byte [], byte []>(Bytes.BYTES_COMPARATOR);
       if(scanners[i] != null && !scanners[i].next(keys[i], resultSets[i])) {
         closeScanner(i);
       }
@@ -105,7 +105,7 @@
   }
 
   /** {@inheritDoc} */
-  public boolean next(HStoreKey key, SortedMap<Text, byte[]> results)
+  public boolean next(HStoreKey key, SortedMap<byte [], byte[]> results)
     throws IOException {
 
     // Filtered flag is set by filters.  If a cell has been 'filtered out'
@@ -114,15 +114,15 @@
     boolean moreToFollow = true;
     while (filtered && moreToFollow) {
       // Find the lowest-possible key.
-      Text chosenRow = null;
+      byte [] chosenRow = null;
       long chosenTimestamp = -1;
       for (int i = 0; i < this.keys.length; i++) {
         if (scanners[i] != null &&
             (chosenRow == null ||
-            (keys[i].getRow().compareTo(chosenRow) < 0) ||
-            ((keys[i].getRow().compareTo(chosenRow) == 0) &&
+            (Bytes.compareTo(keys[i].getRow(), chosenRow) < 0) ||
+            ((Bytes.compareTo(keys[i].getRow(), chosenRow) == 0) &&
             (keys[i].getTimestamp() > chosenTimestamp)))) {
-          chosenRow = new Text(keys[i].getRow());
+          chosenRow = keys[i].getRow();
           chosenTimestamp = keys[i].getTimestamp();
         }
       }
@@ -136,7 +136,7 @@
         // Here we are setting the passed in key with current row+timestamp
         key.setRow(chosenRow);
         key.setVersion(chosenTimestamp);
-        key.setColumn(HConstants.EMPTY_TEXT);
+        key.setColumn(HConstants.EMPTY_BYTE_ARRAY);
         // Keep list of deleted cell keys within this row.  We need this
         // because as we go through scanners, the delete record may be in an
         // early scanner and then the same record with a non-delete, non-null
@@ -150,7 +150,7 @@
           while ((scanners[i] != null
               && !filtered
               && moreToFollow)
-              && (keys[i].getRow().compareTo(chosenRow) == 0)) {
+              && (Bytes.compareTo(keys[i].getRow(), chosenRow) == 0)) {
             // If we are doing a wild card match or there are multiple
             // matchers per column, we need to scan all the older versions of 
             // this row to pick up the rest of the family members
@@ -164,9 +164,9 @@
             // but this had the effect of overwriting newer
             // values with older ones. So now we only insert
             // a result if the map does not contain the key.
-            HStoreKey hsk = new HStoreKey(key.getRow(), HConstants.EMPTY_TEXT,
+            HStoreKey hsk = new HStoreKey(key.getRow(), HConstants.EMPTY_BYTE_ARRAY,
               key.getTimestamp());
-            for (Map.Entry<Text, byte[]> e : resultSets[i].entrySet()) {
+            for (Map.Entry<byte [], byte[]> e : resultSets[i].entrySet()) {
               hsk.setColumn(e.getKey());
               if (HLogEdit.isDeleted(e.getValue())) {
                 if (!deletes.contains(hsk)) {
@@ -202,7 +202,7 @@
         // If the current scanner is non-null AND has a lower-or-equal
         // row label, then its timestamp is bad. We need to advance it.
         while ((scanners[i] != null) &&
-            (keys[i].getRow().compareTo(chosenRow) <= 0)) {
+            (Bytes.compareTo(keys[i].getRow(), chosenRow) <= 0)) {
           resultSets[i].clear();
           if (!scanners[i].next(keys[i], resultSets[i])) {
             closeScanner(i);
@@ -266,7 +266,7 @@
     }
   }
 
-  public Iterator<Map.Entry<HStoreKey, SortedMap<Text, byte[]>>> iterator() {
+  public Iterator<Map.Entry<HStoreKey, SortedMap<byte [], byte[]>>> iterator() {
     throw new UnsupportedOperationException("Unimplemented serverside. " +
       "next(HStoreKey, StortedMap(...) is more efficient");
   }

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/InternalScanner.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/InternalScanner.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/InternalScanner.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/InternalScanner.java Thu May 15 15:10:47 2008
@@ -22,7 +22,6 @@
 import java.io.Closeable;
 import java.io.IOException;
 import java.util.SortedMap;
-import org.apache.hadoop.io.Text;
 import org.apache.hadoop.hbase.HStoreKey;
 
 /**
@@ -50,7 +49,7 @@
    * @return true if data was returned
    * @throws IOException
    */
-  public boolean next(HStoreKey key, SortedMap<Text, byte[]> results)
+  public boolean next(HStoreKey key, SortedMap<byte [], byte[]> results)
   throws IOException;
   
   /**

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/Memcache.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/Memcache.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/Memcache.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/Memcache.java Thu May 15 15:10:47 2008
@@ -24,14 +24,13 @@
 import java.rmi.UnexpectedException;
 import java.util.ArrayList;
 import java.util.Collections;
-import java.util.HashMap;
-import java.util.HashSet;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 import java.util.Set;
 import java.util.SortedMap;
 import java.util.TreeMap;
+import java.util.TreeSet;
 import java.util.concurrent.locks.ReentrantReadWriteLock;
 
 import org.apache.commons.logging.Log;
@@ -39,8 +38,7 @@
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HStoreKey;
 import org.apache.hadoop.hbase.io.Cell;
-import org.apache.hadoop.io.Text;
-import org.apache.hadoop.io.WritableComparable;
+import org.apache.hadoop.hbase.util.Bytes;
 
 
 /**
@@ -200,25 +198,25 @@
    * @return Return lowest of a or b or null if both a and b are null
    */
   @SuppressWarnings("unchecked")
-  private WritableComparable getLowest(final WritableComparable a,
-      final WritableComparable b) {
+  private byte [] getLowest(final byte [] a,
+      final byte [] b) {
     if (a == null) {
       return b;
     }
     if (b == null) {
       return a;
     }
-    return a.compareTo(b) <= 0? a: b;
+    return Bytes.compareTo(a, b) <= 0? a: b;
   }
 
   /**
    * @param row Find the row that comes after this one.
    * @return Next row or null if none found
    */
-  Text getNextRow(final Text row) {
+  byte [] getNextRow(final byte [] row) {
     this.lock.readLock().lock();
     try {
-      return (Text)getLowest(getNextRow(row, this.memcache),
+      return getLowest(getNextRow(row, this.memcache),
         getNextRow(row, this.snapshot));
     } finally {
       this.lock.readLock().unlock();
@@ -231,9 +229,9 @@
    * This method synchronizes on passed map while iterating it.
    * @return Next row or null if none found.
    */
-  private Text getNextRow(final Text row,
+  private byte [] getNextRow(final byte [] row,
       final SortedMap<HStoreKey, byte []> map) {
-    Text result = null;
+    byte [] result = null;
     // Synchronize on the map to make the tailMap making 'safe'.
     synchronized (map) {
       // Make an HSK with maximum timestamp so we get past most of the current
@@ -243,7 +241,7 @@
       // Iterate until we fall into the next row; i.e. move off current row
       for (Map.Entry<HStoreKey, byte []> es: tailMap.entrySet()) {
         HStoreKey itKey = es.getKey();
-        if (itKey.getRow().compareTo(row) <= 0) {
+        if (Bytes.compareTo(itKey.getRow(), row) <= 0) {
           continue;
         }
         // Note: Not suppressing deletes or expired cells.
@@ -262,8 +260,8 @@
    * @param deletes Map to accumulate deletes found.
    * @param results Where to stick row results found.
    */
-  void getFull(HStoreKey key, Set<Text> columns, Map<Text, Long> deletes, 
-    Map<Text, Cell> results) {
+  void getFull(HStoreKey key, Set<byte []> columns, Map<byte [], Long> deletes, 
+    Map<byte [], Cell> results) {
     this.lock.readLock().lock();
     try {
       // The synchronizations here are because internalGet iterates
@@ -279,34 +277,31 @@
   }
 
   private void internalGetFull(SortedMap<HStoreKey, byte[]> map, HStoreKey key, 
-    Set<Text> columns, Map<Text, Long> deletes, Map<Text, Cell> results) {
-
+      Set<byte []> columns, Map<byte [], Long> deletes,
+      Map<byte [], Cell> results) {
     if (map.isEmpty() || key == null) {
       return;
     }
-
     List<HStoreKey> victims = new ArrayList<HStoreKey>();
     SortedMap<HStoreKey, byte[]> tailMap = map.tailMap(key);
     long now = System.currentTimeMillis();
     for (Map.Entry<HStoreKey, byte []> es: tailMap.entrySet()) {
       HStoreKey itKey = es.getKey();
-      Text itCol = itKey.getColumn();
+      byte [] itCol = itKey.getColumn();
       if (results.get(itCol) == null && key.matchesWithoutColumn(itKey)) {
-        byte [] val = tailMap.get(itKey);
-
         if (columns == null || columns.contains(itKey.getColumn())) {
+          byte [] val = tailMap.get(itKey);
           if (HLogEdit.isDeleted(val)) {
             if (!deletes.containsKey(itCol) 
               || deletes.get(itCol).longValue() < itKey.getTimestamp()) {
-              deletes.put(new Text(itCol), Long.valueOf(itKey.getTimestamp()));
+              deletes.put(itCol, Long.valueOf(itKey.getTimestamp()));
             }
           } else if (!(deletes.containsKey(itCol) 
               && deletes.get(itCol).longValue() >= itKey.getTimestamp())) {
             // Skip expired cells
             if (ttl == HConstants.FOREVER ||
                   now < itKey.getTimestamp() + ttl) {
-              results.put(new Text(itCol),
-                new Cell(val, itKey.getTimestamp()));
+              results.put(itCol, new Cell(val, itKey.getTimestamp()));
             } else {
               victims.add(itKey);
               if (LOG.isDebugEnabled()) {
@@ -315,7 +310,7 @@
             }
           }
         }
-      } else if (key.getRow().compareTo(itKey.getRow()) < 0) {
+      } else if (Bytes.compareTo(key.getRow(), itKey.getRow()) < 0) {
         break;
       }
     }
@@ -329,7 +324,7 @@
    * @param candidateKeys Map of candidate keys (Accumulation over lots of
    * lookup over stores and memcaches)
    */
-  void getRowKeyAtOrBefore(final Text row, 
+  void getRowKeyAtOrBefore(final byte [] row, 
     SortedMap<HStoreKey, Long> candidateKeys) {
     this.lock.readLock().lock();
     try {
@@ -345,14 +340,12 @@
   }
 
   private void internalGetRowKeyAtOrBefore(SortedMap<HStoreKey, byte []> map,
-    Text key, SortedMap<HStoreKey, Long> candidateKeys) {
-    
+    byte [] key, SortedMap<HStoreKey, Long> candidateKeys) {
     HStoreKey strippedKey = null;
     
     // we want the earliest possible to start searching from
     HStoreKey search_key = candidateKeys.isEmpty() ? 
       new HStoreKey(key) : new HStoreKey(candidateKeys.firstKey().getRow());
-        
     Iterator<HStoreKey> key_iterator = null;
     HStoreKey found_key = null;
     ArrayList<HStoreKey> victims = new ArrayList<HStoreKey>();
@@ -363,14 +356,15 @@
     // if there are items in the tail map, there's either a direct match to
     // the search key, or a range of values between the first candidate key
     // and the ultimate search key (or the end of the cache)
-    if (!tailMap.isEmpty() && tailMap.firstKey().getRow().compareTo(key) <= 0) {
+    if (!tailMap.isEmpty() &&
+        Bytes.compareTo(tailMap.firstKey().getRow(), key) <= 0) {
       key_iterator = tailMap.keySet().iterator();
 
       // keep looking at cells as long as they are no greater than the 
       // ultimate search key and there's still records left in the map.
       do {
         found_key = key_iterator.next();
-        if (found_key.getRow().compareTo(key) <= 0) {
+        if (Bytes.compareTo(found_key.getRow(), key) <= 0) {
           strippedKey = stripTimestamp(found_key);
           if (HLogEdit.isDeleted(tailMap.get(found_key))) {
             if (candidateKeys.containsKey(strippedKey)) {
@@ -393,7 +387,7 @@
             }
           }
         }
-      } while (found_key.getRow().compareTo(key) <= 0 
+      } while (Bytes.compareTo(found_key.getRow(), key) <= 0 
         && key_iterator.hasNext());
     } else {
       // the tail didn't contain any keys that matched our criteria, or was 
@@ -412,13 +406,14 @@
         HStoreKey[] cells = 
           headMap.keySet().toArray(new HStoreKey[headMap.keySet().size()]);
            
-        Text lastRowFound = null;
+        byte [] lastRowFound = null;
         for(int i = cells.length - 1; i >= 0; i--) {
           HStoreKey thisKey = cells[i];
           
           // if the last row we found a candidate key for is different than
           // the row of the current candidate, we can stop looking.
-          if (lastRowFound != null && !lastRowFound.equals(thisKey.getRow())) {
+          if (lastRowFound != null &&
+              !Bytes.equals(lastRowFound, thisKey.getRow())) {
             break;
           }
           
@@ -587,10 +582,10 @@
       HStoreKey key = es.getKey();
   
       // if there's no column name, then compare rows and timestamps
-      if (origin.getColumn().toString().equals("")) {
+      if (origin.getColumn().length == 0) {
         // if the current and origin row don't match, then we can jump
         // out of the loop entirely.
-        if (!key.getRow().equals(origin.getRow())) {
+        if (!Bytes.equals(key.getRow(), origin.getRow())) {
           break;
         }
         // if the rows match but the timestamp is newer, skip it so we can
@@ -644,7 +639,7 @@
    * @return a scanner over the keys in the Memcache
    */
   InternalScanner getScanner(long timestamp,
-    Text targetCols[], Text firstRow)
+    final byte [][] targetCols, final byte [] firstRow)
   throws IOException {
     this.lock.readLock().lock();
     try {
@@ -660,11 +655,11 @@
   //////////////////////////////////////////////////////////////////////////////
 
   private class MemcacheScanner extends HAbstractScanner {
-    private Text currentRow;
-    private Set<Text> columns = null;
+    private byte [] currentRow;
+    private Set<byte []> columns = null;
     
-    MemcacheScanner(final long timestamp, final Text targetCols[],
-      final Text firstRow)
+    MemcacheScanner(final long timestamp, final byte [] targetCols[],
+      final byte [] firstRow)
     throws IOException {
       // Call to super will create ColumnMatchers and whether this is a regex
       // scanner or not.  Will also save away timestamp.  Also sorts rows.
@@ -675,7 +670,7 @@
       // columns.
       this.columns = null;
       if (!isWildcardScanner()) {
-        this.columns = new HashSet<Text>();
+        this.columns = new TreeSet<byte []>(Bytes.BYTES_COMPARATOR);
         for (int i = 0; i < targetCols.length; i++) {
           this.columns.add(targetCols[i]);
         }
@@ -684,15 +679,19 @@
 
     /** {@inheritDoc} */
     @Override
-    public boolean next(HStoreKey key, SortedMap<Text, byte []> results)
+    public boolean next(HStoreKey key, SortedMap<byte [], byte []> results)
     throws IOException {
       if (this.scannerClosed) {
         return false;
       }
-      Map<Text, Long> deletes = new HashMap<Text, Long>();
+      // This is a treemap rather than a Hashmap because then I can have a
+      // byte array as key -- because I can independently specify a comparator.
+      Map<byte [], Long> deletes =
+        new TreeMap<byte [], Long>(Bytes.BYTES_COMPARATOR);
       // Catch all row results in here.  These results are ten filtered to
       // ensure they match column name regexes, or if none, added to results.
-      Map<Text, Cell> rowResults = new HashMap<Text, Cell>();
+      Map<byte [], Cell> rowResults =
+        new TreeMap<byte [], Cell>(Bytes.BYTES_COMPARATOR);
       if (results.size() > 0) {
         results.clear();
       }
@@ -708,12 +707,12 @@
         key.setVersion(this.timestamp);
         getFull(key, isWildcardScanner() ? null : this.columns, deletes,
             rowResults);
-        for (Map.Entry<Text, Long> e: deletes.entrySet()) {
+        for (Map.Entry<byte [], Long> e: deletes.entrySet()) {
           rowResults.put(e.getKey(),
-              new Cell(HLogEdit.deleteBytes.get(), e.getValue()));
+            new Cell(HLogEdit.deleteBytes.get(), e.getValue().longValue()));
         }
-        for (Map.Entry<Text, Cell> e: rowResults.entrySet()) {
-          Text column = e.getKey();
+        for (Map.Entry<byte [], Cell> e: rowResults.entrySet()) {
+          byte [] column = e.getKey();
           Cell c = e.getValue();
           if (isWildcardScanner()) {
             // Check the results match.  We only check columns, not timestamps.
@@ -736,4 +735,4 @@
       }
     }
   }
-}
+}
\ No newline at end of file

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/RegionUnavailableListener.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/RegionUnavailableListener.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/RegionUnavailableListener.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/RegionUnavailableListener.java Thu May 15 15:10:47 2008
@@ -19,7 +19,6 @@
  */
 package org.apache.hadoop.hbase.regionserver;
 
-import org.apache.hadoop.io.Text;
 
 /**
  * Used as a callback mechanism so that an HRegion can notify the HRegionServer
@@ -33,12 +32,12 @@
    * outstanding transactions.
    * @param regionName
    */
-  public void closing(final Text regionName);
+  public void closing(final byte [] regionName);
   
   /**
    * <code>regionName</code> is closed and no longer available.
    * Listener should clean up any references to <code>regionName</code>
    * @param regionName
    */
-  public void closed(final Text regionName);
+  public void closed(final byte [] regionName);
 }

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/StoreFileScanner.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/StoreFileScanner.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/StoreFileScanner.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/regionserver/StoreFileScanner.java Thu May 15 15:10:47 2008
@@ -27,8 +27,8 @@
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HStoreKey;
 import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
+import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.io.MapFile;
-import org.apache.hadoop.io.Text;
 
 /**
  * A scanner that iterates through HStore files
@@ -50,7 +50,7 @@
   private final ReentrantReadWriteLock lock = new ReentrantReadWriteLock();
   
   public StoreFileScanner(final HStore store, final long timestamp,
-    final Text[] targetCols, final Text firstRow)
+    final byte [][] targetCols, final byte [] firstRow)
   throws IOException {
     super(timestamp, targetCols);
     this.store = store;
@@ -71,7 +71,7 @@
    * @param firstRow
    * @throws IOException
    */
-  private void openReaders(final Text firstRow) throws IOException {
+  private void openReaders(final byte [] firstRow) throws IOException {
     if (this.readers != null) {
       for (int i = 0; i < this.readers.length; i++) {
         this.readers[i].close();
@@ -92,7 +92,7 @@
     // Advance the readers to the first pos.
     for (i = 0; i < readers.length; i++) {
       keys[i] = new HStoreKey();
-      if (firstRow.getLength() != 0) {
+      if (firstRow != null && firstRow.length != 0) {
         if (findFirstRow(i, firstRow)) {
           continue;
         }
@@ -130,7 +130,7 @@
    * @see org.apache.hadoop.hbase.regionserver.InternalScanner#next(org.apache.hadoop.hbase.HStoreKey, java.util.SortedMap)
    */
   @Override
-  public boolean next(HStoreKey key, SortedMap<Text, byte []> results)
+  public boolean next(HStoreKey key, SortedMap<byte [], byte []> results)
   throws IOException {
     if (this.scannerClosed) {
       return false;
@@ -145,12 +145,11 @@
       if (viableRow.getRow() != null) {
         key.setRow(viableRow.getRow());
         key.setVersion(viableRow.getTimestamp());
-        key.setColumn(new Text(""));
 
         for (int i = 0; i < keys.length; i++) {
           // Fetch the data
           while ((keys[i] != null)
-              && (keys[i].getRow().compareTo(viableRow.getRow()) == 0)) {
+              && (Bytes.compareTo(keys[i].getRow(), viableRow.getRow()) == 0)) {
 
             // If we are doing a wild card match or there are multiple matchers
             // per column, we need to scan all the older versions of this row
@@ -164,7 +163,7 @@
             if(columnMatch(i)) {              
               // We only want the first result for any specific family member
               if(!results.containsKey(keys[i].getColumn())) {
-                results.put(new Text(keys[i].getColumn()), vals[i]);
+                results.put(keys[i].getColumn(), vals[i]);
                 insertedItem = true;
               }
             }
@@ -177,7 +176,7 @@
           // Advance the current scanner beyond the chosen row, to
           // a valid timestamp, so we're ready next time.
           while ((keys[i] != null)
-              && ((keys[i].getRow().compareTo(viableRow.getRow()) <= 0)
+              && ((Bytes.compareTo(keys[i].getRow(), viableRow.getRow()) <= 0)
                   || (keys[i].getTimestamp() > this.timestamp)
                   || (! columnMatch(i)))) {
             getNext(i);
@@ -192,19 +191,19 @@
   
   // Data stucture to hold next, viable row (and timestamp).
   class ViableRow {
-    private final Text row;
+    private final byte [] row;
     private final long ts;
 
-    ViableRow(final Text r, final long t) {
+    ViableRow(final byte [] r, final long t) {
       this.row = r;
       this.ts = t;
     }
 
-    public Text getRow() {
+    byte [] getRow() {
       return this.row;
     }
 
-    public long getTimestamp() {
+    long getTimestamp() {
       return this.ts;
     }
   }
@@ -215,7 +214,7 @@
    */
   private ViableRow getNextViableRow() throws IOException {
     // Find the next viable row label (and timestamp).
-    Text viableRow = null;
+    byte [] viableRow = null;
     long viableTimestamp = -1;
     long now = System.currentTimeMillis();
     long ttl = store.ttl;
@@ -224,11 +223,11 @@
           && (columnMatch(i))
           && (keys[i].getTimestamp() <= this.timestamp)
           && ((viableRow == null)
-              || (keys[i].getRow().compareTo(viableRow) < 0)
-              || ((keys[i].getRow().compareTo(viableRow) == 0)
+              || (Bytes.compareTo(keys[i].getRow(), viableRow) < 0)
+              || ((Bytes.compareTo(keys[i].getRow(), viableRow) == 0)
                   && (keys[i].getTimestamp() > viableTimestamp)))) {
         if (ttl == HConstants.FOREVER || now < keys[i].getTimestamp() + ttl) {
-          viableRow = new Text(keys[i].getRow());
+          viableRow = keys[i].getRow();
           viableTimestamp = keys[i].getTimestamp();
         } else {
           if (LOG.isDebugEnabled()) {
@@ -248,7 +247,7 @@
    * @param firstRow seek to this row
    * @return true if this is the first row or if the row was not found
    */
-  boolean findFirstRow(int i, Text firstRow) throws IOException {
+  boolean findFirstRow(int i, final byte [] firstRow) throws IOException {
     ImmutableBytesWritable ibw = new ImmutableBytesWritable();
     HStoreKey firstKey
       = (HStoreKey)readers[i].getClosest(new HStoreKey(firstRow), ibw);
@@ -350,7 +349,8 @@
       // up so future call to next will start here.
       ViableRow viableRow = getNextViableRow();
       openReaders(viableRow.getRow());
-      LOG.debug("Replaced Scanner Readers at row " + viableRow.getRow());
+      LOG.debug("Replaced Scanner Readers at row " +
+        Bytes.toString(viableRow.getRow()));
     } finally {
       this.lock.writeLock().unlock();
     }

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/GenericHandler.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/GenericHandler.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/GenericHandler.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/GenericHandler.java Thu May 15 15:10:47 2008
@@ -32,13 +32,12 @@
 import org.apache.hadoop.hbase.client.HBaseAdmin;
 import org.apache.hadoop.hbase.HBaseConfiguration;
 import org.apache.hadoop.hbase.HConstants;
-import org.apache.hadoop.io.Text;
-import org.mortbay.servlet.MultiPartResponse;
 import org.znerd.xmlenc.LineBreak;
 import org.znerd.xmlenc.XMLOutputter;
 
 import org.apache.hadoop.hbase.client.HTable;
 import org.apache.hadoop.hbase.io.Cell;
+import org.apache.hadoop.hbase.util.Bytes;
 
 /**
  * GenericHandler contains some basic common stuff that all the individual
@@ -228,13 +227,12 @@
    * @throws IOException
    */
   protected void outputColumnsXml(final XMLOutputter outputter,
-    final Map<Text, Cell> m)
+    final Map<byte [], Cell> m)
   throws IllegalStateException, IllegalArgumentException, IOException {
-    for (Map.Entry<Text, Cell> e: m.entrySet()) {
+    for (Map.Entry<byte [], Cell> e: m.entrySet()) {
       outputter.startTag(COLUMN);
       doElement(outputter, "name", 
-        org.apache.hadoop.hbase.util.Base64.encodeBytes(
-          e.getKey().getBytes()));
+        org.apache.hadoop.hbase.util.Base64.encodeBytes(e.getKey()));
       // We don't know String from binary data so we always base64 encode.
       doElement(outputter, "value",
         org.apache.hadoop.hbase.util.Base64.encodeBytes(e.getValue().getValue()));
@@ -259,6 +257,6 @@
    * Get an HTable instance by it's table name.
    */
   protected HTable getTable(final String tableName) throws IOException {
-    return new HTable(this.conf, new Text(tableName));
+    return new HTable(this.conf, Bytes.toBytes(tableName));
   }
 }

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/ScannerHandler.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/ScannerHandler.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/ScannerHandler.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/ScannerHandler.java Thu May 15 15:10:47 2008
@@ -35,6 +35,7 @@
 import org.apache.hadoop.hbase.client.Scanner;
 import org.apache.hadoop.hbase.HBaseConfiguration;
 import org.apache.hadoop.hbase.HConstants;
+import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.hbase.util.JenkinsHash;
 import org.apache.hadoop.hbase.io.Cell;
 import org.apache.hadoop.hbase.io.RowResult;
@@ -178,7 +179,7 @@
     
     // write the row key
     doElement(outputter, "name", 
-      org.apache.hadoop.hbase.util.Base64.encodeBytes(rowResult.getRow().getBytes()));
+      org.apache.hadoop.hbase.util.Base64.encodeBytes(rowResult.getRow()));
     
     outputColumnsXml(outputter, rowResult);
     outputter.endTag();
@@ -244,14 +245,14 @@
     
     // get the list of columns we're supposed to interact with
     String[] raw_columns = request.getParameterValues(COLUMN);
-    Text [] columns = null;
+    byte [][] columns = null;
     
     if (raw_columns != null) {
-      columns = new Text [raw_columns.length];
+      columns = new byte [raw_columns.length][];
       for (int i = 0; i < raw_columns.length; i++) {
         // I think this decoding is redundant.
         columns[i] =
-          new Text(URLDecoder.decode(raw_columns[i], HConstants.UTF8_ENCODING));
+          Bytes.toBytes(URLDecoder.decode(raw_columns[i], HConstants.UTF8_ENCODING));
       }
     } else {
       // TODO: Need to put into the scanner all of the table's column
@@ -264,14 +265,14 @@
     String raw_ts = request.getParameter(TIMESTAMP);
 
     // TODO: Are these decodings redundant?
-    Text startRow = request.getParameter(START_ROW) == null?
+    byte [] startRow = request.getParameter(START_ROW) == null?
       HConstants.EMPTY_START_ROW:
-      new Text(URLDecoder.decode(request.getParameter(START_ROW),
+      Bytes.toBytes(URLDecoder.decode(request.getParameter(START_ROW),
         HConstants.UTF8_ENCODING));
     // Empty start row is same value as empty end row.
-    Text endRow = request.getParameter(END_ROW) == null?
+    byte [] endRow = request.getParameter(END_ROW) == null?
       HConstants.EMPTY_START_ROW:
-      new Text(URLDecoder.decode(request.getParameter(END_ROW),
+      Bytes.toBytes(URLDecoder.decode(request.getParameter(END_ROW),
         HConstants.UTF8_ENCODING));
 
     Scanner scanner = (request.getParameter(END_ROW) == null)?

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/TableHandler.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/TableHandler.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/TableHandler.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/rest/TableHandler.java Thu May 15 15:10:47 2008
@@ -22,10 +22,11 @@
 import java.io.IOException;
 import java.io.PrintWriter;
 import java.net.URLDecoder;
-import java.util.HashMap;
+import java.util.Collection;
 import java.util.HashSet;
 import java.util.Map;
 import java.util.Set;
+import java.util.TreeMap;
 
 import javax.servlet.ServletException;
 import javax.servlet.http.HttpServletRequest;
@@ -33,17 +34,16 @@
 import javax.xml.parsers.DocumentBuilder;
 import javax.xml.parsers.DocumentBuilderFactory;
 
-import org.apache.hadoop.hbase.client.HBaseAdmin;
 import org.apache.hadoop.hbase.HBaseConfiguration;
 import org.apache.hadoop.hbase.HColumnDescriptor;
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HTableDescriptor;
-import org.apache.hadoop.io.Text;
+import org.apache.hadoop.hbase.client.HBaseAdmin;
 import org.apache.hadoop.hbase.client.HTable;
-import org.apache.hadoop.hbase.io.Cell;
 import org.apache.hadoop.hbase.io.BatchUpdate;
-
-import org.mortbay.servlet.MultiPartResponse;
+import org.apache.hadoop.hbase.io.Cell;
+import org.apache.hadoop.hbase.util.Bytes;
+import org.apache.hadoop.io.Text;
 import org.w3c.dom.Document;
 import org.w3c.dom.Element;
 import org.w3c.dom.Node;
@@ -135,9 +135,9 @@
       // They want full row returned. 
 
       // Presumption is that this.table has already been focused on target table.
-      Map<Text, Cell> result = timestampStr == null ? 
-        table.getRow(new Text(row)) 
-        : table.getRow(new Text(row), Long.parseLong(timestampStr));
+      Map<byte [], Cell> result = timestampStr == null ? 
+        table.getRow(Bytes.toBytes(row)) 
+        : table.getRow(Bytes.toBytes(row), Long.parseLong(timestampStr));
         
       if (result == null || result.size() == 0) {
         doNotFound(response, "Row not found!");
@@ -153,7 +153,7 @@
         }
       }
     } else {
-      Map<Text, Cell> prefiltered_result = table.getRow(new Text(row));
+      Map<byte [], Cell> prefiltered_result = table.getRow(Bytes.toBytes(row));
     
       if (prefiltered_result == null || prefiltered_result.size() == 0) {
         doNotFound(response, "Row not found!");
@@ -166,16 +166,14 @@
         }
   
         // output map that will contain the filtered results
-        Map<Text, Cell> m = new HashMap<Text, Cell>();
+        Map<byte [], Cell> m =
+          new TreeMap<byte [], Cell>(Bytes.BYTES_COMPARATOR);
 
         // get an array of all the columns retrieved
-        Text[] columns_retrieved = 
-          prefiltered_result.keySet().toArray(
-            new Text[prefiltered_result.keySet().size()]);
+        Set<byte []> columns_retrieved = prefiltered_result.keySet();
 
         // copy over those cells with requested column names
-        for(int i = 0; i < columns_retrieved.length; i++){
-          Text current_column = (Text)columns_retrieved[i];
+        for(byte [] current_column: columns_retrieved) {
           if(requested_columns_set.contains(current_column.toString())){
             m.put(current_column, prefiltered_result.get(current_column));            
           }
@@ -201,7 +199,7 @@
    * @throws IOException
    */
   private void outputRowXml(final HttpServletResponse response,
-      final Map<Text, Cell> result)
+      final Map<byte [], Cell> result)
   throws IOException {
     setResponseHeader(response, result.size() > 0? 200: 204,
         ContentType.XML.toString());
@@ -349,7 +347,7 @@
     final HttpServletResponse response)
   throws IOException {
     // Presumption is that this.table has already been focused on target table.
-    Text [] startKeys = table.getStartKeys();
+    byte [][] startKeys = table.getStartKeys();
     // Presumption is that this.table has already been set against target table
     switch (ContentType.getContentType(request.getHeader(ACCEPT))) {
       case XML:
@@ -410,18 +408,16 @@
         outputter.startTag("table");
         doElement(outputter, "name", descriptor.getName().toString());
         outputter.startTag("columnfamilies");
-        for (Map.Entry<Text, HColumnDescriptor> e:
-            descriptor.getFamilies().entrySet()) {
+        for (HColumnDescriptor e: descriptor.getFamilies()) {
           outputter.startTag("columnfamily");
-          doElement(outputter, "name", e.getKey().toString());
-          HColumnDescriptor hcd = e.getValue();
-          doElement(outputter, "compression", hcd.getCompression().toString());
+          doElement(outputter, "name", Bytes.toString(e.getName()));
+          doElement(outputter, "compression", e.getCompression().toString());
           doElement(outputter, "bloomfilter",
-            hcd.getBloomFilter() == null? "NONE": hcd.getBloomFilter().toString());
+            e.getBloomFilter() == null? "NONE": e.getBloomFilter().toString());
           doElement(outputter, "max-versions",
-            Integer.toString(hcd.getMaxVersions()));
+            Integer.toString(e.getMaxVersions()));
           doElement(outputter, "maximum-cell-size",
-              Integer.toString(hcd.getMaxValueLength()));
+              Integer.toString(e.getMaxValueLength()));
           outputter.endTag();
         }
         outputter.endTag();

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftServer.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftServer.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftServer.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftServer.java Thu May 15 15:10:47 2008
@@ -22,7 +22,6 @@
 import java.nio.charset.MalformedInputException;
 import java.util.AbstractMap;
 import java.util.ArrayList;
-import java.util.Arrays;
 import java.util.HashMap;
 import java.util.Map;
 import java.util.TreeMap;
@@ -30,13 +29,17 @@
 
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
-import org.apache.hadoop.hbase.client.HBaseAdmin;
 import org.apache.hadoop.hbase.HBaseConfiguration;
 import org.apache.hadoop.hbase.HColumnDescriptor;
 import org.apache.hadoop.hbase.HConstants;
-import org.apache.hadoop.hbase.HStoreKey;
 import org.apache.hadoop.hbase.HTableDescriptor;
 import org.apache.hadoop.hbase.MasterNotRunningException;
+import org.apache.hadoop.hbase.client.HBaseAdmin;
+import org.apache.hadoop.hbase.client.HTable;
+import org.apache.hadoop.hbase.client.Scanner;
+import org.apache.hadoop.hbase.io.BatchUpdate;
+import org.apache.hadoop.hbase.io.Cell;
+import org.apache.hadoop.hbase.io.RowResult;
 import org.apache.hadoop.hbase.thrift.generated.AlreadyExists;
 import org.apache.hadoop.hbase.thrift.generated.ColumnDescriptor;
 import org.apache.hadoop.hbase.thrift.generated.Hbase;
@@ -46,12 +49,8 @@
 import org.apache.hadoop.hbase.thrift.generated.NotFound;
 import org.apache.hadoop.hbase.thrift.generated.RegionDescriptor;
 import org.apache.hadoop.hbase.thrift.generated.ScanEntry;
+import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.io.Text;
-import org.apache.hadoop.hbase.client.HTable;
-import org.apache.hadoop.hbase.client.Scanner;
-import org.apache.hadoop.hbase.io.Cell;
-import org.apache.hadoop.hbase.io.RowResult;
-import org.apache.hadoop.hbase.io.BatchUpdate;
 
 import com.facebook.thrift.TException;
 import com.facebook.thrift.protocol.TBinaryProtocol;
@@ -149,13 +148,13 @@
      * @throws IllegalArgument
      * @throws IOError     
      */
-    Text getText(byte[] buf) throws IOError {
+    byte [] getText(byte[] buf) throws IOError {
       try {
         Text.validateUTF8(buf);
       } catch (MalformedInputException e) {
         throw new IOError("invalid UTF-8 encoding in row or column name");
       }
-      return new Text(buf);
+      return buf;
     }
     
     //
@@ -183,7 +182,7 @@
       LOG.debug("getTableRegions: " + new String(tableName));
       try {
         HTable table = getTable(tableName);
-        Text[] startKeys = table.getStartKeys();
+        byte [][] startKeys = table.getStartKeys();
         ArrayList<RegionDescriptor> regions = new ArrayList<RegionDescriptor>();
         for (int i = 0; i < startKeys.length; i++) {
           RegionDescriptor region = new RegionDescriptor();
@@ -276,12 +275,13 @@
       }
       try {
         HTable table = getTable(tableName);
-        Map<Text, Cell> values = 
+        Map<byte [], Cell> values = 
           table.getRow(getText(row), timestamp);
         // copy the map from type <Text, Cell> to <byte[], byte[]>
-        HashMap<byte[], byte[]> returnValues = new HashMap<byte[], byte[]>();
-        for (Entry<Text, Cell> e : values.entrySet()) {
-          returnValues.put(e.getKey().getBytes(), e.getValue().getValue());
+        TreeMap<byte[], byte[]> returnValues =
+          new TreeMap<byte[], byte[]>(Bytes.BYTES_COMPARATOR);
+        for (Entry<byte [], Cell> e : values.entrySet()) {
+          returnValues.put(e.getKey(), e.getValue().getValue());
         }
         return returnValues;
       } catch (IOException e) {
@@ -353,11 +353,11 @@
         LOG.debug("createTable: table=" + new String(tableName));
       }
       try {
-        Text tableStr = getText(tableName);
+        byte [] tableStr = getText(tableName);
         if (admin.tableExists(tableStr)) {
           throw new AlreadyExists("table name already in use");
         }
-        HTableDescriptor desc = new HTableDescriptor(tableStr.toString());
+        HTableDescriptor desc = new HTableDescriptor(tableStr);
         for (ColumnDescriptor col : columnFamilies) {
           HColumnDescriptor colDesc = ThriftUtilities.colDescFromThrift(col);
           desc.addFamily(colDesc);
@@ -378,7 +378,7 @@
         LOG.debug("deleteTable: table=" + new String(tableName));
       }
       try {
-        Text tableStr = getText(tableName);
+        byte [] tableStr = getText(tableName);
         if (!admin.tableExists(tableStr)) {
           throw new NotFound();
         }
@@ -460,11 +460,11 @@
       }
       
       ScanEntry retval = new ScanEntry();
-      retval.row = results.getRow().getBytes();
-      retval.columns = new HashMap<byte[], byte[]>(results.size());
+      retval.row = results.getRow();
+      retval.columns = new TreeMap<byte[], byte[]>(Bytes.BYTES_COMPARATOR);
       
-      for (Map.Entry<Text, Cell> e : results.entrySet()) {
-        retval.columns.put(e.getKey().getBytes(), e.getValue().getValue());
+      for (Map.Entry<byte [], Cell> e : results.entrySet()) {
+        retval.columns.put(e.getKey(), e.getValue().getValue());
       }
       return retval;
     }
@@ -477,7 +477,7 @@
       }
       try {
         HTable table = getTable(tableName);
-        Text[] columnsText = new Text[columns.size()];
+        byte [][] columnsText = new byte[columns.size()][];
         for (int i = 0; i < columns.size(); ++i) {
           columnsText[i] = getText(columns.get(i));
         }
@@ -498,7 +498,7 @@
       }
       try {
         HTable table = getTable(tableName);
-        Text[] columnsText = new Text[columns.size()];
+        byte [][] columnsText = new byte[columns.size()][];
         for (int i = 0; i < columns.size(); ++i) {
           columnsText[i] = getText(columns.get(i));
         }
@@ -519,7 +519,7 @@
       }
       try {
         HTable table = getTable(tableName);
-        Text[] columnsText = new Text[columns.size()];
+        byte [][] columnsText = new byte[columns.size()][];
         for (int i = 0; i < columns.size(); ++i) {
           columnsText[i] = getText(columns.get(i));
         }
@@ -541,7 +541,7 @@
       }
       try {
         HTable table = getTable(tableName);
-        Text[] columnsText = new Text[columns.size()];
+        byte [][] columnsText = new byte[columns.size()][];
         for (int i = 0; i < columns.size(); ++i) {
           columnsText[i] = getText(columns.get(i));
         }
@@ -559,13 +559,14 @@
         LOG.debug("getColumnDescriptors: table=" + new String(tableName));
       }
       try {
-        HashMap<byte[], ColumnDescriptor> columns = new HashMap<byte[], ColumnDescriptor>();
+        TreeMap<byte[], ColumnDescriptor> columns =
+          new TreeMap<byte[], ColumnDescriptor>(Bytes.BYTES_COMPARATOR);
         
         HTable table = getTable(tableName);
         HTableDescriptor desc = table.getMetadata();
         
-        for (Entry<Text, HColumnDescriptor> e : desc.families().entrySet()) {
-          ColumnDescriptor col = ThriftUtilities.colDescFromHbase(e.getValue());
+        for (HColumnDescriptor e : desc.getFamilies()) {
+          ColumnDescriptor col = ThriftUtilities.colDescFromHbase(e);
           columns.put(col.name, col);
         }
         return columns;

Modified: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftUtilities.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftUtilities.java?rev=656868&r1=656867&r2=656868&view=diff
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftUtilities.java (original)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/thrift/ThriftUtilities.java Thu May 15 15:10:47 2008
@@ -23,7 +23,6 @@
 import org.apache.hadoop.hbase.HColumnDescriptor.CompressionType;
 import org.apache.hadoop.hbase.thrift.generated.ColumnDescriptor;
 import org.apache.hadoop.hbase.thrift.generated.IllegalArgument;
-import org.apache.hadoop.io.Text;
 
 public class ThriftUtilities {
   
@@ -57,7 +56,7 @@
     if (in.name == null || in.name.length <= 0) {
       throw new IllegalArgument("column name is empty");
     }
-    HColumnDescriptor col = new HColumnDescriptor(new Text(in.name),
+    HColumnDescriptor col = new HColumnDescriptor(in.name,
         in.maxVersions, comp, in.inMemory, in.blockCacheEnabled,
         in.maxValueLength, in.timeToLive, bloom);
     return col;
@@ -73,7 +72,7 @@
    */
   static public ColumnDescriptor colDescFromHbase(HColumnDescriptor in) {
     ColumnDescriptor col = new ColumnDescriptor();
-    col.name = in.getName().getBytes();
+    col.name = in.getName();
     col.maxVersions = in.getMaxVersions();
     col.compression = in.getCompression().toString();
     col.inMemory = in.isInMemory();
@@ -88,4 +87,4 @@
     return col;
   }
   
-}
+}
\ No newline at end of file

Added: hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/util/Bytes.java
URL: http://svn.apache.org/viewvc/hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/util/Bytes.java?rev=656868&view=auto
==============================================================================
--- hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/util/Bytes.java (added)
+++ hadoop/hbase/trunk/src/java/org/apache/hadoop/hbase/util/Bytes.java Thu May 15 15:10:47 2008
@@ -0,0 +1,269 @@
+package org.apache.hadoop.hbase.util;
+
+import java.io.DataInput;
+import java.io.DataOutput;
+import java.io.IOException;
+import java.io.UnsupportedEncodingException;
+import java.nio.ByteBuffer;
+import java.util.Comparator;
+
+import org.apache.hadoop.hbase.HConstants;
+import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
+import org.apache.hadoop.io.Text;
+import org.apache.hadoop.io.WritableComparator;
+
+public class Bytes {
+  /**
+   * Size of long in bytes
+   */
+  public static final int SIZEOF_LONG = Long.SIZE/Byte.SIZE;
+
+  /**
+   * Size of int in bytes
+   */
+  public static final int SIZEOF_INT = Integer.SIZE/Byte.SIZE;
+
+  /**
+   * Pass this to TreeMaps where byte [] are keys.
+   */
+  public static Comparator<byte []> BYTES_COMPARATOR =
+      new Comparator<byte []>() {
+    public int compare(byte [] left, byte [] right) {
+      return compareTo(left, right);
+    }
+  };
+  
+  /**
+   * @param in Input to read from.
+   * @return byte array read off <code>in</code>
+   * @throws IOException 
+   */
+  public static byte [] readByteArray(final DataInput in)
+  throws IOException {
+    byte [] result = new byte[in.readInt()];
+    in.readFully(result, 0, result.length);
+    return result;
+  }
+  
+  /**
+   * @param out
+   * @param b
+   * @throws IOException
+   */
+  public static void writeByteArray(final DataOutput out, final byte [] b)
+  throws IOException {
+    out.writeInt(b.length);
+    out.write(b, 0, b.length);
+  }
+  
+  /**
+   * @param b Presumed UTF-8 encoded byte array.
+   * @return String made from <code>b</code>
+   */
+  public static String toString(final byte [] b) {
+    String result = null;
+    try {
+      result = new String(b, HConstants.UTF8_ENCODING);
+    } catch (UnsupportedEncodingException e) {
+      e.printStackTrace();
+    }
+    return result;
+  }
+  
+  
+  /**
+   * Converts a string to a UTF-8 byte array.
+   * @param s
+   * @return the byte array
+   */
+  public static byte[] toBytes(String s) {
+    if (s == null) {
+      throw new IllegalArgumentException("string cannot be null");
+    }
+    byte [] result = null;
+    try {
+      result = s.getBytes(HConstants.UTF8_ENCODING);
+    } catch (UnsupportedEncodingException e) {
+      e.printStackTrace();
+    }
+    return result;
+  }
+
+  /**
+   * Convert a long value to a byte array
+   * @param val
+   * @return the byte array
+   */
+  public static byte[] toBytes(final long val) {
+    ByteBuffer bb = ByteBuffer.allocate(SIZEOF_LONG);
+    bb.putLong(val);
+    return bb.array();
+  }
+
+  /**
+   * Converts a byte array to a long value
+   * @param bytes
+   * @return the long value
+   */
+  public static long toLong(byte[] bytes) {
+    if (bytes == null || bytes.length == 0) {
+      return -1L;
+    }
+    return ByteBuffer.wrap(bytes).getLong();
+  }
+  
+  /**
+   * Convert an int value to a byte array
+   * @param val
+   * @return the byte array
+   */
+  public static byte[] toBytes(final int val) {
+    ByteBuffer bb = ByteBuffer.allocate(SIZEOF_INT);
+    bb.putInt(val);
+    return bb.array();
+  }
+
+  /**
+   * Converts a byte array to a long value
+   * @param bytes
+   * @return the long value
+   */
+  public static long toInt(byte[] bytes) {
+    if (bytes == null || bytes.length == 0) {
+      return -1L;
+    }
+    return ByteBuffer.wrap(bytes).getInt();
+  }
+
+  /**
+   * @param left
+   * @param right
+   * @return 0 if equal, < 0 if left is less than right, etc.
+   */
+  public static int compareTo(final byte [] left, final byte [] right) {
+    return compareTo(left, 0, left.length, right, 0, right.length);
+  }
+
+  /**
+   * @param left
+   * @param right
+   * @param leftOffset Where to start comparing in the left buffer
+   * @param rightOffset Where to start comparing in the right buffer
+   * @param leftLength How much to compare from the left buffer
+   * @param rightLength How much to compare from the right buffer
+   * @return 0 if equal, < 0 if left is less than right, etc.
+   */
+  public static int compareTo(final byte [] left, final int leftOffset,
+      final int leftLength, final byte [] right, final int rightOffset,
+      final int rightLength) {
+    return WritableComparator.compareBytes(left,leftOffset, leftLength,
+        right, rightOffset, rightLength);
+  }
+  
+  /**
+   * @param left
+   * @param right
+   * @return True if equal
+   */
+  public static boolean equals(final byte [] left, final byte [] right) {
+    return left == null && right == null? true:
+      left == null && right != null? false:
+      left != null && right == null? false:
+      left.length != right.length? false:
+        compareTo(left, right) == 0;
+  }
+  
+  /**
+   * @param b
+   * @return Runs {@link WritableComparator#hashBytes(byte[], int)} on the
+   * passed in array.  This method is what {@link Text} and
+   * {@link ImmutableBytesWritable} use calculating hash code.
+   */
+  public static int hashCode(final byte [] b) {
+    return hashCode(b, b.length);
+  }
+
+  /**
+   * @param b
+   * @return Runs {@link WritableComparator#hashBytes(byte[], int)} on the
+   * passed in array.  This method is what {@link Text} and
+   * {@link ImmutableBytesWritable} use calculating hash code.
+   */
+  public static int hashCode(final byte [] b, final int length) {
+    return WritableComparator.hashBytes(b, length);
+  }
+
+  /**
+   * @param b
+   * @return A hash of <code>b</code> as an Integer that can be used as key in
+   * Maps.
+   */
+  public static Integer mapKey(final byte [] b) {
+    return Integer.valueOf(hashCode(b));
+  }
+
+  /**
+   * @param b
+   * @return A hash of <code>b</code> as an Integer that can be used as key in
+   * Maps.
+   */
+  public static Integer mapKey(final byte [] b, final int length) {
+    return Integer.valueOf(hashCode(b, length));
+  }
+
+  /**
+   * @param a
+   * @param b
+   * @return New array that has a in lower half and b in upper half.
+   */
+  public static byte [] add(final byte [] a, final byte [] b) {
+    return add(a, b, HConstants.EMPTY_BYTE_ARRAY);
+  }
+
+  /**
+   * @param a
+   * @param b
+   * @param c
+   * @return New array made from a, b and c
+   */
+  public static byte [] add(final byte [] a, final byte [] b, final byte [] c) {
+    byte [] result = new byte[a.length + b.length + c.length];
+    System.arraycopy(a, 0, result, 0, a.length);
+    System.arraycopy(b, 0, result, a.length, b.length);
+    System.arraycopy(c, 0, result, a.length + b.length, c.length);
+    return result;
+  }
+  
+  
+  /**
+   * @param t
+   * @return Array of byte arrays made from passed array of Text
+   */
+  public static byte [][] toByteArrays(final Text [] t) {
+    byte [][] result = new byte[t.length][];
+    for (int i = 0; i < t.length; i++) {
+      result[i] = t[i].getBytes();
+    }
+    return result;
+  }
+  
+  /**
+   * @param column
+   * @return A byte array of a byte array where first and only entry is
+   * <code>column</code>
+   */
+  public static byte [][] toByteArrays(final String column) {
+    return toByteArrays(toBytes(column));
+  }
+  
+  /**
+   * @param column
+   * @return A byte array of a byte array where first and only entry is
+   * <code>column</code>
+   */
+  public static byte [][] toByteArrays(final byte [] column) {
+    byte [][] result = new byte[1][];
+    result[0] = column;
+    return result;
+  }
+}
\ No newline at end of file



Mime
View raw message