hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From mi...@apache.org
Subject [12/52] [partial] hbase-site git commit: Published site at 1cfd22bf43c9b64afae35d9bf16f764d0da80cab.
Date Tue, 21 Mar 2017 16:36:08 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/22cff34f/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
----------------------------------------------------------------------
diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
index 9c09190..07c69da 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
@@ -145,7 +145,7 @@
 <span class="sourceLineNo">137</span><a name="line.137"></a>
 <span class="sourceLineNo">138</span>  static &lt;V extends Cell&gt; RecordWriter&lt;ImmutableBytesWritable, V&gt;<a name="line.138"></a>
 <span class="sourceLineNo">139</span>  createRecordWriter(final TaskAttemptContext context) throws IOException {<a name="line.139"></a>
-<span class="sourceLineNo">140</span>    return new HFileRecordWriter&lt;V&gt;(context, null);<a name="line.140"></a>
+<span class="sourceLineNo">140</span>    return new HFileRecordWriter&lt;&gt;(context, null);<a name="line.140"></a>
 <span class="sourceLineNo">141</span>  }<a name="line.141"></a>
 <span class="sourceLineNo">142</span><a name="line.142"></a>
 <span class="sourceLineNo">143</span>  protected static class HFileRecordWriter&lt;V extends Cell&gt;<a name="line.143"></a>
@@ -219,7 +219,7 @@
 <span class="sourceLineNo">211</span>        overriddenEncoding = null;<a name="line.211"></a>
 <span class="sourceLineNo">212</span>      }<a name="line.212"></a>
 <span class="sourceLineNo">213</span><a name="line.213"></a>
-<span class="sourceLineNo">214</span>      writers = new TreeMap&lt;byte[], WriterLength&gt;(Bytes.BYTES_COMPARATOR);<a name="line.214"></a>
+<span class="sourceLineNo">214</span>      writers = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.214"></a>
 <span class="sourceLineNo">215</span>      previousRow = HConstants.EMPTY_BYTE_ARRAY;<a name="line.215"></a>
 <span class="sourceLineNo">216</span>      now = Bytes.toBytes(EnvironmentEdgeManager.currentTime());<a name="line.216"></a>
 <span class="sourceLineNo">217</span>      rollRequested = false;<a name="line.217"></a>
@@ -426,435 +426,429 @@
 <span class="sourceLineNo">418</span>  private static List&lt;ImmutableBytesWritable&gt; getRegionStartKeys(RegionLocator table)<a name="line.418"></a>
 <span class="sourceLineNo">419</span>  throws IOException {<a name="line.419"></a>
 <span class="sourceLineNo">420</span>    byte[][] byteKeys = table.getStartKeys();<a name="line.420"></a>
-<span class="sourceLineNo">421</span>    ArrayList&lt;ImmutableBytesWritable&gt; ret =<a name="line.421"></a>
-<span class="sourceLineNo">422</span>      new ArrayList&lt;ImmutableBytesWritable&gt;(byteKeys.length);<a name="line.422"></a>
-<span class="sourceLineNo">423</span>    for (byte[] byteKey : byteKeys) {<a name="line.423"></a>
-<span class="sourceLineNo">424</span>      ret.add(new ImmutableBytesWritable(byteKey));<a name="line.424"></a>
-<span class="sourceLineNo">425</span>    }<a name="line.425"></a>
-<span class="sourceLineNo">426</span>    return ret;<a name="line.426"></a>
-<span class="sourceLineNo">427</span>  }<a name="line.427"></a>
-<span class="sourceLineNo">428</span><a name="line.428"></a>
-<span class="sourceLineNo">429</span>  /**<a name="line.429"></a>
-<span class="sourceLineNo">430</span>   * Write out a {@link SequenceFile} that can be read by<a name="line.430"></a>
-<span class="sourceLineNo">431</span>   * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.431"></a>
-<span class="sourceLineNo">432</span>   */<a name="line.432"></a>
-<span class="sourceLineNo">433</span>  @SuppressWarnings("deprecation")<a name="line.433"></a>
-<span class="sourceLineNo">434</span>  private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.434"></a>
-<span class="sourceLineNo">435</span>      List&lt;ImmutableBytesWritable&gt; startKeys) throws IOException {<a name="line.435"></a>
-<span class="sourceLineNo">436</span>    LOG.info("Writing partition information to " + partitionsPath);<a name="line.436"></a>
-<span class="sourceLineNo">437</span>    if (startKeys.isEmpty()) {<a name="line.437"></a>
-<span class="sourceLineNo">438</span>      throw new IllegalArgumentException("No regions passed");<a name="line.438"></a>
-<span class="sourceLineNo">439</span>    }<a name="line.439"></a>
-<span class="sourceLineNo">440</span><a name="line.440"></a>
-<span class="sourceLineNo">441</span>    // We're generating a list of split points, and we don't ever<a name="line.441"></a>
-<span class="sourceLineNo">442</span>    // have keys &lt; the first region (which has an empty start key)<a name="line.442"></a>
-<span class="sourceLineNo">443</span>    // so we need to remove it. Otherwise we would end up with an<a name="line.443"></a>
-<span class="sourceLineNo">444</span>    // empty reducer with index 0<a name="line.444"></a>
-<span class="sourceLineNo">445</span>    TreeSet&lt;ImmutableBytesWritable&gt; sorted =<a name="line.445"></a>
-<span class="sourceLineNo">446</span>      new TreeSet&lt;ImmutableBytesWritable&gt;(startKeys);<a name="line.446"></a>
-<span class="sourceLineNo">447</span><a name="line.447"></a>
-<span class="sourceLineNo">448</span>    ImmutableBytesWritable first = sorted.first();<a name="line.448"></a>
-<span class="sourceLineNo">449</span>    if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.449"></a>
-<span class="sourceLineNo">450</span>      throw new IllegalArgumentException(<a name="line.450"></a>
-<span class="sourceLineNo">451</span>          "First region of table should have empty start key. Instead has: "<a name="line.451"></a>
-<span class="sourceLineNo">452</span>          + Bytes.toStringBinary(first.get()));<a name="line.452"></a>
-<span class="sourceLineNo">453</span>    }<a name="line.453"></a>
-<span class="sourceLineNo">454</span>    sorted.remove(first);<a name="line.454"></a>
-<span class="sourceLineNo">455</span><a name="line.455"></a>
-<span class="sourceLineNo">456</span>    // Write the actual file<a name="line.456"></a>
-<span class="sourceLineNo">457</span>    FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.457"></a>
-<span class="sourceLineNo">458</span>    SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.458"></a>
-<span class="sourceLineNo">459</span>      fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.459"></a>
-<span class="sourceLineNo">460</span>      NullWritable.class);<a name="line.460"></a>
-<span class="sourceLineNo">461</span><a name="line.461"></a>
-<span class="sourceLineNo">462</span>    try {<a name="line.462"></a>
-<span class="sourceLineNo">463</span>      for (ImmutableBytesWritable startKey : sorted) {<a name="line.463"></a>
-<span class="sourceLineNo">464</span>        writer.append(startKey, NullWritable.get());<a name="line.464"></a>
-<span class="sourceLineNo">465</span>      }<a name="line.465"></a>
-<span class="sourceLineNo">466</span>    } finally {<a name="line.466"></a>
-<span class="sourceLineNo">467</span>      writer.close();<a name="line.467"></a>
-<span class="sourceLineNo">468</span>    }<a name="line.468"></a>
-<span class="sourceLineNo">469</span>  }<a name="line.469"></a>
-<span class="sourceLineNo">470</span><a name="line.470"></a>
-<span class="sourceLineNo">471</span>  /**<a name="line.471"></a>
-<span class="sourceLineNo">472</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.472"></a>
-<span class="sourceLineNo">473</span>   * table. This<a name="line.473"></a>
-<span class="sourceLineNo">474</span>   * &lt;ul&gt;<a name="line.474"></a>
-<span class="sourceLineNo">475</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.475"></a>
-<span class="sourceLineNo">476</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.476"></a>
-<span class="sourceLineNo">477</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.477"></a>
-<span class="sourceLineNo">478</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.478"></a>
-<span class="sourceLineNo">479</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.479"></a>
-<span class="sourceLineNo">480</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.480"></a>
-<span class="sourceLineNo">481</span>   * &lt;/ul&gt;<a name="line.481"></a>
-<span class="sourceLineNo">482</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.482"></a>
-<span class="sourceLineNo">483</span>   * running this function.<a name="line.483"></a>
-<span class="sourceLineNo">484</span>   */<a name="line.484"></a>
-<span class="sourceLineNo">485</span>  public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.485"></a>
-<span class="sourceLineNo">486</span>      throws IOException {<a name="line.486"></a>
-<span class="sourceLineNo">487</span>    configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.487"></a>
-<span class="sourceLineNo">488</span>  }<a name="line.488"></a>
-<span class="sourceLineNo">489</span><a name="line.489"></a>
-<span class="sourceLineNo">490</span>  /**<a name="line.490"></a>
-<span class="sourceLineNo">491</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.491"></a>
-<span class="sourceLineNo">492</span>   * table. This<a name="line.492"></a>
-<span class="sourceLineNo">493</span>   * &lt;ul&gt;<a name="line.493"></a>
-<span class="sourceLineNo">494</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.494"></a>
-<span class="sourceLineNo">495</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.495"></a>
-<span class="sourceLineNo">496</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.496"></a>
-<span class="sourceLineNo">497</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.497"></a>
-<span class="sourceLineNo">498</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.498"></a>
-<span class="sourceLineNo">499</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.499"></a>
-<span class="sourceLineNo">500</span>   * &lt;/ul&gt;<a name="line.500"></a>
-<span class="sourceLineNo">501</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.501"></a>
-<span class="sourceLineNo">502</span>   * running this function.<a name="line.502"></a>
-<span class="sourceLineNo">503</span>   */<a name="line.503"></a>
-<span class="sourceLineNo">504</span>  public static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.504"></a>
-<span class="sourceLineNo">505</span>      RegionLocator regionLocator) throws IOException {<a name="line.505"></a>
-<span class="sourceLineNo">506</span>    configureIncrementalLoad(job, tableDescriptor, regionLocator, HFileOutputFormat2.class);<a name="line.506"></a>
-<span class="sourceLineNo">507</span>  }<a name="line.507"></a>
-<span class="sourceLineNo">508</span><a name="line.508"></a>
-<span class="sourceLineNo">509</span>  static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.509"></a>
-<span class="sourceLineNo">510</span>      RegionLocator regionLocator, Class&lt;? extends OutputFormat&lt;?, ?&gt;&gt; cls) throws IOException,<a name="line.510"></a>
-<span class="sourceLineNo">511</span>      UnsupportedEncodingException {<a name="line.511"></a>
-<span class="sourceLineNo">512</span>    Configuration conf = job.getConfiguration();<a name="line.512"></a>
-<span class="sourceLineNo">513</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.513"></a>
-<span class="sourceLineNo">514</span>    job.setOutputValueClass(KeyValue.class);<a name="line.514"></a>
-<span class="sourceLineNo">515</span>    job.setOutputFormatClass(cls);<a name="line.515"></a>
-<span class="sourceLineNo">516</span><a name="line.516"></a>
-<span class="sourceLineNo">517</span>    // Based on the configured map output class, set the correct reducer to properly<a name="line.517"></a>
-<span class="sourceLineNo">518</span>    // sort the incoming values.<a name="line.518"></a>
-<span class="sourceLineNo">519</span>    // TODO it would be nice to pick one or the other of these formats.<a name="line.519"></a>
-<span class="sourceLineNo">520</span>    if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.520"></a>
-<span class="sourceLineNo">521</span>      job.setReducerClass(KeyValueSortReducer.class);<a name="line.521"></a>
-<span class="sourceLineNo">522</span>    } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span>      job.setReducerClass(PutSortReducer.class);<a name="line.523"></a>
-<span class="sourceLineNo">524</span>    } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.524"></a>
-<span class="sourceLineNo">525</span>      job.setReducerClass(TextSortReducer.class);<a name="line.525"></a>
-<span class="sourceLineNo">526</span>    } else {<a name="line.526"></a>
-<span class="sourceLineNo">527</span>      LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.527"></a>
-<span class="sourceLineNo">528</span>    }<a name="line.528"></a>
-<span class="sourceLineNo">529</span><a name="line.529"></a>
-<span class="sourceLineNo">530</span>    conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.530"></a>
-<span class="sourceLineNo">531</span>        MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.531"></a>
-<span class="sourceLineNo">532</span>        KeyValueSerialization.class.getName());<a name="line.532"></a>
-<span class="sourceLineNo">533</span><a name="line.533"></a>
-<span class="sourceLineNo">534</span>    if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.534"></a>
-<span class="sourceLineNo">535</span>      // record this table name for creating writer by favored nodes<a name="line.535"></a>
-<span class="sourceLineNo">536</span>      LOG.info("bulkload locality sensitive enabled");<a name="line.536"></a>
-<span class="sourceLineNo">537</span>      conf.set(OUTPUT_TABLE_NAME_CONF_KEY, regionLocator.getName().getNameAsString());<a name="line.537"></a>
-<span class="sourceLineNo">538</span>    }<a name="line.538"></a>
-<span class="sourceLineNo">539</span><a name="line.539"></a>
-<span class="sourceLineNo">540</span>    // Use table's region boundaries for TOP split points.<a name="line.540"></a>
-<span class="sourceLineNo">541</span>    LOG.info("Looking up current regions for table " + regionLocator.getName());<a name="line.541"></a>
-<span class="sourceLineNo">542</span>    List&lt;ImmutableBytesWritable&gt; startKeys = getRegionStartKeys(regionLocator);<a name="line.542"></a>
-<span class="sourceLineNo">543</span>    LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.543"></a>
-<span class="sourceLineNo">544</span>        "to match current region count");<a name="line.544"></a>
-<span class="sourceLineNo">545</span>    job.setNumReduceTasks(startKeys.size());<a name="line.545"></a>
-<span class="sourceLineNo">546</span><a name="line.546"></a>
-<span class="sourceLineNo">547</span>    configurePartitioner(job, startKeys);<a name="line.547"></a>
-<span class="sourceLineNo">548</span>    // Set compression algorithms based on column families<a name="line.548"></a>
-<span class="sourceLineNo">549</span>    configureCompression(conf, tableDescriptor);<a name="line.549"></a>
-<span class="sourceLineNo">550</span>    configureBloomType(tableDescriptor, conf);<a name="line.550"></a>
-<span class="sourceLineNo">551</span>    configureBlockSize(tableDescriptor, conf);<a name="line.551"></a>
-<span class="sourceLineNo">552</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.552"></a>
-<span class="sourceLineNo">553</span><a name="line.553"></a>
-<span class="sourceLineNo">554</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.554"></a>
-<span class="sourceLineNo">555</span>    TableMapReduceUtil.initCredentials(job);<a name="line.555"></a>
-<span class="sourceLineNo">556</span>    LOG.info("Incremental table " + regionLocator.getName() + " output configured.");<a name="line.556"></a>
-<span class="sourceLineNo">557</span>  }<a name="line.557"></a>
-<span class="sourceLineNo">558</span><a name="line.558"></a>
-<span class="sourceLineNo">559</span>  public static void configureIncrementalLoadMap(Job job, HTableDescriptor tableDescriptor) throws<a name="line.559"></a>
-<span class="sourceLineNo">560</span>      IOException {<a name="line.560"></a>
-<span class="sourceLineNo">561</span>    Configuration conf = job.getConfiguration();<a name="line.561"></a>
-<span class="sourceLineNo">562</span><a name="line.562"></a>
-<span class="sourceLineNo">563</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.563"></a>
-<span class="sourceLineNo">564</span>    job.setOutputValueClass(KeyValue.class);<a name="line.564"></a>
-<span class="sourceLineNo">565</span>    job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.565"></a>
-<span class="sourceLineNo">566</span><a name="line.566"></a>
-<span class="sourceLineNo">567</span>    // Set compression algorithms based on column families<a name="line.567"></a>
-<span class="sourceLineNo">568</span>    configureCompression(conf, tableDescriptor);<a name="line.568"></a>
-<span class="sourceLineNo">569</span>    configureBloomType(tableDescriptor, conf);<a name="line.569"></a>
-<span class="sourceLineNo">570</span>    configureBlockSize(tableDescriptor, conf);<a name="line.570"></a>
-<span class="sourceLineNo">571</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.571"></a>
-<span class="sourceLineNo">572</span><a name="line.572"></a>
-<span class="sourceLineNo">573</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.573"></a>
-<span class="sourceLineNo">574</span>    TableMapReduceUtil.initCredentials(job);<a name="line.574"></a>
-<span class="sourceLineNo">575</span>    LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.575"></a>
-<span class="sourceLineNo">576</span>  }<a name="line.576"></a>
-<span class="sourceLineNo">577</span><a name="line.577"></a>
-<span class="sourceLineNo">578</span>  /**<a name="line.578"></a>
-<span class="sourceLineNo">579</span>   * Runs inside the task to deserialize column family to compression algorithm<a name="line.579"></a>
-<span class="sourceLineNo">580</span>   * map from the configuration.<a name="line.580"></a>
-<span class="sourceLineNo">581</span>   *<a name="line.581"></a>
-<span class="sourceLineNo">582</span>   * @param conf to read the serialized values from<a name="line.582"></a>
-<span class="sourceLineNo">583</span>   * @return a map from column family to the configured compression algorithm<a name="line.583"></a>
-<span class="sourceLineNo">584</span>   */<a name="line.584"></a>
-<span class="sourceLineNo">585</span>  @VisibleForTesting<a name="line.585"></a>
-<span class="sourceLineNo">586</span>  static Map&lt;byte[], Algorithm&gt; createFamilyCompressionMap(Configuration<a name="line.586"></a>
-<span class="sourceLineNo">587</span>      conf) {<a name="line.587"></a>
-<span class="sourceLineNo">588</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.588"></a>
-<span class="sourceLineNo">589</span>        COMPRESSION_FAMILIES_CONF_KEY);<a name="line.589"></a>
-<span class="sourceLineNo">590</span>    Map&lt;byte[], Algorithm&gt; compressionMap = new TreeMap&lt;byte[],<a name="line.590"></a>
-<span class="sourceLineNo">591</span>        Algorithm&gt;(Bytes.BYTES_COMPARATOR);<a name="line.591"></a>
-<span class="sourceLineNo">592</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.592"></a>
-<span class="sourceLineNo">593</span>      Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.593"></a>
-<span class="sourceLineNo">594</span>      compressionMap.put(e.getKey(), algorithm);<a name="line.594"></a>
-<span class="sourceLineNo">595</span>    }<a name="line.595"></a>
-<span class="sourceLineNo">596</span>    return compressionMap;<a name="line.596"></a>
-<span class="sourceLineNo">597</span>  }<a name="line.597"></a>
-<span class="sourceLineNo">598</span><a name="line.598"></a>
-<span class="sourceLineNo">599</span>  /**<a name="line.599"></a>
-<span class="sourceLineNo">600</span>   * Runs inside the task to deserialize column family to bloom filter type<a name="line.600"></a>
-<span class="sourceLineNo">601</span>   * map from the configuration.<a name="line.601"></a>
-<span class="sourceLineNo">602</span>   *<a name="line.602"></a>
-<span class="sourceLineNo">603</span>   * @param conf to read the serialized values from<a name="line.603"></a>
-<span class="sourceLineNo">604</span>   * @return a map from column family to the the configured bloom filter type<a name="line.604"></a>
-<span class="sourceLineNo">605</span>   */<a name="line.605"></a>
-<span class="sourceLineNo">606</span>  @VisibleForTesting<a name="line.606"></a>
-<span class="sourceLineNo">607</span>  static Map&lt;byte[], BloomType&gt; createFamilyBloomTypeMap(Configuration conf) {<a name="line.607"></a>
-<span class="sourceLineNo">608</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.608"></a>
-<span class="sourceLineNo">609</span>        BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.609"></a>
-<span class="sourceLineNo">610</span>    Map&lt;byte[], BloomType&gt; bloomTypeMap = new TreeMap&lt;byte[],<a name="line.610"></a>
-<span class="sourceLineNo">611</span>        BloomType&gt;(Bytes.BYTES_COMPARATOR);<a name="line.611"></a>
-<span class="sourceLineNo">612</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.612"></a>
-<span class="sourceLineNo">613</span>      BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.613"></a>
-<span class="sourceLineNo">614</span>      bloomTypeMap.put(e.getKey(), bloomType);<a name="line.614"></a>
-<span class="sourceLineNo">615</span>    }<a name="line.615"></a>
-<span class="sourceLineNo">616</span>    return bloomTypeMap;<a name="line.616"></a>
-<span class="sourceLineNo">617</span>  }<a name="line.617"></a>
-<span class="sourceLineNo">618</span><a name="line.618"></a>
-<span class="sourceLineNo">619</span>  /**<a name="line.619"></a>
-<span class="sourceLineNo">620</span>   * Runs inside the task to deserialize column family to block size<a name="line.620"></a>
-<span class="sourceLineNo">621</span>   * map from the configuration.<a name="line.621"></a>
-<span class="sourceLineNo">622</span>   *<a name="line.622"></a>
-<span class="sourceLineNo">623</span>   * @param conf to read the serialized values from<a name="line.623"></a>
-<span class="sourceLineNo">624</span>   * @return a map from column family to the configured block size<a name="line.624"></a>
-<span class="sourceLineNo">625</span>   */<a name="line.625"></a>
-<span class="sourceLineNo">626</span>  @VisibleForTesting<a name="line.626"></a>
-<span class="sourceLineNo">627</span>  static Map&lt;byte[], Integer&gt; createFamilyBlockSizeMap(Configuration conf) {<a name="line.627"></a>
-<span class="sourceLineNo">628</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.628"></a>
-<span class="sourceLineNo">629</span>        BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.629"></a>
-<span class="sourceLineNo">630</span>    Map&lt;byte[], Integer&gt; blockSizeMap = new TreeMap&lt;byte[],<a name="line.630"></a>
-<span class="sourceLineNo">631</span>        Integer&gt;(Bytes.BYTES_COMPARATOR);<a name="line.631"></a>
-<span class="sourceLineNo">632</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.632"></a>
-<span class="sourceLineNo">633</span>      Integer blockSize = Integer.parseInt(e.getValue());<a name="line.633"></a>
-<span class="sourceLineNo">634</span>      blockSizeMap.put(e.getKey(), blockSize);<a name="line.634"></a>
-<span class="sourceLineNo">635</span>    }<a name="line.635"></a>
-<span class="sourceLineNo">636</span>    return blockSizeMap;<a name="line.636"></a>
-<span class="sourceLineNo">637</span>  }<a name="line.637"></a>
-<span class="sourceLineNo">638</span><a name="line.638"></a>
-<span class="sourceLineNo">639</span>  /**<a name="line.639"></a>
-<span class="sourceLineNo">640</span>   * Runs inside the task to deserialize column family to data block encoding<a name="line.640"></a>
-<span class="sourceLineNo">641</span>   * type map from the configuration.<a name="line.641"></a>
-<span class="sourceLineNo">642</span>   *<a name="line.642"></a>
-<span class="sourceLineNo">643</span>   * @param conf to read the serialized values from<a name="line.643"></a>
-<span class="sourceLineNo">644</span>   * @return a map from column family to HFileDataBlockEncoder for the<a name="line.644"></a>
-<span class="sourceLineNo">645</span>   *         configured data block type for the family<a name="line.645"></a>
-<span class="sourceLineNo">646</span>   */<a name="line.646"></a>
-<span class="sourceLineNo">647</span>  @VisibleForTesting<a name="line.647"></a>
-<span class="sourceLineNo">648</span>  static Map&lt;byte[], DataBlockEncoding&gt; createFamilyDataBlockEncodingMap(<a name="line.648"></a>
-<span class="sourceLineNo">649</span>      Configuration conf) {<a name="line.649"></a>
-<span class="sourceLineNo">650</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.650"></a>
-<span class="sourceLineNo">651</span>        DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.651"></a>
-<span class="sourceLineNo">652</span>    Map&lt;byte[], DataBlockEncoding&gt; encoderMap = new TreeMap&lt;byte[],<a name="line.652"></a>
-<span class="sourceLineNo">653</span>        DataBlockEncoding&gt;(Bytes.BYTES_COMPARATOR);<a name="line.653"></a>
-<span class="sourceLineNo">654</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.654"></a>
-<span class="sourceLineNo">655</span>      encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.655"></a>
-<span class="sourceLineNo">656</span>    }<a name="line.656"></a>
-<span class="sourceLineNo">657</span>    return encoderMap;<a name="line.657"></a>
-<span class="sourceLineNo">658</span>  }<a name="line.658"></a>
-<span class="sourceLineNo">659</span><a name="line.659"></a>
-<span class="sourceLineNo">660</span><a name="line.660"></a>
-<span class="sourceLineNo">661</span>  /**<a name="line.661"></a>
-<span class="sourceLineNo">662</span>   * Run inside the task to deserialize column family to given conf value map.<a name="line.662"></a>
-<span class="sourceLineNo">663</span>   *<a name="line.663"></a>
-<span class="sourceLineNo">664</span>   * @param conf to read the serialized values from<a name="line.664"></a>
-<span class="sourceLineNo">665</span>   * @param confName conf key to read from the configuration<a name="line.665"></a>
-<span class="sourceLineNo">666</span>   * @return a map of column family to the given configuration value<a name="line.666"></a>
-<span class="sourceLineNo">667</span>   */<a name="line.667"></a>
-<span class="sourceLineNo">668</span>  private static Map&lt;byte[], String&gt; createFamilyConfValueMap(<a name="line.668"></a>
-<span class="sourceLineNo">669</span>      Configuration conf, String confName) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span>    Map&lt;byte[], String&gt; confValMap = new TreeMap&lt;byte[], String&gt;(Bytes.BYTES_COMPARATOR);<a name="line.670"></a>
-<span class="sourceLineNo">671</span>    String confVal = conf.get(confName, "");<a name="line.671"></a>
-<span class="sourceLineNo">672</span>    for (String familyConf : confVal.split("&amp;")) {<a name="line.672"></a>
-<span class="sourceLineNo">673</span>      String[] familySplit = familyConf.split("=");<a name="line.673"></a>
-<span class="sourceLineNo">674</span>      if (familySplit.length != 2) {<a name="line.674"></a>
-<span class="sourceLineNo">675</span>        continue;<a name="line.675"></a>
-<span class="sourceLineNo">676</span>      }<a name="line.676"></a>
-<span class="sourceLineNo">677</span>      try {<a name="line.677"></a>
-<span class="sourceLineNo">678</span>        confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(),<a name="line.678"></a>
-<span class="sourceLineNo">679</span>            URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.679"></a>
-<span class="sourceLineNo">680</span>      } catch (UnsupportedEncodingException e) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span>        // will not happen with UTF-8 encoding<a name="line.681"></a>
-<span class="sourceLineNo">682</span>        throw new AssertionError(e);<a name="line.682"></a>
-<span class="sourceLineNo">683</span>      }<a name="line.683"></a>
-<span class="sourceLineNo">684</span>    }<a name="line.684"></a>
-<span class="sourceLineNo">685</span>    return confValMap;<a name="line.685"></a>
-<span class="sourceLineNo">686</span>  }<a name="line.686"></a>
-<span class="sourceLineNo">687</span><a name="line.687"></a>
-<span class="sourceLineNo">688</span>  /**<a name="line.688"></a>
-<span class="sourceLineNo">689</span>   * Configure &lt;code&gt;job&lt;/code&gt; with a TotalOrderPartitioner, partitioning against<a name="line.689"></a>
-<span class="sourceLineNo">690</span>   * &lt;code&gt;splitPoints&lt;/code&gt;. Cleans up the partitions file after job exists.<a name="line.690"></a>
-<span class="sourceLineNo">691</span>   */<a name="line.691"></a>
-<span class="sourceLineNo">692</span>  static void configurePartitioner(Job job, List&lt;ImmutableBytesWritable&gt; splitPoints)<a name="line.692"></a>
-<span class="sourceLineNo">693</span>      throws IOException {<a name="line.693"></a>
-<span class="sourceLineNo">694</span>    Configuration conf = job.getConfiguration();<a name="line.694"></a>
-<span class="sourceLineNo">695</span>    // create the partitions file<a name="line.695"></a>
-<span class="sourceLineNo">696</span>    FileSystem fs = FileSystem.get(conf);<a name="line.696"></a>
-<span class="sourceLineNo">697</span>    String hbaseTmpFsDir =<a name="line.697"></a>
-<span class="sourceLineNo">698</span>        conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.698"></a>
-<span class="sourceLineNo">699</span>          HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.699"></a>
-<span class="sourceLineNo">700</span>    Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.700"></a>
-<span class="sourceLineNo">701</span>    fs.makeQualified(partitionsPath);<a name="line.701"></a>
-<span class="sourceLineNo">702</span>    writePartitions(conf, partitionsPath, splitPoints);<a name="line.702"></a>
-<span class="sourceLineNo">703</span>    fs.deleteOnExit(partitionsPath);<a name="line.703"></a>
-<span class="sourceLineNo">704</span><a name="line.704"></a>
-<span class="sourceLineNo">705</span>    // configure job to use it<a name="line.705"></a>
-<span class="sourceLineNo">706</span>    job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.706"></a>
-<span class="sourceLineNo">707</span>    TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.707"></a>
-<span class="sourceLineNo">708</span>  }<a name="line.708"></a>
-<span class="sourceLineNo">709</span><a name="line.709"></a>
-<span class="sourceLineNo">710</span>  /**<a name="line.710"></a>
-<span class="sourceLineNo">711</span>   * Serialize column family to compression algorithm map to configuration.<a name="line.711"></a>
-<span class="sourceLineNo">712</span>   * Invoked while configuring the MR job for incremental load.<a name="line.712"></a>
-<span class="sourceLineNo">713</span>   *<a name="line.713"></a>
-<span class="sourceLineNo">714</span>   * @param tableDescriptor to read the properties from<a name="line.714"></a>
-<span class="sourceLineNo">715</span>   * @param conf to persist serialized values into<a name="line.715"></a>
-<span class="sourceLineNo">716</span>   * @throws IOException<a name="line.716"></a>
-<span class="sourceLineNo">717</span>   *           on failure to read column family descriptors<a name="line.717"></a>
-<span class="sourceLineNo">718</span>   */<a name="line.718"></a>
-<span class="sourceLineNo">719</span>  @edu.umd.cs.findbugs.annotations.SuppressWarnings(<a name="line.719"></a>
-<span class="sourceLineNo">720</span>      value="RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.720"></a>
-<span class="sourceLineNo">721</span>  @VisibleForTesting<a name="line.721"></a>
-<span class="sourceLineNo">722</span>  static void configureCompression(Configuration conf, HTableDescriptor tableDescriptor)<a name="line.722"></a>
-<span class="sourceLineNo">723</span>      throws UnsupportedEncodingException {<a name="line.723"></a>
-<span class="sourceLineNo">724</span>    StringBuilder compressionConfigValue = new StringBuilder();<a name="line.724"></a>
-<span class="sourceLineNo">725</span>    if(tableDescriptor == null){<a name="line.725"></a>
-<span class="sourceLineNo">726</span>      // could happen with mock table instance<a name="line.726"></a>
-<span class="sourceLineNo">727</span>      return;<a name="line.727"></a>
-<span class="sourceLineNo">728</span>    }<a name="line.728"></a>
-<span class="sourceLineNo">729</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.729"></a>
-<span class="sourceLineNo">730</span>    int i = 0;<a name="line.730"></a>
-<span class="sourceLineNo">731</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.731"></a>
-<span class="sourceLineNo">732</span>      if (i++ &gt; 0) {<a name="line.732"></a>
-<span class="sourceLineNo">733</span>        compressionConfigValue.append('&amp;');<a name="line.733"></a>
-<span class="sourceLineNo">734</span>      }<a name="line.734"></a>
-<span class="sourceLineNo">735</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.735"></a>
-<span class="sourceLineNo">736</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.736"></a>
-<span class="sourceLineNo">737</span>      compressionConfigValue.append('=');<a name="line.737"></a>
-<span class="sourceLineNo">738</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.738"></a>
-<span class="sourceLineNo">739</span>        familyDescriptor.getCompressionType().getName(), "UTF-8"));<a name="line.739"></a>
-<span class="sourceLineNo">740</span>    }<a name="line.740"></a>
-<span class="sourceLineNo">741</span>    // Get rid of the last ampersand<a name="line.741"></a>
-<span class="sourceLineNo">742</span>    conf.set(COMPRESSION_FAMILIES_CONF_KEY, compressionConfigValue.toString());<a name="line.742"></a>
-<span class="sourceLineNo">743</span>  }<a name="line.743"></a>
-<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span>  /**<a name="line.745"></a>
-<span class="sourceLineNo">746</span>   * Serialize column family to block size map to configuration.<a name="line.746"></a>
-<span class="sourceLineNo">747</span>   * Invoked while configuring the MR job for incremental load.<a name="line.747"></a>
-<span class="sourceLineNo">748</span>   * @param tableDescriptor to read the properties from<a name="line.748"></a>
-<span class="sourceLineNo">749</span>   * @param conf to persist serialized values into<a name="line.749"></a>
-<span class="sourceLineNo">750</span>   *<a name="line.750"></a>
-<span class="sourceLineNo">751</span>   * @throws IOException<a name="line.751"></a>
-<span class="sourceLineNo">752</span>   *           on failure to read column family descriptors<a name="line.752"></a>
-<span class="sourceLineNo">753</span>   */<a name="line.753"></a>
-<span class="sourceLineNo">754</span>  @VisibleForTesting<a name="line.754"></a>
-<span class="sourceLineNo">755</span>  static void configureBlockSize(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.755"></a>
-<span class="sourceLineNo">756</span>      throws UnsupportedEncodingException {<a name="line.756"></a>
-<span class="sourceLineNo">757</span>    StringBuilder blockSizeConfigValue = new StringBuilder();<a name="line.757"></a>
-<span class="sourceLineNo">758</span>    if (tableDescriptor == null) {<a name="line.758"></a>
-<span class="sourceLineNo">759</span>      // could happen with mock table instance<a name="line.759"></a>
-<span class="sourceLineNo">760</span>      return;<a name="line.760"></a>
-<span class="sourceLineNo">761</span>    }<a name="line.761"></a>
-<span class="sourceLineNo">762</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.762"></a>
-<span class="sourceLineNo">763</span>    int i = 0;<a name="line.763"></a>
-<span class="sourceLineNo">764</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.764"></a>
-<span class="sourceLineNo">765</span>      if (i++ &gt; 0) {<a name="line.765"></a>
-<span class="sourceLineNo">766</span>        blockSizeConfigValue.append('&amp;');<a name="line.766"></a>
-<span class="sourceLineNo">767</span>      }<a name="line.767"></a>
-<span class="sourceLineNo">768</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.768"></a>
-<span class="sourceLineNo">769</span>          familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.769"></a>
-<span class="sourceLineNo">770</span>      blockSizeConfigValue.append('=');<a name="line.770"></a>
-<span class="sourceLineNo">771</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.771"></a>
-<span class="sourceLineNo">772</span>          String.valueOf(familyDescriptor.getBlocksize()), "UTF-8"));<a name="line.772"></a>
-<span class="sourceLineNo">773</span>    }<a name="line.773"></a>
-<span class="sourceLineNo">774</span>    // Get rid of the last ampersand<a name="line.774"></a>
-<span class="sourceLineNo">775</span>    conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, blockSizeConfigValue.toString());<a name="line.775"></a>
-<span class="sourceLineNo">776</span>  }<a name="line.776"></a>
-<span class="sourceLineNo">777</span><a name="line.777"></a>
-<span class="sourceLineNo">778</span>  /**<a name="line.778"></a>
-<span class="sourceLineNo">779</span>   * Serialize column family to bloom type map to configuration.<a name="line.779"></a>
-<span class="sourceLineNo">780</span>   * Invoked while configuring the MR job for incremental load.<a name="line.780"></a>
-<span class="sourceLineNo">781</span>   * @param tableDescriptor to read the properties from<a name="line.781"></a>
-<span class="sourceLineNo">782</span>   * @param conf to persist serialized values into<a name="line.782"></a>
-<span class="sourceLineNo">783</span>   *<a name="line.783"></a>
-<span class="sourceLineNo">784</span>   * @throws IOException<a name="line.784"></a>
-<span class="sourceLineNo">785</span>   *           on failure to read column family descriptors<a name="line.785"></a>
-<span class="sourceLineNo">786</span>   */<a name="line.786"></a>
-<span class="sourceLineNo">787</span>  @VisibleForTesting<a name="line.787"></a>
-<span class="sourceLineNo">788</span>  static void configureBloomType(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.788"></a>
-<span class="sourceLineNo">789</span>      throws UnsupportedEncodingException {<a name="line.789"></a>
-<span class="sourceLineNo">790</span>    if (tableDescriptor == null) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span>      // could happen with mock table instance<a name="line.791"></a>
-<span class="sourceLineNo">792</span>      return;<a name="line.792"></a>
-<span class="sourceLineNo">793</span>    }<a name="line.793"></a>
-<span class="sourceLineNo">794</span>    StringBuilder bloomTypeConfigValue = new StringBuilder();<a name="line.794"></a>
-<span class="sourceLineNo">795</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.795"></a>
-<span class="sourceLineNo">796</span>    int i = 0;<a name="line.796"></a>
-<span class="sourceLineNo">797</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span>      if (i++ &gt; 0) {<a name="line.798"></a>
-<span class="sourceLineNo">799</span>        bloomTypeConfigValue.append('&amp;');<a name="line.799"></a>
-<span class="sourceLineNo">800</span>      }<a name="line.800"></a>
-<span class="sourceLineNo">801</span>      bloomTypeConfigValue.append(URLEncoder.encode(<a name="line.801"></a>
-<span class="sourceLineNo">802</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.802"></a>
-<span class="sourceLineNo">803</span>      bloomTypeConfigValue.append('=');<a name="line.803"></a>
-<span class="sourceLineNo">804</span>      String bloomType = familyDescriptor.getBloomFilterType().toString();<a name="line.804"></a>
-<span class="sourceLineNo">805</span>      if (bloomType == null) {<a name="line.805"></a>
-<span class="sourceLineNo">806</span>        bloomType = HColumnDescriptor.DEFAULT_BLOOMFILTER;<a name="line.806"></a>
-<span class="sourceLineNo">807</span>      }<a name="line.807"></a>
-<span class="sourceLineNo">808</span>      bloomTypeConfigValue.append(URLEncoder.encode(bloomType, "UTF-8"));<a name="line.808"></a>
-<span class="sourceLineNo">809</span>    }<a name="line.809"></a>
-<span class="sourceLineNo">810</span>    conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, bloomTypeConfigValue.toString());<a name="line.810"></a>
-<span class="sourceLineNo">811</span>  }<a name="line.811"></a>
-<span class="sourceLineNo">812</span><a name="line.812"></a>
-<span class="sourceLineNo">813</span>  /**<a name="line.813"></a>
-<span class="sourceLineNo">814</span>   * Serialize column family to data block encoding map to configuration.<a name="line.814"></a>
-<span class="sourceLineNo">815</span>   * Invoked while configuring the MR job for incremental load.<a name="line.815"></a>
-<span class="sourceLineNo">816</span>   *<a name="line.816"></a>
-<span class="sourceLineNo">817</span>   * @param tableDescriptor to read the properties from<a name="line.817"></a>
-<span class="sourceLineNo">818</span>   * @param conf to persist serialized values into<a name="line.818"></a>
-<span class="sourceLineNo">819</span>   * @throws IOException<a name="line.819"></a>
-<span class="sourceLineNo">820</span>   *           on failure to read column family descriptors<a name="line.820"></a>
-<span class="sourceLineNo">821</span>   */<a name="line.821"></a>
-<span class="sourceLineNo">822</span>  @VisibleForTesting<a name="line.822"></a>
-<span class="sourceLineNo">823</span>  static void configureDataBlockEncoding(HTableDescriptor tableDescriptor,<a name="line.823"></a>
-<span class="sourceLineNo">824</span>      Configuration conf) throws UnsupportedEncodingException {<a name="line.824"></a>
-<span class="sourceLineNo">825</span>    if (tableDescriptor == null) {<a name="line.825"></a>
-<span class="sourceLineNo">826</span>      // could happen with mock table instance<a name="line.826"></a>
-<span class="sourceLineNo">827</span>      return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span>    }<a name="line.828"></a>
-<span class="sourceLineNo">829</span>    StringBuilder dataBlockEncodingConfigValue = new StringBuilder();<a name="line.829"></a>
-<span class="sourceLineNo">830</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.830"></a>
-<span class="sourceLineNo">831</span>    int i = 0;<a name="line.831"></a>
-<span class="sourceLineNo">832</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.832"></a>
-<span class="sourceLineNo">833</span>      if (i++ &gt; 0) {<a name="line.833"></a>
-<span class="sourceLineNo">834</span>        dataBlockEncodingConfigValue.append('&amp;');<a name="line.834"></a>
-<span class="sourceLineNo">835</span>      }<a name="line.835"></a>
-<span class="sourceLineNo">836</span>      dataBlockEncodingConfigValue.append(<a name="line.836"></a>
-<span class="sourceLineNo">837</span>          URLEncoder.encode(familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.837"></a>
-<span class="sourceLineNo">838</span>      dataBlockEncodingConfigValue.append('=');<a name="line.838"></a>
-<span class="sourceLineNo">839</span>      DataBlockEncoding encoding = familyDescriptor.getDataBlockEncoding();<a name="line.839"></a>
-<span class="sourceLineNo">840</span>      if (encoding == null) {<a name="line.840"></a>
-<span class="sourceLineNo">841</span>        encoding = DataBlockEncoding.NONE;<a name="line.841"></a>
-<span class="sourceLineNo">842</span>      }<a name="line.842"></a>
-<span class="sourceLineNo">843</span>      dataBlockEncodingConfigValue.append(URLEncoder.encode(encoding.toString(),<a name="line.843"></a>
-<span class="sourceLineNo">844</span>          "UTF-8"));<a name="line.844"></a>
-<span class="sourceLineNo">845</span>    }<a name="line.845"></a>
-<span class="sourceLineNo">846</span>    conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.846"></a>
-<span class="sourceLineNo">847</span>        dataBlockEncodingConfigValue.toString());<a name="line.847"></a>
-<span class="sourceLineNo">848</span>  }<a name="line.848"></a>
-<span class="sourceLineNo">849</span>}<a name="line.849"></a>
+<span class="sourceLineNo">421</span>    ArrayList&lt;ImmutableBytesWritable&gt; ret = new ArrayList&lt;&gt;(byteKeys.length);<a name="line.421"></a>
+<span class="sourceLineNo">422</span>    for (byte[] byteKey : byteKeys) {<a name="line.422"></a>
+<span class="sourceLineNo">423</span>      ret.add(new ImmutableBytesWritable(byteKey));<a name="line.423"></a>
+<span class="sourceLineNo">424</span>    }<a name="line.424"></a>
+<span class="sourceLineNo">425</span>    return ret;<a name="line.425"></a>
+<span class="sourceLineNo">426</span>  }<a name="line.426"></a>
+<span class="sourceLineNo">427</span><a name="line.427"></a>
+<span class="sourceLineNo">428</span>  /**<a name="line.428"></a>
+<span class="sourceLineNo">429</span>   * Write out a {@link SequenceFile} that can be read by<a name="line.429"></a>
+<span class="sourceLineNo">430</span>   * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.430"></a>
+<span class="sourceLineNo">431</span>   */<a name="line.431"></a>
+<span class="sourceLineNo">432</span>  @SuppressWarnings("deprecation")<a name="line.432"></a>
+<span class="sourceLineNo">433</span>  private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.433"></a>
+<span class="sourceLineNo">434</span>      List&lt;ImmutableBytesWritable&gt; startKeys) throws IOException {<a name="line.434"></a>
+<span class="sourceLineNo">435</span>    LOG.info("Writing partition information to " + partitionsPath);<a name="line.435"></a>
+<span class="sourceLineNo">436</span>    if (startKeys.isEmpty()) {<a name="line.436"></a>
+<span class="sourceLineNo">437</span>      throw new IllegalArgumentException("No regions passed");<a name="line.437"></a>
+<span class="sourceLineNo">438</span>    }<a name="line.438"></a>
+<span class="sourceLineNo">439</span><a name="line.439"></a>
+<span class="sourceLineNo">440</span>    // We're generating a list of split points, and we don't ever<a name="line.440"></a>
+<span class="sourceLineNo">441</span>    // have keys &lt; the first region (which has an empty start key)<a name="line.441"></a>
+<span class="sourceLineNo">442</span>    // so we need to remove it. Otherwise we would end up with an<a name="line.442"></a>
+<span class="sourceLineNo">443</span>    // empty reducer with index 0<a name="line.443"></a>
+<span class="sourceLineNo">444</span>    TreeSet&lt;ImmutableBytesWritable&gt; sorted = new TreeSet&lt;&gt;(startKeys);<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span>    ImmutableBytesWritable first = sorted.first();<a name="line.446"></a>
+<span class="sourceLineNo">447</span>    if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.447"></a>
+<span class="sourceLineNo">448</span>      throw new IllegalArgumentException(<a name="line.448"></a>
+<span class="sourceLineNo">449</span>          "First region of table should have empty start key. Instead has: "<a name="line.449"></a>
+<span class="sourceLineNo">450</span>          + Bytes.toStringBinary(first.get()));<a name="line.450"></a>
+<span class="sourceLineNo">451</span>    }<a name="line.451"></a>
+<span class="sourceLineNo">452</span>    sorted.remove(first);<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span>    // Write the actual file<a name="line.454"></a>
+<span class="sourceLineNo">455</span>    FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.455"></a>
+<span class="sourceLineNo">456</span>    SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.456"></a>
+<span class="sourceLineNo">457</span>      fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.457"></a>
+<span class="sourceLineNo">458</span>      NullWritable.class);<a name="line.458"></a>
+<span class="sourceLineNo">459</span><a name="line.459"></a>
+<span class="sourceLineNo">460</span>    try {<a name="line.460"></a>
+<span class="sourceLineNo">461</span>      for (ImmutableBytesWritable startKey : sorted) {<a name="line.461"></a>
+<span class="sourceLineNo">462</span>        writer.append(startKey, NullWritable.get());<a name="line.462"></a>
+<span class="sourceLineNo">463</span>      }<a name="line.463"></a>
+<span class="sourceLineNo">464</span>    } finally {<a name="line.464"></a>
+<span class="sourceLineNo">465</span>      writer.close();<a name="line.465"></a>
+<span class="sourceLineNo">466</span>    }<a name="line.466"></a>
+<span class="sourceLineNo">467</span>  }<a name="line.467"></a>
+<span class="sourceLineNo">468</span><a name="line.468"></a>
+<span class="sourceLineNo">469</span>  /**<a name="line.469"></a>
+<span class="sourceLineNo">470</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.470"></a>
+<span class="sourceLineNo">471</span>   * table. This<a name="line.471"></a>
+<span class="sourceLineNo">472</span>   * &lt;ul&gt;<a name="line.472"></a>
+<span class="sourceLineNo">473</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.473"></a>
+<span class="sourceLineNo">474</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.474"></a>
+<span class="sourceLineNo">475</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.475"></a>
+<span class="sourceLineNo">476</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.476"></a>
+<span class="sourceLineNo">477</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.477"></a>
+<span class="sourceLineNo">478</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.478"></a>
+<span class="sourceLineNo">479</span>   * &lt;/ul&gt;<a name="line.479"></a>
+<span class="sourceLineNo">480</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.480"></a>
+<span class="sourceLineNo">481</span>   * running this function.<a name="line.481"></a>
+<span class="sourceLineNo">482</span>   */<a name="line.482"></a>
+<span class="sourceLineNo">483</span>  public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.483"></a>
+<span class="sourceLineNo">484</span>      throws IOException {<a name="line.484"></a>
+<span class="sourceLineNo">485</span>    configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.485"></a>
+<span class="sourceLineNo">486</span>  }<a name="line.486"></a>
+<span class="sourceLineNo">487</span><a name="line.487"></a>
+<span class="sourceLineNo">488</span>  /**<a name="line.488"></a>
+<span class="sourceLineNo">489</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.489"></a>
+<span class="sourceLineNo">490</span>   * table. This<a name="line.490"></a>
+<span class="sourceLineNo">491</span>   * &lt;ul&gt;<a name="line.491"></a>
+<span class="sourceLineNo">492</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.492"></a>
+<span class="sourceLineNo">493</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.493"></a>
+<span class="sourceLineNo">494</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.494"></a>
+<span class="sourceLineNo">495</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.495"></a>
+<span class="sourceLineNo">496</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.496"></a>
+<span class="sourceLineNo">497</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.497"></a>
+<span class="sourceLineNo">498</span>   * &lt;/ul&gt;<a name="line.498"></a>
+<span class="sourceLineNo">499</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.499"></a>
+<span class="sourceLineNo">500</span>   * running this function.<a name="line.500"></a>
+<span class="sourceLineNo">501</span>   */<a name="line.501"></a>
+<span class="sourceLineNo">502</span>  public static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.502"></a>
+<span class="sourceLineNo">503</span>      RegionLocator regionLocator) throws IOException {<a name="line.503"></a>
+<span class="sourceLineNo">504</span>    configureIncrementalLoad(job, tableDescriptor, regionLocator, HFileOutputFormat2.class);<a name="line.504"></a>
+<span class="sourceLineNo">505</span>  }<a name="line.505"></a>
+<span class="sourceLineNo">506</span><a name="line.506"></a>
+<span class="sourceLineNo">507</span>  static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.507"></a>
+<span class="sourceLineNo">508</span>      RegionLocator regionLocator, Class&lt;? extends OutputFormat&lt;?, ?&gt;&gt; cls) throws IOException,<a name="line.508"></a>
+<span class="sourceLineNo">509</span>      UnsupportedEncodingException {<a name="line.509"></a>
+<span class="sourceLineNo">510</span>    Configuration conf = job.getConfiguration();<a name="line.510"></a>
+<span class="sourceLineNo">511</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.511"></a>
+<span class="sourceLineNo">512</span>    job.setOutputValueClass(KeyValue.class);<a name="line.512"></a>
+<span class="sourceLineNo">513</span>    job.setOutputFormatClass(cls);<a name="line.513"></a>
+<span class="sourceLineNo">514</span><a name="line.514"></a>
+<span class="sourceLineNo">515</span>    // Based on the configured map output class, set the correct reducer to properly<a name="line.515"></a>
+<span class="sourceLineNo">516</span>    // sort the incoming values.<a name="line.516"></a>
+<span class="sourceLineNo">517</span>    // TODO it would be nice to pick one or the other of these formats.<a name="line.517"></a>
+<span class="sourceLineNo">518</span>    if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.518"></a>
+<span class="sourceLineNo">519</span>      job.setReducerClass(KeyValueSortReducer.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span>    } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.520"></a>
+<span class="sourceLineNo">521</span>      job.setReducerClass(PutSortReducer.class);<a name="line.521"></a>
+<span class="sourceLineNo">522</span>    } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.522"></a>
+<span class="sourceLineNo">523</span>      job.setReducerClass(TextSortReducer.class);<a name="line.523"></a>
+<span class="sourceLineNo">524</span>    } else {<a name="line.524"></a>
+<span class="sourceLineNo">525</span>      LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.525"></a>
+<span class="sourceLineNo">526</span>    }<a name="line.526"></a>
+<span class="sourceLineNo">527</span><a name="line.527"></a>
+<span class="sourceLineNo">528</span>    conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.528"></a>
+<span class="sourceLineNo">529</span>        MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.529"></a>
+<span class="sourceLineNo">530</span>        KeyValueSerialization.class.getName());<a name="line.530"></a>
+<span class="sourceLineNo">531</span><a name="line.531"></a>
+<span class="sourceLineNo">532</span>    if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.532"></a>
+<span class="sourceLineNo">533</span>      // record this table name for creating writer by favored nodes<a name="line.533"></a>
+<span class="sourceLineNo">534</span>      LOG.info("bulkload locality sensitive enabled");<a name="line.534"></a>
+<span class="sourceLineNo">535</span>      conf.set(OUTPUT_TABLE_NAME_CONF_KEY, regionLocator.getName().getNameAsString());<a name="line.535"></a>
+<span class="sourceLineNo">536</span>    }<a name="line.536"></a>
+<span class="sourceLineNo">537</span><a name="line.537"></a>
+<span class="sourceLineNo">538</span>    // Use table's region boundaries for TOP split points.<a name="line.538"></a>
+<span class="sourceLineNo">539</span>    LOG.info("Looking up current regions for table " + regionLocator.getName());<a name="line.539"></a>
+<span class="sourceLineNo">540</span>    List&lt;ImmutableBytesWritable&gt; startKeys = getRegionStartKeys(regionLocator);<a name="line.540"></a>
+<span class="sourceLineNo">541</span>    LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.541"></a>
+<span class="sourceLineNo">542</span>        "to match current region count");<a name="line.542"></a>
+<span class="sourceLineNo">543</span>    job.setNumReduceTasks(startKeys.size());<a name="line.543"></a>
+<span class="sourceLineNo">544</span><a name="line.544"></a>
+<span class="sourceLineNo">545</span>    configurePartitioner(job, startKeys);<a name="line.545"></a>
+<span class="sourceLineNo">546</span>    // Set compression algorithms based on column families<a name="line.546"></a>
+<span class="sourceLineNo">547</span>    configureCompression(conf, tableDescriptor);<a name="line.547"></a>
+<span class="sourceLineNo">548</span>    configureBloomType(tableDescriptor, conf);<a name="line.548"></a>
+<span class="sourceLineNo">549</span>    configureBlockSize(tableDescriptor, conf);<a name="line.549"></a>
+<span class="sourceLineNo">550</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.550"></a>
+<span class="sourceLineNo">551</span><a name="line.551"></a>
+<span class="sourceLineNo">552</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.552"></a>
+<span class="sourceLineNo">553</span>    TableMapReduceUtil.initCredentials(job);<a name="line.553"></a>
+<span class="sourceLineNo">554</span>    LOG.info("Incremental table " + regionLocator.getName() + " output configured.");<a name="line.554"></a>
+<span class="sourceLineNo">555</span>  }<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span>  public static void configureIncrementalLoadMap(Job job, HTableDescriptor tableDescriptor) throws<a name="line.557"></a>
+<span class="sourceLineNo">558</span>      IOException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span>    Configuration conf = job.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span><a name="line.560"></a>
+<span class="sourceLineNo">561</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.561"></a>
+<span class="sourceLineNo">562</span>    job.setOutputValueClass(KeyValue.class);<a name="line.562"></a>
+<span class="sourceLineNo">563</span>    job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span>    // Set compression algorithms based on column families<a name="line.565"></a>
+<span class="sourceLineNo">566</span>    configureCompression(conf, tableDescriptor);<a name="line.566"></a>
+<span class="sourceLineNo">567</span>    configureBloomType(tableDescriptor, conf);<a name="line.567"></a>
+<span class="sourceLineNo">568</span>    configureBlockSize(tableDescriptor, conf);<a name="line.568"></a>
+<span class="sourceLineNo">569</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.569"></a>
+<span class="sourceLineNo">570</span><a name="line.570"></a>
+<span class="sourceLineNo">571</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.571"></a>
+<span class="sourceLineNo">572</span>    TableMapReduceUtil.initCredentials(job);<a name="line.572"></a>
+<span class="sourceLineNo">573</span>    LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.573"></a>
+<span class="sourceLineNo">574</span>  }<a name="line.574"></a>
+<span class="sourceLineNo">575</span><a name="line.575"></a>
+<span class="sourceLineNo">576</span>  /**<a name="line.576"></a>
+<span class="sourceLineNo">577</span>   * Runs inside the task to deserialize column family to compression algorithm<a name="line.577"></a>
+<span class="sourceLineNo">578</span>   * map from the configuration.<a name="line.578"></a>
+<span class="sourceLineNo">579</span>   *<a name="line.579"></a>
+<span class="sourceLineNo">580</span>   * @param conf to read the serialized values from<a name="line.580"></a>
+<span class="sourceLineNo">581</span>   * @return a map from column family to the configured compression algorithm<a name="line.581"></a>
+<span class="sourceLineNo">582</span>   */<a name="line.582"></a>
+<span class="sourceLineNo">583</span>  @VisibleForTesting<a name="line.583"></a>
+<span class="sourceLineNo">584</span>  static Map&lt;byte[], Algorithm&gt; createFamilyCompressionMap(Configuration<a name="line.584"></a>
+<span class="sourceLineNo">585</span>      conf) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.586"></a>
+<span class="sourceLineNo">587</span>        COMPRESSION_FAMILIES_CONF_KEY);<a name="line.587"></a>
+<span class="sourceLineNo">588</span>    Map&lt;byte[], Algorithm&gt; compressionMap = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.588"></a>
+<span class="sourceLineNo">589</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.589"></a>
+<span class="sourceLineNo">590</span>      Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.590"></a>
+<span class="sourceLineNo">591</span>      compressionMap.put(e.getKey(), algorithm);<a name="line.591"></a>
+<span class="sourceLineNo">592</span>    }<a name="line.592"></a>
+<span class="sourceLineNo">593</span>    return compressionMap;<a name="line.593"></a>
+<span class="sourceLineNo">594</span>  }<a name="line.594"></a>
+<span class="sourceLineNo">595</span><a name="line.595"></a>
+<span class="sourceLineNo">596</span>  /**<a name="line.596"></a>
+<span class="sourceLineNo">597</span>   * Runs inside the task to deserialize column family to bloom filter type<a name="line.597"></a>
+<span class="sourceLineNo">598</span>   * map from the configuration.<a name="line.598"></a>
+<span class="sourceLineNo">599</span>   *<a name="line.599"></a>
+<span class="sourceLineNo">600</span>   * @param conf to read the serialized values from<a name="line.600"></a>
+<span class="sourceLineNo">601</span>   * @return a map from column family to the the configured bloom filter type<a name="line.601"></a>
+<span class="sourceLineNo">602</span>   */<a name="line.602"></a>
+<span class="sourceLineNo">603</span>  @VisibleForTesting<a name="line.603"></a>
+<span class="sourceLineNo">604</span>  static Map&lt;byte[], BloomType&gt; createFamilyBloomTypeMap(Configuration conf) {<a name="line.604"></a>
+<span class="sourceLineNo">605</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.605"></a>
+<span class="sourceLineNo">606</span>        BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.606"></a>
+<span class="sourceLineNo">607</span>    Map&lt;byte[], BloomType&gt; bloomTypeMap = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.607"></a>
+<span class="sourceLineNo">608</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.608"></a>
+<span class="sourceLineNo">609</span>      BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.609"></a>
+<span class="sourceLineNo">610</span>      bloomTypeMap.put(e.getKey(), bloomType);<a name="line.610"></a>
+<span class="sourceLineNo">611</span>    }<a name="line.611"></a>
+<span class="sourceLineNo">612</span>    return bloomTypeMap;<a name="line.612"></a>
+<span class="sourceLineNo">613</span>  }<a name="line.613"></a>
+<span class="sourceLineNo">614</span><a name="line.614"></a>
+<span class="sourceLineNo">615</span>  /**<a name="line.615"></a>
+<span class="sourceLineNo">616</span>   * Runs inside the task to deserialize column family to block size<a name="line.616"></a>
+<span class="sourceLineNo">617</span>   * map from the configuration.<a name="line.617"></a>
+<span class="sourceLineNo">618</span>   *<a name="line.618"></a>
+<span class="sourceLineNo">619</span>   * @param conf to read the serialized values from<a name="line.619"></a>
+<span class="sourceLineNo">620</span>   * @return a map from column family to the configured block size<a name="line.620"></a>
+<span class="sourceLineNo">621</span>   */<a name="line.621"></a>
+<span class="sourceLineNo">622</span>  @VisibleForTesting<a name="line.622"></a>
+<span class="sourceLineNo">623</span>  static Map&lt;byte[], Integer&gt; createFamilyBlockSizeMap(Configuration conf) {<a name="line.623"></a>
+<span class="sourceLineNo">624</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.624"></a>
+<span class="sourceLineNo">625</span>        BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.625"></a>
+<span class="sourceLineNo">626</span>    Map&lt;byte[], Integer&gt; blockSizeMap = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.626"></a>
+<span class="sourceLineNo">627</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.627"></a>
+<span class="sourceLineNo">628</span>      Integer blockSize = Integer.parseInt(e.getValue());<a name="line.628"></a>
+<span class="sourceLineNo">629</span>      blockSizeMap.put(e.getKey(), blockSize);<a name="line.629"></a>
+<span class="sourceLineNo">630</span>    }<a name="line.630"></a>
+<span class="sourceLineNo">631</span>    return blockSizeMap;<a name="line.631"></a>
+<span class="sourceLineNo">632</span>  }<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span>  /**<a name="line.634"></a>
+<span class="sourceLineNo">635</span>   * Runs inside the task to deserialize column family to data block encoding<a name="line.635"></a>
+<span class="sourceLineNo">636</span>   * type map from the configuration.<a name="line.636"></a>
+<span class="sourceLineNo">637</span>   *<a name="line.637"></a>
+<span class="sourceLineNo">638</span>   * @param conf to read the serialized values from<a name="line.638"></a>
+<span class="sourceLineNo">639</span>   * @return a map from column family to HFileDataBlockEncoder for the<a name="line.639"></a>
+<span class="sourceLineNo">640</span>   *         configured data block type for the family<a name="line.640"></a>
+<span class="sourceLineNo">641</span>   */<a name="line.641"></a>
+<span class="sourceLineNo">642</span>  @VisibleForTesting<a name="line.642"></a>
+<span class="sourceLineNo">643</span>  static Map&lt;byte[], DataBlockEncoding&gt; createFamilyDataBlockEncodingMap(<a name="line.643"></a>
+<span class="sourceLineNo">644</span>      Configuration conf) {<a name="line.644"></a>
+<span class="sourceLineNo">645</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.645"></a>
+<span class="sourceLineNo">646</span>        DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.646"></a>
+<span class="sourceLineNo">647</span>    Map&lt;byte[], DataBlockEncoding&gt; encoderMap = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.647"></a>
+<span class="sourceLineNo">648</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.648"></a>
+<span class="sourceLineNo">649</span>      encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.649"></a>
+<span class="sourceLineNo">650</span>    }<a name="line.650"></a>
+<span class="sourceLineNo">651</span>    return encoderMap;<a name="line.651"></a>
+<span class="sourceLineNo">652</span>  }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span><a name="line.654"></a>
+<span class="sourceLineNo">655</span>  /**<a name="line.655"></a>
+<span class="sourceLineNo">656</span>   * Run inside the task to deserialize column family to given conf value map.<a name="line.656"></a>
+<span class="sourceLineNo">657</span>   *<a name="line.657"></a>
+<span class="sourceLineNo">658</span>   * @param conf to read the serialized values from<a name="line.658"></a>
+<span class="sourceLineNo">659</span>   * @param confName conf key to read from the configuration<a name="line.659"></a>
+<span class="sourceLineNo">660</span>   * @return a map of column family to the given configuration value<a name="line.660"></a>
+<span class="sourceLineNo">661</span>   */<a name="line.661"></a>
+<span class="sourceLineNo">662</span>  private static Map&lt;byte[], String&gt; createFamilyConfValueMap(<a name="line.662"></a>
+<span class="sourceLineNo">663</span>      Configuration conf, String confName) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span>    Map&lt;byte[], String&gt; confValMap = new TreeMap&lt;&gt;(Bytes.BYTES_COMPARATOR);<a name="line.664"></a>
+<span class="sourceLineNo">665</span>    String confVal = conf.get(confName, "");<a name="line.665"></a>
+<span class="sourceLineNo">666</span>    for (String familyConf : confVal.split("&amp;")) {<a name="line.666"></a>
+<span class="sourceLineNo">667</span>      String[] familySplit = familyConf.split("=");<a name="line.667"></a>
+<span class="sourceLineNo">668</span>      if (familySplit.length != 2) {<a name="line.668"></a>
+<span class="sourceLineNo">669</span>        continue;<a name="line.669"></a>
+<span class="sourceLineNo">670</span>      }<a name="line.670"></a>
+<span class="sourceLineNo">671</span>      try {<a name="line.671"></a>
+<span class="sourceLineNo">672</span>        confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(),<a name="line.672"></a>
+<span class="sourceLineNo">673</span>            URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.673"></a>
+<span class="sourceLineNo">674</span>      } catch (UnsupportedEncodingException e) {<a name="line.674"></a>
+<span class="sourceLineNo">675</span>        // will not happen with UTF-8 encoding<a name="line.675"></a>
+<span class="sourceLineNo">676</span>        throw new AssertionError(e);<a name="line.676"></a>
+<span class="sourceLineNo">677</span>      }<a name="line.677"></a>
+<span class="sourceLineNo">678</span>    }<a name="line.678"></a>
+<span class="sourceLineNo">679</span>    return confValMap;<a name="line.679"></a>
+<span class="sourceLineNo">680</span>  }<a name="line.680"></a>
+<span class="sourceLineNo">681</span><a name="line.681"></a>
+<span class="sourceLineNo">682</span>  /**<a name="line.682"></a>
+<span class="sourceLineNo">683</span>   * Configure &lt;code&gt;job&lt;/code&gt; with a TotalOrderPartitioner, partitioning against<a name="line.683"></a>
+<span class="sourceLineNo">684</span>   * &lt;code&gt;splitPoints&lt;/code&gt;. Cleans up the partitions file after job exists.<a name="line.684"></a>
+<span class="sourceLineNo">685</span>   */<a name="line.685"></a>
+<span class="sourceLineNo">686</span>  static void configurePartitioner(Job job, List&lt;ImmutableBytesWritable&gt; splitPoints)<a name="line.686"></a>
+<span class="sourceLineNo">687</span>      throws IOException {<a name="line.687"></a>
+<span class="sourceLineNo">688</span>    Configuration conf = job.getConfiguration();<a name="line.688"></a>
+<span class="sourceLineNo">689</span>    // create the partitions file<a name="line.689"></a>
+<span class="sourceLineNo">690</span>    FileSystem fs = FileSystem.get(conf);<a name="line.690"></a>
+<span class="sourceLineNo">691</span>    String hbaseTmpFsDir =<a name="line.691"></a>
+<span class="sourceLineNo">692</span>        conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.692"></a>
+<span class="sourceLineNo">693</span>          HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.693"></a>
+<span class="sourceLineNo">694</span>    Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.694"></a>
+<span class="sourceLineNo">695</span>    fs.makeQualified(partitionsPath);<a name="line.695"></a>
+<span class="sourceLineNo">696</span>    writePartitions(conf, partitionsPath, splitPoints);<a name="line.696"></a>
+<span class="sourceLineNo">697</span>    fs.deleteOnExit(partitionsPath);<a name="line.697"></a>
+<span class="sourceLineNo">698</span><a name="line.698"></a>
+<span class="sourceLineNo">699</span>    // configure job to use it<a name="line.699"></a>
+<span class="sourceLineNo">700</span>    job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.700"></a>
+<span class="sourceLineNo">701</span>    TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.701"></a>
+<span class="sourceLineNo">702</span>  }<a name="line.702"></a>
+<span class="sourceLineNo">703</span><a name="line.703"></a>
+<span class="sourceLineNo">704</span>  /**<a name="line.704"></a>
+<span class="sourceLineNo">705</span>   * Serialize column family to compression algorithm map to configuration.<a name="line.705"></a>
+<span class="sourceLineNo">706</span>   * Invoked while configuring the MR job for incremental load.<a name="line.706"></a>
+<span class="sourceLineNo">707</span>   *<a name="line.707"></a>
+<span class="sourceLineNo">708</span>   * @param tableDescriptor to read the properties from<a name="line.708"></a>
+<span class="sourceLineNo">709</span>   * @param conf to persist serialized values into<a name="line.709"></a>
+<span class="sourceLineNo">710</span>   * @throws IOException<a name="line.710"></a>
+<span class="sourceLineNo">711</span>   *           on failure to read column family descriptors<a name="line.711"></a>
+<span class="sourceLineNo">712</span>   */<a name="line.712"></a>
+<span class="sourceLineNo">713</span>  @edu.umd.cs.findbugs.annotations.SuppressWarnings(<a name="line.713"></a>
+<span class="sourceLineNo">714</span>      value="RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.714"></a>
+<span class="sourceLineNo">715</span>  @VisibleForTesting<a name="line.715"></a>
+<span class="sourceLineNo">716</span>  static void configureCompression(Configuration conf, HTableDescriptor tableDescriptor)<a name="line.716"></a>
+<span class="sourceLineNo">717</span>      throws UnsupportedEncodingException {<a name="line.717"></a>
+<span class="sourceLineNo">718</span>    StringBuilder compressionConfigValue = new StringBuilder();<a name="line.718"></a>
+<span class="sourceLineNo">719</span>    if(tableDescriptor == null){<a name="line.719"></a>
+<span class="sourceLineNo">720</span>      // could happen with mock table instance<a name="line.720"></a>
+<span class="sourceLineNo">721</span>      return;<a name="line.721"></a>
+<span class="sourceLineNo">722</span>    }<a name="line.722"></a>
+<span class="sourceLineNo">723</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.723"></a>
+<span class="sourceLineNo">724</span>    int i = 0;<a name="line.724"></a>
+<span class="sourceLineNo">725</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span>      if (i++ &gt; 0) {<a name="line.726"></a>
+<span class="sourceLineNo">727</span>        compressionConfigValue.append('&amp;');<a name="line.727"></a>
+<span class="sourceLineNo">728</span>      }<a name="line.728"></a>
+<span class="sourceLineNo">729</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.729"></a>
+<span class="sourceLineNo">730</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.730"></a>
+<span class="sourceLineNo">731</span>      compressionConfigValue.append('=');<a name="line.731"></a>
+<span class="sourceLineNo">732</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.732"></a>
+<span class="sourceLineNo">733</span>        familyDescriptor.getCompressionType().getName(), "UTF-8"));<a name="line.733"></a>
+<span class="sourceLineNo">734</span>    }<a name="line.734"></a>
+<span class="sourceLineNo">735</span>    // Get rid of the last ampersand<a name="line.735"></a>
+<span class="sourceLineNo">736</span>    conf.set(COMPRESSION_FAMILIES_CONF_KEY, compressionConfigValue.toString());<a name="line.736"></a>
+<span class="sourceLineNo">737</span>  }<a name="line.737"></a>
+<span class="sourceLineNo">738</span><a name="line.738"></a>
+<span class="sourceLineNo">739</span>  /**<a name="line.739"></a>
+<span class="sourceLineNo">740</span>   * Serialize column family to block size map to configuration.<a name="line.740"></a>
+<span class="sourceLineNo">741</span>   * Invoked while configuring the MR job for incremental load.<a name="line.741"></a>
+<span class="sourceLineNo">742</span>   * @param tableDescriptor to read the properties from<a name="line.742"></a>
+<span class="sourceLineNo">743</span>   * @param conf to persist serialized values into<a name="line.743"></a>
+<span class="sourceLineNo">744</span>   *<a name="line.744"></a>
+<span class="sourceLineNo">745</span>   * @throws IOException<a name="line.745"></a>
+<span class="sourceLineNo">746</span>   *           on failure to read column family descriptors<a name="line.746"></a>
+<span class="sourceLineNo">747</span>   */<a name="line.747"></a>
+<span class="sourceLineNo">748</span>  @VisibleForTesting<a name="line.748"></a>
+<span class="sourceLineNo">749</span>  static void configureBlockSize(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.749"></a>
+<span class="sourceLineNo">750</span>      throws UnsupportedEncodingException {<a name="line.750"></a>
+<span class="sourceLineNo">751</span>    StringBuilder blockSizeConfigValue = new StringBuilder();<a name="line.751"></a>
+<span class="sourceLineNo">752</span>    if (tableDescriptor == null) {<a name="line.752"></a>
+<span class="sourceLineNo">753</span>      // could happen with mock table instance<a name="line.753"></a>
+<span class="sourceLineNo">754</span>      return;<a name="line.754"></a>
+<span class="sourceLineNo">755</span>    }<a name="line.755"></a>
+<span class="sourceLineNo">756</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.756"></a>
+<span class="sourceLineNo">757</span>    int i = 0;<a name="line.757"></a>
+<span class="sourceLineNo">758</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.758"></a>
+<span class="sourceLineNo">759</span>      if (i++ &gt; 0) {<a name="line.759"></a>
+<span class="sourceLineNo">760</span>        blockSizeConfigValue.append('&amp;');<a name="line.760"></a>
+<span class="sourceLineNo">761</span>      }<a name="line.761"></a>
+<span class="sourceLineNo">762</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.762"></a>
+<span class="sourceLineNo">763</span>          familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.763"></a>
+<span class="sourceLineNo">764</span>      blockSizeConfigValue.append('=');<a name="line.764"></a>
+<span class="sourceLineNo">765</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.765"></a>
+<span class="sourceLineNo">766</span>          String.valueOf(familyDescriptor.getBlocksize()), "UTF-8"));<a name="line.766"></a>
+<span class="sourceLineNo">767</span>    }<a name="line.767"></a>
+<span class="sourceLineNo">768</span>    // Get rid of the last ampersand<a name="line.768"></a>
+<span class="sourceLineNo">769</span>    conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, blockSizeConfigValue.toString());<a name="line.769"></a>
+<span class="sourceLineNo">770</span>  }<a name="line.770"></a>
+<span class="sourceLineNo">771</span><a name="line.771"></a>
+<span class="sourceLineNo">772</span>  /**<a name="line.772"></a>
+<span class="sourceLineNo">773</span>   * Serialize column family to bloom type map to configuration.<a name="line.773"></a>
+<span class="sourceLineNo">774</span>   * Invoked while configuring the MR job for incremental load.<a name="line.774"></a>
+<span class="sourceLineNo">775</span>   * @param tableDescriptor to read the properties from<a name="line.775"></a>
+<span class="sourceLineNo">776</span>   * @param conf to persist serialized values into<a name="line.776"></a>
+<span class="sourceLineNo">777</span>   *<a name="line.777"></a>
+<span class="sourceLineNo">778</span>   * @throws IOException<a name="line.778"></a>
+<span class="sourceLineNo">779</span>   *           on failure to read column family descriptors<a name="line.779"></a>
+<span class="sourceLineNo">780</span>   */<a name="line.780"></a>
+<span class="sourceLineNo">781</span>  @VisibleForTesting<a name="line.781"></a>
+<span class="sourceLineNo">782</span>  static void configureBloomType(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.782"></a>
+<span class="sourceLineNo">783</span>      throws UnsupportedEncodingException {<a name="line.783"></a>
+<span class="sourceLineNo">784</span>    if (tableDescriptor == null) {<a name="line.784"></a>
+<span class="sourceLineNo">785</span>      // could happen with mock table instance<a name="line.785"></a>
+<span class="sourceLineNo">786</span>      return;<a name="line.786"></a>
+<span class="sourceLineNo">787</span>    }<a name="line.787"></a>
+<span class="sourceLineNo">788</span>    StringBuilder bloomTypeConfigValue = new StringBuilder();<a name="line.788"></a>
+<span class="sourceLineNo">789</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.789"></a>
+<span class="sourceLineNo">790</span>    int i = 0;<a name="line.790"></a>
+<span class="sourceLineNo">791</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span>      if (i++ &gt; 0) {<a name="line.792"></a>
+<span class="sourceLineNo">793</span>        bloomTypeConfigValue.append('&amp;');<a name="line.793"></a>
+<span class="sourceLineNo">794</span>      }<a name="line.794"></a>
+<span class="sourceLineNo">795</span>      bloomTypeConfigValue.append(URLEncoder.encode(<a name="line.795"></a>
+<span class="sourceLineNo">796</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.796"></a>
+<span class="sourceLineNo">797</span>      bloomTypeConfigValue.append('=');<a name="line.797"></a>
+<span class="sourceLineNo">798</span>      String bloomType = familyDescriptor.getBloomFilterType().toString();<a name="line.798"></a>
+<span class="sourceLineNo">799</span>      if (bloomType == null) {<a name="line.799"></a>
+<span class="sourceLineNo">800</span>        bloomType = HColumnDescriptor.DEFAULT_BLOOMFILTER;<a name="line.800"></a>
+<span class="sourceLineNo">801</span>      }<a name="line.801"></a>
+<span class="sourceLineNo">802</span>      bloomTypeConfigValue.append(URLEncoder.encode(bloomType, "UTF-8"));<a name="line.802"></a>
+<span class="sourceLineNo">803</span>    }<a name="line.803"></a>
+<span class="sourceLineNo">804</span>    conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, bloomTypeConfigValue.toString());<a name="line.804"></a>
+<span class="sourceLineNo">805</span>  }<a name="line.805"></a>
+<span class="sourceLineNo">806</span><a name="line.806"></a>
+<span class="sourceLineNo">807</span>  /**<a name="line.807"></a>
+<span class="sourceLineNo">808</span>   * Serialize column family to data block encoding map to configuration.<a name="line.808"></a>
+<span class="sourceLineNo">809</span>   * Invoked while configuring the MR job for incremental load.<a name="line.809"></a>
+<span class="sourceLineNo">810</span>   *<a name="line.810"></a>
+<span class="sourceLineNo">811</span>   * @param tableDescriptor to read the properties from<a name="line.811"></a>
+<span class="sourceLineNo">812</span>   * @param conf to persist serialized values into<a name="line.812"></a>
+<span class="sourceLineNo">813</span>   * @throws IOException<a name="line.813"></a>
+<span class="sourceLineNo">814</span>   *           on failure to read column family descriptors<a name="line.814"></a>
+<span class="sourceLineNo">815</span>   */<a name="line.815"></a>
+<span class="sourceLineNo">816</span>  @VisibleForTesting<a name="line.816"></a>
+<span class="sourceLineNo">817</span>  static void configureDataBlockEncoding(HTableDescriptor tableDescriptor,<a name="line.817"></a>
+<span class="sourceLineNo">818</span>      Configuration conf) throws UnsupportedEncodingException {<a name="line.818"></a>
+<span class="sourceLineNo">819</span>    if (tableDescriptor == null) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span>      // could happen with mock table instance<a name="line.820"></a>
+<span class="sourceLineNo">821</span>      return;<a name="line.821"></a>
+<span class="sourceLineNo">822</span>    }<a name="line.822"></a>
+<span class="sourceLineNo">823</span>    StringBuilder dataBlockEncodingConfigValue = new StringBuilder();<a name="line.823"></a>
+<span class="sourceLineNo">824</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.824"></a>
+<span class="sourceLineNo">825</span>    int i = 0;<a name="line.825"></a>
+<span class="sourceLineNo">826</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.826"></a>
+<span class="sourceLineNo">827</span>      if (i++ &gt; 0) {<a name="line.827"></a>
+<span class="sourceLineNo">828</span>        dataBlockEncodingConfigValue.append('&amp;');<a name="line.828"></a>
+<span class="sourceLineNo">829</span>      }<a name="line.829"></a>
+<span class="sourceLineNo">830</span>      dataBlockEncodingConfigValue.append(<a name="line.830"></a>
+<span class="sourceLineNo">831</span>          URLEncoder.encode(familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.831"></a>
+<span class="sourceLineNo">832</span>      dataBlockEncodingConfigValue.append('=');<a name="line.832"></a>
+<span class="sourceLineNo">833</span>      DataBlockEncoding encoding = familyDescriptor.getDataBlockEncoding();<a name="line.833"></a>
+<span class="sourceLineNo">834</span>      if (encoding == null) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span>        encoding = DataBlockEncoding.NONE;<a name="line.835"></a>
+<span class="sourceLineNo">836</span>      }<a nam

<TRUNCATED>

Mime
View raw message