You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by gi...@apache.org on 2018/08/10 14:48:45 UTC
[02/18] hbase-site git commit: Published site at
397388316ead020d005a33e233364d166d4add00.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/5c5e961f/testdevapidocs/src-html/org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.html
----------------------------------------------------------------------
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.html
index 9ddf036..6f6aa09 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.html
@@ -26,715 +26,737 @@
<span class="sourceLineNo">018</span><a name="line.18"></a>
<span class="sourceLineNo">019</span>import java.io.ByteArrayInputStream;<a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.ByteArrayOutputStream;<a name="line.20"></a>
-<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a>
-<span class="sourceLineNo">022</span>import java.io.InputStream;<a name="line.22"></a>
-<span class="sourceLineNo">023</span>import java.text.DecimalFormat;<a name="line.23"></a>
-<span class="sourceLineNo">024</span>import java.util.ArrayList;<a name="line.24"></a>
-<span class="sourceLineNo">025</span>import java.util.Iterator;<a name="line.25"></a>
-<span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a>
-<span class="sourceLineNo">027</span>import java.util.Locale;<a name="line.27"></a>
-<span class="sourceLineNo">028</span><a name="line.28"></a>
-<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a>
-<span class="sourceLineNo">030</span>import org.apache.hadoop.fs.FileSystem;<a name="line.30"></a>
-<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.Path;<a name="line.31"></a>
-<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.Cell;<a name="line.32"></a>
-<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.33"></a>
-<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.34"></a>
-<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.35"></a>
-<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.io.compress.Compression;<a name="line.36"></a>
-<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.37"></a>
-<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoder;<a name="line.38"></a>
-<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.39"></a>
-<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.io.encoding.EncodedDataBlock;<a name="line.40"></a>
-<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.41"></a>
-<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.io.hfile.HFileBlock;<a name="line.42"></a>
-<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.43"></a>
-<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.44"></a>
-<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.io.hfile.HFileReaderImpl;<a name="line.45"></a>
-<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.46"></a>
-<span class="sourceLineNo">047</span>import org.apache.hadoop.io.compress.CompressionOutputStream;<a name="line.47"></a>
-<span class="sourceLineNo">048</span>import org.apache.hadoop.io.compress.Compressor;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.io.compress.Decompressor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.slf4j.Logger;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.slf4j.LoggerFactory;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Options;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.ParseException;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.PosixParser;<a name="line.57"></a>
-<span class="sourceLineNo">058</span><a name="line.58"></a>
-<span class="sourceLineNo">059</span>/**<a name="line.59"></a>
-<span class="sourceLineNo">060</span> * Tests various algorithms for key compression on an existing HFile. Useful<a name="line.60"></a>
-<span class="sourceLineNo">061</span> * for testing, debugging and benchmarking.<a name="line.61"></a>
-<span class="sourceLineNo">062</span> */<a name="line.62"></a>
-<span class="sourceLineNo">063</span>public class DataBlockEncodingTool {<a name="line.63"></a>
-<span class="sourceLineNo">064</span> private static final Logger LOG = LoggerFactory.getLogger(<a name="line.64"></a>
-<span class="sourceLineNo">065</span> DataBlockEncodingTool.class);<a name="line.65"></a>
-<span class="sourceLineNo">066</span><a name="line.66"></a>
-<span class="sourceLineNo">067</span> private static final boolean includesMemstoreTS = true;<a name="line.67"></a>
+<span class="sourceLineNo">021</span>import java.io.DataOutputStream;<a name="line.21"></a>
+<span class="sourceLineNo">022</span>import java.io.IOException;<a name="line.22"></a>
+<span class="sourceLineNo">023</span>import java.io.InputStream;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import java.text.DecimalFormat;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import java.util.Iterator;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import java.util.List;<a name="line.27"></a>
+<span class="sourceLineNo">028</span>import java.util.Locale;<a name="line.28"></a>
+<span class="sourceLineNo">029</span><a name="line.29"></a>
+<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configuration;<a name="line.30"></a>
+<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.FileSystem;<a name="line.31"></a>
+<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.Path;<a name="line.32"></a>
+<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.Cell;<a name="line.33"></a>
+<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.34"></a>
+<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.35"></a>
+<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.36"></a>
+<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.io.compress.Compression;<a name="line.37"></a>
+<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.io.compress.Compression.Algorithm;<a name="line.38"></a>
+<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoder;<a name="line.39"></a>
+<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.40"></a>
+<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.io.encoding.EncodedDataBlock;<a name="line.41"></a>
+<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.42"></a>
+<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.io.hfile.HFileBlock;<a name="line.43"></a>
+<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.io.hfile.HFileContext;<a name="line.44"></a>
+<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.io.hfile.HFileContextBuilder;<a name="line.45"></a>
+<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.io.hfile.HFileReaderImpl;<a name="line.46"></a>
+<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.47"></a>
+<span class="sourceLineNo">048</span>import org.apache.hadoop.io.WritableUtils;<a name="line.48"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.io.compress.CompressionOutputStream;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.io.compress.Compressor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.io.compress.Decompressor;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.slf4j.Logger;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.slf4j.LoggerFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Options;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.ParseException;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.PosixParser;<a name="line.59"></a>
+<span class="sourceLineNo">060</span><a name="line.60"></a>
+<span class="sourceLineNo">061</span>/**<a name="line.61"></a>
+<span class="sourceLineNo">062</span> * Tests various algorithms for key compression on an existing HFile. Useful<a name="line.62"></a>
+<span class="sourceLineNo">063</span> * for testing, debugging and benchmarking.<a name="line.63"></a>
+<span class="sourceLineNo">064</span> */<a name="line.64"></a>
+<span class="sourceLineNo">065</span>public class DataBlockEncodingTool {<a name="line.65"></a>
+<span class="sourceLineNo">066</span> private static final Logger LOG = LoggerFactory.getLogger(<a name="line.66"></a>
+<span class="sourceLineNo">067</span> DataBlockEncodingTool.class);<a name="line.67"></a>
<span class="sourceLineNo">068</span><a name="line.68"></a>
-<span class="sourceLineNo">069</span> /**<a name="line.69"></a>
-<span class="sourceLineNo">070</span> * How many times to run the benchmark. More times means better data in terms<a name="line.70"></a>
-<span class="sourceLineNo">071</span> * of statistics but slower execution. Has to be strictly larger than<a name="line.71"></a>
-<span class="sourceLineNo">072</span> * {@link #DEFAULT_BENCHMARK_N_OMIT}.<a name="line.72"></a>
-<span class="sourceLineNo">073</span> */<a name="line.73"></a>
-<span class="sourceLineNo">074</span> private static final int DEFAULT_BENCHMARK_N_TIMES = 12;<a name="line.74"></a>
-<span class="sourceLineNo">075</span><a name="line.75"></a>
-<span class="sourceLineNo">076</span> /**<a name="line.76"></a>
-<span class="sourceLineNo">077</span> * How many first runs should not be included in the benchmark. Done in order<a name="line.77"></a>
-<span class="sourceLineNo">078</span> * to exclude setup cost.<a name="line.78"></a>
-<span class="sourceLineNo">079</span> */<a name="line.79"></a>
-<span class="sourceLineNo">080</span> private static final int DEFAULT_BENCHMARK_N_OMIT = 2;<a name="line.80"></a>
-<span class="sourceLineNo">081</span><a name="line.81"></a>
-<span class="sourceLineNo">082</span> /** HFile name to be used in benchmark */<a name="line.82"></a>
-<span class="sourceLineNo">083</span> private static final String OPT_HFILE_NAME = "f";<a name="line.83"></a>
-<span class="sourceLineNo">084</span><a name="line.84"></a>
-<span class="sourceLineNo">085</span> /** Maximum number of key/value pairs to process in a single benchmark run */<a name="line.85"></a>
-<span class="sourceLineNo">086</span> private static final String OPT_KV_LIMIT = "n";<a name="line.86"></a>
-<span class="sourceLineNo">087</span><a name="line.87"></a>
-<span class="sourceLineNo">088</span> /** Whether to run a benchmark to measure read throughput */<a name="line.88"></a>
-<span class="sourceLineNo">089</span> private static final String OPT_MEASURE_THROUGHPUT = "b";<a name="line.89"></a>
-<span class="sourceLineNo">090</span><a name="line.90"></a>
-<span class="sourceLineNo">091</span> /** If this is specified, no correctness testing will be done */<a name="line.91"></a>
-<span class="sourceLineNo">092</span> private static final String OPT_OMIT_CORRECTNESS_TEST = "c";<a name="line.92"></a>
-<span class="sourceLineNo">093</span><a name="line.93"></a>
-<span class="sourceLineNo">094</span> /** What encoding algorithm to test */<a name="line.94"></a>
-<span class="sourceLineNo">095</span> private static final String OPT_ENCODING_ALGORITHM = "a";<a name="line.95"></a>
-<span class="sourceLineNo">096</span><a name="line.96"></a>
-<span class="sourceLineNo">097</span> /** Number of times to run each benchmark */<a name="line.97"></a>
-<span class="sourceLineNo">098</span> private static final String OPT_BENCHMARK_N_TIMES = "t";<a name="line.98"></a>
-<span class="sourceLineNo">099</span><a name="line.99"></a>
-<span class="sourceLineNo">100</span> /** Number of first runs of every benchmark to omit from statistics */<a name="line.100"></a>
-<span class="sourceLineNo">101</span> private static final String OPT_BENCHMARK_N_OMIT = "omit";<a name="line.101"></a>
-<span class="sourceLineNo">102</span><a name="line.102"></a>
-<span class="sourceLineNo">103</span> /** Compression algorithm to use if not specified on the command line */<a name="line.103"></a>
-<span class="sourceLineNo">104</span> private static final Algorithm DEFAULT_COMPRESSION =<a name="line.104"></a>
-<span class="sourceLineNo">105</span> Compression.Algorithm.GZ;<a name="line.105"></a>
-<span class="sourceLineNo">106</span><a name="line.106"></a>
-<span class="sourceLineNo">107</span> private static final DecimalFormat DELIMITED_DECIMAL_FORMAT =<a name="line.107"></a>
-<span class="sourceLineNo">108</span> new DecimalFormat();<a name="line.108"></a>
-<span class="sourceLineNo">109</span><a name="line.109"></a>
-<span class="sourceLineNo">110</span> static {<a name="line.110"></a>
-<span class="sourceLineNo">111</span> DELIMITED_DECIMAL_FORMAT.setGroupingSize(3);<a name="line.111"></a>
-<span class="sourceLineNo">112</span> }<a name="line.112"></a>
-<span class="sourceLineNo">113</span><a name="line.113"></a>
-<span class="sourceLineNo">114</span> private static final String PCT_FORMAT = "%.2f %%";<a name="line.114"></a>
-<span class="sourceLineNo">115</span> private static final String INT_FORMAT = "%d";<a name="line.115"></a>
-<span class="sourceLineNo">116</span><a name="line.116"></a>
-<span class="sourceLineNo">117</span> private static int benchmarkNTimes = DEFAULT_BENCHMARK_N_TIMES;<a name="line.117"></a>
-<span class="sourceLineNo">118</span> private static int benchmarkNOmit = DEFAULT_BENCHMARK_N_OMIT;<a name="line.118"></a>
-<span class="sourceLineNo">119</span><a name="line.119"></a>
-<span class="sourceLineNo">120</span> private List<EncodedDataBlock> codecs = new ArrayList<>();<a name="line.120"></a>
-<span class="sourceLineNo">121</span> private long totalPrefixLength = 0;<a name="line.121"></a>
-<span class="sourceLineNo">122</span> private long totalKeyLength = 0;<a name="line.122"></a>
-<span class="sourceLineNo">123</span> private long totalValueLength = 0;<a name="line.123"></a>
-<span class="sourceLineNo">124</span> private long totalKeyRedundancyLength = 0;<a name="line.124"></a>
-<span class="sourceLineNo">125</span> private long totalCFLength = 0;<a name="line.125"></a>
-<span class="sourceLineNo">126</span><a name="line.126"></a>
-<span class="sourceLineNo">127</span> private byte[] rawKVs;<a name="line.127"></a>
-<span class="sourceLineNo">128</span> private boolean useHBaseChecksum = false;<a name="line.128"></a>
-<span class="sourceLineNo">129</span><a name="line.129"></a>
-<span class="sourceLineNo">130</span> private final String compressionAlgorithmName;<a name="line.130"></a>
-<span class="sourceLineNo">131</span> private final Algorithm compressionAlgorithm;<a name="line.131"></a>
-<span class="sourceLineNo">132</span> private final Compressor compressor;<a name="line.132"></a>
-<span class="sourceLineNo">133</span> private final Decompressor decompressor;<a name="line.133"></a>
-<span class="sourceLineNo">134</span><a name="line.134"></a>
-<span class="sourceLineNo">135</span> private static enum Manipulation {<a name="line.135"></a>
-<span class="sourceLineNo">136</span> ENCODING,<a name="line.136"></a>
-<span class="sourceLineNo">137</span> DECODING,<a name="line.137"></a>
-<span class="sourceLineNo">138</span> COMPRESSION,<a name="line.138"></a>
-<span class="sourceLineNo">139</span> DECOMPRESSION;<a name="line.139"></a>
-<span class="sourceLineNo">140</span><a name="line.140"></a>
-<span class="sourceLineNo">141</span> @Override<a name="line.141"></a>
-<span class="sourceLineNo">142</span> public String toString() {<a name="line.142"></a>
-<span class="sourceLineNo">143</span> String s = super.toString();<a name="line.143"></a>
-<span class="sourceLineNo">144</span> StringBuilder sb = new StringBuilder();<a name="line.144"></a>
-<span class="sourceLineNo">145</span> sb.append(s.charAt(0));<a name="line.145"></a>
-<span class="sourceLineNo">146</span> sb.append(s.substring(1).toLowerCase(Locale.ROOT));<a name="line.146"></a>
-<span class="sourceLineNo">147</span> return sb.toString();<a name="line.147"></a>
-<span class="sourceLineNo">148</span> }<a name="line.148"></a>
-<span class="sourceLineNo">149</span> }<a name="line.149"></a>
-<span class="sourceLineNo">150</span><a name="line.150"></a>
-<span class="sourceLineNo">151</span> /**<a name="line.151"></a>
-<span class="sourceLineNo">152</span> * @param compressionAlgorithmName What kind of algorithm should be used<a name="line.152"></a>
-<span class="sourceLineNo">153</span> * as baseline for comparison (e.g. lzo, gz).<a name="line.153"></a>
-<span class="sourceLineNo">154</span> */<a name="line.154"></a>
-<span class="sourceLineNo">155</span> public DataBlockEncodingTool(String compressionAlgorithmName) {<a name="line.155"></a>
-<span class="sourceLineNo">156</span> this.compressionAlgorithmName = compressionAlgorithmName;<a name="line.156"></a>
-<span class="sourceLineNo">157</span> this.compressionAlgorithm = Compression.getCompressionAlgorithmByName(<a name="line.157"></a>
-<span class="sourceLineNo">158</span> compressionAlgorithmName);<a name="line.158"></a>
-<span class="sourceLineNo">159</span> this.compressor = this.compressionAlgorithm.getCompressor();<a name="line.159"></a>
-<span class="sourceLineNo">160</span> this.decompressor = this.compressionAlgorithm.getDecompressor();<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> /**<a name="line.163"></a>
-<span class="sourceLineNo">164</span> * Check statistics for given HFile for different data block encoders.<a name="line.164"></a>
-<span class="sourceLineNo">165</span> * @param scanner Of file which will be compressed.<a name="line.165"></a>
-<span class="sourceLineNo">166</span> * @param kvLimit Maximal count of KeyValue which will be processed.<a name="line.166"></a>
-<span class="sourceLineNo">167</span> * @throws IOException thrown if scanner is invalid<a name="line.167"></a>
-<span class="sourceLineNo">168</span> */<a name="line.168"></a>
-<span class="sourceLineNo">169</span> public void checkStatistics(final KeyValueScanner scanner, final int kvLimit)<a name="line.169"></a>
-<span class="sourceLineNo">170</span> throws IOException {<a name="line.170"></a>
-<span class="sourceLineNo">171</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> KeyValue currentKV;<a name="line.173"></a>
-<span class="sourceLineNo">174</span><a name="line.174"></a>
-<span class="sourceLineNo">175</span> byte[] previousKey = null;<a name="line.175"></a>
-<span class="sourceLineNo">176</span> byte[] currentKey;<a name="line.176"></a>
+<span class="sourceLineNo">069</span> private static final boolean includesMemstoreTS = true;<a name="line.69"></a>
+<span class="sourceLineNo">070</span><a name="line.70"></a>
+<span class="sourceLineNo">071</span> /**<a name="line.71"></a>
+<span class="sourceLineNo">072</span> * How many times to run the benchmark. More times means better data in terms<a name="line.72"></a>
+<span class="sourceLineNo">073</span> * of statistics but slower execution. Has to be strictly larger than<a name="line.73"></a>
+<span class="sourceLineNo">074</span> * {@link #DEFAULT_BENCHMARK_N_OMIT}.<a name="line.74"></a>
+<span class="sourceLineNo">075</span> */<a name="line.75"></a>
+<span class="sourceLineNo">076</span> private static final int DEFAULT_BENCHMARK_N_TIMES = 12;<a name="line.76"></a>
+<span class="sourceLineNo">077</span><a name="line.77"></a>
+<span class="sourceLineNo">078</span> /**<a name="line.78"></a>
+<span class="sourceLineNo">079</span> * How many first runs should not be included in the benchmark. Done in order<a name="line.79"></a>
+<span class="sourceLineNo">080</span> * to exclude setup cost.<a name="line.80"></a>
+<span class="sourceLineNo">081</span> */<a name="line.81"></a>
+<span class="sourceLineNo">082</span> private static final int DEFAULT_BENCHMARK_N_OMIT = 2;<a name="line.82"></a>
+<span class="sourceLineNo">083</span><a name="line.83"></a>
+<span class="sourceLineNo">084</span> /** HFile name to be used in benchmark */<a name="line.84"></a>
+<span class="sourceLineNo">085</span> private static final String OPT_HFILE_NAME = "f";<a name="line.85"></a>
+<span class="sourceLineNo">086</span><a name="line.86"></a>
+<span class="sourceLineNo">087</span> /** Maximum number of key/value pairs to process in a single benchmark run */<a name="line.87"></a>
+<span class="sourceLineNo">088</span> private static final String OPT_KV_LIMIT = "n";<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span> /** Whether to run a benchmark to measure read throughput */<a name="line.90"></a>
+<span class="sourceLineNo">091</span> private static final String OPT_MEASURE_THROUGHPUT = "b";<a name="line.91"></a>
+<span class="sourceLineNo">092</span><a name="line.92"></a>
+<span class="sourceLineNo">093</span> /** If this is specified, no correctness testing will be done */<a name="line.93"></a>
+<span class="sourceLineNo">094</span> private static final String OPT_OMIT_CORRECTNESS_TEST = "c";<a name="line.94"></a>
+<span class="sourceLineNo">095</span><a name="line.95"></a>
+<span class="sourceLineNo">096</span> /** What compression algorithm to test */<a name="line.96"></a>
+<span class="sourceLineNo">097</span> private static final String OPT_COMPRESSION_ALGORITHM = "a";<a name="line.97"></a>
+<span class="sourceLineNo">098</span><a name="line.98"></a>
+<span class="sourceLineNo">099</span> /** Number of times to run each benchmark */<a name="line.99"></a>
+<span class="sourceLineNo">100</span> private static final String OPT_BENCHMARK_N_TIMES = "t";<a name="line.100"></a>
+<span class="sourceLineNo">101</span><a name="line.101"></a>
+<span class="sourceLineNo">102</span> /** Number of first runs of every benchmark to omit from statistics */<a name="line.102"></a>
+<span class="sourceLineNo">103</span> private static final String OPT_BENCHMARK_N_OMIT = "omit";<a name="line.103"></a>
+<span class="sourceLineNo">104</span><a name="line.104"></a>
+<span class="sourceLineNo">105</span> /** Compression algorithm to use if not specified on the command line */<a name="line.105"></a>
+<span class="sourceLineNo">106</span> private static final Algorithm DEFAULT_COMPRESSION =<a name="line.106"></a>
+<span class="sourceLineNo">107</span> Compression.Algorithm.GZ;<a name="line.107"></a>
+<span class="sourceLineNo">108</span><a name="line.108"></a>
+<span class="sourceLineNo">109</span> private static final DecimalFormat DELIMITED_DECIMAL_FORMAT =<a name="line.109"></a>
+<span class="sourceLineNo">110</span> new DecimalFormat();<a name="line.110"></a>
+<span class="sourceLineNo">111</span><a name="line.111"></a>
+<span class="sourceLineNo">112</span> static {<a name="line.112"></a>
+<span class="sourceLineNo">113</span> DELIMITED_DECIMAL_FORMAT.setGroupingSize(3);<a name="line.113"></a>
+<span class="sourceLineNo">114</span> }<a name="line.114"></a>
+<span class="sourceLineNo">115</span><a name="line.115"></a>
+<span class="sourceLineNo">116</span> private static final String PCT_FORMAT = "%.2f %%";<a name="line.116"></a>
+<span class="sourceLineNo">117</span> private static final String INT_FORMAT = "%d";<a name="line.117"></a>
+<span class="sourceLineNo">118</span><a name="line.118"></a>
+<span class="sourceLineNo">119</span> private static int benchmarkNTimes = DEFAULT_BENCHMARK_N_TIMES;<a name="line.119"></a>
+<span class="sourceLineNo">120</span> private static int benchmarkNOmit = DEFAULT_BENCHMARK_N_OMIT;<a name="line.120"></a>
+<span class="sourceLineNo">121</span><a name="line.121"></a>
+<span class="sourceLineNo">122</span> private List<EncodedDataBlock> codecs = new ArrayList<>();<a name="line.122"></a>
+<span class="sourceLineNo">123</span> private long totalPrefixLength = 0;<a name="line.123"></a>
+<span class="sourceLineNo">124</span> private long totalKeyLength = 0;<a name="line.124"></a>
+<span class="sourceLineNo">125</span> private long totalValueLength = 0;<a name="line.125"></a>
+<span class="sourceLineNo">126</span> private long totalKeyRedundancyLength = 0;<a name="line.126"></a>
+<span class="sourceLineNo">127</span> private long totalCFLength = 0;<a name="line.127"></a>
+<span class="sourceLineNo">128</span><a name="line.128"></a>
+<span class="sourceLineNo">129</span> private byte[] rawKVs;<a name="line.129"></a>
+<span class="sourceLineNo">130</span> private boolean useHBaseChecksum = false;<a name="line.130"></a>
+<span class="sourceLineNo">131</span><a name="line.131"></a>
+<span class="sourceLineNo">132</span> private final String compressionAlgorithmName;<a name="line.132"></a>
+<span class="sourceLineNo">133</span> private final Algorithm compressionAlgorithm;<a name="line.133"></a>
+<span class="sourceLineNo">134</span> private final Compressor compressor;<a name="line.134"></a>
+<span class="sourceLineNo">135</span> private final Decompressor decompressor;<a name="line.135"></a>
+<span class="sourceLineNo">136</span><a name="line.136"></a>
+<span class="sourceLineNo">137</span> // Check if HFile use Tag.<a name="line.137"></a>
+<span class="sourceLineNo">138</span> private static boolean USE_TAG = false;<a name="line.138"></a>
+<span class="sourceLineNo">139</span><a name="line.139"></a>
+<span class="sourceLineNo">140</span> private enum Manipulation {<a name="line.140"></a>
+<span class="sourceLineNo">141</span> ENCODING,<a name="line.141"></a>
+<span class="sourceLineNo">142</span> DECODING,<a name="line.142"></a>
+<span class="sourceLineNo">143</span> COMPRESSION,<a name="line.143"></a>
+<span class="sourceLineNo">144</span> DECOMPRESSION;<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> @Override<a name="line.146"></a>
+<span class="sourceLineNo">147</span> public String toString() {<a name="line.147"></a>
+<span class="sourceLineNo">148</span> String s = super.toString();<a name="line.148"></a>
+<span class="sourceLineNo">149</span> StringBuilder sb = new StringBuilder();<a name="line.149"></a>
+<span class="sourceLineNo">150</span> sb.append(s.charAt(0));<a name="line.150"></a>
+<span class="sourceLineNo">151</span> sb.append(s.substring(1).toLowerCase(Locale.ROOT));<a name="line.151"></a>
+<span class="sourceLineNo">152</span> return sb.toString();<a name="line.152"></a>
+<span class="sourceLineNo">153</span> }<a name="line.153"></a>
+<span class="sourceLineNo">154</span> }<a name="line.154"></a>
+<span class="sourceLineNo">155</span><a name="line.155"></a>
+<span class="sourceLineNo">156</span> /**<a name="line.156"></a>
+<span class="sourceLineNo">157</span> * @param compressionAlgorithmName What kind of algorithm should be used<a name="line.157"></a>
+<span class="sourceLineNo">158</span> * as baseline for comparison (e.g. lzo, gz).<a name="line.158"></a>
+<span class="sourceLineNo">159</span> */<a name="line.159"></a>
+<span class="sourceLineNo">160</span> public DataBlockEncodingTool(String compressionAlgorithmName) {<a name="line.160"></a>
+<span class="sourceLineNo">161</span> this.compressionAlgorithmName = compressionAlgorithmName;<a name="line.161"></a>
+<span class="sourceLineNo">162</span> this.compressionAlgorithm = Compression.getCompressionAlgorithmByName(<a name="line.162"></a>
+<span class="sourceLineNo">163</span> compressionAlgorithmName);<a name="line.163"></a>
+<span class="sourceLineNo">164</span> this.compressor = this.compressionAlgorithm.getCompressor();<a name="line.164"></a>
+<span class="sourceLineNo">165</span> this.decompressor = this.compressionAlgorithm.getDecompressor();<a name="line.165"></a>
+<span class="sourceLineNo">166</span> }<a name="line.166"></a>
+<span class="sourceLineNo">167</span><a name="line.167"></a>
+<span class="sourceLineNo">168</span> /**<a name="line.168"></a>
+<span class="sourceLineNo">169</span> * Check statistics for given HFile for different data block encoders.<a name="line.169"></a>
+<span class="sourceLineNo">170</span> * @param scanner Of file which will be compressed.<a name="line.170"></a>
+<span class="sourceLineNo">171</span> * @param kvLimit Maximal count of KeyValue which will be processed.<a name="line.171"></a>
+<span class="sourceLineNo">172</span> * @throws IOException thrown if scanner is invalid<a name="line.172"></a>
+<span class="sourceLineNo">173</span> */<a name="line.173"></a>
+<span class="sourceLineNo">174</span> public void checkStatistics(final KeyValueScanner scanner, final int kvLimit)<a name="line.174"></a>
+<span class="sourceLineNo">175</span> throws IOException {<a name="line.175"></a>
+<span class="sourceLineNo">176</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.176"></a>
<span class="sourceLineNo">177</span><a name="line.177"></a>
-<span class="sourceLineNo">178</span> DataBlockEncoding[] encodings = DataBlockEncoding.values();<a name="line.178"></a>
+<span class="sourceLineNo">178</span> KeyValue currentKV;<a name="line.178"></a>
<span class="sourceLineNo">179</span><a name="line.179"></a>
-<span class="sourceLineNo">180</span> ByteArrayOutputStream uncompressedOutputStream =<a name="line.180"></a>
-<span class="sourceLineNo">181</span> new ByteArrayOutputStream();<a name="line.181"></a>
+<span class="sourceLineNo">180</span> byte[] previousKey = null;<a name="line.180"></a>
+<span class="sourceLineNo">181</span> byte[] currentKey;<a name="line.181"></a>
<span class="sourceLineNo">182</span><a name="line.182"></a>
-<span class="sourceLineNo">183</span> int j = 0;<a name="line.183"></a>
-<span class="sourceLineNo">184</span> while ((currentKV = KeyValueUtil.ensureKeyValue(scanner.next())) != null && j < kvLimit) {<a name="line.184"></a>
-<span class="sourceLineNo">185</span> // Iterates through key/value pairs<a name="line.185"></a>
-<span class="sourceLineNo">186</span> j++;<a name="line.186"></a>
-<span class="sourceLineNo">187</span> currentKey = currentKV.getKey();<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (previousKey != null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> for (int i = 0; i < previousKey.length && i < currentKey.length &&<a name="line.189"></a>
-<span class="sourceLineNo">190</span> previousKey[i] == currentKey[i]; ++i) {<a name="line.190"></a>
-<span class="sourceLineNo">191</span> totalKeyRedundancyLength++;<a name="line.191"></a>
-<span class="sourceLineNo">192</span> }<a name="line.192"></a>
-<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span><a name="line.194"></a>
-<span class="sourceLineNo">195</span> uncompressedOutputStream.write(currentKV.getBuffer(),<a name="line.195"></a>
-<span class="sourceLineNo">196</span> currentKV.getOffset(), currentKV.getLength());<a name="line.196"></a>
-<span class="sourceLineNo">197</span><a name="line.197"></a>
-<span class="sourceLineNo">198</span> previousKey = currentKey;<a name="line.198"></a>
+<span class="sourceLineNo">183</span> DataBlockEncoding[] encodings = DataBlockEncoding.values();<a name="line.183"></a>
+<span class="sourceLineNo">184</span><a name="line.184"></a>
+<span class="sourceLineNo">185</span> ByteArrayOutputStream uncompressedOutputStream =<a name="line.185"></a>
+<span class="sourceLineNo">186</span> new ByteArrayOutputStream();<a name="line.186"></a>
+<span class="sourceLineNo">187</span><a name="line.187"></a>
+<span class="sourceLineNo">188</span> int j = 0;<a name="line.188"></a>
+<span class="sourceLineNo">189</span> while ((currentKV = KeyValueUtil.ensureKeyValue(scanner.next())) != null && j < kvLimit) {<a name="line.189"></a>
+<span class="sourceLineNo">190</span> // Iterates through key/value pairs<a name="line.190"></a>
+<span class="sourceLineNo">191</span> j++;<a name="line.191"></a>
+<span class="sourceLineNo">192</span> currentKey = currentKV.getKey();<a name="line.192"></a>
+<span class="sourceLineNo">193</span> if (previousKey != null) {<a name="line.193"></a>
+<span class="sourceLineNo">194</span> for (int i = 0; i < previousKey.length && i < currentKey.length &&<a name="line.194"></a>
+<span class="sourceLineNo">195</span> previousKey[i] == currentKey[i]; ++i) {<a name="line.195"></a>
+<span class="sourceLineNo">196</span> totalKeyRedundancyLength++;<a name="line.196"></a>
+<span class="sourceLineNo">197</span> }<a name="line.197"></a>
+<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span><a name="line.199"></a>
-<span class="sourceLineNo">200</span> int kLen = currentKV.getKeyLength();<a name="line.200"></a>
-<span class="sourceLineNo">201</span> int vLen = currentKV.getValueLength();<a name="line.201"></a>
-<span class="sourceLineNo">202</span> int cfLen = currentKV.getFamilyLength(currentKV.getFamilyOffset());<a name="line.202"></a>
-<span class="sourceLineNo">203</span> int restLen = currentKV.getLength() - kLen - vLen;<a name="line.203"></a>
-<span class="sourceLineNo">204</span><a name="line.204"></a>
-<span class="sourceLineNo">205</span> totalKeyLength += kLen;<a name="line.205"></a>
-<span class="sourceLineNo">206</span> totalValueLength += vLen;<a name="line.206"></a>
-<span class="sourceLineNo">207</span> totalPrefixLength += restLen;<a name="line.207"></a>
-<span class="sourceLineNo">208</span> totalCFLength += cfLen;<a name="line.208"></a>
-<span class="sourceLineNo">209</span> }<a name="line.209"></a>
-<span class="sourceLineNo">210</span><a name="line.210"></a>
-<span class="sourceLineNo">211</span> rawKVs = uncompressedOutputStream.toByteArray();<a name="line.211"></a>
-<span class="sourceLineNo">212</span> boolean useTag = (currentKV.getTagsLength() > 0);<a name="line.212"></a>
-<span class="sourceLineNo">213</span> for (DataBlockEncoding encoding : encodings) {<a name="line.213"></a>
-<span class="sourceLineNo">214</span> if (encoding == DataBlockEncoding.NONE) {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> continue;<a name="line.215"></a>
-<span class="sourceLineNo">216</span> }<a name="line.216"></a>
-<span class="sourceLineNo">217</span> DataBlockEncoder d = encoding.getEncoder();<a name="line.217"></a>
-<span class="sourceLineNo">218</span> HFileContext meta = new HFileContextBuilder()<a name="line.218"></a>
-<span class="sourceLineNo">219</span> .withCompression(Compression.Algorithm.NONE)<a name="line.219"></a>
-<span class="sourceLineNo">220</span> .withIncludesMvcc(includesMemstoreTS)<a name="line.220"></a>
-<span class="sourceLineNo">221</span> .withIncludesTags(useTag).build();<a name="line.221"></a>
-<span class="sourceLineNo">222</span> codecs.add(new EncodedDataBlock(d, encoding, rawKVs, meta ));<a name="line.222"></a>
-<span class="sourceLineNo">223</span> }<a name="line.223"></a>
-<span class="sourceLineNo">224</span> }<a name="line.224"></a>
-<span class="sourceLineNo">225</span><a name="line.225"></a>
-<span class="sourceLineNo">226</span> /**<a name="line.226"></a>
-<span class="sourceLineNo">227</span> * Verify if all data block encoders are working properly.<a name="line.227"></a>
-<span class="sourceLineNo">228</span> *<a name="line.228"></a>
-<span class="sourceLineNo">229</span> * @param scanner Of file which was compressed.<a name="line.229"></a>
-<span class="sourceLineNo">230</span> * @param kvLimit Maximal count of KeyValue which will be processed.<a name="line.230"></a>
-<span class="sourceLineNo">231</span> * @return true if all data block encoders compressed/decompressed correctly.<a name="line.231"></a>
-<span class="sourceLineNo">232</span> * @throws IOException thrown if scanner is invalid<a name="line.232"></a>
-<span class="sourceLineNo">233</span> */<a name="line.233"></a>
-<span class="sourceLineNo">234</span> public boolean verifyCodecs(final KeyValueScanner scanner, final int kvLimit)<a name="line.234"></a>
-<span class="sourceLineNo">235</span> throws IOException {<a name="line.235"></a>
-<span class="sourceLineNo">236</span> KeyValue currentKv;<a name="line.236"></a>
-<span class="sourceLineNo">237</span><a name="line.237"></a>
-<span class="sourceLineNo">238</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.238"></a>
-<span class="sourceLineNo">239</span> List<Iterator<Cell>> codecIterators = new ArrayList<>();<a name="line.239"></a>
-<span class="sourceLineNo">240</span> for(EncodedDataBlock codec : codecs) {<a name="line.240"></a>
-<span class="sourceLineNo">241</span> codecIterators.add(codec.getIterator(HFileBlock.headerSize(useHBaseChecksum)));<a name="line.241"></a>
-<span class="sourceLineNo">242</span> }<a name="line.242"></a>
-<span class="sourceLineNo">243</span><a name="line.243"></a>
-<span class="sourceLineNo">244</span> int j = 0;<a name="line.244"></a>
-<span class="sourceLineNo">245</span> while ((currentKv = KeyValueUtil.ensureKeyValue(scanner.next())) != null && j < kvLimit) {<a name="line.245"></a>
-<span class="sourceLineNo">246</span> // Iterates through key/value pairs<a name="line.246"></a>
-<span class="sourceLineNo">247</span> ++j;<a name="line.247"></a>
-<span class="sourceLineNo">248</span> for (Iterator<Cell> it : codecIterators) {<a name="line.248"></a>
-<span class="sourceLineNo">249</span> Cell c = it.next();<a name="line.249"></a>
-<span class="sourceLineNo">250</span> KeyValue codecKv = KeyValueUtil.ensureKeyValue(c);<a name="line.250"></a>
-<span class="sourceLineNo">251</span> if (codecKv == null || 0 != Bytes.compareTo(<a name="line.251"></a>
-<span class="sourceLineNo">252</span> codecKv.getBuffer(), codecKv.getOffset(), codecKv.getLength(),<a name="line.252"></a>
-<span class="sourceLineNo">253</span> currentKv.getBuffer(), currentKv.getOffset(),<a name="line.253"></a>
-<span class="sourceLineNo">254</span> currentKv.getLength())) {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> if (codecKv == null) {<a name="line.255"></a>
-<span class="sourceLineNo">256</span> LOG.error("There is a bug in codec " + it +<a name="line.256"></a>
-<span class="sourceLineNo">257</span> " it returned null KeyValue,");<a name="line.257"></a>
-<span class="sourceLineNo">258</span> } else {<a name="line.258"></a>
-<span class="sourceLineNo">259</span> int prefix = 0;<a name="line.259"></a>
-<span class="sourceLineNo">260</span> int limitLength = 2 * Bytes.SIZEOF_INT +<a name="line.260"></a>
-<span class="sourceLineNo">261</span> Math.min(codecKv.getLength(), currentKv.getLength());<a name="line.261"></a>
-<span class="sourceLineNo">262</span> while (prefix < limitLength &&<a name="line.262"></a>
-<span class="sourceLineNo">263</span> codecKv.getBuffer()[prefix + codecKv.getOffset()] ==<a name="line.263"></a>
-<span class="sourceLineNo">264</span> currentKv.getBuffer()[prefix + currentKv.getOffset()]) {<a name="line.264"></a>
-<span class="sourceLineNo">265</span> prefix++;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> }<a name="line.266"></a>
-<span class="sourceLineNo">267</span><a name="line.267"></a>
-<span class="sourceLineNo">268</span> LOG.error("There is bug in codec " + it.toString() +<a name="line.268"></a>
-<span class="sourceLineNo">269</span> "\n on element " + j +<a name="line.269"></a>
-<span class="sourceLineNo">270</span> "\n codecKv.getKeyLength() " + codecKv.getKeyLength() +<a name="line.270"></a>
-<span class="sourceLineNo">271</span> "\n codecKv.getValueLength() " + codecKv.getValueLength() +<a name="line.271"></a>
-<span class="sourceLineNo">272</span> "\n codecKv.getLength() " + codecKv.getLength() +<a name="line.272"></a>
-<span class="sourceLineNo">273</span> "\n currentKv.getKeyLength() " + currentKv.getKeyLength() +<a name="line.273"></a>
-<span class="sourceLineNo">274</span> "\n currentKv.getValueLength() " + currentKv.getValueLength() +<a name="line.274"></a>
-<span class="sourceLineNo">275</span> "\n codecKv.getLength() " + currentKv.getLength() +<a name="line.275"></a>
-<span class="sourceLineNo">276</span> "\n currentKV rowLength " + currentKv.getRowLength() +<a name="line.276"></a>
-<span class="sourceLineNo">277</span> " familyName " + currentKv.getFamilyLength() +<a name="line.277"></a>
-<span class="sourceLineNo">278</span> " qualifier " + currentKv.getQualifierLength() +<a name="line.278"></a>
-<span class="sourceLineNo">279</span> "\n prefix " + prefix +<a name="line.279"></a>
-<span class="sourceLineNo">280</span> "\n codecKv '" + Bytes.toStringBinary(codecKv.getBuffer(),<a name="line.280"></a>
-<span class="sourceLineNo">281</span> codecKv.getOffset(), prefix) + "' diff '" +<a name="line.281"></a>
-<span class="sourceLineNo">282</span> Bytes.toStringBinary(codecKv.getBuffer(),<a name="line.282"></a>
-<span class="sourceLineNo">283</span> codecKv.getOffset() + prefix, codecKv.getLength() -<a name="line.283"></a>
-<span class="sourceLineNo">284</span> prefix) + "'" +<a name="line.284"></a>
-<span class="sourceLineNo">285</span> "\n currentKv '" + Bytes.toStringBinary(<a name="line.285"></a>
-<span class="sourceLineNo">286</span> currentKv.getBuffer(),<a name="line.286"></a>
-<span class="sourceLineNo">287</span> currentKv.getOffset(), prefix) + "' diff '" +<a name="line.287"></a>
-<span class="sourceLineNo">288</span> Bytes.toStringBinary(currentKv.getBuffer(),<a name="line.288"></a>
-<span class="sourceLineNo">289</span> currentKv.getOffset() + prefix, currentKv.getLength() -<a name="line.289"></a>
-<span class="sourceLineNo">290</span> prefix) + "'"<a name="line.290"></a>
-<span class="sourceLineNo">291</span> );<a name="line.291"></a>
-<span class="sourceLineNo">292</span> }<a name="line.292"></a>
-<span class="sourceLineNo">293</span> return false;<a name="line.293"></a>
-<span class="sourceLineNo">294</span> }<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span> }<a name="line.296"></a>
-<span class="sourceLineNo">297</span><a name="line.297"></a>
-<span class="sourceLineNo">298</span> LOG.info("Verification was successful!");<a name="line.298"></a>
-<span class="sourceLineNo">299</span><a name="line.299"></a>
-<span class="sourceLineNo">300</span> return true;<a name="line.300"></a>
-<span class="sourceLineNo">301</span> }<a name="line.301"></a>
-<span class="sourceLineNo">302</span><a name="line.302"></a>
-<span class="sourceLineNo">303</span> /**<a name="line.303"></a>
-<span class="sourceLineNo">304</span> * Benchmark codec's speed.<a name="line.304"></a>
-<span class="sourceLineNo">305</span> */<a name="line.305"></a>
-<span class="sourceLineNo">306</span> public void benchmarkCodecs() throws IOException {<a name="line.306"></a>
-<span class="sourceLineNo">307</span> LOG.info("Starting a throughput benchmark for data block encoding codecs");<a name="line.307"></a>
-<span class="sourceLineNo">308</span> int prevTotalSize = -1;<a name="line.308"></a>
-<span class="sourceLineNo">309</span> for (EncodedDataBlock codec : codecs) {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> prevTotalSize = benchmarkEncoder(prevTotalSize, codec);<a name="line.310"></a>
-<span class="sourceLineNo">311</span> }<a name="line.311"></a>
-<span class="sourceLineNo">312</span><a name="line.312"></a>
-<span class="sourceLineNo">313</span> benchmarkDefaultCompression(prevTotalSize, rawKVs);<a name="line.313"></a>
-<span class="sourceLineNo">314</span> }<a name="line.314"></a>
-<span class="sourceLineNo">315</span><a name="line.315"></a>
-<span class="sourceLineNo">316</span> /**<a name="line.316"></a>
-<span class="sourceLineNo">317</span> * Benchmark compression/decompression throughput.<a name="line.317"></a>
-<span class="sourceLineNo">318</span> * @param previousTotalSize Total size used for verification. Use -1 if<a name="line.318"></a>
-<span class="sourceLineNo">319</span> * unknown.<a name="line.319"></a>
-<span class="sourceLineNo">320</span> * @param codec Tested encoder.<a name="line.320"></a>
-<span class="sourceLineNo">321</span> * @return Size of uncompressed data.<a name="line.321"></a>
-<span class="sourceLineNo">322</span> */<a name="line.322"></a>
-<span class="sourceLineNo">323</span> private int benchmarkEncoder(int previousTotalSize, EncodedDataBlock codec) {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> int prevTotalSize = previousTotalSize;<a name="line.324"></a>
-<span class="sourceLineNo">325</span> int totalSize = 0;<a name="line.325"></a>
-<span class="sourceLineNo">326</span><a name="line.326"></a>
-<span class="sourceLineNo">327</span> // decompression time<a name="line.327"></a>
-<span class="sourceLineNo">328</span> List<Long> durations = new ArrayList<>();<a name="line.328"></a>
-<span class="sourceLineNo">329</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> totalSize = 0;<a name="line.330"></a>
-<span class="sourceLineNo">331</span><a name="line.331"></a>
-<span class="sourceLineNo">332</span> Iterator<Cell> it;<a name="line.332"></a>
-<span class="sourceLineNo">333</span><a name="line.333"></a>
-<span class="sourceLineNo">334</span> it = codec.getIterator(HFileBlock.headerSize(useHBaseChecksum));<a name="line.334"></a>
-<span class="sourceLineNo">335</span><a name="line.335"></a>
-<span class="sourceLineNo">336</span> // count only the algorithm time, without memory allocations<a name="line.336"></a>
-<span class="sourceLineNo">337</span> // (expect first time)<a name="line.337"></a>
-<span class="sourceLineNo">338</span> final long startTime = System.nanoTime();<a name="line.338"></a>
-<span class="sourceLineNo">339</span> while (it.hasNext()) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> totalSize += KeyValueUtil.ensureKeyValue(it.next()).getLength();<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span> final long finishTime = System.nanoTime();<a name="line.342"></a>
-<span class="sourceLineNo">343</span> if (itTime >= benchmarkNOmit) {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> durations.add(finishTime - startTime);<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> if (prevTotalSize != -1 && prevTotalSize != totalSize) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> throw new IllegalStateException(String.format(<a name="line.348"></a>
-<span class="sourceLineNo">349</span> "Algorithm '%s' decoded data to different size", codec.toString()));<a name="line.349"></a>
-<span class="sourceLineNo">350</span> }<a name="line.350"></a>
-<span class="sourceLineNo">351</span> prevTotalSize = totalSize;<a name="line.351"></a>
-<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">200</span> // Add tagsLen zero to cells don't include tags. Since the process of<a name="line.200"></a>
+<span class="sourceLineNo">201</span> // scanner converts byte array to KV would abandon tagsLen part if tagsLen<a name="line.201"></a>
+<span class="sourceLineNo">202</span> // is zero. But we still needs the tagsLen part to check if current cell<a name="line.202"></a>
+<span class="sourceLineNo">203</span> // include tags. If USE_TAG is true, HFile contains cells with tags,<a name="line.203"></a>
+<span class="sourceLineNo">204</span> // if the cell tagsLen equals 0, it means other cells may have tags.<a name="line.204"></a>
+<span class="sourceLineNo">205</span> if (USE_TAG && currentKV.getTagsLength() == 0) {<a name="line.205"></a>
+<span class="sourceLineNo">206</span> uncompressedOutputStream.write(currentKV.getBuffer(),<a name="line.206"></a>
+<span class="sourceLineNo">207</span> currentKV.getOffset(), currentKV.getLength());<a name="line.207"></a>
+<span class="sourceLineNo">208</span> // write tagsLen = 0.<a name="line.208"></a>
+<span class="sourceLineNo">209</span> uncompressedOutputStream.write(Bytes.toBytes((short) 0));<a name="line.209"></a>
+<span class="sourceLineNo">210</span> } else {<a name="line.210"></a>
+<span class="sourceLineNo">211</span> uncompressedOutputStream.write(currentKV.getBuffer(),<a name="line.211"></a>
+<span class="sourceLineNo">212</span> currentKV.getOffset(), currentKV.getLength());<a name="line.212"></a>
+<span class="sourceLineNo">213</span> }<a name="line.213"></a>
+<span class="sourceLineNo">214</span><a name="line.214"></a>
+<span class="sourceLineNo">215</span> if(includesMemstoreTS) {<a name="line.215"></a>
+<span class="sourceLineNo">216</span> WritableUtils.writeVLong(<a name="line.216"></a>
+<span class="sourceLineNo">217</span> new DataOutputStream(uncompressedOutputStream), currentKV.getSequenceId());<a name="line.217"></a>
+<span class="sourceLineNo">218</span> }<a name="line.218"></a>
+<span class="sourceLineNo">219</span><a name="line.219"></a>
+<span class="sourceLineNo">220</span> previousKey = currentKey;<a name="line.220"></a>
+<span class="sourceLineNo">221</span><a name="line.221"></a>
+<span class="sourceLineNo">222</span> int kLen = currentKV.getKeyLength();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> int vLen = currentKV.getValueLength();<a name="line.223"></a>
+<span class="sourceLineNo">224</span> int cfLen = currentKV.getFamilyLength(currentKV.getFamilyOffset());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> int restLen = currentKV.getLength() - kLen - vLen;<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> totalKeyLength += kLen;<a name="line.227"></a>
+<span class="sourceLineNo">228</span> totalValueLength += vLen;<a name="line.228"></a>
+<span class="sourceLineNo">229</span> totalPrefixLength += restLen;<a name="line.229"></a>
+<span class="sourceLineNo">230</span> totalCFLength += cfLen;<a name="line.230"></a>
+<span class="sourceLineNo">231</span> }<a name="line.231"></a>
+<span class="sourceLineNo">232</span><a name="line.232"></a>
+<span class="sourceLineNo">233</span> rawKVs = uncompressedOutputStream.toByteArray();<a name="line.233"></a>
+<span class="sourceLineNo">234</span> for (DataBlockEncoding encoding : encodings) {<a name="line.234"></a>
+<span class="sourceLineNo">235</span> if (encoding == DataBlockEncoding.NONE) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> continue;<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span> DataBlockEncoder d = encoding.getEncoder();<a name="line.238"></a>
+<span class="sourceLineNo">239</span> HFileContext meta = new HFileContextBuilder()<a name="line.239"></a>
+<span class="sourceLineNo">240</span> .withDataBlockEncoding(encoding)<a name="line.240"></a>
+<span class="sourceLineNo">241</span> .withCompression(Compression.Algorithm.NONE)<a name="line.241"></a>
+<span class="sourceLineNo">242</span> .withIncludesMvcc(includesMemstoreTS)<a name="line.242"></a>
+<span class="sourceLineNo">243</span> .withIncludesTags(USE_TAG).build();<a name="line.243"></a>
+<span class="sourceLineNo">244</span> codecs.add(new EncodedDataBlock(d, encoding, rawKVs, meta ));<a name="line.244"></a>
+<span class="sourceLineNo">245</span> }<a name="line.245"></a>
+<span class="sourceLineNo">246</span> }<a name="line.246"></a>
+<span class="sourceLineNo">247</span><a name="line.247"></a>
+<span class="sourceLineNo">248</span> /**<a name="line.248"></a>
+<span class="sourceLineNo">249</span> * Verify if all data block encoders are working properly.<a name="line.249"></a>
+<span class="sourceLineNo">250</span> *<a name="line.250"></a>
+<span class="sourceLineNo">251</span> * @param scanner Of file which was compressed.<a name="line.251"></a>
+<span class="sourceLineNo">252</span> * @param kvLimit Maximal count of KeyValue which will be processed.<a name="line.252"></a>
+<span class="sourceLineNo">253</span> * @return true if all data block encoders compressed/decompressed correctly.<a name="line.253"></a>
+<span class="sourceLineNo">254</span> * @throws IOException thrown if scanner is invalid<a name="line.254"></a>
+<span class="sourceLineNo">255</span> */<a name="line.255"></a>
+<span class="sourceLineNo">256</span> public boolean verifyCodecs(final KeyValueScanner scanner, final int kvLimit)<a name="line.256"></a>
+<span class="sourceLineNo">257</span> throws IOException {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> KeyValue currentKv;<a name="line.258"></a>
+<span class="sourceLineNo">259</span><a name="line.259"></a>
+<span class="sourceLineNo">260</span> scanner.seek(KeyValue.LOWESTKEY);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> List<Iterator<Cell>> codecIterators = new ArrayList<>();<a name="line.261"></a>
+<span class="sourceLineNo">262</span> for(EncodedDataBlock codec : codecs) {<a name="line.262"></a>
+<span class="sourceLineNo">263</span> codecIterators.add(codec.getIterator(HFileBlock.headerSize(useHBaseChecksum)));<a name="line.263"></a>
+<span class="sourceLineNo">264</span> }<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> int j = 0;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> while ((currentKv = KeyValueUtil.ensureKeyValue(scanner.next())) != null && j < kvLimit) {<a name="line.267"></a>
+<span class="sourceLineNo">268</span> // Iterates through key/value pairs<a name="line.268"></a>
+<span class="sourceLineNo">269</span> ++j;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> for (Iterator<Cell> it : codecIterators) {<a name="line.270"></a>
+<span class="sourceLineNo">271</span> Cell c = it.next();<a name="line.271"></a>
+<span class="sourceLineNo">272</span> KeyValue codecKv = KeyValueUtil.ensureKeyValue(c);<a name="line.272"></a>
+<span class="sourceLineNo">273</span> if (codecKv == null || 0 != Bytes.compareTo(<a name="line.273"></a>
+<span class="sourceLineNo">274</span> codecKv.getBuffer(), codecKv.getOffset(), codecKv.getLength(),<a name="line.274"></a>
+<span class="sourceLineNo">275</span> currentKv.getBuffer(), currentKv.getOffset(),<a name="line.275"></a>
+<span class="sourceLineNo">276</span> currentKv.getLength())) {<a name="line.276"></a>
+<span class="sourceLineNo">277</span> if (codecKv == null) {<a name="line.277"></a>
+<span class="sourceLineNo">278</span> LOG.error("There is a bug in codec " + it +<a name="line.278"></a>
+<span class="sourceLineNo">279</span> " it returned null KeyValue,");<a name="line.279"></a>
+<span class="sourceLineNo">280</span> } else {<a name="line.280"></a>
+<span class="sourceLineNo">281</span> int prefix = 0;<a name="line.281"></a>
+<span class="sourceLineNo">282</span> int limitLength = 2 * Bytes.SIZEOF_INT +<a name="line.282"></a>
+<span class="sourceLineNo">283</span> Math.min(codecKv.getLength(), currentKv.getLength());<a name="line.283"></a>
+<span class="sourceLineNo">284</span> while (prefix < limitLength &&<a name="line.284"></a>
+<span class="sourceLineNo">285</span> codecKv.getBuffer()[prefix + codecKv.getOffset()] ==<a name="line.285"></a>
+<span class="sourceLineNo">286</span> currentKv.getBuffer()[prefix + currentKv.getOffset()]) {<a name="line.286"></a>
+<span class="sourceLineNo">287</span> prefix++;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> }<a name="line.288"></a>
+<span class="sourceLineNo">289</span><a name="line.289"></a>
+<span class="sourceLineNo">290</span> LOG.error("There is bug in codec " + it.toString() +<a name="line.290"></a>
+<span class="sourceLineNo">291</span> "\n on element " + j +<a name="line.291"></a>
+<span class="sourceLineNo">292</span> "\n codecKv.getKeyLength() " + codecKv.getKeyLength() +<a name="line.292"></a>
+<span class="sourceLineNo">293</span> "\n codecKv.getValueLength() " + codecKv.getValueLength() +<a name="line.293"></a>
+<span class="sourceLineNo">294</span> "\n codecKv.getLength() " + codecKv.getLength() +<a name="line.294"></a>
+<span class="sourceLineNo">295</span> "\n currentKv.getKeyLength() " + currentKv.getKeyLength() +<a name="line.295"></a>
+<span class="sourceLineNo">296</span> "\n currentKv.getValueLength() " + currentKv.getValueLength() +<a name="line.296"></a>
+<span class="sourceLineNo">297</span> "\n codecKv.getLength() " + currentKv.getLength() +<a name="line.297"></a>
+<span class="sourceLineNo">298</span> "\n currentKV rowLength " + currentKv.getRowLength() +<a name="line.298"></a>
+<span class="sourceLineNo">299</span> " familyName " + currentKv.getFamilyLength() +<a name="line.299"></a>
+<span class="sourceLineNo">300</span> " qualifier " + currentKv.getQualifierLength() +<a name="line.300"></a>
+<span class="sourceLineNo">301</span> "\n prefix " + prefix +<a name="line.301"></a>
+<span class="sourceLineNo">302</span> "\n codecKv '" + Bytes.toStringBinary(codecKv.getBuffer(),<a name="line.302"></a>
+<span class="sourceLineNo">303</span> codecKv.getOffset(), prefix) + "' diff '" +<a name="line.303"></a>
+<span class="sourceLineNo">304</span> Bytes.toStringBinary(codecKv.getBuffer(),<a name="line.304"></a>
+<span class="sourceLineNo">305</span> codecKv.getOffset() + prefix, codecKv.getLength() -<a name="line.305"></a>
+<span class="sourceLineNo">306</span> prefix) + "'" +<a name="line.306"></a>
+<span class="sourceLineNo">307</span> "\n currentKv '" + Bytes.toStringBinary(<a name="line.307"></a>
+<span class="sourceLineNo">308</span> currentKv.getBuffer(),<a name="line.308"></a>
+<span class="sourceLineNo">309</span> currentKv.getOffset(), prefix) + "' diff '" +<a name="line.309"></a>
+<span class="sourceLineNo">310</span> Bytes.toStringBinary(currentKv.getBuffer(),<a name="line.310"></a>
+<span class="sourceLineNo">311</span> currentKv.getOffset() + prefix, currentKv.getLength() -<a name="line.311"></a>
+<span class="sourceLineNo">312</span> prefix) + "'"<a name="line.312"></a>
+<span class="sourceLineNo">313</span> );<a name="line.313"></a>
+<span class="sourceLineNo">314</span> }<a name="line.314"></a>
+<span class="sourceLineNo">315</span> return false;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> }<a name="line.316"></a>
+<span class="sourceLineNo">317</span> }<a name="line.317"></a>
+<span class="sourceLineNo">318</span> }<a name="line.318"></a>
+<span class="sourceLineNo">319</span><a name="line.319"></a>
+<span class="sourceLineNo">320</span> LOG.info("Verification was successful!");<a name="line.320"></a>
+<span class="sourceLineNo">321</span><a name="line.321"></a>
+<span class="sourceLineNo">322</span> return true;<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> /**<a name="line.325"></a>
+<span class="sourceLineNo">326</span> * Benchmark codec's speed.<a name="line.326"></a>
+<span class="sourceLineNo">327</span> */<a name="line.327"></a>
+<span class="sourceLineNo">328</span> public void benchmarkCodecs() throws IOException {<a name="line.328"></a>
+<span class="sourceLineNo">329</span> LOG.info("Starting a throughput benchmark for data block encoding codecs");<a name="line.329"></a>
+<span class="sourceLineNo">330</span> int prevTotalSize = -1;<a name="line.330"></a>
+<span class="sourceLineNo">331</span> for (EncodedDataBlock codec : codecs) {<a name="line.331"></a>
+<span class="sourceLineNo">332</span> prevTotalSize = benchmarkEncoder(prevTotalSize, codec);<a name="line.332"></a>
+<span class="sourceLineNo">333</span> }<a name="line.333"></a>
+<span class="sourceLineNo">334</span><a name="line.334"></a>
+<span class="sourceLineNo">335</span> benchmarkDefaultCompression(prevTotalSize, rawKVs);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> /**<a name="line.338"></a>
+<span class="sourceLineNo">339</span> * Benchmark compression/decompression throughput.<a name="line.339"></a>
+<span class="sourceLineNo">340</span> * @param previousTotalSize Total size used for verification. Use -1 if<a name="line.340"></a>
+<span class="sourceLineNo">341</span> * unknown.<a name="line.341"></a>
+<span class="sourceLineNo">342</span> * @param codec Tested encoder.<a name="line.342"></a>
+<span class="sourceLineNo">343</span> * @return Size of uncompressed data.<a name="line.343"></a>
+<span class="sourceLineNo">344</span> */<a name="line.344"></a>
+<span class="sourceLineNo">345</span> private int benchmarkEncoder(int previousTotalSize, EncodedDataBlock codec) {<a name="line.345"></a>
+<span class="sourceLineNo">346</span> int prevTotalSize = previousTotalSize;<a name="line.346"></a>
+<span class="sourceLineNo">347</span> int totalSize = 0;<a name="line.347"></a>
+<span class="sourceLineNo">348</span><a name="line.348"></a>
+<span class="sourceLineNo">349</span> // decompression time<a name="line.349"></a>
+<span class="sourceLineNo">350</span> List<Long> durations = new ArrayList<>();<a name="line.350"></a>
+<span class="sourceLineNo">351</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.351"></a>
+<span class="sourceLineNo">352</span> totalSize = 0;<a name="line.352"></a>
<span class="sourceLineNo">353</span><a name="line.353"></a>
-<span class="sourceLineNo">354</span> List<Long> encodingDurations = new ArrayList<>();<a name="line.354"></a>
-<span class="sourceLineNo">355</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> final long startTime = System.nanoTime();<a name="line.356"></a>
-<span class="sourceLineNo">357</span> codec.encodeData();<a name="line.357"></a>
-<span class="sourceLineNo">358</span> final long finishTime = System.nanoTime();<a name="line.358"></a>
-<span class="sourceLineNo">359</span> if (itTime >= benchmarkNOmit) {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> encodingDurations.add(finishTime - startTime);<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span> }<a name="line.362"></a>
-<span class="sourceLineNo">363</span><a name="line.363"></a>
-<span class="sourceLineNo">364</span> System.out.println(codec.toString() + ":");<a name="line.364"></a>
-<span class="sourceLineNo">365</span> printBenchmarkResult(totalSize, encodingDurations, Manipulation.ENCODING);<a name="line.365"></a>
-<span class="sourceLineNo">366</span> printBenchmarkResult(totalSize, durations, Manipulation.DECODING);<a name="line.366"></a>
-<span class="sourceLineNo">367</span> System.out.println();<a name="line.367"></a>
+<span class="sourceLineNo">354</span> Iterator<Cell> it;<a name="line.354"></a>
+<span class="sourceLineNo">355</span><a name="line.355"></a>
+<span class="sourceLineNo">356</span> it = codec.getIterator(HFileBlock.headerSize(useHBaseChecksum));<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> // count only the algorithm time, without memory allocations<a name="line.358"></a>
+<span class="sourceLineNo">359</span> // (expect first time)<a name="line.359"></a>
+<span class="sourceLineNo">360</span> final long startTime = System.nanoTime();<a name="line.360"></a>
+<span class="sourceLineNo">361</span> while (it.hasNext()) {<a name="line.361"></a>
+<span class="sourceLineNo">362</span> totalSize += KeyValueUtil.ensureKeyValue(it.next()).getLength();<a name="line.362"></a>
+<span class="sourceLineNo">363</span> }<a name="line.363"></a>
+<span class="sourceLineNo">364</span> final long finishTime = System.nanoTime();<a name="line.364"></a>
+<span class="sourceLineNo">365</span> if (itTime >= benchmarkNOmit) {<a name="line.365"></a>
+<span class="sourceLineNo">366</span> durations.add(finishTime - startTime);<a name="line.366"></a>
+<span class="sourceLineNo">367</span> }<a name="line.367"></a>
<span class="sourceLineNo">368</span><a name="line.368"></a>
-<span class="sourceLineNo">369</span> return prevTotalSize;<a name="line.369"></a>
-<span class="sourceLineNo">370</span> }<a name="line.370"></a>
-<span class="sourceLineNo">371</span><a name="line.371"></a>
-<span class="sourceLineNo">372</span> private void benchmarkDefaultCompression(int totalSize, byte[] rawBuffer)<a name="line.372"></a>
-<span class="sourceLineNo">373</span> throws IOException {<a name="line.373"></a>
-<span class="sourceLineNo">374</span> benchmarkAlgorithm(compressionAlgorithm,<a name="line.374"></a>
-<span class="sourceLineNo">375</span> compressionAlgorithmName.toUpperCase(Locale.ROOT), rawBuffer, 0, totalSize);<a name="line.375"></a>
-<span class="sourceLineNo">376</span> }<a name="line.376"></a>
-<span class="sourceLineNo">377</span><a name="line.377"></a>
-<span class="sourceLineNo">378</span> /**<a name="line.378"></a>
-<span class="sourceLineNo">379</span> * Check decompress performance of a given algorithm and print it.<a name="line.379"></a>
-<span class="sourceLineNo">380</span> * @param algorithm Compression algorithm.<a name="line.380"></a>
-<span class="sourceLineNo">381</span> * @param name Name of algorithm.<a name="line.381"></a>
-<span class="sourceLineNo">382</span> * @param buffer Buffer to be compressed.<a name="line.382"></a>
-<span class="sourceLineNo">383</span> * @param offset Position of the beginning of the data.<a name="line.383"></a>
-<span class="sourceLineNo">384</span> * @param length Length of data in buffer.<a name="line.384"></a>
-<span class="sourceLineNo">385</span> * @throws IOException<a name="line.385"></a>
-<span class="sourceLineNo">386</span> */<a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void benchmarkAlgorithm(Compression.Algorithm algorithm, String name,<a name="line.387"></a>
-<span class="sourceLineNo">388</span> byte[] buffer, int offset, int length) throws IOException {<a name="line.388"></a>
-<span class="sourceLineNo">389</span> System.out.println(name + ":");<a name="line.389"></a>
+<span class="sourceLineNo">369</span> if (prevTotalSize != -1 && prevTotalSize != totalSize) {<a name="line.369"></a>
+<span class="sourceLineNo">370</span> throw new IllegalStateException(String.format(<a name="line.370"></a>
+<span class="sourceLineNo">371</span> "Algorithm '%s' decoded data to different size", codec.toString()));<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span> prevTotalSize = totalSize;<a name="line.373"></a>
+<span class="sourceLineNo">374</span> }<a name="line.374"></a>
+<span class="sourceLineNo">375</span><a name="line.375"></a>
+<span class="sourceLineNo">376</span> List<Long> encodingDurations = new ArrayList<>();<a name="line.376"></a>
+<span class="sourceLineNo">377</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.377"></a>
+<span class="sourceLineNo">378</span> final long startTime = System.nanoTime();<a name="line.378"></a>
+<span class="sourceLineNo">379</span> codec.encodeData();<a name="line.379"></a>
+<span class="sourceLineNo">380</span> final long finishTime = System.nanoTime();<a name="line.380"></a>
+<span class="sourceLineNo">381</span> if (itTime >= benchmarkNOmit) {<a name="line.381"></a>
+<span class="sourceLineNo">382</span> encodingDurations.add(finishTime - startTime);<a name="line.382"></a>
+<span class="sourceLineNo">383</span> }<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> System.out.println(codec.toString() + ":");<a name="line.386"></a>
+<span class="sourceLineNo">387</span> printBenchmarkResult(totalSize, encodingDurations, Manipulation.ENCODING);<a name="line.387"></a>
+<span class="sourceLineNo">388</span> printBenchmarkResult(totalSize, durations, Manipulation.DECODING);<a name="line.388"></a>
+<span class="sourceLineNo">389</span> System.out.println();<a name="line.389"></a>
<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> // compress it<a name="line.391"></a>
-<span class="sourceLineNo">392</span> List<Long> compressDurations = new ArrayList<>();<a name="line.392"></a>
-<span class="sourceLineNo">393</span> ByteArrayOutputStream compressedStream = new ByteArrayOutputStream();<a name="line.393"></a>
-<span class="sourceLineNo">394</span> CompressionOutputStream compressingStream =<a name="line.394"></a>
-<span class="sourceLineNo">395</span> algorithm.createPlainCompressionStream(compressedStream, compressor);<a name="line.395"></a>
-<span class="sourceLineNo">396</span> try {<a name="line.396"></a>
-<span class="sourceLineNo">397</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.397"></a>
-<span class="sourceLineNo">398</span> final long startTime = System.nanoTime();<a name="line.398"></a>
-<span class="sourceLineNo">399</span> compressingStream.resetState();<a name="line.399"></a>
-<span class="sourceLineNo">400</span> compressedStream.reset();<a name="line.400"></a>
-<span class="sourceLineNo">401</span> compressingStream.write(buffer, offset, length);<a name="line.401"></a>
-<span class="sourceLineNo">402</span> compressingStream.flush();<a name="line.402"></a>
-<span class="sourceLineNo">403</span> compressedStream.toByteArray();<a name="line.403"></a>
-<span class="sourceLineNo">404</span><a name="line.404"></a>
-<span class="sourceLineNo">405</span> final long finishTime = System.nanoTime();<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> // add time record<a name="line.407"></a>
-<span class="sourceLineNo">408</span> if (itTime >= benchmarkNOmit) {<a name="line.408"></a>
-<span class="sourceLineNo">409</span> compressDurations.add(finishTime - startTime);<a name="line.409"></a>
-<span class="sourceLineNo">410</span> }<a name="line.410"></a>
-<span class="sourceLineNo">411</span> }<a name="line.411"></a>
-<span class="sourceLineNo">412</span> } catch (IOException e) {<a name="line.412"></a>
-<span class="sourceLineNo">413</span> throw new RuntimeException(String.format(<a name="line.413"></a>
-<span class="sourceLineNo">414</span> "Benchmark, or encoding algorithm '%s' cause some stream problems",<a name="line.414"></a>
-<span class="sourceLineNo">415</span> name), e);<a name="line.415"></a>
-<span class="sourceLineNo">416</span> }<a name="line.416"></a>
-<span class="sourceLineNo">417</span> compressingStream.close();<a name="line.417"></a>
-<span class="sourceLineNo">418</span> printBenchmarkResult(length, compressDurations, Manipulation.COMPRESSION);<a name="line.418"></a>
-<span class="sourceLineNo">419</span><a name="line.419"></a>
-<span class="sourceLineNo">420</span> byte[] compBuffer = compressedStream.toByteArray();<a name="line.420"></a>
-<span class="sourceLineNo">421</span><a name="line.421"></a>
-<span class="sourceLineNo">422</span> // uncompress it several times and measure performance<a name="line.422"></a>
-<span class="sourceLineNo">423</span> List<Long> durations = new ArrayList<>();<a name="line.423"></a>
-<span class="sourceLineNo">424</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> final long startTime = System.nanoTime();<a name="line.425"></a>
-<span class="sourceLineNo">426</span> byte[] newBuf = new byte[length + 1];<a name="line.426"></a>
-<span class="sourceLineNo">427</span><a name="line.427"></a>
-<span class="sourceLineNo">428</span> try {<a name="line.428"></a>
-<span class="sourceLineNo">429</span> ByteArrayInputStream downStream = new ByteArrayInputStream(compBuffer,<a name="line.429"></a>
-<span class="sourceLineNo">430</span> 0, compBuffer.length);<a name="line.430"></a>
-<span class="sourceLineNo">431</span> InputStream decompressedStream = algorithm.createDecompressionStream(<a name="line.431"></a>
-<span class="sourceLineNo">432</span> downStream, decompressor, 0);<a name="line.432"></a>
-<span class="sourceLineNo">433</span><a name="line.433"></a>
-<span class="sourceLineNo">434</span> int destOffset = 0;<a name="line.434"></a>
-<span class="sourceLineNo">435</span> int nextChunk;<a name="line.435"></a>
-<span class="sourceLineNo">436</span> while ((nextChunk = decompressedStream.available()) > 0) {<a name="line.436"></a>
-<span class="sourceLineNo">437</span> destOffset += decompressedStream.read(newBuf, destOffset, nextChunk);<a name="line.437"></a>
-<span class="sourceLineNo">438</span> }<a name="line.438"></a>
-<span class="sourceLineNo">439</span> decompressedStream.close();<a name="line.439"></a>
-<span class="sourceLineNo">440</span><a name="line.440"></a>
-<span class="sourceLineNo">441</span> // iterate over KeyValues<a name="line.441"></a>
-<span class="sourceLineNo">442</span> KeyValue kv;<a name="line.442"></a>
-<span class="sourceLineNo">443</span> for (int pos = 0; pos < length; pos += kv.getLength()) {<a name="line.443"></a>
-<span class="sourceLineNo">444</span> kv = new KeyValue(newBuf, pos);<a name="line.444"></a>
-<span class="sourceLineNo">445</span> }<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> } catch (IOException e) {<a name="line.447"></a>
-<span class="sourceLineNo">448</span> throw new RuntimeException(String.format(<a name="line.448"></a>
-<span class="sourceLineNo">449</span> "Decoding path in '%s' algorithm cause exception ", name), e);<a name="line.449"></a>
-<span class="sourceLineNo">450</span> }<a name="line.450"></a>
+<span class="sourceLineNo">391</span> return prevTotalSize;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> private void benchmarkDefaultCompression(int totalSize, byte[] rawBuffer)<a name="line.394"></a>
+<span class="sourceLineNo">395</span> throws IOException {<a name="line.395"></a>
+<span class="sourceLineNo">396</span> benchmarkAlgorithm(compressionAlgorithm,<a name="line.396"></a>
+<span class="sourceLineNo">397</span> compressionAlgorithmName.toUpperCase(Locale.ROOT), rawBuffer, 0, totalSize);<a name="line.397"></a>
+<span class="sourceLineNo">398</span> }<a name="line.398"></a>
+<span class="sourceLineNo">399</span><a name="line.399"></a>
+<span class="sourceLineNo">400</span> /**<a name="line.400"></a>
+<span class="sourceLineNo">401</span> * Check decompress performance of a given algorithm and print it.<a name="line.401"></a>
+<span class="sourceLineNo">402</span> * @param algorithm Compression algorithm.<a name="line.402"></a>
+<span class="sourceLineNo">403</span> * @param name Name of algorithm.<a name="line.403"></a>
+<span class="sourceLineNo">404</span> * @param buffer Buffer to be compressed.<a name="line.404"></a>
+<span class="sourceLineNo">405</span> * @param offset Position of the beginning of the data.<a name="line.405"></a>
+<span class="sourceLineNo">406</span> * @param length Length of data in buffer.<a name="line.406"></a>
+<span class="sourceLineNo">407</span> * @throws IOException<a name="line.407"></a>
+<span class="sourceLineNo">408</span> */<a name="line.408"></a>
+<span class="sourceLineNo">409</span> public void benchmarkAlgorithm(Compression.Algorithm algorithm, String name,<a name="line.409"></a>
+<span class="sourceLineNo">410</span> byte[] buffer, int offset, int length) throws IOException {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> System.out.println(name + ":");<a name="line.411"></a>
+<span class="sourceLineNo">412</span><a name="line.412"></a>
+<span class="sourceLineNo">413</span> // compress it<a name="line.413"></a>
+<span class="sourceLineNo">414</span> List<Long> compressDurations = new ArrayList<>();<a name="line.414"></a>
+<span class="sourceLineNo">415</span> ByteArrayOutputStream compressedStream = new ByteArrayOutputStream();<a name="line.415"></a>
+<span class="sourceLineNo">416</span> CompressionOutputStream compressingStream =<a name="line.416"></a>
+<span class="sourceLineNo">417</span> algorithm.createPlainCompressionStream(compressedStream, compressor);<a name="line.417"></a>
+<span class="sourceLineNo">418</span> try {<a name="line.418"></a>
+<span class="sourceLineNo">419</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.419"></a>
+<span class="sourceLineNo">420</span> final long startTime = System.nanoTime();<a name="line.420"></a>
+<span class="sourceLineNo">421</span> // The compressedStream should reset before compressingStream resetState since in GZ<a name="line.421"></a>
+<span class="sourceLineNo">422</span> // resetStatue will write header in the outputstream.<a name="line.422"></a>
+<span class="sourceLineNo">423</span> compressedStream.reset();<a name="line.423"></a>
+<span class="sourceLineNo">424</span> compressingStream.resetState();<a name="line.424"></a>
+<span class="sourceLineNo">425</span> compressingStream.write(buffer, offset, length);<a name="line.425"></a>
+<span class="sourceLineNo">426</span> compressingStream.flush();<a name="line.426"></a>
+<span class="sourceLineNo">427</span> compressedStream.toByteArray();<a name="line.427"></a>
+<span class="sourceLineNo">428</span><a name="line.428"></a>
+<span class="sourceLineNo">429</span> final long finishTime = System.nanoTime();<a name="line.429"></a>
+<span class="sourceLineNo">430</span><a name="line.430"></a>
+<span class="sourceLineNo">431</span> // add time record<a name="line.431"></a>
+<span class="sourceLineNo">432</span> if (itTime >= benchmarkNOmit) {<a name="line.432"></a>
+<span class="sourceLineNo">433</span> compressDurations.add(finishTime - startTime);<a name="line.433"></a>
+<span class="sourceLineNo">434</span> }<a name="line.434"></a>
+<span class="sourceLineNo">435</span> }<a name="line.435"></a>
+<span class="sourceLineNo">436</span> } catch (IOException e) {<a name="line.436"></a>
+<span class="sourceLineNo">437</span> throw new RuntimeException(String.format(<a name="line.437"></a>
+<span class="sourceLineNo">438</span> "Benchmark, or encoding algorithm '%s' cause some stream problems",<a name="line.438"></a>
+<span class="sourceLineNo">439</span> name), e);<a name="line.439"></a>
+<span class="sourceLineNo">440</span> }<a name="line.440"></a>
+<span class="sourceLineNo">441</span> compressingStream.close();<a name="line.441"></a>
+<span class="sourceLineNo">442</span> printBenchmarkResult(length, compressDurations, Manipulation.COMPRESSION);<a name="line.442"></a>
+<span class="sourceLineNo">443</span><a name="line.443"></a>
+<span class="sourceLineNo">444</span> byte[] compBuffer = compressedStream.toByteArray();<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> // uncompress it several times and measure performance<a name="line.446"></a>
+<span class="sourceLineNo">447</span> List<Long> durations = new ArrayList<>();<a name="line.447"></a>
+<span class="sourceLineNo">448</span> for (int itTime = 0; itTime < benchmarkNTimes; ++itTime) {<a name="line.448"></a>
+<span class="sourceLineNo">449</span> final long startTime = System.nanoTime();<a name="line.449"></a>
+<span class="sourceLineNo">450</span> byte[] newBuf = new byte[length + 1];<a name="line.450"></a>
<span class="sourceLineNo">451</span><a name="line.451"></a>
-<span class="sourceLineNo">452</span> final long finishTime = System.nanoTime();<a name="line.452"></a>
-<span class="sourceLineNo">453</span><a name="line.453"></a>
-<span class="sourceLineNo">454</span> // check correctness<a name="line.454"></a>
-<span class="sourceLineNo">455</span> if (0 != Bytes.compareTo(buffer, 0, length, newBuf, 0, length)) {<a name="line.455"></a>
-<span class="sourceLineNo">456</span> int prefix = 0;<a name="line.456"></a>
-<span class="sourceLineNo">457</span> for(; prefix < buffer.length && prefix < newBuf.length; ++prefix) {<a name="line.457"></a>
-<span class="sourceLineNo">458</span> if (buffer[prefix] != newBuf[prefix]) {<a name="line.458"></a>
-<span class="sourceLineNo">459</span> break;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> }<a name="line.460"></a>
-<span class="sourceLineNo">461</span> }<a name="line.461"></a>
-<span class="sourceLineNo">462</span> throw new RuntimeException(String.format(<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "Algorithm '%s' is corrupting the data", name));<a name="line.463"></a>
-<span class="sourceLineNo">464</span> }<a name="line.464"></a>
-<span class="sourceLineNo">465</span><a name="line.465"></a>
-<span class="sourceLineNo">466</span> // add time record<a name="line.466"></a>
-<span class="sourceLineNo">467</span> if (itTime >= benchmarkNOmit) {<a name="line.467"></a>
-<span class="sourceLineNo">468</span> durations.add(finishTime - startTime);<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> printBenchmarkResult(length, durations, Manipulation.DECOMPRESSION);<a name="line.471"></a>
-<span class="sourceLineNo">472</span> System.out.println();<a name="line.472"></a>
-<span class="sourceLineNo">473</span> }<a name="line.473"></a>
-<span class="sourceLineNo">474</span><a name="line.474"></a>
-<span class="sourceLineNo">475</span> private static final double BYTES_IN_MB = 1024 * 1024.0;<a name="line.475"></a>
-<span class="sourceLineNo">476</span> private static final double NS_IN_SEC = 1000.0 * 1000.0 * 1000.0;<a name="line.476"></a>
-<span class="sourceLineNo">477</span> private static final double MB_SEC_COEF = NS_IN_SEC / BYTES_IN_MB;<a name="line.477"></a>
-<span class="sourceLineNo">478</span><a name="line.478"></a>
-<span class="sourceLineNo">479</span> private static void printBenchmarkResult(int totalSize,<a name="line.479"></a>
-<span class="sourceLineNo">480</span> List<Long> durationsInNanoSec, Manipulation manipulation) {<a name="line.480"></a>
-<span class="sourceLineNo">481</span> final int n = durationsInNanoSec.size();<a name="line.481"></a>
-<span class="sourceLineNo">482</span> long meanTime = 0;<a name="line.482"></a>
-<span class="sourceLineNo">483</span> for (long time : durationsInNanoSec) {<a name="line.483"></a>
-<span class="sourceLineNo">484</span> meanTime += time;<a name="line.484"></a>
-<span class="sourceLineNo">485</span> }<a name="line.485"></a>
-<span class="sourceLineNo">486</span> meanTime /= n;<a name="line.486"></a>
-<span class="sourceLineNo">487</span><a name="line.487"></a>
-<span class="sourceLineNo">488</span> double meanMBPerSec = totalSize * MB_SEC_COEF / meanTime;<a name="line.488"></a>
-<span class="sourceLineNo">489</span> double mbPerSecSTD = 0;<a name="line.489"></a>
-<span class="sourceLineNo">490</span> if (n > 0) {<a name="line.490"></a>
-<span class="sourceLineNo">491</span> for (long time : durationsInNanoSec) {<a name="line.491"></a>
-<span class="sourceLineNo">492</span> double mbPerSec = totalSize * MB_SEC_COEF / time;<a name="line.492"></a>
-
<TRUNCATED>