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 2018/09/22 14:51:18 UTC
[15/43] hbase-site git commit: Published site at
7ab77518a2569e2416a50020393aa386e7734501.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/bbe517ae/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterChunk.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterChunk.html b/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterChunk.html
index 3464580..b1a17d7 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterChunk.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterChunk.html
@@ -195,12 +195,12 @@
<span class="sourceLineNo">187</span> int hash1;<a name="line.187"></a>
<span class="sourceLineNo">188</span> int hash2;<a name="line.188"></a>
<span class="sourceLineNo">189</span> HashKey<Cell> hashKey;<a name="line.189"></a>
-<span class="sourceLineNo">190</span> if (this.bloomType == BloomType.ROW) {<a name="line.190"></a>
-<span class="sourceLineNo">191</span> hashKey = new RowBloomHashKey(cell);<a name="line.191"></a>
+<span class="sourceLineNo">190</span> if (this.bloomType == BloomType.ROWCOL) {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> hashKey = new RowColBloomHashKey(cell);<a name="line.191"></a>
<span class="sourceLineNo">192</span> hash1 = this.hash.hash(hashKey, 0);<a name="line.192"></a>
<span class="sourceLineNo">193</span> hash2 = this.hash.hash(hashKey, hash1);<a name="line.193"></a>
<span class="sourceLineNo">194</span> } else {<a name="line.194"></a>
-<span class="sourceLineNo">195</span> hashKey = new RowColBloomHashKey(cell);<a name="line.195"></a>
+<span class="sourceLineNo">195</span> hashKey = new RowBloomHashKey(cell);<a name="line.195"></a>
<span class="sourceLineNo">196</span> hash1 = this.hash.hash(hashKey, 0);<a name="line.196"></a>
<span class="sourceLineNo">197</span> hash2 = this.hash.hash(hashKey, hash1);<a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/bbe517ae/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterUtil.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterUtil.html b/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterUtil.html
index ae6ba83..4f3415b 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterUtil.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/util/BloomFilterUtil.html
@@ -28,271 +28,317 @@
<span class="sourceLineNo">020</span>import java.text.NumberFormat;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.util.Random;<a name="line.21"></a>
<span class="sourceLineNo">022</span><a name="line.22"></a>
-<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.Cell;<a name="line.23"></a>
-<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.nio.ByteBuff;<a name="line.24"></a>
-<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.25"></a>
-<span class="sourceLineNo">026</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.26"></a>
-<span class="sourceLineNo">027</span><a name="line.27"></a>
-<span class="sourceLineNo">028</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.28"></a>
+<span class="sourceLineNo">023</span>import org.apache.hadoop.conf.Configuration;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.Cell;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.HConstants;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.nio.ByteBuff;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.27"></a>
+<span class="sourceLineNo">028</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.28"></a>
<span class="sourceLineNo">029</span><a name="line.29"></a>
-<span class="sourceLineNo">030</span>/**<a name="line.30"></a>
-<span class="sourceLineNo">031</span> * Utility methods related to BloomFilters<a name="line.31"></a>
-<span class="sourceLineNo">032</span> */<a name="line.32"></a>
-<span class="sourceLineNo">033</span>@InterfaceAudience.Private<a name="line.33"></a>
-<span class="sourceLineNo">034</span>public final class BloomFilterUtil {<a name="line.34"></a>
-<span class="sourceLineNo">035</span><a name="line.35"></a>
-<span class="sourceLineNo">036</span> /** Record separator for the Bloom filter statistics human-readable string */<a name="line.36"></a>
-<span class="sourceLineNo">037</span> public static final String STATS_RECORD_SEP = "; ";<a name="line.37"></a>
-<span class="sourceLineNo">038</span> /**<a name="line.38"></a>
-<span class="sourceLineNo">039</span> * Used in computing the optimal Bloom filter size. This approximately equals<a name="line.39"></a>
-<span class="sourceLineNo">040</span> * 0.480453.<a name="line.40"></a>
-<span class="sourceLineNo">041</span> */<a name="line.41"></a>
-<span class="sourceLineNo">042</span> public static final double LOG2_SQUARED = Math.log(2) * Math.log(2);<a name="line.42"></a>
-<span class="sourceLineNo">043</span> <a name="line.43"></a>
-<span class="sourceLineNo">044</span> /**<a name="line.44"></a>
-<span class="sourceLineNo">045</span> * A random number generator to use for "fake lookups" when testing to<a name="line.45"></a>
-<span class="sourceLineNo">046</span> * estimate the ideal false positive rate.<a name="line.46"></a>
-<span class="sourceLineNo">047</span> */<a name="line.47"></a>
-<span class="sourceLineNo">048</span> private static Random randomGeneratorForTest;<a name="line.48"></a>
-<span class="sourceLineNo">049</span> <a name="line.49"></a>
-<span class="sourceLineNo">050</span> /** Bit-value lookup array to prevent doing the same work over and over */<a name="line.50"></a>
-<span class="sourceLineNo">051</span> public static final byte [] bitvals = {<a name="line.51"></a>
-<span class="sourceLineNo">052</span> (byte) 0x01,<a name="line.52"></a>
-<span class="sourceLineNo">053</span> (byte) 0x02,<a name="line.53"></a>
-<span class="sourceLineNo">054</span> (byte) 0x04,<a name="line.54"></a>
-<span class="sourceLineNo">055</span> (byte) 0x08,<a name="line.55"></a>
-<span class="sourceLineNo">056</span> (byte) 0x10,<a name="line.56"></a>
-<span class="sourceLineNo">057</span> (byte) 0x20,<a name="line.57"></a>
-<span class="sourceLineNo">058</span> (byte) 0x40,<a name="line.58"></a>
-<span class="sourceLineNo">059</span> (byte) 0x80<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> /**<a name="line.62"></a>
-<span class="sourceLineNo">063</span> * Private constructor to keep this class from being instantiated.<a name="line.63"></a>
-<span class="sourceLineNo">064</span> */<a name="line.64"></a>
-<span class="sourceLineNo">065</span> private BloomFilterUtil() {<a name="line.65"></a>
-<span class="sourceLineNo">066</span> }<a name="line.66"></a>
-<span class="sourceLineNo">067</span><a name="line.67"></a>
-<span class="sourceLineNo">068</span> /**<a name="line.68"></a>
-<span class="sourceLineNo">069</span> * @param maxKeys<a name="line.69"></a>
-<span class="sourceLineNo">070</span> * @param errorRate<a name="line.70"></a>
-<span class="sourceLineNo">071</span> * @return the number of bits for a Bloom filter than can hold the given<a name="line.71"></a>
-<span class="sourceLineNo">072</span> * number of keys and provide the given error rate, assuming that the<a name="line.72"></a>
-<span class="sourceLineNo">073</span> * optimal number of hash functions is used and it does not have to<a name="line.73"></a>
-<span class="sourceLineNo">074</span> * be an integer.<a name="line.74"></a>
-<span class="sourceLineNo">075</span> */<a name="line.75"></a>
-<span class="sourceLineNo">076</span> public static long computeBitSize(long maxKeys, double errorRate) {<a name="line.76"></a>
-<span class="sourceLineNo">077</span> return (long) Math.ceil(maxKeys * (-Math.log(errorRate) / LOG2_SQUARED));<a name="line.77"></a>
-<span class="sourceLineNo">078</span> }<a name="line.78"></a>
-<span class="sourceLineNo">079</span><a name="line.79"></a>
-<span class="sourceLineNo">080</span> /**<a name="line.80"></a>
-<span class="sourceLineNo">081</span> * Sets a random generator to be used for look-ups instead of computing hashes. Can be used to<a name="line.81"></a>
-<span class="sourceLineNo">082</span> * simulate uniformity of accesses better in a test environment. Should not be set in a real<a name="line.82"></a>
-<span class="sourceLineNo">083</span> * environment where correctness matters!<a name="line.83"></a>
-<span class="sourceLineNo">084</span> * <p><a name="line.84"></a>
-<span class="sourceLineNo">085</span> * This gets used in {@link #contains(ByteBuff, int, int, Hash, int, HashKey)}<a name="line.85"></a>
-<span class="sourceLineNo">086</span> * @param random The random number source to use, or null to compute actual hashes<a name="line.86"></a>
-<span class="sourceLineNo">087</span> */<a name="line.87"></a>
-<span class="sourceLineNo">088</span> @VisibleForTesting<a name="line.88"></a>
-<span class="sourceLineNo">089</span> public static void setRandomGeneratorForTest(Random random) {<a name="line.89"></a>
-<span class="sourceLineNo">090</span> randomGeneratorForTest = random;<a name="line.90"></a>
-<span class="sourceLineNo">091</span> }<a name="line.91"></a>
-<span class="sourceLineNo">092</span><a name="line.92"></a>
-<span class="sourceLineNo">093</span> /**<a name="line.93"></a>
-<span class="sourceLineNo">094</span> * The maximum number of keys we can put into a Bloom filter of a certain<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * size to maintain the given error rate, assuming the number of hash<a name="line.95"></a>
-<span class="sourceLineNo">096</span> * functions is chosen optimally and does not even have to be an integer<a name="line.96"></a>
-<span class="sourceLineNo">097</span> * (hence the "ideal" in the function name).<a name="line.97"></a>
-<span class="sourceLineNo">098</span> *<a name="line.98"></a>
-<span class="sourceLineNo">099</span> * @param bitSize<a name="line.99"></a>
-<span class="sourceLineNo">100</span> * @param errorRate<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * @return maximum number of keys that can be inserted into the Bloom filter<a name="line.101"></a>
-<span class="sourceLineNo">102</span> * @see #computeMaxKeys(long, double, int) for a more precise estimate<a name="line.102"></a>
-<span class="sourceLineNo">103</span> */<a name="line.103"></a>
-<span class="sourceLineNo">104</span> public static long idealMaxKeys(long bitSize, double errorRate) {<a name="line.104"></a>
-<span class="sourceLineNo">105</span> // The reason we need to use floor here is that otherwise we might put<a name="line.105"></a>
-<span class="sourceLineNo">106</span> // more keys in a Bloom filter than is allowed by the target error rate.<a name="line.106"></a>
-<span class="sourceLineNo">107</span> return (long) (bitSize * (LOG2_SQUARED / -Math.log(errorRate)));<a name="line.107"></a>
-<span class="sourceLineNo">108</span> }<a name="line.108"></a>
-<span class="sourceLineNo">109</span><a name="line.109"></a>
-<span class="sourceLineNo">110</span> /**<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * The maximum number of keys we can put into a Bloom filter of a certain<a name="line.111"></a>
-<span class="sourceLineNo">112</span> * size to get the given error rate, with the given number of hash functions.<a name="line.112"></a>
-<span class="sourceLineNo">113</span> *<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * @param bitSize<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * @param errorRate<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * @param hashCount<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * @return the maximum number of keys that can be inserted in a Bloom filter<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * to maintain the target error rate, if the number of hash functions<a name="line.118"></a>
-<span class="sourceLineNo">119</span> * is provided.<a name="line.119"></a>
-<span class="sourceLineNo">120</span> */<a name="line.120"></a>
-<span class="sourceLineNo">121</span> public static long computeMaxKeys(long bitSize, double errorRate,<a name="line.121"></a>
-<span class="sourceLineNo">122</span> int hashCount) {<a name="line.122"></a>
-<span class="sourceLineNo">123</span> return (long) (-bitSize * 1.0 / hashCount *<a name="line.123"></a>
-<span class="sourceLineNo">124</span> Math.log(1 - Math.exp(Math.log(errorRate) / hashCount)));<a name="line.124"></a>
-<span class="sourceLineNo">125</span> }<a name="line.125"></a>
-<span class="sourceLineNo">126</span><a name="line.126"></a>
-<span class="sourceLineNo">127</span> /**<a name="line.127"></a>
-<span class="sourceLineNo">128</span> * Computes the actual error rate for the given number of elements, number<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * of bits, and number of hash functions. Taken directly from the<a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <a href=<a name="line.130"></a>
-<span class="sourceLineNo">131</span> * "http://en.wikipedia.org/wiki/Bloom_filter#Probability_of_false_positives"<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * > Wikipedia Bloom filter article</a>.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> *<a name="line.133"></a>
-<span class="sourceLineNo">134</span> * @param maxKeys<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * @param bitSize<a name="line.135"></a>
-<span class="sourceLineNo">136</span> * @param functionCount<a name="line.136"></a>
-<span class="sourceLineNo">137</span> * @return the actual error rate<a name="line.137"></a>
-<span class="sourceLineNo">138</span> */<a name="line.138"></a>
-<span class="sourceLineNo">139</span> public static double actualErrorRate(long maxKeys, long bitSize,<a name="line.139"></a>
-<span class="sourceLineNo">140</span> int functionCount) {<a name="line.140"></a>
-<span class="sourceLineNo">141</span> return Math.exp(Math.log(1 - Math.exp(-functionCount * maxKeys * 1.0<a name="line.141"></a>
-<span class="sourceLineNo">142</span> / bitSize)) * functionCount);<a name="line.142"></a>
-<span class="sourceLineNo">143</span> }<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> /**<a name="line.145"></a>
-<span class="sourceLineNo">146</span> * Increases the given byte size of a Bloom filter until it can be folded by<a name="line.146"></a>
-<span class="sourceLineNo">147</span> * the given factor.<a name="line.147"></a>
-<span class="sourceLineNo">148</span> *<a name="line.148"></a>
-<span class="sourceLineNo">149</span> * @param bitSize<a name="line.149"></a>
-<span class="sourceLineNo">150</span> * @param foldFactor<a name="line.150"></a>
-<span class="sourceLineNo">151</span> * @return Foldable byte size<a name="line.151"></a>
-<span class="sourceLineNo">152</span> */<a name="line.152"></a>
-<span class="sourceLineNo">153</span> public static int computeFoldableByteSize(long bitSize, int foldFactor) {<a name="line.153"></a>
-<span class="sourceLineNo">154</span> long byteSizeLong = (bitSize + 7) / 8;<a name="line.154"></a>
-<span class="sourceLineNo">155</span> int mask = (1 << foldFactor) - 1;<a name="line.155"></a>
-<span class="sourceLineNo">156</span> if ((mask & byteSizeLong) != 0) {<a name="line.156"></a>
-<span class="sourceLineNo">157</span> byteSizeLong >>= foldFactor;<a name="line.157"></a>
-<span class="sourceLineNo">158</span> ++byteSizeLong;<a name="line.158"></a>
-<span class="sourceLineNo">159</span> byteSizeLong <<= foldFactor;<a name="line.159"></a>
-<span class="sourceLineNo">160</span> }<a name="line.160"></a>
-<span class="sourceLineNo">161</span> if (byteSizeLong > Integer.MAX_VALUE) {<a name="line.161"></a>
-<span class="sourceLineNo">162</span> throw new IllegalArgumentException("byteSize=" + byteSizeLong + " too "<a name="line.162"></a>
-<span class="sourceLineNo">163</span> + "large for bitSize=" + bitSize + ", foldFactor=" + foldFactor);<a name="line.163"></a>
-<span class="sourceLineNo">164</span> }<a name="line.164"></a>
-<span class="sourceLineNo">165</span> return (int) byteSizeLong;<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> public static int optimalFunctionCount(int maxKeys, long bitSize) {<a name="line.168"></a>
-<span class="sourceLineNo">169</span> long i = bitSize / maxKeys;<a name="line.169"></a>
-<span class="sourceLineNo">170</span> double result = Math.ceil(Math.log(2) * i);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> if (result > Integer.MAX_VALUE){<a name="line.171"></a>
-<span class="sourceLineNo">172</span> throw new IllegalArgumentException("result too large for integer value.");<a name="line.172"></a>
-<span class="sourceLineNo">173</span> }<a name="line.173"></a>
-<span class="sourceLineNo">174</span> return (int)result;<a name="line.174"></a>
-<span class="sourceLineNo">175</span> }<a name="line.175"></a>
-<span class="sourceLineNo">176</span> <a name="line.176"></a>
-<span class="sourceLineNo">177</span> /**<a name="line.177"></a>
-<span class="sourceLineNo">178</span> * Creates a Bloom filter chunk of the given size.<a name="line.178"></a>
-<span class="sourceLineNo">179</span> *<a name="line.179"></a>
-<span class="sourceLineNo">180</span> * @param byteSizeHint the desired number of bytes for the Bloom filter bit<a name="line.180"></a>
-<span class="sourceLineNo">181</span> * array. Will be increased so that folding is possible.<a name="line.181"></a>
-<span class="sourceLineNo">182</span> * @param errorRate target false positive rate of the Bloom filter<a name="line.182"></a>
-<span class="sourceLineNo">183</span> * @param hashType Bloom filter hash function type<a name="line.183"></a>
-<span class="sourceLineNo">184</span> * @param foldFactor<a name="line.184"></a>
-<span class="sourceLineNo">185</span> * @param bloomType<a name="line.185"></a>
-<span class="sourceLineNo">186</span> * @return the new Bloom filter of the desired size<a name="line.186"></a>
-<span class="sourceLineNo">187</span> */<a name="line.187"></a>
-<span class="sourceLineNo">188</span> public static BloomFilterChunk createBySize(int byteSizeHint,<a name="line.188"></a>
-<span class="sourceLineNo">189</span> double errorRate, int hashType, int foldFactor, BloomType bloomType) {<a name="line.189"></a>
-<span class="sourceLineNo">190</span> BloomFilterChunk bbf = new BloomFilterChunk(hashType, bloomType);<a name="line.190"></a>
-<span class="sourceLineNo">191</span><a name="line.191"></a>
-<span class="sourceLineNo">192</span> bbf.byteSize = computeFoldableByteSize(byteSizeHint * 8L, foldFactor);<a name="line.192"></a>
-<span class="sourceLineNo">193</span> long bitSize = bbf.byteSize * 8;<a name="line.193"></a>
-<span class="sourceLineNo">194</span> bbf.maxKeys = (int) idealMaxKeys(bitSize, errorRate);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> bbf.hashCount = optimalFunctionCount(bbf.maxKeys, bitSize);<a name="line.195"></a>
+<span class="sourceLineNo">030</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.30"></a>
+<span class="sourceLineNo">031</span><a name="line.31"></a>
+<span class="sourceLineNo">032</span>/**<a name="line.32"></a>
+<span class="sourceLineNo">033</span> * Utility methods related to BloomFilters<a name="line.33"></a>
+<span class="sourceLineNo">034</span> */<a name="line.34"></a>
+<span class="sourceLineNo">035</span>@InterfaceAudience.Private<a name="line.35"></a>
+<span class="sourceLineNo">036</span>public final class BloomFilterUtil {<a name="line.36"></a>
+<span class="sourceLineNo">037</span><a name="line.37"></a>
+<span class="sourceLineNo">038</span> /** Record separator for the Bloom filter statistics human-readable string */<a name="line.38"></a>
+<span class="sourceLineNo">039</span> public static final String STATS_RECORD_SEP = "; ";<a name="line.39"></a>
+<span class="sourceLineNo">040</span> /**<a name="line.40"></a>
+<span class="sourceLineNo">041</span> * Used in computing the optimal Bloom filter size. This approximately equals<a name="line.41"></a>
+<span class="sourceLineNo">042</span> * 0.480453.<a name="line.42"></a>
+<span class="sourceLineNo">043</span> */<a name="line.43"></a>
+<span class="sourceLineNo">044</span> public static final double LOG2_SQUARED = Math.log(2) * Math.log(2);<a name="line.44"></a>
+<span class="sourceLineNo">045</span> <a name="line.45"></a>
+<span class="sourceLineNo">046</span> /**<a name="line.46"></a>
+<span class="sourceLineNo">047</span> * A random number generator to use for "fake lookups" when testing to<a name="line.47"></a>
+<span class="sourceLineNo">048</span> * estimate the ideal false positive rate.<a name="line.48"></a>
+<span class="sourceLineNo">049</span> */<a name="line.49"></a>
+<span class="sourceLineNo">050</span> private static Random randomGeneratorForTest;<a name="line.50"></a>
+<span class="sourceLineNo">051</span><a name="line.51"></a>
+<span class="sourceLineNo">052</span> public static final String PREFIX_LENGTH_KEY = "RowPrefixBloomFilter.prefix_length";<a name="line.52"></a>
+<span class="sourceLineNo">053</span> public static final String DELIMITER_KEY = "RowPrefixDelimitedBloomFilter.delimiter";<a name="line.53"></a>
+<span class="sourceLineNo">054</span> <a name="line.54"></a>
+<span class="sourceLineNo">055</span> /** Bit-value lookup array to prevent doing the same work over and over */<a name="line.55"></a>
+<span class="sourceLineNo">056</span> public static final byte [] bitvals = {<a name="line.56"></a>
+<span class="sourceLineNo">057</span> (byte) 0x01,<a name="line.57"></a>
+<span class="sourceLineNo">058</span> (byte) 0x02,<a name="line.58"></a>
+<span class="sourceLineNo">059</span> (byte) 0x04,<a name="line.59"></a>
+<span class="sourceLineNo">060</span> (byte) 0x08,<a name="line.60"></a>
+<span class="sourceLineNo">061</span> (byte) 0x10,<a name="line.61"></a>
+<span class="sourceLineNo">062</span> (byte) 0x20,<a name="line.62"></a>
+<span class="sourceLineNo">063</span> (byte) 0x40,<a name="line.63"></a>
+<span class="sourceLineNo">064</span> (byte) 0x80<a name="line.64"></a>
+<span class="sourceLineNo">065</span> };<a name="line.65"></a>
+<span class="sourceLineNo">066</span><a name="line.66"></a>
+<span class="sourceLineNo">067</span> /**<a name="line.67"></a>
+<span class="sourceLineNo">068</span> * Private constructor to keep this class from being instantiated.<a name="line.68"></a>
+<span class="sourceLineNo">069</span> */<a name="line.69"></a>
+<span class="sourceLineNo">070</span> private BloomFilterUtil() {<a name="line.70"></a>
+<span class="sourceLineNo">071</span> }<a name="line.71"></a>
+<span class="sourceLineNo">072</span><a name="line.72"></a>
+<span class="sourceLineNo">073</span> /**<a name="line.73"></a>
+<span class="sourceLineNo">074</span> * @param maxKeys<a name="line.74"></a>
+<span class="sourceLineNo">075</span> * @param errorRate<a name="line.75"></a>
+<span class="sourceLineNo">076</span> * @return the number of bits for a Bloom filter than can hold the given<a name="line.76"></a>
+<span class="sourceLineNo">077</span> * number of keys and provide the given error rate, assuming that the<a name="line.77"></a>
+<span class="sourceLineNo">078</span> * optimal number of hash functions is used and it does not have to<a name="line.78"></a>
+<span class="sourceLineNo">079</span> * be an integer.<a name="line.79"></a>
+<span class="sourceLineNo">080</span> */<a name="line.80"></a>
+<span class="sourceLineNo">081</span> public static long computeBitSize(long maxKeys, double errorRate) {<a name="line.81"></a>
+<span class="sourceLineNo">082</span> return (long) Math.ceil(maxKeys * (-Math.log(errorRate) / LOG2_SQUARED));<a name="line.82"></a>
+<span class="sourceLineNo">083</span> }<a name="line.83"></a>
+<span class="sourceLineNo">084</span><a name="line.84"></a>
+<span class="sourceLineNo">085</span> /**<a name="line.85"></a>
+<span class="sourceLineNo">086</span> * Sets a random generator to be used for look-ups instead of computing hashes. Can be used to<a name="line.86"></a>
+<span class="sourceLineNo">087</span> * simulate uniformity of accesses better in a test environment. Should not be set in a real<a name="line.87"></a>
+<span class="sourceLineNo">088</span> * environment where correctness matters!<a name="line.88"></a>
+<span class="sourceLineNo">089</span> * <p><a name="line.89"></a>
+<span class="sourceLineNo">090</span> * This gets used in {@link #contains(ByteBuff, int, int, Hash, int, HashKey)}<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * @param random The random number source to use, or null to compute actual hashes<a name="line.91"></a>
+<span class="sourceLineNo">092</span> */<a name="line.92"></a>
+<span class="sourceLineNo">093</span> @VisibleForTesting<a name="line.93"></a>
+<span class="sourceLineNo">094</span> public static void setRandomGeneratorForTest(Random random) {<a name="line.94"></a>
+<span class="sourceLineNo">095</span> randomGeneratorForTest = random;<a name="line.95"></a>
+<span class="sourceLineNo">096</span> }<a name="line.96"></a>
+<span class="sourceLineNo">097</span><a name="line.97"></a>
+<span class="sourceLineNo">098</span> /**<a name="line.98"></a>
+<span class="sourceLineNo">099</span> * The maximum number of keys we can put into a Bloom filter of a certain<a name="line.99"></a>
+<span class="sourceLineNo">100</span> * size to maintain the given error rate, assuming the number of hash<a name="line.100"></a>
+<span class="sourceLineNo">101</span> * functions is chosen optimally and does not even have to be an integer<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * (hence the "ideal" in the function name).<a name="line.102"></a>
+<span class="sourceLineNo">103</span> *<a name="line.103"></a>
+<span class="sourceLineNo">104</span> * @param bitSize<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * @param errorRate<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * @return maximum number of keys that can be inserted into the Bloom filter<a name="line.106"></a>
+<span class="sourceLineNo">107</span> * @see #computeMaxKeys(long, double, int) for a more precise estimate<a name="line.107"></a>
+<span class="sourceLineNo">108</span> */<a name="line.108"></a>
+<span class="sourceLineNo">109</span> public static long idealMaxKeys(long bitSize, double errorRate) {<a name="line.109"></a>
+<span class="sourceLineNo">110</span> // The reason we need to use floor here is that otherwise we might put<a name="line.110"></a>
+<span class="sourceLineNo">111</span> // more keys in a Bloom filter than is allowed by the target error rate.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> return (long) (bitSize * (LOG2_SQUARED / -Math.log(errorRate)));<a name="line.112"></a>
+<span class="sourceLineNo">113</span> }<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> * The maximum number of keys we can put into a Bloom filter of a certain<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * size to get the given error rate, with the given number of hash functions.<a name="line.117"></a>
+<span class="sourceLineNo">118</span> *<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * @param bitSize<a name="line.119"></a>
+<span class="sourceLineNo">120</span> * @param errorRate<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * @param hashCount<a name="line.121"></a>
+<span class="sourceLineNo">122</span> * @return the maximum number of keys that can be inserted in a Bloom filter<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * to maintain the target error rate, if the number of hash functions<a name="line.123"></a>
+<span class="sourceLineNo">124</span> * is provided.<a name="line.124"></a>
+<span class="sourceLineNo">125</span> */<a name="line.125"></a>
+<span class="sourceLineNo">126</span> public static long computeMaxKeys(long bitSize, double errorRate,<a name="line.126"></a>
+<span class="sourceLineNo">127</span> int hashCount) {<a name="line.127"></a>
+<span class="sourceLineNo">128</span> return (long) (-bitSize * 1.0 / hashCount *<a name="line.128"></a>
+<span class="sourceLineNo">129</span> Math.log(1 - Math.exp(Math.log(errorRate) / hashCount)));<a name="line.129"></a>
+<span class="sourceLineNo">130</span> }<a name="line.130"></a>
+<span class="sourceLineNo">131</span><a name="line.131"></a>
+<span class="sourceLineNo">132</span> /**<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * Computes the actual error rate for the given number of elements, number<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * of bits, and number of hash functions. Taken directly from the<a name="line.134"></a>
+<span class="sourceLineNo">135</span> * <a href=<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * "http://en.wikipedia.org/wiki/Bloom_filter#Probability_of_false_positives"<a name="line.136"></a>
+<span class="sourceLineNo">137</span> * > Wikipedia Bloom filter article</a>.<a name="line.137"></a>
+<span class="sourceLineNo">138</span> *<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * @param maxKeys<a name="line.139"></a>
+<span class="sourceLineNo">140</span> * @param bitSize<a name="line.140"></a>
+<span class="sourceLineNo">141</span> * @param functionCount<a name="line.141"></a>
+<span class="sourceLineNo">142</span> * @return the actual error rate<a name="line.142"></a>
+<span class="sourceLineNo">143</span> */<a name="line.143"></a>
+<span class="sourceLineNo">144</span> public static double actualErrorRate(long maxKeys, long bitSize,<a name="line.144"></a>
+<span class="sourceLineNo">145</span> int functionCount) {<a name="line.145"></a>
+<span class="sourceLineNo">146</span> return Math.exp(Math.log(1 - Math.exp(-functionCount * maxKeys * 1.0<a name="line.146"></a>
+<span class="sourceLineNo">147</span> / bitSize)) * functionCount);<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> * Increases the given byte size of a Bloom filter until it can be folded by<a name="line.151"></a>
+<span class="sourceLineNo">152</span> * the given factor.<a name="line.152"></a>
+<span class="sourceLineNo">153</span> *<a name="line.153"></a>
+<span class="sourceLineNo">154</span> * @param bitSize<a name="line.154"></a>
+<span class="sourceLineNo">155</span> * @param foldFactor<a name="line.155"></a>
+<span class="sourceLineNo">156</span> * @return Foldable byte size<a name="line.156"></a>
+<span class="sourceLineNo">157</span> */<a name="line.157"></a>
+<span class="sourceLineNo">158</span> public static int computeFoldableByteSize(long bitSize, int foldFactor) {<a name="line.158"></a>
+<span class="sourceLineNo">159</span> long byteSizeLong = (bitSize + 7) / 8;<a name="line.159"></a>
+<span class="sourceLineNo">160</span> int mask = (1 << foldFactor) - 1;<a name="line.160"></a>
+<span class="sourceLineNo">161</span> if ((mask & byteSizeLong) != 0) {<a name="line.161"></a>
+<span class="sourceLineNo">162</span> byteSizeLong >>= foldFactor;<a name="line.162"></a>
+<span class="sourceLineNo">163</span> ++byteSizeLong;<a name="line.163"></a>
+<span class="sourceLineNo">164</span> byteSizeLong <<= foldFactor;<a name="line.164"></a>
+<span class="sourceLineNo">165</span> }<a name="line.165"></a>
+<span class="sourceLineNo">166</span> if (byteSizeLong > Integer.MAX_VALUE) {<a name="line.166"></a>
+<span class="sourceLineNo">167</span> throw new IllegalArgumentException("byteSize=" + byteSizeLong + " too "<a name="line.167"></a>
+<span class="sourceLineNo">168</span> + "large for bitSize=" + bitSize + ", foldFactor=" + foldFactor);<a name="line.168"></a>
+<span class="sourceLineNo">169</span> }<a name="line.169"></a>
+<span class="sourceLineNo">170</span> return (int) byteSizeLong;<a name="line.170"></a>
+<span class="sourceLineNo">171</span> }<a name="line.171"></a>
+<span class="sourceLineNo">172</span><a name="line.172"></a>
+<span class="sourceLineNo">173</span> public static int optimalFunctionCount(int maxKeys, long bitSize) {<a name="line.173"></a>
+<span class="sourceLineNo">174</span> long i = bitSize / maxKeys;<a name="line.174"></a>
+<span class="sourceLineNo">175</span> double result = Math.ceil(Math.log(2) * i);<a name="line.175"></a>
+<span class="sourceLineNo">176</span> if (result > Integer.MAX_VALUE){<a name="line.176"></a>
+<span class="sourceLineNo">177</span> throw new IllegalArgumentException("result too large for integer value.");<a name="line.177"></a>
+<span class="sourceLineNo">178</span> }<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return (int)result;<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span> <a name="line.181"></a>
+<span class="sourceLineNo">182</span> /**<a name="line.182"></a>
+<span class="sourceLineNo">183</span> * Creates a Bloom filter chunk of the given size.<a name="line.183"></a>
+<span class="sourceLineNo">184</span> *<a name="line.184"></a>
+<span class="sourceLineNo">185</span> * @param byteSizeHint the desired number of bytes for the Bloom filter bit<a name="line.185"></a>
+<span class="sourceLineNo">186</span> * array. Will be increased so that folding is possible.<a name="line.186"></a>
+<span class="sourceLineNo">187</span> * @param errorRate target false positive rate of the Bloom filter<a name="line.187"></a>
+<span class="sourceLineNo">188</span> * @param hashType Bloom filter hash function type<a name="line.188"></a>
+<span class="sourceLineNo">189</span> * @param foldFactor<a name="line.189"></a>
+<span class="sourceLineNo">190</span> * @param bloomType<a name="line.190"></a>
+<span class="sourceLineNo">191</span> * @return the new Bloom filter of the desired size<a name="line.191"></a>
+<span class="sourceLineNo">192</span> */<a name="line.192"></a>
+<span class="sourceLineNo">193</span> public static BloomFilterChunk createBySize(int byteSizeHint,<a name="line.193"></a>
+<span class="sourceLineNo">194</span> double errorRate, int hashType, int foldFactor, BloomType bloomType) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> BloomFilterChunk bbf = new BloomFilterChunk(hashType, bloomType);<a name="line.195"></a>
<span class="sourceLineNo">196</span><a name="line.196"></a>
-<span class="sourceLineNo">197</span> // Adjust max keys to bring error rate closer to what was requested,<a name="line.197"></a>
-<span class="sourceLineNo">198</span> // because byteSize was adjusted to allow for folding, and hashCount was<a name="line.198"></a>
-<span class="sourceLineNo">199</span> // rounded.<a name="line.199"></a>
-<span class="sourceLineNo">200</span> bbf.maxKeys = (int) computeMaxKeys(bitSize, errorRate, bbf.hashCount);<a name="line.200"></a>
+<span class="sourceLineNo">197</span> bbf.byteSize = computeFoldableByteSize(byteSizeHint * 8L, foldFactor);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> long bitSize = bbf.byteSize * 8;<a name="line.198"></a>
+<span class="sourceLineNo">199</span> bbf.maxKeys = (int) idealMaxKeys(bitSize, errorRate);<a name="line.199"></a>
+<span class="sourceLineNo">200</span> bbf.hashCount = optimalFunctionCount(bbf.maxKeys, bitSize);<a name="line.200"></a>
<span class="sourceLineNo">201</span><a name="line.201"></a>
-<span class="sourceLineNo">202</span> return bbf;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span><a name="line.204"></a>
-<span class="sourceLineNo">205</span> public static boolean contains(byte[] buf, int offset, int length,<a name="line.205"></a>
-<span class="sourceLineNo">206</span> ByteBuff bloomBuf, int bloomOffset, int bloomSize, Hash hash,<a name="line.206"></a>
-<span class="sourceLineNo">207</span> int hashCount) {<a name="line.207"></a>
-<span class="sourceLineNo">208</span> HashKey<byte[]> hashKey = new ByteArrayHashKey(buf, offset, length);<a name="line.208"></a>
-<span class="sourceLineNo">209</span> return contains(bloomBuf, bloomOffset, bloomSize, hash, hashCount, hashKey);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> }<a name="line.210"></a>
-<span class="sourceLineNo">211</span><a name="line.211"></a>
-<span class="sourceLineNo">212</span> private static <T> boolean contains(ByteBuff bloomBuf, int bloomOffset, int bloomSize, Hash hash,<a name="line.212"></a>
-<span class="sourceLineNo">213</span> int hashCount, HashKey<T> hashKey) {<a name="line.213"></a>
-<span class="sourceLineNo">214</span> int hash1 = hash.hash(hashKey, 0);<a name="line.214"></a>
-<span class="sourceLineNo">215</span> int bloomBitSize = bloomSize << 3;<a name="line.215"></a>
+<span class="sourceLineNo">202</span> // Adjust max keys to bring error rate closer to what was requested,<a name="line.202"></a>
+<span class="sourceLineNo">203</span> // because byteSize was adjusted to allow for folding, and hashCount was<a name="line.203"></a>
+<span class="sourceLineNo">204</span> // rounded.<a name="line.204"></a>
+<span class="sourceLineNo">205</span> bbf.maxKeys = (int) computeMaxKeys(bitSize, errorRate, bbf.hashCount);<a name="line.205"></a>
+<span class="sourceLineNo">206</span><a name="line.206"></a>
+<span class="sourceLineNo">207</span> return bbf;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> public static boolean contains(byte[] buf, int offset, int length,<a name="line.210"></a>
+<span class="sourceLineNo">211</span> ByteBuff bloomBuf, int bloomOffset, int bloomSize, Hash hash,<a name="line.211"></a>
+<span class="sourceLineNo">212</span> int hashCount) {<a name="line.212"></a>
+<span class="sourceLineNo">213</span> HashKey<byte[]> hashKey = new ByteArrayHashKey(buf, offset, length);<a name="line.213"></a>
+<span class="sourceLineNo">214</span> return contains(bloomBuf, bloomOffset, bloomSize, hash, hashCount, hashKey);<a name="line.214"></a>
+<span class="sourceLineNo">215</span> }<a name="line.215"></a>
<span class="sourceLineNo">216</span><a name="line.216"></a>
-<span class="sourceLineNo">217</span> int hash2 = 0;<a name="line.217"></a>
-<span class="sourceLineNo">218</span> int compositeHash = 0;<a name="line.218"></a>
-<span class="sourceLineNo">219</span><a name="line.219"></a>
-<span class="sourceLineNo">220</span> if (randomGeneratorForTest == null) {<a name="line.220"></a>
-<span class="sourceLineNo">221</span> // Production mode<a name="line.221"></a>
-<span class="sourceLineNo">222</span> compositeHash = hash1;<a name="line.222"></a>
-<span class="sourceLineNo">223</span> hash2 = hash.hash(hashKey, hash1);<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> for (int i = 0; i < hashCount; i++) {<a name="line.226"></a>
-<span class="sourceLineNo">227</span> int hashLoc = (randomGeneratorForTest == null<a name="line.227"></a>
-<span class="sourceLineNo">228</span> // Production mode<a name="line.228"></a>
-<span class="sourceLineNo">229</span> ? Math.abs(compositeHash % bloomBitSize)<a name="line.229"></a>
-<span class="sourceLineNo">230</span> // Test mode with "fake look-ups" to estimate "ideal false positive rate"<a name="line.230"></a>
-<span class="sourceLineNo">231</span> : randomGeneratorForTest.nextInt(bloomBitSize));<a name="line.231"></a>
-<span class="sourceLineNo">232</span> compositeHash += hash2;<a name="line.232"></a>
-<span class="sourceLineNo">233</span> if (!checkBit(hashLoc, bloomBuf, bloomOffset)) {<a name="line.233"></a>
-<span class="sourceLineNo">234</span> return false;<a name="line.234"></a>
-<span class="sourceLineNo">235</span> }<a name="line.235"></a>
-<span class="sourceLineNo">236</span> }<a name="line.236"></a>
-<span class="sourceLineNo">237</span> return true;<a name="line.237"></a>
-<span class="sourceLineNo">238</span> }<a name="line.238"></a>
-<span class="sourceLineNo">239</span><a name="line.239"></a>
-<span class="sourceLineNo">240</span> public static boolean contains(Cell cell, ByteBuff bloomBuf, int bloomOffset, int bloomSize,<a name="line.240"></a>
-<span class="sourceLineNo">241</span> Hash hash, int hashCount, BloomType type) {<a name="line.241"></a>
-<span class="sourceLineNo">242</span> HashKey<Cell> hashKey = type == BloomType.ROW ? new RowBloomHashKey(cell)<a name="line.242"></a>
-<span class="sourceLineNo">243</span> : new RowColBloomHashKey(cell);<a name="line.243"></a>
-<span class="sourceLineNo">244</span> return contains(bloomBuf, bloomOffset, bloomSize, hash, hashCount, hashKey);<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> * Check if bit at specified index is 1.<a name="line.248"></a>
-<span class="sourceLineNo">249</span> *<a name="line.249"></a>
-<span class="sourceLineNo">250</span> * @param pos index of bit<a name="line.250"></a>
-<span class="sourceLineNo">251</span> * @return true if bit at specified index is 1, false if 0.<a name="line.251"></a>
-<span class="sourceLineNo">252</span> */<a name="line.252"></a>
-<span class="sourceLineNo">253</span> static boolean checkBit(int pos, ByteBuff bloomBuf, int bloomOffset) {<a name="line.253"></a>
-<span class="sourceLineNo">254</span> int bytePos = pos >> 3; //pos / 8<a name="line.254"></a>
-<span class="sourceLineNo">255</span> int bitPos = pos & 0x7; //pos % 8<a name="line.255"></a>
-<span class="sourceLineNo">256</span> byte curByte = bloomBuf.get(bloomOffset + bytePos);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> curByte &= bitvals[bitPos];<a name="line.257"></a>
-<span class="sourceLineNo">258</span> return (curByte != 0);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> /**<a name="line.261"></a>
-<span class="sourceLineNo">262</span> * A human-readable string with statistics for the given Bloom filter.<a name="line.262"></a>
-<span class="sourceLineNo">263</span> *<a name="line.263"></a>
-<span class="sourceLineNo">264</span> * @param bloomFilter the Bloom filter to output statistics for;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> * @return a string consisting of "&lt;key&gt;: &lt;value&gt;" parts<a name="line.265"></a>
-<span class="sourceLineNo">266</span> * separated by {@link #STATS_RECORD_SEP}.<a name="line.266"></a>
-<span class="sourceLineNo">267</span> */<a name="line.267"></a>
-<span class="sourceLineNo">268</span> public static String formatStats(BloomFilterBase bloomFilter) {<a name="line.268"></a>
-<span class="sourceLineNo">269</span> StringBuilder sb = new StringBuilder();<a name="line.269"></a>
-<span class="sourceLineNo">270</span> long k = bloomFilter.getKeyCount();<a name="line.270"></a>
-<span class="sourceLineNo">271</span> long m = bloomFilter.getMaxKeys();<a name="line.271"></a>
-<span class="sourceLineNo">272</span><a name="line.272"></a>
-<span class="sourceLineNo">273</span> sb.append("BloomSize: " + bloomFilter.getByteSize() + STATS_RECORD_SEP);<a name="line.273"></a>
-<span class="sourceLineNo">274</span> sb.append("No of Keys in bloom: " + k + STATS_RECORD_SEP);<a name="line.274"></a>
-<span class="sourceLineNo">275</span> sb.append("Max Keys for bloom: " + m);<a name="line.275"></a>
-<span class="sourceLineNo">276</span> if (m > 0) {<a name="line.276"></a>
-<span class="sourceLineNo">277</span> sb.append(STATS_RECORD_SEP + "Percentage filled: "<a name="line.277"></a>
-<span class="sourceLineNo">278</span> + NumberFormat.getPercentInstance().format(k * 1.0 / m));<a name="line.278"></a>
-<span class="sourceLineNo">279</span> }<a name="line.279"></a>
-<span class="sourceLineNo">280</span> return sb.toString();<a name="line.280"></a>
-<span class="sourceLineNo">281</span> }<a name="line.281"></a>
-<span class="sourceLineNo">282</span><a name="line.282"></a>
-<span class="sourceLineNo">283</span> public static String toString(BloomFilterChunk bloomFilter) {<a name="line.283"></a>
-<span class="sourceLineNo">284</span> return formatStats(bloomFilter) + STATS_RECORD_SEP + "Actual error rate: "<a name="line.284"></a>
-<span class="sourceLineNo">285</span> + String.format("%.8f", bloomFilter.actualErrorRate());<a name="line.285"></a>
+<span class="sourceLineNo">217</span> private static <T> boolean contains(ByteBuff bloomBuf, int bloomOffset, int bloomSize, Hash hash,<a name="line.217"></a>
+<span class="sourceLineNo">218</span> int hashCount, HashKey<T> hashKey) {<a name="line.218"></a>
+<span class="sourceLineNo">219</span> int hash1 = hash.hash(hashKey, 0);<a name="line.219"></a>
+<span class="sourceLineNo">220</span> int bloomBitSize = bloomSize << 3;<a name="line.220"></a>
+<span class="sourceLineNo">221</span><a name="line.221"></a>
+<span class="sourceLineNo">222</span> int hash2 = 0;<a name="line.222"></a>
+<span class="sourceLineNo">223</span> int compositeHash = 0;<a name="line.223"></a>
+<span class="sourceLineNo">224</span><a name="line.224"></a>
+<span class="sourceLineNo">225</span> if (randomGeneratorForTest == null) {<a name="line.225"></a>
+<span class="sourceLineNo">226</span> // Production mode<a name="line.226"></a>
+<span class="sourceLineNo">227</span> compositeHash = hash1;<a name="line.227"></a>
+<span class="sourceLineNo">228</span> hash2 = hash.hash(hashKey, hash1);<a name="line.228"></a>
+<span class="sourceLineNo">229</span> }<a name="line.229"></a>
+<span class="sourceLineNo">230</span><a name="line.230"></a>
+<span class="sourceLineNo">231</span> for (int i = 0; i < hashCount; i++) {<a name="line.231"></a>
+<span class="sourceLineNo">232</span> int hashLoc = (randomGeneratorForTest == null<a name="line.232"></a>
+<span class="sourceLineNo">233</span> // Production mode<a name="line.233"></a>
+<span class="sourceLineNo">234</span> ? Math.abs(compositeHash % bloomBitSize)<a name="line.234"></a>
+<span class="sourceLineNo">235</span> // Test mode with "fake look-ups" to estimate "ideal false positive rate"<a name="line.235"></a>
+<span class="sourceLineNo">236</span> : randomGeneratorForTest.nextInt(bloomBitSize));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> compositeHash += hash2;<a name="line.237"></a>
+<span class="sourceLineNo">238</span> if (!checkBit(hashLoc, bloomBuf, bloomOffset)) {<a name="line.238"></a>
+<span class="sourceLineNo">239</span> return false;<a name="line.239"></a>
+<span class="sourceLineNo">240</span> }<a name="line.240"></a>
+<span class="sourceLineNo">241</span> }<a name="line.241"></a>
+<span class="sourceLineNo">242</span> return true;<a name="line.242"></a>
+<span class="sourceLineNo">243</span> }<a name="line.243"></a>
+<span class="sourceLineNo">244</span><a name="line.244"></a>
+<span class="sourceLineNo">245</span> public static boolean contains(Cell cell, ByteBuff bloomBuf, int bloomOffset, int bloomSize,<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Hash hash, int hashCount, BloomType type) {<a name="line.246"></a>
+<span class="sourceLineNo">247</span> HashKey<Cell> hashKey = type == BloomType.ROWCOL ? new RowColBloomHashKey(cell)<a name="line.247"></a>
+<span class="sourceLineNo">248</span> : new RowBloomHashKey(cell);<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return contains(bloomBuf, bloomOffset, bloomSize, hash, hashCount, hashKey);<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> /**<a name="line.252"></a>
+<span class="sourceLineNo">253</span> * Check if bit at specified index is 1.<a name="line.253"></a>
+<span class="sourceLineNo">254</span> *<a name="line.254"></a>
+<span class="sourceLineNo">255</span> * @param pos index of bit<a name="line.255"></a>
+<span class="sourceLineNo">256</span> * @return true if bit at specified index is 1, false if 0.<a name="line.256"></a>
+<span class="sourceLineNo">257</span> */<a name="line.257"></a>
+<span class="sourceLineNo">258</span> static boolean checkBit(int pos, ByteBuff bloomBuf, int bloomOffset) {<a name="line.258"></a>
+<span class="sourceLineNo">259</span> int bytePos = pos >> 3; //pos / 8<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int bitPos = pos & 0x7; //pos % 8<a name="line.260"></a>
+<span class="sourceLineNo">261</span> byte curByte = bloomBuf.get(bloomOffset + bytePos);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> curByte &= bitvals[bitPos];<a name="line.262"></a>
+<span class="sourceLineNo">263</span> return (curByte != 0);<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> /**<a name="line.266"></a>
+<span class="sourceLineNo">267</span> * A human-readable string with statistics for the given Bloom filter.<a name="line.267"></a>
+<span class="sourceLineNo">268</span> *<a name="line.268"></a>
+<span class="sourceLineNo">269</span> * @param bloomFilter the Bloom filter to output statistics for;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> * @return a string consisting of "&lt;key&gt;: &lt;value&gt;" parts<a name="line.270"></a>
+<span class="sourceLineNo">271</span> * separated by {@link #STATS_RECORD_SEP}.<a name="line.271"></a>
+<span class="sourceLineNo">272</span> */<a name="line.272"></a>
+<span class="sourceLineNo">273</span> public static String formatStats(BloomFilterBase bloomFilter) {<a name="line.273"></a>
+<span class="sourceLineNo">274</span> StringBuilder sb = new StringBuilder();<a name="line.274"></a>
+<span class="sourceLineNo">275</span> long k = bloomFilter.getKeyCount();<a name="line.275"></a>
+<span class="sourceLineNo">276</span> long m = bloomFilter.getMaxKeys();<a name="line.276"></a>
+<span class="sourceLineNo">277</span><a name="line.277"></a>
+<span class="sourceLineNo">278</span> sb.append("BloomSize: " + bloomFilter.getByteSize() + STATS_RECORD_SEP);<a name="line.278"></a>
+<span class="sourceLineNo">279</span> sb.append("No of Keys in bloom: " + k + STATS_RECORD_SEP);<a name="line.279"></a>
+<span class="sourceLineNo">280</span> sb.append("Max Keys for bloom: " + m);<a name="line.280"></a>
+<span class="sourceLineNo">281</span> if (m > 0) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> sb.append(STATS_RECORD_SEP + "Percentage filled: "<a name="line.282"></a>
+<span class="sourceLineNo">283</span> + NumberFormat.getPercentInstance().format(k * 1.0 / m));<a name="line.283"></a>
+<span class="sourceLineNo">284</span> }<a name="line.284"></a>
+<span class="sourceLineNo">285</span> return sb.toString();<a name="line.285"></a>
<span class="sourceLineNo">286</span> }<a name="line.286"></a>
-<span class="sourceLineNo">287</span>}<a name="line.287"></a>
+<span class="sourceLineNo">287</span><a name="line.287"></a>
+<span class="sourceLineNo">288</span> public static String toString(BloomFilterChunk bloomFilter) {<a name="line.288"></a>
+<span class="sourceLineNo">289</span> return formatStats(bloomFilter) + STATS_RECORD_SEP + "Actual error rate: "<a name="line.289"></a>
+<span class="sourceLineNo">290</span> + String.format("%.8f", bloomFilter.actualErrorRate());<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> public static byte[] getBloomFilterParam(BloomType bloomFilterType, Configuration conf)<a name="line.293"></a>
+<span class="sourceLineNo">294</span> throws IllegalArgumentException{<a name="line.294"></a>
+<span class="sourceLineNo">295</span> byte[] bloomParam = null;<a name="line.295"></a>
+<span class="sourceLineNo">296</span> String message = "Bloom filter type is " + bloomFilterType + ", ";<a name="line.296"></a>
+<span class="sourceLineNo">297</span> switch (bloomFilterType) {<a name="line.297"></a>
+<span class="sourceLineNo">298</span> case ROWPREFIX_FIXED_LENGTH:<a name="line.298"></a>
+<span class="sourceLineNo">299</span> String prefixLengthString = conf.get(PREFIX_LENGTH_KEY);<a name="line.299"></a>
+<span class="sourceLineNo">300</span> if (prefixLengthString == null) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> message += PREFIX_LENGTH_KEY + " not specified.";<a name="line.301"></a>
+<span class="sourceLineNo">302</span> throw new IllegalArgumentException(message);<a name="line.302"></a>
+<span class="sourceLineNo">303</span> }<a name="line.303"></a>
+<span class="sourceLineNo">304</span> int prefixLength;<a name="line.304"></a>
+<span class="sourceLineNo">305</span> try {<a name="line.305"></a>
+<span class="sourceLineNo">306</span> prefixLength = Integer.parseInt(prefixLengthString);<a name="line.306"></a>
+<span class="sourceLineNo">307</span> if (prefixLength <= 0 || prefixLength > HConstants.MAX_ROW_LENGTH) {<a name="line.307"></a>
+<span class="sourceLineNo">308</span> message += "the value of " + PREFIX_LENGTH_KEY<a name="line.308"></a>
+<span class="sourceLineNo">309</span> + " must >=0 and < " + HConstants.MAX_ROW_LENGTH;<a name="line.309"></a>
+<span class="sourceLineNo">310</span> throw new IllegalArgumentException(message);<a name="line.310"></a>
+<span class="sourceLineNo">311</span> }<a name="line.311"></a>
+<span class="sourceLineNo">312</span> } catch (NumberFormatException nfe) {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> message = "Number format exception when parsing " + PREFIX_LENGTH_KEY + " for BloomType "<a name="line.313"></a>
+<span class="sourceLineNo">314</span> + bloomFilterType.toString() + ":"<a name="line.314"></a>
+<span class="sourceLineNo">315</span> + prefixLengthString;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> throw new IllegalArgumentException(message, nfe);<a name="line.316"></a>
+<span class="sourceLineNo">317</span> }<a name="line.317"></a>
+<span class="sourceLineNo">318</span> bloomParam = Bytes.toBytes(prefixLength);<a name="line.318"></a>
+<span class="sourceLineNo">319</span> break;<a name="line.319"></a>
+<span class="sourceLineNo">320</span> case ROWPREFIX_DELIMITED:<a name="line.320"></a>
+<span class="sourceLineNo">321</span> String delimiterString = conf.get(DELIMITER_KEY);<a name="line.321"></a>
+<span class="sourceLineNo">322</span> if (delimiterString == null || delimiterString.length() == 0) {<a name="line.322"></a>
+<span class="sourceLineNo">323</span> message += DELIMITER_KEY + " not specified.";<a name="line.323"></a>
+<span class="sourceLineNo">324</span> throw new IllegalArgumentException(message);<a name="line.324"></a>
+<span class="sourceLineNo">325</span> }<a name="line.325"></a>
+<span class="sourceLineNo">326</span> bloomParam = Bytes.toBytes(delimiterString);<a name="line.326"></a>
+<span class="sourceLineNo">327</span> break;<a name="line.327"></a>
+<span class="sourceLineNo">328</span> default:<a name="line.328"></a>
+<span class="sourceLineNo">329</span> break;<a name="line.329"></a>
+<span class="sourceLineNo">330</span> }<a name="line.330"></a>
+<span class="sourceLineNo">331</span> return bloomParam;<a name="line.331"></a>
+<span class="sourceLineNo">332</span> }<a name="line.332"></a>
+<span class="sourceLineNo">333</span>}<a name="line.333"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/bbe517ae/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixDelimiterBloomContext.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixDelimiterBloomContext.html b/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixDelimiterBloomContext.html
new file mode 100644
index 0000000..9f92a72
--- /dev/null
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixDelimiterBloomContext.html
@@ -0,0 +1,134 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+<title>Source code</title>
+<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
+</head>
+<body>
+<div class="sourceContainer">
+<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a>
+<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
+<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
+<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
+<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
+<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
+<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
+<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
+<span class="sourceLineNo">009</span> *<a name="line.9"></a>
+<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
+<span class="sourceLineNo">011</span> *<a name="line.11"></a>
+<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
+<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
+<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
+<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
+<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
+<span class="sourceLineNo">017</span> */<a name="line.17"></a>
+<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.util;<a name="line.18"></a>
+<span class="sourceLineNo">019</span><a name="line.19"></a>
+<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
+<span class="sourceLineNo">021</span><a name="line.21"></a>
+<span class="sourceLineNo">022</span>import org.apache.hadoop.hbase.Cell;<a name="line.22"></a>
+<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.CellBuilderType;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.ExtendedCellBuilderFactory;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.27"></a>
+<span class="sourceLineNo">028</span><a name="line.28"></a>
+<span class="sourceLineNo">029</span>/**<a name="line.29"></a>
+<span class="sourceLineNo">030</span> * Handles ROWPREFIX_DELIMITED bloom related context.<a name="line.30"></a>
+<span class="sourceLineNo">031</span> * It works with both ByteBufferedCell and byte[] backed cells<a name="line.31"></a>
+<span class="sourceLineNo">032</span> */<a name="line.32"></a>
+<span class="sourceLineNo">033</span>@InterfaceAudience.Private<a name="line.33"></a>
+<span class="sourceLineNo">034</span>public class RowPrefixDelimiterBloomContext extends RowBloomContext {<a name="line.34"></a>
+<span class="sourceLineNo">035</span> private final byte[] delimiter;<a name="line.35"></a>
+<span class="sourceLineNo">036</span><a name="line.36"></a>
+<span class="sourceLineNo">037</span> public RowPrefixDelimiterBloomContext(BloomFilterWriter bloomFilterWriter,<a name="line.37"></a>
+<span class="sourceLineNo">038</span> CellComparator comparator, byte[] delimiter) {<a name="line.38"></a>
+<span class="sourceLineNo">039</span> super(bloomFilterWriter, comparator);<a name="line.39"></a>
+<span class="sourceLineNo">040</span> this.delimiter = delimiter;<a name="line.40"></a>
+<span class="sourceLineNo">041</span> }<a name="line.41"></a>
+<span class="sourceLineNo">042</span><a name="line.42"></a>
+<span class="sourceLineNo">043</span> public void writeBloom(Cell cell) throws IOException {<a name="line.43"></a>
+<span class="sourceLineNo">044</span> super.writeBloom(getDelimitedRowPrefixCell(cell));<a name="line.44"></a>
+<span class="sourceLineNo">045</span> }<a name="line.45"></a>
+<span class="sourceLineNo">046</span><a name="line.46"></a>
+<span class="sourceLineNo">047</span> /**<a name="line.47"></a>
+<span class="sourceLineNo">048</span> * @param cell the new cell<a name="line.48"></a>
+<span class="sourceLineNo">049</span> * @return the new cell created by delimited row prefix<a name="line.49"></a>
+<span class="sourceLineNo">050</span> */<a name="line.50"></a>
+<span class="sourceLineNo">051</span> private Cell getDelimitedRowPrefixCell(Cell cell) {<a name="line.51"></a>
+<span class="sourceLineNo">052</span> byte[] row = CellUtil.copyRow(cell);<a name="line.52"></a>
+<span class="sourceLineNo">053</span> int prefixLength = Bytes.indexOf(row, delimiter);<a name="line.53"></a>
+<span class="sourceLineNo">054</span> if (prefixLength <= 0) {<a name="line.54"></a>
+<span class="sourceLineNo">055</span> return cell;<a name="line.55"></a>
+<span class="sourceLineNo">056</span> }<a name="line.56"></a>
+<span class="sourceLineNo">057</span> return ExtendedCellBuilderFactory.create(CellBuilderType.DEEP_COPY)<a name="line.57"></a>
+<span class="sourceLineNo">058</span> .setRow(row, 0, Math.min(prefixLength, row.length))<a name="line.58"></a>
+<span class="sourceLineNo">059</span> .setType(Cell.Type.Put)<a name="line.59"></a>
+<span class="sourceLineNo">060</span> .build();<a name="line.60"></a>
+<span class="sourceLineNo">061</span> }<a name="line.61"></a>
+<span class="sourceLineNo">062</span>}<a name="line.62"></a>
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+</pre>
+</div>
+</body>
+</html>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/bbe517ae/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixFixedLengthBloomContext.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixFixedLengthBloomContext.html b/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixFixedLengthBloomContext.html
new file mode 100644
index 0000000..b76323a
--- /dev/null
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/util/RowPrefixFixedLengthBloomContext.html
@@ -0,0 +1,130 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+<title>Source code</title>
+<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
+</head>
+<body>
+<div class="sourceContainer">
+<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a>
+<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
+<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
+<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
+<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
+<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
+<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
+<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
+<span class="sourceLineNo">009</span> *<a name="line.9"></a>
+<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
+<span class="sourceLineNo">011</span> *<a name="line.11"></a>
+<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
+<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
+<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
+<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
+<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
+<span class="sourceLineNo">017</span> */<a name="line.17"></a>
+<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.util;<a name="line.18"></a>
+<span class="sourceLineNo">019</span><a name="line.19"></a>
+<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
+<span class="sourceLineNo">021</span><a name="line.21"></a>
+<span class="sourceLineNo">022</span>import org.apache.hadoop.hbase.Cell;<a name="line.22"></a>
+<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.CellBuilderType;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.ExtendedCellBuilderFactory;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.27"></a>
+<span class="sourceLineNo">028</span><a name="line.28"></a>
+<span class="sourceLineNo">029</span>/**<a name="line.29"></a>
+<span class="sourceLineNo">030</span> * Handles ROWPREFIX bloom related context.<a name="line.30"></a>
+<span class="sourceLineNo">031</span> * It works with both ByteBufferedCell and byte[] backed cells<a name="line.31"></a>
+<span class="sourceLineNo">032</span> */<a name="line.32"></a>
+<span class="sourceLineNo">033</span>@InterfaceAudience.Private<a name="line.33"></a>
+<span class="sourceLineNo">034</span>public class RowPrefixFixedLengthBloomContext extends RowBloomContext {<a name="line.34"></a>
+<span class="sourceLineNo">035</span> private final int prefixLength;<a name="line.35"></a>
+<span class="sourceLineNo">036</span><a name="line.36"></a>
+<span class="sourceLineNo">037</span> public RowPrefixFixedLengthBloomContext(BloomFilterWriter bloomFilterWriter,<a name="line.37"></a>
+<span class="sourceLineNo">038</span> CellComparator comparator, int prefixLength) {<a name="line.38"></a>
+<span class="sourceLineNo">039</span> super(bloomFilterWriter, comparator);<a name="line.39"></a>
+<span class="sourceLineNo">040</span> this.prefixLength = prefixLength;<a name="line.40"></a>
+<span class="sourceLineNo">041</span> }<a name="line.41"></a>
+<span class="sourceLineNo">042</span><a name="line.42"></a>
+<span class="sourceLineNo">043</span> public void writeBloom(Cell cell) throws IOException {<a name="line.43"></a>
+<span class="sourceLineNo">044</span> super.writeBloom(getRowPrefixCell(cell));<a name="line.44"></a>
+<span class="sourceLineNo">045</span> }<a name="line.45"></a>
+<span class="sourceLineNo">046</span><a name="line.46"></a>
+<span class="sourceLineNo">047</span> /**<a name="line.47"></a>
+<span class="sourceLineNo">048</span> * @param cell the cell<a name="line.48"></a>
+<span class="sourceLineNo">049</span> * @return the new cell created by row prefix<a name="line.49"></a>
+<span class="sourceLineNo">050</span> */<a name="line.50"></a>
+<span class="sourceLineNo">051</span> private Cell getRowPrefixCell(Cell cell) {<a name="line.51"></a>
+<span class="sourceLineNo">052</span> byte[] row = CellUtil.copyRow(cell);<a name="line.52"></a>
+<span class="sourceLineNo">053</span> return ExtendedCellBuilderFactory.create(CellBuilderType.DEEP_COPY)<a name="line.53"></a>
+<span class="sourceLineNo">054</span> .setRow(row, 0, Math.min(prefixLength, row.length))<a name="line.54"></a>
+<span class="sourceLineNo">055</span> .setType(Cell.Type.Put)<a name="line.55"></a>
+<span class="sourceLineNo">056</span> .build();<a name="line.56"></a>
+<span class="sourceLineNo">057</span> }<a name="line.57"></a>
+<span class="sourceLineNo">058</span>}<a name="line.58"></a>
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+</pre>
+</div>
+</body>
+</html>