http://git-wip-us.apache.org/repos/asf/hbase-site/blob/0cf79db0/apidocs/src-html/org/apache/hadoop/hbase/filter/MultiRowRangeFilter.RowRange.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/filter/MultiRowRangeFilter.RowRange.html b/apidocs/src-html/org/apache/hadoop/hbase/filter/MultiRowRangeFilter.RowRange.html index 4804996..c3cf972 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/filter/MultiRowRangeFilter.RowRange.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/filter/MultiRowRangeFilter.RowRange.html @@ -28,500 +28,538 @@ <span class="sourceLineNo">020</span>import java.util.ArrayList;<a name="line.20"></a> <span class="sourceLineNo">021</span>import java.util.Collections;<a name="line.21"></a> <span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a> -<span class="sourceLineNo">023</span><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.CellUtil;<a name="line.25"></a> -<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.HConstants;<a name="line.26"></a> -<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.PrivateCellUtil;<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>import org.apache.hadoop.hbase.exceptions.DeserializationException;<a name="line.29"></a> -<span class="sourceLineNo">030</span>import org.apache.hbase.thirdparty.com.google.protobuf.InvalidProtocolBufferException;<a name="line.30"></a> -<span class="sourceLineNo">031</span>import org.apache.hbase.thirdparty.com.google.protobuf.UnsafeByteOperations;<a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.FilterProtos;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.33"></a> -<span class="sourceLineNo">034</span><a name="line.34"></a> -<span class="sourceLineNo">035</span>/**<a name="line.35"></a> -<span class="sourceLineNo">036</span> * Filter to support scan multiple row key ranges. It can construct the row key ranges from the<a name="line.36"></a> -<span class="sourceLineNo">037</span> * passed list which can be accessed by each region server.<a name="line.37"></a> -<span class="sourceLineNo">038</span> *<a name="line.38"></a> -<span class="sourceLineNo">039</span> * HBase is quite efficient when scanning only one small row key range. If user needs to specify<a name="line.39"></a> -<span class="sourceLineNo">040</span> * multiple row key ranges in one scan, the typical solutions are: 1. through FilterList which is a<a name="line.40"></a> -<span class="sourceLineNo">041</span> * list of row key Filters, 2. using the SQL layer over HBase to join with two table, such as hive,<a name="line.41"></a> -<span class="sourceLineNo">042</span> * phoenix etc. However, both solutions are inefficient. Both of them can't utilize the range info<a name="line.42"></a> -<span class="sourceLineNo">043</span> * to perform fast forwarding during scan which is quite time consuming. If the number of ranges<a name="line.43"></a> -<span class="sourceLineNo">044</span> * are quite big (e.g. millions), join is a proper solution though it is slow. However, there are<a name="line.44"></a> -<span class="sourceLineNo">045</span> * cases that user wants to specify a small number of ranges to scan (e.g. &lt;1000 ranges). Both<a name="line.45"></a> -<span class="sourceLineNo">046</span> * solutions can't provide satisfactory performance in such case. MultiRowRangeFilter is to support<a name="line.46"></a> -<span class="sourceLineNo">047</span> * such usec ase (scan multiple row key ranges), which can construct the row key ranges from user<a name="line.47"></a> -<span class="sourceLineNo">048</span> * specified list and perform fast-forwarding during scan. Thus, the scan will be quite efficient.<a name="line.48"></a> -<span class="sourceLineNo">049</span> */<a name="line.49"></a> -<span class="sourceLineNo">050</span>@InterfaceAudience.Public<a name="line.50"></a> -<span class="sourceLineNo">051</span>public class MultiRowRangeFilter extends FilterBase {<a name="line.51"></a> -<span class="sourceLineNo">052</span><a name="line.52"></a> -<span class="sourceLineNo">053</span> private List<RowRange> rangeList;<a name="line.53"></a> -<span class="sourceLineNo">054</span><a name="line.54"></a> -<span class="sourceLineNo">055</span> private static final int ROW_BEFORE_FIRST_RANGE = -1;<a name="line.55"></a> -<span class="sourceLineNo">056</span> private boolean EXCLUSIVE = false;<a name="line.56"></a> -<span class="sourceLineNo">057</span> private boolean done = false;<a name="line.57"></a> -<span class="sourceLineNo">058</span> private boolean initialized = false;<a name="line.58"></a> -<span class="sourceLineNo">059</span> private int index;<a name="line.59"></a> -<span class="sourceLineNo">060</span> private RowRange range;<a name="line.60"></a> -<span class="sourceLineNo">061</span> private ReturnCode currentReturnCode;<a name="line.61"></a> -<span class="sourceLineNo">062</span><a name="line.62"></a> -<span class="sourceLineNo">063</span> /**<a name="line.63"></a> -<span class="sourceLineNo">064</span> * @param list A list of <code>RowRange</code><a name="line.64"></a> -<span class="sourceLineNo">065</span> */<a name="line.65"></a> -<span class="sourceLineNo">066</span> public MultiRowRangeFilter(List<RowRange> list) {<a name="line.66"></a> -<span class="sourceLineNo">067</span> this.rangeList = sortAndMerge(list);<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> @Override<a name="line.70"></a> -<span class="sourceLineNo">071</span> public boolean filterAllRemaining() {<a name="line.71"></a> -<span class="sourceLineNo">072</span> return done;<a name="line.72"></a> -<span class="sourceLineNo">073</span> }<a name="line.73"></a> -<span class="sourceLineNo">074</span><a name="line.74"></a> -<span class="sourceLineNo">075</span> public List<RowRange> getRowRanges() {<a name="line.75"></a> -<span class="sourceLineNo">076</span> return this.rangeList;<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> @Override<a name="line.79"></a> -<span class="sourceLineNo">080</span> public boolean filterRowKey(Cell firstRowCell) {<a name="line.80"></a> -<span class="sourceLineNo">081</span> if (filterAllRemaining()) return true;<a name="line.81"></a> -<span class="sourceLineNo">082</span> // If it is the first time of running, calculate the current range index for<a name="line.82"></a> -<span class="sourceLineNo">083</span> // the row key. If index is out of bound which happens when the start row<a name="line.83"></a> -<span class="sourceLineNo">084</span> // user sets is after the largest stop row of the ranges, stop the scan.<a name="line.84"></a> -<span class="sourceLineNo">085</span> // If row key is after the current range, find the next range and update index.<a name="line.85"></a> -<span class="sourceLineNo">086</span> byte[] rowArr = firstRowCell.getRowArray();<a name="line.86"></a> -<span class="sourceLineNo">087</span> int length = firstRowCell.getRowLength();<a name="line.87"></a> -<span class="sourceLineNo">088</span> int offset = firstRowCell.getRowOffset();<a name="line.88"></a> -<span class="sourceLineNo">089</span> if (!initialized<a name="line.89"></a> -<span class="sourceLineNo">090</span> || !range.contains(rowArr, offset, length)) {<a name="line.90"></a> -<span class="sourceLineNo">091</span> byte[] rowkey = CellUtil.cloneRow(firstRowCell);<a name="line.91"></a> -<span class="sourceLineNo">092</span> index = getNextRangeIndex(rowkey);<a name="line.92"></a> -<span class="sourceLineNo">093</span> if (index >= rangeList.size()) {<a name="line.93"></a> -<span class="sourceLineNo">094</span> done = true;<a name="line.94"></a> -<span class="sourceLineNo">095</span> currentReturnCode = ReturnCode.NEXT_ROW;<a name="line.95"></a> -<span class="sourceLineNo">096</span> return false;<a name="line.96"></a> -<span class="sourceLineNo">097</span> }<a name="line.97"></a> -<span class="sourceLineNo">098</span> if(index != ROW_BEFORE_FIRST_RANGE) {<a name="line.98"></a> -<span class="sourceLineNo">099</span> range = rangeList.get(index);<a name="line.99"></a> -<span class="sourceLineNo">100</span> } else {<a name="line.100"></a> -<span class="sourceLineNo">101</span> range = rangeList.get(0);<a name="line.101"></a> -<span class="sourceLineNo">102</span> }<a name="line.102"></a> -<span class="sourceLineNo">103</span> if (EXCLUSIVE) {<a name="line.103"></a> -<span class="sourceLineNo">104</span> EXCLUSIVE = false;<a name="line.104"></a> -<span class="sourceLineNo">105</span> currentReturnCode = ReturnCode.NEXT_ROW;<a name="line.105"></a> -<span class="sourceLineNo">106</span> return false;<a name="line.106"></a> -<span class="sourceLineNo">107</span> }<a name="line.107"></a> -<span class="sourceLineNo">108</span> if (!initialized) {<a name="line.108"></a> -<span class="sourceLineNo">109</span> if(index != ROW_BEFORE_FIRST_RANGE) {<a name="line.109"></a> -<span class="sourceLineNo">110</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.110"></a> -<span class="sourceLineNo">111</span> } else {<a name="line.111"></a> -<span class="sourceLineNo">112</span> currentReturnCode = ReturnCode.SEEK_NEXT_USING_HINT;<a name="line.112"></a> -<span class="sourceLineNo">113</span> }<a name="line.113"></a> -<span class="sourceLineNo">114</span> initialized = true;<a name="line.114"></a> -<span class="sourceLineNo">115</span> } else {<a name="line.115"></a> -<span class="sourceLineNo">116</span> if (range.contains(rowArr, offset, length)) {<a name="line.116"></a> -<span class="sourceLineNo">117</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.117"></a> -<span class="sourceLineNo">118</span> } else {<a name="line.118"></a> -<span class="sourceLineNo">119</span> currentReturnCode = ReturnCode.SEEK_NEXT_USING_HINT;<a name="line.119"></a> -<span class="sourceLineNo">120</span> }<a name="line.120"></a> -<span class="sourceLineNo">121</span> }<a name="line.121"></a> -<span class="sourceLineNo">122</span> } else {<a name="line.122"></a> -<span class="sourceLineNo">123</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.123"></a> -<span class="sourceLineNo">124</span> }<a name="line.124"></a> -<span class="sourceLineNo">125</span> return false;<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> @Deprecated<a name="line.128"></a> -<span class="sourceLineNo">129</span> @Override<a name="line.129"></a> -<span class="sourceLineNo">130</span> public ReturnCode filterKeyValue(final Cell ignored) {<a name="line.130"></a> -<span class="sourceLineNo">131</span> return filterCell(ignored);<a name="line.131"></a> -<span class="sourceLineNo">132</span> }<a name="line.132"></a> -<span class="sourceLineNo">133</span><a name="line.133"></a> -<span class="sourceLineNo">134</span> @Override<a name="line.134"></a> -<span class="sourceLineNo">135</span> public ReturnCode filterCell(final Cell ignored) {<a name="line.135"></a> -<span class="sourceLineNo">136</span> return currentReturnCode;<a name="line.136"></a> -<span class="sourceLineNo">137</span> }<a name="line.137"></a> -<span class="sourceLineNo">138</span><a name="line.138"></a> -<span class="sourceLineNo">139</span> @Override<a name="line.139"></a> -<span class="sourceLineNo">140</span> public Cell getNextCellHint(Cell currentKV) {<a name="line.140"></a> -<span class="sourceLineNo">141</span> // skip to the next range's start row<a name="line.141"></a> -<span class="sourceLineNo">142</span> return PrivateCellUtil.createFirstOnRow(range.startRow, 0,<a name="line.142"></a> -<span class="sourceLineNo">143</span> (short) range.startRow.length);<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> /**<a name="line.146"></a> -<span class="sourceLineNo">147</span> * @return The filter serialized using pb<a name="line.147"></a> -<span class="sourceLineNo">148</span> */<a name="line.148"></a> -<span class="sourceLineNo">149</span> @Override<a name="line.149"></a> -<span class="sourceLineNo">150</span> public byte[] toByteArray() {<a name="line.150"></a> -<span class="sourceLineNo">151</span> FilterProtos.MultiRowRangeFilter.Builder builder = FilterProtos.MultiRowRangeFilter<a name="line.151"></a> -<span class="sourceLineNo">152</span> .newBuilder();<a name="line.152"></a> -<span class="sourceLineNo">153</span> for (RowRange range : rangeList) {<a name="line.153"></a> -<span class="sourceLineNo">154</span> if (range != null) {<a name="line.154"></a> -<span class="sourceLineNo">155</span> FilterProtos.RowRange.Builder rangebuilder = FilterProtos.RowRange.newBuilder();<a name="line.155"></a> -<span class="sourceLineNo">156</span> if (range.startRow != null)<a name="line.156"></a> -<span class="sourceLineNo">157</span> rangebuilder.setStartRow(UnsafeByteOperations.unsafeWrap(range.startRow));<a name="line.157"></a> -<span class="sourceLineNo">158</span> rangebuilder.setStartRowInclusive(range.startRowInclusive);<a name="line.158"></a> -<span class="sourceLineNo">159</span> if (range.stopRow != null)<a name="line.159"></a> -<span class="sourceLineNo">160</span> rangebuilder.setStopRow(UnsafeByteOperations.unsafeWrap(range.stopRow));<a name="line.160"></a> -<span class="sourceLineNo">161</span> rangebuilder.setStopRowInclusive(range.stopRowInclusive);<a name="line.161"></a> -<span class="sourceLineNo">162</span> builder.addRowRangeList(rangebuilder.build());<a name="line.162"></a> -<span class="sourceLineNo">163</span> }<a name="line.163"></a> -<span class="sourceLineNo">164</span> }<a name="line.164"></a> -<span class="sourceLineNo">165</span> return builder.build().toByteArray();<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> * @param pbBytes A pb serialized instance<a name="line.169"></a> -<span class="sourceLineNo">170</span> * @return An instance of MultiRowRangeFilter<a name="line.170"></a> -<span class="sourceLineNo">171</span> * @throws org.apache.hadoop.hbase.exceptions.DeserializationException<a name="line.171"></a> -<span class="sourceLineNo">172</span> */<a name="line.172"></a> -<span class="sourceLineNo">173</span> public static MultiRowRangeFilter parseFrom(final byte[] pbBytes)<a name="line.173"></a> -<span class="sourceLineNo">174</span> throws DeserializationException {<a name="line.174"></a> -<span class="sourceLineNo">175</span> FilterProtos.MultiRowRangeFilter proto;<a name="line.175"></a> -<span class="sourceLineNo">176</span> try {<a name="line.176"></a> -<span class="sourceLineNo">177</span> proto = FilterProtos.MultiRowRangeFilter.parseFrom(pbBytes);<a name="line.177"></a> -<span class="sourceLineNo">178</span> } catch (InvalidProtocolBufferException e) {<a name="line.178"></a> -<span class="sourceLineNo">179</span> throw new DeserializationException(e);<a name="line.179"></a> -<span class="sourceLineNo">180</span> }<a name="line.180"></a> -<span class="sourceLineNo">181</span> int length = proto.getRowRangeListCount();<a name="line.181"></a> -<span class="sourceLineNo">182</span> List<FilterProtos.RowRange> rangeProtos = proto.getRowRangeListList();<a name="line.182"></a> -<span class="sourceLineNo">183</span> List<RowRange> rangeList = new ArrayList<>(length);<a name="line.183"></a> -<span class="sourceLineNo">184</span> for (FilterProtos.RowRange rangeProto : rangeProtos) {<a name="line.184"></a> -<span class="sourceLineNo">185</span> RowRange range = new RowRange(rangeProto.hasStartRow() ? rangeProto.getStartRow()<a name="line.185"></a> -<span class="sourceLineNo">186</span> .toByteArray() : null, rangeProto.getStartRowInclusive(), rangeProto.hasStopRow() ?<a name="line.186"></a> -<span class="sourceLineNo">187</span> rangeProto.getStopRow().toByteArray() : null, rangeProto.getStopRowInclusive());<a name="line.187"></a> -<span class="sourceLineNo">188</span> rangeList.add(range);<a name="line.188"></a> -<span class="sourceLineNo">189</span> }<a name="line.189"></a> -<span class="sourceLineNo">190</span> return new MultiRowRangeFilter(rangeList);<a name="line.190"></a> -<span class="sourceLineNo">191</span> }<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> * @param o the filter to compare<a name="line.194"></a> -<span class="sourceLineNo">195</span> * @return true if and only if the fields of the filter that are serialized are equal to the<a name="line.195"></a> -<span class="sourceLineNo">196</span> * corresponding fields in other. Used for testing.<a name="line.196"></a> -<span class="sourceLineNo">197</span> */<a name="line.197"></a> -<span class="sourceLineNo">198</span> @Override<a name="line.198"></a> -<span class="sourceLineNo">199</span> boolean areSerializedFieldsEqual(Filter o) {<a name="line.199"></a> -<span class="sourceLineNo">200</span> if (o == this)<a name="line.200"></a> -<span class="sourceLineNo">201</span> return true;<a name="line.201"></a> -<span class="sourceLineNo">202</span> if (!(o instanceof MultiRowRangeFilter))<a name="line.202"></a> -<span class="sourceLineNo">203</span> return false;<a name="line.203"></a> -<span class="sourceLineNo">204</span><a name="line.204"></a> -<span class="sourceLineNo">205</span> MultiRowRangeFilter other = (MultiRowRangeFilter) o;<a name="line.205"></a> -<span class="sourceLineNo">206</span> if (this.rangeList.size() != other.rangeList.size())<a name="line.206"></a> -<span class="sourceLineNo">207</span> return false;<a name="line.207"></a> -<span class="sourceLineNo">208</span> for (int i = 0; i < rangeList.size(); ++i) {<a name="line.208"></a> -<span class="sourceLineNo">209</span> RowRange thisRange = this.rangeList.get(i);<a name="line.209"></a> -<span class="sourceLineNo">210</span> RowRange otherRange = other.rangeList.get(i);<a name="line.210"></a> -<span class="sourceLineNo">211</span> if (!(Bytes.equals(thisRange.startRow, otherRange.startRow) && Bytes.equals(<a name="line.211"></a> -<span class="sourceLineNo">212</span> thisRange.stopRow, otherRange.stopRow) && (thisRange.startRowInclusive ==<a name="line.212"></a> -<span class="sourceLineNo">213</span> otherRange.startRowInclusive) && (thisRange.stopRowInclusive ==<a name="line.213"></a> -<span class="sourceLineNo">214</span> otherRange.stopRowInclusive))) {<a name="line.214"></a> -<span class="sourceLineNo">215</span> return false;<a name="line.215"></a> -<span class="sourceLineNo">216</span> }<a name="line.216"></a> -<span class="sourceLineNo">217</span> }<a name="line.217"></a> -<span class="sourceLineNo">218</span> return true;<a name="line.218"></a> -<span class="sourceLineNo">219</span> }<a name="line.219"></a> -<span class="sourceLineNo">220</span><a name="line.220"></a> -<span class="sourceLineNo">221</span> /**<a name="line.221"></a> -<span class="sourceLineNo">222</span> * calculate the position where the row key in the ranges list.<a name="line.222"></a> -<span class="sourceLineNo">223</span> *<a name="line.223"></a> -<span class="sourceLineNo">224</span> * @param rowKey the row key to calculate<a name="line.224"></a> -<span class="sourceLineNo">225</span> * @return index the position of the row key<a name="line.225"></a> -<span class="sourceLineNo">226</span> */<a name="line.226"></a> -<span class="sourceLineNo">227</span> private int getNextRangeIndex(byte[] rowKey) {<a name="line.227"></a> -<span class="sourceLineNo">228</span> RowRange temp = new RowRange(rowKey, true, null, true);<a name="line.228"></a> -<span class="sourceLineNo">229</span> int index = Collections.binarySearch(rangeList, temp);<a name="line.229"></a> -<span class="sourceLineNo">230</span> if (index < 0) {<a name="line.230"></a> -<span class="sourceLineNo">231</span> int insertionPosition = -index - 1;<a name="line.231"></a> -<span class="sourceLineNo">232</span> // check if the row key in the range before the insertion position<a name="line.232"></a> -<span class="sourceLineNo">233</span> if (insertionPosition != 0 && rangeList.get(insertionPosition - 1).contains(rowKey)) {<a name="line.233"></a> -<span class="sourceLineNo">234</span> return insertionPosition - 1;<a name="line.234"></a> -<span class="sourceLineNo">235</span> }<a name="line.235"></a> -<span class="sourceLineNo">236</span> // check if the row key is before the first range<a name="line.236"></a> -<span class="sourceLineNo">237</span> if (insertionPosition == 0 && !rangeList.get(insertionPosition).contains(rowKey)) {<a name="line.237"></a> -<span class="sourceLineNo">238</span> return ROW_BEFORE_FIRST_RANGE;<a name="line.238"></a> -<span class="sourceLineNo">239</span> }<a name="line.239"></a> -<span class="sourceLineNo">240</span> if (!initialized) {<a name="line.240"></a> -<span class="sourceLineNo">241</span> initialized = true;<a name="line.241"></a> -<span class="sourceLineNo">242</span> }<a name="line.242"></a> -<span class="sourceLineNo">243</span> return insertionPosition;<a name="line.243"></a> -<span class="sourceLineNo">244</span> }<a name="line.244"></a> -<span class="sourceLineNo">245</span> // the row key equals one of the start keys, and the the range exclude the start key<a name="line.245"></a> -<span class="sourceLineNo">246</span> if(rangeList.get(index).startRowInclusive == false) {<a name="line.246"></a> -<span class="sourceLineNo">247</span> EXCLUSIVE = true;<a name="line.247"></a> -<span class="sourceLineNo">248</span> }<a name="line.248"></a> -<span class="sourceLineNo">249</span> return index;<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> * sort the ranges and if the ranges with overlap, then merge them.<a name="line.253"></a> -<span class="sourceLineNo">254</span> *<a name="line.254"></a> -<span class="sourceLineNo">255</span> * @param ranges the list of ranges to sort and merge.<a name="line.255"></a> -<span class="sourceLineNo">256</span> * @return the ranges after sort and merge.<a name="line.256"></a> -<span class="sourceLineNo">257</span> */<a name="line.257"></a> -<span class="sourceLineNo">258</span> public static List<RowRange> sortAndMerge(List<RowRange> ranges) {<a name="line.258"></a> -<span class="sourceLineNo">259</span> if (ranges.isEmpty()) {<a name="line.259"></a> -<span class="sourceLineNo">260</span> throw new IllegalArgumentException("No ranges found.");<a name="line.260"></a> -<span class="sourceLineNo">261</span> }<a name="line.261"></a> -<span class="sourceLineNo">262</span> List<RowRange> invalidRanges = new ArrayList<>();<a name="line.262"></a> -<span class="sourceLineNo">263</span> List<RowRange> newRanges = new ArrayList<>(ranges.size());<a name="line.263"></a> -<span class="sourceLineNo">264</span> Collections.sort(ranges);<a name="line.264"></a> -<span class="sourceLineNo">265</span> if(ranges.get(0).isValid()) {<a name="line.265"></a> -<span class="sourceLineNo">266</span> if (ranges.size() == 1) {<a name="line.266"></a> -<span class="sourceLineNo">267</span> newRanges.add(ranges.get(0));<a name="line.267"></a> -<span class="sourceLineNo">268</span> }<a name="line.268"></a> -<span class="sourceLineNo">269</span> } else {<a name="line.269"></a> -<span class="sourceLineNo">270</span> invalidRanges.add(ranges.get(0));<a name="line.270"></a> -<span class="sourceLineNo">271</span> }<a name="line.271"></a> -<span class="sourceLineNo">272</span><a name="line.272"></a> -<span class="sourceLineNo">273</span> byte[] lastStartRow = ranges.get(0).startRow;<a name="line.273"></a> -<span class="sourceLineNo">274</span> boolean lastStartRowInclusive = ranges.get(0).startRowInclusive;<a name="line.274"></a> -<span class="sourceLineNo">275</span> byte[] lastStopRow = ranges.get(0).stopRow;<a name="line.275"></a> -<span class="sourceLineNo">276</span> boolean lastStopRowInclusive = ranges.get(0).stopRowInclusive;<a name="line.276"></a> -<span class="sourceLineNo">277</span> int i = 1;<a name="line.277"></a> -<span class="sourceLineNo">278</span> for (; i < ranges.size(); i++) {<a name="line.278"></a> -<span class="sourceLineNo">279</span> RowRange range = ranges.get(i);<a name="line.279"></a> -<span class="sourceLineNo">280</span> if (!range.isValid()) {<a name="line.280"></a> -<span class="sourceLineNo">281</span> invalidRanges.add(range);<a name="line.281"></a> -<span class="sourceLineNo">282</span> }<a name="line.282"></a> -<span class="sourceLineNo">283</span> if(Bytes.equals(lastStopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.283"></a> -<span class="sourceLineNo">284</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.284"></a> -<span class="sourceLineNo">285</span> lastStopRowInclusive));<a name="line.285"></a> -<span class="sourceLineNo">286</span> break;<a name="line.286"></a> -<span class="sourceLineNo">287</span> }<a name="line.287"></a> -<span class="sourceLineNo">288</span> // with overlap in the ranges<a name="line.288"></a> -<span class="sourceLineNo">289</span> if ((Bytes.compareTo(lastStopRow, range.startRow) > 0) ||<a name="line.289"></a> -<span class="sourceLineNo">290</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 && !(lastStopRowInclusive == false &&<a name="line.290"></a> -<span class="sourceLineNo">291</span> range.isStartRowInclusive() == false))) {<a name="line.291"></a> -<span class="sourceLineNo">292</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.292"></a> -<span class="sourceLineNo">293</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, range.stopRow,<a name="line.293"></a> -<span class="sourceLineNo">294</span> range.stopRowInclusive));<a name="line.294"></a> -<span class="sourceLineNo">295</span> break;<a name="line.295"></a> -<span class="sourceLineNo">296</span> }<a name="line.296"></a> -<span class="sourceLineNo">297</span> // if first range contains second range, ignore the second range<a name="line.297"></a> -<span class="sourceLineNo">298</span> if (Bytes.compareTo(lastStopRow, range.stopRow) >= 0) {<a name="line.298"></a> -<span class="sourceLineNo">299</span> if((Bytes.compareTo(lastStopRow, range.stopRow) == 0)) {<a name="line.299"></a> -<span class="sourceLineNo">300</span> if(lastStopRowInclusive == true || range.stopRowInclusive == true) {<a name="line.300"></a> -<span class="sourceLineNo">301</span> lastStopRowInclusive = true;<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> if ((i + 1) == ranges.size()) {<a name="line.304"></a> -<span class="sourceLineNo">305</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.305"></a> -<span class="sourceLineNo">306</span> lastStopRowInclusive));<a name="line.306"></a> -<span class="sourceLineNo">307</span> }<a name="line.307"></a> -<span class="sourceLineNo">308</span> } else {<a name="line.308"></a> -<span class="sourceLineNo">309</span> lastStopRow = range.stopRow;<a name="line.309"></a> -<span class="sourceLineNo">310</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.310"></a> -<span class="sourceLineNo">311</span> if ((i + 1) < ranges.size()) {<a name="line.311"></a> -<span class="sourceLineNo">312</span> i++;<a name="line.312"></a> -<span class="sourceLineNo">313</span> range = ranges.get(i);<a name="line.313"></a> -<span class="sourceLineNo">314</span> if (!range.isValid()) {<a name="line.314"></a> -<span class="sourceLineNo">315</span> invalidRanges.add(range);<a name="line.315"></a> -<span class="sourceLineNo">316</span> }<a name="line.316"></a> -<span class="sourceLineNo">317</span> } else {<a name="line.317"></a> -<span class="sourceLineNo">318</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.318"></a> -<span class="sourceLineNo">319</span> lastStopRowInclusive));<a name="line.319"></a> -<span class="sourceLineNo">320</span> break;<a name="line.320"></a> -<span class="sourceLineNo">321</span> }<a name="line.321"></a> -<span class="sourceLineNo">322</span> while ((Bytes.compareTo(lastStopRow, range.startRow) > 0) ||<a name="line.322"></a> -<span class="sourceLineNo">323</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 &&<a name="line.323"></a> -<span class="sourceLineNo">324</span> (lastStopRowInclusive == true || range.startRowInclusive==true))) {<a name="line.324"></a> -<span class="sourceLineNo">325</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.325"></a> -<span class="sourceLineNo">326</span> break;<a name="line.326"></a> -<span class="sourceLineNo">327</span> }<a name="line.327"></a> -<span class="sourceLineNo">328</span> // if this first range contain second range, ignore the second range<a name="line.328"></a> -<span class="sourceLineNo">329</span> if (Bytes.compareTo(lastStopRow, range.stopRow) >= 0) {<a name="line.329"></a> -<span class="sourceLineNo">330</span> if(lastStopRowInclusive == true || range.stopRowInclusive == true) {<a name="line.330"></a> -<span class="sourceLineNo">331</span> lastStopRowInclusive = true;<a name="line.331"></a> -<span class="sourceLineNo">332</span> }<a name="line.332"></a> -<span class="sourceLineNo">333</span> i++;<a name="line.333"></a> -<span class="sourceLineNo">334</span> if (i < ranges.size()) {<a name="line.334"></a> -<span class="sourceLineNo">335</span> range = ranges.get(i);<a name="line.335"></a> -<span class="sourceLineNo">336</span> if (!range.isValid()) {<a name="line.336"></a> -<span class="sourceLineNo">337</span> invalidRanges.add(range);<a name="line.337"></a> -<span class="sourceLineNo">338</span> }<a name="line.338"></a> -<span class="sourceLineNo">339</span> } else {<a name="line.339"></a> -<span class="sourceLineNo">340</span> break;<a name="line.340"></a> -<span class="sourceLineNo">341</span> }<a name="line.341"></a> -<span class="sourceLineNo">342</span> } else {<a name="line.342"></a> -<span class="sourceLineNo">343</span> lastStopRow = range.stopRow;<a name="line.343"></a> -<span class="sourceLineNo">344</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.344"></a> -<span class="sourceLineNo">345</span> i++;<a name="line.345"></a> -<span class="sourceLineNo">346</span> if (i < ranges.size()) {<a name="line.346"></a> -<span class="sourceLineNo">347</span> range = ranges.get(i);<a name="line.347"></a> -<span class="sourceLineNo">348</span> if (!range.isValid()) {<a name="line.348"></a> -<span class="sourceLineNo">349</span> invalidRanges.add(range);<a name="line.349"></a> -<span class="sourceLineNo">350</span> }<a name="line.350"></a> -<span class="sourceLineNo">351</span> } else {<a name="line.351"></a> -<span class="sourceLineNo">352</span> break;<a name="line.352"></a> -<span class="sourceLineNo">353</span> }<a name="line.353"></a> -<span class="sourceLineNo">354</span> }<a name="line.354"></a> -<span class="sourceLineNo">355</span> }<a name="line.355"></a> -<span class="sourceLineNo">356</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.356"></a> -<span class="sourceLineNo">357</span> if((Bytes.compareTo(lastStopRow, range.startRow) < 0) ||<a name="line.357"></a> -<span class="sourceLineNo">358</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 &&<a name="line.358"></a> -<span class="sourceLineNo">359</span> lastStopRowInclusive == false && range.startRowInclusive == false)) {<a name="line.359"></a> -<span class="sourceLineNo">360</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.360"></a> -<span class="sourceLineNo">361</span> lastStopRowInclusive));<a name="line.361"></a> -<span class="sourceLineNo">362</span> newRanges.add(range);<a name="line.362"></a> -<span class="sourceLineNo">363</span> } else {<a name="line.363"></a> -<span class="sourceLineNo">364</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, range.stopRow,<a name="line.364"></a> -<span class="sourceLineNo">365</span> range.stopRowInclusive));<a name="line.365"></a> -<span class="sourceLineNo">366</span> break;<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> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.369"></a> -<span class="sourceLineNo">370</span> lastStopRowInclusive));<a name="line.370"></a> -<span class="sourceLineNo">371</span> if ((i + 1) == ranges.size()) {<a name="line.371"></a> -<span class="sourceLineNo">372</span> newRanges.add(range);<a name="line.372"></a> -<span class="sourceLineNo">373</span> }<a name="line.373"></a> -<span class="sourceLineNo">374</span> lastStartRow = range.startRow;<a name="line.374"></a> -<span class="sourceLineNo">375</span> lastStartRowInclusive = range.startRowInclusive;<a name="line.375"></a> -<span class="sourceLineNo">376</span> lastStopRow = range.stopRow;<a name="line.376"></a> -<span class="sourceLineNo">377</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.377"></a> -<span class="sourceLineNo">378</span> }<a name="line.378"></a> -<span class="sourceLineNo">379</span> } else {<a name="line.379"></a> -<span class="sourceLineNo">380</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.380"></a> -<span class="sourceLineNo">381</span> lastStopRowInclusive));<a name="line.381"></a> -<span class="sourceLineNo">382</span> if ((i + 1) == ranges.size()) {<a name="line.382"></a> -<span class="sourceLineNo">383</span> newRanges.add(range);<a name="line.383"></a> -<span class="sourceLineNo">384</span> }<a name="line.384"></a> -<span class="sourceLineNo">385</span> lastStartRow = range.startRow;<a name="line.385"></a> -<span class="sourceLineNo">386</span> lastStartRowInclusive = range.startRowInclusive;<a name="line.386"></a> -<span class="sourceLineNo">387</span> lastStopRow = range.stopRow;<a name="line.387"></a> -<span class="sourceLineNo">388</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.388"></a> -<span class="sourceLineNo">389</span> }<a name="line.389"></a> -<span class="sourceLineNo">390</span> }<a name="line.390"></a> -<span class="sourceLineNo">391</span> // check the remaining ranges<a name="line.391"></a> -<span class="sourceLineNo">392</span> for(int j=i; j < ranges.size(); j++) {<a name="line.392"></a> -<span class="sourceLineNo">393</span> if(!ranges.get(j).isValid()) {<a name="line.393"></a> -<span class="sourceLineNo">394</span> invalidRanges.add(ranges.get(j));<a name="line.394"></a> -<span class="sourceLineNo">395</span> }<a name="line.395"></a> -<span class="sourceLineNo">396</span> }<a name="line.396"></a> -<span class="sourceLineNo">397</span> // if invalid range exists, throw the exception<a name="line.397"></a> -<span class="sourceLineNo">398</span> if (invalidRanges.size() != 0) {<a name="line.398"></a> -<span class="sourceLineNo">399</span> throwExceptionForInvalidRanges(invalidRanges, true);<a name="line.399"></a> -<span class="sourceLineNo">400</span> }<a name="line.400"></a> -<span class="sourceLineNo">401</span> // If no valid ranges found, throw the exception<a name="line.401"></a> -<span class="sourceLineNo">402</span> if(newRanges.isEmpty()) {<a name="line.402"></a> -<span class="sourceLineNo">403</span> throw new IllegalArgumentException("No valid ranges found.");<a name="line.403"></a> -<span class="sourceLineNo">404</span> }<a name="line.404"></a> -<span class="sourceLineNo">405</span> return newRanges;<a name="line.405"></a> -<span class="sourceLineNo">406</span> }<a name="line.406"></a> -<span class="sourceLineNo">407</span><a name="line.407"></a> -<span class="sourceLineNo">408</span> private static void throwExceptionForInvalidRanges(List<RowRange> invalidRanges,<a name="line.408"></a> -<span class="sourceLineNo">409</span> boolean details) {<a name="line.409"></a> -<span class="sourceLineNo">410</span> StringBuilder sb = new StringBuilder();<a name="line.410"></a> -<span class="sourceLineNo">411</span> sb.append(invalidRanges.size()).append(" invaild ranges.\n");<a name="line.411"></a> -<span class="sourceLineNo">412</span> if (details) {<a name="line.412"></a> -<span class="sourceLineNo">413</span> for (RowRange range : invalidRanges) {<a name="line.413"></a> -<span class="sourceLineNo">414</span> sb.append(<a name="line.414"></a> -<span class="sourceLineNo">415</span> "Invalid range: start row => " + Bytes.toString(range.startRow) + ", stop row => "<a name="line.415"></a> -<span class="sourceLineNo">416</span> + Bytes.toString(range.stopRow)).append('\n');<a name="line.416"></a> -<span class="sourceLineNo">417</span> }<a name="line.417"></a> -<span class="sourceLineNo">418</span> }<a name="line.418"></a> -<span class="sourceLineNo">419</span> throw new IllegalArgumentException(sb.toString());<a name="line.419"></a> -<span class="sourceLineNo">420</span> }<a name="line.420"></a> -<span class="sourceLineNo">421</span><a name="line.421"></a> -<span class="sourceLineNo">422</span> @InterfaceAudience.Public<a name="line.422"></a> -<span class="sourceLineNo">423</span> public static class RowRange implements Comparable<RowRange> {<a name="line.423"></a> -<span class="sourceLineNo">424</span> private byte[] startRow;<a name="line.424"></a> -<span class="sourceLineNo">425</span> private boolean startRowInclusive = true;<a name="line.425"></a> -<span class="sourceLineNo">426</span> private byte[] stopRow;<a name="line.426"></a> -<span class="sourceLineNo">427</span> private boolean stopRowInclusive = false;<a name="line.427"></a> -<span class="sourceLineNo">428</span><a name="line.428"></a> -<span class="sourceLineNo">429</span> public RowRange() {<a name="line.429"></a> -<span class="sourceLineNo">430</span> }<a name="line.430"></a> -<span class="sourceLineNo">431</span> /**<a name="line.431"></a> -<span class="sourceLineNo">432</span> * If the startRow is empty or null, set it to HConstants.EMPTY_BYTE_ARRAY, means begin at the<a name="line.432"></a> -<span class="sourceLineNo">433</span> * start row of the table. If the stopRow is empty or null, set it to<a name="line.433"></a> -<span class="sourceLineNo">434</span> * HConstants.EMPTY_BYTE_ARRAY, means end of the last row of table.<a name="line.434"></a> -<span class="sourceLineNo">435</span> */<a name="line.435"></a> -<span class="sourceLineNo">436</span> public RowRange(String startRow, boolean startRowInclusive, String stopRow,<a name="line.436"></a> -<span class="sourceLineNo">437</span> boolean stopRowInclusive) {<a name="line.437"></a> -<span class="sourceLineNo">438</span> this((startRow == null || startRow.isEmpty()) ? HConstants.EMPTY_BYTE_ARRAY :<a name="line.438"></a> -<span class="sourceLineNo">439</span> Bytes.toBytes(startRow), startRowInclusive,<a name="line.439"></a> -<span class="sourceLineNo">440</span> (stopRow == null || stopRow.isEmpty()) ? HConstants.EMPTY_BYTE_ARRAY :<a name="line.440"></a> -<span class="sourceLineNo">441</span> Bytes.toBytes(stopRow), stopRowInclusive);<a name="line.441"></a> -<span class="sourceLineNo">442</span> }<a name="line.442"></a> -<span class="sourceLineNo">443</span><a name="line.443"></a> -<span class="sourceLineNo">444</span> public RowRange(byte[] startRow, boolean startRowInclusive, byte[] stopRow,<a name="line.444"></a> -<span class="sourceLineNo">445</span> boolean stopRowInclusive) {<a name="line.445"></a> -<span class="sourceLineNo">446</span> this.startRow = (startRow == null) ? HConstants.EMPTY_BYTE_ARRAY : startRow;<a name="line.446"></a> -<span class="sourceLineNo">447</span> this.startRowInclusive = startRowInclusive;<a name="line.447"></a> -<span class="sourceLineNo">448</span> this.stopRow = (stopRow == null) ? HConstants.EMPTY_BYTE_ARRAY :stopRow;<a name="line.448"></a> -<span class="sourceLineNo">449</span> this.stopRowInclusive = stopRowInclusive;<a name="line.449"></a> -<span class="sourceLineNo">450</span> }<a name="line.450"></a> -<span class="sourceLineNo">451</span><a name="line.451"></a> -<span class="sourceLineNo">452</span> public byte[] getStartRow() {<a name="line.452"></a> -<span class="sourceLineNo">453</span> return startRow;<a name="line.453"></a> -<span class="sourceLineNo">454</span> }<a name="line.454"></a> -<span class="sourceLineNo">455</span><a name="line.455"></a> -<span class="sourceLineNo">456</span> public byte[] getStopRow() {<a name="line.456"></a> -<span class="sourceLineNo">457</span> return stopRow;<a name="line.457"></a> -<span class="sourceLineNo">458</span> }<a name="line.458"></a> -<span class="sourceLineNo">459</span><a name="line.459"></a> -<span class="sourceLineNo">460</span> /**<a name="line.460"></a> -<span class="sourceLineNo">461</span> * @return if start row is inclusive.<a name="line.461"></a> -<span class="sourceLineNo">462</span> */<a name="line.462"></a> -<span class="sourceLineNo">463</span> public boolean isStartRowInclusive() {<a name="line.463"></a> -<span class="sourceLineNo">464</span> return startRowInclusive;<a name="line.464"></a> -<span class="sourceLineNo">465</span> }<a name="line.465"></a> -<span class="sourceLineNo">466</span><a name="line.466"></a> -<span class="sourceLineNo">467</span> /**<a name="line.467"></a> -<span class="sourceLineNo">468</span> * @return if stop row is inclusive.<a name="line.468"></a> -<span class="sourceLineNo">469</span> */<a name="line.469"></a> -<span class="sourceLineNo">470</span> public boolean isStopRowInclusive() {<a name="line.470"></a> -<span class="sourceLineNo">471</span> return stopRowInclusive;<a name="line.471"></a> -<span class="sourceLineNo">472</span> }<a name="line.472"></a> -<span class="sourceLineNo">473</span><a name="line.473"></a> -<span class="sourceLineNo">474</span> public boolean contains(byte[] row) {<a name="line.474"></a> -<span class="sourceLineNo">475</span> return contains(row, 0, row.length);<a name="line.475"></a> -<span class="sourceLineNo">476</span> }<a name="line.476"></a> -<span class="sourceLineNo">477</span><a name="line.477"></a> -<span class="sourceLineNo">478</span> public boolean contains(byte[] buffer, int offset, int length) {<a name="line.478"></a> -<span class="sourceLineNo">479</span> if(startRowInclusive) {<a name="line.479"></a> -<span class="sourceLineNo">480</span> if(stopRowInclusive) {<a name="line.480"></a> -<span class="sourceLineNo">481</span> return Bytes.compareTo(buffer, offset, length, startRow, 0, startRow.length) >= 0<a name="line.481"></a> -<span class="sourceLineNo">482</span> && (Bytes.equals(stopRow, HConstants.EMPTY_BYTE_ARRAY) ||<a name="line.482"></a> -<span class="sourceLineNo">483</span> Bytes.compareTo(buffer, offset, length, stopRow, 0, stopRow.length) <= 0);<a name="line.483"></a> -<span class="sourceLineNo">484</span> } else {<a name="line.484"></a> -<span class="sourceLineNo">485</span> return Bytes.compareTo(buffer, offset, length, startRow, 0, startRow.length) >= 0<a name="line.485"></a> -<span class="sourceLineNo">486</span> && (Bytes.equals(stopRow, HConstants.EMPTY_BYTE_ARRAY) ||<a name="line.486"></a> -<span class="sourceLineNo">487</span> Bytes.compareTo(buffer, offset, length, stopRow, 0, stopRow.length) < 0);<a name="line.487"></a> -<span class="sourceLineNo">488</span> }<a name="line.488"></a> -<span class="sourceLineNo">489</span> } else {<a name="line.489"></a> -<span class="sourceLineNo">490</span> if(stopRowInclusive) {<a name="line.490"></a> -<span class="sourceLineNo">491</span> return Bytes.compareTo(buffer, offset, length, startRow, 0, startRow.length) > 0<a name="line.491"></a> -<span class="sourceLineNo">492</span> && (Bytes.equals(stopRow, HConstants.EMPTY_BYTE_ARRAY) ||<a name="line.492"></a> -<span class="sourceLineNo">493</span> Bytes.compareTo(buffer, offset, length, stopRow, 0, stopRow.length) <= 0);<a name="line.493"></a> -<span class="sourceLineNo">494</span> } else {<a name="line.494"></a> -<span class="sourceLineNo">495</span> return Bytes.compareTo(buffer, offset, length, startRow, 0, startRow.length) > 0<a name="line.495"></a> -<span class="sourceLineNo">496</span> && (Bytes.equals(stopRow, HConstants.EMPTY_BYTE_ARRAY) ||<a name="line.496"></a> -<span class="sourceLineNo">497</span> Bytes.compareTo(buffer, offset, length, stopRow, 0, stopRow.length) < 0);<a name="line.497"></a> -<span class="sourceLineNo">498</span> }<a name="line.498"></a> -<span class="sourceLineNo">499</span> }<a name="line.499"></a> -<span class="sourceLineNo">500</span> }<a name="line.500"></a> -<span class="sourceLineNo">501</span><a name="line.501"></a> -<span class="sourceLineNo">502</span> @Override<a name="line.502"></a> -<span class="sourceLineNo">503</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value="EQ_COMPARETO_USE_OBJECT_EQUALS",<a name="line.503"></a> -<span class="sourceLineNo">504</span> justification="This compareTo is not of this Object, but of referenced RowRange")<a name="line.504"></a> -<span class="sourceLineNo">505</span> public int compareTo(RowRange other) {<a name="line.505"></a> -<span class="sourceLineNo">506</span> return Bytes.compareTo(this.startRow, other.startRow);<a name="line.506"></a> -<span class="sourceLineNo">507</span> }<a name="line.507"></a> -<span class="sourceLineNo">508</span><a name="line.508"></a> -<span class="sourceLineNo">509</span> public boolean isValid() {<a name="line.509"></a> -<span class="sourceLineNo">510</span> return Bytes.equals(startRow, HConstants.EMPTY_BYTE_ARRAY)<a name="line.510"></a> -<span class="sourceLineNo">511</span> || Bytes.equals(stopRow, HConstants.EMPTY_BYTE_ARRAY)<a name="line.511"></a> -<span class="sourceLineNo">512</span> || Bytes.compareTo(startRow, stopRow) < 0<a name="line.512"></a> -<span class="sourceLineNo">513</span> || (Bytes.compareTo(startRow, stopRow) == 0 && stopRowInclusive == true);<a name="line.513"></a> -<span class="sourceLineNo">514</span> }<a name="line.514"></a> -<span class="sourceLineNo">515</span> }<a name="line.515"></a> -<span class="sourceLineNo">516</span>}<a name="line.516"></a> +<span class="sourceLineNo">023</span>import java.util.Objects;<a name="line.23"></a> +<span class="sourceLineNo">024</span><a name="line.24"></a> +<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.Cell;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.HConstants;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.PrivateCellUtil;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.exceptions.DeserializationException;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import org.apache.hbase.thirdparty.com.google.protobuf.InvalidProtocolBufferException;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import org.apache.hbase.thirdparty.com.google.protobuf.UnsafeByteOperations;<a name="line.32"></a> +<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.shaded.protobuf.generated.FilterProtos;<a name="line.33"></a> +<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.34"></a> +<span class="sourceLineNo">035</span><a name="line.35"></a> +<span class="sourceLineNo">036</span>/**<a name="line.36"></a> +<span class="sourceLineNo">037</span> * Filter to support scan multiple row key ranges. It can construct the row key ranges from the<a name="line.37"></a> +<span class="sourceLineNo">038</span> * passed list which can be accessed by each region server.<a name="line.38"></a> +<span class="sourceLineNo">039</span> *<a name="line.39"></a> +<span class="sourceLineNo">040</span> * HBase is quite efficient when scanning only one small row key range. If user needs to specify<a name="line.40"></a> +<span class="sourceLineNo">041</span> * multiple row key ranges in one scan, the typical solutions are: 1. through FilterList which is a<a name="line.41"></a> +<span class="sourceLineNo">042</span> * list of row key Filters, 2. using the SQL layer over HBase to join with two table, such as hive,<a name="line.42"></a> +<span class="sourceLineNo">043</span> * phoenix etc. However, both solutions are inefficient. Both of them can't utilize the range info<a name="line.43"></a> +<span class="sourceLineNo">044</span> * to perform fast forwarding during scan which is quite time consuming. If the number of ranges<a name="line.44"></a> +<span class="sourceLineNo">045</span> * are quite big (e.g. millions), join is a proper solution though it is slow. However, there are<a name="line.45"></a> +<span class="sourceLineNo">046</span> * cases that user wants to specify a small number of ranges to scan (e.g. &lt;1000 ranges). Both<a name="line.46"></a> +<span class="sourceLineNo">047</span> * solutions can't provide satisfactory performance in such case. MultiRowRangeFilter is to support<a name="line.47"></a> +<span class="sourceLineNo">048</span> * such usec ase (scan multiple row key ranges), which can construct the row key ranges from user<a name="line.48"></a> +<span class="sourceLineNo">049</span> * specified list and perform fast-forwarding during scan. Thus, the scan will be quite efficient.<a name="line.49"></a> +<span class="sourceLineNo">050</span> */<a name="line.50"></a> +<span class="sourceLineNo">051</span>@InterfaceAudience.Public<a name="line.51"></a> +<span class="sourceLineNo">052</span>public class MultiRowRangeFilter extends FilterBase {<a name="line.52"></a> +<span class="sourceLineNo">053</span><a name="line.53"></a> +<span class="sourceLineNo">054</span> private List<RowRange> rangeList;<a name="line.54"></a> +<span class="sourceLineNo">055</span><a name="line.55"></a> +<span class="sourceLineNo">056</span> private static final int ROW_BEFORE_FIRST_RANGE = -1;<a name="line.56"></a> +<span class="sourceLineNo">057</span> private boolean EXCLUSIVE = false;<a name="line.57"></a> +<span class="sourceLineNo">058</span> private boolean done = false;<a name="line.58"></a> +<span class="sourceLineNo">059</span> private boolean initialized = false;<a name="line.59"></a> +<span class="sourceLineNo">060</span> private int index;<a name="line.60"></a> +<span class="sourceLineNo">061</span> private RowRange range;<a name="line.61"></a> +<span class="sourceLineNo">062</span> private ReturnCode currentReturnCode;<a name="line.62"></a> +<span class="sourceLineNo">063</span><a name="line.63"></a> +<span class="sourceLineNo">064</span> /**<a name="line.64"></a> +<span class="sourceLineNo">065</span> * @param list A list of <code>RowRange</code><a name="line.65"></a> +<span class="sourceLineNo">066</span> */<a name="line.66"></a> +<span class="sourceLineNo">067</span> public MultiRowRangeFilter(List<RowRange> list) {<a name="line.67"></a> +<span class="sourceLineNo">068</span> this.rangeList = sortAndMerge(list);<a name="line.68"></a> +<span class="sourceLineNo">069</span> }<a name="line.69"></a> +<span class="sourceLineNo">070</span><a name="line.70"></a> +<span class="sourceLineNo">071</span> @Override<a name="line.71"></a> +<span class="sourceLineNo">072</span> public boolean filterAllRemaining() {<a name="line.72"></a> +<span class="sourceLineNo">073</span> return done;<a name="line.73"></a> +<span class="sourceLineNo">074</span> }<a name="line.74"></a> +<span class="sourceLineNo">075</span><a name="line.75"></a> +<span class="sourceLineNo">076</span> public List<RowRange> getRowRanges() {<a name="line.76"></a> +<span class="sourceLineNo">077</span> return this.rangeList;<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> @Override<a name="line.80"></a> +<span class="sourceLineNo">081</span> public boolean filterRowKey(Cell firstRowCell) {<a name="line.81"></a> +<span class="sourceLineNo">082</span> if (filterAllRemaining()) return true;<a name="line.82"></a> +<span class="sourceLineNo">083</span> // If it is the first time of running, calculate the current range index for<a name="line.83"></a> +<span class="sourceLineNo">084</span> // the row key. If index is out of bound which happens when the start row<a name="line.84"></a> +<span class="sourceLineNo">085</span> // user sets is after the largest stop row of the ranges, stop the scan.<a name="line.85"></a> +<span class="sourceLineNo">086</span> // If row key is after the current range, find the next range and update index.<a name="line.86"></a> +<span class="sourceLineNo">087</span> byte[] rowArr = firstRowCell.getRowArray();<a name="line.87"></a> +<span class="sourceLineNo">088</span> int length = firstRowCell.getRowLength();<a name="line.88"></a> +<span class="sourceLineNo">089</span> int offset = firstRowCell.getRowOffset();<a name="line.89"></a> +<span class="sourceLineNo">090</span> if (!initialized<a name="line.90"></a> +<span class="sourceLineNo">091</span> || !range.contains(rowArr, offset, length)) {<a name="line.91"></a> +<span class="sourceLineNo">092</span> byte[] rowkey = CellUtil.cloneRow(firstRowCell);<a name="line.92"></a> +<span class="sourceLineNo">093</span> index = getNextRangeIndex(rowkey);<a name="line.93"></a> +<span class="sourceLineNo">094</span> if (index >= rangeList.size()) {<a name="line.94"></a> +<span class="sourceLineNo">095</span> done = true;<a name="line.95"></a> +<span class="sourceLineNo">096</span> currentReturnCode = ReturnCode.NEXT_ROW;<a name="line.96"></a> +<span class="sourceLineNo">097</span> return false;<a name="line.97"></a> +<span class="sourceLineNo">098</span> }<a name="line.98"></a> +<span class="sourceLineNo">099</span> if(index != ROW_BEFORE_FIRST_RANGE) {<a name="line.99"></a> +<span class="sourceLineNo">100</span> range = rangeList.get(index);<a name="line.100"></a> +<span class="sourceLineNo">101</span> } else {<a name="line.101"></a> +<span class="sourceLineNo">102</span> range = rangeList.get(0);<a name="line.102"></a> +<span class="sourceLineNo">103</span> }<a name="line.103"></a> +<span class="sourceLineNo">104</span> if (EXCLUSIVE) {<a name="line.104"></a> +<span class="sourceLineNo">105</span> EXCLUSIVE = false;<a name="line.105"></a> +<span class="sourceLineNo">106</span> currentReturnCode = ReturnCode.NEXT_ROW;<a name="line.106"></a> +<span class="sourceLineNo">107</span> return false;<a name="line.107"></a> +<span class="sourceLineNo">108</span> }<a name="line.108"></a> +<span class="sourceLineNo">109</span> if (!initialized) {<a name="line.109"></a> +<span class="sourceLineNo">110</span> if(index != ROW_BEFORE_FIRST_RANGE) {<a name="line.110"></a> +<span class="sourceLineNo">111</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.111"></a> +<span class="sourceLineNo">112</span> } else {<a name="line.112"></a> +<span class="sourceLineNo">113</span> currentReturnCode = ReturnCode.SEEK_NEXT_USING_HINT;<a name="line.113"></a> +<span class="sourceLineNo">114</span> }<a name="line.114"></a> +<span class="sourceLineNo">115</span> initialized = true;<a name="line.115"></a> +<span class="sourceLineNo">116</span> } else {<a name="line.116"></a> +<span class="sourceLineNo">117</span> if (range.contains(rowArr, offset, length)) {<a name="line.117"></a> +<span class="sourceLineNo">118</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.118"></a> +<span class="sourceLineNo">119</span> } else {<a name="line.119"></a> +<span class="sourceLineNo">120</span> currentReturnCode = ReturnCode.SEEK_NEXT_USING_HINT;<a name="line.120"></a> +<span class="sourceLineNo">121</span> }<a name="line.121"></a> +<span class="sourceLineNo">122</span> }<a name="line.122"></a> +<span class="sourceLineNo">123</span> } else {<a name="line.123"></a> +<span class="sourceLineNo">124</span> currentReturnCode = ReturnCode.INCLUDE;<a name="line.124"></a> +<span class="sourceLineNo">125</span> }<a name="line.125"></a> +<span class="sourceLineNo">126</span> return false;<a name="line.126"></a> +<span class="sourceLineNo">127</span> }<a name="line.127"></a> +<span class="sourceLineNo">128</span><a name="line.128"></a> +<span class="sourceLineNo">129</span> @Deprecated<a name="line.129"></a> +<span class="sourceLineNo">130</span> @Override<a name="line.130"></a> +<span class="sourceLineNo">131</span> public ReturnCode filterKeyValue(final Cell ignored) {<a name="line.131"></a> +<span class="sourceLineNo">132</span> return filterCell(ignored);<a name="line.132"></a> +<span class="sourceLineNo">133</span> }<a name="line.133"></a> +<span class="sourceLineNo">134</span><a name="line.134"></a> +<span class="sourceLineNo">135</span> @Override<a name="line.135"></a> +<span class="sourceLineNo">136</span> public ReturnCode filterCell(final Cell ignored) {<a name="line.136"></a> +<span class="sourceLineNo">137</span> return currentReturnCode;<a name="line.137"></a> +<span class="sourceLineNo">138</span> }<a name="line.138"></a> +<span class="sourceLineNo">139</span><a name="line.139"></a> +<span class="sourceLineNo">140</span> @Override<a name="line.140"></a> +<span class="sourceLineNo">141</span> public Cell getNextCellHint(Cell currentKV) {<a name="line.141"></a> +<span class="sourceLineNo">142</span> // skip to the next range's start row<a name="line.142"></a> +<span class="sourceLineNo">143</span> return PrivateCellUtil.createFirstOnRow(range.startRow, 0,<a name="line.143"></a> +<span class="sourceLineNo">144</span> (short) range.startRow.length);<a name="line.144"></a> +<span class="sourceLineNo">145</span> }<a name="line.145"></a> +<span class="sourceLineNo">146</span><a name="line.146"></a> +<span class="sourceLineNo">147</span> /**<a name="line.147"></a> +<span class="sourceLineNo">148</span> * @return The filter serialized using pb<a name="line.148"></a> +<span class="sourceLineNo">149</span> */<a name="line.149"></a> +<span class="sourceLineNo">150</span> @Override<a name="line.150"></a> +<span class="sourceLineNo">151</span> public byte[] toByteArray() {<a name="line.151"></a> +<span class="sourceLineNo">152</span> FilterProtos.MultiRowRangeFilter.Builder builder = FilterProtos.MultiRowRangeFilter<a name="line.152"></a> +<span class="sourceLineNo">153</span> .newBuilder();<a name="line.153"></a> +<span class="sourceLineNo">154</span> for (RowRange range : rangeList) {<a name="line.154"></a> +<span class="sourceLineNo">155</span> if (range != null) {<a name="line.155"></a> +<span class="sourceLineNo">156</span> FilterProtos.RowRange.Builder rangebuilder = FilterProtos.RowRange.newBuilder();<a name="line.156"></a> +<span class="sourceLineNo">157</span> if (range.startRow != null)<a name="line.157"></a> +<span class="sourceLineNo">158</span> rangebuilder.setStartRow(UnsafeByteOperations.unsafeWrap(range.startRow));<a name="line.158"></a> +<span class="sourceLineNo">159</span> rangebuilder.setStartRowInclusive(range.startRowInclusive);<a name="line.159"></a> +<span class="sourceLineNo">160</span> if (range.stopRow != null)<a name="line.160"></a> +<span class="sourceLineNo">161</span> rangebuilder.setStopRow(UnsafeByteOperations.unsafeWrap(range.stopRow));<a name="line.161"></a> +<span class="sourceLineNo">162</span> rangebuilder.setStopRowInclusive(range.stopRowInclusive);<a name="line.162"></a> +<span class="sourceLineNo">163</span> builder.addRowRangeList(rangebuilder.build());<a name="line.163"></a> +<span class="sourceLineNo">164</span> }<a name="line.164"></a> +<span class="sourceLineNo">165</span> }<a name="line.165"></a> +<span class="sourceLineNo">166</span> return builder.build().toByteArray();<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> /**<a name="line.169"></a> +<span class="sourceLineNo">170</span> * @param pbBytes A pb serialized instance<a name="line.170"></a> +<span class="sourceLineNo">171</span> * @return An instance of MultiRowRangeFilter<a name="line.171"></a> +<span class="sourceLineNo">172</span> * @throws org.apache.hadoop.hbase.exceptions.DeserializationException<a name="line.172"></a> +<span class="sourceLineNo">173</span> */<a name="line.173"></a> +<span class="sourceLineNo">174</span> public static MultiRowRangeFilter parseFrom(final byte[] pbBytes)<a name="line.174"></a> +<span class="sourceLineNo">175</span> throws DeserializationException {<a name="line.175"></a> +<span class="sourceLineNo">176</span> FilterProtos.MultiRowRangeFilter proto;<a name="line.176"></a> +<span class="sourceLineNo">177</span> try {<a name="line.177"></a> +<span class="sourceLineNo">178</span> proto = FilterProtos.MultiRowRangeFilter.parseFrom(pbBytes);<a name="line.178"></a> +<span class="sourceLineNo">179</span> } catch (InvalidProtocolBufferException e) {<a name="line.179"></a> +<span class="sourceLineNo">180</span> throw new DeserializationException(e);<a name="line.180"></a> +<span class="sourceLineNo">181</span> }<a name="line.181"></a> +<span class="sourceLineNo">182</span> int length = proto.getRowRangeListCount();<a name="line.182"></a> +<span class="sourceLineNo">183</span> List<FilterProtos.RowRange> rangeProtos = proto.getRowRangeListList();<a name="line.183"></a> +<span class="sourceLineNo">184</span> List<RowRange> rangeList = new ArrayList<>(length);<a name="line.184"></a> +<span class="sourceLineNo">185</span> for (FilterProtos.RowRange rangeProto : rangeProtos) {<a name="line.185"></a> +<span class="sourceLineNo">186</span> RowRange range = new RowRange(rangeProto.hasStartRow() ? rangeProto.getStartRow()<a name="line.186"></a> +<span class="sourceLineNo">187</span> .toByteArray() : null, rangeProto.getStartRowInclusive(), rangeProto.hasStopRow() ?<a name="line.187"></a> +<span class="sourceLineNo">188</span> rangeProto.getStopRow().toByteArray() : null, rangeProto.getStopRowInclusive());<a name="line.188"></a> +<span class="sourceLineNo">189</span> rangeList.add(range);<a name="line.189"></a> +<span class="sourceLineNo">190</span> }<a name="line.190"></a> +<span class="sourceLineNo">191</span> return new MultiRowRangeFilter(rangeList);<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> * @param o the filter to compare<a name="line.195"></a> +<span class="sourceLineNo">196</span> * @return true if and only if the fields of the filter that are serialized are equal to the<a name="line.196"></a> +<span class="sourceLineNo">197</span> * corresponding fields in other. Used for testing.<a name="line.197"></a> +<span class="sourceLineNo">198</span> */<a name="line.198"></a> +<span class="sourceLineNo">199</span> @Override<a name="line.199"></a> +<span class="sourceLineNo">200</span> boolean areSerializedFieldsEqual(Filter o) {<a name="line.200"></a> +<span class="sourceLineNo">201</span> if (o == this)<a name="line.201"></a> +<span class="sourceLineNo">202</span> return true;<a name="line.202"></a> +<span class="sourceLineNo">203</span> if (!(o instanceof MultiRowRangeFilter))<a name="line.203"></a> +<span class="sourceLineNo">204</span> return false;<a name="line.204"></a> +<span class="sourceLineNo">205</span><a name="line.205"></a> +<span class="sourceLineNo">206</span> MultiRowRangeFilter other = (MultiRowRangeFilter) o;<a name="line.206"></a> +<span class="sourceLineNo">207</span> if (this.rangeList.size() != other.rangeList.size())<a name="line.207"></a> +<span class="sourceLineNo">208</span> return false;<a name="line.208"></a> +<span class="sourceLineNo">209</span> for (int i = 0; i < rangeList.size(); ++i) {<a name="line.209"></a> +<span class="sourceLineNo">210</span> RowRange thisRange = this.rangeList.get(i);<a name="line.210"></a> +<span class="sourceLineNo">211</span> RowRange otherRange = other.rangeList.get(i);<a name="line.211"></a> +<span class="sourceLineNo">212</span> if (!(Bytes.equals(thisRange.startRow, otherRange.startRow) && Bytes.equals(<a name="line.212"></a> +<span class="sourceLineNo">213</span> thisRange.stopRow, otherRange.stopRow) && (thisRange.startRowInclusive ==<a name="line.213"></a> +<span class="sourceLineNo">214</span> otherRange.startRowInclusive) && (thisRange.stopRowInclusive ==<a name="line.214"></a> +<span class="sourceLineNo">215</span> otherRange.stopRowInclusive))) {<a name="line.215"></a> +<span class="sourceLineNo">216</span> return false;<a name="line.216"></a> +<span class="sourceLineNo">217</span> }<a name="line.217"></a> +<span class="sourceLineNo">218</span> }<a name="line.218"></a> +<span class="sourceLineNo">219</span> return true;<a name="line.219"></a> +<span class="sourceLineNo">220</span> }<a name="line.220"></a> +<span class="sourceLineNo">221</span><a name="line.221"></a> +<span class="sourceLineNo">222</span> /**<a name="line.222"></a> +<span class="sourceLineNo">223</span> * calculate the position where the row key in the ranges list.<a name="line.223"></a> +<span class="sourceLineNo">224</span> *<a name="line.224"></a> +<span class="sourceLineNo">225</span> * @param rowKey the row key to calculate<a name="line.225"></a> +<span class="sourceLineNo">226</span> * @return index the position of the row key<a name="line.226"></a> +<span class="sourceLineNo">227</span> */<a name="line.227"></a> +<span class="sourceLineNo">228</span> private int getNextRangeIndex(byte[] rowKey) {<a name="line.228"></a> +<span class="sourceLineNo">229</span> RowRange temp = new RowRange(rowKey, true, null, true);<a name="line.229"></a> +<span class="sourceLineNo">230</span> int index = Collections.binarySearch(rangeList, temp);<a name="line.230"></a> +<span class="sourceLineNo">231</span> if (index < 0) {<a name="line.231"></a> +<span class="sourceLineNo">232</span> int insertionPosition = -index - 1;<a name="line.232"></a> +<span class="sourceLineNo">233</span> // check if the row key in the range before the insertion position<a name="line.233"></a> +<span class="sourceLineNo">234</span> if (insertionPosition != 0 && rangeList.get(insertionPosition - 1).contains(rowKey)) {<a name="line.234"></a> +<span class="sourceLineNo">235</span> return insertionPosition - 1;<a name="line.235"></a> +<span class="sourceLineNo">236</span> }<a name="line.236"></a> +<span class="sourceLineNo">237</span> // check if the row key is before the first range<a name="line.237"></a> +<span class="sourceLineNo">238</span> if (insertionPosition == 0 && !rangeList.get(insertionPosition).contains(rowKey)) {<a name="line.238"></a> +<span class="sourceLineNo">239</span> return ROW_BEFORE_FIRST_RANGE;<a name="line.239"></a> +<span class="sourceLineNo">240</span> }<a name="line.240"></a> +<span class="sourceLineNo">241</span> if (!initialized) {<a name="line.241"></a> +<span class="sourceLineNo">242</span> initialized = true;<a name="line.242"></a> +<span class="sourceLineNo">243</span> }<a name="line.243"></a> +<span class="sourceLineNo">244</span> return insertionPosition;<a name="line.244"></a> +<span class="sourceLineNo">245</span> }<a name="line.245"></a> +<span class="sourceLineNo">246</span> // the row key equals one of the start keys, and the the range exclude the start key<a name="line.246"></a> +<span class="sourceLineNo">247</span> if(rangeList.get(index).startRowInclusive == false) {<a name="line.247"></a> +<span class="sourceLineNo">248</span> EXCLUSIVE = true;<a name="line.248"></a> +<span class="sourceLineNo">249</span> }<a name="line.249"></a> +<span class="sourceLineNo">250</span> return index;<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> /**<a name="line.253"></a> +<span class="sourceLineNo">254</span> * sort the ranges and if the ranges with overlap, then merge them.<a name="line.254"></a> +<span class="sourceLineNo">255</span> *<a name="line.255"></a> +<span class="sourceLineNo">256</span> * @param ranges the list of ranges to sort and merge.<a name="line.256"></a> +<span class="sourceLineNo">257</span> * @return the ranges after sort and merge.<a name="line.257"></a> +<span class="sourceLineNo">258</span> */<a name="line.258"></a> +<span class="sourceLineNo">259</span> public static List<RowRange> sortAndMerge(List<RowRange> ranges) {<a name="line.259"></a> +<span class="sourceLineNo">260</span> if (ranges.isEmpty()) {<a name="line.260"></a> +<span class="sourceLineNo">261</span> throw new IllegalArgumentException("No ranges found.");<a name="line.261"></a> +<span class="sourceLineNo">262</span> }<a name="line.262"></a> +<span class="sourceLineNo">263</span> List<RowRange> invalidRanges = new ArrayList<>();<a name="line.263"></a> +<span class="sourceLineNo">264</span> List<RowRange> newRanges = new ArrayList<>(ranges.size());<a name="line.264"></a> +<span class="sourceLineNo">265</span> Collections.sort(ranges);<a name="line.265"></a> +<span class="sourceLineNo">266</span> if(ranges.get(0).isValid()) {<a name="line.266"></a> +<span class="sourceLineNo">267</span> if (ranges.size() == 1) {<a name="line.267"></a> +<span class="sourceLineNo">268</span> newRanges.add(ranges.get(0));<a name="line.268"></a> +<span class="sourceLineNo">269</span> }<a name="line.269"></a> +<span class="sourceLineNo">270</span> } else {<a name="line.270"></a> +<span class="sourceLineNo">271</span> invalidRanges.add(ranges.get(0));<a name="line.271"></a> +<span class="sourceLineNo">272</span> }<a name="line.272"></a> +<span class="sourceLineNo">273</span><a name="line.273"></a> +<span class="sourceLineNo">274</span> byte[] lastStartRow = ranges.get(0).startRow;<a name="line.274"></a> +<span class="sourceLineNo">275</span> boolean lastStartRowInclusive = ranges.get(0).startRowInclusive;<a name="line.275"></a> +<span class="sourceLineNo">276</span> byte[] lastStopRow = ranges.get(0).stopRow;<a name="line.276"></a> +<span class="sourceLineNo">277</span> boolean lastStopRowInclusive = ranges.get(0).stopRowInclusive;<a name="line.277"></a> +<span class="sourceLineNo">278</span> int i = 1;<a name="line.278"></a> +<span class="sourceLineNo">279</span> for (; i < ranges.size(); i++) {<a name="line.279"></a> +<span class="sourceLineNo">280</span> RowRange range = ranges.get(i);<a name="line.280"></a> +<span class="sourceLineNo">281</span> if (!range.isValid()) {<a name="line.281"></a> +<span class="sourceLineNo">282</span> invalidRanges.add(range);<a name="line.282"></a> +<span class="sourceLineNo">283</span> }<a name="line.283"></a> +<span class="sourceLineNo">284</span> if(Bytes.equals(lastStopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.284"></a> +<span class="sourceLineNo">285</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.285"></a> +<span class="sourceLineNo">286</span> lastStopRowInclusive));<a name="line.286"></a> +<span class="sourceLineNo">287</span> break;<a name="line.287"></a> +<span class="sourceLineNo">288</span> }<a name="line.288"></a> +<span class="sourceLineNo">289</span> // with overlap in the ranges<a name="line.289"></a> +<span class="sourceLineNo">290</span> if ((Bytes.compareTo(lastStopRow, range.startRow) > 0) ||<a name="line.290"></a> +<span class="sourceLineNo">291</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 && !(lastStopRowInclusive == false &&<a name="line.291"></a> +<span class="sourceLineNo">292</span> range.isStartRowInclusive() == false))) {<a name="line.292"></a> +<span class="sourceLineNo">293</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.293"></a> +<span class="sourceLineNo">294</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, range.stopRow,<a name="line.294"></a> +<span class="sourceLineNo">295</span> range.stopRowInclusive));<a name="line.295"></a> +<span class="sourceLineNo">296</span> break;<a name="line.296"></a> +<span class="sourceLineNo">297</span> }<a name="line.297"></a> +<span class="sourceLineNo">298</span> // if first range contains second range, ignore the second range<a name="line.298"></a> +<span class="sourceLineNo">299</span> if (Bytes.compareTo(lastStopRow, range.stopRow) >= 0) {<a name="line.299"></a> +<span class="sourceLineNo">300</span> if((Bytes.compareTo(lastStopRow, range.stopRow) == 0)) {<a name="line.300"></a> +<span class="sourceLineNo">301</span> if(lastStopRowInclusive == true || range.stopRowInclusive == true) {<a name="line.301"></a> +<span class="sourceLineNo">302</span> lastStopRowInclusive = true;<a name="line.302"></a> +<span class="sourceLineNo">303</span> }<a name="line.303"></a> +<span class="sourceLineNo">304</span> }<a name="line.304"></a> +<span class="sourceLineNo">305</span> if ((i + 1) == ranges.size()) {<a name="line.305"></a> +<span class="sourceLineNo">306</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.306"></a> +<span class="sourceLineNo">307</span> lastStopRowInclusive));<a name="line.307"></a> +<span class="sourceLineNo">308</span> }<a name="line.308"></a> +<span class="sourceLineNo">309</span> } else {<a name="line.309"></a> +<span class="sourceLineNo">310</span> lastStopRow = range.stopRow;<a name="line.310"></a> +<span class="sourceLineNo">311</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.311"></a> +<span class="sourceLineNo">312</span> if ((i + 1) < ranges.size()) {<a name="line.312"></a> +<span class="sourceLineNo">313</span> i++;<a name="line.313"></a> +<span class="sourceLineNo">314</span> range = ranges.get(i);<a name="line.314"></a> +<span class="sourceLineNo">315</span> if (!range.isValid()) {<a name="line.315"></a> +<span class="sourceLineNo">316</span> invalidRanges.add(range);<a name="line.316"></a> +<span class="sourceLineNo">317</span> }<a name="line.317"></a> +<span class="sourceLineNo">318</span> } else {<a name="line.318"></a> +<span class="sourceLineNo">319</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.319"></a> +<span class="sourceLineNo">320</span> lastStopRowInclusive));<a name="line.320"></a> +<span class="sourceLineNo">321</span> break;<a name="line.321"></a> +<span class="sourceLineNo">322</span> }<a name="line.322"></a> +<span class="sourceLineNo">323</span> while ((Bytes.compareTo(lastStopRow, range.startRow) > 0) ||<a name="line.323"></a> +<span class="sourceLineNo">324</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 &&<a name="line.324"></a> +<span class="sourceLineNo">325</span> (lastStopRowInclusive == true || range.startRowInclusive==true))) {<a name="line.325"></a> +<span class="sourceLineNo">326</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.326"></a> +<span class="sourceLineNo">327</span> break;<a name="line.327"></a> +<span class="sourceLineNo">328</span> }<a name="line.328"></a> +<span class="sourceLineNo">329</span> // if this first range contain second range, ignore the second range<a name="line.329"></a> +<span class="sourceLineNo">330</span> if (Bytes.compareTo(lastStopRow, range.stopRow) >= 0) {<a name="line.330"></a> +<span class="sourceLineNo">331</span> if(lastStopRowInclusive == true || range.stopRowInclusive == true) {<a name="line.331"></a> +<span class="sourceLineNo">332</span> lastStopRowInclusive = true;<a name="line.332"></a> +<span class="sourceLineNo">333</span> }<a name="line.333"></a> +<span class="sourceLineNo">334</span> i++;<a name="line.334"></a> +<span class="sourceLineNo">335</span> if (i < ranges.size()) {<a name="line.335"></a> +<span class="sourceLineNo">336</span> range = ranges.get(i);<a name="line.336"></a> +<span class="sourceLineNo">337</span> if (!range.isValid()) {<a name="line.337"></a> +<span class="sourceLineNo">338</span> invalidRanges.add(range);<a name="line.338"></a> +<span class="sourceLineNo">339</span> }<a name="line.339"></a> +<span class="sourceLineNo">340</span> } else {<a name="line.340"></a> +<span class="sourceLineNo">341</span> break;<a name="line.341"></a> +<span class="sourceLineNo">342</span> }<a name="line.342"></a> +<span class="sourceLineNo">343</span> } else {<a name="line.343"></a> +<span class="sourceLineNo">344</span> lastStopRow = range.stopRow;<a name="line.344"></a> +<span class="sourceLineNo">345</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.345"></a> +<span class="sourceLineNo">346</span> i++;<a name="line.346"></a> +<span class="sourceLineNo">347</span> if (i < ranges.size()) {<a name="line.347"></a> +<span class="sourceLineNo">348</span> range = ranges.get(i);<a name="line.348"></a> +<span class="sourceLineNo">349</span> if (!range.isValid()) {<a name="line.349"></a> +<span class="sourceLineNo">350</span> invalidRanges.add(range);<a name="line.350"></a> +<span class="sourceLineNo">351</span> }<a name="line.351"></a> +<span class="sourceLineNo">352</span> } else {<a name="line.352"></a> +<span class="sourceLineNo">353</span> break;<a name="line.353"></a> +<span class="sourceLineNo">354</span> }<a name="line.354"></a> +<span class="sourceLineNo">355</span> }<a name="line.355"></a> +<span class="sourceLineNo">356</span> }<a name="line.356"></a> +<span class="sourceLineNo">357</span> if(Bytes.equals(range.stopRow, HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.357"></a> +<span class="sourceLineNo">358</span> if((Bytes.compareTo(lastStopRow, range.startRow) < 0) ||<a name="line.358"></a> +<span class="sourceLineNo">359</span> (Bytes.compareTo(lastStopRow, range.startRow) == 0 &&<a name="line.359"></a> +<span class="sourceLineNo">360</span> lastStopRowInclusive == false && range.startRowInclusive == false)) {<a name="line.360"></a> +<span class="sourceLineNo">361</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.361"></a> +<span class="sourceLineNo">362</span> lastStopRowInclusive));<a name="line.362"></a> +<span class="sourceLineNo">363</span> newRanges.add(range);<a name="line.363"></a> +<span class="sourceLineNo">364</span> } else {<a name="line.364"></a> +<span class="sourceLineNo">365</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, range.stopRow,<a name="line.365"></a> +<span class="sourceLineNo">366</span> range.stopRowInclusive));<a name="line.366"></a> +<span class="sourceLineNo">367</span> break;<a name="line.367"></a> +<span class="sourceLineNo">368</span> }<a name="line.368"></a> +<span class="sourceLineNo">369</span> }<a name="line.369"></a> +<span class="sourceLineNo">370</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.370"></a> +<span class="sourceLineNo">371</span> lastStopRowInclusive));<a name="line.371"></a> +<span class="sourceLineNo">372</span> if ((i + 1) == ranges.size()) {<a name="line.372"></a> +<span class="sourceLineNo">373</span> newRanges.add(range);<a name="line.373"></a> +<span class="sourceLineNo">374</span> }<a name="line.374"></a> +<span class="sourceLineNo">375</span> lastStartRow = range.startRow;<a name="line.375"></a> +<span class="sourceLineNo">376</span> lastStartRowInclusive = range.startRowInclusive;<a name="line.376"></a> +<span class="sourceLineNo">377</span> lastStopRow = range.stopRow;<a name="line.377"></a> +<span class="sourceLineNo">378</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.378"></a> +<span class="sourceLineNo">379</span> }<a name="line.379"></a> +<span class="sourceLineNo">380</span> } else {<a name="line.380"></a> +<span class="sourceLineNo">381</span> newRanges.add(new RowRange(lastStartRow, lastStartRowInclusive, lastStopRow,<a name="line.381"></a> +<span class="sourceLineNo">382</span> lastStopRowInclusive));<a name="line.382"></a> +<span class="sourceLineNo">383</span> if ((i + 1) == ranges.size()) {<a name="line.383"></a> +<span class="sourceLineNo">384</span> newRanges.add(range);<a name="line.384"></a> +<span class="sourceLineNo">385</span> }<a name="line.385"></a> +<span class="sourceLineNo">386</span> lastStartRow = range.startRow;<a name="line.386"></a> +<span class="sourceLineNo">387</span> lastStartRowInclusive = range.startRowInclusive;<a name="line.387"></a> +<span class="sourceLineNo">388</span> lastStopRow = range.stopRow;<a name="line.388"></a> +<span class="sourceLineNo">389</span> lastStopRowInclusive = range.stopRowInclusive;<a name="line.389"></a> +<span class="sourceLineNo">390</span> }<a name="line.390"></a> +<span class="sourceLineNo">391</span> }<a name="line.391"></a> +<span class="sourceLineNo">392</span> // check the remaining ranges<a name="line.392"></a> +<span class="sourceLineNo">393</span> for(int j=i; j < ranges.size(); j++) {<a name="line.393"></a> +<span class="sourceLineNo">394</span> if(!ranges.get(j).isValid()) {<a name="line.394"></a> +<span class="sourceLineNo">395</span> invalidRanges.add(ranges.get(j));<a name="line.395"></a> +<span class="sourceLineNo">396</span> }<a name="line.396"></a> +<span class="sourceLineNo">397</span> }<a name="line.397"></a> +<span class="sourceLineNo">398</span> // if invalid range exists, throw the exception<a name="line.398"></a> +<span class="sourceLineNo">399</span> if (invalidRanges.size() != 0) {<a name="line.399"></a> +<span class="sourceLineNo">400</span> throwExceptionForInvalidRanges(invalidRanges, true);<a name="line.400"></a> +<span class="sourceLineNo">401</span> }<a name="line.401"></a> +<span class="sourceLineNo">402</span> // If no valid ranges found, throw the exception<a name="line.402"></a> +<span class="sourceLineNo">403</span> if(newRanges.isEmpty()) {<a name="line.403"></a> +<span class="sourceLineNo">404</span> throw new IllegalArgumentException("No valid ranges found.");<a name="line.404"></a> +<span class="sourceLineNo">405</span> }<a name="line.405"></a> +<span class="sourceLineNo">406</span> return newRanges;<a name="line.406"></a> +<span class="sourceLineNo">407</span> }<a name="line.407"></a> +<span class="sourceLineNo">408</span><a name="line.408"></a> +<span class="sourceLineNo">409</span> private static void throwExceptionForInvalidRanges(List<RowRange> invalidRanges,<a name="line.409"></a> +<span class="sourceLineNo">410</span> boolean details) {<a name="line.410"></a> +<span class="sourceLineNo">411</span> StringBuilder sb = new StringBuilder();<a name="line.411"></a> +<span class="sourceLineNo">412</span> sb.append(invalidRanges.size()).append(" invaild ranges.\n");<a name="line.412"></a> +<span class="sourceLineNo">413</span> if (details) {<a name="line.413"></a> +<span class="sourceLineNo">414</span> for (RowRange range : invalidRanges) {<a name="line.414"></a> +<span class="sourceLineNo">415</span> sb.append(<a name="line.415"></a> +<span class="sourceLineNo">416</span> "Invalid range: start row => " + Bytes.toString(range.startRow) + ", stop row => "<a name="line.416"></a> +<span class="sourceLineNo">417</span> + Bytes.toString(range.stopRow)).append('
<TRUNCATED>
