http://git-wip-us.apache.org/repos/asf/hbase-site/blob/75eda567/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html index cf443ff..1adf99a 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/MultiTableInputFormatBase.html @@ -118,195 +118,191 @@ <span class="sourceLineNo">110</span> @Override<a name="line.110"></a> <span class="sourceLineNo">111</span> public void close() throws IOException {<a name="line.111"></a> <span class="sourceLineNo">112</span> trr.close();<a name="line.112"></a> -<span class="sourceLineNo">113</span> if (connection != null) {<a name="line.113"></a> -<span class="sourceLineNo">114</span> connection.close();<a name="line.114"></a> -<span class="sourceLineNo">115</span> }<a name="line.115"></a> -<span class="sourceLineNo">116</span> }<a name="line.116"></a> -<span class="sourceLineNo">117</span><a name="line.117"></a> -<span class="sourceLineNo">118</span> @Override<a name="line.118"></a> -<span class="sourceLineNo">119</span> public ImmutableBytesWritable getCurrentKey() throws IOException, InterruptedException {<a name="line.119"></a> -<span class="sourceLineNo">120</span> return trr.getCurrentKey();<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> @Override<a name="line.123"></a> -<span class="sourceLineNo">124</span> public Result getCurrentValue() throws IOException, InterruptedException {<a name="line.124"></a> -<span class="sourceLineNo">125</span> return trr.getCurrentValue();<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> @Override<a name="line.128"></a> -<span class="sourceLineNo">129</span> public float getProgress() throws IOException, InterruptedException {<a name="line.129"></a> -<span class="sourceLineNo">130</span> return trr.getProgress();<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> @Override<a name="line.133"></a> -<span class="sourceLineNo">134</span> public void initialize(InputSplit inputsplit, TaskAttemptContext context)<a name="line.134"></a> -<span class="sourceLineNo">135</span> throws IOException, InterruptedException {<a name="line.135"></a> -<span class="sourceLineNo">136</span> trr.initialize(inputsplit, context);<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 boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.140"></a> -<span class="sourceLineNo">141</span> return trr.nextKeyValue();<a name="line.141"></a> -<span class="sourceLineNo">142</span> }<a name="line.142"></a> -<span class="sourceLineNo">143</span> };<a name="line.143"></a> -<span class="sourceLineNo">144</span> } catch (IOException ioe) {<a name="line.144"></a> -<span class="sourceLineNo">145</span> // If there is an exception make sure that all<a name="line.145"></a> -<span class="sourceLineNo">146</span> // resources are closed and released.<a name="line.146"></a> -<span class="sourceLineNo">147</span> trr.close();<a name="line.147"></a> -<span class="sourceLineNo">148</span> if (connection != null) {<a name="line.148"></a> -<span class="sourceLineNo">149</span> connection.close();<a name="line.149"></a> -<span class="sourceLineNo">150</span> }<a name="line.150"></a> -<span class="sourceLineNo">151</span> throw ioe;<a name="line.151"></a> -<span class="sourceLineNo">152</span> }<a name="line.152"></a> -<span class="sourceLineNo">153</span> }<a name="line.153"></a> -<span class="sourceLineNo">154</span><a name="line.154"></a> -<span class="sourceLineNo">155</span> /**<a name="line.155"></a> -<span class="sourceLineNo">156</span> * Calculates the splits that will serve as input for the map tasks. The<a name="line.156"></a> -<span class="sourceLineNo">157</span> * number of splits matches the number of regions in a table.<a name="line.157"></a> -<span class="sourceLineNo">158</span> *<a name="line.158"></a> -<span class="sourceLineNo">159</span> * @param context The current job context.<a name="line.159"></a> -<span class="sourceLineNo">160</span> * @return The list of input splits.<a name="line.160"></a> -<span class="sourceLineNo">161</span> * @throws IOException When creating the list of splits fails.<a name="line.161"></a> -<span class="sourceLineNo">162</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(org.apache.hadoop.mapreduce.JobContext)<a name="line.162"></a> -<span class="sourceLineNo">163</span> */<a name="line.163"></a> -<span class="sourceLineNo">164</span> @Override<a name="line.164"></a> -<span class="sourceLineNo">165</span> public List<InputSplit> getSplits(JobContext context) throws IOException {<a name="line.165"></a> -<span class="sourceLineNo">166</span> if (scans.isEmpty()) {<a name="line.166"></a> -<span class="sourceLineNo">167</span> throw new IOException("No scans were provided.");<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> Map<TableName, List<Scan>> tableMaps = new HashMap<TableName, List<Scan>>();<a name="line.170"></a> -<span class="sourceLineNo">171</span> for (Scan scan : scans) {<a name="line.171"></a> -<span class="sourceLineNo">172</span> byte[] tableNameBytes = scan.getAttribute(Scan.SCAN_ATTRIBUTES_TABLE_NAME);<a name="line.172"></a> -<span class="sourceLineNo">173</span> if (tableNameBytes == null)<a name="line.173"></a> -<span class="sourceLineNo">174</span> throw new IOException("A scan object did not have a table name");<a name="line.174"></a> -<span class="sourceLineNo">175</span><a name="line.175"></a> -<span class="sourceLineNo">176</span> TableName tableName = TableName.valueOf(tableNameBytes);<a name="line.176"></a> -<span class="sourceLineNo">177</span><a name="line.177"></a> -<span class="sourceLineNo">178</span> List<Scan> scanList = tableMaps.get(tableName);<a name="line.178"></a> -<span class="sourceLineNo">179</span> if (scanList == null) {<a name="line.179"></a> -<span class="sourceLineNo">180</span> scanList = new ArrayList<Scan>();<a name="line.180"></a> -<span class="sourceLineNo">181</span> tableMaps.put(tableName, scanList);<a name="line.181"></a> -<span class="sourceLineNo">182</span> }<a name="line.182"></a> -<span class="sourceLineNo">183</span> scanList.add(scan);<a name="line.183"></a> -<span class="sourceLineNo">184</span> }<a name="line.184"></a> -<span class="sourceLineNo">185</span><a name="line.185"></a> -<span class="sourceLineNo">186</span> List<InputSplit> splits = new ArrayList<InputSplit>();<a name="line.186"></a> -<span class="sourceLineNo">187</span> Iterator iter = tableMaps.entrySet().iterator();<a name="line.187"></a> -<span class="sourceLineNo">188</span> while (iter.hasNext()) {<a name="line.188"></a> -<span class="sourceLineNo">189</span> Map.Entry<TableName, List<Scan>> entry = (Map.Entry<TableName, List<Scan>>) iter.next();<a name="line.189"></a> -<span class="sourceLineNo">190</span> TableName tableName = entry.getKey();<a name="line.190"></a> -<span class="sourceLineNo">191</span> List<Scan> scanList = entry.getValue();<a name="line.191"></a> -<span class="sourceLineNo">192</span><a name="line.192"></a> -<span class="sourceLineNo">193</span> try (Connection conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.193"></a> -<span class="sourceLineNo">194</span> Table table = conn.getTable(tableName);<a name="line.194"></a> -<span class="sourceLineNo">195</span> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {<a name="line.195"></a> -<span class="sourceLineNo">196</span> RegionSizeCalculator sizeCalculator = new RegionSizeCalculator(<a name="line.196"></a> -<span class="sourceLineNo">197</span> regionLocator, conn.getAdmin());<a name="line.197"></a> -<span class="sourceLineNo">198</span> Pair<byte[][], byte[][]> keys = regionLocator.getStartEndKeys();<a name="line.198"></a> -<span class="sourceLineNo">199</span> for (Scan scan : scanList) {<a name="line.199"></a> -<span class="sourceLineNo">200</span> if (keys == null || keys.getFirst() == null || keys.getFirst().length == 0) {<a name="line.200"></a> -<span class="sourceLineNo">201</span> throw new IOException("Expecting at least one region for table : "<a name="line.201"></a> -<span class="sourceLineNo">202</span> + tableName.getNameAsString());<a name="line.202"></a> -<span class="sourceLineNo">203</span> }<a name="line.203"></a> -<span class="sourceLineNo">204</span> int count = 0;<a name="line.204"></a> -<span class="sourceLineNo">205</span><a name="line.205"></a> -<span class="sourceLineNo">206</span> byte[] startRow = scan.getStartRow();<a name="line.206"></a> -<span class="sourceLineNo">207</span> byte[] stopRow = scan.getStopRow();<a name="line.207"></a> -<span class="sourceLineNo">208</span><a name="line.208"></a> -<span class="sourceLineNo">209</span> for (int i = 0; i < keys.getFirst().length; i++) {<a name="line.209"></a> -<span class="sourceLineNo">210</span> if (!includeRegionInSplit(keys.getFirst()[i], keys.getSecond()[i])) {<a name="line.210"></a> -<span class="sourceLineNo">211</span> continue;<a name="line.211"></a> -<span class="sourceLineNo">212</span> }<a name="line.212"></a> -<span class="sourceLineNo">213</span><a name="line.213"></a> -<span class="sourceLineNo">214</span> if ((startRow.length == 0 || keys.getSecond()[i].length == 0 ||<a name="line.214"></a> -<span class="sourceLineNo">215</span> Bytes.compareTo(startRow, keys.getSecond()[i]) < 0) &&<a name="line.215"></a> -<span class="sourceLineNo">216</span> (stopRow.length == 0 || Bytes.compareTo(stopRow,<a name="line.216"></a> -<span class="sourceLineNo">217</span> keys.getFirst()[i]) > 0)) {<a name="line.217"></a> -<span class="sourceLineNo">218</span> byte[] splitStart = startRow.length == 0 ||<a name="line.218"></a> -<span class="sourceLineNo">219</span> Bytes.compareTo(keys.getFirst()[i], startRow) >= 0 ?<a name="line.219"></a> -<span class="sourceLineNo">220</span> keys.getFirst()[i] : startRow;<a name="line.220"></a> -<span class="sourceLineNo">221</span> byte[] splitStop = (stopRow.length == 0 ||<a name="line.221"></a> -<span class="sourceLineNo">222</span> Bytes.compareTo(keys.getSecond()[i], stopRow) <= 0) &&<a name="line.222"></a> -<span class="sourceLineNo">223</span> keys.getSecond()[i].length > 0 ?<a name="line.223"></a> -<span class="sourceLineNo">224</span> keys.getSecond()[i] : stopRow;<a name="line.224"></a> -<span class="sourceLineNo">225</span><a name="line.225"></a> -<span class="sourceLineNo">226</span> HRegionLocation hregionLocation = regionLocator.getRegionLocation(<a name="line.226"></a> -<span class="sourceLineNo">227</span> keys.getFirst()[i], false);<a name="line.227"></a> -<span class="sourceLineNo">228</span> String regionHostname = hregionLocation.getHostname();<a name="line.228"></a> -<span class="sourceLineNo">229</span> HRegionInfo regionInfo = hregionLocation.getRegionInfo();<a name="line.229"></a> -<span class="sourceLineNo">230</span> long regionSize = sizeCalculator.getRegionSize(<a name="line.230"></a> -<span class="sourceLineNo">231</span> regionInfo.getRegionName());<a name="line.231"></a> -<span class="sourceLineNo">232</span><a name="line.232"></a> -<span class="sourceLineNo">233</span> TableSplit split = new TableSplit(table.getName(),<a name="line.233"></a> -<span class="sourceLineNo">234</span> scan, splitStart, splitStop, regionHostname, regionSize);<a name="line.234"></a> -<span class="sourceLineNo">235</span><a name="line.235"></a> -<span class="sourceLineNo">236</span> splits.add(split);<a name="line.236"></a> -<span class="sourceLineNo">237</span><a name="line.237"></a> -<span class="sourceLineNo">238</span> if (LOG.isDebugEnabled())<a name="line.238"></a> -<span class="sourceLineNo">239</span> LOG.debug("getSplits: split -> " + (count++) + " -> " + split);<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> }<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><a name="line.245"></a> -<span class="sourceLineNo">246</span> return splits;<a name="line.246"></a> -<span class="sourceLineNo">247</span> }<a name="line.247"></a> -<span class="sourceLineNo">248</span><a name="line.248"></a> -<span class="sourceLineNo">249</span> /**<a name="line.249"></a> -<span class="sourceLineNo">250</span> * Test if the given region is to be included in the InputSplit while<a name="line.250"></a> -<span class="sourceLineNo">251</span> * splitting the regions of a table.<a name="line.251"></a> -<span class="sourceLineNo">252</span> * <p><a name="line.252"></a> -<span class="sourceLineNo">253</span> * This optimization is effective when there is a specific reasoning to<a name="line.253"></a> -<span class="sourceLineNo">254</span> * exclude an entire region from the M-R job, (and hence, not contributing to<a name="line.254"></a> -<span class="sourceLineNo">255</span> * the InputSplit), given the start and end keys of the same. <br><a name="line.255"></a> -<span class="sourceLineNo">256</span> * Useful when we need to remember the last-processed top record and revisit<a name="line.256"></a> -<span class="sourceLineNo">257</span> * the [last, current) interval for M-R processing, continuously. In addition<a name="line.257"></a> -<span class="sourceLineNo">258</span> * to reducing InputSplits, reduces the load on the region server as well, due<a name="line.258"></a> -<span class="sourceLineNo">259</span> * to the ordering of the keys. <br><a name="line.259"></a> -<span class="sourceLineNo">260</span> * <br><a name="line.260"></a> -<span class="sourceLineNo">261</span> * Note: It is possible that <code>endKey.length() == 0 </code> , for the last<a name="line.261"></a> -<span class="sourceLineNo">262</span> * (recent) region. <br><a name="line.262"></a> -<span class="sourceLineNo">263</span> * Override this method, if you want to bulk exclude regions altogether from<a name="line.263"></a> -<span class="sourceLineNo">264</span> * M-R. By default, no region is excluded( i.e. all regions are included).<a name="line.264"></a> -<span class="sourceLineNo">265</span> *<a name="line.265"></a> -<span class="sourceLineNo">266</span> * @param startKey Start key of the region<a name="line.266"></a> -<span class="sourceLineNo">267</span> * @param endKey End key of the region<a name="line.267"></a> -<span class="sourceLineNo">268</span> * @return true, if this region needs to be included as part of the input<a name="line.268"></a> -<span class="sourceLineNo">269</span> * (default).<a name="line.269"></a> -<span class="sourceLineNo">270</span> */<a name="line.270"></a> -<span class="sourceLineNo">271</span> protected boolean includeRegionInSplit(final byte[] startKey,<a name="line.271"></a> -<span class="sourceLineNo">272</span> final byte[] endKey) {<a name="line.272"></a> -<span class="sourceLineNo">273</span> return true;<a name="line.273"></a> -<span class="sourceLineNo">274</span> }<a name="line.274"></a> -<span class="sourceLineNo">275</span><a name="line.275"></a> -<span class="sourceLineNo">276</span> /**<a name="line.276"></a> -<span class="sourceLineNo">277</span> * Allows subclasses to get the list of {@link Scan} objects.<a name="line.277"></a> -<span class="sourceLineNo">278</span> */<a name="line.278"></a> -<span class="sourceLineNo">279</span> protected List<Scan> getScans() {<a name="line.279"></a> -<span class="sourceLineNo">280</span> return this.scans;<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> /**<a name="line.283"></a> -<span class="sourceLineNo">284</span> * Allows subclasses to set the list of {@link Scan} objects.<a name="line.284"></a> -<span class="sourceLineNo">285</span> *<a name="line.285"></a> -<span class="sourceLineNo">286</span> * @param scans The list of {@link Scan} used to define the input<a name="line.286"></a> -<span class="sourceLineNo">287</span> */<a name="line.287"></a> -<span class="sourceLineNo">288</span> protected void setScans(List<Scan> scans) {<a name="line.288"></a> -<span class="sourceLineNo">289</span> this.scans = scans;<a name="line.289"></a> -<span class="sourceLineNo">290</span> }<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> * Allows subclasses to set the {@link TableRecordReader}.<a name="line.293"></a> -<span class="sourceLineNo">294</span> *<a name="line.294"></a> -<span class="sourceLineNo">295</span> * @param tableRecordReader A different {@link TableRecordReader}<a name="line.295"></a> -<span class="sourceLineNo">296</span> * implementation.<a name="line.296"></a> -<span class="sourceLineNo">297</span> */<a name="line.297"></a> -<span class="sourceLineNo">298</span> protected void setTableRecordReader(TableRecordReader tableRecordReader) {<a name="line.298"></a> -<span class="sourceLineNo">299</span> this.tableRecordReader = tableRecordReader;<a name="line.299"></a> -<span class="sourceLineNo">300</span> }<a name="line.300"></a> -<span class="sourceLineNo">301</span>}<a name="line.301"></a> +<span class="sourceLineNo">113</span> connection.close();<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> @Override<a name="line.116"></a> +<span class="sourceLineNo">117</span> public ImmutableBytesWritable getCurrentKey() throws IOException, InterruptedException {<a name="line.117"></a> +<span class="sourceLineNo">118</span> return trr.getCurrentKey();<a name="line.118"></a> +<span class="sourceLineNo">119</span> }<a name="line.119"></a> +<span class="sourceLineNo">120</span><a name="line.120"></a> +<span class="sourceLineNo">121</span> @Override<a name="line.121"></a> +<span class="sourceLineNo">122</span> public Result getCurrentValue() throws IOException, InterruptedException {<a name="line.122"></a> +<span class="sourceLineNo">123</span> return trr.getCurrentValue();<a name="line.123"></a> +<span class="sourceLineNo">124</span> }<a name="line.124"></a> +<span class="sourceLineNo">125</span><a name="line.125"></a> +<span class="sourceLineNo">126</span> @Override<a name="line.126"></a> +<span class="sourceLineNo">127</span> public float getProgress() throws IOException, InterruptedException {<a name="line.127"></a> +<span class="sourceLineNo">128</span> return trr.getProgress();<a name="line.128"></a> +<span class="sourceLineNo">129</span> }<a name="line.129"></a> +<span class="sourceLineNo">130</span><a name="line.130"></a> +<span class="sourceLineNo">131</span> @Override<a name="line.131"></a> +<span class="sourceLineNo">132</span> public void initialize(InputSplit inputsplit, TaskAttemptContext context)<a name="line.132"></a> +<span class="sourceLineNo">133</span> throws IOException, InterruptedException {<a name="line.133"></a> +<span class="sourceLineNo">134</span> trr.initialize(inputsplit, context);<a name="line.134"></a> +<span class="sourceLineNo">135</span> }<a name="line.135"></a> +<span class="sourceLineNo">136</span><a name="line.136"></a> +<span class="sourceLineNo">137</span> @Override<a name="line.137"></a> +<span class="sourceLineNo">138</span> public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.138"></a> +<span class="sourceLineNo">139</span> return trr.nextKeyValue();<a name="line.139"></a> +<span class="sourceLineNo">140</span> }<a name="line.140"></a> +<span class="sourceLineNo">141</span> };<a name="line.141"></a> +<span class="sourceLineNo">142</span> } catch (IOException ioe) {<a name="line.142"></a> +<span class="sourceLineNo">143</span> // If there is an exception make sure that all<a name="line.143"></a> +<span class="sourceLineNo">144</span> // resources are closed and released.<a name="line.144"></a> +<span class="sourceLineNo">145</span> trr.close();<a name="line.145"></a> +<span class="sourceLineNo">146</span> connection.close();<a name="line.146"></a> +<span class="sourceLineNo">147</span> throw ioe;<a name="line.147"></a> +<span class="sourceLineNo">148</span> }<a name="line.148"></a> +<span class="sourceLineNo">149</span> }<a name="line.149"></a> +<span class="sourceLineNo">150</span><a name="line.150"></a> +<span class="sourceLineNo">151</span> /**<a name="line.151"></a> +<span class="sourceLineNo">152</span> * Calculates the splits that will serve as input for the map tasks. The<a name="line.152"></a> +<span class="sourceLineNo">153</span> * number of splits matches the number of regions in a table.<a name="line.153"></a> +<span class="sourceLineNo">154</span> *<a name="line.154"></a> +<span class="sourceLineNo">155</span> * @param context The current job context.<a name="line.155"></a> +<span class="sourceLineNo">156</span> * @return The list of input splits.<a name="line.156"></a> +<span class="sourceLineNo">157</span> * @throws IOException When creating the list of splits fails.<a name="line.157"></a> +<span class="sourceLineNo">158</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(org.apache.hadoop.mapreduce.JobContext)<a name="line.158"></a> +<span class="sourceLineNo">159</span> */<a name="line.159"></a> +<span class="sourceLineNo">160</span> @Override<a name="line.160"></a> +<span class="sourceLineNo">161</span> public List<InputSplit> getSplits(JobContext context) throws IOException {<a name="line.161"></a> +<span class="sourceLineNo">162</span> if (scans.isEmpty()) {<a name="line.162"></a> +<span class="sourceLineNo">163</span> throw new IOException("No scans were provided.");<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> Map<TableName, List<Scan>> tableMaps = new HashMap<TableName, List<Scan>>();<a name="line.166"></a> +<span class="sourceLineNo">167</span> for (Scan scan : scans) {<a name="line.167"></a> +<span class="sourceLineNo">168</span> byte[] tableNameBytes = scan.getAttribute(Scan.SCAN_ATTRIBUTES_TABLE_NAME);<a name="line.168"></a> +<span class="sourceLineNo">169</span> if (tableNameBytes == null)<a name="line.169"></a> +<span class="sourceLineNo">170</span> throw new IOException("A scan object did not have a table name");<a name="line.170"></a> +<span class="sourceLineNo">171</span><a name="line.171"></a> +<span class="sourceLineNo">172</span> TableName tableName = TableName.valueOf(tableNameBytes);<a name="line.172"></a> +<span class="sourceLineNo">173</span><a name="line.173"></a> +<span class="sourceLineNo">174</span> List<Scan> scanList = tableMaps.get(tableName);<a name="line.174"></a> +<span class="sourceLineNo">175</span> if (scanList == null) {<a name="line.175"></a> +<span class="sourceLineNo">176</span> scanList = new ArrayList<Scan>();<a name="line.176"></a> +<span class="sourceLineNo">177</span> tableMaps.put(tableName, scanList);<a name="line.177"></a> +<span class="sourceLineNo">178</span> }<a name="line.178"></a> +<span class="sourceLineNo">179</span> scanList.add(scan);<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> List<InputSplit> splits = new ArrayList<InputSplit>();<a name="line.182"></a> +<span class="sourceLineNo">183</span> Iterator iter = tableMaps.entrySet().iterator();<a name="line.183"></a> +<span class="sourceLineNo">184</span> while (iter.hasNext()) {<a name="line.184"></a> +<span class="sourceLineNo">185</span> Map.Entry<TableName, List<Scan>> entry = (Map.Entry<TableName, List<Scan>>) iter.next();<a name="line.185"></a> +<span class="sourceLineNo">186</span> TableName tableName = entry.getKey();<a name="line.186"></a> +<span class="sourceLineNo">187</span> List<Scan> scanList = entry.getValue();<a name="line.187"></a> +<span class="sourceLineNo">188</span><a name="line.188"></a> +<span class="sourceLineNo">189</span> try (Connection conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.189"></a> +<span class="sourceLineNo">190</span> Table table = conn.getTable(tableName);<a name="line.190"></a> +<span class="sourceLineNo">191</span> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {<a name="line.191"></a> +<span class="sourceLineNo">192</span> RegionSizeCalculator sizeCalculator = new RegionSizeCalculator(<a name="line.192"></a> +<span class="sourceLineNo">193</span> regionLocator, conn.getAdmin());<a name="line.193"></a> +<span class="sourceLineNo">194</span> Pair<byte[][], byte[][]> keys = regionLocator.getStartEndKeys();<a name="line.194"></a> +<span class="sourceLineNo">195</span> for (Scan scan : scanList) {<a name="line.195"></a> +<span class="sourceLineNo">196</span> if (keys == null || keys.getFirst() == null || keys.getFirst().length == 0) {<a name="line.196"></a> +<span class="sourceLineNo">197</span> throw new IOException("Expecting at least one region for table : "<a name="line.197"></a> +<span class="sourceLineNo">198</span> + tableName.getNameAsString());<a name="line.198"></a> +<span class="sourceLineNo">199</span> }<a name="line.199"></a> +<span class="sourceLineNo">200</span> int count = 0;<a name="line.200"></a> +<span class="sourceLineNo">201</span><a name="line.201"></a> +<span class="sourceLineNo">202</span> byte[] startRow = scan.getStartRow();<a name="line.202"></a> +<span class="sourceLineNo">203</span> byte[] stopRow = scan.getStopRow();<a name="line.203"></a> +<span class="sourceLineNo">204</span><a name="line.204"></a> +<span class="sourceLineNo">205</span> for (int i = 0; i < keys.getFirst().length; i++) {<a name="line.205"></a> +<span class="sourceLineNo">206</span> if (!includeRegionInSplit(keys.getFirst()[i], keys.getSecond()[i])) {<a name="line.206"></a> +<span class="sourceLineNo">207</span> continue;<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> if ((startRow.length == 0 || keys.getSecond()[i].length == 0 ||<a name="line.210"></a> +<span class="sourceLineNo">211</span> Bytes.compareTo(startRow, keys.getSecond()[i]) < 0) &&<a name="line.211"></a> +<span class="sourceLineNo">212</span> (stopRow.length == 0 || Bytes.compareTo(stopRow,<a name="line.212"></a> +<span class="sourceLineNo">213</span> keys.getFirst()[i]) > 0)) {<a name="line.213"></a> +<span class="sourceLineNo">214</span> byte[] splitStart = startRow.length == 0 ||<a name="line.214"></a> +<span class="sourceLineNo">215</span> Bytes.compareTo(keys.getFirst()[i], startRow) >= 0 ?<a name="line.215"></a> +<span class="sourceLineNo">216</span> keys.getFirst()[i] : startRow;<a name="line.216"></a> +<span class="sourceLineNo">217</span> byte[] splitStop = (stopRow.length == 0 ||<a name="line.217"></a> +<span class="sourceLineNo">218</span> Bytes.compareTo(keys.getSecond()[i], stopRow) <= 0) &&<a name="line.218"></a> +<span class="sourceLineNo">219</span> keys.getSecond()[i].length > 0 ?<a name="line.219"></a> +<span class="sourceLineNo">220</span> keys.getSecond()[i] : stopRow;<a name="line.220"></a> +<span class="sourceLineNo">221</span><a name="line.221"></a> +<span class="sourceLineNo">222</span> HRegionLocation hregionLocation = regionLocator.getRegionLocation(<a name="line.222"></a> +<span class="sourceLineNo">223</span> keys.getFirst()[i], false);<a name="line.223"></a> +<span class="sourceLineNo">224</span> String regionHostname = hregionLocation.getHostname();<a name="line.224"></a> +<span class="sourceLineNo">225</span> HRegionInfo regionInfo = hregionLocation.getRegionInfo();<a name="line.225"></a> +<span class="sourceLineNo">226</span> long regionSize = sizeCalculator.getRegionSize(<a name="line.226"></a> +<span class="sourceLineNo">227</span> regionInfo.getRegionName());<a name="line.227"></a> +<span class="sourceLineNo">228</span><a name="line.228"></a> +<span class="sourceLineNo">229</span> TableSplit split = new TableSplit(table.getName(),<a name="line.229"></a> +<span class="sourceLineNo">230</span> scan, splitStart, splitStop, regionHostname, regionSize);<a name="line.230"></a> +<span class="sourceLineNo">231</span><a name="line.231"></a> +<span class="sourceLineNo">232</span> splits.add(split);<a name="line.232"></a> +<span class="sourceLineNo">233</span><a name="line.233"></a> +<span class="sourceLineNo">234</span> if (LOG.isDebugEnabled())<a name="line.234"></a> +<span class="sourceLineNo">235</span> LOG.debug("getSplits: split -> " + (count++) + " -> " + split);<a name="line.235"></a> +<span class="sourceLineNo">236</span> }<a name="line.236"></a> +<span class="sourceLineNo">237</span> }<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> }<a name="line.240"></a> +<span class="sourceLineNo">241</span><a name="line.241"></a> +<span class="sourceLineNo">242</span> return splits;<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> /**<a name="line.245"></a> +<span class="sourceLineNo">246</span> * Test if the given region is to be included in the InputSplit while<a name="line.246"></a> +<span class="sourceLineNo">247</span> * splitting the regions of a table.<a name="line.247"></a> +<span class="sourceLineNo">248</span> * <p><a name="line.248"></a> +<span class="sourceLineNo">249</span> * This optimization is effective when there is a specific reasoning to<a name="line.249"></a> +<span class="sourceLineNo">250</span> * exclude an entire region from the M-R job, (and hence, not contributing to<a name="line.250"></a> +<span class="sourceLineNo">251</span> * the InputSplit), given the start and end keys of the same. <br><a name="line.251"></a> +<span class="sourceLineNo">252</span> * Useful when we need to remember the last-processed top record and revisit<a name="line.252"></a> +<span class="sourceLineNo">253</span> * the [last, current) interval for M-R processing, continuously. In addition<a name="line.253"></a> +<span class="sourceLineNo">254</span> * to reducing InputSplits, reduces the load on the region server as well, due<a name="line.254"></a> +<span class="sourceLineNo">255</span> * to the ordering of the keys. <br><a name="line.255"></a> +<span class="sourceLineNo">256</span> * <br><a name="line.256"></a> +<span class="sourceLineNo">257</span> * Note: It is possible that <code>endKey.length() == 0 </code> , for the last<a name="line.257"></a> +<span class="sourceLineNo">258</span> * (recent) region. <br><a name="line.258"></a> +<span class="sourceLineNo">259</span> * Override this method, if you want to bulk exclude regions altogether from<a name="line.259"></a> +<span class="sourceLineNo">260</span> * M-R. By default, no region is excluded( i.e. all regions are included).<a name="line.260"></a> +<span class="sourceLineNo">261</span> *<a name="line.261"></a> +<span class="sourceLineNo">262</span> * @param startKey Start key of the region<a name="line.262"></a> +<span class="sourceLineNo">263</span> * @param endKey End key of the region<a name="line.263"></a> +<span class="sourceLineNo">264</span> * @return true, if this region needs to be included as part of the input<a name="line.264"></a> +<span class="sourceLineNo">265</span> * (default).<a name="line.265"></a> +<span class="sourceLineNo">266</span> */<a name="line.266"></a> +<span class="sourceLineNo">267</span> protected boolean includeRegionInSplit(final byte[] startKey,<a name="line.267"></a> +<span class="sourceLineNo">268</span> final byte[] endKey) {<a name="line.268"></a> +<span class="sourceLineNo">269</span> return true;<a name="line.269"></a> +<span class="sourceLineNo">270</span> }<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> * Allows subclasses to get the list of {@link Scan} objects.<a name="line.273"></a> +<span class="sourceLineNo">274</span> */<a name="line.274"></a> +<span class="sourceLineNo">275</span> protected List<Scan> getScans() {<a name="line.275"></a> +<span class="sourceLineNo">276</span> return this.scans;<a name="line.276"></a> +<span class="sourceLineNo">277</span> }<a name="line.277"></a> +<span class="sourceLineNo">278</span><a name="line.278"></a> +<span class="sourceLineNo">279</span> /**<a name="line.279"></a> +<span class="sourceLineNo">280</span> * Allows subclasses to set the list of {@link Scan} objects.<a name="line.280"></a> +<span class="sourceLineNo">281</span> *<a name="line.281"></a> +<span class="sourceLineNo">282</span> * @param scans The list of {@link Scan} used to define the input<a name="line.282"></a> +<span class="sourceLineNo">283</span> */<a name="line.283"></a> +<span class="sourceLineNo">284</span> protected void setScans(List<Scan> scans) {<a name="line.284"></a> +<span class="sourceLineNo">285</span> this.scans = scans;<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">288</span> /**<a name="line.288"></a> +<span class="sourceLineNo">289</span> * Allows subclasses to set the {@link TableRecordReader}.<a name="line.289"></a> +<span class="sourceLineNo">290</span> *<a name="line.290"></a> +<span class="sourceLineNo">291</span> * @param tableRecordReader A different {@link TableRecordReader}<a name="line.291"></a> +<span class="sourceLineNo">292</span> * implementation.<a name="line.292"></a> +<span class="sourceLineNo">293</span> */<a name="line.293"></a> +<span class="sourceLineNo">294</span> protected void setTableRecordReader(TableRecordReader tableRecordReader) {<a name="line.294"></a> +<span class="sourceLineNo">295</span> this.tableRecordReader = tableRecordReader;<a name="line.295"></a> +<span class="sourceLineNo">296</span> }<a name="line.296"></a> +<span class="sourceLineNo">297</span>}<a name="line.297"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/75eda567/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html index 79ac839..4c12385 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html @@ -119,170 +119,172 @@ <span class="sourceLineNo">111</span> * org.apache.hadoop.conf.Configuration)<a name="line.111"></a> <span class="sourceLineNo">112</span> */<a name="line.112"></a> <span class="sourceLineNo">113</span> @Override<a name="line.113"></a> -<span class="sourceLineNo">114</span> public void setConf(Configuration configuration) {<a name="line.114"></a> -<span class="sourceLineNo">115</span> this.conf = configuration;<a name="line.115"></a> -<span class="sourceLineNo">116</span><a name="line.116"></a> -<span class="sourceLineNo">117</span> Scan scan = null;<a name="line.117"></a> +<span class="sourceLineNo">114</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value="REC_CATCH_EXCEPTION",<a name="line.114"></a> +<span class="sourceLineNo">115</span> justification="Intentional")<a name="line.115"></a> +<span class="sourceLineNo">116</span> public void setConf(Configuration configuration) {<a name="line.116"></a> +<span class="sourceLineNo">117</span> this.conf = configuration;<a name="line.117"></a> <span class="sourceLineNo">118</span><a name="line.118"></a> -<span class="sourceLineNo">119</span> if (conf.get(SCAN) != null) {<a name="line.119"></a> -<span class="sourceLineNo">120</span> try {<a name="line.120"></a> -<span class="sourceLineNo">121</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(SCAN));<a name="line.121"></a> -<span class="sourceLineNo">122</span> } catch (IOException e) {<a name="line.122"></a> -<span class="sourceLineNo">123</span> LOG.error("An error occurred.", e);<a name="line.123"></a> -<span class="sourceLineNo">124</span> }<a name="line.124"></a> -<span class="sourceLineNo">125</span> } else {<a name="line.125"></a> -<span class="sourceLineNo">126</span> try {<a name="line.126"></a> -<span class="sourceLineNo">127</span> scan = new Scan();<a name="line.127"></a> -<span class="sourceLineNo">128</span><a name="line.128"></a> -<span class="sourceLineNo">129</span> if (conf.get(SCAN_ROW_START) != null) {<a name="line.129"></a> -<span class="sourceLineNo">130</span> scan.setStartRow(Bytes.toBytes(conf.get(SCAN_ROW_START)));<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> if (conf.get(SCAN_ROW_STOP) != null) {<a name="line.133"></a> -<span class="sourceLineNo">134</span> scan.setStopRow(Bytes.toBytes(conf.get(SCAN_ROW_STOP)));<a name="line.134"></a> -<span class="sourceLineNo">135</span> }<a name="line.135"></a> -<span class="sourceLineNo">136</span><a name="line.136"></a> -<span class="sourceLineNo">137</span> if (conf.get(SCAN_COLUMNS) != null) {<a name="line.137"></a> -<span class="sourceLineNo">138</span> addColumns(scan, conf.get(SCAN_COLUMNS));<a name="line.138"></a> -<span class="sourceLineNo">139</span> }<a name="line.139"></a> -<span class="sourceLineNo">140</span><a name="line.140"></a> -<span class="sourceLineNo">141</span> if (conf.get(SCAN_COLUMN_FAMILY) != null) {<a name="line.141"></a> -<span class="sourceLineNo">142</span> scan.addFamily(Bytes.toBytes(conf.get(SCAN_COLUMN_FAMILY)));<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> if (conf.get(SCAN_TIMESTAMP) != null) {<a name="line.145"></a> -<span class="sourceLineNo">146</span> scan.setTimeStamp(Long.parseLong(conf.get(SCAN_TIMESTAMP)));<a name="line.146"></a> -<span class="sourceLineNo">147</span> }<a name="line.147"></a> -<span class="sourceLineNo">148</span><a name="line.148"></a> -<span class="sourceLineNo">149</span> if (conf.get(SCAN_TIMERANGE_START) != null && conf.get(SCAN_TIMERANGE_END) != null) {<a name="line.149"></a> -<span class="sourceLineNo">150</span> scan.setTimeRange(<a name="line.150"></a> -<span class="sourceLineNo">151</span> Long.parseLong(conf.get(SCAN_TIMERANGE_START)),<a name="line.151"></a> -<span class="sourceLineNo">152</span> Long.parseLong(conf.get(SCAN_TIMERANGE_END)));<a name="line.152"></a> -<span class="sourceLineNo">153</span> }<a name="line.153"></a> -<span class="sourceLineNo">154</span><a name="line.154"></a> -<span class="sourceLineNo">155</span> if (conf.get(SCAN_MAXVERSIONS) != null) {<a name="line.155"></a> -<span class="sourceLineNo">156</span> scan.setMaxVersions(Integer.parseInt(conf.get(SCAN_MAXVERSIONS)));<a name="line.156"></a> -<span class="sourceLineNo">157</span> }<a name="line.157"></a> -<span class="sourceLineNo">158</span><a name="line.158"></a> -<span class="sourceLineNo">159</span> if (conf.get(SCAN_CACHEDROWS) != null) {<a name="line.159"></a> -<span class="sourceLineNo">160</span> scan.setCaching(Integer.parseInt(conf.get(SCAN_CACHEDROWS)));<a name="line.160"></a> -<span class="sourceLineNo">161</span> }<a name="line.161"></a> -<span class="sourceLineNo">162</span><a name="line.162"></a> -<span class="sourceLineNo">163</span> if (conf.get(SCAN_BATCHSIZE) != null) {<a name="line.163"></a> -<span class="sourceLineNo">164</span> scan.setBatch(Integer.parseInt(conf.get(SCAN_BATCHSIZE)));<a name="line.164"></a> -<span class="sourceLineNo">165</span> }<a name="line.165"></a> -<span class="sourceLineNo">166</span><a name="line.166"></a> -<span class="sourceLineNo">167</span> // false by default, full table scans generate too much BC churn<a name="line.167"></a> -<span class="sourceLineNo">168</span> scan.setCacheBlocks((conf.getBoolean(SCAN_CACHEBLOCKS, false)));<a name="line.168"></a> -<span class="sourceLineNo">169</span> } catch (Exception e) {<a name="line.169"></a> -<span class="sourceLineNo">170</span> LOG.error(StringUtils.stringifyException(e));<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><a name="line.173"></a> -<span class="sourceLineNo">174</span> setScan(scan);<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> @Override<a name="line.177"></a> -<span class="sourceLineNo">178</span> protected void initialize(JobContext context) throws IOException {<a name="line.178"></a> -<span class="sourceLineNo">179</span> // Do we have to worry about mis-matches between the Configuration from setConf and the one<a name="line.179"></a> -<span class="sourceLineNo">180</span> // in this context?<a name="line.180"></a> -<span class="sourceLineNo">181</span> TableName tableName = TableName.valueOf(conf.get(INPUT_TABLE));<a name="line.181"></a> -<span class="sourceLineNo">182</span> try {<a name="line.182"></a> -<span class="sourceLineNo">183</span> initializeTable(ConnectionFactory.createConnection(new Configuration(conf)), tableName);<a name="line.183"></a> -<span class="sourceLineNo">184</span> } catch (Exception e) {<a name="line.184"></a> -<span class="sourceLineNo">185</span> LOG.error(StringUtils.stringifyException(e));<a name="line.185"></a> -<span class="sourceLineNo">186</span> }<a name="line.186"></a> -<span class="sourceLineNo">187</span> }<a name="line.187"></a> -<span class="sourceLineNo">188</span><a name="line.188"></a> -<span class="sourceLineNo">189</span> /**<a name="line.189"></a> -<span class="sourceLineNo">190</span> * Parses a combined family and qualifier and adds either both or just the<a name="line.190"></a> -<span class="sourceLineNo">191</span> * family in case there is no qualifier. This assumes the older colon<a name="line.191"></a> -<span class="sourceLineNo">192</span> * divided notation, e.g. "family:qualifier".<a name="line.192"></a> -<span class="sourceLineNo">193</span> *<a name="line.193"></a> -<span class="sourceLineNo">194</span> * @param scan The Scan to update.<a name="line.194"></a> -<span class="sourceLineNo">195</span> * @param familyAndQualifier family and qualifier<a name="line.195"></a> -<span class="sourceLineNo">196</span> * @throws IllegalArgumentException When familyAndQualifier is invalid.<a name="line.196"></a> -<span class="sourceLineNo">197</span> */<a name="line.197"></a> -<span class="sourceLineNo">198</span> private static void addColumn(Scan scan, byte[] familyAndQualifier) {<a name="line.198"></a> -<span class="sourceLineNo">199</span> byte [][] fq = KeyValue.parseColumn(familyAndQualifier);<a name="line.199"></a> -<span class="sourceLineNo">200</span> if (fq.length == 1) {<a name="line.200"></a> -<span class="sourceLineNo">201</span> scan.addFamily(fq[0]);<a name="line.201"></a> -<span class="sourceLineNo">202</span> } else if (fq.length == 2) {<a name="line.202"></a> -<span class="sourceLineNo">203</span> scan.addColumn(fq[0], fq[1]);<a name="line.203"></a> -<span class="sourceLineNo">204</span> } else {<a name="line.204"></a> -<span class="sourceLineNo">205</span> throw new IllegalArgumentException("Invalid familyAndQualifier provided.");<a name="line.205"></a> -<span class="sourceLineNo">206</span> }<a name="line.206"></a> -<span class="sourceLineNo">207</span> }<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> * Adds an array of columns specified using old format, family:qualifier.<a name="line.210"></a> -<span class="sourceLineNo">211</span> * <p><a name="line.211"></a> -<span class="sourceLineNo">212</span> * Overrides previous calls to {@link Scan#addColumn(byte[], byte[])}for any families in the<a name="line.212"></a> -<span class="sourceLineNo">213</span> * input.<a name="line.213"></a> -<span class="sourceLineNo">214</span> *<a name="line.214"></a> -<span class="sourceLineNo">215</span> * @param scan The Scan to update.<a name="line.215"></a> -<span class="sourceLineNo">216</span> * @param columns array of columns, formatted as <code>family:qualifier</code><a name="line.216"></a> -<span class="sourceLineNo">217</span> * @see Scan#addColumn(byte[], byte[])<a name="line.217"></a> -<span class="sourceLineNo">218</span> */<a name="line.218"></a> -<span class="sourceLineNo">219</span> public static void addColumns(Scan scan, byte [][] columns) {<a name="line.219"></a> -<span class="sourceLineNo">220</span> for (byte[] column : columns) {<a name="line.220"></a> -<span class="sourceLineNo">221</span> addColumn(scan, column);<a name="line.221"></a> -<span class="sourceLineNo">222</span> }<a name="line.222"></a> -<span class="sourceLineNo">223</span> }<a name="line.223"></a> -<span class="sourceLineNo">224</span><a name="line.224"></a> -<span class="sourceLineNo">225</span> /**<a name="line.225"></a> -<span class="sourceLineNo">226</span> * Calculates the splits that will serve as input for the map tasks. The<a name="line.226"></a> -<span class="sourceLineNo">227</span> * number of splits matches the number of regions in a table. Splits are shuffled if<a name="line.227"></a> -<span class="sourceLineNo">228</span> * required.<a name="line.228"></a> -<span class="sourceLineNo">229</span> * @param context The current job context.<a name="line.229"></a> -<span class="sourceLineNo">230</span> * @return The list of input splits.<a name="line.230"></a> -<span class="sourceLineNo">231</span> * @throws IOException When creating the list of splits fails.<a name="line.231"></a> -<span class="sourceLineNo">232</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(<a name="line.232"></a> -<span class="sourceLineNo">233</span> * org.apache.hadoop.mapreduce.JobContext)<a name="line.233"></a> -<span class="sourceLineNo">234</span> */<a name="line.234"></a> -<span class="sourceLineNo">235</span> @Override<a name="line.235"></a> -<span class="sourceLineNo">236</span> public List<InputSplit> getSplits(JobContext context) throws IOException {<a name="line.236"></a> -<span class="sourceLineNo">237</span> List<InputSplit> splits = super.getSplits(context);<a name="line.237"></a> -<span class="sourceLineNo">238</span> if ((conf.get(SHUFFLE_MAPS) != null) && "true".equals(conf.get(SHUFFLE_MAPS).toLowerCase())) {<a name="line.238"></a> -<span class="sourceLineNo">239</span> Collections.shuffle(splits);<a name="line.239"></a> -<span class="sourceLineNo">240</span> }<a name="line.240"></a> -<span class="sourceLineNo">241</span> return splits;<a name="line.241"></a> -<span class="sourceLineNo">242</span> }<a name="line.242"></a> -<span class="sourceLineNo">243</span><a name="line.243"></a> -<span class="sourceLineNo">244</span> /**<a name="line.244"></a> -<span class="sourceLineNo">245</span> * Convenience method to parse a string representation of an array of column specifiers.<a name="line.245"></a> -<span class="sourceLineNo">246</span> *<a name="line.246"></a> -<span class="sourceLineNo">247</span> * @param scan The Scan to update.<a name="line.247"></a> -<span class="sourceLineNo">248</span> * @param columns The columns to parse.<a name="line.248"></a> -<span class="sourceLineNo">249</span> */<a name="line.249"></a> -<span class="sourceLineNo">250</span> private static void addColumns(Scan scan, String columns) {<a name="line.250"></a> -<span class="sourceLineNo">251</span> String[] cols = columns.split(" ");<a name="line.251"></a> -<span class="sourceLineNo">252</span> for (String col : cols) {<a name="line.252"></a> -<span class="sourceLineNo">253</span> addColumn(scan, Bytes.toBytes(col));<a name="line.253"></a> -<span class="sourceLineNo">254</span> }<a name="line.254"></a> -<span class="sourceLineNo">255</span> }<a name="line.255"></a> -<span class="sourceLineNo">256</span><a name="line.256"></a> -<span class="sourceLineNo">257</span> @Override<a name="line.257"></a> -<span class="sourceLineNo">258</span> protected Pair<byte[][], byte[][]> getStartEndKeys() throws IOException {<a name="line.258"></a> -<span class="sourceLineNo">259</span> if (conf.get(SPLIT_TABLE) != null) {<a name="line.259"></a> -<span class="sourceLineNo">260</span> TableName splitTableName = TableName.valueOf(conf.get(SPLIT_TABLE));<a name="line.260"></a> -<span class="sourceLineNo">261</span> try (Connection conn = ConnectionFactory.createConnection(getConf())) {<a name="line.261"></a> -<span class="sourceLineNo">262</span> try (RegionLocator rl = conn.getRegionLocator(splitTableName)) {<a name="line.262"></a> -<span class="sourceLineNo">263</span> return rl.getStartEndKeys();<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 name="line.267"></a> -<span class="sourceLineNo">268</span> return super.getStartEndKeys();<a name="line.268"></a> -<span class="sourceLineNo">269</span> }<a name="line.269"></a> -<span class="sourceLineNo">270</span><a name="line.270"></a> -<span class="sourceLineNo">271</span> /**<a name="line.271"></a> -<span class="sourceLineNo">272</span> * Sets split table in map-reduce job.<a name="line.272"></a> -<span class="sourceLineNo">273</span> */<a name="line.273"></a> -<span class="sourceLineNo">274</span> public static void configureSplitTable(Job job, TableName tableName) {<a name="line.274"></a> -<span class="sourceLineNo">275</span> job.getConfiguration().set(SPLIT_TABLE, tableName.getNameAsString());<a name="line.275"></a> -<span class="sourceLineNo">276</span> }<a name="line.276"></a> -<span class="sourceLineNo">277</span>}<a name="line.277"></a> +<span class="sourceLineNo">119</span> Scan scan = null;<a name="line.119"></a> +<span class="sourceLineNo">120</span><a name="line.120"></a> +<span class="sourceLineNo">121</span> if (conf.get(SCAN) != null) {<a name="line.121"></a> +<span class="sourceLineNo">122</span> try {<a name="line.122"></a> +<span class="sourceLineNo">123</span> scan = TableMapReduceUtil.convertStringToScan(conf.get(SCAN));<a name="line.123"></a> +<span class="sourceLineNo">124</span> } catch (IOException e) {<a name="line.124"></a> +<span class="sourceLineNo">125</span> LOG.error("An error occurred.", e);<a name="line.125"></a> +<span class="sourceLineNo">126</span> }<a name="line.126"></a> +<span class="sourceLineNo">127</span> } else {<a name="line.127"></a> +<span class="sourceLineNo">128</span> try {<a name="line.128"></a> +<span class="sourceLineNo">129</span> scan = new Scan();<a name="line.129"></a> +<span class="sourceLineNo">130</span><a name="line.130"></a> +<span class="sourceLineNo">131</span> if (conf.get(SCAN_ROW_START) != null) {<a name="line.131"></a> +<span class="sourceLineNo">132</span> scan.setStartRow(Bytes.toBytes(conf.get(SCAN_ROW_START)));<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> if (conf.get(SCAN_ROW_STOP) != null) {<a name="line.135"></a> +<span class="sourceLineNo">136</span> scan.setStopRow(Bytes.toBytes(conf.get(SCAN_ROW_STOP)));<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> if (conf.get(SCAN_COLUMNS) != null) {<a name="line.139"></a> +<span class="sourceLineNo">140</span> addColumns(scan, conf.get(SCAN_COLUMNS));<a name="line.140"></a> +<span class="sourceLineNo">141</span> }<a name="line.141"></a> +<span class="sourceLineNo">142</span><a name="line.142"></a> +<span class="sourceLineNo">143</span> if (conf.get(SCAN_COLUMN_FAMILY) != null) {<a name="line.143"></a> +<span class="sourceLineNo">144</span> scan.addFamily(Bytes.toBytes(conf.get(SCAN_COLUMN_FAMILY)));<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> if (conf.get(SCAN_TIMESTAMP) != null) {<a name="line.147"></a> +<span class="sourceLineNo">148</span> scan.setTimeStamp(Long.parseLong(conf.get(SCAN_TIMESTAMP)));<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> if (conf.get(SCAN_TIMERANGE_START) != null && conf.get(SCAN_TIMERANGE_END) != null) {<a name="line.151"></a> +<span class="sourceLineNo">152</span> scan.setTimeRange(<a name="line.152"></a> +<span class="sourceLineNo">153</span> Long.parseLong(conf.get(SCAN_TIMERANGE_START)),<a name="line.153"></a> +<span class="sourceLineNo">154</span> Long.parseLong(conf.get(SCAN_TIMERANGE_END)));<a name="line.154"></a> +<span class="sourceLineNo">155</span> }<a name="line.155"></a> +<span class="sourceLineNo">156</span><a name="line.156"></a> +<span class="sourceLineNo">157</span> if (conf.get(SCAN_MAXVERSIONS) != null) {<a name="line.157"></a> +<span class="sourceLineNo">158</span> scan.setMaxVersions(Integer.parseInt(conf.get(SCAN_MAXVERSIONS)));<a name="line.158"></a> +<span class="sourceLineNo">159</span> }<a name="line.159"></a> +<span class="sourceLineNo">160</span><a name="line.160"></a> +<span class="sourceLineNo">161</span> if (conf.get(SCAN_CACHEDROWS) != null) {<a name="line.161"></a> +<span class="sourceLineNo">162</span> scan.setCaching(Integer.parseInt(conf.get(SCAN_CACHEDROWS)));<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> if (conf.get(SCAN_BATCHSIZE) != null) {<a name="line.165"></a> +<span class="sourceLineNo">166</span> scan.setBatch(Integer.parseInt(conf.get(SCAN_BATCHSIZE)));<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> // false by default, full table scans generate too much BC churn<a name="line.169"></a> +<span class="sourceLineNo">170</span> scan.setCacheBlocks((conf.getBoolean(SCAN_CACHEBLOCKS, false)));<a name="line.170"></a> +<span class="sourceLineNo">171</span> } catch (Exception e) {<a name="line.171"></a> +<span class="sourceLineNo">172</span> LOG.error(StringUtils.stringifyException(e));<a name="line.172"></a> +<span class="sourceLineNo">173</span> }<a name="line.173"></a> +<span class="sourceLineNo">174</span> }<a name="line.174"></a> +<span class="sourceLineNo">175</span><a name="line.175"></a> +<span class="sourceLineNo">176</span> setScan(scan);<a name="line.176"></a> +<span class="sourceLineNo">177</span> }<a name="line.177"></a> +<span class="sourceLineNo">178</span><a name="line.178"></a> +<span class="sourceLineNo">179</span> @Override<a name="line.179"></a> +<span class="sourceLineNo">180</span> protected void initialize(JobContext context) throws IOException {<a name="line.180"></a> +<span class="sourceLineNo">181</span> // Do we have to worry about mis-matches between the Configuration from setConf and the one<a name="line.181"></a> +<span class="sourceLineNo">182</span> // in this context?<a name="line.182"></a> +<span class="sourceLineNo">183</span> TableName tableName = TableName.valueOf(conf.get(INPUT_TABLE));<a name="line.183"></a> +<span class="sourceLineNo">184</span> try {<a name="line.184"></a> +<span class="sourceLineNo">185</span> initializeTable(ConnectionFactory.createConnection(new Configuration(conf)), tableName);<a name="line.185"></a> +<span class="sourceLineNo">186</span> } catch (Exception e) {<a name="line.186"></a> +<span class="sourceLineNo">187</span> LOG.error(StringUtils.stringifyException(e));<a name="line.187"></a> +<span class="sourceLineNo">188</span> }<a name="line.188"></a> +<span class="sourceLineNo">189</span> }<a name="line.189"></a> +<span class="sourceLineNo">190</span><a name="line.190"></a> +<span class="sourceLineNo">191</span> /**<a name="line.191"></a> +<span class="sourceLineNo">192</span> * Parses a combined family and qualifier and adds either both or just the<a name="line.192"></a> +<span class="sourceLineNo">193</span> * family in case there is no qualifier. This assumes the older colon<a name="line.193"></a> +<span class="sourceLineNo">194</span> * divided notation, e.g. "family:qualifier".<a name="line.194"></a> +<span class="sourceLineNo">195</span> *<a name="line.195"></a> +<span class="sourceLineNo">196</span> * @param scan The Scan to update.<a name="line.196"></a> +<span class="sourceLineNo">197</span> * @param familyAndQualifier family and qualifier<a name="line.197"></a> +<span class="sourceLineNo">198</span> * @throws IllegalArgumentException When familyAndQualifier is invalid.<a name="line.198"></a> +<span class="sourceLineNo">199</span> */<a name="line.199"></a> +<span class="sourceLineNo">200</span> private static void addColumn(Scan scan, byte[] familyAndQualifier) {<a name="line.200"></a> +<span class="sourceLineNo">201</span> byte [][] fq = KeyValue.parseColumn(familyAndQualifier);<a name="line.201"></a> +<span class="sourceLineNo">202</span> if (fq.length == 1) {<a name="line.202"></a> +<span class="sourceLineNo">203</span> scan.addFamily(fq[0]);<a name="line.203"></a> +<span class="sourceLineNo">204</span> } else if (fq.length == 2) {<a name="line.204"></a> +<span class="sourceLineNo">205</span> scan.addColumn(fq[0], fq[1]);<a name="line.205"></a> +<span class="sourceLineNo">206</span> } else {<a name="line.206"></a> +<span class="sourceLineNo">207</span> throw new IllegalArgumentException("Invalid familyAndQualifier provided.");<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><a name="line.210"></a> +<span class="sourceLineNo">211</span> /**<a name="line.211"></a> +<span class="sourceLineNo">212</span> * Adds an array of columns specified using old format, family:qualifier.<a name="line.212"></a> +<span class="sourceLineNo">213</span> * <p><a name="line.213"></a> +<span class="sourceLineNo">214</span> * Overrides previous calls to {@link Scan#addColumn(byte[], byte[])}for any families in the<a name="line.214"></a> +<span class="sourceLineNo">215</span> * input.<a name="line.215"></a> +<span class="sourceLineNo">216</span> *<a name="line.216"></a> +<span class="sourceLineNo">217</span> * @param scan The Scan to update.<a name="line.217"></a> +<span class="sourceLineNo">218</span> * @param columns array of columns, formatted as <code>family:qualifier</code><a name="line.218"></a> +<span class="sourceLineNo">219</span> * @see Scan#addColumn(byte[], byte[])<a name="line.219"></a> +<span class="sourceLineNo">220</span> */<a name="line.220"></a> +<span class="sourceLineNo">221</span> public static void addColumns(Scan scan, byte [][] columns) {<a name="line.221"></a> +<span class="sourceLineNo">222</span> for (byte[] column : columns) {<a name="line.222"></a> +<span class="sourceLineNo">223</span> addColumn(scan, column);<a name="line.223"></a> +<span class="sourceLineNo">224</span> }<a name="line.224"></a> +<span class="sourceLineNo">225</span> }<a name="line.225"></a> +<span class="sourceLineNo">226</span><a name="line.226"></a> +<span class="sourceLineNo">227</span> /**<a name="line.227"></a> +<span class="sourceLineNo">228</span> * Calculates the splits that will serve as input for the map tasks. The<a name="line.228"></a> +<span class="sourceLineNo">229</span> * number of splits matches the number of regions in a table. Splits are shuffled if<a name="line.229"></a> +<span class="sourceLineNo">230</span> * required.<a name="line.230"></a> +<span class="sourceLineNo">231</span> * @param context The current job context.<a name="line.231"></a> +<span class="sourceLineNo">232</span> * @return The list of input splits.<a name="line.232"></a> +<span class="sourceLineNo">233</span> * @throws IOException When creating the list of splits fails.<a name="line.233"></a> +<span class="sourceLineNo">234</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(<a name="line.234"></a> +<span class="sourceLineNo">235</span> * org.apache.hadoop.mapreduce.JobContext)<a name="line.235"></a> +<span class="sourceLineNo">236</span> */<a name="line.236"></a> +<span class="sourceLineNo">237</span> @Override<a name="line.237"></a> +<span class="sourceLineNo">238</span> public List<InputSplit> getSplits(JobContext context) throws IOException {<a name="line.238"></a> +<span class="sourceLineNo">239</span> List<InputSplit> splits = super.getSplits(context);<a name="line.239"></a> +<span class="sourceLineNo">240</span> if ((conf.get(SHUFFLE_MAPS) != null) && "true".equals(conf.get(SHUFFLE_MAPS).toLowerCase())) {<a name="line.240"></a> +<span class="sourceLineNo">241</span> Collections.shuffle(splits);<a name="line.241"></a> +<span class="sourceLineNo">242</span> }<a name="line.242"></a> +<span class="sourceLineNo">243</span> return splits;<a name="line.243"></a> +<span class="sourceLineNo">244</span> }<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> * Convenience method to parse a string representation of an array of column specifiers.<a name="line.247"></a> +<span class="sourceLineNo">248</span> *<a name="line.248"></a> +<span class="sourceLineNo">249</span> * @param scan The Scan to update.<a name="line.249"></a> +<span class="sourceLineNo">250</span> * @param columns The columns to parse.<a name="line.250"></a> +<span class="sourceLineNo">251</span> */<a name="line.251"></a> +<span class="sourceLineNo">252</span> private static void addColumns(Scan scan, String columns) {<a name="line.252"></a> +<span class="sourceLineNo">253</span> String[] cols = columns.split(" ");<a name="line.253"></a> +<span class="sourceLineNo">254</span> for (String col : cols) {<a name="line.254"></a> +<span class="sourceLineNo">255</span> addColumn(scan, Bytes.toBytes(col));<a name="line.255"></a> +<span class="sourceLineNo">256</span> }<a name="line.256"></a> +<span class="sourceLineNo">257</span> }<a name="line.257"></a> +<span class="sourceLineNo">258</span><a name="line.258"></a> +<span class="sourceLineNo">259</span> @Override<a name="line.259"></a> +<span class="sourceLineNo">260</span> protected Pair<byte[][], byte[][]> getStartEndKeys() throws IOException {<a name="line.260"></a> +<span class="sourceLineNo">261</span> if (conf.get(SPLIT_TABLE) != null) {<a name="line.261"></a> +<span class="sourceLineNo">262</span> TableName splitTableName = TableName.valueOf(conf.get(SPLIT_TABLE));<a name="line.262"></a> +<span class="sourceLineNo">263</span> try (Connection conn = ConnectionFactory.createConnection(getConf())) {<a name="line.263"></a> +<span class="sourceLineNo">264</span> try (RegionLocator rl = conn.getRegionLocator(splitTableName)) {<a name="line.264"></a> +<span class="sourceLineNo">265</span> return rl.getStartEndKeys();<a name="line.265"></a> +<span class="sourceLineNo">266</span> }<a name="line.266"></a> +<span class="sourceLineNo">267</span> }<a name="line.267"></a> +<span class="sourceLineNo">268</span> }<a name="line.268"></a> +<span class="sourceLineNo">269</span><a name="line.269"></a> +<span class="sourceLineNo">270</span> return super.getStartEndKeys();<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> /**<a name="line.273"></a> +<span class="sourceLineNo">274</span> * Sets split table in map-reduce job.<a name="line.274"></a> +<span class="sourceLineNo">275</span> */<a name="line.275"></a> +<span class="sourceLineNo">276</span> public static void configureSplitTable(Job job, TableName tableName) {<a name="line.276"></a> +<span class="sourceLineNo">277</span> job.getConfiguration().set(SPLIT_TABLE, tableName.getNameAsString());<a name="line.277"></a> +<span class="sourceLineNo">278</span> }<a name="line.278"></a> +<span class="sourceLineNo">279</span>}<a name="line.279"></a>