http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f3d62514/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.Counters.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.Counters.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.Counters.html index 6f1fc1b..fbfe15d 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.Counters.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.Counters.html @@ -184,165 +184,166 @@ <span class="sourceLineNo">176</span> context.getCounter(Counters.CELLS).increment(cellCount);<a name="line.176"></a> <span class="sourceLineNo">177</span> }<a name="line.177"></a> <span class="sourceLineNo">178</span> } catch (InterruptedException e) {<a name="line.178"></a> -<span class="sourceLineNo">179</span> e.printStackTrace();<a name="line.179"></a> -<span class="sourceLineNo">180</span> }<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span> }<a name="line.182"></a> -<span class="sourceLineNo">183</span><a name="line.183"></a> -<span class="sourceLineNo">184</span> static class IntSumReducer<Key> extends Reducer<Key, IntWritable,<a name="line.184"></a> -<span class="sourceLineNo">185</span> Key, IntWritable> {<a name="line.185"></a> -<span class="sourceLineNo">186</span><a name="line.186"></a> -<span class="sourceLineNo">187</span> private IntWritable result = new IntWritable();<a name="line.187"></a> -<span class="sourceLineNo">188</span> public void reduce(Key key, Iterable<IntWritable> values,<a name="line.188"></a> -<span class="sourceLineNo">189</span> Context context)<a name="line.189"></a> -<span class="sourceLineNo">190</span> throws IOException, InterruptedException {<a name="line.190"></a> -<span class="sourceLineNo">191</span> int sum = 0;<a name="line.191"></a> -<span class="sourceLineNo">192</span> for (IntWritable val : values) {<a name="line.192"></a> -<span class="sourceLineNo">193</span> sum += val.get();<a name="line.193"></a> -<span class="sourceLineNo">194</span> }<a name="line.194"></a> -<span class="sourceLineNo">195</span> result.set(sum);<a name="line.195"></a> -<span class="sourceLineNo">196</span> context.write(key, result);<a name="line.196"></a> -<span class="sourceLineNo">197</span> }<a name="line.197"></a> -<span class="sourceLineNo">198</span> }<a name="line.198"></a> -<span class="sourceLineNo">199</span><a name="line.199"></a> -<span class="sourceLineNo">200</span> /**<a name="line.200"></a> -<span class="sourceLineNo">201</span> * Sets up the actual job.<a name="line.201"></a> -<span class="sourceLineNo">202</span> *<a name="line.202"></a> -<span class="sourceLineNo">203</span> * @param conf The current configuration.<a name="line.203"></a> -<span class="sourceLineNo">204</span> * @param args The command line parameters.<a name="line.204"></a> -<span class="sourceLineNo">205</span> * @return The newly created job.<a name="line.205"></a> -<span class="sourceLineNo">206</span> * @throws IOException When setting up the job fails.<a name="line.206"></a> -<span class="sourceLineNo">207</span> */<a name="line.207"></a> -<span class="sourceLineNo">208</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.208"></a> -<span class="sourceLineNo">209</span> throws IOException {<a name="line.209"></a> -<span class="sourceLineNo">210</span> String tableName = args[0];<a name="line.210"></a> -<span class="sourceLineNo">211</span> Path outputDir = new Path(args[1]);<a name="line.211"></a> -<span class="sourceLineNo">212</span> String reportSeparatorString = (args.length > 2) ? args[2]: ":";<a name="line.212"></a> -<span class="sourceLineNo">213</span> conf.set("ReportSeparator", reportSeparatorString);<a name="line.213"></a> -<span class="sourceLineNo">214</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.214"></a> -<span class="sourceLineNo">215</span> job.setJarByClass(CellCounter.class);<a name="line.215"></a> -<span class="sourceLineNo">216</span> Scan scan = getConfiguredScanForJob(conf, args);<a name="line.216"></a> -<span class="sourceLineNo">217</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.217"></a> -<span class="sourceLineNo">218</span> CellCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.218"></a> -<span class="sourceLineNo">219</span> job.setNumReduceTasks(1);<a name="line.219"></a> -<span class="sourceLineNo">220</span> job.setMapOutputKeyClass(Text.class);<a name="line.220"></a> -<span class="sourceLineNo">221</span> job.setMapOutputValueClass(IntWritable.class);<a name="line.221"></a> -<span class="sourceLineNo">222</span> job.setOutputFormatClass(TextOutputFormat.class);<a name="line.222"></a> -<span class="sourceLineNo">223</span> job.setOutputKeyClass(Text.class);<a name="line.223"></a> -<span class="sourceLineNo">224</span> job.setOutputValueClass(IntWritable.class);<a name="line.224"></a> -<span class="sourceLineNo">225</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.225"></a> -<span class="sourceLineNo">226</span> job.setReducerClass(IntSumReducer.class);<a name="line.226"></a> -<span class="sourceLineNo">227</span> return job;<a name="line.227"></a> -<span class="sourceLineNo">228</span> }<a name="line.228"></a> -<span class="sourceLineNo">229</span><a name="line.229"></a> -<span class="sourceLineNo">230</span> private static Scan getConfiguredScanForJob(Configuration conf, String[] args)<a name="line.230"></a> -<span class="sourceLineNo">231</span> throws IOException {<a name="line.231"></a> -<span class="sourceLineNo">232</span> // create scan with any properties set from TableInputFormat<a name="line.232"></a> -<span class="sourceLineNo">233</span> Scan s = TableInputFormat.createScanFromConfiguration(conf);<a name="line.233"></a> -<span class="sourceLineNo">234</span> // Set Scan Versions<a name="line.234"></a> -<span class="sourceLineNo">235</span> if (conf.get(TableInputFormat.SCAN_MAXVERSIONS) == null) {<a name="line.235"></a> -<span class="sourceLineNo">236</span> // default to all versions unless explicitly set<a name="line.236"></a> -<span class="sourceLineNo">237</span> s.setMaxVersions(Integer.MAX_VALUE);<a name="line.237"></a> -<span class="sourceLineNo">238</span> }<a name="line.238"></a> -<span class="sourceLineNo">239</span> s.setCacheBlocks(false);<a name="line.239"></a> -<span class="sourceLineNo">240</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.240"></a> -<span class="sourceLineNo">241</span> Filter rowFilter = getRowFilter(args);<a name="line.241"></a> -<span class="sourceLineNo">242</span> if (rowFilter!= null) {<a name="line.242"></a> -<span class="sourceLineNo">243</span> LOG.info("Setting Row Filter for counter.");<a name="line.243"></a> -<span class="sourceLineNo">244</span> s.setFilter(rowFilter);<a name="line.244"></a> -<span class="sourceLineNo">245</span> }<a name="line.245"></a> -<span class="sourceLineNo">246</span> // Set TimeRange if defined<a name="line.246"></a> -<span class="sourceLineNo">247</span> long timeRange[] = getTimeRange(args);<a name="line.247"></a> -<span class="sourceLineNo">248</span> if (timeRange != null) {<a name="line.248"></a> -<span class="sourceLineNo">249</span> LOG.info("Setting TimeRange for counter.");<a name="line.249"></a> -<span class="sourceLineNo">250</span> s.setTimeRange(timeRange[0], timeRange[1]);<a name="line.250"></a> -<span class="sourceLineNo">251</span> }<a name="line.251"></a> -<span class="sourceLineNo">252</span> return s;<a name="line.252"></a> -<span class="sourceLineNo">253</span> }<a name="line.253"></a> -<span class="sourceLineNo">254</span><a name="line.254"></a> +<span class="sourceLineNo">179</span> LOG.error("Interrupted while writing cellCount", e);<a name="line.179"></a> +<span class="sourceLineNo">180</span> Thread.currentThread().interrupt();<a name="line.180"></a> +<span class="sourceLineNo">181</span> }<a name="line.181"></a> +<span class="sourceLineNo">182</span> }<a name="line.182"></a> +<span class="sourceLineNo">183</span> }<a name="line.183"></a> +<span class="sourceLineNo">184</span><a name="line.184"></a> +<span class="sourceLineNo">185</span> static class IntSumReducer<Key> extends Reducer<Key, IntWritable,<a name="line.185"></a> +<span class="sourceLineNo">186</span> Key, IntWritable> {<a name="line.186"></a> +<span class="sourceLineNo">187</span><a name="line.187"></a> +<span class="sourceLineNo">188</span> private IntWritable result = new IntWritable();<a name="line.188"></a> +<span class="sourceLineNo">189</span> public void reduce(Key key, Iterable<IntWritable> values,<a name="line.189"></a> +<span class="sourceLineNo">190</span> Context context)<a name="line.190"></a> +<span class="sourceLineNo">191</span> throws IOException, InterruptedException {<a name="line.191"></a> +<span class="sourceLineNo">192</span> int sum = 0;<a name="line.192"></a> +<span class="sourceLineNo">193</span> for (IntWritable val : values) {<a name="line.193"></a> +<span class="sourceLineNo">194</span> sum += val.get();<a name="line.194"></a> +<span class="sourceLineNo">195</span> }<a name="line.195"></a> +<span class="sourceLineNo">196</span> result.set(sum);<a name="line.196"></a> +<span class="sourceLineNo">197</span> context.write(key, result);<a name="line.197"></a> +<span class="sourceLineNo">198</span> }<a name="line.198"></a> +<span class="sourceLineNo">199</span> }<a name="line.199"></a> +<span class="sourceLineNo">200</span><a name="line.200"></a> +<span class="sourceLineNo">201</span> /**<a name="line.201"></a> +<span class="sourceLineNo">202</span> * Sets up the actual job.<a name="line.202"></a> +<span class="sourceLineNo">203</span> *<a name="line.203"></a> +<span class="sourceLineNo">204</span> * @param conf The current configuration.<a name="line.204"></a> +<span class="sourceLineNo">205</span> * @param args The command line parameters.<a name="line.205"></a> +<span class="sourceLineNo">206</span> * @return The newly created job.<a name="line.206"></a> +<span class="sourceLineNo">207</span> * @throws IOException When setting up the job fails.<a name="line.207"></a> +<span class="sourceLineNo">208</span> */<a name="line.208"></a> +<span class="sourceLineNo">209</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.209"></a> +<span class="sourceLineNo">210</span> throws IOException {<a name="line.210"></a> +<span class="sourceLineNo">211</span> String tableName = args[0];<a name="line.211"></a> +<span class="sourceLineNo">212</span> Path outputDir = new Path(args[1]);<a name="line.212"></a> +<span class="sourceLineNo">213</span> String reportSeparatorString = (args.length > 2) ? args[2]: ":";<a name="line.213"></a> +<span class="sourceLineNo">214</span> conf.set("ReportSeparator", reportSeparatorString);<a name="line.214"></a> +<span class="sourceLineNo">215</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.215"></a> +<span class="sourceLineNo">216</span> job.setJarByClass(CellCounter.class);<a name="line.216"></a> +<span class="sourceLineNo">217</span> Scan scan = getConfiguredScanForJob(conf, args);<a name="line.217"></a> +<span class="sourceLineNo">218</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.218"></a> +<span class="sourceLineNo">219</span> CellCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.219"></a> +<span class="sourceLineNo">220</span> job.setNumReduceTasks(1);<a name="line.220"></a> +<span class="sourceLineNo">221</span> job.setMapOutputKeyClass(Text.class);<a name="line.221"></a> +<span class="sourceLineNo">222</span> job.setMapOutputValueClass(IntWritable.class);<a name="line.222"></a> +<span class="sourceLineNo">223</span> job.setOutputFormatClass(TextOutputFormat.class);<a name="line.223"></a> +<span class="sourceLineNo">224</span> job.setOutputKeyClass(Text.class);<a name="line.224"></a> +<span class="sourceLineNo">225</span> job.setOutputValueClass(IntWritable.class);<a name="line.225"></a> +<span class="sourceLineNo">226</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.226"></a> +<span class="sourceLineNo">227</span> job.setReducerClass(IntSumReducer.class);<a name="line.227"></a> +<span class="sourceLineNo">228</span> return job;<a name="line.228"></a> +<span class="sourceLineNo">229</span> }<a name="line.229"></a> +<span class="sourceLineNo">230</span><a name="line.230"></a> +<span class="sourceLineNo">231</span> private static Scan getConfiguredScanForJob(Configuration conf, String[] args)<a name="line.231"></a> +<span class="sourceLineNo">232</span> throws IOException {<a name="line.232"></a> +<span class="sourceLineNo">233</span> // create scan with any properties set from TableInputFormat<a name="line.233"></a> +<span class="sourceLineNo">234</span> Scan s = TableInputFormat.createScanFromConfiguration(conf);<a name="line.234"></a> +<span class="sourceLineNo">235</span> // Set Scan Versions<a name="line.235"></a> +<span class="sourceLineNo">236</span> if (conf.get(TableInputFormat.SCAN_MAXVERSIONS) == null) {<a name="line.236"></a> +<span class="sourceLineNo">237</span> // default to all versions unless explicitly set<a name="line.237"></a> +<span class="sourceLineNo">238</span> s.setMaxVersions(Integer.MAX_VALUE);<a name="line.238"></a> +<span class="sourceLineNo">239</span> }<a name="line.239"></a> +<span class="sourceLineNo">240</span> s.setCacheBlocks(false);<a name="line.240"></a> +<span class="sourceLineNo">241</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.241"></a> +<span class="sourceLineNo">242</span> Filter rowFilter = getRowFilter(args);<a name="line.242"></a> +<span class="sourceLineNo">243</span> if (rowFilter!= null) {<a name="line.243"></a> +<span class="sourceLineNo">244</span> LOG.info("Setting Row Filter for counter.");<a name="line.244"></a> +<span class="sourceLineNo">245</span> s.setFilter(rowFilter);<a name="line.245"></a> +<span class="sourceLineNo">246</span> }<a name="line.246"></a> +<span class="sourceLineNo">247</span> // Set TimeRange if defined<a name="line.247"></a> +<span class="sourceLineNo">248</span> long timeRange[] = getTimeRange(args);<a name="line.248"></a> +<span class="sourceLineNo">249</span> if (timeRange != null) {<a name="line.249"></a> +<span class="sourceLineNo">250</span> LOG.info("Setting TimeRange for counter.");<a name="line.250"></a> +<span class="sourceLineNo">251</span> s.setTimeRange(timeRange[0], timeRange[1]);<a name="line.251"></a> +<span class="sourceLineNo">252</span> }<a name="line.252"></a> +<span class="sourceLineNo">253</span> return s;<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> private static Filter getRowFilter(String[] args) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> Filter rowFilter = null;<a name="line.257"></a> -<span class="sourceLineNo">258</span> String filterCriteria = (args.length > 3) ? args[3]: null;<a name="line.258"></a> -<span class="sourceLineNo">259</span> if (filterCriteria == null) return null;<a name="line.259"></a> -<span class="sourceLineNo">260</span> if (filterCriteria.startsWith("^")) {<a name="line.260"></a> -<span class="sourceLineNo">261</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.261"></a> -<span class="sourceLineNo">262</span> rowFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.262"></a> -<span class="sourceLineNo">263</span> } else {<a name="line.263"></a> -<span class="sourceLineNo">264</span> rowFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.264"></a> -<span class="sourceLineNo">265</span> }<a name="line.265"></a> -<span class="sourceLineNo">266</span> return rowFilter;<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> private static long[] getTimeRange(String[] args) throws IOException {<a name="line.269"></a> -<span class="sourceLineNo">270</span> final String startTimeArgKey = "--starttime=";<a name="line.270"></a> -<span class="sourceLineNo">271</span> final String endTimeArgKey = "--endtime=";<a name="line.271"></a> -<span class="sourceLineNo">272</span> long startTime = 0L;<a name="line.272"></a> -<span class="sourceLineNo">273</span> long endTime = 0L;<a name="line.273"></a> -<span class="sourceLineNo">274</span><a name="line.274"></a> -<span class="sourceLineNo">275</span> for (int i = 1; i < args.length; i++) {<a name="line.275"></a> -<span class="sourceLineNo">276</span> System.out.println("i:" + i + "arg[i]" + args[i]);<a name="line.276"></a> -<span class="sourceLineNo">277</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.277"></a> -<span class="sourceLineNo">278</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.278"></a> -<span class="sourceLineNo">279</span> }<a name="line.279"></a> -<span class="sourceLineNo">280</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.280"></a> -<span class="sourceLineNo">281</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<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><a name="line.284"></a> -<span class="sourceLineNo">285</span> if (startTime == 0 && endTime == 0)<a name="line.285"></a> -<span class="sourceLineNo">286</span> return null;<a name="line.286"></a> -<span class="sourceLineNo">287</span><a name="line.287"></a> -<span class="sourceLineNo">288</span> endTime = endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime;<a name="line.288"></a> -<span class="sourceLineNo">289</span> return new long [] {startTime, endTime};<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> @Override<a name="line.292"></a> -<span class="sourceLineNo">293</span> public int run(String[] args) throws Exception {<a name="line.293"></a> -<span class="sourceLineNo">294</span> if (args.length < 2) {<a name="line.294"></a> -<span class="sourceLineNo">295</span> printUsage(args.length);<a name="line.295"></a> -<span class="sourceLineNo">296</span> return -1;<a name="line.296"></a> -<span class="sourceLineNo">297</span> }<a name="line.297"></a> -<span class="sourceLineNo">298</span> Job job = createSubmittableJob(getConf(), args);<a name="line.298"></a> -<span class="sourceLineNo">299</span> return (job.waitForCompletion(true) ? 0 : 1);<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">302</span> private void printUsage(int parameterCount) {<a name="line.302"></a> -<span class="sourceLineNo">303</span> System.err.println("ERROR: Wrong number of parameters: " + parameterCount);<a name="line.303"></a> -<span class="sourceLineNo">304</span> System.err.println("Usage: hbase cellcounter <tablename> <outputDir> [reportSeparator] "<a name="line.304"></a> -<span class="sourceLineNo">305</span> + "[^[regex pattern] or [Prefix]] [--starttime=<starttime> --endtime=<endtime>]");<a name="line.305"></a> -<span class="sourceLineNo">306</span> System.err.println(" Note: -D properties will be applied to the conf used.");<a name="line.306"></a> -<span class="sourceLineNo">307</span> System.err.println(" Additionally, all of the SCAN properties from TableInputFormat can be "<a name="line.307"></a> -<span class="sourceLineNo">308</span> + "specified to get fine grained control on what is counted.");<a name="line.308"></a> -<span class="sourceLineNo">309</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_START + "=<rowkey>");<a name="line.309"></a> -<span class="sourceLineNo">310</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_STOP + "=<rowkey>");<a name="line.310"></a> -<span class="sourceLineNo">311</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMNS + "=\"<col1> <col2>...\"");<a name="line.311"></a> -<span class="sourceLineNo">312</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMN_FAMILY<a name="line.312"></a> -<span class="sourceLineNo">313</span> + "=<family1>,<family2>, ...");<a name="line.313"></a> -<span class="sourceLineNo">314</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMESTAMP + "=<timestamp>");<a name="line.314"></a> -<span class="sourceLineNo">315</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_START + "=<timestamp>");<a name="line.315"></a> -<span class="sourceLineNo">316</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_END + "=<timestamp>");<a name="line.316"></a> -<span class="sourceLineNo">317</span> System.err.println(" -D" + TableInputFormat.SCAN_MAXVERSIONS + "=<count>");<a name="line.317"></a> -<span class="sourceLineNo">318</span> System.err.println(" -D" + TableInputFormat.SCAN_CACHEDROWS + "=<count>");<a name="line.318"></a> -<span class="sourceLineNo">319</span> System.err.println(" -D" + TableInputFormat.SCAN_BATCHSIZE + "=<count>");<a name="line.319"></a> -<span class="sourceLineNo">320</span> System.err.println(" <reportSeparator> parameter can be used to override the default report "<a name="line.320"></a> -<span class="sourceLineNo">321</span> + "separator string : used to separate the rowId/column family name and qualifier name.");<a name="line.321"></a> -<span class="sourceLineNo">322</span> System.err.println(" [^[regex pattern] or [Prefix] parameter can be used to limit the cell "<a name="line.322"></a> -<span class="sourceLineNo">323</span> + "counter count operation to a limited subset of rows from the table based on regex or "<a name="line.323"></a> -<span class="sourceLineNo">324</span> + "prefix pattern.");<a name="line.324"></a> -<span class="sourceLineNo">325</span> }<a name="line.325"></a> -<span class="sourceLineNo">326</span><a name="line.326"></a> -<span class="sourceLineNo">327</span> /**<a name="line.327"></a> -<span class="sourceLineNo">328</span> * Main entry point.<a name="line.328"></a> -<span class="sourceLineNo">329</span> * @param args The command line parameters.<a name="line.329"></a> -<span class="sourceLineNo">330</span> * @throws Exception When running the job fails.<a name="line.330"></a> -<span class="sourceLineNo">331</span> */<a name="line.331"></a> -<span class="sourceLineNo">332</span> public static void main(String[] args) throws Exception {<a name="line.332"></a> -<span class="sourceLineNo">333</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new CellCounter(), args);<a name="line.333"></a> -<span class="sourceLineNo">334</span> System.exit(errCode);<a name="line.334"></a> -<span class="sourceLineNo">335</span> }<a name="line.335"></a> -<span class="sourceLineNo">336</span><a name="line.336"></a> -<span class="sourceLineNo">337</span>}<a name="line.337"></a> +<span class="sourceLineNo">256</span><a name="line.256"></a> +<span class="sourceLineNo">257</span> private static Filter getRowFilter(String[] args) {<a name="line.257"></a> +<span class="sourceLineNo">258</span> Filter rowFilter = null;<a name="line.258"></a> +<span class="sourceLineNo">259</span> String filterCriteria = (args.length > 3) ? args[3]: null;<a name="line.259"></a> +<span class="sourceLineNo">260</span> if (filterCriteria == null) return null;<a name="line.260"></a> +<span class="sourceLineNo">261</span> if (filterCriteria.startsWith("^")) {<a name="line.261"></a> +<span class="sourceLineNo">262</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.262"></a> +<span class="sourceLineNo">263</span> rowFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.263"></a> +<span class="sourceLineNo">264</span> } else {<a name="line.264"></a> +<span class="sourceLineNo">265</span> rowFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.265"></a> +<span class="sourceLineNo">266</span> }<a name="line.266"></a> +<span class="sourceLineNo">267</span> return rowFilter;<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> private static long[] getTimeRange(String[] args) throws IOException {<a name="line.270"></a> +<span class="sourceLineNo">271</span> final String startTimeArgKey = "--starttime=";<a name="line.271"></a> +<span class="sourceLineNo">272</span> final String endTimeArgKey = "--endtime=";<a name="line.272"></a> +<span class="sourceLineNo">273</span> long startTime = 0L;<a name="line.273"></a> +<span class="sourceLineNo">274</span> long endTime = 0L;<a name="line.274"></a> +<span class="sourceLineNo">275</span><a name="line.275"></a> +<span class="sourceLineNo">276</span> for (int i = 1; i < args.length; i++) {<a name="line.276"></a> +<span class="sourceLineNo">277</span> System.out.println("i:" + i + "arg[i]" + args[i]);<a name="line.277"></a> +<span class="sourceLineNo">278</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.278"></a> +<span class="sourceLineNo">279</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.279"></a> +<span class="sourceLineNo">280</span> }<a name="line.280"></a> +<span class="sourceLineNo">281</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.281"></a> +<span class="sourceLineNo">282</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.282"></a> +<span class="sourceLineNo">283</span> }<a name="line.283"></a> +<span class="sourceLineNo">284</span> }<a name="line.284"></a> +<span class="sourceLineNo">285</span><a name="line.285"></a> +<span class="sourceLineNo">286</span> if (startTime == 0 && endTime == 0)<a name="line.286"></a> +<span class="sourceLineNo">287</span> return null;<a name="line.287"></a> +<span class="sourceLineNo">288</span><a name="line.288"></a> +<span class="sourceLineNo">289</span> endTime = endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime;<a name="line.289"></a> +<span class="sourceLineNo">290</span> return new long [] {startTime, endTime};<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> @Override<a name="line.293"></a> +<span class="sourceLineNo">294</span> public int run(String[] args) throws Exception {<a name="line.294"></a> +<span class="sourceLineNo">295</span> if (args.length < 2) {<a name="line.295"></a> +<span class="sourceLineNo">296</span> printUsage(args.length);<a name="line.296"></a> +<span class="sourceLineNo">297</span> return -1;<a name="line.297"></a> +<span class="sourceLineNo">298</span> }<a name="line.298"></a> +<span class="sourceLineNo">299</span> Job job = createSubmittableJob(getConf(), args);<a name="line.299"></a> +<span class="sourceLineNo">300</span> return (job.waitForCompletion(true) ? 0 : 1);<a name="line.300"></a> +<span class="sourceLineNo">301</span> }<a name="line.301"></a> +<span class="sourceLineNo">302</span><a name="line.302"></a> +<span class="sourceLineNo">303</span> private void printUsage(int parameterCount) {<a name="line.303"></a> +<span class="sourceLineNo">304</span> System.err.println("ERROR: Wrong number of parameters: " + parameterCount);<a name="line.304"></a> +<span class="sourceLineNo">305</span> System.err.println("Usage: hbase cellcounter <tablename> <outputDir> [reportSeparator] "<a name="line.305"></a> +<span class="sourceLineNo">306</span> + "[^[regex pattern] or [Prefix]] [--starttime=<starttime> --endtime=<endtime>]");<a name="line.306"></a> +<span class="sourceLineNo">307</span> System.err.println(" Note: -D properties will be applied to the conf used.");<a name="line.307"></a> +<span class="sourceLineNo">308</span> System.err.println(" Additionally, all of the SCAN properties from TableInputFormat can be "<a name="line.308"></a> +<span class="sourceLineNo">309</span> + "specified to get fine grained control on what is counted.");<a name="line.309"></a> +<span class="sourceLineNo">310</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_START + "=<rowkey>");<a name="line.310"></a> +<span class="sourceLineNo">311</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_STOP + "=<rowkey>");<a name="line.311"></a> +<span class="sourceLineNo">312</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMNS + "=\"<col1> <col2>...\"");<a name="line.312"></a> +<span class="sourceLineNo">313</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMN_FAMILY<a name="line.313"></a> +<span class="sourceLineNo">314</span> + "=<family1>,<family2>, ...");<a name="line.314"></a> +<span class="sourceLineNo">315</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMESTAMP + "=<timestamp>");<a name="line.315"></a> +<span class="sourceLineNo">316</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_START + "=<timestamp>");<a name="line.316"></a> +<span class="sourceLineNo">317</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_END + "=<timestamp>");<a name="line.317"></a> +<span class="sourceLineNo">318</span> System.err.println(" -D" + TableInputFormat.SCAN_MAXVERSIONS + "=<count>");<a name="line.318"></a> +<span class="sourceLineNo">319</span> System.err.println(" -D" + TableInputFormat.SCAN_CACHEDROWS + "=<count>");<a name="line.319"></a> +<span class="sourceLineNo">320</span> System.err.println(" -D" + TableInputFormat.SCAN_BATCHSIZE + "=<count>");<a name="line.320"></a> +<span class="sourceLineNo">321</span> System.err.println(" <reportSeparator> parameter can be used to override the default report "<a name="line.321"></a> +<span class="sourceLineNo">322</span> + "separator string : used to separate the rowId/column family name and qualifier name.");<a name="line.322"></a> +<span class="sourceLineNo">323</span> System.err.println(" [^[regex pattern] or [Prefix] parameter can be used to limit the cell "<a name="line.323"></a> +<span class="sourceLineNo">324</span> + "counter count operation to a limited subset of rows from the table based on regex or "<a name="line.324"></a> +<span class="sourceLineNo">325</span> + "prefix pattern.");<a name="line.325"></a> +<span class="sourceLineNo">326</span> }<a name="line.326"></a> +<span class="sourceLineNo">327</span><a name="line.327"></a> +<span class="sourceLineNo">328</span> /**<a name="line.328"></a> +<span class="sourceLineNo">329</span> * Main entry point.<a name="line.329"></a> +<span class="sourceLineNo">330</span> * @param args The command line parameters.<a name="line.330"></a> +<span class="sourceLineNo">331</span> * @throws Exception When running the job fails.<a name="line.331"></a> +<span class="sourceLineNo">332</span> */<a name="line.332"></a> +<span class="sourceLineNo">333</span> public static void main(String[] args) throws Exception {<a name="line.333"></a> +<span class="sourceLineNo">334</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new CellCounter(), args);<a name="line.334"></a> +<span class="sourceLineNo">335</span> System.exit(errCode);<a name="line.335"></a> +<span class="sourceLineNo">336</span> }<a name="line.336"></a> +<span class="sourceLineNo">337</span><a name="line.337"></a> +<span class="sourceLineNo">338</span>}<a name="line.338"></a>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f3d62514/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.html index 6f1fc1b..fbfe15d 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/CellCounter.CellCounterMapper.html @@ -184,165 +184,166 @@ <span class="sourceLineNo">176</span> context.getCounter(Counters.CELLS).increment(cellCount);<a name="line.176"></a> <span class="sourceLineNo">177</span> }<a name="line.177"></a> <span class="sourceLineNo">178</span> } catch (InterruptedException e) {<a name="line.178"></a> -<span class="sourceLineNo">179</span> e.printStackTrace();<a name="line.179"></a> -<span class="sourceLineNo">180</span> }<a name="line.180"></a> -<span class="sourceLineNo">181</span> }<a name="line.181"></a> -<span class="sourceLineNo">182</span> }<a name="line.182"></a> -<span class="sourceLineNo">183</span><a name="line.183"></a> -<span class="sourceLineNo">184</span> static class IntSumReducer<Key> extends Reducer<Key, IntWritable,<a name="line.184"></a> -<span class="sourceLineNo">185</span> Key, IntWritable> {<a name="line.185"></a> -<span class="sourceLineNo">186</span><a name="line.186"></a> -<span class="sourceLineNo">187</span> private IntWritable result = new IntWritable();<a name="line.187"></a> -<span class="sourceLineNo">188</span> public void reduce(Key key, Iterable<IntWritable> values,<a name="line.188"></a> -<span class="sourceLineNo">189</span> Context context)<a name="line.189"></a> -<span class="sourceLineNo">190</span> throws IOException, InterruptedException {<a name="line.190"></a> -<span class="sourceLineNo">191</span> int sum = 0;<a name="line.191"></a> -<span class="sourceLineNo">192</span> for (IntWritable val : values) {<a name="line.192"></a> -<span class="sourceLineNo">193</span> sum += val.get();<a name="line.193"></a> -<span class="sourceLineNo">194</span> }<a name="line.194"></a> -<span class="sourceLineNo">195</span> result.set(sum);<a name="line.195"></a> -<span class="sourceLineNo">196</span> context.write(key, result);<a name="line.196"></a> -<span class="sourceLineNo">197</span> }<a name="line.197"></a> -<span class="sourceLineNo">198</span> }<a name="line.198"></a> -<span class="sourceLineNo">199</span><a name="line.199"></a> -<span class="sourceLineNo">200</span> /**<a name="line.200"></a> -<span class="sourceLineNo">201</span> * Sets up the actual job.<a name="line.201"></a> -<span class="sourceLineNo">202</span> *<a name="line.202"></a> -<span class="sourceLineNo">203</span> * @param conf The current configuration.<a name="line.203"></a> -<span class="sourceLineNo">204</span> * @param args The command line parameters.<a name="line.204"></a> -<span class="sourceLineNo">205</span> * @return The newly created job.<a name="line.205"></a> -<span class="sourceLineNo">206</span> * @throws IOException When setting up the job fails.<a name="line.206"></a> -<span class="sourceLineNo">207</span> */<a name="line.207"></a> -<span class="sourceLineNo">208</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.208"></a> -<span class="sourceLineNo">209</span> throws IOException {<a name="line.209"></a> -<span class="sourceLineNo">210</span> String tableName = args[0];<a name="line.210"></a> -<span class="sourceLineNo">211</span> Path outputDir = new Path(args[1]);<a name="line.211"></a> -<span class="sourceLineNo">212</span> String reportSeparatorString = (args.length > 2) ? args[2]: ":";<a name="line.212"></a> -<span class="sourceLineNo">213</span> conf.set("ReportSeparator", reportSeparatorString);<a name="line.213"></a> -<span class="sourceLineNo">214</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.214"></a> -<span class="sourceLineNo">215</span> job.setJarByClass(CellCounter.class);<a name="line.215"></a> -<span class="sourceLineNo">216</span> Scan scan = getConfiguredScanForJob(conf, args);<a name="line.216"></a> -<span class="sourceLineNo">217</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.217"></a> -<span class="sourceLineNo">218</span> CellCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.218"></a> -<span class="sourceLineNo">219</span> job.setNumReduceTasks(1);<a name="line.219"></a> -<span class="sourceLineNo">220</span> job.setMapOutputKeyClass(Text.class);<a name="line.220"></a> -<span class="sourceLineNo">221</span> job.setMapOutputValueClass(IntWritable.class);<a name="line.221"></a> -<span class="sourceLineNo">222</span> job.setOutputFormatClass(TextOutputFormat.class);<a name="line.222"></a> -<span class="sourceLineNo">223</span> job.setOutputKeyClass(Text.class);<a name="line.223"></a> -<span class="sourceLineNo">224</span> job.setOutputValueClass(IntWritable.class);<a name="line.224"></a> -<span class="sourceLineNo">225</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.225"></a> -<span class="sourceLineNo">226</span> job.setReducerClass(IntSumReducer.class);<a name="line.226"></a> -<span class="sourceLineNo">227</span> return job;<a name="line.227"></a> -<span class="sourceLineNo">228</span> }<a name="line.228"></a> -<span class="sourceLineNo">229</span><a name="line.229"></a> -<span class="sourceLineNo">230</span> private static Scan getConfiguredScanForJob(Configuration conf, String[] args)<a name="line.230"></a> -<span class="sourceLineNo">231</span> throws IOException {<a name="line.231"></a> -<span class="sourceLineNo">232</span> // create scan with any properties set from TableInputFormat<a name="line.232"></a> -<span class="sourceLineNo">233</span> Scan s = TableInputFormat.createScanFromConfiguration(conf);<a name="line.233"></a> -<span class="sourceLineNo">234</span> // Set Scan Versions<a name="line.234"></a> -<span class="sourceLineNo">235</span> if (conf.get(TableInputFormat.SCAN_MAXVERSIONS) == null) {<a name="line.235"></a> -<span class="sourceLineNo">236</span> // default to all versions unless explicitly set<a name="line.236"></a> -<span class="sourceLineNo">237</span> s.setMaxVersions(Integer.MAX_VALUE);<a name="line.237"></a> -<span class="sourceLineNo">238</span> }<a name="line.238"></a> -<span class="sourceLineNo">239</span> s.setCacheBlocks(false);<a name="line.239"></a> -<span class="sourceLineNo">240</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.240"></a> -<span class="sourceLineNo">241</span> Filter rowFilter = getRowFilter(args);<a name="line.241"></a> -<span class="sourceLineNo">242</span> if (rowFilter!= null) {<a name="line.242"></a> -<span class="sourceLineNo">243</span> LOG.info("Setting Row Filter for counter.");<a name="line.243"></a> -<span class="sourceLineNo">244</span> s.setFilter(rowFilter);<a name="line.244"></a> -<span class="sourceLineNo">245</span> }<a name="line.245"></a> -<span class="sourceLineNo">246</span> // Set TimeRange if defined<a name="line.246"></a> -<span class="sourceLineNo">247</span> long timeRange[] = getTimeRange(args);<a name="line.247"></a> -<span class="sourceLineNo">248</span> if (timeRange != null) {<a name="line.248"></a> -<span class="sourceLineNo">249</span> LOG.info("Setting TimeRange for counter.");<a name="line.249"></a> -<span class="sourceLineNo">250</span> s.setTimeRange(timeRange[0], timeRange[1]);<a name="line.250"></a> -<span class="sourceLineNo">251</span> }<a name="line.251"></a> -<span class="sourceLineNo">252</span> return s;<a name="line.252"></a> -<span class="sourceLineNo">253</span> }<a name="line.253"></a> -<span class="sourceLineNo">254</span><a name="line.254"></a> +<span class="sourceLineNo">179</span> LOG.error("Interrupted while writing cellCount", e);<a name="line.179"></a> +<span class="sourceLineNo">180</span> Thread.currentThread().interrupt();<a name="line.180"></a> +<span class="sourceLineNo">181</span> }<a name="line.181"></a> +<span class="sourceLineNo">182</span> }<a name="line.182"></a> +<span class="sourceLineNo">183</span> }<a name="line.183"></a> +<span class="sourceLineNo">184</span><a name="line.184"></a> +<span class="sourceLineNo">185</span> static class IntSumReducer<Key> extends Reducer<Key, IntWritable,<a name="line.185"></a> +<span class="sourceLineNo">186</span> Key, IntWritable> {<a name="line.186"></a> +<span class="sourceLineNo">187</span><a name="line.187"></a> +<span class="sourceLineNo">188</span> private IntWritable result = new IntWritable();<a name="line.188"></a> +<span class="sourceLineNo">189</span> public void reduce(Key key, Iterable<IntWritable> values,<a name="line.189"></a> +<span class="sourceLineNo">190</span> Context context)<a name="line.190"></a> +<span class="sourceLineNo">191</span> throws IOException, InterruptedException {<a name="line.191"></a> +<span class="sourceLineNo">192</span> int sum = 0;<a name="line.192"></a> +<span class="sourceLineNo">193</span> for (IntWritable val : values) {<a name="line.193"></a> +<span class="sourceLineNo">194</span> sum += val.get();<a name="line.194"></a> +<span class="sourceLineNo">195</span> }<a name="line.195"></a> +<span class="sourceLineNo">196</span> result.set(sum);<a name="line.196"></a> +<span class="sourceLineNo">197</span> context.write(key, result);<a name="line.197"></a> +<span class="sourceLineNo">198</span> }<a name="line.198"></a> +<span class="sourceLineNo">199</span> }<a name="line.199"></a> +<span class="sourceLineNo">200</span><a name="line.200"></a> +<span class="sourceLineNo">201</span> /**<a name="line.201"></a> +<span class="sourceLineNo">202</span> * Sets up the actual job.<a name="line.202"></a> +<span class="sourceLineNo">203</span> *<a name="line.203"></a> +<span class="sourceLineNo">204</span> * @param conf The current configuration.<a name="line.204"></a> +<span class="sourceLineNo">205</span> * @param args The command line parameters.<a name="line.205"></a> +<span class="sourceLineNo">206</span> * @return The newly created job.<a name="line.206"></a> +<span class="sourceLineNo">207</span> * @throws IOException When setting up the job fails.<a name="line.207"></a> +<span class="sourceLineNo">208</span> */<a name="line.208"></a> +<span class="sourceLineNo">209</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.209"></a> +<span class="sourceLineNo">210</span> throws IOException {<a name="line.210"></a> +<span class="sourceLineNo">211</span> String tableName = args[0];<a name="line.211"></a> +<span class="sourceLineNo">212</span> Path outputDir = new Path(args[1]);<a name="line.212"></a> +<span class="sourceLineNo">213</span> String reportSeparatorString = (args.length > 2) ? args[2]: ":";<a name="line.213"></a> +<span class="sourceLineNo">214</span> conf.set("ReportSeparator", reportSeparatorString);<a name="line.214"></a> +<span class="sourceLineNo">215</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.215"></a> +<span class="sourceLineNo">216</span> job.setJarByClass(CellCounter.class);<a name="line.216"></a> +<span class="sourceLineNo">217</span> Scan scan = getConfiguredScanForJob(conf, args);<a name="line.217"></a> +<span class="sourceLineNo">218</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.218"></a> +<span class="sourceLineNo">219</span> CellCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.219"></a> +<span class="sourceLineNo">220</span> job.setNumReduceTasks(1);<a name="line.220"></a> +<span class="sourceLineNo">221</span> job.setMapOutputKeyClass(Text.class);<a name="line.221"></a> +<span class="sourceLineNo">222</span> job.setMapOutputValueClass(IntWritable.class);<a name="line.222"></a> +<span class="sourceLineNo">223</span> job.setOutputFormatClass(TextOutputFormat.class);<a name="line.223"></a> +<span class="sourceLineNo">224</span> job.setOutputKeyClass(Text.class);<a name="line.224"></a> +<span class="sourceLineNo">225</span> job.setOutputValueClass(IntWritable.class);<a name="line.225"></a> +<span class="sourceLineNo">226</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.226"></a> +<span class="sourceLineNo">227</span> job.setReducerClass(IntSumReducer.class);<a name="line.227"></a> +<span class="sourceLineNo">228</span> return job;<a name="line.228"></a> +<span class="sourceLineNo">229</span> }<a name="line.229"></a> +<span class="sourceLineNo">230</span><a name="line.230"></a> +<span class="sourceLineNo">231</span> private static Scan getConfiguredScanForJob(Configuration conf, String[] args)<a name="line.231"></a> +<span class="sourceLineNo">232</span> throws IOException {<a name="line.232"></a> +<span class="sourceLineNo">233</span> // create scan with any properties set from TableInputFormat<a name="line.233"></a> +<span class="sourceLineNo">234</span> Scan s = TableInputFormat.createScanFromConfiguration(conf);<a name="line.234"></a> +<span class="sourceLineNo">235</span> // Set Scan Versions<a name="line.235"></a> +<span class="sourceLineNo">236</span> if (conf.get(TableInputFormat.SCAN_MAXVERSIONS) == null) {<a name="line.236"></a> +<span class="sourceLineNo">237</span> // default to all versions unless explicitly set<a name="line.237"></a> +<span class="sourceLineNo">238</span> s.setMaxVersions(Integer.MAX_VALUE);<a name="line.238"></a> +<span class="sourceLineNo">239</span> }<a name="line.239"></a> +<span class="sourceLineNo">240</span> s.setCacheBlocks(false);<a name="line.240"></a> +<span class="sourceLineNo">241</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.241"></a> +<span class="sourceLineNo">242</span> Filter rowFilter = getRowFilter(args);<a name="line.242"></a> +<span class="sourceLineNo">243</span> if (rowFilter!= null) {<a name="line.243"></a> +<span class="sourceLineNo">244</span> LOG.info("Setting Row Filter for counter.");<a name="line.244"></a> +<span class="sourceLineNo">245</span> s.setFilter(rowFilter);<a name="line.245"></a> +<span class="sourceLineNo">246</span> }<a name="line.246"></a> +<span class="sourceLineNo">247</span> // Set TimeRange if defined<a name="line.247"></a> +<span class="sourceLineNo">248</span> long timeRange[] = getTimeRange(args);<a name="line.248"></a> +<span class="sourceLineNo">249</span> if (timeRange != null) {<a name="line.249"></a> +<span class="sourceLineNo">250</span> LOG.info("Setting TimeRange for counter.");<a name="line.250"></a> +<span class="sourceLineNo">251</span> s.setTimeRange(timeRange[0], timeRange[1]);<a name="line.251"></a> +<span class="sourceLineNo">252</span> }<a name="line.252"></a> +<span class="sourceLineNo">253</span> return s;<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> private static Filter getRowFilter(String[] args) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> Filter rowFilter = null;<a name="line.257"></a> -<span class="sourceLineNo">258</span> String filterCriteria = (args.length > 3) ? args[3]: null;<a name="line.258"></a> -<span class="sourceLineNo">259</span> if (filterCriteria == null) return null;<a name="line.259"></a> -<span class="sourceLineNo">260</span> if (filterCriteria.startsWith("^")) {<a name="line.260"></a> -<span class="sourceLineNo">261</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.261"></a> -<span class="sourceLineNo">262</span> rowFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.262"></a> -<span class="sourceLineNo">263</span> } else {<a name="line.263"></a> -<span class="sourceLineNo">264</span> rowFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.264"></a> -<span class="sourceLineNo">265</span> }<a name="line.265"></a> -<span class="sourceLineNo">266</span> return rowFilter;<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> private static long[] getTimeRange(String[] args) throws IOException {<a name="line.269"></a> -<span class="sourceLineNo">270</span> final String startTimeArgKey = "--starttime=";<a name="line.270"></a> -<span class="sourceLineNo">271</span> final String endTimeArgKey = "--endtime=";<a name="line.271"></a> -<span class="sourceLineNo">272</span> long startTime = 0L;<a name="line.272"></a> -<span class="sourceLineNo">273</span> long endTime = 0L;<a name="line.273"></a> -<span class="sourceLineNo">274</span><a name="line.274"></a> -<span class="sourceLineNo">275</span> for (int i = 1; i < args.length; i++) {<a name="line.275"></a> -<span class="sourceLineNo">276</span> System.out.println("i:" + i + "arg[i]" + args[i]);<a name="line.276"></a> -<span class="sourceLineNo">277</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.277"></a> -<span class="sourceLineNo">278</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.278"></a> -<span class="sourceLineNo">279</span> }<a name="line.279"></a> -<span class="sourceLineNo">280</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.280"></a> -<span class="sourceLineNo">281</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<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><a name="line.284"></a> -<span class="sourceLineNo">285</span> if (startTime == 0 && endTime == 0)<a name="line.285"></a> -<span class="sourceLineNo">286</span> return null;<a name="line.286"></a> -<span class="sourceLineNo">287</span><a name="line.287"></a> -<span class="sourceLineNo">288</span> endTime = endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime;<a name="line.288"></a> -<span class="sourceLineNo">289</span> return new long [] {startTime, endTime};<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> @Override<a name="line.292"></a> -<span class="sourceLineNo">293</span> public int run(String[] args) throws Exception {<a name="line.293"></a> -<span class="sourceLineNo">294</span> if (args.length < 2) {<a name="line.294"></a> -<span class="sourceLineNo">295</span> printUsage(args.length);<a name="line.295"></a> -<span class="sourceLineNo">296</span> return -1;<a name="line.296"></a> -<span class="sourceLineNo">297</span> }<a name="line.297"></a> -<span class="sourceLineNo">298</span> Job job = createSubmittableJob(getConf(), args);<a name="line.298"></a> -<span class="sourceLineNo">299</span> return (job.waitForCompletion(true) ? 0 : 1);<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">302</span> private void printUsage(int parameterCount) {<a name="line.302"></a> -<span class="sourceLineNo">303</span> System.err.println("ERROR: Wrong number of parameters: " + parameterCount);<a name="line.303"></a> -<span class="sourceLineNo">304</span> System.err.println("Usage: hbase cellcounter <tablename> <outputDir> [reportSeparator] "<a name="line.304"></a> -<span class="sourceLineNo">305</span> + "[^[regex pattern] or [Prefix]] [--starttime=<starttime> --endtime=<endtime>]");<a name="line.305"></a> -<span class="sourceLineNo">306</span> System.err.println(" Note: -D properties will be applied to the conf used.");<a name="line.306"></a> -<span class="sourceLineNo">307</span> System.err.println(" Additionally, all of the SCAN properties from TableInputFormat can be "<a name="line.307"></a> -<span class="sourceLineNo">308</span> + "specified to get fine grained control on what is counted.");<a name="line.308"></a> -<span class="sourceLineNo">309</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_START + "=<rowkey>");<a name="line.309"></a> -<span class="sourceLineNo">310</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_STOP + "=<rowkey>");<a name="line.310"></a> -<span class="sourceLineNo">311</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMNS + "=\"<col1> <col2>...\"");<a name="line.311"></a> -<span class="sourceLineNo">312</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMN_FAMILY<a name="line.312"></a> -<span class="sourceLineNo">313</span> + "=<family1>,<family2>, ...");<a name="line.313"></a> -<span class="sourceLineNo">314</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMESTAMP + "=<timestamp>");<a name="line.314"></a> -<span class="sourceLineNo">315</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_START + "=<timestamp>");<a name="line.315"></a> -<span class="sourceLineNo">316</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_END + "=<timestamp>");<a name="line.316"></a> -<span class="sourceLineNo">317</span> System.err.println(" -D" + TableInputFormat.SCAN_MAXVERSIONS + "=<count>");<a name="line.317"></a> -<span class="sourceLineNo">318</span> System.err.println(" -D" + TableInputFormat.SCAN_CACHEDROWS + "=<count>");<a name="line.318"></a> -<span class="sourceLineNo">319</span> System.err.println(" -D" + TableInputFormat.SCAN_BATCHSIZE + "=<count>");<a name="line.319"></a> -<span class="sourceLineNo">320</span> System.err.println(" <reportSeparator> parameter can be used to override the default report "<a name="line.320"></a> -<span class="sourceLineNo">321</span> + "separator string : used to separate the rowId/column family name and qualifier name.");<a name="line.321"></a> -<span class="sourceLineNo">322</span> System.err.println(" [^[regex pattern] or [Prefix] parameter can be used to limit the cell "<a name="line.322"></a> -<span class="sourceLineNo">323</span> + "counter count operation to a limited subset of rows from the table based on regex or "<a name="line.323"></a> -<span class="sourceLineNo">324</span> + "prefix pattern.");<a name="line.324"></a> -<span class="sourceLineNo">325</span> }<a name="line.325"></a> -<span class="sourceLineNo">326</span><a name="line.326"></a> -<span class="sourceLineNo">327</span> /**<a name="line.327"></a> -<span class="sourceLineNo">328</span> * Main entry point.<a name="line.328"></a> -<span class="sourceLineNo">329</span> * @param args The command line parameters.<a name="line.329"></a> -<span class="sourceLineNo">330</span> * @throws Exception When running the job fails.<a name="line.330"></a> -<span class="sourceLineNo">331</span> */<a name="line.331"></a> -<span class="sourceLineNo">332</span> public static void main(String[] args) throws Exception {<a name="line.332"></a> -<span class="sourceLineNo">333</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new CellCounter(), args);<a name="line.333"></a> -<span class="sourceLineNo">334</span> System.exit(errCode);<a name="line.334"></a> -<span class="sourceLineNo">335</span> }<a name="line.335"></a> -<span class="sourceLineNo">336</span><a name="line.336"></a> -<span class="sourceLineNo">337</span>}<a name="line.337"></a> +<span class="sourceLineNo">256</span><a name="line.256"></a> +<span class="sourceLineNo">257</span> private static Filter getRowFilter(String[] args) {<a name="line.257"></a> +<span class="sourceLineNo">258</span> Filter rowFilter = null;<a name="line.258"></a> +<span class="sourceLineNo">259</span> String filterCriteria = (args.length > 3) ? args[3]: null;<a name="line.259"></a> +<span class="sourceLineNo">260</span> if (filterCriteria == null) return null;<a name="line.260"></a> +<span class="sourceLineNo">261</span> if (filterCriteria.startsWith("^")) {<a name="line.261"></a> +<span class="sourceLineNo">262</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.262"></a> +<span class="sourceLineNo">263</span> rowFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.263"></a> +<span class="sourceLineNo">264</span> } else {<a name="line.264"></a> +<span class="sourceLineNo">265</span> rowFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.265"></a> +<span class="sourceLineNo">266</span> }<a name="line.266"></a> +<span class="sourceLineNo">267</span> return rowFilter;<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> private static long[] getTimeRange(String[] args) throws IOException {<a name="line.270"></a> +<span class="sourceLineNo">271</span> final String startTimeArgKey = "--starttime=";<a name="line.271"></a> +<span class="sourceLineNo">272</span> final String endTimeArgKey = "--endtime=";<a name="line.272"></a> +<span class="sourceLineNo">273</span> long startTime = 0L;<a name="line.273"></a> +<span class="sourceLineNo">274</span> long endTime = 0L;<a name="line.274"></a> +<span class="sourceLineNo">275</span><a name="line.275"></a> +<span class="sourceLineNo">276</span> for (int i = 1; i < args.length; i++) {<a name="line.276"></a> +<span class="sourceLineNo">277</span> System.out.println("i:" + i + "arg[i]" + args[i]);<a name="line.277"></a> +<span class="sourceLineNo">278</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.278"></a> +<span class="sourceLineNo">279</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.279"></a> +<span class="sourceLineNo">280</span> }<a name="line.280"></a> +<span class="sourceLineNo">281</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.281"></a> +<span class="sourceLineNo">282</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.282"></a> +<span class="sourceLineNo">283</span> }<a name="line.283"></a> +<span class="sourceLineNo">284</span> }<a name="line.284"></a> +<span class="sourceLineNo">285</span><a name="line.285"></a> +<span class="sourceLineNo">286</span> if (startTime == 0 && endTime == 0)<a name="line.286"></a> +<span class="sourceLineNo">287</span> return null;<a name="line.287"></a> +<span class="sourceLineNo">288</span><a name="line.288"></a> +<span class="sourceLineNo">289</span> endTime = endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime;<a name="line.289"></a> +<span class="sourceLineNo">290</span> return new long [] {startTime, endTime};<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> @Override<a name="line.293"></a> +<span class="sourceLineNo">294</span> public int run(String[] args) throws Exception {<a name="line.294"></a> +<span class="sourceLineNo">295</span> if (args.length < 2) {<a name="line.295"></a> +<span class="sourceLineNo">296</span> printUsage(args.length);<a name="line.296"></a> +<span class="sourceLineNo">297</span> return -1;<a name="line.297"></a> +<span class="sourceLineNo">298</span> }<a name="line.298"></a> +<span class="sourceLineNo">299</span> Job job = createSubmittableJob(getConf(), args);<a name="line.299"></a> +<span class="sourceLineNo">300</span> return (job.waitForCompletion(true) ? 0 : 1);<a name="line.300"></a> +<span class="sourceLineNo">301</span> }<a name="line.301"></a> +<span class="sourceLineNo">302</span><a name="line.302"></a> +<span class="sourceLineNo">303</span> private void printUsage(int parameterCount) {<a name="line.303"></a> +<span class="sourceLineNo">304</span> System.err.println("ERROR: Wrong number of parameters: " + parameterCount);<a name="line.304"></a> +<span class="sourceLineNo">305</span> System.err.println("Usage: hbase cellcounter <tablename> <outputDir> [reportSeparator] "<a name="line.305"></a> +<span class="sourceLineNo">306</span> + "[^[regex pattern] or [Prefix]] [--starttime=<starttime> --endtime=<endtime>]");<a name="line.306"></a> +<span class="sourceLineNo">307</span> System.err.println(" Note: -D properties will be applied to the conf used.");<a name="line.307"></a> +<span class="sourceLineNo">308</span> System.err.println(" Additionally, all of the SCAN properties from TableInputFormat can be "<a name="line.308"></a> +<span class="sourceLineNo">309</span> + "specified to get fine grained control on what is counted.");<a name="line.309"></a> +<span class="sourceLineNo">310</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_START + "=<rowkey>");<a name="line.310"></a> +<span class="sourceLineNo">311</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_STOP + "=<rowkey>");<a name="line.311"></a> +<span class="sourceLineNo">312</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMNS + "=\"<col1> <col2>...\"");<a name="line.312"></a> +<span class="sourceLineNo">313</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMN_FAMILY<a name="line.313"></a> +<span class="sourceLineNo">314</span> + "=<family1>,<family2>, ...");<a name="line.314"></a> +<span class="sourceLineNo">315</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMESTAMP + "=<timestamp>");<a name="line.315"></a> +<span class="sourceLineNo">316</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_START + "=<timestamp>");<a name="line.316"></a> +<span class="sourceLineNo">317</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_END + "=<timestamp>");<a name="line.317"></a> +<span class="sourceLineNo">318</span> System.err.println(" -D" + TableInputFormat.SCAN_MAXVERSIONS + "=<count>");<a name="line.318"></a> +<span class="sourceLineNo">319</span> System.err.println(" -D" + TableInputFormat.SCAN_CACHEDROWS + "=<count>");<a name="line.319"></a> +<span class="sourceLineNo">320</span> System.err.println(" -D" + TableInputFormat.SCAN_BATCHSIZE + "=<count>");<a name="line.320"></a> +<span class="sourceLineNo">321</span> System.err.println(" <reportSeparator> parameter can be used to override the default report "<a name="line.321"></a> +<span class="sourceLineNo">322</span> + "separator string : used to separate the rowId/column family name and qualifier name.");<a name="line.322"></a> +<span class="sourceLineNo">323</span> System.err.println(" [^[regex pattern] or [Prefix] parameter can be used to limit the cell "<a name="line.323"></a> +<span class="sourceLineNo">324</span> + "counter count operation to a limited subset of rows from the table based on regex or "<a name="line.324"></a> +<span class="sourceLineNo">325</span> + "prefix pattern.");<a name="line.325"></a> +<span class="sourceLineNo">326</span> }<a name="line.326"></a> +<span class="sourceLineNo">327</span><a name="line.327"></a> +<span class="sourceLineNo">328</span> /**<a name="line.328"></a> +<span class="sourceLineNo">329</span> * Main entry point.<a name="line.329"></a> +<span class="sourceLineNo">330</span> * @param args The command line parameters.<a name="line.330"></a> +<span class="sourceLineNo">331</span> * @throws Exception When running the job fails.<a name="line.331"></a> +<span class="sourceLineNo">332</span> */<a name="line.332"></a> +<span class="sourceLineNo">333</span> public static void main(String[] args) throws Exception {<a name="line.333"></a> +<span class="sourceLineNo">334</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new CellCounter(), args);<a name="line.334"></a> +<span class="sourceLineNo">335</span> System.exit(errCode);<a name="line.335"></a> +<span class="sourceLineNo">336</span> }<a name="line.336"></a> +<span class="sourceLineNo">337</span><a name="line.337"></a> +<span class="sourceLineNo">338</span>}<a name="line.338"></a>