http://git-wip-us.apache.org/repos/asf/hbase-site/blob/6444c276/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableOutputFormat.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableOutputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableOutputFormat.html index 2cffd19..126b093 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableOutputFormat.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapred/TableOutputFormat.html @@ -61,66 +61,71 @@ <span class="sourceLineNo">053</span> */<a name="line.53"></a> <span class="sourceLineNo">054</span> protected static class TableRecordWriter implements RecordWriter<ImmutableBytesWritable, Put> {<a name="line.54"></a> <span class="sourceLineNo">055</span> private BufferedMutator m_mutator;<a name="line.55"></a> -<span class="sourceLineNo">056</span> private Connection connection;<a name="line.56"></a> -<span class="sourceLineNo">057</span> /**<a name="line.57"></a> -<span class="sourceLineNo">058</span> * Instantiate a TableRecordWriter with the HBase HClient for writing. Assumes control over the<a name="line.58"></a> -<span class="sourceLineNo">059</span> * lifecycle of {@code conn}.<a name="line.59"></a> +<span class="sourceLineNo">056</span> private Connection conn;<a name="line.56"></a> +<span class="sourceLineNo">057</span><a name="line.57"></a> +<span class="sourceLineNo">058</span> /**<a name="line.58"></a> +<span class="sourceLineNo">059</span> * Instantiate a TableRecordWriter with a BufferedMutator for batch writing.<a name="line.59"></a> <span class="sourceLineNo">060</span> */<a name="line.60"></a> -<span class="sourceLineNo">061</span> public TableRecordWriter(final BufferedMutator mutator) throws IOException {<a name="line.61"></a> -<span class="sourceLineNo">062</span> this.m_mutator = mutator;<a name="line.62"></a> -<span class="sourceLineNo">063</span> }<a name="line.63"></a> -<span class="sourceLineNo">064</span><a name="line.64"></a> -<span class="sourceLineNo">065</span> public TableRecordWriter(JobConf job) throws IOException {<a name="line.65"></a> -<span class="sourceLineNo">066</span> // expecting exactly one path<a name="line.66"></a> -<span class="sourceLineNo">067</span> TableName tableName = TableName.valueOf(job.get(OUTPUT_TABLE));<a name="line.67"></a> -<span class="sourceLineNo">068</span> connection = ConnectionFactory.createConnection(job);<a name="line.68"></a> -<span class="sourceLineNo">069</span> m_mutator = connection.getBufferedMutator(tableName);<a name="line.69"></a> -<span class="sourceLineNo">070</span> }<a name="line.70"></a> -<span class="sourceLineNo">071</span><a name="line.71"></a> -<span class="sourceLineNo">072</span> public void close(Reporter reporter) throws IOException {<a name="line.72"></a> -<span class="sourceLineNo">073</span> this.m_mutator.close();<a name="line.73"></a> -<span class="sourceLineNo">074</span> if (connection != null) {<a name="line.74"></a> -<span class="sourceLineNo">075</span> connection.close();<a name="line.75"></a> -<span class="sourceLineNo">076</span> connection = null;<a name="line.76"></a> -<span class="sourceLineNo">077</span> }<a name="line.77"></a> -<span class="sourceLineNo">078</span> }<a name="line.78"></a> -<span class="sourceLineNo">079</span><a name="line.79"></a> -<span class="sourceLineNo">080</span> public void write(ImmutableBytesWritable key, Put value) throws IOException {<a name="line.80"></a> -<span class="sourceLineNo">081</span> m_mutator.mutate(new Put(value));<a name="line.81"></a> +<span class="sourceLineNo">061</span> public TableRecordWriter(JobConf job) throws IOException {<a name="line.61"></a> +<span class="sourceLineNo">062</span> // expecting exactly one path<a name="line.62"></a> +<span class="sourceLineNo">063</span> TableName tableName = TableName.valueOf(job.get(OUTPUT_TABLE));<a name="line.63"></a> +<span class="sourceLineNo">064</span> try {<a name="line.64"></a> +<span class="sourceLineNo">065</span> this.conn = ConnectionFactory.createConnection(job);<a name="line.65"></a> +<span class="sourceLineNo">066</span> this.m_mutator = conn.getBufferedMutator(tableName);<a name="line.66"></a> +<span class="sourceLineNo">067</span> } finally {<a name="line.67"></a> +<span class="sourceLineNo">068</span> if (this.m_mutator == null) {<a name="line.68"></a> +<span class="sourceLineNo">069</span> conn.close();<a name="line.69"></a> +<span class="sourceLineNo">070</span> conn = null;<a name="line.70"></a> +<span class="sourceLineNo">071</span> }<a name="line.71"></a> +<span class="sourceLineNo">072</span> }<a name="line.72"></a> +<span class="sourceLineNo">073</span> }<a name="line.73"></a> +<span class="sourceLineNo">074</span><a name="line.74"></a> +<span class="sourceLineNo">075</span> public void close(Reporter reporter) throws IOException {<a name="line.75"></a> +<span class="sourceLineNo">076</span> if (this.m_mutator != null) {<a name="line.76"></a> +<span class="sourceLineNo">077</span> this.m_mutator.close();<a name="line.77"></a> +<span class="sourceLineNo">078</span> }<a name="line.78"></a> +<span class="sourceLineNo">079</span> if (conn != null) {<a name="line.79"></a> +<span class="sourceLineNo">080</span> this.conn.close();<a name="line.80"></a> +<span class="sourceLineNo">081</span> }<a name="line.81"></a> <span class="sourceLineNo">082</span> }<a name="line.82"></a> -<span class="sourceLineNo">083</span> }<a name="line.83"></a> -<span class="sourceLineNo">084</span><a name="line.84"></a> -<span class="sourceLineNo">085</span> /**<a name="line.85"></a> -<span class="sourceLineNo">086</span> * Creates a new record writer.<a name="line.86"></a> -<span class="sourceLineNo">087</span> * <a name="line.87"></a> -<span class="sourceLineNo">088</span> * Be aware that the baseline javadoc gives the impression that there is a single<a name="line.88"></a> -<span class="sourceLineNo">089</span> * {@link RecordWriter} per job but in HBase, it is more natural if we give you a new<a name="line.89"></a> -<span class="sourceLineNo">090</span> * RecordWriter per call of this method. You must close the returned RecordWriter when done.<a name="line.90"></a> -<span class="sourceLineNo">091</span> * Failure to do so will drop writes.<a name="line.91"></a> -<span class="sourceLineNo">092</span> *<a name="line.92"></a> -<span class="sourceLineNo">093</span> * @param ignored Ignored filesystem<a name="line.93"></a> -<span class="sourceLineNo">094</span> * @param job Current JobConf<a name="line.94"></a> -<span class="sourceLineNo">095</span> * @param name Name of the job<a name="line.95"></a> -<span class="sourceLineNo">096</span> * @param progress<a name="line.96"></a> -<span class="sourceLineNo">097</span> * @return The newly created writer instance.<a name="line.97"></a> -<span class="sourceLineNo">098</span> * @throws IOException When creating the writer fails.<a name="line.98"></a> -<span class="sourceLineNo">099</span> */<a name="line.99"></a> -<span class="sourceLineNo">100</span> @Override<a name="line.100"></a> -<span class="sourceLineNo">101</span> public RecordWriter getRecordWriter(FileSystem ignored, JobConf job, String name,<a name="line.101"></a> -<span class="sourceLineNo">102</span> Progressable progress)<a name="line.102"></a> -<span class="sourceLineNo">103</span> throws IOException {<a name="line.103"></a> -<span class="sourceLineNo">104</span> return new TableRecordWriter(job);<a name="line.104"></a> -<span class="sourceLineNo">105</span> }<a name="line.105"></a> -<span class="sourceLineNo">106</span><a name="line.106"></a> -<span class="sourceLineNo">107</span> @Override<a name="line.107"></a> -<span class="sourceLineNo">108</span> public void checkOutputSpecs(FileSystem ignored, JobConf job)<a name="line.108"></a> -<span class="sourceLineNo">109</span> throws FileAlreadyExistsException, InvalidJobConfException, IOException {<a name="line.109"></a> -<span class="sourceLineNo">110</span> String tableName = job.get(OUTPUT_TABLE);<a name="line.110"></a> -<span class="sourceLineNo">111</span> if (tableName == null) {<a name="line.111"></a> -<span class="sourceLineNo">112</span> throw new IOException("Must specify table name");<a name="line.112"></a> -<span class="sourceLineNo">113</span> }<a name="line.113"></a> -<span class="sourceLineNo">114</span> }<a name="line.114"></a> -<span class="sourceLineNo">115</span>}<a name="line.115"></a> +<span class="sourceLineNo">083</span><a name="line.83"></a> +<span class="sourceLineNo">084</span> public void write(ImmutableBytesWritable key, Put value) throws IOException {<a name="line.84"></a> +<span class="sourceLineNo">085</span> m_mutator.mutate(new Put(value));<a name="line.85"></a> +<span class="sourceLineNo">086</span> }<a name="line.86"></a> +<span class="sourceLineNo">087</span> }<a name="line.87"></a> +<span class="sourceLineNo">088</span><a name="line.88"></a> +<span class="sourceLineNo">089</span> /**<a name="line.89"></a> +<span class="sourceLineNo">090</span> * Creates a new record writer.<a name="line.90"></a> +<span class="sourceLineNo">091</span> * <a name="line.91"></a> +<span class="sourceLineNo">092</span> * Be aware that the baseline javadoc gives the impression that there is a single<a name="line.92"></a> +<span class="sourceLineNo">093</span> * {@link RecordWriter} per job but in HBase, it is more natural if we give you a new<a name="line.93"></a> +<span class="sourceLineNo">094</span> * RecordWriter per call of this method. You must close the returned RecordWriter when done.<a name="line.94"></a> +<span class="sourceLineNo">095</span> * Failure to do so will drop writes.<a name="line.95"></a> +<span class="sourceLineNo">096</span> *<a name="line.96"></a> +<span class="sourceLineNo">097</span> * @param ignored Ignored filesystem<a name="line.97"></a> +<span class="sourceLineNo">098</span> * @param job Current JobConf<a name="line.98"></a> +<span class="sourceLineNo">099</span> * @param name Name of the job<a name="line.99"></a> +<span class="sourceLineNo">100</span> * @param progress<a name="line.100"></a> +<span class="sourceLineNo">101</span> * @return The newly created writer instance.<a name="line.101"></a> +<span class="sourceLineNo">102</span> * @throws IOException When creating the writer fails.<a name="line.102"></a> +<span class="sourceLineNo">103</span> */<a name="line.103"></a> +<span class="sourceLineNo">104</span> @Override<a name="line.104"></a> +<span class="sourceLineNo">105</span> public RecordWriter getRecordWriter(FileSystem ignored, JobConf job, String name,<a name="line.105"></a> +<span class="sourceLineNo">106</span> Progressable progress)<a name="line.106"></a> +<span class="sourceLineNo">107</span> throws IOException {<a name="line.107"></a> +<span class="sourceLineNo">108</span> // Clear write buffer on fail is true by default so no need to reset it.<a name="line.108"></a> +<span class="sourceLineNo">109</span> return new TableRecordWriter(job);<a name="line.109"></a> +<span class="sourceLineNo">110</span> }<a name="line.110"></a> +<span class="sourceLineNo">111</span><a name="line.111"></a> +<span class="sourceLineNo">112</span> @Override<a name="line.112"></a> +<span class="sourceLineNo">113</span> public void checkOutputSpecs(FileSystem ignored, JobConf job)<a name="line.113"></a> +<span class="sourceLineNo">114</span> throws FileAlreadyExistsException, InvalidJobConfException, IOException {<a name="line.114"></a> +<span class="sourceLineNo">115</span> String tableName = job.get(OUTPUT_TABLE);<a name="line.115"></a> +<span class="sourceLineNo">116</span> if (tableName == null) {<a name="line.116"></a> +<span class="sourceLineNo">117</span> throw new IOException("Must specify table name");<a name="line.117"></a> +<span class="sourceLineNo">118</span> }<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>
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/6444c276/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/Import.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/Import.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/Import.html index 654bc34..c72a9ce 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/Import.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/Import.html @@ -217,7 +217,7 @@ <span class="sourceLineNo">209</span> + Bytes.toString(row.get(), row.getOffset(), row.getLength()));<a name="line.209"></a> <span class="sourceLineNo">210</span> }<a name="line.210"></a> <span class="sourceLineNo">211</span> if (filter == null<a name="line.211"></a> -<span class="sourceLineNo">212</span> || !filter.filterRowKey(KeyValueUtil.createFirstOnRow(row.get(), row.getOffset(),<a name="line.212"></a> +<span class="sourceLineNo">212</span> || !filter.filterRowKey(CellUtil.createFirstOnRow(row.get(), row.getOffset(),<a name="line.212"></a> <span class="sourceLineNo">213</span> (short) row.getLength()))) {<a name="line.213"></a> <span class="sourceLineNo">214</span> for (Cell kv : value.rawCells()) {<a name="line.214"></a> <span class="sourceLineNo">215</span> kv = filterKv(filter, kv);<a name="line.215"></a> @@ -283,7 +283,7 @@ <span class="sourceLineNo">275</span> + Bytes.toString(row.get(), row.getOffset(), row.getLength()));<a name="line.275"></a> <span class="sourceLineNo">276</span> }<a name="line.276"></a> <span class="sourceLineNo">277</span> if (filter == null<a name="line.277"></a> -<span class="sourceLineNo">278</span> || !filter.filterRowKey(KeyValueUtil.createFirstOnRow(row.get(), row.getOffset(),<a name="line.278"></a> +<span class="sourceLineNo">278</span> || !filter.filterRowKey(CellUtil.createFirstOnRow(row.get(), row.getOffset(),<a name="line.278"></a> <span class="sourceLineNo">279</span> (short) row.getLength()))) {<a name="line.279"></a> <span class="sourceLineNo">280</span> for (Cell kv : value.rawCells()) {<a name="line.280"></a> <span class="sourceLineNo">281</span> kv = filterKv(filter, kv);<a name="line.281"></a> @@ -340,7 +340,7 @@ <span class="sourceLineNo">332</span> + Bytes.toString(key.get(), key.getOffset(), key.getLength()));<a name="line.332"></a> <span class="sourceLineNo">333</span> }<a name="line.333"></a> <span class="sourceLineNo">334</span> if (filter == null<a name="line.334"></a> -<span class="sourceLineNo">335</span> || !filter.filterRowKey(KeyValueUtil.createFirstOnRow(key.get(), key.getOffset(),<a name="line.335"></a> +<span class="sourceLineNo">335</span> || !filter.filterRowKey(CellUtil.createFirstOnRow(key.get(), key.getOffset(),<a name="line.335"></a> <span class="sourceLineNo">336</span> (short) key.getLength()))) {<a name="line.336"></a> <span class="sourceLineNo">337</span> processKV(key, result, context, put, delete);<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/6444c276/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html index ca3ad7f..e209a48 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/RowCounter.html @@ -27,105 +27,105 @@ <span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.19"></a> <span class="sourceLineNo">020</span><a name="line.20"></a> <span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a> -<span class="sourceLineNo">022</span><a name="line.22"></a> -<span class="sourceLineNo">023</span>import org.apache.commons.logging.Log;<a name="line.23"></a> -<span class="sourceLineNo">024</span>import org.apache.commons.logging.LogFactory;<a name="line.24"></a> -<span class="sourceLineNo">025</span>import org.apache.commons.lang.StringUtils;<a name="line.25"></a> -<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.HConstants;<a name="line.26"></a> -<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.27"></a> -<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.28"></a> -<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a> -<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configured;<a name="line.30"></a> -<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.31"></a> -<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.Result;<a name="line.32"></a> -<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.33"></a> -<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.filter.FirstKeyOnlyFilter;<a name="line.34"></a> -<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.35"></a> -<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.36"></a> -<span class="sourceLineNo">037</span>import org.apache.hadoop.mapreduce.Counter;<a name="line.37"></a> -<span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.Job;<a name="line.38"></a> -<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.39"></a> -<span class="sourceLineNo">040</span>import org.apache.hadoop.util.Tool;<a name="line.40"></a> -<span class="sourceLineNo">041</span>import org.apache.hadoop.util.ToolRunner;<a name="line.41"></a> -<span class="sourceLineNo">042</span><a name="line.42"></a> -<span class="sourceLineNo">043</span>/**<a name="line.43"></a> -<span class="sourceLineNo">044</span> * A job with a just a map phase to count rows. Map outputs table rows IF the<a name="line.44"></a> -<span class="sourceLineNo">045</span> * input row has columns that have content.<a name="line.45"></a> -<span class="sourceLineNo">046</span> */<a name="line.46"></a> -<span class="sourceLineNo">047</span>@InterfaceAudience.Public<a name="line.47"></a> -<span class="sourceLineNo">048</span>@InterfaceStability.Stable<a name="line.48"></a> -<span class="sourceLineNo">049</span>public class RowCounter extends Configured implements Tool {<a name="line.49"></a> -<span class="sourceLineNo">050</span><a name="line.50"></a> -<span class="sourceLineNo">051</span> private static final Log LOG = LogFactory.getLog(RowCounter.class);<a name="line.51"></a> -<span class="sourceLineNo">052</span><a name="line.52"></a> -<span class="sourceLineNo">053</span> /** Name of this 'program'. */<a name="line.53"></a> -<span class="sourceLineNo">054</span> static final String NAME = "rowcounter";<a name="line.54"></a> -<span class="sourceLineNo">055</span><a name="line.55"></a> -<span class="sourceLineNo">056</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.56"></a> -<span class="sourceLineNo">057</span> private final static String EXPECTED_COUNT_KEY = RowCounter.class.getName() + ".expected_count";<a name="line.57"></a> -<span class="sourceLineNo">058</span><a name="line.58"></a> -<span class="sourceLineNo">059</span> /**<a name="line.59"></a> -<span class="sourceLineNo">060</span> * Mapper that runs the count.<a name="line.60"></a> -<span class="sourceLineNo">061</span> */<a name="line.61"></a> -<span class="sourceLineNo">062</span> static class RowCounterMapper<a name="line.62"></a> -<span class="sourceLineNo">063</span> extends TableMapper<ImmutableBytesWritable, Result> {<a name="line.63"></a> -<span class="sourceLineNo">064</span><a name="line.64"></a> -<span class="sourceLineNo">065</span> /** Counter enumeration to count the actual rows. */<a name="line.65"></a> -<span class="sourceLineNo">066</span> public static enum Counters {ROWS}<a name="line.66"></a> -<span class="sourceLineNo">067</span><a name="line.67"></a> -<span class="sourceLineNo">068</span> /**<a name="line.68"></a> -<span class="sourceLineNo">069</span> * Maps the data.<a name="line.69"></a> -<span class="sourceLineNo">070</span> *<a name="line.70"></a> -<span class="sourceLineNo">071</span> * @param row The current table row key.<a name="line.71"></a> -<span class="sourceLineNo">072</span> * @param values The columns.<a name="line.72"></a> -<span class="sourceLineNo">073</span> * @param context The current context.<a name="line.73"></a> -<span class="sourceLineNo">074</span> * @throws IOException When something is broken with the data.<a name="line.74"></a> -<span class="sourceLineNo">075</span> * @see org.apache.hadoop.mapreduce.Mapper#map(KEYIN, VALUEIN,<a name="line.75"></a> -<span class="sourceLineNo">076</span> * org.apache.hadoop.mapreduce.Mapper.Context)<a name="line.76"></a> -<span class="sourceLineNo">077</span> */<a name="line.77"></a> -<span class="sourceLineNo">078</span> @Override<a name="line.78"></a> -<span class="sourceLineNo">079</span> public void map(ImmutableBytesWritable row, Result values,<a name="line.79"></a> -<span class="sourceLineNo">080</span> Context context)<a name="line.80"></a> -<span class="sourceLineNo">081</span> throws IOException {<a name="line.81"></a> -<span class="sourceLineNo">082</span> // Count every row containing data, whether it's in qualifiers or values<a name="line.82"></a> -<span class="sourceLineNo">083</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.83"></a> -<span class="sourceLineNo">084</span> }<a name="line.84"></a> -<span class="sourceLineNo">085</span> }<a name="line.85"></a> -<span class="sourceLineNo">086</span><a name="line.86"></a> -<span class="sourceLineNo">087</span> /**<a name="line.87"></a> -<span class="sourceLineNo">088</span> * Sets up the actual job.<a name="line.88"></a> -<span class="sourceLineNo">089</span> *<a name="line.89"></a> -<span class="sourceLineNo">090</span> * @param conf The current configuration.<a name="line.90"></a> -<span class="sourceLineNo">091</span> * @param args The command line parameters.<a name="line.91"></a> -<span class="sourceLineNo">092</span> * @return The newly created job.<a name="line.92"></a> -<span class="sourceLineNo">093</span> * @throws IOException When setting up the job fails.<a name="line.93"></a> -<span class="sourceLineNo">094</span> */<a name="line.94"></a> -<span class="sourceLineNo">095</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.95"></a> -<span class="sourceLineNo">096</span> throws IOException {<a name="line.96"></a> -<span class="sourceLineNo">097</span> String tableName = args[0];<a name="line.97"></a> -<span class="sourceLineNo">098</span> String startKey = null;<a name="line.98"></a> -<span class="sourceLineNo">099</span> String endKey = null;<a name="line.99"></a> -<span class="sourceLineNo">100</span> long startTime = 0;<a name="line.100"></a> -<span class="sourceLineNo">101</span> long endTime = 0;<a name="line.101"></a> -<span class="sourceLineNo">102</span><a name="line.102"></a> -<span class="sourceLineNo">103</span> StringBuilder sb = new StringBuilder();<a name="line.103"></a> -<span class="sourceLineNo">104</span><a name="line.104"></a> -<span class="sourceLineNo">105</span> final String rangeSwitch = "--range=";<a name="line.105"></a> -<span class="sourceLineNo">106</span> final String startTimeArgKey = "--starttime=";<a name="line.106"></a> -<span class="sourceLineNo">107</span> final String endTimeArgKey = "--endtime=";<a name="line.107"></a> -<span class="sourceLineNo">108</span> final String expectedCountArg = "--expected-count=";<a name="line.108"></a> -<span class="sourceLineNo">109</span><a name="line.109"></a> -<span class="sourceLineNo">110</span> // First argument is table name, starting from second<a name="line.110"></a> -<span class="sourceLineNo">111</span> for (int i = 1; i < args.length; i++) {<a name="line.111"></a> -<span class="sourceLineNo">112</span> if (args[i].startsWith(rangeSwitch)) {<a name="line.112"></a> -<span class="sourceLineNo">113</span> String[] startEnd = args[i].substring(rangeSwitch.length()).split(",", 2);<a name="line.113"></a> -<span class="sourceLineNo">114</span> if (startEnd.length != 2 || startEnd[1].contains(",")) {<a name="line.114"></a> -<span class="sourceLineNo">115</span> printUsage("Please specify range in such format as \"--range=a,b\" " +<a name="line.115"></a> -<span class="sourceLineNo">116</span> "or, with only one boundary, \"--range=,b\" or \"--range=a,\"");<a name="line.116"></a> -<span class="sourceLineNo">117</span> return null;<a name="line.117"></a> -<span class="sourceLineNo">118</span> }<a name="line.118"></a> -<span class="sourceLineNo">119</span> startKey = startEnd[0];<a name="line.119"></a> -<span class="sourceLineNo">120</span> endKey = startEnd[1];<a name="line.120"></a> +<span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a> +<span class="sourceLineNo">023</span>import java.util.ArrayList;<a name="line.23"></a> +<span class="sourceLineNo">024</span><a name="line.24"></a> +<span class="sourceLineNo">025</span>import org.apache.commons.logging.Log;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import org.apache.commons.logging.LogFactory;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import org.apache.commons.lang.StringUtils;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HConstants;<a name="line.28"></a> +<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.29"></a> +<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.classification.InterfaceStability;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import org.apache.hadoop.conf.Configuration;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import org.apache.hadoop.conf.Configured;<a name="line.32"></a> +<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.33"></a> +<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Result;<a name="line.34"></a> +<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.filter.FilterBase;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.filter.FirstKeyOnlyFilter;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.filter.MultiRowRangeFilter;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.Counter;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.Job;<a name="line.42"></a> +<span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.43"></a> +<span class="sourceLineNo">044</span>import org.apache.hadoop.util.Tool;<a name="line.44"></a> +<span class="sourceLineNo">045</span>import org.apache.hadoop.util.ToolRunner;<a name="line.45"></a> +<span class="sourceLineNo">046</span><a name="line.46"></a> +<span class="sourceLineNo">047</span>/**<a name="line.47"></a> +<span class="sourceLineNo">048</span> * A job with a just a map phase to count rows. Map outputs table rows IF the<a name="line.48"></a> +<span class="sourceLineNo">049</span> * input row has columns that have content.<a name="line.49"></a> +<span class="sourceLineNo">050</span> */<a name="line.50"></a> +<span class="sourceLineNo">051</span>@InterfaceAudience.Public<a name="line.51"></a> +<span class="sourceLineNo">052</span>@InterfaceStability.Stable<a name="line.52"></a> +<span class="sourceLineNo">053</span>public class RowCounter extends Configured implements Tool {<a name="line.53"></a> +<span class="sourceLineNo">054</span><a name="line.54"></a> +<span class="sourceLineNo">055</span> private static final Log LOG = LogFactory.getLog(RowCounter.class);<a name="line.55"></a> +<span class="sourceLineNo">056</span><a name="line.56"></a> +<span class="sourceLineNo">057</span> /** Name of this 'program'. */<a name="line.57"></a> +<span class="sourceLineNo">058</span> static final String NAME = "rowcounter";<a name="line.58"></a> +<span class="sourceLineNo">059</span><a name="line.59"></a> +<span class="sourceLineNo">060</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.60"></a> +<span class="sourceLineNo">061</span> private final static String EXPECTED_COUNT_KEY = RowCounter.class.getName() + ".expected_count";<a name="line.61"></a> +<span class="sourceLineNo">062</span><a name="line.62"></a> +<span class="sourceLineNo">063</span> /**<a name="line.63"></a> +<span class="sourceLineNo">064</span> * Mapper that runs the count.<a name="line.64"></a> +<span class="sourceLineNo">065</span> */<a name="line.65"></a> +<span class="sourceLineNo">066</span> static class RowCounterMapper<a name="line.66"></a> +<span class="sourceLineNo">067</span> extends TableMapper<ImmutableBytesWritable, Result> {<a name="line.67"></a> +<span class="sourceLineNo">068</span><a name="line.68"></a> +<span class="sourceLineNo">069</span> /** Counter enumeration to count the actual rows. */<a name="line.69"></a> +<span class="sourceLineNo">070</span> public static enum Counters {ROWS}<a name="line.70"></a> +<span class="sourceLineNo">071</span><a name="line.71"></a> +<span class="sourceLineNo">072</span> /**<a name="line.72"></a> +<span class="sourceLineNo">073</span> * Maps the data.<a name="line.73"></a> +<span class="sourceLineNo">074</span> *<a name="line.74"></a> +<span class="sourceLineNo">075</span> * @param row The current table row key.<a name="line.75"></a> +<span class="sourceLineNo">076</span> * @param values The columns.<a name="line.76"></a> +<span class="sourceLineNo">077</span> * @param context The current context.<a name="line.77"></a> +<span class="sourceLineNo">078</span> * @throws IOException When something is broken with the data.<a name="line.78"></a> +<span class="sourceLineNo">079</span> * @see org.apache.hadoop.mapreduce.Mapper#map(KEYIN, VALUEIN,<a name="line.79"></a> +<span class="sourceLineNo">080</span> * org.apache.hadoop.mapreduce.Mapper.Context)<a name="line.80"></a> +<span class="sourceLineNo">081</span> */<a name="line.81"></a> +<span class="sourceLineNo">082</span> @Override<a name="line.82"></a> +<span class="sourceLineNo">083</span> public void map(ImmutableBytesWritable row, Result values,<a name="line.83"></a> +<span class="sourceLineNo">084</span> Context context)<a name="line.84"></a> +<span class="sourceLineNo">085</span> throws IOException {<a name="line.85"></a> +<span class="sourceLineNo">086</span> // Count every row containing data, whether it's in qualifiers or values<a name="line.86"></a> +<span class="sourceLineNo">087</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.87"></a> +<span class="sourceLineNo">088</span> }<a name="line.88"></a> +<span class="sourceLineNo">089</span> }<a name="line.89"></a> +<span class="sourceLineNo">090</span><a name="line.90"></a> +<span class="sourceLineNo">091</span> /**<a name="line.91"></a> +<span class="sourceLineNo">092</span> * Sets up the actual job.<a name="line.92"></a> +<span class="sourceLineNo">093</span> *<a name="line.93"></a> +<span class="sourceLineNo">094</span> * @param conf The current configuration.<a name="line.94"></a> +<span class="sourceLineNo">095</span> * @param args The command line parameters.<a name="line.95"></a> +<span class="sourceLineNo">096</span> * @return The newly created job.<a name="line.96"></a> +<span class="sourceLineNo">097</span> * @throws IOException When setting up the job fails.<a name="line.97"></a> +<span class="sourceLineNo">098</span> */<a name="line.98"></a> +<span class="sourceLineNo">099</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.99"></a> +<span class="sourceLineNo">100</span> throws IOException {<a name="line.100"></a> +<span class="sourceLineNo">101</span> String tableName = args[0];<a name="line.101"></a> +<span class="sourceLineNo">102</span> List<MultiRowRangeFilter.RowRange> rowRangeList = null;<a name="line.102"></a> +<span class="sourceLineNo">103</span> long startTime = 0;<a name="line.103"></a> +<span class="sourceLineNo">104</span> long endTime = 0;<a name="line.104"></a> +<span class="sourceLineNo">105</span><a name="line.105"></a> +<span class="sourceLineNo">106</span> StringBuilder sb = new StringBuilder();<a name="line.106"></a> +<span class="sourceLineNo">107</span><a name="line.107"></a> +<span class="sourceLineNo">108</span> final String rangeSwitch = "--range=";<a name="line.108"></a> +<span class="sourceLineNo">109</span> final String startTimeArgKey = "--starttime=";<a name="line.109"></a> +<span class="sourceLineNo">110</span> final String endTimeArgKey = "--endtime=";<a name="line.110"></a> +<span class="sourceLineNo">111</span> final String expectedCountArg = "--expected-count=";<a name="line.111"></a> +<span class="sourceLineNo">112</span><a name="line.112"></a> +<span class="sourceLineNo">113</span> // First argument is table name, starting from second<a name="line.113"></a> +<span class="sourceLineNo">114</span> for (int i = 1; i < args.length; i++) {<a name="line.114"></a> +<span class="sourceLineNo">115</span> if (args[i].startsWith(rangeSwitch)) {<a name="line.115"></a> +<span class="sourceLineNo">116</span> try {<a name="line.116"></a> +<span class="sourceLineNo">117</span> rowRangeList = parseRowRangeParameter(args[i], rangeSwitch);<a name="line.117"></a> +<span class="sourceLineNo">118</span> } catch (IllegalArgumentException e) {<a name="line.118"></a> +<span class="sourceLineNo">119</span> return null;<a name="line.119"></a> +<span class="sourceLineNo">120</span> }<a name="line.120"></a> <span class="sourceLineNo">121</span> continue;<a name="line.121"></a> <span class="sourceLineNo">122</span> }<a name="line.122"></a> <span class="sourceLineNo">123</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.123"></a> @@ -154,89 +154,131 @@ <span class="sourceLineNo">146</span> job.setJarByClass(RowCounter.class);<a name="line.146"></a> <span class="sourceLineNo">147</span> Scan scan = new Scan();<a name="line.147"></a> <span class="sourceLineNo">148</span> scan.setCacheBlocks(false);<a name="line.148"></a> -<span class="sourceLineNo">149</span> if (startKey != null && !startKey.equals("")) {<a name="line.149"></a> -<span class="sourceLineNo">150</span> scan.setStartRow(Bytes.toBytesBinary(startKey));<a name="line.150"></a> -<span class="sourceLineNo">151</span> }<a name="line.151"></a> -<span class="sourceLineNo">152</span> if (endKey != null && !endKey.equals("")) {<a name="line.152"></a> -<span class="sourceLineNo">153</span> scan.setStopRow(Bytes.toBytesBinary(endKey));<a name="line.153"></a> -<span class="sourceLineNo">154</span> }<a name="line.154"></a> -<span class="sourceLineNo">155</span> if (sb.length() > 0) {<a name="line.155"></a> -<span class="sourceLineNo">156</span> for (String columnName : sb.toString().trim().split(" ")) {<a name="line.156"></a> -<span class="sourceLineNo">157</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.157"></a> -<span class="sourceLineNo">158</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.158"></a> -<span class="sourceLineNo">159</span><a name="line.159"></a> -<span class="sourceLineNo">160</span> if (StringUtils.isBlank(qualifier)) {<a name="line.160"></a> -<span class="sourceLineNo">161</span> scan.addFamily(Bytes.toBytes(family));<a name="line.161"></a> -<span class="sourceLineNo">162</span> }<a name="line.162"></a> -<span class="sourceLineNo">163</span> else {<a name="line.163"></a> -<span class="sourceLineNo">164</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<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> }<a name="line.167"></a> -<span class="sourceLineNo">168</span> scan.setFilter(new FirstKeyOnlyFilter());<a name="line.168"></a> -<span class="sourceLineNo">169</span> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);<a name="line.169"></a> -<span class="sourceLineNo">170</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.170"></a> -<span class="sourceLineNo">171</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.171"></a> -<span class="sourceLineNo">172</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.172"></a> -<span class="sourceLineNo">173</span> job.setNumReduceTasks(0);<a name="line.173"></a> -<span class="sourceLineNo">174</span> return job;<a name="line.174"></a> -<span class="sourceLineNo">175</span> }<a name="line.175"></a> -<span class="sourceLineNo">176</span><a name="line.176"></a> -<span class="sourceLineNo">177</span> /*<a name="line.177"></a> -<span class="sourceLineNo">178</span> * @param errorMessage Can attach a message when error occurs.<a name="line.178"></a> -<span class="sourceLineNo">179</span> */<a name="line.179"></a> -<span class="sourceLineNo">180</span> private static void printUsage(String errorMessage) {<a name="line.180"></a> -<span class="sourceLineNo">181</span> System.err.println("ERROR: " + errorMessage);<a name="line.181"></a> -<span class="sourceLineNo">182</span> printUsage();<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> /**<a name="line.185"></a> -<span class="sourceLineNo">186</span> * Prints usage without error message.<a name="line.186"></a> -<span class="sourceLineNo">187</span> * Note that we don't document --expected-count, because it's intended for test.<a name="line.187"></a> -<span class="sourceLineNo">188</span> */<a name="line.188"></a> -<span class="sourceLineNo">189</span> private static void printUsage() {<a name="line.189"></a> -<span class="sourceLineNo">190</span> System.err.println("Usage: RowCounter [options] <tablename> " +<a name="line.190"></a> -<span class="sourceLineNo">191</span> "[--starttime=[start] --endtime=[end] " +<a name="line.191"></a> -<span class="sourceLineNo">192</span> "[--range=[startKey],[endKey]] [<column1> <column2>...]");<a name="line.192"></a> -<span class="sourceLineNo">193</span> System.err.println("For performance consider the following options:\n"<a name="line.193"></a> -<span class="sourceLineNo">194</span> + "-Dhbase.client.scanner.caching=100\n"<a name="line.194"></a> -<span class="sourceLineNo">195</span> + "-Dmapreduce.map.speculative=false");<a name="line.195"></a> -<span class="sourceLineNo">196</span> }<a name="line.196"></a> -<span class="sourceLineNo">197</span><a name="line.197"></a> -<span class="sourceLineNo">198</span> @Override<a name="line.198"></a> -<span class="sourceLineNo">199</span> public int run(String[] args) throws Exception {<a name="line.199"></a> -<span class="sourceLineNo">200</span> if (args.length < 1) {<a name="line.200"></a> -<span class="sourceLineNo">201</span> printUsage("Wrong number of parameters: " + args.length);<a name="line.201"></a> -<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a> -<span class="sourceLineNo">203</span> }<a name="line.203"></a> -<span class="sourceLineNo">204</span> Job job = createSubmittableJob(getConf(), args);<a name="line.204"></a> -<span class="sourceLineNo">205</span> if (job == null) {<a name="line.205"></a> -<span class="sourceLineNo">206</span> return -1;<a name="line.206"></a> -<span class="sourceLineNo">207</span> }<a name="line.207"></a> -<span class="sourceLineNo">208</span> boolean success = job.waitForCompletion(true);<a name="line.208"></a> -<span class="sourceLineNo">209</span> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);<a name="line.209"></a> -<span class="sourceLineNo">210</span> if (success && expectedCount != -1) {<a name="line.210"></a> -<span class="sourceLineNo">211</span> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);<a name="line.211"></a> -<span class="sourceLineNo">212</span> success = expectedCount == counter.getValue();<a name="line.212"></a> -<span class="sourceLineNo">213</span> if (!success) {<a name="line.213"></a> -<span class="sourceLineNo">214</span> LOG.error("Failing job because count of '" + counter.getValue() +<a name="line.214"></a> -<span class="sourceLineNo">215</span> "' does not match expected count of '" + expectedCount + "'");<a name="line.215"></a> -<span class="sourceLineNo">216</span> }<a name="line.216"></a> -<span class="sourceLineNo">217</span> }<a name="line.217"></a> -<span class="sourceLineNo">218</span> return (success ? 0 : 1);<a name="line.218"></a> -<span class="sourceLineNo">219</span> }<a name="line.219"></a> -<span class="sourceLineNo">220</span><a name="line.220"></a> -<span class="sourceLineNo">221</span> /**<a name="line.221"></a> -<span class="sourceLineNo">222</span> * Main entry point.<a name="line.222"></a> -<span class="sourceLineNo">223</span> * @param args The command line parameters.<a name="line.223"></a> -<span class="sourceLineNo">224</span> * @throws Exception When running the job fails.<a name="line.224"></a> -<span class="sourceLineNo">225</span> */<a name="line.225"></a> -<span class="sourceLineNo">226</span> public static void main(String[] args) throws Exception {<a name="line.226"></a> -<span class="sourceLineNo">227</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new RowCounter(), args);<a name="line.227"></a> -<span class="sourceLineNo">228</span> System.exit(errCode);<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>}<a name="line.231"></a> +<span class="sourceLineNo">149</span> setScanFilter(scan, rowRangeList);<a name="line.149"></a> +<span class="sourceLineNo">150</span> if (sb.length() > 0) {<a name="line.150"></a> +<span class="sourceLineNo">151</span> for (String columnName : sb.toString().trim().split(" ")) {<a name="line.151"></a> +<span class="sourceLineNo">152</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.152"></a> +<span class="sourceLineNo">153</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.153"></a> +<span class="sourceLineNo">154</span><a name="line.154"></a> +<span class="sourceLineNo">155</span> if (StringUtils.isBlank(qualifier)) {<a name="line.155"></a> +<span class="sourceLineNo">156</span> scan.addFamily(Bytes.toBytes(family));<a name="line.156"></a> +<span class="sourceLineNo">157</span> }<a name="line.157"></a> +<span class="sourceLineNo">158</span> else {<a name="line.158"></a> +<span class="sourceLineNo">159</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<a name="line.159"></a> +<span class="sourceLineNo">160</span> }<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> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);<a name="line.163"></a> +<span class="sourceLineNo">164</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.164"></a> +<span class="sourceLineNo">165</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.165"></a> +<span class="sourceLineNo">166</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.166"></a> +<span class="sourceLineNo">167</span> job.setNumReduceTasks(0);<a name="line.167"></a> +<span class="sourceLineNo">168</span> return job;<a name="line.168"></a> +<span class="sourceLineNo">169</span> }<a name="line.169"></a> +<span class="sourceLineNo">170</span><a name="line.170"></a> +<span class="sourceLineNo">171</span> private static List<MultiRowRangeFilter.RowRange> parseRowRangeParameter(<a name="line.171"></a> +<span class="sourceLineNo">172</span> String arg, String rangeSwitch) {<a name="line.172"></a> +<span class="sourceLineNo">173</span> final String[] ranges = arg.substring(rangeSwitch.length()).split(";");<a name="line.173"></a> +<span class="sourceLineNo">174</span> final List<MultiRowRangeFilter.RowRange> rangeList = new ArrayList<>();<a name="line.174"></a> +<span class="sourceLineNo">175</span> for (String range : ranges) {<a name="line.175"></a> +<span class="sourceLineNo">176</span> String[] startEnd = range.split(",", 2);<a name="line.176"></a> +<span class="sourceLineNo">177</span> if (startEnd.length != 2 || startEnd[1].contains(",")) {<a name="line.177"></a> +<span class="sourceLineNo">178</span> printUsage("Please specify range in such format as \"--range=a,b\" " +<a name="line.178"></a> +<span class="sourceLineNo">179</span> "or, with only one boundary, \"--range=,b\" or \"--range=a,\"");<a name="line.179"></a> +<span class="sourceLineNo">180</span> throw new IllegalArgumentException("Wrong range specification: " + range);<a name="line.180"></a> +<span class="sourceLineNo">181</span> }<a name="line.181"></a> +<span class="sourceLineNo">182</span> String startKey = startEnd[0];<a name="line.182"></a> +<span class="sourceLineNo">183</span> String endKey = startEnd[1];<a name="line.183"></a> +<span class="sourceLineNo">184</span> rangeList.add(new MultiRowRangeFilter.RowRange(<a name="line.184"></a> +<span class="sourceLineNo">185</span> Bytes.toBytesBinary(startKey), true,<a name="line.185"></a> +<span class="sourceLineNo">186</span> Bytes.toBytesBinary(endKey), false));<a name="line.186"></a> +<span class="sourceLineNo">187</span> }<a name="line.187"></a> +<span class="sourceLineNo">188</span> return rangeList;<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> * Sets filter {@link FilterBase} to the {@link Scan} instance.<a name="line.192"></a> +<span class="sourceLineNo">193</span> * If provided rowRangeList contains more than one element,<a name="line.193"></a> +<span class="sourceLineNo">194</span> * method sets filter which is instance of {@link MultiRowRangeFilter}.<a name="line.194"></a> +<span class="sourceLineNo">195</span> * Otherwise, method sets filter which is instance of {@link FirstKeyOnlyFilter}.<a name="line.195"></a> +<span class="sourceLineNo">196</span> * If rowRangeList contains exactly one element, startRow and stopRow are set to the scan.<a name="line.196"></a> +<span class="sourceLineNo">197</span> * @param scan<a name="line.197"></a> +<span class="sourceLineNo">198</span> * @param rowRangeList<a name="line.198"></a> +<span class="sourceLineNo">199</span> */<a name="line.199"></a> +<span class="sourceLineNo">200</span> private static void setScanFilter(Scan scan, List<MultiRowRangeFilter.RowRange> rowRangeList) {<a name="line.200"></a> +<span class="sourceLineNo">201</span> final int size = rowRangeList == null ? 0 : rowRangeList.size();<a name="line.201"></a> +<span class="sourceLineNo">202</span> if (size <= 1) {<a name="line.202"></a> +<span class="sourceLineNo">203</span> scan.setFilter(new FirstKeyOnlyFilter());<a name="line.203"></a> +<span class="sourceLineNo">204</span> }<a name="line.204"></a> +<span class="sourceLineNo">205</span> if (size == 1) {<a name="line.205"></a> +<span class="sourceLineNo">206</span> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);<a name="line.206"></a> +<span class="sourceLineNo">207</span> scan.setStartRow(range.getStartRow()); //inclusive<a name="line.207"></a> +<span class="sourceLineNo">208</span> scan.setStopRow(range.getStopRow()); //exclusive<a name="line.208"></a> +<span class="sourceLineNo">209</span> } else if (size > 1) {<a name="line.209"></a> +<span class="sourceLineNo">210</span> try {<a name="line.210"></a> +<span class="sourceLineNo">211</span> scan.setFilter(new MultiRowRangeFilter(rowRangeList));<a name="line.211"></a> +<span class="sourceLineNo">212</span> } catch (IOException e) {<a name="line.212"></a> +<span class="sourceLineNo">213</span> //the IOException should never be thrown. see HBASE-16145<a name="line.213"></a> +<span class="sourceLineNo">214</span> throw new RuntimeException("Cannot instantiate MultiRowRangeFilter");<a name="line.214"></a> +<span class="sourceLineNo">215</span> }<a name="line.215"></a> +<span class="sourceLineNo">216</span> }<a name="line.216"></a> +<span class="sourceLineNo">217</span> }<a name="line.217"></a> +<span class="sourceLineNo">218</span><a name="line.218"></a> +<span class="sourceLineNo">219</span> /*<a name="line.219"></a> +<span class="sourceLineNo">220</span> * @param errorMessage Can attach a message when error occurs.<a name="line.220"></a> +<span class="sourceLineNo">221</span> */<a name="line.221"></a> +<span class="sourceLineNo">222</span> private static void printUsage(String errorMessage) {<a name="line.222"></a> +<span class="sourceLineNo">223</span> System.err.println("ERROR: " + errorMessage);<a name="line.223"></a> +<span class="sourceLineNo">224</span> printUsage();<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> * Prints usage without error message.<a name="line.228"></a> +<span class="sourceLineNo">229</span> * Note that we don't document --expected-count, because it's intended for test.<a name="line.229"></a> +<span class="sourceLineNo">230</span> */<a name="line.230"></a> +<span class="sourceLineNo">231</span> private static void printUsage() {<a name="line.231"></a> +<span class="sourceLineNo">232</span> System.err.println("Usage: RowCounter [options] <tablename> " +<a name="line.232"></a> +<span class="sourceLineNo">233</span> "[--starttime=[start] --endtime=[end] " +<a name="line.233"></a> +<span class="sourceLineNo">234</span> "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [<column1> <column2>...]");<a name="line.234"></a> +<span class="sourceLineNo">235</span> System.err.println("For performance consider the following options:\n"<a name="line.235"></a> +<span class="sourceLineNo">236</span> + "-Dhbase.client.scanner.caching=100\n"<a name="line.236"></a> +<span class="sourceLineNo">237</span> + "-Dmapreduce.map.speculative=false");<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> @Override<a name="line.240"></a> +<span class="sourceLineNo">241</span> public int run(String[] args) throws Exception {<a name="line.241"></a> +<span class="sourceLineNo">242</span> if (args.length < 1) {<a name="line.242"></a> +<span class="sourceLineNo">243</span> printUsage("Wrong number of parameters: " + args.length);<a name="line.243"></a> +<span class="sourceLineNo">244</span> return -1;<a name="line.244"></a> +<span class="sourceLineNo">245</span> }<a name="line.245"></a> +<span class="sourceLineNo">246</span> Job job = createSubmittableJob(getConf(), args);<a name="line.246"></a> +<span class="sourceLineNo">247</span> if (job == null) {<a name="line.247"></a> +<span class="sourceLineNo">248</span> return -1;<a name="line.248"></a> +<span class="sourceLineNo">249</span> }<a name="line.249"></a> +<span class="sourceLineNo">250</span> boolean success = job.waitForCompletion(true);<a name="line.250"></a> +<span class="sourceLineNo">251</span> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);<a name="line.251"></a> +<span class="sourceLineNo">252</span> if (success && expectedCount != -1) {<a name="line.252"></a> +<span class="sourceLineNo">253</span> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);<a name="line.253"></a> +<span class="sourceLineNo">254</span> success = expectedCount == counter.getValue();<a name="line.254"></a> +<span class="sourceLineNo">255</span> if (!success) {<a name="line.255"></a> +<span class="sourceLineNo">256</span> LOG.error("Failing job because count of '" + counter.getValue() +<a name="line.256"></a> +<span class="sourceLineNo">257</span> "' does not match expected count of '" + expectedCount + "'");<a name="line.257"></a> +<span class="sourceLineNo">258</span> }<a name="line.258"></a> +<span class="sourceLineNo">259</span> }<a name="line.259"></a> +<span class="sourceLineNo">260</span> return (success ? 0 : 1);<a name="line.260"></a> +<span class="sourceLineNo">261</span> }<a name="line.261"></a> +<span class="sourceLineNo">262</span><a name="line.262"></a> +<span class="sourceLineNo">263</span> /**<a name="line.263"></a> +<span class="sourceLineNo">264</span> * Main entry point.<a name="line.264"></a> +<span class="sourceLineNo">265</span> * @param args The command line parameters.<a name="line.265"></a> +<span class="sourceLineNo">266</span> * @throws Exception When running the job fails.<a name="line.266"></a> +<span class="sourceLineNo">267</span> */<a name="line.267"></a> +<span class="sourceLineNo">268</span> public static void main(String[] args) throws Exception {<a name="line.268"></a> +<span class="sourceLineNo">269</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new RowCounter(), args);<a name="line.269"></a> +<span class="sourceLineNo">270</span> System.exit(errCode);<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>
