hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From git-site-r...@apache.org
Subject [23/24] hbase-site git commit: Published site at 910b68082c8f200f0ba6395a76b7ee1c8917e401.
Date Wed, 05 Apr 2017 01:50:37 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/f5caf0f9/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/WALInputFormat.html
----------------------------------------------------------------------
diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/WALInputFormat.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/WALInputFormat.html
index 05e46b2..12db0bb 100644
--- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/WALInputFormat.html
+++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/WALInputFormat.html
@@ -28,283 +28,328 @@
 <span class="sourceLineNo">020</span>import java.io.DataInput;<a name="line.20"></a>
 <span class="sourceLineNo">021</span>import java.io.DataOutput;<a name="line.21"></a>
 <span class="sourceLineNo">022</span>import java.io.EOFException;<a name="line.22"></a>
-<span class="sourceLineNo">023</span>import java.io.IOException;<a name="line.23"></a>
-<span class="sourceLineNo">024</span>import java.util.ArrayList;<a name="line.24"></a>
-<span class="sourceLineNo">025</span>import java.util.Collections;<a name="line.25"></a>
-<span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a>
-<span class="sourceLineNo">027</span><a name="line.27"></a>
-<span class="sourceLineNo">028</span>import org.apache.commons.logging.Log;<a name="line.28"></a>
-<span class="sourceLineNo">029</span>import org.apache.commons.logging.LogFactory;<a name="line.29"></a>
-<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configuration;<a name="line.30"></a>
-<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.FileStatus;<a name="line.31"></a>
-<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.FileSystem;<a name="line.32"></a>
-<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.LocatedFileStatus;<a name="line.33"></a>
-<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.Path;<a name="line.34"></a>
-<span class="sourceLineNo">035</span>import org.apache.hadoop.fs.RemoteIterator;<a name="line.35"></a>
-<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.36"></a>
-<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.regionserver.wal.WALEdit;<a name="line.37"></a>
-<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.wal.WAL;<a name="line.38"></a>
-<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.39"></a>
-<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.wal.WAL.Reader;<a name="line.40"></a>
-<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.wal.WALFactory;<a name="line.41"></a>
-<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.wal.WALKey;<a name="line.42"></a>
-<span class="sourceLineNo">043</span>import org.apache.hadoop.io.Writable;<a name="line.43"></a>
-<span class="sourceLineNo">044</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.44"></a>
-<span class="sourceLineNo">045</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.45"></a>
-<span class="sourceLineNo">046</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.46"></a>
-<span class="sourceLineNo">047</span>import org.apache.hadoop.mapreduce.RecordReader;<a name="line.47"></a>
-<span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.util.StringUtils;<a name="line.49"></a>
-<span class="sourceLineNo">050</span><a name="line.50"></a>
-<span class="sourceLineNo">051</span>/**<a name="line.51"></a>
-<span class="sourceLineNo">052</span> * Simple {@link InputFormat} for {@link org.apache.hadoop.hbase.wal.WAL} files.<a name="line.52"></a>
-<span class="sourceLineNo">053</span> */<a name="line.53"></a>
-<span class="sourceLineNo">054</span>@InterfaceAudience.Public<a name="line.54"></a>
-<span class="sourceLineNo">055</span>public class WALInputFormat extends InputFormat&lt;WALKey, WALEdit&gt; {<a name="line.55"></a>
-<span class="sourceLineNo">056</span>  private static final Log LOG = LogFactory.getLog(WALInputFormat.class);<a name="line.56"></a>
-<span class="sourceLineNo">057</span><a name="line.57"></a>
-<span class="sourceLineNo">058</span>  public static final String START_TIME_KEY = "wal.start.time";<a name="line.58"></a>
-<span class="sourceLineNo">059</span>  public static final String END_TIME_KEY = "wal.end.time";<a name="line.59"></a>
-<span class="sourceLineNo">060</span><a name="line.60"></a>
-<span class="sourceLineNo">061</span>  /**<a name="line.61"></a>
-<span class="sourceLineNo">062</span>   * {@link InputSplit} for {@link WAL} files. Each split represent<a name="line.62"></a>
-<span class="sourceLineNo">063</span>   * exactly one log file.<a name="line.63"></a>
-<span class="sourceLineNo">064</span>   */<a name="line.64"></a>
-<span class="sourceLineNo">065</span>  static class WALSplit extends InputSplit implements Writable {<a name="line.65"></a>
-<span class="sourceLineNo">066</span>    private String logFileName;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>    private long fileSize;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>    private long startTime;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>    private long endTime;<a name="line.69"></a>
-<span class="sourceLineNo">070</span><a name="line.70"></a>
-<span class="sourceLineNo">071</span>    /** for serialization */<a name="line.71"></a>
-<span class="sourceLineNo">072</span>    public WALSplit() {}<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>     * Represent an WALSplit, i.e. a single WAL file.<a name="line.75"></a>
-<span class="sourceLineNo">076</span>     * Start- and EndTime are managed by the split, so that WAL files can be<a name="line.76"></a>
-<span class="sourceLineNo">077</span>     * filtered before WALEdits are passed to the mapper(s).<a name="line.77"></a>
-<span class="sourceLineNo">078</span>     * @param logFileName<a name="line.78"></a>
-<span class="sourceLineNo">079</span>     * @param fileSize<a name="line.79"></a>
-<span class="sourceLineNo">080</span>     * @param startTime<a name="line.80"></a>
-<span class="sourceLineNo">081</span>     * @param endTime<a name="line.81"></a>
-<span class="sourceLineNo">082</span>     */<a name="line.82"></a>
-<span class="sourceLineNo">083</span>    public WALSplit(String logFileName, long fileSize, long startTime, long endTime) {<a name="line.83"></a>
-<span class="sourceLineNo">084</span>      this.logFileName = logFileName;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>      this.fileSize = fileSize;<a name="line.85"></a>
-<span class="sourceLineNo">086</span>      this.startTime = startTime;<a name="line.86"></a>
-<span class="sourceLineNo">087</span>      this.endTime = endTime;<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>    @Override<a name="line.90"></a>
-<span class="sourceLineNo">091</span>    public long getLength() throws IOException, InterruptedException {<a name="line.91"></a>
-<span class="sourceLineNo">092</span>      return fileSize;<a name="line.92"></a>
-<span class="sourceLineNo">093</span>    }<a name="line.93"></a>
-<span class="sourceLineNo">094</span><a name="line.94"></a>
-<span class="sourceLineNo">095</span>    @Override<a name="line.95"></a>
-<span class="sourceLineNo">096</span>    public String[] getLocations() throws IOException, InterruptedException {<a name="line.96"></a>
-<span class="sourceLineNo">097</span>      // TODO: Find the data node with the most blocks for this WAL?<a name="line.97"></a>
-<span class="sourceLineNo">098</span>      return new String[] {};<a name="line.98"></a>
-<span class="sourceLineNo">099</span>    }<a name="line.99"></a>
-<span class="sourceLineNo">100</span><a name="line.100"></a>
-<span class="sourceLineNo">101</span>    public String getLogFileName() {<a name="line.101"></a>
-<span class="sourceLineNo">102</span>      return logFileName;<a name="line.102"></a>
-<span class="sourceLineNo">103</span>    }<a name="line.103"></a>
-<span class="sourceLineNo">104</span><a name="line.104"></a>
-<span class="sourceLineNo">105</span>    public long getStartTime() {<a name="line.105"></a>
-<span class="sourceLineNo">106</span>      return startTime;<a name="line.106"></a>
-<span class="sourceLineNo">107</span>    }<a name="line.107"></a>
-<span class="sourceLineNo">108</span><a name="line.108"></a>
-<span class="sourceLineNo">109</span>    public long getEndTime() {<a name="line.109"></a>
-<span class="sourceLineNo">110</span>      return endTime;<a name="line.110"></a>
-<span class="sourceLineNo">111</span>    }<a name="line.111"></a>
-<span class="sourceLineNo">112</span><a name="line.112"></a>
-<span class="sourceLineNo">113</span>    @Override<a name="line.113"></a>
-<span class="sourceLineNo">114</span>    public void readFields(DataInput in) throws IOException {<a name="line.114"></a>
-<span class="sourceLineNo">115</span>      logFileName = in.readUTF();<a name="line.115"></a>
-<span class="sourceLineNo">116</span>      fileSize = in.readLong();<a name="line.116"></a>
-<span class="sourceLineNo">117</span>      startTime = in.readLong();<a name="line.117"></a>
-<span class="sourceLineNo">118</span>      endTime = in.readLong();<a name="line.118"></a>
-<span class="sourceLineNo">119</span>    }<a name="line.119"></a>
-<span class="sourceLineNo">120</span><a name="line.120"></a>
-<span class="sourceLineNo">121</span>    @Override<a name="line.121"></a>
-<span class="sourceLineNo">122</span>    public void write(DataOutput out) throws IOException {<a name="line.122"></a>
-<span class="sourceLineNo">123</span>      out.writeUTF(logFileName);<a name="line.123"></a>
-<span class="sourceLineNo">124</span>      out.writeLong(fileSize);<a name="line.124"></a>
-<span class="sourceLineNo">125</span>      out.writeLong(startTime);<a name="line.125"></a>
-<span class="sourceLineNo">126</span>      out.writeLong(endTime);<a name="line.126"></a>
-<span class="sourceLineNo">127</span>    }<a name="line.127"></a>
-<span class="sourceLineNo">128</span><a name="line.128"></a>
-<span class="sourceLineNo">129</span>    @Override<a name="line.129"></a>
-<span class="sourceLineNo">130</span>    public String toString() {<a name="line.130"></a>
-<span class="sourceLineNo">131</span>      return logFileName + " (" + startTime + ":" + endTime + ") length:" + fileSize;<a name="line.131"></a>
-<span class="sourceLineNo">132</span>    }<a name="line.132"></a>
-<span class="sourceLineNo">133</span>  }<a name="line.133"></a>
-<span class="sourceLineNo">134</span><a name="line.134"></a>
-<span class="sourceLineNo">135</span>  /**<a name="line.135"></a>
-<span class="sourceLineNo">136</span>   * {@link RecordReader} for an {@link WAL} file.<a name="line.136"></a>
-<span class="sourceLineNo">137</span>   * Implementation shared with deprecated HLogInputFormat.<a name="line.137"></a>
-<span class="sourceLineNo">138</span>   */<a name="line.138"></a>
-<span class="sourceLineNo">139</span>  static abstract class WALRecordReader&lt;K extends WALKey&gt; extends RecordReader&lt;K, WALEdit&gt; {<a name="line.139"></a>
-<span class="sourceLineNo">140</span>    private Reader reader = null;<a name="line.140"></a>
-<span class="sourceLineNo">141</span>    // visible until we can remove the deprecated HLogInputFormat<a name="line.141"></a>
-<span class="sourceLineNo">142</span>    Entry currentEntry = new Entry();<a name="line.142"></a>
-<span class="sourceLineNo">143</span>    private long startTime;<a name="line.143"></a>
-<span class="sourceLineNo">144</span>    private long endTime;<a name="line.144"></a>
-<span class="sourceLineNo">145</span><a name="line.145"></a>
-<span class="sourceLineNo">146</span>    @Override<a name="line.146"></a>
-<span class="sourceLineNo">147</span>    public void initialize(InputSplit split, TaskAttemptContext context)<a name="line.147"></a>
-<span class="sourceLineNo">148</span>        throws IOException, InterruptedException {<a name="line.148"></a>
-<span class="sourceLineNo">149</span>      WALSplit hsplit = (WALSplit)split;<a name="line.149"></a>
-<span class="sourceLineNo">150</span>      Path logFile = new Path(hsplit.getLogFileName());<a name="line.150"></a>
-<span class="sourceLineNo">151</span>      Configuration conf = context.getConfiguration();<a name="line.151"></a>
-<span class="sourceLineNo">152</span>      LOG.info("Opening reader for "+split);<a name="line.152"></a>
-<span class="sourceLineNo">153</span>      try {<a name="line.153"></a>
-<span class="sourceLineNo">154</span>        this.reader = WALFactory.createReader(logFile.getFileSystem(conf), logFile, conf);<a name="line.154"></a>
-<span class="sourceLineNo">155</span>      } catch (EOFException x) {<a name="line.155"></a>
-<span class="sourceLineNo">156</span>        LOG.info("Ignoring corrupted WAL file: " + logFile<a name="line.156"></a>
-<span class="sourceLineNo">157</span>            + " (This is normal when a RegionServer crashed.)");<a name="line.157"></a>
-<span class="sourceLineNo">158</span>        this.reader = null;<a name="line.158"></a>
-<span class="sourceLineNo">159</span>      }<a name="line.159"></a>
-<span class="sourceLineNo">160</span>      this.startTime = hsplit.getStartTime();<a name="line.160"></a>
-<span class="sourceLineNo">161</span>      this.endTime = hsplit.getEndTime();<a name="line.161"></a>
-<span class="sourceLineNo">162</span>    }<a name="line.162"></a>
-<span class="sourceLineNo">163</span><a name="line.163"></a>
-<span class="sourceLineNo">164</span>    @Override<a name="line.164"></a>
-<span class="sourceLineNo">165</span>    public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.165"></a>
-<span class="sourceLineNo">166</span>      if (reader == null) return false;<a name="line.166"></a>
-<span class="sourceLineNo">167</span><a name="line.167"></a>
-<span class="sourceLineNo">168</span>      Entry temp;<a name="line.168"></a>
-<span class="sourceLineNo">169</span>      long i = -1;<a name="line.169"></a>
-<span class="sourceLineNo">170</span>      do {<a name="line.170"></a>
-<span class="sourceLineNo">171</span>        // skip older entries<a name="line.171"></a>
-<span class="sourceLineNo">172</span>        try {<a name="line.172"></a>
-<span class="sourceLineNo">173</span>          temp = reader.next(currentEntry);<a name="line.173"></a>
-<span class="sourceLineNo">174</span>          i++;<a name="line.174"></a>
-<span class="sourceLineNo">175</span>        } catch (EOFException x) {<a name="line.175"></a>
-<span class="sourceLineNo">176</span>          LOG.warn("Corrupted entry detected. Ignoring the rest of the file."<a name="line.176"></a>
-<span class="sourceLineNo">177</span>              + " (This is normal when a RegionServer crashed.)");<a name="line.177"></a>
-<span class="sourceLineNo">178</span>          return false;<a name="line.178"></a>
-<span class="sourceLineNo">179</span>        }<a name="line.179"></a>
-<span class="sourceLineNo">180</span>      }<a name="line.180"></a>
-<span class="sourceLineNo">181</span>      while(temp != null &amp;&amp; temp.getKey().getWriteTime() &lt; startTime);<a name="line.181"></a>
-<span class="sourceLineNo">182</span><a name="line.182"></a>
-<span class="sourceLineNo">183</span>      if (temp == null) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span>        if (i &gt; 0) LOG.info("Skipped " + i + " entries.");<a name="line.184"></a>
-<span class="sourceLineNo">185</span>        LOG.info("Reached end of file.");<a name="line.185"></a>
-<span class="sourceLineNo">186</span>        return false;<a name="line.186"></a>
-<span class="sourceLineNo">187</span>      } else if (i &gt; 0) {<a name="line.187"></a>
-<span class="sourceLineNo">188</span>        LOG.info("Skipped " + i + " entries, until ts: " + temp.getKey().getWriteTime() + ".");<a name="line.188"></a>
-<span class="sourceLineNo">189</span>      }<a name="line.189"></a>
-<span class="sourceLineNo">190</span>      boolean res = temp.getKey().getWriteTime() &lt;= endTime;<a name="line.190"></a>
-<span class="sourceLineNo">191</span>      if (!res) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span>        LOG.info("Reached ts: " + temp.getKey().getWriteTime() + " ignoring the rest of the file.");<a name="line.192"></a>
-<span class="sourceLineNo">193</span>      }<a name="line.193"></a>
-<span class="sourceLineNo">194</span>      return res;<a name="line.194"></a>
-<span class="sourceLineNo">195</span>    }<a name="line.195"></a>
-<span class="sourceLineNo">196</span><a name="line.196"></a>
-<span class="sourceLineNo">197</span>    @Override<a name="line.197"></a>
-<span class="sourceLineNo">198</span>    public WALEdit getCurrentValue() throws IOException, InterruptedException {<a name="line.198"></a>
-<span class="sourceLineNo">199</span>      return currentEntry.getEdit();<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>    @Override<a name="line.202"></a>
-<span class="sourceLineNo">203</span>    public float getProgress() throws IOException, InterruptedException {<a name="line.203"></a>
-<span class="sourceLineNo">204</span>      // N/A depends on total number of entries, which is unknown<a name="line.204"></a>
-<span class="sourceLineNo">205</span>      return 0;<a name="line.205"></a>
-<span class="sourceLineNo">206</span>    }<a name="line.206"></a>
-<span class="sourceLineNo">207</span><a name="line.207"></a>
-<span class="sourceLineNo">208</span>    @Override<a name="line.208"></a>
-<span class="sourceLineNo">209</span>    public void close() throws IOException {<a name="line.209"></a>
-<span class="sourceLineNo">210</span>      LOG.info("Closing reader");<a name="line.210"></a>
-<span class="sourceLineNo">211</span>      if (reader != null) this.reader.close();<a name="line.211"></a>
-<span class="sourceLineNo">212</span>    }<a name="line.212"></a>
-<span class="sourceLineNo">213</span>  }<a name="line.213"></a>
-<span class="sourceLineNo">214</span><a name="line.214"></a>
-<span class="sourceLineNo">215</span>  /**<a name="line.215"></a>
-<span class="sourceLineNo">216</span>   * handler for non-deprecated WALKey version. fold into WALRecordReader once we no longer<a name="line.216"></a>
-<span class="sourceLineNo">217</span>   * need to support HLogInputFormat.<a name="line.217"></a>
-<span class="sourceLineNo">218</span>   */<a name="line.218"></a>
-<span class="sourceLineNo">219</span>  static class WALKeyRecordReader extends WALRecordReader&lt;WALKey&gt; {<a name="line.219"></a>
-<span class="sourceLineNo">220</span>    @Override<a name="line.220"></a>
-<span class="sourceLineNo">221</span>    public WALKey getCurrentKey() throws IOException, InterruptedException {<a name="line.221"></a>
-<span class="sourceLineNo">222</span>      return currentEntry.getKey();<a name="line.222"></a>
-<span class="sourceLineNo">223</span>    }<a name="line.223"></a>
-<span class="sourceLineNo">224</span>  }<a name="line.224"></a>
-<span class="sourceLineNo">225</span><a name="line.225"></a>
-<span class="sourceLineNo">226</span>  @Override<a name="line.226"></a>
-<span class="sourceLineNo">227</span>  public List&lt;InputSplit&gt; getSplits(JobContext context) throws IOException,<a name="line.227"></a>
-<span class="sourceLineNo">228</span>      InterruptedException {<a name="line.228"></a>
-<span class="sourceLineNo">229</span>    return getSplits(context, START_TIME_KEY, END_TIME_KEY);<a name="line.229"></a>
-<span class="sourceLineNo">230</span>  }<a name="line.230"></a>
+<span class="sourceLineNo">023</span>import java.io.FileNotFoundException;<a name="line.23"></a>
+<span class="sourceLineNo">024</span>import java.io.IOException;<a name="line.24"></a>
+<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a>
+<span class="sourceLineNo">026</span>import java.util.Collections;<a name="line.26"></a>
+<span class="sourceLineNo">027</span>import java.util.List;<a name="line.27"></a>
+<span class="sourceLineNo">028</span><a name="line.28"></a>
+<span class="sourceLineNo">029</span>import org.apache.commons.logging.Log;<a name="line.29"></a>
+<span class="sourceLineNo">030</span>import org.apache.commons.logging.LogFactory;<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.fs.FileStatus;<a name="line.32"></a>
+<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.FileSystem;<a name="line.33"></a>
+<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.LocatedFileStatus;<a name="line.34"></a>
+<span class="sourceLineNo">035</span>import org.apache.hadoop.fs.Path;<a name="line.35"></a>
+<span class="sourceLineNo">036</span>import org.apache.hadoop.fs.RemoteIterator;<a name="line.36"></a>
+<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.classification.InterfaceAudience;<a name="line.37"></a>
+<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.regionserver.wal.WALEdit;<a name="line.38"></a>
+<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.wal.AbstractFSWALProvider;<a name="line.39"></a>
+<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.wal.WAL;<a name="line.40"></a>
+<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.41"></a>
+<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.wal.WAL.Reader;<a name="line.42"></a>
+<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.wal.WALKey;<a name="line.43"></a>
+<span class="sourceLineNo">044</span>import org.apache.hadoop.io.Writable;<a name="line.44"></a>
+<span class="sourceLineNo">045</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.45"></a>
+<span class="sourceLineNo">046</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.46"></a>
+<span class="sourceLineNo">047</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.47"></a>
+<span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.RecordReader;<a name="line.48"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.util.StringUtils;<a name="line.51"></a>
+<span class="sourceLineNo">052</span><a name="line.52"></a>
+<span class="sourceLineNo">053</span>/**<a name="line.53"></a>
+<span class="sourceLineNo">054</span> * Simple {@link InputFormat} for {@link org.apache.hadoop.hbase.wal.WAL} files.<a name="line.54"></a>
+<span class="sourceLineNo">055</span> */<a name="line.55"></a>
+<span class="sourceLineNo">056</span>@InterfaceAudience.Public<a name="line.56"></a>
+<span class="sourceLineNo">057</span>public class WALInputFormat extends InputFormat&lt;WALKey, WALEdit&gt; {<a name="line.57"></a>
+<span class="sourceLineNo">058</span>  private static final Log LOG = LogFactory.getLog(WALInputFormat.class);<a name="line.58"></a>
+<span class="sourceLineNo">059</span><a name="line.59"></a>
+<span class="sourceLineNo">060</span>  public static final String START_TIME_KEY = "wal.start.time";<a name="line.60"></a>
+<span class="sourceLineNo">061</span>  public static final String END_TIME_KEY = "wal.end.time";<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>   * {@link InputSplit} for {@link WAL} files. Each split represent<a name="line.64"></a>
+<span class="sourceLineNo">065</span>   * exactly one log file.<a name="line.65"></a>
+<span class="sourceLineNo">066</span>   */<a name="line.66"></a>
+<span class="sourceLineNo">067</span>  static class WALSplit extends InputSplit implements Writable {<a name="line.67"></a>
+<span class="sourceLineNo">068</span>    private String logFileName;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>    private long fileSize;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>    private long startTime;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>    private long endTime;<a name="line.71"></a>
+<span class="sourceLineNo">072</span><a name="line.72"></a>
+<span class="sourceLineNo">073</span>    /** for serialization */<a name="line.73"></a>
+<span class="sourceLineNo">074</span>    public WALSplit() {}<a name="line.74"></a>
+<span class="sourceLineNo">075</span><a name="line.75"></a>
+<span class="sourceLineNo">076</span>    /**<a name="line.76"></a>
+<span class="sourceLineNo">077</span>     * Represent an WALSplit, i.e. a single WAL file.<a name="line.77"></a>
+<span class="sourceLineNo">078</span>     * Start- and EndTime are managed by the split, so that WAL files can be<a name="line.78"></a>
+<span class="sourceLineNo">079</span>     * filtered before WALEdits are passed to the mapper(s).<a name="line.79"></a>
+<span class="sourceLineNo">080</span>     * @param logFileName<a name="line.80"></a>
+<span class="sourceLineNo">081</span>     * @param fileSize<a name="line.81"></a>
+<span class="sourceLineNo">082</span>     * @param startTime<a name="line.82"></a>
+<span class="sourceLineNo">083</span>     * @param endTime<a name="line.83"></a>
+<span class="sourceLineNo">084</span>     */<a name="line.84"></a>
+<span class="sourceLineNo">085</span>    public WALSplit(String logFileName, long fileSize, long startTime, long endTime) {<a name="line.85"></a>
+<span class="sourceLineNo">086</span>      this.logFileName = logFileName;<a name="line.86"></a>
+<span class="sourceLineNo">087</span>      this.fileSize = fileSize;<a name="line.87"></a>
+<span class="sourceLineNo">088</span>      this.startTime = startTime;<a name="line.88"></a>
+<span class="sourceLineNo">089</span>      this.endTime = endTime;<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>    @Override<a name="line.92"></a>
+<span class="sourceLineNo">093</span>    public long getLength() throws IOException, InterruptedException {<a name="line.93"></a>
+<span class="sourceLineNo">094</span>      return fileSize;<a name="line.94"></a>
+<span class="sourceLineNo">095</span>    }<a name="line.95"></a>
+<span class="sourceLineNo">096</span><a name="line.96"></a>
+<span class="sourceLineNo">097</span>    @Override<a name="line.97"></a>
+<span class="sourceLineNo">098</span>    public String[] getLocations() throws IOException, InterruptedException {<a name="line.98"></a>
+<span class="sourceLineNo">099</span>      // TODO: Find the data node with the most blocks for this WAL?<a name="line.99"></a>
+<span class="sourceLineNo">100</span>      return new String[] {};<a name="line.100"></a>
+<span class="sourceLineNo">101</span>    }<a name="line.101"></a>
+<span class="sourceLineNo">102</span><a name="line.102"></a>
+<span class="sourceLineNo">103</span>    public String getLogFileName() {<a name="line.103"></a>
+<span class="sourceLineNo">104</span>      return logFileName;<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>    public long getStartTime() {<a name="line.107"></a>
+<span class="sourceLineNo">108</span>      return startTime;<a name="line.108"></a>
+<span class="sourceLineNo">109</span>    }<a name="line.109"></a>
+<span class="sourceLineNo">110</span><a name="line.110"></a>
+<span class="sourceLineNo">111</span>    public long getEndTime() {<a name="line.111"></a>
+<span class="sourceLineNo">112</span>      return endTime;<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>    @Override<a name="line.115"></a>
+<span class="sourceLineNo">116</span>    public void readFields(DataInput in) throws IOException {<a name="line.116"></a>
+<span class="sourceLineNo">117</span>      logFileName = in.readUTF();<a name="line.117"></a>
+<span class="sourceLineNo">118</span>      fileSize = in.readLong();<a name="line.118"></a>
+<span class="sourceLineNo">119</span>      startTime = in.readLong();<a name="line.119"></a>
+<span class="sourceLineNo">120</span>      endTime = in.readLong();<a name="line.120"></a>
+<span class="sourceLineNo">121</span>    }<a name="line.121"></a>
+<span class="sourceLineNo">122</span><a name="line.122"></a>
+<span class="sourceLineNo">123</span>    @Override<a name="line.123"></a>
+<span class="sourceLineNo">124</span>    public void write(DataOutput out) throws IOException {<a name="line.124"></a>
+<span class="sourceLineNo">125</span>      out.writeUTF(logFileName);<a name="line.125"></a>
+<span class="sourceLineNo">126</span>      out.writeLong(fileSize);<a name="line.126"></a>
+<span class="sourceLineNo">127</span>      out.writeLong(startTime);<a name="line.127"></a>
+<span class="sourceLineNo">128</span>      out.writeLong(endTime);<a name="line.128"></a>
+<span class="sourceLineNo">129</span>    }<a name="line.129"></a>
+<span class="sourceLineNo">130</span><a name="line.130"></a>
+<span class="sourceLineNo">131</span>    @Override<a name="line.131"></a>
+<span class="sourceLineNo">132</span>    public String toString() {<a name="line.132"></a>
+<span class="sourceLineNo">133</span>      return logFileName + " (" + startTime + ":" + endTime + ") length:" + fileSize;<a name="line.133"></a>
+<span class="sourceLineNo">134</span>    }<a name="line.134"></a>
+<span class="sourceLineNo">135</span>  }<a name="line.135"></a>
+<span class="sourceLineNo">136</span><a name="line.136"></a>
+<span class="sourceLineNo">137</span>  /**<a name="line.137"></a>
+<span class="sourceLineNo">138</span>   * {@link RecordReader} for an {@link WAL} file.<a name="line.138"></a>
+<span class="sourceLineNo">139</span>   * Implementation shared with deprecated HLogInputFormat.<a name="line.139"></a>
+<span class="sourceLineNo">140</span>   */<a name="line.140"></a>
+<span class="sourceLineNo">141</span>  static abstract class WALRecordReader&lt;K extends WALKey&gt; extends RecordReader&lt;K, WALEdit&gt; {<a name="line.141"></a>
+<span class="sourceLineNo">142</span>    private Reader reader = null;<a name="line.142"></a>
+<span class="sourceLineNo">143</span>    // visible until we can remove the deprecated HLogInputFormat<a name="line.143"></a>
+<span class="sourceLineNo">144</span>    Entry currentEntry = new Entry();<a name="line.144"></a>
+<span class="sourceLineNo">145</span>    private long startTime;<a name="line.145"></a>
+<span class="sourceLineNo">146</span>    private long endTime;<a name="line.146"></a>
+<span class="sourceLineNo">147</span>    private Configuration conf;<a name="line.147"></a>
+<span class="sourceLineNo">148</span>    private Path logFile;<a name="line.148"></a>
+<span class="sourceLineNo">149</span>    private long currentPos;<a name="line.149"></a>
+<span class="sourceLineNo">150</span><a name="line.150"></a>
+<span class="sourceLineNo">151</span>    @Override<a name="line.151"></a>
+<span class="sourceLineNo">152</span>    public void initialize(InputSplit split, TaskAttemptContext context)<a name="line.152"></a>
+<span class="sourceLineNo">153</span>        throws IOException, InterruptedException {<a name="line.153"></a>
+<span class="sourceLineNo">154</span>      WALSplit hsplit = (WALSplit)split;<a name="line.154"></a>
+<span class="sourceLineNo">155</span>      logFile = new Path(hsplit.getLogFileName());<a name="line.155"></a>
+<span class="sourceLineNo">156</span>      conf = context.getConfiguration();<a name="line.156"></a>
+<span class="sourceLineNo">157</span>      LOG.info("Opening reader for "+split);<a name="line.157"></a>
+<span class="sourceLineNo">158</span>      openReader(logFile);<a name="line.158"></a>
+<span class="sourceLineNo">159</span>      this.startTime = hsplit.getStartTime();<a name="line.159"></a>
+<span class="sourceLineNo">160</span>      this.endTime = hsplit.getEndTime();<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>    private void openReader(Path path) throws IOException<a name="line.163"></a>
+<span class="sourceLineNo">164</span>    {<a name="line.164"></a>
+<span class="sourceLineNo">165</span>      closeReader();<a name="line.165"></a>
+<span class="sourceLineNo">166</span>      reader = AbstractFSWALProvider.openReader(path, conf);<a name="line.166"></a>
+<span class="sourceLineNo">167</span>      seek();<a name="line.167"></a>
+<span class="sourceLineNo">168</span>      setCurrentPath(path);<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 void setCurrentPath(Path path) {<a name="line.171"></a>
+<span class="sourceLineNo">172</span>      this.logFile = path;<a name="line.172"></a>
+<span class="sourceLineNo">173</span>    }<a name="line.173"></a>
+<span class="sourceLineNo">174</span><a name="line.174"></a>
+<span class="sourceLineNo">175</span>    private void closeReader() throws IOException {<a name="line.175"></a>
+<span class="sourceLineNo">176</span>      if (reader != null) {<a name="line.176"></a>
+<span class="sourceLineNo">177</span>        reader.close();<a name="line.177"></a>
+<span class="sourceLineNo">178</span>        reader = null;<a name="line.178"></a>
+<span class="sourceLineNo">179</span>      }<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>    private void seek() throws IOException {<a name="line.182"></a>
+<span class="sourceLineNo">183</span>      if (currentPos != 0) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span>        reader.seek(currentPos);<a name="line.184"></a>
+<span class="sourceLineNo">185</span>      }<a name="line.185"></a>
+<span class="sourceLineNo">186</span>    }<a name="line.186"></a>
+<span class="sourceLineNo">187</span><a name="line.187"></a>
+<span class="sourceLineNo">188</span>    @Override<a name="line.188"></a>
+<span class="sourceLineNo">189</span>    public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.189"></a>
+<span class="sourceLineNo">190</span>      if (reader == null) return false;<a name="line.190"></a>
+<span class="sourceLineNo">191</span>      this.currentPos = reader.getPosition();<a name="line.191"></a>
+<span class="sourceLineNo">192</span>      Entry temp;<a name="line.192"></a>
+<span class="sourceLineNo">193</span>      long i = -1;<a name="line.193"></a>
+<span class="sourceLineNo">194</span>      try {<a name="line.194"></a>
+<span class="sourceLineNo">195</span>        do {<a name="line.195"></a>
+<span class="sourceLineNo">196</span>          // skip older entries<a name="line.196"></a>
+<span class="sourceLineNo">197</span>          try {<a name="line.197"></a>
+<span class="sourceLineNo">198</span>            temp = reader.next(currentEntry);<a name="line.198"></a>
+<span class="sourceLineNo">199</span>            i++;<a name="line.199"></a>
+<span class="sourceLineNo">200</span>          } catch (EOFException x) {<a name="line.200"></a>
+<span class="sourceLineNo">201</span>            LOG.warn("Corrupted entry detected. Ignoring the rest of the file."<a name="line.201"></a>
+<span class="sourceLineNo">202</span>                + " (This is normal when a RegionServer crashed.)");<a name="line.202"></a>
+<span class="sourceLineNo">203</span>            return false;<a name="line.203"></a>
+<span class="sourceLineNo">204</span>          }<a name="line.204"></a>
+<span class="sourceLineNo">205</span>        } while (temp != null &amp;&amp; temp.getKey().getWriteTime() &lt; startTime);<a name="line.205"></a>
+<span class="sourceLineNo">206</span><a name="line.206"></a>
+<span class="sourceLineNo">207</span>        if (temp == null) {<a name="line.207"></a>
+<span class="sourceLineNo">208</span>          if (i &gt; 0) LOG.info("Skipped " + i + " entries.");<a name="line.208"></a>
+<span class="sourceLineNo">209</span>          LOG.info("Reached end of file.");<a name="line.209"></a>
+<span class="sourceLineNo">210</span>          return false;<a name="line.210"></a>
+<span class="sourceLineNo">211</span>        } else if (i &gt; 0) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span>          LOG.info("Skipped " + i + " entries, until ts: " + temp.getKey().getWriteTime() + ".");<a name="line.212"></a>
+<span class="sourceLineNo">213</span>        }<a name="line.213"></a>
+<span class="sourceLineNo">214</span>        boolean res = temp.getKey().getWriteTime() &lt;= endTime;<a name="line.214"></a>
+<span class="sourceLineNo">215</span>        if (!res) {<a name="line.215"></a>
+<span class="sourceLineNo">216</span>          LOG.info("Reached ts: " + temp.getKey().getWriteTime()<a name="line.216"></a>
+<span class="sourceLineNo">217</span>              + " ignoring the rest of the file.");<a name="line.217"></a>
+<span class="sourceLineNo">218</span>        }<a name="line.218"></a>
+<span class="sourceLineNo">219</span>        return res;<a name="line.219"></a>
+<span class="sourceLineNo">220</span>      } catch (IOException e) {<a name="line.220"></a>
+<span class="sourceLineNo">221</span>        Path archivedLog = AbstractFSWALProvider.getArchivedLogPath(logFile, conf);<a name="line.221"></a>
+<span class="sourceLineNo">222</span>        if (logFile != archivedLog) {<a name="line.222"></a>
+<span class="sourceLineNo">223</span>          openReader(archivedLog);<a name="line.223"></a>
+<span class="sourceLineNo">224</span>          // Try call again in recursion<a name="line.224"></a>
+<span class="sourceLineNo">225</span>          return nextKeyValue();<a name="line.225"></a>
+<span class="sourceLineNo">226</span>        } else {<a name="line.226"></a>
+<span class="sourceLineNo">227</span>          throw e;<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>    }<a name="line.230"></a>
 <span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span>  /**<a name="line.232"></a>
-<span class="sourceLineNo">233</span>   * implementation shared with deprecated HLogInputFormat<a name="line.233"></a>
-<span class="sourceLineNo">234</span>   */<a name="line.234"></a>
-<span class="sourceLineNo">235</span>  List&lt;InputSplit&gt; getSplits(final JobContext context, final String startKey, final String endKey)<a name="line.235"></a>
-<span class="sourceLineNo">236</span>      throws IOException, InterruptedException {<a name="line.236"></a>
-<span class="sourceLineNo">237</span>    Configuration conf = context.getConfiguration();<a name="line.237"></a>
-<span class="sourceLineNo">238</span>    Path[] inputPaths = getInputPaths(conf);<a name="line.238"></a>
-<span class="sourceLineNo">239</span>    long startTime = conf.getLong(startKey, Long.MIN_VALUE);<a name="line.239"></a>
-<span class="sourceLineNo">240</span>    long endTime = conf.getLong(endKey, Long.MAX_VALUE);<a name="line.240"></a>
-<span class="sourceLineNo">241</span><a name="line.241"></a>
-<span class="sourceLineNo">242</span>    List&lt;FileStatus&gt; allFiles = new ArrayList&lt;FileStatus&gt;();<a name="line.242"></a>
-<span class="sourceLineNo">243</span>    for(Path inputPath: inputPaths){<a name="line.243"></a>
-<span class="sourceLineNo">244</span>      FileSystem fs = inputPath.getFileSystem(conf);<a name="line.244"></a>
-<span class="sourceLineNo">245</span>      List&lt;FileStatus&gt; files = getFiles(fs, inputPath, startTime, endTime);<a name="line.245"></a>
-<span class="sourceLineNo">246</span>      allFiles.addAll(files);<a name="line.246"></a>
+<span class="sourceLineNo">232</span>    @Override<a name="line.232"></a>
+<span class="sourceLineNo">233</span>    public WALEdit getCurrentValue() throws IOException, InterruptedException {<a name="line.233"></a>
+<span class="sourceLineNo">234</span>      return currentEntry.getEdit();<a name="line.234"></a>
+<span class="sourceLineNo">235</span>    }<a name="line.235"></a>
+<span class="sourceLineNo">236</span><a name="line.236"></a>
+<span class="sourceLineNo">237</span>    @Override<a name="line.237"></a>
+<span class="sourceLineNo">238</span>    public float getProgress() throws IOException, InterruptedException {<a name="line.238"></a>
+<span class="sourceLineNo">239</span>      // N/A depends on total number of entries, which is unknown<a name="line.239"></a>
+<span class="sourceLineNo">240</span>      return 0;<a name="line.240"></a>
+<span class="sourceLineNo">241</span>    }<a name="line.241"></a>
+<span class="sourceLineNo">242</span><a name="line.242"></a>
+<span class="sourceLineNo">243</span>    @Override<a name="line.243"></a>
+<span class="sourceLineNo">244</span>    public void close() throws IOException {<a name="line.244"></a>
+<span class="sourceLineNo">245</span>      LOG.info("Closing reader");<a name="line.245"></a>
+<span class="sourceLineNo">246</span>      if (reader != null) this.reader.close();<a name="line.246"></a>
 <span class="sourceLineNo">247</span>    }<a name="line.247"></a>
-<span class="sourceLineNo">248</span>    List&lt;InputSplit&gt; splits = new ArrayList&lt;InputSplit&gt;(allFiles.size());<a name="line.248"></a>
-<span class="sourceLineNo">249</span>    for (FileStatus file : allFiles) {<a name="line.249"></a>
-<span class="sourceLineNo">250</span>      splits.add(new WALSplit(file.getPath().toString(), file.getLen(), startTime, endTime));<a name="line.250"></a>
-<span class="sourceLineNo">251</span>    }<a name="line.251"></a>
-<span class="sourceLineNo">252</span>    return splits;<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">255</span>  private Path[] getInputPaths(Configuration conf) {<a name="line.255"></a>
-<span class="sourceLineNo">256</span>    String inpDirs = conf.get("mapreduce.input.fileinputformat.inputdir");<a name="line.256"></a>
-<span class="sourceLineNo">257</span>    return StringUtils.stringToPath(inpDirs.split(","));<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>  private List&lt;FileStatus&gt; getFiles(FileSystem fs, Path dir, long startTime, long endTime)<a name="line.260"></a>
-<span class="sourceLineNo">261</span>      throws IOException {<a name="line.261"></a>
-<span class="sourceLineNo">262</span>    List&lt;FileStatus&gt; result = new ArrayList&lt;&gt;();<a name="line.262"></a>
-<span class="sourceLineNo">263</span>    LOG.debug("Scanning " + dir.toString() + " for WAL files");<a name="line.263"></a>
-<span class="sourceLineNo">264</span><a name="line.264"></a>
-<span class="sourceLineNo">265</span>    RemoteIterator&lt;LocatedFileStatus&gt; iter = fs.listLocatedStatus(dir);<a name="line.265"></a>
-<span class="sourceLineNo">266</span>    if (!iter.hasNext()) return Collections.emptyList();<a name="line.266"></a>
-<span class="sourceLineNo">267</span>    while (iter.hasNext()) {<a name="line.267"></a>
-<span class="sourceLineNo">268</span>      LocatedFileStatus file = iter.next();<a name="line.268"></a>
-<span class="sourceLineNo">269</span>      if (file.isDirectory()) {<a name="line.269"></a>
-<span class="sourceLineNo">270</span>        // recurse into sub directories<a name="line.270"></a>
-<span class="sourceLineNo">271</span>        result.addAll(getFiles(fs, file.getPath(), startTime, endTime));<a name="line.271"></a>
-<span class="sourceLineNo">272</span>      } else {<a name="line.272"></a>
-<span class="sourceLineNo">273</span>        String name = file.getPath().toString();<a name="line.273"></a>
-<span class="sourceLineNo">274</span>        int idx = name.lastIndexOf('.');<a name="line.274"></a>
-<span class="sourceLineNo">275</span>        if (idx &gt; 0) {<a name="line.275"></a>
-<span class="sourceLineNo">276</span>          try {<a name="line.276"></a>
-<span class="sourceLineNo">277</span>            long fileStartTime = Long.parseLong(name.substring(idx+1));<a name="line.277"></a>
-<span class="sourceLineNo">278</span>            if (fileStartTime &lt;= endTime) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span>              LOG.info("Found: " + file);<a name="line.279"></a>
-<span class="sourceLineNo">280</span>              result.add(file);<a name="line.280"></a>
-<span class="sourceLineNo">281</span>            }<a name="line.281"></a>
-<span class="sourceLineNo">282</span>          } catch (NumberFormatException x) {<a name="line.282"></a>
-<span class="sourceLineNo">283</span>            idx = 0;<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 (idx == 0) {<a name="line.286"></a>
-<span class="sourceLineNo">287</span>          LOG.warn("File " + name + " does not appear to be an WAL file. Skipping...");<a name="line.287"></a>
+<span class="sourceLineNo">248</span>  }<a name="line.248"></a>
+<span class="sourceLineNo">249</span><a name="line.249"></a>
+<span class="sourceLineNo">250</span>  /**<a name="line.250"></a>
+<span class="sourceLineNo">251</span>   * handler for non-deprecated WALKey version. fold into WALRecordReader once we no longer<a name="line.251"></a>
+<span class="sourceLineNo">252</span>   * need to support HLogInputFormat.<a name="line.252"></a>
+<span class="sourceLineNo">253</span>   */<a name="line.253"></a>
+<span class="sourceLineNo">254</span>  static class WALKeyRecordReader extends WALRecordReader&lt;WALKey&gt; {<a name="line.254"></a>
+<span class="sourceLineNo">255</span>    @Override<a name="line.255"></a>
+<span class="sourceLineNo">256</span>    public WALKey getCurrentKey() throws IOException, InterruptedException {<a name="line.256"></a>
+<span class="sourceLineNo">257</span>      return currentEntry.getKey();<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><a name="line.260"></a>
+<span class="sourceLineNo">261</span>  @Override<a name="line.261"></a>
+<span class="sourceLineNo">262</span>  public List&lt;InputSplit&gt; getSplits(JobContext context) throws IOException,<a name="line.262"></a>
+<span class="sourceLineNo">263</span>      InterruptedException {<a name="line.263"></a>
+<span class="sourceLineNo">264</span>    return getSplits(context, START_TIME_KEY, END_TIME_KEY);<a name="line.264"></a>
+<span class="sourceLineNo">265</span>  }<a name="line.265"></a>
+<span class="sourceLineNo">266</span><a name="line.266"></a>
+<span class="sourceLineNo">267</span>  /**<a name="line.267"></a>
+<span class="sourceLineNo">268</span>   * implementation shared with deprecated HLogInputFormat<a name="line.268"></a>
+<span class="sourceLineNo">269</span>   */<a name="line.269"></a>
+<span class="sourceLineNo">270</span>  List&lt;InputSplit&gt; getSplits(final JobContext context, final String startKey, final String endKey)<a name="line.270"></a>
+<span class="sourceLineNo">271</span>      throws IOException, InterruptedException {<a name="line.271"></a>
+<span class="sourceLineNo">272</span>    Configuration conf = context.getConfiguration();<a name="line.272"></a>
+<span class="sourceLineNo">273</span>    boolean ignoreMissing = conf.getBoolean(WALPlayer.IGNORE_MISSING_FILES, false);<a name="line.273"></a>
+<span class="sourceLineNo">274</span>    Path[] inputPaths = getInputPaths(conf);<a name="line.274"></a>
+<span class="sourceLineNo">275</span>    long startTime = conf.getLong(startKey, Long.MIN_VALUE);<a name="line.275"></a>
+<span class="sourceLineNo">276</span>    long endTime = conf.getLong(endKey, Long.MAX_VALUE);<a name="line.276"></a>
+<span class="sourceLineNo">277</span><a name="line.277"></a>
+<span class="sourceLineNo">278</span>    List&lt;FileStatus&gt; allFiles = new ArrayList&lt;FileStatus&gt;();<a name="line.278"></a>
+<span class="sourceLineNo">279</span>    for(Path inputPath: inputPaths){<a name="line.279"></a>
+<span class="sourceLineNo">280</span>      FileSystem fs = inputPath.getFileSystem(conf);<a name="line.280"></a>
+<span class="sourceLineNo">281</span>      try {<a name="line.281"></a>
+<span class="sourceLineNo">282</span>        List&lt;FileStatus&gt; files = getFiles(fs, inputPath, startTime, endTime);<a name="line.282"></a>
+<span class="sourceLineNo">283</span>        allFiles.addAll(files);<a name="line.283"></a>
+<span class="sourceLineNo">284</span>      } catch (FileNotFoundException e) {<a name="line.284"></a>
+<span class="sourceLineNo">285</span>        if (ignoreMissing) {<a name="line.285"></a>
+<span class="sourceLineNo">286</span>          LOG.warn("File "+ inputPath +" is missing. Skipping it.");<a name="line.286"></a>
+<span class="sourceLineNo">287</span>          continue;<a name="line.287"></a>
 <span class="sourceLineNo">288</span>        }<a name="line.288"></a>
-<span class="sourceLineNo">289</span>      }<a name="line.289"></a>
-<span class="sourceLineNo">290</span>    }<a name="line.290"></a>
-<span class="sourceLineNo">291</span>    return result;<a name="line.291"></a>
-<span class="sourceLineNo">292</span>  }<a name="line.292"></a>
-<span class="sourceLineNo">293</span><a name="line.293"></a>
-<span class="sourceLineNo">294</span>  @Override<a name="line.294"></a>
-<span class="sourceLineNo">295</span>  public RecordReader&lt;WALKey, WALEdit&gt; createRecordReader(InputSplit split,<a name="line.295"></a>
-<span class="sourceLineNo">296</span>      TaskAttemptContext context) throws IOException, InterruptedException {<a name="line.296"></a>
-<span class="sourceLineNo">297</span>    return new WALKeyRecordReader();<a name="line.297"></a>
-<span class="sourceLineNo">298</span>  }<a name="line.298"></a>
-<span class="sourceLineNo">299</span>}<a name="line.299"></a>
+<span class="sourceLineNo">289</span>        throw e;<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>    List&lt;InputSplit&gt; splits = new ArrayList&lt;InputSplit&gt;(allFiles.size());<a name="line.292"></a>
+<span class="sourceLineNo">293</span>    for (FileStatus file : allFiles) {<a name="line.293"></a>
+<span class="sourceLineNo">294</span>      splits.add(new WALSplit(file.getPath().toString(), file.getLen(), startTime, endTime));<a name="line.294"></a>
+<span class="sourceLineNo">295</span>    }<a name="line.295"></a>
+<span class="sourceLineNo">296</span>    return splits;<a name="line.296"></a>
+<span class="sourceLineNo">297</span>  }<a name="line.297"></a>
+<span class="sourceLineNo">298</span><a name="line.298"></a>
+<span class="sourceLineNo">299</span>  private Path[] getInputPaths(Configuration conf) {<a name="line.299"></a>
+<span class="sourceLineNo">300</span>    String inpDirs = conf.get(FileInputFormat.INPUT_DIR);<a name="line.300"></a>
+<span class="sourceLineNo">301</span>    return StringUtils.stringToPath(<a name="line.301"></a>
+<span class="sourceLineNo">302</span>      inpDirs.split(conf.get(WALPlayer.INPUT_FILES_SEPARATOR_KEY, ",")));<a name="line.302"></a>
+<span class="sourceLineNo">303</span>  }<a name="line.303"></a>
+<span class="sourceLineNo">304</span><a name="line.304"></a>
+<span class="sourceLineNo">305</span>  private List&lt;FileStatus&gt; getFiles(FileSystem fs, Path dir, long startTime, long endTime)<a name="line.305"></a>
+<span class="sourceLineNo">306</span>      throws IOException {<a name="line.306"></a>
+<span class="sourceLineNo">307</span>    List&lt;FileStatus&gt; result = new ArrayList&lt;&gt;();<a name="line.307"></a>
+<span class="sourceLineNo">308</span>    LOG.debug("Scanning " + dir.toString() + " for WAL files");<a name="line.308"></a>
+<span class="sourceLineNo">309</span><a name="line.309"></a>
+<span class="sourceLineNo">310</span>    RemoteIterator&lt;LocatedFileStatus&gt; iter = fs.listLocatedStatus(dir);<a name="line.310"></a>
+<span class="sourceLineNo">311</span>    if (!iter.hasNext()) return Collections.emptyList();<a name="line.311"></a>
+<span class="sourceLineNo">312</span>    while (iter.hasNext()) {<a name="line.312"></a>
+<span class="sourceLineNo">313</span>      LocatedFileStatus file = iter.next();<a name="line.313"></a>
+<span class="sourceLineNo">314</span>      if (file.isDirectory()) {<a name="line.314"></a>
+<span class="sourceLineNo">315</span>        // recurse into sub directories<a name="line.315"></a>
+<span class="sourceLineNo">316</span>        result.addAll(getFiles(fs, file.getPath(), startTime, endTime));<a name="line.316"></a>
+<span class="sourceLineNo">317</span>      } else {<a name="line.317"></a>
+<span class="sourceLineNo">318</span>        String name = file.getPath().toString();<a name="line.318"></a>
+<span class="sourceLineNo">319</span>        int idx = name.lastIndexOf('.');<a name="line.319"></a>
+<span class="sourceLineNo">320</span>        if (idx &gt; 0) {<a name="line.320"></a>
+<span class="sourceLineNo">321</span>          try {<a name="line.321"></a>
+<span class="sourceLineNo">322</span>            long fileStartTime = Long.parseLong(name.substring(idx+1));<a name="line.322"></a>
+<span class="sourceLineNo">323</span>            if (fileStartTime &lt;= endTime) {<a name="line.323"></a>
+<span class="sourceLineNo">324</span>              LOG.info("Found: " + file);<a name="line.324"></a>
+<span class="sourceLineNo">325</span>              result.add(file);<a name="line.325"></a>
+<span class="sourceLineNo">326</span>            }<a name="line.326"></a>
+<span class="sourceLineNo">327</span>          } catch (NumberFormatException x) {<a name="line.327"></a>
+<span class="sourceLineNo">328</span>            idx = 0;<a name="line.328"></a>
+<span class="sourceLineNo">329</span>          }<a name="line.329"></a>
+<span class="sourceLineNo">330</span>        }<a name="line.330"></a>
+<span class="sourceLineNo">331</span>        if (idx == 0) {<a name="line.331"></a>
+<span class="sourceLineNo">332</span>          LOG.warn("File " + name + " does not appear to be an WAL file. Skipping...");<a name="line.332"></a>
+<span class="sourceLineNo">333</span>        }<a name="line.333"></a>
+<span class="sourceLineNo">334</span>      }<a name="line.334"></a>
+<span class="sourceLineNo">335</span>    }<a name="line.335"></a>
+<span class="sourceLineNo">336</span>    return result;<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>
+<span class="sourceLineNo">339</span>  @Override<a name="line.339"></a>
+<span class="sourceLineNo">340</span>  public RecordReader&lt;WALKey, WALEdit&gt; createRecordReader(InputSplit split,<a name="line.340"></a>
+<span class="sourceLineNo">341</span>      TaskAttemptContext context) throws IOException, InterruptedException {<a name="line.341"></a>
+<span class="sourceLineNo">342</span>    return new WALKeyRecordReader();<a name="line.342"></a>
+<span class="sourceLineNo">343</span>  }<a name="line.343"></a>
+<span class="sourceLineNo">344</span>}<a name="line.344"></a>
 
 
 


Mime
View raw message