blob: 0e2cb76badfad3ba89323e4b2bbf399a7aa9d8aa [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.test;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.BufferedReader;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.ByteArrayOutputStream;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.DataInput;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.DataOutput;<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.io.InputStreamReader;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.net.URI;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.net.URISyntaxException;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.nio.charset.StandardCharsets;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import java.util.ArrayList;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import java.util.Collection;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import java.util.HashSet;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import java.util.List;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import java.util.Set;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import java.util.concurrent.CompletableFuture;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import java.util.concurrent.ExecutionException;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import java.util.concurrent.ExecutorService;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import java.util.concurrent.Executors;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import java.util.concurrent.TimeUnit;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import java.util.concurrent.atomic.AtomicLong;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import java.util.regex.Matcher;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import java.util.regex.Pattern;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import java.util.zip.GZIPInputStream;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.conf.Configuration;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.conf.Configured;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.fs.FSDataInputStream;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.fs.FileSystem;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.fs.Path;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.Cell;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.HBaseTestingUtil;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.HConstants;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.IntegrationTestBase;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.IntegrationTestingUtility;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.TableName;<a name="line.55"></a>
<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.AsyncConnection;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.AsyncTable;<a name="line.58"></a>
<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.59"></a>
<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.60"></a>
<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.61"></a>
<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.62"></a>
<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.client.Get;<a name="line.63"></a>
<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.client.Increment;<a name="line.64"></a>
<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.client.Put;<a name="line.65"></a>
<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.client.Result;<a name="line.66"></a>
<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.client.ScanResultConsumer;<a name="line.67"></a>
<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.client.Table;<a name="line.68"></a>
<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.69"></a>
<span class="sourceLineNo">070</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.70"></a>
<span class="sourceLineNo">071</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.71"></a>
<span class="sourceLineNo">072</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.72"></a>
<span class="sourceLineNo">073</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.73"></a>
<span class="sourceLineNo">074</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.74"></a>
<span class="sourceLineNo">075</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.75"></a>
<span class="sourceLineNo">076</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.76"></a>
<span class="sourceLineNo">077</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.77"></a>
<span class="sourceLineNo">078</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.78"></a>
<span class="sourceLineNo">079</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.79"></a>
<span class="sourceLineNo">080</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.80"></a>
<span class="sourceLineNo">081</span>import org.apache.hadoop.io.BytesWritable;<a name="line.81"></a>
<span class="sourceLineNo">082</span>import org.apache.hadoop.io.LongWritable;<a name="line.82"></a>
<span class="sourceLineNo">083</span>import org.apache.hadoop.io.NullWritable;<a name="line.83"></a>
<span class="sourceLineNo">084</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.84"></a>
<span class="sourceLineNo">085</span>import org.apache.hadoop.io.Writable;<a name="line.85"></a>
<span class="sourceLineNo">086</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.86"></a>
<span class="sourceLineNo">087</span>import org.apache.hadoop.mapreduce.Job;<a name="line.87"></a>
<span class="sourceLineNo">088</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.88"></a>
<span class="sourceLineNo">089</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.89"></a>
<span class="sourceLineNo">090</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.90"></a>
<span class="sourceLineNo">091</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.91"></a>
<span class="sourceLineNo">092</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.92"></a>
<span class="sourceLineNo">093</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.93"></a>
<span class="sourceLineNo">094</span>import org.apache.hadoop.util.Tool;<a name="line.94"></a>
<span class="sourceLineNo">095</span>import org.apache.hadoop.util.ToolRunner;<a name="line.95"></a>
<span class="sourceLineNo">096</span>import org.slf4j.Logger;<a name="line.96"></a>
<span class="sourceLineNo">097</span>import org.slf4j.LoggerFactory;<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.99"></a>
<span class="sourceLineNo">100</span><a name="line.100"></a>
<span class="sourceLineNo">101</span>/**<a name="line.101"></a>
<span class="sourceLineNo">102</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.102"></a>
<span class="sourceLineNo">103</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be used<a name="line.103"></a>
<span class="sourceLineNo">104</span> * to later verify the presence and integrity of those records.<a name="line.104"></a>
<span class="sourceLineNo">105</span> * &lt;p&gt;<a name="line.105"></a>
<span class="sourceLineNo">106</span> * Run like: &lt;blockquote&gt; ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl<a name="line.106"></a>
<span class="sourceLineNo">107</span> * \&lt;br&gt;<a name="line.107"></a>
<span class="sourceLineNo">108</span> * &amp;nbsp;&amp;nbsp; -Dfs.s3n.awsAccessKeyId=&amp;lt;AWS access key&amp;gt; \&lt;br&gt;<a name="line.108"></a>
<span class="sourceLineNo">109</span> * &amp;nbsp;&amp;nbsp; -Dfs.s3n.awsSecretAccessKey=&amp;lt;AWS secret key&amp;gt; \&lt;br&gt;<a name="line.109"></a>
<span class="sourceLineNo">110</span> * &amp;nbsp;&amp;nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \&lt;br&gt;<a name="line.110"></a>
<span class="sourceLineNo">111</span> * &amp;nbsp;&amp;nbsp; /path/to/tmp/warc-loader-output &lt;/blockquote&gt;<a name="line.111"></a>
<span class="sourceLineNo">112</span> * &lt;p&gt;<a name="line.112"></a>
<span class="sourceLineNo">113</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but Hadoop's<a name="line.113"></a>
<span class="sourceLineNo">114</span> * S3N filesystem still requires valid access credentials to initialize.<a name="line.114"></a>
<span class="sourceLineNo">115</span> * &lt;p&gt;<a name="line.115"></a>
<span class="sourceLineNo">116</span> * The input path can either specify a directory or a file. The file may optionally be compressed<a name="line.116"></a>
<span class="sourceLineNo">117</span> * with gzip. If a directory, the loader expects the directory to contain one or more WARC files<a name="line.117"></a>
<span class="sourceLineNo">118</span> * from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N URIs which<a name="line.118"></a>
<span class="sourceLineNo">119</span> * point to S3 locations for one or more WARC files from the Common Crawl dataset, one URI per line.<a name="line.119"></a>
<span class="sourceLineNo">120</span> * Lines should be terminated with the UNIX line terminator.<a name="line.120"></a>
<span class="sourceLineNo">121</span> * &lt;p&gt;<a name="line.121"></a>
<span class="sourceLineNo">122</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC files<a name="line.122"></a>
<span class="sourceLineNo">123</span> * comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.123"></a>
<span class="sourceLineNo">124</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.124"></a>
<span class="sourceLineNo">125</span> * request, and response, but we only load the response record types. If the HBase table schema does<a name="line.125"></a>
<span class="sourceLineNo">126</span> * not specify compression (by default) there is roughly a 10x expansion. Loading the full crawl<a name="line.126"></a>
<span class="sourceLineNo">127</span> * archive results in a table approximately 640 TB in size.<a name="line.127"></a>
<span class="sourceLineNo">128</span> * &lt;p&gt;<a name="line.128"></a>
<span class="sourceLineNo">129</span> * The loader can optionally drive read load during ingest by incrementing counters for each URL<a name="line.129"></a>
<span class="sourceLineNo">130</span> * discovered in content. Add &lt;tt&gt;-DIntegrationTestLoadCommonCrawl.increments=true&lt;/tt&gt; to the<a name="line.130"></a>
<span class="sourceLineNo">131</span> * command line to enable.<a name="line.131"></a>
<span class="sourceLineNo">132</span> * &lt;p&gt;<a name="line.132"></a>
<span class="sourceLineNo">133</span> * You can also split the Loader and Verify stages:<a name="line.133"></a>
<span class="sourceLineNo">134</span> * &lt;p&gt;<a name="line.134"></a>
<span class="sourceLineNo">135</span> * Load with: &lt;blockquote&gt; ./bin/hbase<a name="line.135"></a>
<span class="sourceLineNo">136</span> * 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \&lt;br&gt;<a name="line.136"></a>
<span class="sourceLineNo">137</span> * &amp;nbsp;&amp;nbsp; -files /path/to/hadoop-aws.jar \&lt;br&gt;<a name="line.137"></a>
<span class="sourceLineNo">138</span> * &amp;nbsp;&amp;nbsp; -Dfs.s3n.awsAccessKeyId=&amp;lt;AWS access key&amp;gt; \&lt;br&gt;<a name="line.138"></a>
<span class="sourceLineNo">139</span> * &amp;nbsp;&amp;nbsp; -Dfs.s3n.awsSecretAccessKey=&amp;lt;AWS secret key&amp;gt; \&lt;br&gt;<a name="line.139"></a>
<span class="sourceLineNo">140</span> * &amp;nbsp;&amp;nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \&lt;br&gt;<a name="line.140"></a>
<span class="sourceLineNo">141</span> * &amp;nbsp;&amp;nbsp; /path/to/tmp/warc-loader-output &lt;/blockquote&gt;<a name="line.141"></a>
<span class="sourceLineNo">142</span> * &lt;p&gt;<a name="line.142"></a>
<span class="sourceLineNo">143</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use the<a name="line.143"></a>
<span class="sourceLineNo">144</span> * &lt;tt&gt;-files&lt;/tt&gt; ToolRunner argument to add it.<a name="line.144"></a>
<span class="sourceLineNo">145</span> * &lt;p&gt;<a name="line.145"></a>
<span class="sourceLineNo">146</span> * Verify with: &lt;blockquote&gt; ./bin/hbase<a name="line.146"></a>
<span class="sourceLineNo">147</span> * 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \&lt;br&gt;<a name="line.147"></a>
<span class="sourceLineNo">148</span> * &amp;nbsp;&amp;nbsp; /path/to/tmp/warc-loader-output &lt;/blockquote&gt;<a name="line.148"></a>
<span class="sourceLineNo">149</span> * &lt;p&gt;<a name="line.149"></a>
<span class="sourceLineNo">150</span> */<a name="line.150"></a>
<span class="sourceLineNo">151</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.151"></a>
<span class="sourceLineNo">152</span><a name="line.152"></a>
<span class="sourceLineNo">153</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> static final String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.155"></a>
<span class="sourceLineNo">156</span> static final String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.156"></a>
<span class="sourceLineNo">157</span><a name="line.157"></a>
<span class="sourceLineNo">158</span> static final String INCREMENTS_NAME_KEY = "IntegrationTestLoadCommonCrawl.increments";<a name="line.158"></a>
<span class="sourceLineNo">159</span> static final boolean DEFAULT_INCREMENTS = false;<a name="line.159"></a>
<span class="sourceLineNo">160</span><a name="line.160"></a>
<span class="sourceLineNo">161</span> static final int MAX_INFLIGHT = 1000;<a name="line.161"></a>
<span class="sourceLineNo">162</span> static final int INFLIGHT_PAUSE_MS = 100;<a name="line.162"></a>
<span class="sourceLineNo">163</span><a name="line.163"></a>
<span class="sourceLineNo">164</span> static final byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.164"></a>
<span class="sourceLineNo">165</span> static final byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.165"></a>
<span class="sourceLineNo">166</span> static final byte[] URL_FAMILY_NAME = Bytes.toBytes("u");<a name="line.166"></a>
<span class="sourceLineNo">167</span> static final byte[] SEP = Bytes.toBytes(":");<a name="line.167"></a>
<span class="sourceLineNo">168</span> static final byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.168"></a>
<span class="sourceLineNo">169</span> static final byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.169"></a>
<span class="sourceLineNo">170</span> static final byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.170"></a>
<span class="sourceLineNo">171</span> static final byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.171"></a>
<span class="sourceLineNo">172</span> static final byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.172"></a>
<span class="sourceLineNo">173</span> static final byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.173"></a>
<span class="sourceLineNo">174</span> static final byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.174"></a>
<span class="sourceLineNo">175</span> static final byte[] REF_QUALIFIER = Bytes.toBytes("ref");<a name="line.175"></a>
<span class="sourceLineNo">176</span><a name="line.176"></a>
<span class="sourceLineNo">177</span> public static enum Counts {<a name="line.177"></a>
<span class="sourceLineNo">178</span> REFERENCED,<a name="line.178"></a>
<span class="sourceLineNo">179</span> UNREFERENCED,<a name="line.179"></a>
<span class="sourceLineNo">180</span> CORRUPT,<a name="line.180"></a>
<span class="sourceLineNo">181</span> RPC_BYTES_WRITTEN,<a name="line.181"></a>
<span class="sourceLineNo">182</span> RPC_TIME_MS,<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> protected Path warcFileInputDir = null;<a name="line.185"></a>
<span class="sourceLineNo">186</span> protected Path outputDir = null;<a name="line.186"></a>
<span class="sourceLineNo">187</span> protected String[] args;<a name="line.187"></a>
<span class="sourceLineNo">188</span><a name="line.188"></a>
<span class="sourceLineNo">189</span> protected int runLoader(final Path warcFileInputDir, final Path outputDir) throws Exception {<a name="line.189"></a>
<span class="sourceLineNo">190</span> Loader loader = new Loader();<a name="line.190"></a>
<span class="sourceLineNo">191</span> loader.setConf(conf);<a name="line.191"></a>
<span class="sourceLineNo">192</span> return loader.run(warcFileInputDir, outputDir);<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
<span class="sourceLineNo">194</span><a name="line.194"></a>
<span class="sourceLineNo">195</span> protected int runVerify(final Path inputDir) throws Exception {<a name="line.195"></a>
<span class="sourceLineNo">196</span> Verify verify = new Verify();<a name="line.196"></a>
<span class="sourceLineNo">197</span> verify.setConf(conf);<a name="line.197"></a>
<span class="sourceLineNo">198</span> return verify.run(inputDir);<a name="line.198"></a>
<span class="sourceLineNo">199</span> }<a name="line.199"></a>
<span class="sourceLineNo">200</span><a name="line.200"></a>
<span class="sourceLineNo">201</span> @Override<a name="line.201"></a>
<span class="sourceLineNo">202</span> public int run(String[] args) {<a name="line.202"></a>
<span class="sourceLineNo">203</span> if (args.length &gt; 0) {<a name="line.203"></a>
<span class="sourceLineNo">204</span> warcFileInputDir = new Path(args[0]);<a name="line.204"></a>
<span class="sourceLineNo">205</span> if (args.length &gt; 1) {<a name="line.205"></a>
<span class="sourceLineNo">206</span> outputDir = new Path(args[1]);<a name="line.206"></a>
<span class="sourceLineNo">207</span> }<a name="line.207"></a>
<span class="sourceLineNo">208</span> }<a name="line.208"></a>
<span class="sourceLineNo">209</span> try {<a name="line.209"></a>
<span class="sourceLineNo">210</span> if (warcFileInputDir == null) {<a name="line.210"></a>
<span class="sourceLineNo">211</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.211"></a>
<span class="sourceLineNo">212</span> }<a name="line.212"></a>
<span class="sourceLineNo">213</span> if (outputDir == null) {<a name="line.213"></a>
<span class="sourceLineNo">214</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.214"></a>
<span class="sourceLineNo">215</span> }<a name="line.215"></a>
<span class="sourceLineNo">216</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.216"></a>
<span class="sourceLineNo">217</span> if (res != 0) {<a name="line.217"></a>
<span class="sourceLineNo">218</span> LOG.error("Loader failed");<a name="line.218"></a>
<span class="sourceLineNo">219</span> return -1;<a name="line.219"></a>
<span class="sourceLineNo">220</span> }<a name="line.220"></a>
<span class="sourceLineNo">221</span> res = runVerify(outputDir);<a name="line.221"></a>
<span class="sourceLineNo">222</span> } catch (Exception e) {<a name="line.222"></a>
<span class="sourceLineNo">223</span> LOG.error("Tool failed with exception", e);<a name="line.223"></a>
<span class="sourceLineNo">224</span> return -1;<a name="line.224"></a>
<span class="sourceLineNo">225</span> }<a name="line.225"></a>
<span class="sourceLineNo">226</span> return 0;<a name="line.226"></a>
<span class="sourceLineNo">227</span> }<a name="line.227"></a>
<span class="sourceLineNo">228</span><a name="line.228"></a>
<span class="sourceLineNo">229</span> @Override<a name="line.229"></a>
<span class="sourceLineNo">230</span> protected void processOptions(final CommandLine cmd) {<a name="line.230"></a>
<span class="sourceLineNo">231</span> processBaseOptions(cmd);<a name="line.231"></a>
<span class="sourceLineNo">232</span> args = cmd.getArgs();<a name="line.232"></a>
<span class="sourceLineNo">233</span> }<a name="line.233"></a>
<span class="sourceLineNo">234</span><a name="line.234"></a>
<span class="sourceLineNo">235</span> @Override<a name="line.235"></a>
<span class="sourceLineNo">236</span> public void setUpCluster() throws Exception {<a name="line.236"></a>
<span class="sourceLineNo">237</span> util = getTestingUtil(getConf());<a name="line.237"></a>
<span class="sourceLineNo">238</span> boolean isDistributed = util.isDistributedCluster();<a name="line.238"></a>
<span class="sourceLineNo">239</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.239"></a>
<span class="sourceLineNo">240</span> if (!isDistributed) {<a name="line.240"></a>
<span class="sourceLineNo">241</span> util.startMiniMapReduceCluster();<a name="line.241"></a>
<span class="sourceLineNo">242</span> }<a name="line.242"></a>
<span class="sourceLineNo">243</span> this.setConf(util.getConfiguration());<a name="line.243"></a>
<span class="sourceLineNo">244</span> }<a name="line.244"></a>
<span class="sourceLineNo">245</span><a name="line.245"></a>
<span class="sourceLineNo">246</span> @Override<a name="line.246"></a>
<span class="sourceLineNo">247</span> public void cleanUpCluster() throws Exception {<a name="line.247"></a>
<span class="sourceLineNo">248</span> super.cleanUpCluster();<a name="line.248"></a>
<span class="sourceLineNo">249</span> if (util.isDistributedCluster()) {<a name="line.249"></a>
<span class="sourceLineNo">250</span> util.shutdownMiniMapReduceCluster();<a name="line.250"></a>
<span class="sourceLineNo">251</span> }<a name="line.251"></a>
<span class="sourceLineNo">252</span> }<a name="line.252"></a>
<span class="sourceLineNo">253</span><a name="line.253"></a>
<span class="sourceLineNo">254</span> static TableName getTablename(final Configuration c) {<a name="line.254"></a>
<span class="sourceLineNo">255</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.255"></a>
<span class="sourceLineNo">256</span> }<a name="line.256"></a>
<span class="sourceLineNo">257</span><a name="line.257"></a>
<span class="sourceLineNo">258</span> @Override<a name="line.258"></a>
<span class="sourceLineNo">259</span> public TableName getTablename() {<a name="line.259"></a>
<span class="sourceLineNo">260</span> return getTablename(getConf());<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> @Override<a name="line.263"></a>
<span class="sourceLineNo">264</span> protected Set&lt;String&gt; getColumnFamilies() {<a name="line.264"></a>
<span class="sourceLineNo">265</span> Set&lt;String&gt; families = new HashSet&lt;&gt;();<a name="line.265"></a>
<span class="sourceLineNo">266</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.266"></a>
<span class="sourceLineNo">267</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.267"></a>
<span class="sourceLineNo">268</span> families.add(Bytes.toString(URL_FAMILY_NAME));<a name="line.268"></a>
<span class="sourceLineNo">269</span> return families;<a name="line.269"></a>
<span class="sourceLineNo">270</span> }<a name="line.270"></a>
<span class="sourceLineNo">271</span><a name="line.271"></a>
<span class="sourceLineNo">272</span> @Override<a name="line.272"></a>
<span class="sourceLineNo">273</span> public int runTestFromCommandLine() throws Exception {<a name="line.273"></a>
<span class="sourceLineNo">274</span> return ToolRunner.run(getConf(), this, args);<a name="line.274"></a>
<span class="sourceLineNo">275</span> }<a name="line.275"></a>
<span class="sourceLineNo">276</span><a name="line.276"></a>
<span class="sourceLineNo">277</span> public static void main(String[] args) throws Exception {<a name="line.277"></a>
<span class="sourceLineNo">278</span> Configuration conf = HBaseConfiguration.create();<a name="line.278"></a>
<span class="sourceLineNo">279</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.279"></a>
<span class="sourceLineNo">280</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.280"></a>
<span class="sourceLineNo">281</span> System.exit(ret);<a name="line.281"></a>
<span class="sourceLineNo">282</span> }<a name="line.282"></a>
<span class="sourceLineNo">283</span><a name="line.283"></a>
<span class="sourceLineNo">284</span> public static class HBaseKeyWritable implements Writable {<a name="line.284"></a>
<span class="sourceLineNo">285</span><a name="line.285"></a>
<span class="sourceLineNo">286</span> private byte[] row;<a name="line.286"></a>
<span class="sourceLineNo">287</span> private int rowOffset;<a name="line.287"></a>
<span class="sourceLineNo">288</span> private int rowLength;<a name="line.288"></a>
<span class="sourceLineNo">289</span> private byte[] family;<a name="line.289"></a>
<span class="sourceLineNo">290</span> private int familyOffset;<a name="line.290"></a>
<span class="sourceLineNo">291</span> private int familyLength;<a name="line.291"></a>
<span class="sourceLineNo">292</span> private byte[] qualifier;<a name="line.292"></a>
<span class="sourceLineNo">293</span> private int qualifierOffset;<a name="line.293"></a>
<span class="sourceLineNo">294</span> private int qualifierLength;<a name="line.294"></a>
<span class="sourceLineNo">295</span> private long ts;<a name="line.295"></a>
<span class="sourceLineNo">296</span><a name="line.296"></a>
<span class="sourceLineNo">297</span> public HBaseKeyWritable() {<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">300</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength, byte[] family,<a name="line.300"></a>
<span class="sourceLineNo">301</span> int familyOffset, int familyLength, byte[] qualifier, int qualifierOffset,<a name="line.301"></a>
<span class="sourceLineNo">302</span> int qualifierLength, long ts) {<a name="line.302"></a>
<span class="sourceLineNo">303</span> this.row = row;<a name="line.303"></a>
<span class="sourceLineNo">304</span> this.rowOffset = rowOffset;<a name="line.304"></a>
<span class="sourceLineNo">305</span> this.rowLength = rowLength;<a name="line.305"></a>
<span class="sourceLineNo">306</span> this.family = family;<a name="line.306"></a>
<span class="sourceLineNo">307</span> this.familyOffset = familyOffset;<a name="line.307"></a>
<span class="sourceLineNo">308</span> this.familyLength = familyLength;<a name="line.308"></a>
<span class="sourceLineNo">309</span> this.qualifier = qualifier;<a name="line.309"></a>
<span class="sourceLineNo">310</span> this.qualifierOffset = qualifierOffset;<a name="line.310"></a>
<span class="sourceLineNo">311</span> this.qualifierLength = qualifierLength;<a name="line.311"></a>
<span class="sourceLineNo">312</span> this.ts = ts;<a name="line.312"></a>
<span class="sourceLineNo">313</span> }<a name="line.313"></a>
<span class="sourceLineNo">314</span><a name="line.314"></a>
<span class="sourceLineNo">315</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.315"></a>
<span class="sourceLineNo">316</span> this(row, 0, row.length, family, 0, family.length, qualifier, 0,<a name="line.316"></a>
<span class="sourceLineNo">317</span> qualifier != null ? qualifier.length : 0, ts);<a name="line.317"></a>
<span class="sourceLineNo">318</span> }<a name="line.318"></a>
<span class="sourceLineNo">319</span><a name="line.319"></a>
<span class="sourceLineNo">320</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier) {<a name="line.320"></a>
<span class="sourceLineNo">321</span> this(row, family, qualifier, Long.MAX_VALUE);<a name="line.321"></a>
<span class="sourceLineNo">322</span> }<a name="line.322"></a>
<span class="sourceLineNo">323</span><a name="line.323"></a>
<span class="sourceLineNo">324</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.324"></a>
<span class="sourceLineNo">325</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.325"></a>
<span class="sourceLineNo">326</span> }<a name="line.326"></a>
<span class="sourceLineNo">327</span><a name="line.327"></a>
<span class="sourceLineNo">328</span> public HBaseKeyWritable(byte[] row, byte[] family) {<a name="line.328"></a>
<span class="sourceLineNo">329</span> this(row, family, Long.MAX_VALUE);<a name="line.329"></a>
<span class="sourceLineNo">330</span> }<a name="line.330"></a>
<span class="sourceLineNo">331</span><a name="line.331"></a>
<span class="sourceLineNo">332</span> public HBaseKeyWritable(Cell cell) {<a name="line.332"></a>
<span class="sourceLineNo">333</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(), cell.getFamilyArray(),<a name="line.333"></a>
<span class="sourceLineNo">334</span> cell.getFamilyOffset(), cell.getFamilyLength(), cell.getQualifierArray(),<a name="line.334"></a>
<span class="sourceLineNo">335</span> cell.getQualifierOffset(), cell.getQualifierLength(), cell.getTimestamp());<a name="line.335"></a>
<span class="sourceLineNo">336</span> }<a name="line.336"></a>
<span class="sourceLineNo">337</span><a name="line.337"></a>
<span class="sourceLineNo">338</span> @Override<a name="line.338"></a>
<span class="sourceLineNo">339</span> public void readFields(DataInput in) throws IOException {<a name="line.339"></a>
<span class="sourceLineNo">340</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.340"></a>
<span class="sourceLineNo">341</span> this.rowOffset = 0;<a name="line.341"></a>
<span class="sourceLineNo">342</span> this.rowLength = row.length;<a name="line.342"></a>
<span class="sourceLineNo">343</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.343"></a>
<span class="sourceLineNo">344</span> this.familyOffset = 0;<a name="line.344"></a>
<span class="sourceLineNo">345</span> this.familyLength = family.length;<a name="line.345"></a>
<span class="sourceLineNo">346</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.346"></a>
<span class="sourceLineNo">347</span> this.qualifierOffset = 0;<a name="line.347"></a>
<span class="sourceLineNo">348</span> this.qualifierLength = qualifier.length;<a name="line.348"></a>
<span class="sourceLineNo">349</span> this.ts = in.readLong();<a name="line.349"></a>
<span class="sourceLineNo">350</span> }<a name="line.350"></a>
<span class="sourceLineNo">351</span><a name="line.351"></a>
<span class="sourceLineNo">352</span> @Override<a name="line.352"></a>
<span class="sourceLineNo">353</span> public void write(DataOutput out) throws IOException {<a name="line.353"></a>
<span class="sourceLineNo">354</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.354"></a>
<span class="sourceLineNo">355</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.355"></a>
<span class="sourceLineNo">356</span> if (qualifier != null) {<a name="line.356"></a>
<span class="sourceLineNo">357</span> out.writeUTF(<a name="line.357"></a>
<span class="sourceLineNo">358</span> new String(qualifier, qualifierOffset, qualifierLength, StandardCharsets.UTF_8));<a name="line.358"></a>
<span class="sourceLineNo">359</span> } else {<a name="line.359"></a>
<span class="sourceLineNo">360</span> out.writeUTF("");<a name="line.360"></a>
<span class="sourceLineNo">361</span> }<a name="line.361"></a>
<span class="sourceLineNo">362</span> out.writeLong(ts);<a name="line.362"></a>
<span class="sourceLineNo">363</span> }<a name="line.363"></a>
<span class="sourceLineNo">364</span><a name="line.364"></a>
<span class="sourceLineNo">365</span> public byte[] getRowArray() {<a name="line.365"></a>
<span class="sourceLineNo">366</span> return row;<a name="line.366"></a>
<span class="sourceLineNo">367</span> }<a name="line.367"></a>
<span class="sourceLineNo">368</span><a name="line.368"></a>
<span class="sourceLineNo">369</span> public void setRow(byte[] row) {<a name="line.369"></a>
<span class="sourceLineNo">370</span> this.row = row;<a name="line.370"></a>
<span class="sourceLineNo">371</span> }<a name="line.371"></a>
<span class="sourceLineNo">372</span><a name="line.372"></a>
<span class="sourceLineNo">373</span> public int getRowOffset() {<a name="line.373"></a>
<span class="sourceLineNo">374</span> return rowOffset;<a name="line.374"></a>
<span class="sourceLineNo">375</span> }<a name="line.375"></a>
<span class="sourceLineNo">376</span><a name="line.376"></a>
<span class="sourceLineNo">377</span> public void setRowOffset(int rowOffset) {<a name="line.377"></a>
<span class="sourceLineNo">378</span> this.rowOffset = rowOffset;<a name="line.378"></a>
<span class="sourceLineNo">379</span> }<a name="line.379"></a>
<span class="sourceLineNo">380</span><a name="line.380"></a>
<span class="sourceLineNo">381</span> public int getRowLength() {<a name="line.381"></a>
<span class="sourceLineNo">382</span> return rowLength;<a name="line.382"></a>
<span class="sourceLineNo">383</span> }<a name="line.383"></a>
<span class="sourceLineNo">384</span><a name="line.384"></a>
<span class="sourceLineNo">385</span> public void setRowLength(int rowLength) {<a name="line.385"></a>
<span class="sourceLineNo">386</span> this.rowLength = rowLength;<a name="line.386"></a>
<span class="sourceLineNo">387</span> }<a name="line.387"></a>
<span class="sourceLineNo">388</span><a name="line.388"></a>
<span class="sourceLineNo">389</span> public byte[] getFamilyArray() {<a name="line.389"></a>
<span class="sourceLineNo">390</span> return family;<a name="line.390"></a>
<span class="sourceLineNo">391</span> }<a name="line.391"></a>
<span class="sourceLineNo">392</span><a name="line.392"></a>
<span class="sourceLineNo">393</span> public void setFamily(byte[] family) {<a name="line.393"></a>
<span class="sourceLineNo">394</span> this.family = family;<a name="line.394"></a>
<span class="sourceLineNo">395</span> }<a name="line.395"></a>
<span class="sourceLineNo">396</span><a name="line.396"></a>
<span class="sourceLineNo">397</span> public int getFamilyOffset() {<a name="line.397"></a>
<span class="sourceLineNo">398</span> return familyOffset;<a name="line.398"></a>
<span class="sourceLineNo">399</span> }<a name="line.399"></a>
<span class="sourceLineNo">400</span><a name="line.400"></a>
<span class="sourceLineNo">401</span> public void setFamilyOffset(int familyOffset) {<a name="line.401"></a>
<span class="sourceLineNo">402</span> this.familyOffset = familyOffset;<a name="line.402"></a>
<span class="sourceLineNo">403</span> }<a name="line.403"></a>
<span class="sourceLineNo">404</span><a name="line.404"></a>
<span class="sourceLineNo">405</span> public int getFamilyLength() {<a name="line.405"></a>
<span class="sourceLineNo">406</span> return familyLength;<a name="line.406"></a>
<span class="sourceLineNo">407</span> }<a name="line.407"></a>
<span class="sourceLineNo">408</span><a name="line.408"></a>
<span class="sourceLineNo">409</span> public void setFamilyLength(int familyLength) {<a name="line.409"></a>
<span class="sourceLineNo">410</span> this.familyLength = familyLength;<a name="line.410"></a>
<span class="sourceLineNo">411</span> }<a name="line.411"></a>
<span class="sourceLineNo">412</span><a name="line.412"></a>
<span class="sourceLineNo">413</span> public byte[] getQualifierArray() {<a name="line.413"></a>
<span class="sourceLineNo">414</span> return qualifier;<a name="line.414"></a>
<span class="sourceLineNo">415</span> }<a name="line.415"></a>
<span class="sourceLineNo">416</span><a name="line.416"></a>
<span class="sourceLineNo">417</span> public void setQualifier(byte[] qualifier) {<a name="line.417"></a>
<span class="sourceLineNo">418</span> this.qualifier = qualifier;<a name="line.418"></a>
<span class="sourceLineNo">419</span> }<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
<span class="sourceLineNo">421</span> public int getQualifierOffset() {<a name="line.421"></a>
<span class="sourceLineNo">422</span> return qualifierOffset;<a name="line.422"></a>
<span class="sourceLineNo">423</span> }<a name="line.423"></a>
<span class="sourceLineNo">424</span><a name="line.424"></a>
<span class="sourceLineNo">425</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.425"></a>
<span class="sourceLineNo">426</span> this.qualifierOffset = qualifierOffset;<a name="line.426"></a>
<span class="sourceLineNo">427</span> }<a name="line.427"></a>
<span class="sourceLineNo">428</span><a name="line.428"></a>
<span class="sourceLineNo">429</span> public int getQualifierLength() {<a name="line.429"></a>
<span class="sourceLineNo">430</span> return qualifierLength;<a name="line.430"></a>
<span class="sourceLineNo">431</span> }<a name="line.431"></a>
<span class="sourceLineNo">432</span><a name="line.432"></a>
<span class="sourceLineNo">433</span> public void setQualifierLength(int qualifierLength) {<a name="line.433"></a>
<span class="sourceLineNo">434</span> this.qualifierLength = qualifierLength;<a name="line.434"></a>
<span class="sourceLineNo">435</span> }<a name="line.435"></a>
<span class="sourceLineNo">436</span><a name="line.436"></a>
<span class="sourceLineNo">437</span> public long getTimestamp() {<a name="line.437"></a>
<span class="sourceLineNo">438</span> return ts;<a name="line.438"></a>
<span class="sourceLineNo">439</span> }<a name="line.439"></a>
<span class="sourceLineNo">440</span><a name="line.440"></a>
<span class="sourceLineNo">441</span> public void setTimestamp(long ts) {<a name="line.441"></a>
<span class="sourceLineNo">442</span> this.ts = ts;<a name="line.442"></a>
<span class="sourceLineNo">443</span> }<a name="line.443"></a>
<span class="sourceLineNo">444</span> }<a name="line.444"></a>
<span class="sourceLineNo">445</span><a name="line.445"></a>
<span class="sourceLineNo">446</span> public static class Loader extends Configured implements Tool {<a name="line.446"></a>
<span class="sourceLineNo">447</span><a name="line.447"></a>
<span class="sourceLineNo">448</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.448"></a>
<span class="sourceLineNo">449</span> private static final String USAGE = "Loader &lt;warInputDir | warFileList&gt; &lt;outputDir&gt;";<a name="line.449"></a>
<span class="sourceLineNo">450</span><a name="line.450"></a>
<span class="sourceLineNo">451</span> void createSchema(final TableName tableName) throws IOException {<a name="line.451"></a>
<span class="sourceLineNo">452</span><a name="line.452"></a>
<span class="sourceLineNo">453</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.453"></a>
<span class="sourceLineNo">454</span> Admin admin = conn.getAdmin()) {<a name="line.454"></a>
<span class="sourceLineNo">455</span> if (!admin.tableExists(tableName)) {<a name="line.455"></a>
<span class="sourceLineNo">456</span><a name="line.456"></a>
<span class="sourceLineNo">457</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.457"></a>
<span class="sourceLineNo">458</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME).setMaxVersions(1000)<a name="line.458"></a>
<span class="sourceLineNo">459</span> .setDataBlockEncoding(DataBlockEncoding.NONE).setBloomFilterType(BloomType.ROW);<a name="line.459"></a>
<span class="sourceLineNo">460</span><a name="line.460"></a>
<span class="sourceLineNo">461</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.461"></a>
<span class="sourceLineNo">462</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME).setMaxVersions(1000)<a name="line.462"></a>
<span class="sourceLineNo">463</span> .setDataBlockEncoding(DataBlockEncoding.ROW_INDEX_V1)<a name="line.463"></a>
<span class="sourceLineNo">464</span> .setBloomFilterType(BloomType.ROWCOL).setBlocksize(8 * 1024);<a name="line.464"></a>
<span class="sourceLineNo">465</span><a name="line.465"></a>
<span class="sourceLineNo">466</span> ColumnFamilyDescriptorBuilder urlFamilyBuilder =<a name="line.466"></a>
<span class="sourceLineNo">467</span> ColumnFamilyDescriptorBuilder.newBuilder(URL_FAMILY_NAME).setMaxVersions(1000)<a name="line.467"></a>
<span class="sourceLineNo">468</span> .setDataBlockEncoding(DataBlockEncoding.ROW_INDEX_V1)<a name="line.468"></a>
<span class="sourceLineNo">469</span> .setBloomFilterType(BloomType.ROWCOL).setBlocksize(8 * 1024);<a name="line.469"></a>
<span class="sourceLineNo">470</span><a name="line.470"></a>
<span class="sourceLineNo">471</span> Set&lt;ColumnFamilyDescriptor&gt; families = new HashSet&lt;&gt;();<a name="line.471"></a>
<span class="sourceLineNo">472</span> families.add(contentFamilyBuilder.build());<a name="line.472"></a>
<span class="sourceLineNo">473</span> families.add(infoFamilyBuilder.build());<a name="line.473"></a>
<span class="sourceLineNo">474</span> families.add(urlFamilyBuilder.build());<a name="line.474"></a>
<span class="sourceLineNo">475</span><a name="line.475"></a>
<span class="sourceLineNo">476</span> TableDescriptor tableDescriptor =<a name="line.476"></a>
<span class="sourceLineNo">477</span> TableDescriptorBuilder.newBuilder(tableName).setColumnFamilies(families).build();<a name="line.477"></a>
<span class="sourceLineNo">478</span><a name="line.478"></a>
<span class="sourceLineNo">479</span> if (<a name="line.479"></a>
<span class="sourceLineNo">480</span> getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.480"></a>
<span class="sourceLineNo">481</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)<a name="line.481"></a>
<span class="sourceLineNo">482</span> ) {<a name="line.482"></a>
<span class="sourceLineNo">483</span> int numberOfServers = admin.getRegionServers().size();<a name="line.483"></a>
<span class="sourceLineNo">484</span> if (numberOfServers == 0) {<a name="line.484"></a>
<span class="sourceLineNo">485</span> throw new IllegalStateException("No live regionservers");<a name="line.485"></a>
<span class="sourceLineNo">486</span> }<a name="line.486"></a>
<span class="sourceLineNo">487</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.487"></a>
<span class="sourceLineNo">488</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.488"></a>
<span class="sourceLineNo">489</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.489"></a>
<span class="sourceLineNo">490</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.490"></a>
<span class="sourceLineNo">491</span> LOG.info("Number of live regionservers: " + numberOfServers + ", "<a name="line.491"></a>
<span class="sourceLineNo">492</span> + "pre-splitting table into " + totalNumberOfRegions + " regions "<a name="line.492"></a>
<span class="sourceLineNo">493</span> + "(default regions per server: " + regionsPerServer + ")");<a name="line.493"></a>
<span class="sourceLineNo">494</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.494"></a>
<span class="sourceLineNo">495</span> admin.createTable(tableDescriptor, splits);<a name="line.495"></a>
<span class="sourceLineNo">496</span> } else {<a name="line.496"></a>
<span class="sourceLineNo">497</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.497"></a>
<span class="sourceLineNo">498</span> admin.createTable(tableDescriptor);<a name="line.498"></a>
<span class="sourceLineNo">499</span> }<a name="line.499"></a>
<span class="sourceLineNo">500</span> }<a name="line.500"></a>
<span class="sourceLineNo">501</span> } catch (MasterNotRunningException e) {<a name="line.501"></a>
<span class="sourceLineNo">502</span> LOG.error("Master not running", e);<a name="line.502"></a>
<span class="sourceLineNo">503</span> throw new IOException(e);<a name="line.503"></a>
<span class="sourceLineNo">504</span> }<a name="line.504"></a>
<span class="sourceLineNo">505</span> }<a name="line.505"></a>
<span class="sourceLineNo">506</span><a name="line.506"></a>
<span class="sourceLineNo">507</span> int run(final Path warcFileInput, final Path outputDir)<a name="line.507"></a>
<span class="sourceLineNo">508</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.508"></a>
<span class="sourceLineNo">509</span><a name="line.509"></a>
<span class="sourceLineNo">510</span> createSchema(getTablename(getConf()));<a name="line.510"></a>
<span class="sourceLineNo">511</span><a name="line.511"></a>
<span class="sourceLineNo">512</span> final Job job = Job.getInstance(getConf());<a name="line.512"></a>
<span class="sourceLineNo">513</span> job.setJobName(Loader.class.getName());<a name="line.513"></a>
<span class="sourceLineNo">514</span> job.setNumReduceTasks(0);<a name="line.514"></a>
<span class="sourceLineNo">515</span> job.setJarByClass(getClass());<a name="line.515"></a>
<span class="sourceLineNo">516</span> job.setMapperClass(LoaderMapper.class);<a name="line.516"></a>
<span class="sourceLineNo">517</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.517"></a>
<span class="sourceLineNo">518</span> final FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.518"></a>
<span class="sourceLineNo">519</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.519"></a>
<span class="sourceLineNo">520</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.520"></a>
<span class="sourceLineNo">521</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.521"></a>
<span class="sourceLineNo">522</span> } else if (warcFileInput.toUri().getScheme().equals("file")) {<a name="line.522"></a>
<span class="sourceLineNo">523</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.523"></a>
<span class="sourceLineNo">524</span> final List&lt;Path&gt; paths = new ArrayList&lt;Path&gt;();<a name="line.524"></a>
<span class="sourceLineNo">525</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.525"></a>
<span class="sourceLineNo">526</span> InputStreamReader reader;<a name="line.526"></a>
<span class="sourceLineNo">527</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.527"></a>
<span class="sourceLineNo">528</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.528"></a>
<span class="sourceLineNo">529</span> } else {<a name="line.529"></a>
<span class="sourceLineNo">530</span> reader = new InputStreamReader(is);<a name="line.530"></a>
<span class="sourceLineNo">531</span> }<a name="line.531"></a>
<span class="sourceLineNo">532</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.532"></a>
<span class="sourceLineNo">533</span> String line;<a name="line.533"></a>
<span class="sourceLineNo">534</span> while ((line = br.readLine()) != null) {<a name="line.534"></a>
<span class="sourceLineNo">535</span> paths.add(new Path(line));<a name="line.535"></a>
<span class="sourceLineNo">536</span> }<a name="line.536"></a>
<span class="sourceLineNo">537</span> }<a name="line.537"></a>
<span class="sourceLineNo">538</span> }<a name="line.538"></a>
<span class="sourceLineNo">539</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.539"></a>
<span class="sourceLineNo">540</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.540"></a>
<span class="sourceLineNo">541</span> } else {<a name="line.541"></a>
<span class="sourceLineNo">542</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.542"></a>
<span class="sourceLineNo">543</span> }<a name="line.543"></a>
<span class="sourceLineNo">544</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.544"></a>
<span class="sourceLineNo">545</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.545"></a>
<span class="sourceLineNo">546</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.546"></a>
<span class="sourceLineNo">547</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.547"></a>
<span class="sourceLineNo">548</span> job.setOutputValueClass(BytesWritable.class);<a name="line.548"></a>
<span class="sourceLineNo">549</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.549"></a>
<span class="sourceLineNo">550</span> // Increase max attempts because S3 might throttle aggressively and ultimately fail a task<a name="line.550"></a>
<span class="sourceLineNo">551</span> job.getConfiguration().setInt("mapred.map.max.attempts", 100);<a name="line.551"></a>
<span class="sourceLineNo">552</span> job.getConfiguration().setInt("mapreduce.map.maxattempts", 100);<a name="line.552"></a>
<span class="sourceLineNo">553</span><a name="line.553"></a>
<span class="sourceLineNo">554</span> boolean success = job.waitForCompletion(true);<a name="line.554"></a>
<span class="sourceLineNo">555</span> if (!success) {<a name="line.555"></a>
<span class="sourceLineNo">556</span> LOG.error("Failure during job " + job.getJobID());<a name="line.556"></a>
<span class="sourceLineNo">557</span> }<a name="line.557"></a>
<span class="sourceLineNo">558</span><a name="line.558"></a>
<span class="sourceLineNo">559</span> final Counters counters = job.getCounters();<a name="line.559"></a>
<span class="sourceLineNo">560</span> for (Counts c : Counts.values()) {<a name="line.560"></a>
<span class="sourceLineNo">561</span> long value = counters.findCounter(c).getValue();<a name="line.561"></a>
<span class="sourceLineNo">562</span> if (value != 0) {<a name="line.562"></a>
<span class="sourceLineNo">563</span> LOG.info(c + ": " + value);<a name="line.563"></a>
<span class="sourceLineNo">564</span> }<a name="line.564"></a>
<span class="sourceLineNo">565</span> }<a name="line.565"></a>
<span class="sourceLineNo">566</span><a name="line.566"></a>
<span class="sourceLineNo">567</span> return success ? 0 : 1;<a name="line.567"></a>
<span class="sourceLineNo">568</span> }<a name="line.568"></a>
<span class="sourceLineNo">569</span><a name="line.569"></a>
<span class="sourceLineNo">570</span> @Override<a name="line.570"></a>
<span class="sourceLineNo">571</span> public int run(String[] args) throws Exception {<a name="line.571"></a>
<span class="sourceLineNo">572</span> if (args.length &lt; 2) {<a name="line.572"></a>
<span class="sourceLineNo">573</span> System.err.println(USAGE);<a name="line.573"></a>
<span class="sourceLineNo">574</span> return 1;<a name="line.574"></a>
<span class="sourceLineNo">575</span> }<a name="line.575"></a>
<span class="sourceLineNo">576</span> try {<a name="line.576"></a>
<span class="sourceLineNo">577</span> Path warcFileInput = new Path(args[0]);<a name="line.577"></a>
<span class="sourceLineNo">578</span> Path outputDir = new Path(args[1]);<a name="line.578"></a>
<span class="sourceLineNo">579</span> return run(warcFileInput, outputDir);<a name="line.579"></a>
<span class="sourceLineNo">580</span> } catch (NumberFormatException e) {<a name="line.580"></a>
<span class="sourceLineNo">581</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.581"></a>
<span class="sourceLineNo">582</span> System.err.println(USAGE);<a name="line.582"></a>
<span class="sourceLineNo">583</span> return 1;<a name="line.583"></a>
<span class="sourceLineNo">584</span> }<a name="line.584"></a>
<span class="sourceLineNo">585</span> }<a name="line.585"></a>
<span class="sourceLineNo">586</span><a name="line.586"></a>
<span class="sourceLineNo">587</span> public static void main(String[] args) throws Exception {<a name="line.587"></a>
<span class="sourceLineNo">588</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.588"></a>
<span class="sourceLineNo">589</span> }<a name="line.589"></a>
<span class="sourceLineNo">590</span><a name="line.590"></a>
<span class="sourceLineNo">591</span> public static class LoaderMapper<a name="line.591"></a>
<span class="sourceLineNo">592</span> extends Mapper&lt;LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable&gt; {<a name="line.592"></a>
<span class="sourceLineNo">593</span><a name="line.593"></a>
<span class="sourceLineNo">594</span> protected AsyncConnection conn;<a name="line.594"></a>
<span class="sourceLineNo">595</span> protected AsyncTable&lt;ScanResultConsumer&gt; table;<a name="line.595"></a>
<span class="sourceLineNo">596</span> protected ExecutorService executor;<a name="line.596"></a>
<span class="sourceLineNo">597</span> protected AtomicLong inflight = new AtomicLong();<a name="line.597"></a>
<span class="sourceLineNo">598</span> protected boolean doIncrements;<a name="line.598"></a>
<span class="sourceLineNo">599</span><a name="line.599"></a>
<span class="sourceLineNo">600</span> @Override<a name="line.600"></a>
<span class="sourceLineNo">601</span> protected void setup(final Context context) throws IOException, InterruptedException {<a name="line.601"></a>
<span class="sourceLineNo">602</span> executor = Executors.newWorkStealingPool(Runtime.getRuntime().availableProcessors());<a name="line.602"></a>
<span class="sourceLineNo">603</span> Configuration conf = context.getConfiguration();<a name="line.603"></a>
<span class="sourceLineNo">604</span> doIncrements = conf.getBoolean(INCREMENTS_NAME_KEY, DEFAULT_INCREMENTS);<a name="line.604"></a>
<span class="sourceLineNo">605</span> try {<a name="line.605"></a>
<span class="sourceLineNo">606</span> conn = ConnectionFactory.createAsyncConnection(conf).get();<a name="line.606"></a>
<span class="sourceLineNo">607</span> table = conn.getTable(getTablename(conf), executor);<a name="line.607"></a>
<span class="sourceLineNo">608</span> } catch (ExecutionException e) {<a name="line.608"></a>
<span class="sourceLineNo">609</span> throw new IOException(e);<a name="line.609"></a>
<span class="sourceLineNo">610</span> }<a name="line.610"></a>
<span class="sourceLineNo">611</span> }<a name="line.611"></a>
<span class="sourceLineNo">612</span><a name="line.612"></a>
<span class="sourceLineNo">613</span> @Override<a name="line.613"></a>
<span class="sourceLineNo">614</span> protected void cleanup(final Context context) throws IOException, InterruptedException {<a name="line.614"></a>
<span class="sourceLineNo">615</span><a name="line.615"></a>
<span class="sourceLineNo">616</span> while (inflight.get() != 0) {<a name="line.616"></a>
<span class="sourceLineNo">617</span> LOG.info("Operations in flight, waiting");<a name="line.617"></a>
<span class="sourceLineNo">618</span> Thread.sleep(INFLIGHT_PAUSE_MS);<a name="line.618"></a>
<span class="sourceLineNo">619</span> }<a name="line.619"></a>
<span class="sourceLineNo">620</span><a name="line.620"></a>
<span class="sourceLineNo">621</span> // Shut down the executor<a name="line.621"></a>
<span class="sourceLineNo">622</span> executor.shutdown();<a name="line.622"></a>
<span class="sourceLineNo">623</span> if (!executor.awaitTermination(1, TimeUnit.MINUTES)) {<a name="line.623"></a>
<span class="sourceLineNo">624</span> LOG.warn("Pool did not shut down cleanly");<a name="line.624"></a>
<span class="sourceLineNo">625</span> }<a name="line.625"></a>
<span class="sourceLineNo">626</span> // Close the connection<a name="line.626"></a>
<span class="sourceLineNo">627</span> try {<a name="line.627"></a>
<span class="sourceLineNo">628</span> conn.close();<a name="line.628"></a>
<span class="sourceLineNo">629</span> } catch (Exception e) {<a name="line.629"></a>
<span class="sourceLineNo">630</span> LOG.warn("Exception closing Connection", e);<a name="line.630"></a>
<span class="sourceLineNo">631</span> }<a name="line.631"></a>
<span class="sourceLineNo">632</span> }<a name="line.632"></a>
<span class="sourceLineNo">633</span><a name="line.633"></a>
<span class="sourceLineNo">634</span> @Override<a name="line.634"></a>
<span class="sourceLineNo">635</span> protected void map(final LongWritable key, final WARCWritable value, final Context output)<a name="line.635"></a>
<span class="sourceLineNo">636</span> throws IOException, InterruptedException {<a name="line.636"></a>
<span class="sourceLineNo">637</span> final WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.637"></a>
<span class="sourceLineNo">638</span> final String recordID = warcHeader.getRecordID();<a name="line.638"></a>
<span class="sourceLineNo">639</span> final String targetURI = warcHeader.getTargetURI();<a name="line.639"></a>
<span class="sourceLineNo">640</span> if (warcHeader.getRecordType().equals("response") &amp;&amp; targetURI != null) {<a name="line.640"></a>
<span class="sourceLineNo">641</span> final String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.641"></a>
<span class="sourceLineNo">642</span> if (contentType != null) {<a name="line.642"></a>
<span class="sourceLineNo">643</span> // Make row key<a name="line.643"></a>
<span class="sourceLineNo">644</span> byte[] rowKey;<a name="line.644"></a>
<span class="sourceLineNo">645</span> try {<a name="line.645"></a>
<span class="sourceLineNo">646</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.646"></a>
<span class="sourceLineNo">647</span> } catch (IllegalArgumentException e) {<a name="line.647"></a>
<span class="sourceLineNo">648</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.648"></a>
<span class="sourceLineNo">649</span> return;<a name="line.649"></a>
<span class="sourceLineNo">650</span> } catch (URISyntaxException e) {<a name="line.650"></a>
<span class="sourceLineNo">651</span> LOG.warn(<a name="line.651"></a>
<span class="sourceLineNo">652</span> "Could not parse URI \"" + targetURI + "\" for record " + recordID + ", ignoring");<a name="line.652"></a>
<span class="sourceLineNo">653</span> return;<a name="line.653"></a>
<span class="sourceLineNo">654</span> }<a name="line.654"></a>
<span class="sourceLineNo">655</span><a name="line.655"></a>
<span class="sourceLineNo">656</span> // Get the content and calculate the CRC64<a name="line.656"></a>
<span class="sourceLineNo">657</span> final byte[] content = value.getRecord().getContent();<a name="line.657"></a>
<span class="sourceLineNo">658</span> final CRC64 crc = new CRC64();<a name="line.658"></a>
<span class="sourceLineNo">659</span> crc.update(content);<a name="line.659"></a>
<span class="sourceLineNo">660</span> final long crc64 = crc.getValue();<a name="line.660"></a>
<span class="sourceLineNo">661</span> LOG.info("{}: content {} bytes, crc64={}", targetURI, content.length,<a name="line.661"></a>
<span class="sourceLineNo">662</span> Bytes.toHex(Bytes.toBytes(crc64)));<a name="line.662"></a>
<span class="sourceLineNo">663</span><a name="line.663"></a>
<span class="sourceLineNo">664</span> // Store to HBase<a name="line.664"></a>
<span class="sourceLineNo">665</span> final long ts = getSequence();<a name="line.665"></a>
<span class="sourceLineNo">666</span> final Put put = new Put(rowKey);<a name="line.666"></a>
<span class="sourceLineNo">667</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, ts, content);<a name="line.667"></a>
<span class="sourceLineNo">668</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, ts,<a name="line.668"></a>
<span class="sourceLineNo">669</span> Bytes.toBytes(content.length));<a name="line.669"></a>
<span class="sourceLineNo">670</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, ts, Bytes.toBytes(contentType));<a name="line.670"></a>
<span class="sourceLineNo">671</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, ts, Bytes.toBytes(crc64));<a name="line.671"></a>
<span class="sourceLineNo">672</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, ts, Bytes.toBytes(targetURI));<a name="line.672"></a>
<span class="sourceLineNo">673</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, ts,<a name="line.673"></a>
<span class="sourceLineNo">674</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.674"></a>
<span class="sourceLineNo">675</span> final String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.675"></a>
<span class="sourceLineNo">676</span> if (ipAddr != null) {<a name="line.676"></a>
<span class="sourceLineNo">677</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, ts, Bytes.toBytes(ipAddr));<a name="line.677"></a>
<span class="sourceLineNo">678</span> }<a name="line.678"></a>
<span class="sourceLineNo">679</span> long pending = inflight.incrementAndGet();<a name="line.679"></a>
<span class="sourceLineNo">680</span> while (pending &gt; MAX_INFLIGHT) {<a name="line.680"></a>
<span class="sourceLineNo">681</span> LOG.info("Too many operations in flight, waiting");<a name="line.681"></a>
<span class="sourceLineNo">682</span> Thread.sleep(INFLIGHT_PAUSE_MS);<a name="line.682"></a>
<span class="sourceLineNo">683</span> pending = inflight.get();<a name="line.683"></a>
<span class="sourceLineNo">684</span> }<a name="line.684"></a>
<span class="sourceLineNo">685</span> final long putStartTime = System.currentTimeMillis();<a name="line.685"></a>
<span class="sourceLineNo">686</span> final CompletableFuture&lt;Void&gt; putFuture = table.put(put);<a name="line.686"></a>
<span class="sourceLineNo">687</span> putFuture.thenRun(() -&gt; {<a name="line.687"></a>
<span class="sourceLineNo">688</span> inflight.decrementAndGet();<a name="line.688"></a>
<span class="sourceLineNo">689</span> if (!putFuture.isCompletedExceptionally()) {<a name="line.689"></a>
<span class="sourceLineNo">690</span> output.getCounter(Counts.RPC_TIME_MS)<a name="line.690"></a>
<span class="sourceLineNo">691</span> .increment(System.currentTimeMillis() - putStartTime);<a name="line.691"></a>
<span class="sourceLineNo">692</span> output.getCounter(Counts.RPC_BYTES_WRITTEN).increment(put.heapSize());<a name="line.692"></a>
<span class="sourceLineNo">693</span> }<a name="line.693"></a>
<span class="sourceLineNo">694</span> });<a name="line.694"></a>
<span class="sourceLineNo">695</span><a name="line.695"></a>
<span class="sourceLineNo">696</span> // Write records out for later verification, one per HBase field except for the<a name="line.696"></a>
<span class="sourceLineNo">697</span> // content record, which will be verified by CRC64.<a name="line.697"></a>
<span class="sourceLineNo">698</span> output.write(<a name="line.698"></a>
<span class="sourceLineNo">699</span> new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, ts),<a name="line.699"></a>
<span class="sourceLineNo">700</span> new BytesWritable(Bytes.toBytes(content.length)));<a name="line.700"></a>
<span class="sourceLineNo">701</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, ts),<a name="line.701"></a>
<span class="sourceLineNo">702</span> new BytesWritable(Bytes.toBytes(contentType)));<a name="line.702"></a>
<span class="sourceLineNo">703</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, ts),<a name="line.703"></a>
<span class="sourceLineNo">704</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.704"></a>
<span class="sourceLineNo">705</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, ts),<a name="line.705"></a>
<span class="sourceLineNo">706</span> new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.706"></a>
<span class="sourceLineNo">707</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, ts),<a name="line.707"></a>
<span class="sourceLineNo">708</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.708"></a>
<span class="sourceLineNo">709</span> if (ipAddr != null) {<a name="line.709"></a>
<span class="sourceLineNo">710</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, ts),<a name="line.710"></a>
<span class="sourceLineNo">711</span> new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.711"></a>
<span class="sourceLineNo">712</span> }<a name="line.712"></a>
<span class="sourceLineNo">713</span><a name="line.713"></a>
<span class="sourceLineNo">714</span> if (doIncrements) {<a name="line.714"></a>
<span class="sourceLineNo">715</span> // The URLs cf is not tracked for correctness. For now it is used only to exercise<a name="line.715"></a>
<span class="sourceLineNo">716</span> // Increments, to drive some read load during ingest. They can be verified with a<a name="line.716"></a>
<span class="sourceLineNo">717</span> // reducer to sum increments per row and then compare the final count to the table<a name="line.717"></a>
<span class="sourceLineNo">718</span> // data. This is left as a future exercise.<a name="line.718"></a>
<span class="sourceLineNo">719</span> final byte[] refQual = Bytes.add(REF_QUALIFIER, SEP, rowKey);<a name="line.719"></a>
<span class="sourceLineNo">720</span> for (String refUri : extractUrls(content)) {<a name="line.720"></a>
<span class="sourceLineNo">721</span> try {<a name="line.721"></a>
<span class="sourceLineNo">722</span> byte[] urlRowKey = rowKeyFromTargetURI(refUri);<a name="line.722"></a>
<span class="sourceLineNo">723</span> LOG.debug(" -&gt; {}", refUri);<a name="line.723"></a>
<span class="sourceLineNo">724</span> final Increment increment = new Increment(urlRowKey);<a name="line.724"></a>
<span class="sourceLineNo">725</span> increment.setTimestamp(ts);<a name="line.725"></a>
<span class="sourceLineNo">726</span> increment.addColumn(URL_FAMILY_NAME, refQual, 1);<a name="line.726"></a>
<span class="sourceLineNo">727</span> pending = inflight.incrementAndGet();<a name="line.727"></a>
<span class="sourceLineNo">728</span> while (pending &gt; MAX_INFLIGHT) {<a name="line.728"></a>
<span class="sourceLineNo">729</span> LOG.info("Too many operations in flight, waiting");<a name="line.729"></a>
<span class="sourceLineNo">730</span> Thread.sleep(INFLIGHT_PAUSE_MS);<a name="line.730"></a>
<span class="sourceLineNo">731</span> pending = inflight.get();<a name="line.731"></a>
<span class="sourceLineNo">732</span> }<a name="line.732"></a>
<span class="sourceLineNo">733</span> final long incrStartTime = System.currentTimeMillis();<a name="line.733"></a>
<span class="sourceLineNo">734</span> final CompletableFuture&lt;Result&gt; incrFuture = table.increment(increment);<a name="line.734"></a>
<span class="sourceLineNo">735</span> incrFuture.thenRun(() -&gt; {<a name="line.735"></a>
<span class="sourceLineNo">736</span> inflight.decrementAndGet();<a name="line.736"></a>
<span class="sourceLineNo">737</span> if (!incrFuture.isCompletedExceptionally()) {<a name="line.737"></a>
<span class="sourceLineNo">738</span> output.getCounter(Counts.RPC_TIME_MS)<a name="line.738"></a>
<span class="sourceLineNo">739</span> .increment(System.currentTimeMillis() - incrStartTime);<a name="line.739"></a>
<span class="sourceLineNo">740</span> output.getCounter(Counts.RPC_BYTES_WRITTEN).increment(increment.heapSize());<a name="line.740"></a>
<span class="sourceLineNo">741</span> }<a name="line.741"></a>
<span class="sourceLineNo">742</span> });<a name="line.742"></a>
<span class="sourceLineNo">743</span> } catch (IllegalArgumentException | URISyntaxException e) {<a name="line.743"></a>
<span class="sourceLineNo">744</span> LOG.debug("Could not make a row key for URI " + refUri + ", ignoring", e);<a name="line.744"></a>
<span class="sourceLineNo">745</span> }<a name="line.745"></a>
<span class="sourceLineNo">746</span> }<a name="line.746"></a>
<span class="sourceLineNo">747</span> }<a name="line.747"></a>
<span class="sourceLineNo">748</span> }<a name="line.748"></a>
<span class="sourceLineNo">749</span> }<a name="line.749"></a>
<span class="sourceLineNo">750</span> }<a name="line.750"></a>
<span class="sourceLineNo">751</span> }<a name="line.751"></a>
<span class="sourceLineNo">752</span> }<a name="line.752"></a>
<span class="sourceLineNo">753</span><a name="line.753"></a>
<span class="sourceLineNo">754</span> public static class OneFilePerMapperSFIF&lt;K, V&gt; extends SequenceFileInputFormat&lt;K, V&gt; {<a name="line.754"></a>
<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
<span class="sourceLineNo">756</span> protected boolean isSplitable(final JobContext context, final Path filename) {<a name="line.756"></a>
<span class="sourceLineNo">757</span> return false;<a name="line.757"></a>
<span class="sourceLineNo">758</span> }<a name="line.758"></a>
<span class="sourceLineNo">759</span> }<a name="line.759"></a>
<span class="sourceLineNo">760</span><a name="line.760"></a>
<span class="sourceLineNo">761</span> public static class Verify extends Configured implements Tool {<a name="line.761"></a>
<span class="sourceLineNo">762</span><a name="line.762"></a>
<span class="sourceLineNo">763</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.763"></a>
<span class="sourceLineNo">764</span> public static final String USAGE = "Verify &lt;inputDir&gt;";<a name="line.764"></a>
<span class="sourceLineNo">765</span><a name="line.765"></a>
<span class="sourceLineNo">766</span> int run(final Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.766"></a>
<span class="sourceLineNo">767</span> Job job = Job.getInstance(getConf());<a name="line.767"></a>
<span class="sourceLineNo">768</span> job.setJobName(Verify.class.getName());<a name="line.768"></a>
<span class="sourceLineNo">769</span> job.setJarByClass(getClass());<a name="line.769"></a>
<span class="sourceLineNo">770</span> job.setMapperClass(VerifyMapper.class);<a name="line.770"></a>
<span class="sourceLineNo">771</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.771"></a>
<span class="sourceLineNo">772</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.772"></a>
<span class="sourceLineNo">773</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.773"></a>
<span class="sourceLineNo">774</span> job.setOutputKeyClass(NullWritable.class);<a name="line.774"></a>
<span class="sourceLineNo">775</span> job.setOutputValueClass(NullWritable.class);<a name="line.775"></a>
<span class="sourceLineNo">776</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.776"></a>
<span class="sourceLineNo">777</span><a name="line.777"></a>
<span class="sourceLineNo">778</span> boolean success = job.waitForCompletion(true);<a name="line.778"></a>
<span class="sourceLineNo">779</span> if (!success) {<a name="line.779"></a>
<span class="sourceLineNo">780</span> LOG.error("Failure during job " + job.getJobID());<a name="line.780"></a>
<span class="sourceLineNo">781</span> }<a name="line.781"></a>
<span class="sourceLineNo">782</span><a name="line.782"></a>
<span class="sourceLineNo">783</span> final Counters counters = job.getCounters();<a name="line.783"></a>
<span class="sourceLineNo">784</span> for (Counts c : Counts.values()) {<a name="line.784"></a>
<span class="sourceLineNo">785</span> long value = counters.findCounter(c).getValue();<a name="line.785"></a>
<span class="sourceLineNo">786</span> if (value != 0) {<a name="line.786"></a>
<span class="sourceLineNo">787</span> LOG.info(c + ": " + value);<a name="line.787"></a>
<span class="sourceLineNo">788</span> }<a name="line.788"></a>
<span class="sourceLineNo">789</span> }<a name="line.789"></a>
<span class="sourceLineNo">790</span> if (counters.findCounter(Counts.UNREFERENCED).getValue() &gt; 0) {<a name="line.790"></a>
<span class="sourceLineNo">791</span> LOG.error("Nonzero UNREFERENCED count from job " + job.getJobID());<a name="line.791"></a>
<span class="sourceLineNo">792</span> success = false;<a name="line.792"></a>
<span class="sourceLineNo">793</span> }<a name="line.793"></a>
<span class="sourceLineNo">794</span> if (counters.findCounter(Counts.CORRUPT).getValue() &gt; 0) {<a name="line.794"></a>
<span class="sourceLineNo">795</span> LOG.error("Nonzero CORRUPT count from job " + job.getJobID());<a name="line.795"></a>
<span class="sourceLineNo">796</span> success = false;<a name="line.796"></a>
<span class="sourceLineNo">797</span> }<a name="line.797"></a>
<span class="sourceLineNo">798</span><a name="line.798"></a>
<span class="sourceLineNo">799</span> return success ? 0 : 1;<a name="line.799"></a>
<span class="sourceLineNo">800</span> }<a name="line.800"></a>
<span class="sourceLineNo">801</span><a name="line.801"></a>
<span class="sourceLineNo">802</span> @Override<a name="line.802"></a>
<span class="sourceLineNo">803</span> public int run(String[] args) throws Exception {<a name="line.803"></a>
<span class="sourceLineNo">804</span> if (args.length &lt; 2) {<a name="line.804"></a>
<span class="sourceLineNo">805</span> System.err.println(USAGE);<a name="line.805"></a>
<span class="sourceLineNo">806</span> return 1;<a name="line.806"></a>
<span class="sourceLineNo">807</span> }<a name="line.807"></a>
<span class="sourceLineNo">808</span> Path loaderOutput = new Path(args[0]);<a name="line.808"></a>
<span class="sourceLineNo">809</span> return run(loaderOutput);<a name="line.809"></a>
<span class="sourceLineNo">810</span> }<a name="line.810"></a>
<span class="sourceLineNo">811</span><a name="line.811"></a>
<span class="sourceLineNo">812</span> public static void main(String[] args) throws Exception {<a name="line.812"></a>
<span class="sourceLineNo">813</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.813"></a>
<span class="sourceLineNo">814</span> }<a name="line.814"></a>
<span class="sourceLineNo">815</span><a name="line.815"></a>
<span class="sourceLineNo">816</span> public static class VerifyMapper<a name="line.816"></a>
<span class="sourceLineNo">817</span> extends Mapper&lt;HBaseKeyWritable, BytesWritable, NullWritable, NullWritable&gt; {<a name="line.817"></a>
<span class="sourceLineNo">818</span><a name="line.818"></a>
<span class="sourceLineNo">819</span> protected Connection conn;<a name="line.819"></a>
<span class="sourceLineNo">820</span> protected Table table;<a name="line.820"></a>
<span class="sourceLineNo">821</span><a name="line.821"></a>
<span class="sourceLineNo">822</span> @Override<a name="line.822"></a>
<span class="sourceLineNo">823</span> protected void setup(final Context context) throws IOException, InterruptedException {<a name="line.823"></a>
<span class="sourceLineNo">824</span> Configuration conf = context.getConfiguration();<a name="line.824"></a>
<span class="sourceLineNo">825</span> conn = ConnectionFactory.createConnection(conf);<a name="line.825"></a>
<span class="sourceLineNo">826</span> table = conn.getTable(getTablename(conf));<a name="line.826"></a>
<span class="sourceLineNo">827</span> }<a name="line.827"></a>
<span class="sourceLineNo">828</span><a name="line.828"></a>
<span class="sourceLineNo">829</span> @Override<a name="line.829"></a>
<span class="sourceLineNo">830</span> protected void cleanup(final Context context) throws IOException, InterruptedException {<a name="line.830"></a>
<span class="sourceLineNo">831</span> // Close the table<a name="line.831"></a>
<span class="sourceLineNo">832</span> try {<a name="line.832"></a>
<span class="sourceLineNo">833</span> table.close();<a name="line.833"></a>
<span class="sourceLineNo">834</span> } catch (Exception e) {<a name="line.834"></a>
<span class="sourceLineNo">835</span> LOG.warn("Exception closing table", e);<a name="line.835"></a>
<span class="sourceLineNo">836</span> }<a name="line.836"></a>
<span class="sourceLineNo">837</span> // Close the connection<a name="line.837"></a>
<span class="sourceLineNo">838</span> try {<a name="line.838"></a>
<span class="sourceLineNo">839</span> conn.close();<a name="line.839"></a>
<span class="sourceLineNo">840</span> } catch (Exception e) {<a name="line.840"></a>
<span class="sourceLineNo">841</span> LOG.warn("Exception closing Connection", e);<a name="line.841"></a>
<span class="sourceLineNo">842</span> }<a name="line.842"></a>
<span class="sourceLineNo">843</span> }<a name="line.843"></a>
<span class="sourceLineNo">844</span><a name="line.844"></a>
<span class="sourceLineNo">845</span> @Override<a name="line.845"></a>
<span class="sourceLineNo">846</span> protected void map(final HBaseKeyWritable key, final BytesWritable value,<a name="line.846"></a>
<span class="sourceLineNo">847</span> final Context output) throws IOException, InterruptedException {<a name="line.847"></a>
<span class="sourceLineNo">848</span> final byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.848"></a>
<span class="sourceLineNo">849</span> final byte[] family =<a name="line.849"></a>
<span class="sourceLineNo">850</span> Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(), key.getFamilyLength());<a name="line.850"></a>
<span class="sourceLineNo">851</span> final byte[] qualifier =<a name="line.851"></a>
<span class="sourceLineNo">852</span> Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(), key.getQualifierLength());<a name="line.852"></a>
<span class="sourceLineNo">853</span> final long ts = key.getTimestamp();<a name="line.853"></a>
<span class="sourceLineNo">854</span><a name="line.854"></a>
<span class="sourceLineNo">855</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &amp;&amp; Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.855"></a>
<span class="sourceLineNo">856</span> final long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.856"></a>
<span class="sourceLineNo">857</span> final Get get = new Get(row).setTimestamp(ts).addFamily(CONTENT_FAMILY_NAME)<a name="line.857"></a>
<span class="sourceLineNo">858</span> .addFamily(INFO_FAMILY_NAME);<a name="line.858"></a>
<span class="sourceLineNo">859</span> final long startTime = System.currentTimeMillis();<a name="line.859"></a>
<span class="sourceLineNo">860</span> Result r;<a name="line.860"></a>
<span class="sourceLineNo">861</span> try {<a name="line.861"></a>
<span class="sourceLineNo">862</span> r = table.get(get);<a name="line.862"></a>
<span class="sourceLineNo">863</span> output.getCounter(Counts.RPC_TIME_MS).increment(System.currentTimeMillis() - startTime);<a name="line.863"></a>
<span class="sourceLineNo">864</span> } catch (Exception e) {<a name="line.864"></a>
<span class="sourceLineNo">865</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": exception", e);<a name="line.865"></a>
<span class="sourceLineNo">866</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.866"></a>
<span class="sourceLineNo">867</span> return;<a name="line.867"></a>
<span class="sourceLineNo">868</span> }<a name="line.868"></a>
<span class="sourceLineNo">869</span> final byte[] crcBytes = r.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.869"></a>
<span class="sourceLineNo">870</span> if (crcBytes == null) {<a name="line.870"></a>
<span class="sourceLineNo">871</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.871"></a>
<span class="sourceLineNo">872</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.872"></a>
<span class="sourceLineNo">873</span> return;<a name="line.873"></a>
<span class="sourceLineNo">874</span> }<a name="line.874"></a>
<span class="sourceLineNo">875</span> if (Bytes.toLong(crcBytes) != expectedCRC64) {<a name="line.875"></a>
<span class="sourceLineNo">876</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.876"></a>
<span class="sourceLineNo">877</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.877"></a>
<span class="sourceLineNo">878</span> return;<a name="line.878"></a>
<span class="sourceLineNo">879</span> }<a name="line.879"></a>
<span class="sourceLineNo">880</span> // If we fell through to here all verification checks have succeeded for the info<a name="line.880"></a>
<span class="sourceLineNo">881</span> // record.<a name="line.881"></a>
<span class="sourceLineNo">882</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.882"></a>
<span class="sourceLineNo">883</span> final byte[] content = r.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.883"></a>
<span class="sourceLineNo">884</span> if (content == null) {<a name="line.884"></a>
<span class="sourceLineNo">885</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.885"></a>
<span class="sourceLineNo">886</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.886"></a>
<span class="sourceLineNo">887</span> return;<a name="line.887"></a>
<span class="sourceLineNo">888</span> } else {<a name="line.888"></a>
<span class="sourceLineNo">889</span> final CRC64 crc = new CRC64();<a name="line.889"></a>
<span class="sourceLineNo">890</span> crc.update(content);<a name="line.890"></a>
<span class="sourceLineNo">891</span> if (crc.getValue() != expectedCRC64) {<a name="line.891"></a>
<span class="sourceLineNo">892</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.892"></a>
<span class="sourceLineNo">893</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.893"></a>
<span class="sourceLineNo">894</span> return;<a name="line.894"></a>
<span class="sourceLineNo">895</span> }<a name="line.895"></a>
<span class="sourceLineNo">896</span> }<a name="line.896"></a>
<span class="sourceLineNo">897</span> // If we fell through to here all verification checks have succeeded for the content<a name="line.897"></a>
<span class="sourceLineNo">898</span> // record.<a name="line.898"></a>
<span class="sourceLineNo">899</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.899"></a>
<span class="sourceLineNo">900</span> } else {<a name="line.900"></a>
<span class="sourceLineNo">901</span> final long startTime = System.currentTimeMillis();<a name="line.901"></a>
<span class="sourceLineNo">902</span> final Get get = new Get(row).setTimestamp(ts).addColumn(family, qualifier);<a name="line.902"></a>
<span class="sourceLineNo">903</span> Result r;<a name="line.903"></a>
<span class="sourceLineNo">904</span> try {<a name="line.904"></a>
<span class="sourceLineNo">905</span> r = table.get(get);<a name="line.905"></a>
<span class="sourceLineNo">906</span> output.getCounter(Counts.RPC_TIME_MS).increment(System.currentTimeMillis() - startTime);<a name="line.906"></a>
<span class="sourceLineNo">907</span> } catch (Exception e) {<a name="line.907"></a>
<span class="sourceLineNo">908</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": exception", e);<a name="line.908"></a>
<span class="sourceLineNo">909</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.909"></a>
<span class="sourceLineNo">910</span> return;<a name="line.910"></a>
<span class="sourceLineNo">911</span> }<a name="line.911"></a>
<span class="sourceLineNo">912</span> final byte[] bytes = r.getValue(family, qualifier);<a name="line.912"></a>
<span class="sourceLineNo">913</span> if (bytes == null) {<a name="line.913"></a>
<span class="sourceLineNo">914</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing "<a name="line.914"></a>
<span class="sourceLineNo">915</span> + Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.915"></a>
<span class="sourceLineNo">916</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.916"></a>
<span class="sourceLineNo">917</span> return;<a name="line.917"></a>
<span class="sourceLineNo">918</span> }<a name="line.918"></a>
<span class="sourceLineNo">919</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.919"></a>
<span class="sourceLineNo">920</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " + Bytes.toStringBinary(family) + ":"<a name="line.920"></a>
<span class="sourceLineNo">921</span> + Bytes.toStringBinary(qualifier) + " mismatch");<a name="line.921"></a>
<span class="sourceLineNo">922</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.922"></a>
<span class="sourceLineNo">923</span> return;<a name="line.923"></a>
<span class="sourceLineNo">924</span> }<a name="line.924"></a>
<span class="sourceLineNo">925</span> // If we fell through to here all verification checks have succeeded for the info<a name="line.925"></a>
<span class="sourceLineNo">926</span> // record.<a name="line.926"></a>
<span class="sourceLineNo">927</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.927"></a>
<span class="sourceLineNo">928</span> }<a name="line.928"></a>
<span class="sourceLineNo">929</span> }<a name="line.929"></a>
<span class="sourceLineNo">930</span> }<a name="line.930"></a>
<span class="sourceLineNo">931</span> }<a name="line.931"></a>
<span class="sourceLineNo">932</span><a name="line.932"></a>
<span class="sourceLineNo">933</span> private static final AtomicLong counter = new AtomicLong();<a name="line.933"></a>
<span class="sourceLineNo">934</span> private static final int shift = 8;<a name="line.934"></a>
<span class="sourceLineNo">935</span><a name="line.935"></a>
<span class="sourceLineNo">936</span> private static long getSequence() {<a name="line.936"></a>
<span class="sourceLineNo">937</span> long t = EnvironmentEdgeManager.currentTime();<a name="line.937"></a>
<span class="sourceLineNo">938</span> t &lt;&lt;= shift;<a name="line.938"></a>
<span class="sourceLineNo">939</span> t |= (counter.getAndIncrement() % (1 &lt;&lt; shift));<a name="line.939"></a>
<span class="sourceLineNo">940</span> return t;<a name="line.940"></a>
<span class="sourceLineNo">941</span> }<a name="line.941"></a>
<span class="sourceLineNo">942</span><a name="line.942"></a>
<span class="sourceLineNo">943</span> private static byte[] rowKeyFromTargetURI(final String targetUri)<a name="line.943"></a>
<span class="sourceLineNo">944</span> throws IOException, URISyntaxException, IllegalArgumentException {<a name="line.944"></a>
<span class="sourceLineNo">945</span> final URI uri = new URI(targetUri);<a name="line.945"></a>
<span class="sourceLineNo">946</span> // Ignore the scheme<a name="line.946"></a>
<span class="sourceLineNo">947</span> // Reverse the components of the hostname<a name="line.947"></a>
<span class="sourceLineNo">948</span> String reversedHost;<a name="line.948"></a>
<span class="sourceLineNo">949</span> if (uri.getHost() != null) {<a name="line.949"></a>
<span class="sourceLineNo">950</span> final StringBuilder sb = new StringBuilder();<a name="line.950"></a>
<span class="sourceLineNo">951</span> final String[] hostComponents = uri.getHost().split("\\.");<a name="line.951"></a>
<span class="sourceLineNo">952</span> for (int i = hostComponents.length - 1; i &gt;= 0; i--) {<a name="line.952"></a>
<span class="sourceLineNo">953</span> sb.append(hostComponents[i]);<a name="line.953"></a>
<span class="sourceLineNo">954</span> if (i != 0) {<a name="line.954"></a>
<span class="sourceLineNo">955</span> sb.append('.');<a name="line.955"></a>
<span class="sourceLineNo">956</span> }<a name="line.956"></a>
<span class="sourceLineNo">957</span> }<a name="line.957"></a>
<span class="sourceLineNo">958</span> reversedHost = sb.toString();<a name="line.958"></a>
<span class="sourceLineNo">959</span> } else {<a name="line.959"></a>
<span class="sourceLineNo">960</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.960"></a>
<span class="sourceLineNo">961</span> }<a name="line.961"></a>
<span class="sourceLineNo">962</span> final ByteArrayOutputStream os = new ByteArrayOutputStream();<a name="line.962"></a>
<span class="sourceLineNo">963</span> os.write(reversedHost.getBytes(StandardCharsets.UTF_8));<a name="line.963"></a>
<span class="sourceLineNo">964</span> if (uri.getPort() &gt;= 0) {<a name="line.964"></a>
<span class="sourceLineNo">965</span> os.write(String.format(":%d", uri.getPort()).getBytes(StandardCharsets.UTF_8));<a name="line.965"></a>
<span class="sourceLineNo">966</span> }<a name="line.966"></a>
<span class="sourceLineNo">967</span> os.write((byte) '|');<a name="line.967"></a>
<span class="sourceLineNo">968</span> if (uri.getPath() != null) {<a name="line.968"></a>
<span class="sourceLineNo">969</span> os.write(uri.getPath().getBytes(StandardCharsets.UTF_8));<a name="line.969"></a>
<span class="sourceLineNo">970</span> }<a name="line.970"></a>
<span class="sourceLineNo">971</span> if (uri.getQuery() != null) {<a name="line.971"></a>
<span class="sourceLineNo">972</span> os.write(String.format("?%s", uri.getQuery()).getBytes(StandardCharsets.UTF_8));<a name="line.972"></a>
<span class="sourceLineNo">973</span> }<a name="line.973"></a>
<span class="sourceLineNo">974</span> if (uri.getFragment() != null) {<a name="line.974"></a>
<span class="sourceLineNo">975</span> os.write(String.format("#%s", uri.getFragment()).getBytes(StandardCharsets.UTF_8));<a name="line.975"></a>
<span class="sourceLineNo">976</span> }<a name="line.976"></a>
<span class="sourceLineNo">977</span> if (os.size() &gt; HConstants.MAX_ROW_LENGTH) {<a name="line.977"></a>
<span class="sourceLineNo">978</span> throw new IllegalArgumentException(<a name="line.978"></a>
<span class="sourceLineNo">979</span> "Key would be too large (length=" + os.size() + ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.979"></a>
<span class="sourceLineNo">980</span> }<a name="line.980"></a>
<span class="sourceLineNo">981</span> return os.toByteArray();<a name="line.981"></a>
<span class="sourceLineNo">982</span> }<a name="line.982"></a>
<span class="sourceLineNo">983</span><a name="line.983"></a>
<span class="sourceLineNo">984</span> static final Pattern URL_PATTERN = Pattern.compile(<a name="line.984"></a>
<span class="sourceLineNo">985</span> "\\b((https?|ftp|file)://|(www|ftp)\\.)" + "[\\-A-Z0-9+&amp;@#/%?=~_|$!:,\\.;]*[A-Z0-9+&amp;@#/%=~_|$]",<a name="line.985"></a>
<span class="sourceLineNo">986</span> Pattern.CASE_INSENSITIVE);<a name="line.986"></a>
<span class="sourceLineNo">987</span><a name="line.987"></a>
<span class="sourceLineNo">988</span> private static Collection&lt;String&gt; extractUrls(byte[] content) {<a name="line.988"></a>
<span class="sourceLineNo">989</span> final Set&lt;String&gt; list = new HashSet&lt;&gt;(); // uniques<a name="line.989"></a>
<span class="sourceLineNo">990</span> final Matcher m = URL_PATTERN.matcher(new String(content, StandardCharsets.UTF_8));<a name="line.990"></a>
<span class="sourceLineNo">991</span> while (m.find()) {<a name="line.991"></a>
<span class="sourceLineNo">992</span> list.add(m.group());<a name="line.992"></a>
<span class="sourceLineNo">993</span> }<a name="line.993"></a>
<span class="sourceLineNo">994</span> return list;<a name="line.994"></a>
<span class="sourceLineNo">995</span> }<a name="line.995"></a>
<span class="sourceLineNo">996</span><a name="line.996"></a>
<span class="sourceLineNo">997</span>}<a name="line.997"></a>
</pre>
</div>
</body>
</html>