http://git-wip-us.apache.org/repos/asf/hbase-site/blob/9fb0764b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ImportTsv.html ---------------------------------------------------------------------- diff --git a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ImportTsv.html b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ImportTsv.html index 1d5c1db..84104c6 100644 --- a/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ImportTsv.html +++ b/apidocs/src-html/org/apache/hadoop/hbase/mapreduce/ImportTsv.html @@ -71,732 +71,734 @@ <span class="sourceLineNo">063</span>import com.google.common.base.Preconditions;<a name="line.63"></a> <span class="sourceLineNo">064</span>import com.google.common.base.Splitter;<a name="line.64"></a> <span class="sourceLineNo">065</span>import com.google.common.collect.Lists;<a name="line.65"></a> -<span class="sourceLineNo">066</span><a name="line.66"></a> -<span class="sourceLineNo">067</span>/**<a name="line.67"></a> -<span class="sourceLineNo">068</span> * Tool to import data from a TSV file.<a name="line.68"></a> -<span class="sourceLineNo">069</span> *<a name="line.69"></a> -<span class="sourceLineNo">070</span> * This tool is rather simplistic - it doesn't do any quoting or<a name="line.70"></a> -<span class="sourceLineNo">071</span> * escaping, but is useful for many data loads.<a name="line.71"></a> -<span class="sourceLineNo">072</span> *<a name="line.72"></a> -<span class="sourceLineNo">073</span> * @see ImportTsv#usage(String)<a name="line.73"></a> -<span class="sourceLineNo">074</span> */<a name="line.74"></a> -<span class="sourceLineNo">075</span>@InterfaceAudience.Public<a name="line.75"></a> -<span class="sourceLineNo">076</span>public class ImportTsv extends Configured implements Tool {<a name="line.76"></a> -<span class="sourceLineNo">077</span><a name="line.77"></a> -<span class="sourceLineNo">078</span> protected static final Log LOG = LogFactory.getLog(ImportTsv.class);<a name="line.78"></a> +<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.66"></a> +<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.67"></a> +<span class="sourceLineNo">068</span><a name="line.68"></a> +<span class="sourceLineNo">069</span>/**<a name="line.69"></a> +<span class="sourceLineNo">070</span> * Tool to import data from a TSV file.<a name="line.70"></a> +<span class="sourceLineNo">071</span> *<a name="line.71"></a> +<span class="sourceLineNo">072</span> * This tool is rather simplistic - it doesn't do any quoting or<a name="line.72"></a> +<span class="sourceLineNo">073</span> * escaping, but is useful for many data loads.<a name="line.73"></a> +<span class="sourceLineNo">074</span> *<a name="line.74"></a> +<span class="sourceLineNo">075</span> * @see ImportTsv#usage(String)<a name="line.75"></a> +<span class="sourceLineNo">076</span> */<a name="line.76"></a> +<span class="sourceLineNo">077</span>@InterfaceAudience.Public<a name="line.77"></a> +<span class="sourceLineNo">078</span>public class ImportTsv extends Configured implements Tool {<a name="line.78"></a> <span class="sourceLineNo">079</span><a name="line.79"></a> -<span class="sourceLineNo">080</span> final static String NAME = "importtsv";<a name="line.80"></a> +<span class="sourceLineNo">080</span> protected static final Log LOG = LogFactory.getLog(ImportTsv.class);<a name="line.80"></a> <span class="sourceLineNo">081</span><a name="line.81"></a> -<span class="sourceLineNo">082</span> public final static String MAPPER_CONF_KEY = "importtsv.mapper.class";<a name="line.82"></a> -<span class="sourceLineNo">083</span> public final static String BULK_OUTPUT_CONF_KEY = "importtsv.bulk.output";<a name="line.83"></a> -<span class="sourceLineNo">084</span> public final static String TIMESTAMP_CONF_KEY = "importtsv.timestamp";<a name="line.84"></a> -<span class="sourceLineNo">085</span> public final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.85"></a> -<span class="sourceLineNo">086</span> // TODO: the rest of these configs are used exclusively by TsvImporterMapper.<a name="line.86"></a> -<span class="sourceLineNo">087</span> // Move them out of the tool and let the mapper handle its own validation.<a name="line.87"></a> -<span class="sourceLineNo">088</span> public final static String DRY_RUN_CONF_KEY = "importtsv.dry.run";<a name="line.88"></a> -<span class="sourceLineNo">089</span> // If true, bad lines are logged to stderr. Default: false.<a name="line.89"></a> -<span class="sourceLineNo">090</span> public final static String LOG_BAD_LINES_CONF_KEY = "importtsv.log.bad.lines";<a name="line.90"></a> -<span class="sourceLineNo">091</span> public final static String SKIP_LINES_CONF_KEY = "importtsv.skip.bad.lines";<a name="line.91"></a> -<span class="sourceLineNo">092</span> public final static String SKIP_EMPTY_COLUMNS = "importtsv.skip.empty.columns";<a name="line.92"></a> -<span class="sourceLineNo">093</span> public final static String COLUMNS_CONF_KEY = "importtsv.columns";<a name="line.93"></a> -<span class="sourceLineNo">094</span> public final static String SEPARATOR_CONF_KEY = "importtsv.separator";<a name="line.94"></a> -<span class="sourceLineNo">095</span> public final static String ATTRIBUTE_SEPERATOR_CONF_KEY = "attributes.seperator";<a name="line.95"></a> -<span class="sourceLineNo">096</span> //This config is used to propagate credentials from parent MR jobs which launch<a name="line.96"></a> -<span class="sourceLineNo">097</span> //ImportTSV jobs. SEE IntegrationTestImportTsv.<a name="line.97"></a> -<span class="sourceLineNo">098</span> public final static String CREDENTIALS_LOCATION = "credentials_location";<a name="line.98"></a> -<span class="sourceLineNo">099</span> final static String DEFAULT_SEPARATOR = "\t";<a name="line.99"></a> -<span class="sourceLineNo">100</span> final static String DEFAULT_ATTRIBUTES_SEPERATOR = "=>";<a name="line.100"></a> -<span class="sourceLineNo">101</span> final static String DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR = ",";<a name="line.101"></a> -<span class="sourceLineNo">102</span> final static Class DEFAULT_MAPPER = TsvImporterMapper.class;<a name="line.102"></a> -<span class="sourceLineNo">103</span> public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.103"></a> -<span class="sourceLineNo">104</span> public final static String NO_STRICT_COL_FAMILY = "no.strict";<a name="line.104"></a> -<span class="sourceLineNo">105</span> /**<a name="line.105"></a> -<span class="sourceLineNo">106</span> * If table didn't exist and was created in dry-run mode, this flag is<a name="line.106"></a> -<span class="sourceLineNo">107</span> * flipped to delete it when MR ends.<a name="line.107"></a> -<span class="sourceLineNo">108</span> */<a name="line.108"></a> -<span class="sourceLineNo">109</span> private static boolean DRY_RUN_TABLE_CREATED;<a name="line.109"></a> -<span class="sourceLineNo">110</span><a name="line.110"></a> -<span class="sourceLineNo">111</span> public static class TsvParser {<a name="line.111"></a> -<span class="sourceLineNo">112</span> /**<a name="line.112"></a> -<span class="sourceLineNo">113</span> * Column families and qualifiers mapped to the TSV columns<a name="line.113"></a> -<span class="sourceLineNo">114</span> */<a name="line.114"></a> -<span class="sourceLineNo">115</span> private final byte[][] families;<a name="line.115"></a> -<span class="sourceLineNo">116</span> private final byte[][] qualifiers;<a name="line.116"></a> -<span class="sourceLineNo">117</span><a name="line.117"></a> -<span class="sourceLineNo">118</span> private final byte separatorByte;<a name="line.118"></a> +<span class="sourceLineNo">082</span> final static String NAME = "importtsv";<a name="line.82"></a> +<span class="sourceLineNo">083</span><a name="line.83"></a> +<span class="sourceLineNo">084</span> public final static String MAPPER_CONF_KEY = "importtsv.mapper.class";<a name="line.84"></a> +<span class="sourceLineNo">085</span> public final static String BULK_OUTPUT_CONF_KEY = "importtsv.bulk.output";<a name="line.85"></a> +<span class="sourceLineNo">086</span> public final static String TIMESTAMP_CONF_KEY = "importtsv.timestamp";<a name="line.86"></a> +<span class="sourceLineNo">087</span> public final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.87"></a> +<span class="sourceLineNo">088</span> // TODO: the rest of these configs are used exclusively by TsvImporterMapper.<a name="line.88"></a> +<span class="sourceLineNo">089</span> // Move them out of the tool and let the mapper handle its own validation.<a name="line.89"></a> +<span class="sourceLineNo">090</span> public final static String DRY_RUN_CONF_KEY = "importtsv.dry.run";<a name="line.90"></a> +<span class="sourceLineNo">091</span> // If true, bad lines are logged to stderr. Default: false.<a name="line.91"></a> +<span class="sourceLineNo">092</span> public final static String LOG_BAD_LINES_CONF_KEY = "importtsv.log.bad.lines";<a name="line.92"></a> +<span class="sourceLineNo">093</span> public final static String SKIP_LINES_CONF_KEY = "importtsv.skip.bad.lines";<a name="line.93"></a> +<span class="sourceLineNo">094</span> public final static String SKIP_EMPTY_COLUMNS = "importtsv.skip.empty.columns";<a name="line.94"></a> +<span class="sourceLineNo">095</span> public final static String COLUMNS_CONF_KEY = "importtsv.columns";<a name="line.95"></a> +<span class="sourceLineNo">096</span> public final static String SEPARATOR_CONF_KEY = "importtsv.separator";<a name="line.96"></a> +<span class="sourceLineNo">097</span> public final static String ATTRIBUTE_SEPERATOR_CONF_KEY = "attributes.seperator";<a name="line.97"></a> +<span class="sourceLineNo">098</span> //This config is used to propagate credentials from parent MR jobs which launch<a name="line.98"></a> +<span class="sourceLineNo">099</span> //ImportTSV jobs. SEE IntegrationTestImportTsv.<a name="line.99"></a> +<span class="sourceLineNo">100</span> public final static String CREDENTIALS_LOCATION = "credentials_location";<a name="line.100"></a> +<span class="sourceLineNo">101</span> final static String DEFAULT_SEPARATOR = "\t";<a name="line.101"></a> +<span class="sourceLineNo">102</span> final static String DEFAULT_ATTRIBUTES_SEPERATOR = "=>";<a name="line.102"></a> +<span class="sourceLineNo">103</span> final static String DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR = ",";<a name="line.103"></a> +<span class="sourceLineNo">104</span> final static Class DEFAULT_MAPPER = TsvImporterMapper.class;<a name="line.104"></a> +<span class="sourceLineNo">105</span> public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.105"></a> +<span class="sourceLineNo">106</span> public final static String NO_STRICT_COL_FAMILY = "no.strict";<a name="line.106"></a> +<span class="sourceLineNo">107</span> /**<a name="line.107"></a> +<span class="sourceLineNo">108</span> * If table didn't exist and was created in dry-run mode, this flag is<a name="line.108"></a> +<span class="sourceLineNo">109</span> * flipped to delete it when MR ends.<a name="line.109"></a> +<span class="sourceLineNo">110</span> */<a name="line.110"></a> +<span class="sourceLineNo">111</span> private static boolean DRY_RUN_TABLE_CREATED;<a name="line.111"></a> +<span class="sourceLineNo">112</span><a name="line.112"></a> +<span class="sourceLineNo">113</span> public static class TsvParser {<a name="line.113"></a> +<span class="sourceLineNo">114</span> /**<a name="line.114"></a> +<span class="sourceLineNo">115</span> * Column families and qualifiers mapped to the TSV columns<a name="line.115"></a> +<span class="sourceLineNo">116</span> */<a name="line.116"></a> +<span class="sourceLineNo">117</span> private final byte[][] families;<a name="line.117"></a> +<span class="sourceLineNo">118</span> private final byte[][] qualifiers;<a name="line.118"></a> <span class="sourceLineNo">119</span><a name="line.119"></a> -<span class="sourceLineNo">120</span> private int rowKeyColumnIndex;<a name="line.120"></a> +<span class="sourceLineNo">120</span> private final byte separatorByte;<a name="line.120"></a> <span class="sourceLineNo">121</span><a name="line.121"></a> -<span class="sourceLineNo">122</span> private int maxColumnCount;<a name="line.122"></a> +<span class="sourceLineNo">122</span> private int rowKeyColumnIndex;<a name="line.122"></a> <span class="sourceLineNo">123</span><a name="line.123"></a> -<span class="sourceLineNo">124</span> // Default value must be negative<a name="line.124"></a> -<span class="sourceLineNo">125</span> public static final int DEFAULT_TIMESTAMP_COLUMN_INDEX = -1;<a name="line.125"></a> -<span class="sourceLineNo">126</span><a name="line.126"></a> -<span class="sourceLineNo">127</span> private int timestampKeyColumnIndex = DEFAULT_TIMESTAMP_COLUMN_INDEX;<a name="line.127"></a> +<span class="sourceLineNo">124</span> private int maxColumnCount;<a name="line.124"></a> +<span class="sourceLineNo">125</span><a name="line.125"></a> +<span class="sourceLineNo">126</span> // Default value must be negative<a name="line.126"></a> +<span class="sourceLineNo">127</span> public static final int DEFAULT_TIMESTAMP_COLUMN_INDEX = -1;<a name="line.127"></a> <span class="sourceLineNo">128</span><a name="line.128"></a> -<span class="sourceLineNo">129</span> public static final String ROWKEY_COLUMN_SPEC = "HBASE_ROW_KEY";<a name="line.129"></a> +<span class="sourceLineNo">129</span> private int timestampKeyColumnIndex = DEFAULT_TIMESTAMP_COLUMN_INDEX;<a name="line.129"></a> <span class="sourceLineNo">130</span><a name="line.130"></a> -<span class="sourceLineNo">131</span> public static final String TIMESTAMPKEY_COLUMN_SPEC = "HBASE_TS_KEY";<a name="line.131"></a> +<span class="sourceLineNo">131</span> public static final String ROWKEY_COLUMN_SPEC = "HBASE_ROW_KEY";<a name="line.131"></a> <span class="sourceLineNo">132</span><a name="line.132"></a> -<span class="sourceLineNo">133</span> public static final String ATTRIBUTES_COLUMN_SPEC = "HBASE_ATTRIBUTES_KEY";<a name="line.133"></a> +<span class="sourceLineNo">133</span> public static final String TIMESTAMPKEY_COLUMN_SPEC = "HBASE_TS_KEY";<a name="line.133"></a> <span class="sourceLineNo">134</span><a name="line.134"></a> -<span class="sourceLineNo">135</span> public static final String CELL_VISIBILITY_COLUMN_SPEC = "HBASE_CELL_VISIBILITY";<a name="line.135"></a> +<span class="sourceLineNo">135</span> public static final String ATTRIBUTES_COLUMN_SPEC = "HBASE_ATTRIBUTES_KEY";<a name="line.135"></a> <span class="sourceLineNo">136</span><a name="line.136"></a> -<span class="sourceLineNo">137</span> public static final String CELL_TTL_COLUMN_SPEC = "HBASE_CELL_TTL";<a name="line.137"></a> +<span class="sourceLineNo">137</span> public static final String CELL_VISIBILITY_COLUMN_SPEC = "HBASE_CELL_VISIBILITY";<a name="line.137"></a> <span class="sourceLineNo">138</span><a name="line.138"></a> -<span class="sourceLineNo">139</span> private int attrKeyColumnIndex = DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.139"></a> +<span class="sourceLineNo">139</span> public static final String CELL_TTL_COLUMN_SPEC = "HBASE_CELL_TTL";<a name="line.139"></a> <span class="sourceLineNo">140</span><a name="line.140"></a> -<span class="sourceLineNo">141</span> public static final int DEFAULT_ATTRIBUTES_COLUMN_INDEX = -1;<a name="line.141"></a> +<span class="sourceLineNo">141</span> private int attrKeyColumnIndex = DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.141"></a> <span class="sourceLineNo">142</span><a name="line.142"></a> -<span class="sourceLineNo">143</span> public static final int DEFAULT_CELL_VISIBILITY_COLUMN_INDEX = -1;<a name="line.143"></a> +<span class="sourceLineNo">143</span> public static final int DEFAULT_ATTRIBUTES_COLUMN_INDEX = -1;<a name="line.143"></a> <span class="sourceLineNo">144</span><a name="line.144"></a> -<span class="sourceLineNo">145</span> public static final int DEFAULT_CELL_TTL_COLUMN_INDEX = -1;<a name="line.145"></a> +<span class="sourceLineNo">145</span> public static final int DEFAULT_CELL_VISIBILITY_COLUMN_INDEX = -1;<a name="line.145"></a> <span class="sourceLineNo">146</span><a name="line.146"></a> -<span class="sourceLineNo">147</span> private int cellVisibilityColumnIndex = DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.147"></a> +<span class="sourceLineNo">147</span> public static final int DEFAULT_CELL_TTL_COLUMN_INDEX = -1;<a name="line.147"></a> <span class="sourceLineNo">148</span><a name="line.148"></a> -<span class="sourceLineNo">149</span> private int cellTTLColumnIndex = DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.149"></a> +<span class="sourceLineNo">149</span> private int cellVisibilityColumnIndex = DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.149"></a> <span class="sourceLineNo">150</span><a name="line.150"></a> -<span class="sourceLineNo">151</span> /**<a name="line.151"></a> -<span class="sourceLineNo">152</span> * @param columnsSpecification the list of columns to parser out, comma separated.<a name="line.152"></a> -<span class="sourceLineNo">153</span> * The row key should be the special token TsvParser.ROWKEY_COLUMN_SPEC<a name="line.153"></a> -<span class="sourceLineNo">154</span> * @param separatorStr<a name="line.154"></a> -<span class="sourceLineNo">155</span> */<a name="line.155"></a> -<span class="sourceLineNo">156</span> public TsvParser(String columnsSpecification, String separatorStr) {<a name="line.156"></a> -<span class="sourceLineNo">157</span> // Configure separator<a name="line.157"></a> -<span class="sourceLineNo">158</span> byte[] separator = Bytes.toBytes(separatorStr);<a name="line.158"></a> -<span class="sourceLineNo">159</span> Preconditions.checkArgument(separator.length == 1,<a name="line.159"></a> -<span class="sourceLineNo">160</span> "TsvParser only supports single-byte separators");<a name="line.160"></a> -<span class="sourceLineNo">161</span> separatorByte = separator[0];<a name="line.161"></a> -<span class="sourceLineNo">162</span><a name="line.162"></a> -<span class="sourceLineNo">163</span> // Configure columns<a name="line.163"></a> -<span class="sourceLineNo">164</span> ArrayList<String> columnStrings = Lists.newArrayList(<a name="line.164"></a> -<span class="sourceLineNo">165</span> Splitter.on(',').trimResults().split(columnsSpecification));<a name="line.165"></a> -<span class="sourceLineNo">166</span><a name="line.166"></a> -<span class="sourceLineNo">167</span> maxColumnCount = columnStrings.size();<a name="line.167"></a> -<span class="sourceLineNo">168</span> families = new byte[maxColumnCount][];<a name="line.168"></a> -<span class="sourceLineNo">169</span> qualifiers = new byte[maxColumnCount][];<a name="line.169"></a> -<span class="sourceLineNo">170</span><a name="line.170"></a> -<span class="sourceLineNo">171</span> for (int i = 0; i < columnStrings.size(); i++) {<a name="line.171"></a> -<span class="sourceLineNo">172</span> String str = columnStrings.get(i);<a name="line.172"></a> -<span class="sourceLineNo">173</span> if (ROWKEY_COLUMN_SPEC.equals(str)) {<a name="line.173"></a> -<span class="sourceLineNo">174</span> rowKeyColumnIndex = i;<a name="line.174"></a> -<span class="sourceLineNo">175</span> continue;<a name="line.175"></a> -<span class="sourceLineNo">176</span> }<a name="line.176"></a> -<span class="sourceLineNo">177</span> if (TIMESTAMPKEY_COLUMN_SPEC.equals(str)) {<a name="line.177"></a> -<span class="sourceLineNo">178</span> timestampKeyColumnIndex = i;<a name="line.178"></a> -<span class="sourceLineNo">179</span> continue;<a name="line.179"></a> -<span class="sourceLineNo">180</span> }<a name="line.180"></a> -<span class="sourceLineNo">181</span> if (ATTRIBUTES_COLUMN_SPEC.equals(str)) {<a name="line.181"></a> -<span class="sourceLineNo">182</span> attrKeyColumnIndex = i;<a name="line.182"></a> -<span class="sourceLineNo">183</span> continue;<a name="line.183"></a> -<span class="sourceLineNo">184</span> }<a name="line.184"></a> -<span class="sourceLineNo">185</span> if (CELL_VISIBILITY_COLUMN_SPEC.equals(str)) {<a name="line.185"></a> -<span class="sourceLineNo">186</span> cellVisibilityColumnIndex = i;<a name="line.186"></a> -<span class="sourceLineNo">187</span> continue;<a name="line.187"></a> -<span class="sourceLineNo">188</span> }<a name="line.188"></a> -<span class="sourceLineNo">189</span> if (CELL_TTL_COLUMN_SPEC.equals(str)) {<a name="line.189"></a> -<span class="sourceLineNo">190</span> cellTTLColumnIndex = i;<a name="line.190"></a> -<span class="sourceLineNo">191</span> continue;<a name="line.191"></a> -<span class="sourceLineNo">192</span> }<a name="line.192"></a> -<span class="sourceLineNo">193</span> String[] parts = str.split(":", 2);<a name="line.193"></a> -<span class="sourceLineNo">194</span> if (parts.length == 1) {<a name="line.194"></a> -<span class="sourceLineNo">195</span> families[i] = str.getBytes();<a name="line.195"></a> -<span class="sourceLineNo">196</span> qualifiers[i] = HConstants.EMPTY_BYTE_ARRAY;<a name="line.196"></a> -<span class="sourceLineNo">197</span> } else {<a name="line.197"></a> -<span class="sourceLineNo">198</span> families[i] = parts[0].getBytes();<a name="line.198"></a> -<span class="sourceLineNo">199</span> qualifiers[i] = parts[1].getBytes();<a name="line.199"></a> -<span class="sourceLineNo">200</span> }<a name="line.200"></a> -<span class="sourceLineNo">201</span> }<a name="line.201"></a> -<span class="sourceLineNo">202</span> }<a name="line.202"></a> -<span class="sourceLineNo">203</span><a name="line.203"></a> -<span class="sourceLineNo">204</span> public boolean hasTimestamp() {<a name="line.204"></a> -<span class="sourceLineNo">205</span> return timestampKeyColumnIndex != DEFAULT_TIMESTAMP_COLUMN_INDEX;<a name="line.205"></a> -<span class="sourceLineNo">206</span> }<a name="line.206"></a> -<span class="sourceLineNo">207</span><a name="line.207"></a> -<span class="sourceLineNo">208</span> public int getTimestampKeyColumnIndex() {<a name="line.208"></a> -<span class="sourceLineNo">209</span> return timestampKeyColumnIndex;<a name="line.209"></a> -<span class="sourceLineNo">210</span> }<a name="line.210"></a> -<span class="sourceLineNo">211</span><a name="line.211"></a> -<span class="sourceLineNo">212</span> public boolean hasAttributes() {<a name="line.212"></a> -<span class="sourceLineNo">213</span> return attrKeyColumnIndex != DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.213"></a> -<span class="sourceLineNo">214</span> }<a name="line.214"></a> -<span class="sourceLineNo">215</span><a name="line.215"></a> -<span class="sourceLineNo">216</span> public boolean hasCellVisibility() {<a name="line.216"></a> -<span class="sourceLineNo">217</span> return cellVisibilityColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.217"></a> -<span class="sourceLineNo">218</span> }<a name="line.218"></a> -<span class="sourceLineNo">219</span><a name="line.219"></a> -<span class="sourceLineNo">220</span> public boolean hasCellTTL() {<a name="line.220"></a> -<span class="sourceLineNo">221</span> return cellTTLColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.221"></a> -<span class="sourceLineNo">222</span> }<a name="line.222"></a> -<span class="sourceLineNo">223</span><a name="line.223"></a> -<span class="sourceLineNo">224</span> public int getAttributesKeyColumnIndex() {<a name="line.224"></a> -<span class="sourceLineNo">225</span> return attrKeyColumnIndex;<a name="line.225"></a> -<span class="sourceLineNo">226</span> }<a name="line.226"></a> -<span class="sourceLineNo">227</span><a name="line.227"></a> -<span class="sourceLineNo">228</span> public int getCellVisibilityColumnIndex() {<a name="line.228"></a> -<span class="sourceLineNo">229</span> return cellVisibilityColumnIndex;<a name="line.229"></a> -<span class="sourceLineNo">230</span> }<a name="line.230"></a> -<span class="sourceLineNo">231</span><a name="line.231"></a> -<span class="sourceLineNo">232</span> public int getCellTTLColumnIndex() {<a name="line.232"></a> -<span class="sourceLineNo">233</span> return cellTTLColumnIndex;<a name="line.233"></a> -<span class="sourceLineNo">234</span> }<a name="line.234"></a> -<span class="sourceLineNo">235</span><a name="line.235"></a> -<span class="sourceLineNo">236</span> public int getRowKeyColumnIndex() {<a name="line.236"></a> -<span class="sourceLineNo">237</span> return rowKeyColumnIndex;<a name="line.237"></a> -<span class="sourceLineNo">238</span> }<a name="line.238"></a> -<span class="sourceLineNo">239</span><a name="line.239"></a> -<span class="sourceLineNo">240</span> public byte[] getFamily(int idx) {<a name="line.240"></a> -<span class="sourceLineNo">241</span> return families[idx];<a name="line.241"></a> -<span class="sourceLineNo">242</span> }<a name="line.242"></a> -<span class="sourceLineNo">243</span> public byte[] getQualifier(int idx) {<a name="line.243"></a> -<span class="sourceLineNo">244</span> return qualifiers[idx];<a name="line.244"></a> -<span class="sourceLineNo">245</span> }<a name="line.245"></a> -<span class="sourceLineNo">246</span><a name="line.246"></a> -<span class="sourceLineNo">247</span> public ParsedLine parse(byte[] lineBytes, int length)<a name="line.247"></a> -<span class="sourceLineNo">248</span> throws BadTsvLineException {<a name="line.248"></a> -<span class="sourceLineNo">249</span> // Enumerate separator offsets<a name="line.249"></a> -<span class="sourceLineNo">250</span> ArrayList<Integer> tabOffsets = new ArrayList<>(maxColumnCount);<a name="line.250"></a> -<span class="sourceLineNo">251</span> for (int i = 0; i < length; i++) {<a name="line.251"></a> -<span class="sourceLineNo">252</span> if (lineBytes[i] == separatorByte) {<a name="line.252"></a> -<span class="sourceLineNo">253</span> tabOffsets.add(i);<a name="line.253"></a> -<span class="sourceLineNo">254</span> }<a name="line.254"></a> -<span class="sourceLineNo">255</span> }<a name="line.255"></a> -<span class="sourceLineNo">256</span> if (tabOffsets.isEmpty()) {<a name="line.256"></a> -<span class="sourceLineNo">257</span> throw new BadTsvLineException("No delimiter");<a name="line.257"></a> -<span class="sourceLineNo">258</span> }<a name="line.258"></a> -<span class="sourceLineNo">259</span><a name="line.259"></a> -<span class="sourceLineNo">260</span> tabOffsets.add(length);<a name="line.260"></a> +<span class="sourceLineNo">151</span> private int cellTTLColumnIndex = DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.151"></a> +<span class="sourceLineNo">152</span><a name="line.152"></a> +<span class="sourceLineNo">153</span> /**<a name="line.153"></a> +<span class="sourceLineNo">154</span> * @param columnsSpecification the list of columns to parser out, comma separated.<a name="line.154"></a> +<span class="sourceLineNo">155</span> * The row key should be the special token TsvParser.ROWKEY_COLUMN_SPEC<a name="line.155"></a> +<span class="sourceLineNo">156</span> * @param separatorStr<a name="line.156"></a> +<span class="sourceLineNo">157</span> */<a name="line.157"></a> +<span class="sourceLineNo">158</span> public TsvParser(String columnsSpecification, String separatorStr) {<a name="line.158"></a> +<span class="sourceLineNo">159</span> // Configure separator<a name="line.159"></a> +<span class="sourceLineNo">160</span> byte[] separator = Bytes.toBytes(separatorStr);<a name="line.160"></a> +<span class="sourceLineNo">161</span> Preconditions.checkArgument(separator.length == 1,<a name="line.161"></a> +<span class="sourceLineNo">162</span> "TsvParser only supports single-byte separators");<a name="line.162"></a> +<span class="sourceLineNo">163</span> separatorByte = separator[0];<a name="line.163"></a> +<span class="sourceLineNo">164</span><a name="line.164"></a> +<span class="sourceLineNo">165</span> // Configure columns<a name="line.165"></a> +<span class="sourceLineNo">166</span> ArrayList<String> columnStrings = Lists.newArrayList(<a name="line.166"></a> +<span class="sourceLineNo">167</span> Splitter.on(',').trimResults().split(columnsSpecification));<a name="line.167"></a> +<span class="sourceLineNo">168</span><a name="line.168"></a> +<span class="sourceLineNo">169</span> maxColumnCount = columnStrings.size();<a name="line.169"></a> +<span class="sourceLineNo">170</span> families = new byte[maxColumnCount][];<a name="line.170"></a> +<span class="sourceLineNo">171</span> qualifiers = new byte[maxColumnCount][];<a name="line.171"></a> +<span class="sourceLineNo">172</span><a name="line.172"></a> +<span class="sourceLineNo">173</span> for (int i = 0; i < columnStrings.size(); i++) {<a name="line.173"></a> +<span class="sourceLineNo">174</span> String str = columnStrings.get(i);<a name="line.174"></a> +<span class="sourceLineNo">175</span> if (ROWKEY_COLUMN_SPEC.equals(str)) {<a name="line.175"></a> +<span class="sourceLineNo">176</span> rowKeyColumnIndex = i;<a name="line.176"></a> +<span class="sourceLineNo">177</span> continue;<a name="line.177"></a> +<span class="sourceLineNo">178</span> }<a name="line.178"></a> +<span class="sourceLineNo">179</span> if (TIMESTAMPKEY_COLUMN_SPEC.equals(str)) {<a name="line.179"></a> +<span class="sourceLineNo">180</span> timestampKeyColumnIndex = i;<a name="line.180"></a> +<span class="sourceLineNo">181</span> continue;<a name="line.181"></a> +<span class="sourceLineNo">182</span> }<a name="line.182"></a> +<span class="sourceLineNo">183</span> if (ATTRIBUTES_COLUMN_SPEC.equals(str)) {<a name="line.183"></a> +<span class="sourceLineNo">184</span> attrKeyColumnIndex = i;<a name="line.184"></a> +<span class="sourceLineNo">185</span> continue;<a name="line.185"></a> +<span class="sourceLineNo">186</span> }<a name="line.186"></a> +<span class="sourceLineNo">187</span> if (CELL_VISIBILITY_COLUMN_SPEC.equals(str)) {<a name="line.187"></a> +<span class="sourceLineNo">188</span> cellVisibilityColumnIndex = i;<a name="line.188"></a> +<span class="sourceLineNo">189</span> continue;<a name="line.189"></a> +<span class="sourceLineNo">190</span> }<a name="line.190"></a> +<span class="sourceLineNo">191</span> if (CELL_TTL_COLUMN_SPEC.equals(str)) {<a name="line.191"></a> +<span class="sourceLineNo">192</span> cellTTLColumnIndex = i;<a name="line.192"></a> +<span class="sourceLineNo">193</span> continue;<a name="line.193"></a> +<span class="sourceLineNo">194</span> }<a name="line.194"></a> +<span class="sourceLineNo">195</span> String[] parts = str.split(":", 2);<a name="line.195"></a> +<span class="sourceLineNo">196</span> if (parts.length == 1) {<a name="line.196"></a> +<span class="sourceLineNo">197</span> families[i] = str.getBytes();<a name="line.197"></a> +<span class="sourceLineNo">198</span> qualifiers[i] = HConstants.EMPTY_BYTE_ARRAY;<a name="line.198"></a> +<span class="sourceLineNo">199</span> } else {<a name="line.199"></a> +<span class="sourceLineNo">200</span> families[i] = parts[0].getBytes();<a name="line.200"></a> +<span class="sourceLineNo">201</span> qualifiers[i] = parts[1].getBytes();<a name="line.201"></a> +<span class="sourceLineNo">202</span> }<a name="line.202"></a> +<span class="sourceLineNo">203</span> }<a name="line.203"></a> +<span class="sourceLineNo">204</span> }<a name="line.204"></a> +<span class="sourceLineNo">205</span><a name="line.205"></a> +<span class="sourceLineNo">206</span> public boolean hasTimestamp() {<a name="line.206"></a> +<span class="sourceLineNo">207</span> return timestampKeyColumnIndex != DEFAULT_TIMESTAMP_COLUMN_INDEX;<a name="line.207"></a> +<span class="sourceLineNo">208</span> }<a name="line.208"></a> +<span class="sourceLineNo">209</span><a name="line.209"></a> +<span class="sourceLineNo">210</span> public int getTimestampKeyColumnIndex() {<a name="line.210"></a> +<span class="sourceLineNo">211</span> return timestampKeyColumnIndex;<a name="line.211"></a> +<span class="sourceLineNo">212</span> }<a name="line.212"></a> +<span class="sourceLineNo">213</span><a name="line.213"></a> +<span class="sourceLineNo">214</span> public boolean hasAttributes() {<a name="line.214"></a> +<span class="sourceLineNo">215</span> return attrKeyColumnIndex != DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.215"></a> +<span class="sourceLineNo">216</span> }<a name="line.216"></a> +<span class="sourceLineNo">217</span><a name="line.217"></a> +<span class="sourceLineNo">218</span> public boolean hasCellVisibility() {<a name="line.218"></a> +<span class="sourceLineNo">219</span> return cellVisibilityColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.219"></a> +<span class="sourceLineNo">220</span> }<a name="line.220"></a> +<span class="sourceLineNo">221</span><a name="line.221"></a> +<span class="sourceLineNo">222</span> public boolean hasCellTTL() {<a name="line.222"></a> +<span class="sourceLineNo">223</span> return cellTTLColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.223"></a> +<span class="sourceLineNo">224</span> }<a name="line.224"></a> +<span class="sourceLineNo">225</span><a name="line.225"></a> +<span class="sourceLineNo">226</span> public int getAttributesKeyColumnIndex() {<a name="line.226"></a> +<span class="sourceLineNo">227</span> return attrKeyColumnIndex;<a name="line.227"></a> +<span class="sourceLineNo">228</span> }<a name="line.228"></a> +<span class="sourceLineNo">229</span><a name="line.229"></a> +<span class="sourceLineNo">230</span> public int getCellVisibilityColumnIndex() {<a name="line.230"></a> +<span class="sourceLineNo">231</span> return cellVisibilityColumnIndex;<a name="line.231"></a> +<span class="sourceLineNo">232</span> }<a name="line.232"></a> +<span class="sourceLineNo">233</span><a name="line.233"></a> +<span class="sourceLineNo">234</span> public int getCellTTLColumnIndex() {<a name="line.234"></a> +<span class="sourceLineNo">235</span> return cellTTLColumnIndex;<a name="line.235"></a> +<span class="sourceLineNo">236</span> }<a name="line.236"></a> +<span class="sourceLineNo">237</span><a name="line.237"></a> +<span class="sourceLineNo">238</span> public int getRowKeyColumnIndex() {<a name="line.238"></a> +<span class="sourceLineNo">239</span> return rowKeyColumnIndex;<a name="line.239"></a> +<span class="sourceLineNo">240</span> }<a name="line.240"></a> +<span class="sourceLineNo">241</span><a name="line.241"></a> +<span class="sourceLineNo">242</span> public byte[] getFamily(int idx) {<a name="line.242"></a> +<span class="sourceLineNo">243</span> return families[idx];<a name="line.243"></a> +<span class="sourceLineNo">244</span> }<a name="line.244"></a> +<span class="sourceLineNo">245</span> public byte[] getQualifier(int idx) {<a name="line.245"></a> +<span class="sourceLineNo">246</span> return qualifiers[idx];<a name="line.246"></a> +<span class="sourceLineNo">247</span> }<a name="line.247"></a> +<span class="sourceLineNo">248</span><a name="line.248"></a> +<span class="sourceLineNo">249</span> public ParsedLine parse(byte[] lineBytes, int length)<a name="line.249"></a> +<span class="sourceLineNo">250</span> throws BadTsvLineException {<a name="line.250"></a> +<span class="sourceLineNo">251</span> // Enumerate separator offsets<a name="line.251"></a> +<span class="sourceLineNo">252</span> ArrayList<Integer> tabOffsets = new ArrayList<>(maxColumnCount);<a name="line.252"></a> +<span class="sourceLineNo">253</span> for (int i = 0; i < length; i++) {<a name="line.253"></a> +<span class="sourceLineNo">254</span> if (lineBytes[i] == separatorByte) {<a name="line.254"></a> +<span class="sourceLineNo">255</span> tabOffsets.add(i);<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> if (tabOffsets.isEmpty()) {<a name="line.258"></a> +<span class="sourceLineNo">259</span> throw new BadTsvLineException("No delimiter");<a name="line.259"></a> +<span class="sourceLineNo">260</span> }<a name="line.260"></a> <span class="sourceLineNo">261</span><a name="line.261"></a> -<span class="sourceLineNo">262</span> if (tabOffsets.size() > maxColumnCount) {<a name="line.262"></a> -<span class="sourceLineNo">263</span> throw new BadTsvLineException("Excessive columns");<a name="line.263"></a> -<span class="sourceLineNo">264</span> } else if (tabOffsets.size() <= getRowKeyColumnIndex()) {<a name="line.264"></a> -<span class="sourceLineNo">265</span> throw new BadTsvLineException("No row key");<a name="line.265"></a> -<span class="sourceLineNo">266</span> } else if (hasTimestamp()<a name="line.266"></a> -<span class="sourceLineNo">267</span> && tabOffsets.size() <= getTimestampKeyColumnIndex()) {<a name="line.267"></a> -<span class="sourceLineNo">268</span> throw new BadTsvLineException("No timestamp");<a name="line.268"></a> -<span class="sourceLineNo">269</span> } else if (hasAttributes() && tabOffsets.size() <= getAttributesKeyColumnIndex()) {<a name="line.269"></a> -<span class="sourceLineNo">270</span> throw new BadTsvLineException("No attributes specified");<a name="line.270"></a> -<span class="sourceLineNo">271</span> } else if (hasCellVisibility() && tabOffsets.size() <= getCellVisibilityColumnIndex()) {<a name="line.271"></a> -<span class="sourceLineNo">272</span> throw new BadTsvLineException("No cell visibility specified");<a name="line.272"></a> -<span class="sourceLineNo">273</span> } else if (hasCellTTL() && tabOffsets.size() <= getCellTTLColumnIndex()) {<a name="line.273"></a> -<span class="sourceLineNo">274</span> throw new BadTsvLineException("No cell TTL specified");<a name="line.274"></a> -<span class="sourceLineNo">275</span> }<a name="line.275"></a> -<span class="sourceLineNo">276</span> return new ParsedLine(tabOffsets, lineBytes);<a name="line.276"></a> -<span class="sourceLineNo">277</span> }<a name="line.277"></a> -<span class="sourceLineNo">278</span><a name="line.278"></a> -<span class="sourceLineNo">279</span> class ParsedLine {<a name="line.279"></a> -<span class="sourceLineNo">280</span> private final ArrayList<Integer> tabOffsets;<a name="line.280"></a> -<span class="sourceLineNo">281</span> private byte[] lineBytes;<a name="line.281"></a> -<span class="sourceLineNo">282</span><a name="line.282"></a> -<span class="sourceLineNo">283</span> ParsedLine(ArrayList<Integer> tabOffsets, byte[] lineBytes) {<a name="line.283"></a> -<span class="sourceLineNo">284</span> this.tabOffsets = tabOffsets;<a name="line.284"></a> -<span class="sourceLineNo">285</span> this.lineBytes = lineBytes;<a name="line.285"></a> -<span class="sourceLineNo">286</span> }<a name="line.286"></a> -<span class="sourceLineNo">287</span><a name="line.287"></a> -<span class="sourceLineNo">288</span> public int getRowKeyOffset() {<a name="line.288"></a> -<span class="sourceLineNo">289</span> return getColumnOffset(rowKeyColumnIndex);<a name="line.289"></a> -<span class="sourceLineNo">290</span> }<a name="line.290"></a> -<span class="sourceLineNo">291</span> public int getRowKeyLength() {<a name="line.291"></a> -<span class="sourceLineNo">292</span> return getColumnLength(rowKeyColumnIndex);<a name="line.292"></a> -<span class="sourceLineNo">293</span> }<a name="line.293"></a> -<span class="sourceLineNo">294</span><a name="line.294"></a> -<span class="sourceLineNo">295</span> public long getTimestamp(long ts) throws BadTsvLineException {<a name="line.295"></a> -<span class="sourceLineNo">296</span> // Return ts if HBASE_TS_KEY is not configured in column spec<a name="line.296"></a> -<span class="sourceLineNo">297</span> if (!hasTimestamp()) {<a name="line.297"></a> -<span class="sourceLineNo">298</span> return ts;<a name="line.298"></a> -<span class="sourceLineNo">299</span> }<a name="line.299"></a> -<span class="sourceLineNo">300</span><a name="line.300"></a> -<span class="sourceLineNo">301</span> String timeStampStr = Bytes.toString(lineBytes,<a name="line.301"></a> -<span class="sourceLineNo">302</span> getColumnOffset(timestampKeyColumnIndex),<a name="line.302"></a> -<span class="sourceLineNo">303</span> getColumnLength(timestampKeyColumnIndex));<a name="line.303"></a> -<span class="sourceLineNo">304</span> try {<a name="line.304"></a> -<span class="sourceLineNo">305</span> return Long.parseLong(timeStampStr);<a name="line.305"></a> -<span class="sourceLineNo">306</span> } catch (NumberFormatException nfe) {<a name="line.306"></a> -<span class="sourceLineNo">307</span> // treat this record as bad record<a name="line.307"></a> -<span class="sourceLineNo">308</span> throw new BadTsvLineException("Invalid timestamp " + timeStampStr);<a name="line.308"></a> -<span class="sourceLineNo">309</span> }<a name="line.309"></a> -<span class="sourceLineNo">310</span> }<a name="line.310"></a> -<span class="sourceLineNo">311</span><a name="line.311"></a> -<span class="sourceLineNo">312</span> private String getAttributes() {<a name="line.312"></a> -<span class="sourceLineNo">313</span> if (!hasAttributes()) {<a name="line.313"></a> -<span class="sourceLineNo">314</span> return null;<a name="line.314"></a> -<span class="sourceLineNo">315</span> } else {<a name="line.315"></a> -<span class="sourceLineNo">316</span> return Bytes.toString(lineBytes, getColumnOffset(attrKeyColumnIndex),<a name="line.316"></a> -<span class="sourceLineNo">317</span> getColumnLength(attrKeyColumnIndex));<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><a name="line.320"></a> -<span class="sourceLineNo">321</span> public String[] getIndividualAttributes() {<a name="line.321"></a> -<span class="sourceLineNo">322</span> String attributes = getAttributes();<a name="line.322"></a> -<span class="sourceLineNo">323</span> if (attributes != null) {<a name="line.323"></a> -<span class="sourceLineNo">324</span> return attributes.split(DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR);<a name="line.324"></a> -<span class="sourceLineNo">325</span> } else {<a name="line.325"></a> -<span class="sourceLineNo">326</span> return null;<a name="line.326"></a> -<span class="sourceLineNo">327</span> }<a name="line.327"></a> -<span class="sourceLineNo">328</span> }<a name="line.328"></a> -<span class="sourceLineNo">329</span><a name="line.329"></a> -<span class="sourceLineNo">330</span> public int getAttributeKeyOffset() {<a name="line.330"></a> -<span class="sourceLineNo">331</span> if (hasAttributes()) {<a name="line.331"></a> -<span class="sourceLineNo">332</span> return getColumnOffset(attrKeyColumnIndex);<a name="line.332"></a> -<span class="sourceLineNo">333</span> } else {<a name="line.333"></a> -<span class="sourceLineNo">334</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.334"></a> -<span class="sourceLineNo">335</span> }<a name="line.335"></a> -<span class="sourceLineNo">336</span> }<a name="line.336"></a> -<span class="sourceLineNo">337</span><a name="line.337"></a> -<span class="sourceLineNo">338</span> public int getAttributeKeyLength() {<a name="line.338"></a> -<span class="sourceLineNo">339</span> if (hasAttributes()) {<a name="line.339"></a> -<span class="sourceLineNo">340</span> return getColumnLength(attrKeyColumnIndex);<a name="line.340"></a> -<span class="sourceLineNo">341</span> } else {<a name="line.341"></a> -<span class="sourceLineNo">342</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.342"></a> -<span class="sourceLineNo">343</span> }<a name="line.343"></a> -<span class="sourceLineNo">344</span> }<a name="line.344"></a> -<span class="sourceLineNo">345</span><a name="line.345"></a> -<span class="sourceLineNo">346</span> public int getCellVisibilityColumnOffset() {<a name="line.346"></a> -<span class="sourceLineNo">347</span> if (hasCellVisibility()) {<a name="line.347"></a> -<span class="sourceLineNo">348</span> return getColumnOffset(cellVisibilityColumnIndex);<a name="line.348"></a> -<span class="sourceLineNo">349</span> } else {<a name="line.349"></a> -<span class="sourceLineNo">350</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.350"></a> -<span class="sourceLineNo">351</span> }<a name="line.351"></a> -<span class="sourceLineNo">352</span> }<a name="line.352"></a> -<span class="sourceLineNo">353</span><a name="line.353"></a> -<span class="sourceLineNo">354</span> public int getCellVisibilityColumnLength() {<a name="line.354"></a> -<span class="sourceLineNo">355</span> if (hasCellVisibility()) {<a name="line.355"></a> -<span class="sourceLineNo">356</span> return getColumnLength(cellVisibilityColumnIndex);<a name="line.356"></a> -<span class="sourceLineNo">357</span> } else {<a name="line.357"></a> -<span class="sourceLineNo">358</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.358"></a> -<span class="sourceLineNo">359</span> }<a name="line.359"></a> -<span class="sourceLineNo">360</span> }<a name="line.360"></a> -<span class="sourceLineNo">361</span><a name="line.361"></a> -<span class="sourceLineNo">362</span> public String getCellVisibility() {<a name="line.362"></a> -<span class="sourceLineNo">363</span> if (!hasCellVisibility()) {<a name="line.363"></a> -<span class="sourceLineNo">364</span> return null;<a name="line.364"></a> -<span class="sourceLineNo">365</span> } else {<a name="line.365"></a> -<span class="sourceLineNo">366</span> return Bytes.toString(lineBytes, getColumnOffset(cellVisibilityColumnIndex),<a name="line.366"></a> -<span class="sourceLineNo">367</span> getColumnLength(cellVisibilityColumnIndex));<a name="line.367"></a> -<span class="sourceLineNo">368</span> }<a name="line.368"></a> -<span class="sourceLineNo">369</span> }<a name="line.369"></a> -<span class="sourceLineNo">370</span><a name="line.370"></a> -<span class="sourceLineNo">371</span> public int getCellTTLColumnOffset() {<a name="line.371"></a> -<span class="sourceLineNo">372</span> if (hasCellTTL()) {<a name="line.372"></a> -<span class="sourceLineNo">373</span> return getColumnOffset(cellTTLColumnIndex);<a name="line.373"></a> -<span class="sourceLineNo">374</span> } else {<a name="line.374"></a> -<span class="sourceLineNo">375</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.375"></a> -<span class="sourceLineNo">376</span> }<a name="line.376"></a> -<span class="sourceLineNo">377</span> }<a name="line.377"></a> -<span class="sourceLineNo">378</span><a name="line.378"></a> -<span class="sourceLineNo">379</span> public int getCellTTLColumnLength() {<a name="line.379"></a> -<span class="sourceLineNo">380</span> if (hasCellTTL()) {<a name="line.380"></a> -<span class="sourceLineNo">381</span> return getColumnLength(cellTTLColumnIndex);<a name="line.381"></a> -<span class="sourceLineNo">382</span> } else {<a name="line.382"></a> -<span class="sourceLineNo">383</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.383"></a> -<span class="sourceLineNo">384</span> }<a name="line.384"></a> -<span class="sourceLineNo">385</span> }<a name="line.385"></a> -<span class="sourceLineNo">386</span><a name="line.386"></a> -<span class="sourceLineNo">387</span> public long getCellTTL() {<a name="line.387"></a> -<span class="sourceLineNo">388</span> if (!hasCellTTL()) {<a name="line.388"></a> -<span class="sourceLineNo">389</span> return 0;<a name="line.389"></a> -<span class="sourceLineNo">390</span> } else {<a name="line.390"></a> -<span class="sourceLineNo">391</span> return Bytes.toLong(lineBytes, getColumnOffset(cellTTLColumnIndex),<a name="line.391"></a> -<span class="sourceLineNo">392</span> getColumnLength(cellTTLColumnIndex));<a name="line.392"></a> -<span class="sourceLineNo">393</span> }<a name="line.393"></a> -<span class="sourceLineNo">394</span> }<a name="line.394"></a> -<span class="sourceLineNo">395</span><a name="line.395"></a> -<span class="sourceLineNo">396</span> public int getColumnOffset(int idx) {<a name="line.396"></a> -<span class="sourceLineNo">397</span> if (idx > 0)<a name="line.397"></a> -<span class="sourceLineNo">398</span> return tabOffsets.get(idx - 1) + 1;<a name="line.398"></a> -<span class="sourceLineNo">399</span> else<a name="line.399"></a> -<span class="sourceLineNo">400</span> return 0;<a name="line.400"></a> -<span class="sourceLineNo">401</span> }<a name="line.401"></a> -<span class="sourceLineNo">402</span> public int getColumnLength(int idx) {<a name="line.402"></a> -<span class="sourceLineNo">403</span> return tabOffsets.get(idx) - getColumnOffset(idx);<a name="line.403"></a> -<span class="sourceLineNo">404</span> }<a name="line.404"></a> -<span class="sourceLineNo">405</span> public int getColumnCount() {<a name="line.405"></a> -<span class="sourceLineNo">406</span> return tabOffsets.size();<a name="line.406"></a> -<span class="sourceLineNo">407</span> }<a name="line.407"></a> -<span class="sourceLineNo">408</span> public byte[] getLineBytes() {<a name="line.408"></a> -<span class="sourceLineNo">409</span> return lineBytes;<a name="line.409"></a> -<span class="sourceLineNo">410</span> }<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 static class BadTsvLineException extends Exception {<a name="line.413"></a> -<span class="sourceLineNo">414</span> public BadTsvLineException(String err) {<a name="line.414"></a> -<span class="sourceLineNo">415</span> super(err);<a name="line.415"></a> -<span class="sourceLineNo">416</span> }<a name="line.416"></a> -<span class="sourceLineNo">417</span> private static final long serialVersionUID = 1L;<a name="line.417"></a> -<span class="sourceLineNo">418</span> }<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> * Return starting position and length of row key from the specified line bytes.<a name="line.421"></a> -<span class="sourceLineNo">422</span> * @param lineBytes<a name="line.422"></a> -<span class="sourceLineNo">423</span> * @param length<a name="line.423"></a> -<span class="sourceLineNo">424</span> * @return Pair of row key offset and length.<a name="line.424"></a> -<span class="sourceLineNo">425</span> * @throws BadTsvLineException<a name="line.425"></a> -<span class="sourceLineNo">426</span> */<a name="line.426"></a> -<span class="sourceLineNo">427</span> public Pair<Integer, Integer> parseRowKey(byte[] lineBytes, int length)<a name="line.427"></a> -<span class="sourceLineNo">428</span> throws BadTsvLineException {<a name="line.428"></a> -<span class="sourceLineNo">429</span> int rkColumnIndex = 0;<a name="line.429"></a> -<span class="sourceLineNo">430</span> int startPos = 0, endPos = 0;<a name="line.430"></a> -<span class="sourceLineNo">431</span> for (int i = 0; i <= length; i++) {<a name="line.431"></a> -<span class="sourceLineNo">432</span> if (i == length || lineBytes[i] == separatorByte) {<a name="line.432"></a> -<span class="sourceLineNo">433</span> endPos = i - 1;<a name="line.433"></a> -<span class="sourceLineNo">434</span> if (rkColumnIndex++ == getRowKeyColumnIndex()) {<a name="line.434"></a> -<span class="sourceLineNo">435</span> if ((endPos + 1) == startPos) {<a name="line.435"></a> -<span class="sourceLineNo">436</span> throw new BadTsvLineException("Empty value for ROW KEY.");<a name="line.436"></a> -<span class="sourceLineNo">437</span> }<a name="line.437"></a> -<span class="sourceLineNo">438</span> break;<a name="line.438"></a> -<span class="sourceLineNo">439</span> } else {<a name="line.439"></a> -<span class="sourceLineNo">440</span> startPos = endPos + 2;<a name="line.440"></a> -<span class="sourceLineNo">441</span> }<a name="line.441"></a> -<span class="sourceLineNo">442</span> }<a name="line.442"></a> -<span class="sourceLineNo">443</span> if (i == length) {<a name="line.443"></a> -<span class="sourceLineNo">444</span> throw new BadTsvLineException(<a name="line.444"></a> -<span class="sourceLineNo">445</span> "Row key does not exist as number of columns in the line"<a name="line.445"></a> -<span class="sourceLineNo">446</span> + " are less than row key position.");<a name="line.446"></a> -<span class="sourceLineNo">447</span> }<a name="line.447"></a> -<span class="sourceLineNo">448</span> }<a name="line.448"></a> -<span class="sourceLineNo">449</span> return new Pair<>(startPos, endPos - startPos + 1);<a name="line.449"></a> -<span class="sourceLineNo">450</span> }<a name="line.450"></a> -<span class="sourceLineNo">451</span> }<a name="line.451"></a> -<span class="sourceLineNo">452</span><a name="line.452"></a> -<span class="sourceLineNo">453</span> /**<a name="line.453"></a> -<span class="sourceLineNo">454</span> * Sets up the actual job.<a name="line.454"></a> -<span class="sourceLineNo">455</span> *<a name="line.455"></a> -<span class="sourceLineNo">456</span> * @param conf The current configuration.<a name="line.456"></a> -<span class="sourceLineNo">457</span> * @param args The command line parameters.<a name="line.457"></a> -<span class="sourceLineNo">458</span> * @return The newly created job.<a name="line.458"></a> -<span class="sourceLineNo">459</span> * @throws IOException When setting up the job fails.<a name="line.459"></a> -<span class="sourceLineNo">460</span> */<a name="line.460"></a> -<span class="sourceLineNo">461</span> protected static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.461"></a> -<span class="sourceLineNo">462</span> throws IOException, ClassNotFoundException {<a name="line.462"></a> -<span class="sourceLineNo">463</span> Job job = null;<a name="line.463"></a> -<span class="sourceLineNo">464</span> boolean isDryRun = conf.getBoolean(DRY_RUN_CONF_KEY, false);<a name="line.464"></a> -<span class="sourceLineNo">465</span> try (Connection connection = ConnectionFactory.createConnection(conf)) {<a name="line.465"></a> -<span class="sourceLineNo">466</span> try (Admin admin = connection.getAdmin()) {<a name="line.466"></a> -<span class="sourceLineNo">467</span> // Support non-XML supported characters<a name="line.467"></a> -<span class="sourceLineNo">468</span> // by re-encoding the passed separator as a Base64 string.<a name="line.468"></a> -<span class="sourceLineNo">469</span> String actualSeparator = conf.get(SEPARATOR_CONF_KEY);<a name="line.469"></a> -<span class="sourceLineNo">470</span> if (actualSeparator != null) {<a name="line.470"></a> -<span class="sourceLineNo">471</span> conf.set(SEPARATOR_CONF_KEY,<a name="line.471"></a> -<span class="sourceLineNo">472</span> Base64.encodeBytes(actualSeparator.getBytes()));<a name="line.472"></a> -<span class="sourceLineNo">473</span> }<a name="line.473"></a> -<span class="sourceLineNo">474</span><a name="line.474"></a> -<span class="sourceLineNo">475</span> // See if a non-default Mapper was set<a name="line.475"></a> -<span class="sourceLineNo">476</span> String mapperClassName = conf.get(MAPPER_CONF_KEY);<a name="line.476"></a> -<span class="sourceLineNo">477</span> Class mapperClass = mapperClassName != null? Class.forName(mapperClassName): DEFAULT_MAPPER;<a name="line.477"></a> -<span class="sourceLineNo">478</span><a name="line.478"></a> -<span class="sourceLineNo">479</span> TableName tableName = TableName.valueOf(args[0]);<a name="line.479"></a> -<span class="sourceLineNo">480</span> Path inputDir = new Path(args[1]);<a name="line.480"></a> -<span class="sourceLineNo">481</span> String jobName = conf.get(JOB_NAME_CONF_KEY,NAME + "_" + tableName.getNameAsString());<a name="line.481"></a> -<span class="sourceLineNo">482</span> job = Job.getInstance(conf, jobName);<a name="line.482"></a> -<span class="sourceLineNo">483</span> job.setJarByClass(mapperClass);<a name="line.483"></a> -<span class="sourceLineNo">484</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.484"></a> -<span class="sourceLineNo">485</span> job.setInputFormatClass(TextInputFormat.class);<a name="line.485"></a> -<span class="sourceLineNo">486</span> job.setMapperClass(mapperClass);<a name="line.486"></a> -<span class="sourceLineNo">487</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.487"></a> -<span class="sourceLineNo">488</span> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);<a name="line.488"></a> -<span class="sourceLineNo">489</span> String[] columns = conf.getStrings(COLUMNS_CONF_KEY);<a name="line.489"></a> -<span class="sourceLineNo">490</span> if(StringUtils.isNotEmpty(conf.get(CREDENTIALS_LOCATION))) {<a name="line.490"></a> -<span class="sourceLineNo">491</span> String fileLoc = conf.get(CREDENTIALS_LOCATION);<a name="line.491"></a> -<span class="sourceLineNo">492</span> Credentials cred = Credentials.readTokenStorageFile(new File(fileLoc), conf);<a name="line.492"></a> -<span class="sourceLineNo">493</span> job.getCredentials().addAll(cred);<a name="line.493"></a> -<span class="sourceLineNo">494</span> }<a name="line.494"></a> -<span class="sourceLineNo">495</span><a name="line.495"></a> -<span class="sourceLineNo">496</span> if (hfileOutPath != null) {<a name="line.496"></a> -<span class="sourceLineNo">497</span> if (!admin.tableExists(tableName)) {<a name="line.497"></a> -<span class="sourceLineNo">498</span> LOG.warn(format("Table '%s' does not exist.", tableName));<a name="line.498"></a> -<span class="sourceLineNo">499</span> if ("yes".equalsIgnoreCase(conf.get(CREATE_TABLE_CONF_KEY, "yes"))) {<a name="line.499"></a> -<span class="sourceLineNo">500</span> // TODO: this is backwards. Instead of depending on the existence of a table,<a name="line.500"></a> -<span class="sourceLineNo">501</span> // create a sane splits file for HFileOutputFormat based on data sampling.<a name="line.501"></a> -<span class="sourceLineNo">502</span> createTable(admin, tableName, columns);<a name="line.502"></a> -<span class="sourceLineNo">503</span> if (isDryRun) {<a name="line.503"></a> -<span class="sourceLineNo">504</span> LOG.warn("Dry run: Table will be deleted at end of dry run.");<a name="line.504"></a> -<span class="sourceLineNo">505</span> synchronized (ImportTsv.class) {<a name="line.505"></a> -<span class="sourceLineNo">506</span> DRY_RUN_TABLE_CREATED = true;<a name="line.506"></a> -<span class="sourceLineNo">507</span> }<a name="line.507"></a> -<span class="sourceLineNo">508</span> }<a name="line.508"></a> -<span class="sourceLineNo">509</span> } else {<a name="line.509"></a> -<span class="sourceLineNo">510</span> String errorMsg =<a name="line.510"></a> -<span class="sourceLineNo">511</span> format("Table '%s' does not exist and '%s' is set to no.", tableName,<a name="line.511"></a> -<span class="sourceLineNo">512</span> CREATE_TABLE_CONF_KEY);<a name="line.512"></a> -<span class="sourceLineNo">513</span> LOG.error(errorMsg);<a name="line.513"></a> -<span class="sourceLineNo">514</span> throw new TableNotFoundException(errorMsg);<a name="line.514"></a> -<span class="sourceLineNo">515</span> }<a name="line.515"></a> -<span class="sourceLineNo">516</span> }<a name="line.516"></a> -<span class="sourceLineNo">517</span> try (Table table = connection.getTable(tableName);<a name="line.517"></a> -<span class="sourceLineNo">518</span> RegionLocator regionLocator = connection.getRegionLocator(tableName)) {<a name="line.518"></a> -<span class="sourceLineNo">519</span> boolean noStrict = conf.getBoolean(NO_STRICT_COL_FAMILY, false);<a name="line.519"></a> -<span class="sourceLineNo">520</span> // if no.strict is false then check column family<a name="line.520"></a> -<span class="sourceLineNo">521</span> if(!noStrict) {<a name="line.521"></a> -<span class="sourceLineNo">522</span> ArrayList<String> unmatchedFamilies = new ArrayList<>();<a name="line.522"></a> -<span class="sourceLineNo">523</span> Set<String> cfSet = getColumnFamilies(columns);<a name="line.523"></a> -<span class="sourceLineNo">524</span> HTableDescriptor tDesc = table.getTableDescriptor();<a name="line.524"></a> -<span class="sourceLineNo">525</span> for (String cf : cfSet) {<a name="line.525"></a> -<span class="sourceLineNo">526</span> if(tDesc.getFamily(Bytes.toBytes(cf)) == null) {<a name="line.526"></a> -<span class="sourceLineNo">527</span> unmatchedFamilies.add(cf);<a name="line.527"></a> -<span class="sourceLineNo">528</span> }<a name="line.528"></a> -<span class="sourceLineNo">529</span> }<a name="line.529"></a> -<span class="sourceLineNo">530</span> if(unmatchedFamilies.size() > 0) {<a name="line.530"></a> -<span class="sourceLineNo">531</span> ArrayList<String> familyNames = new ArrayList<>();<a name="line.531"></a> -<span class="sourceLineNo">532</span> for (HColumnDescriptor family : table.getTableDescriptor().getFamilies()) {<a name="line.532"></a> -<span class="sourceLineNo">533</span> familyNames.add(family.getNameAsString());<a name="line.533"></a> -<span class="sourceLineNo">534</span> }<a name="line.534"></a> -<span class="sourceLineNo">535</span> String msg =<a name="line.535"></a> -<span class="sourceLineNo">536</span> "Column Families " + unmatchedFamilies + " specified in " + COLUMNS_CONF_KEY<a name="line.536"></a> -<span class="sourceLineNo">537</span> + " does not match with any of the table " + tableName<a name="line.537"></a> -<span class="sourceLineNo">538</span> + " column families " + familyNames + ".\n"<a name="line.538"></a> -<span class="sourceLineNo">539</span> + "To disable column family check, use -D" + NO_STRICT_COL_FAMILY<a name="line.539"></a> -<span class="sourceLineNo">540</span> + "=true.\n";<a name="line.540"></a> -<span class="sourceLineNo">541</span> usage(msg);<a name="line.541"></a> -<span class="sourceLineNo">542</span> System.exit(-1);<a name="line.542"></a> -<span class="sourceLineNo">543</span> }<a name="line.543"></a> -<span class="sourceLineNo">544</span> }<a name="line.544"></a> -<span class="sourceLineNo">545</span> if (mapperClass.equals(TsvImporterTextMapper.class)) {<a name="line.545"></a> -<span class="sourceLineNo">546</span> job.setMapOutputValueClass(Text.class);<a name="line.546"></a> -<span class="sourceLineNo">547</span> job.setReducerClass(TextSortReducer.class);<a name="line.547"></a> -<span class="sourceLineNo">548</span> } else {<a name="line.548"></a> -<span class="sourceLineNo">549</span> job.setMapOutputValueClass(Put.class);<a name="line.549"></a> -<span class="sourceLineNo">550</span> job.setCombinerClass(PutCombiner.class);<a name="line.550"></a> -<span class="sourceLineNo">551</span> job.setReducerClass(PutSortReducer.class);<a name="line.551"></a> -<span class="sourceLineNo">552</span> }<a name="line.552"></a> -<span class="sourceLineNo">553</span> if (!isDryRun) {<a name="line.553"></a> -<span class="sourceLineNo">554</span> Path outputDir = new Path(hfileOutPath);<a name="line.554"></a> -<span class="sourceLineNo">555</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.555"></a> -<span class="sourceLineNo">556</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getTableDescriptor(),<a name="line.556"></a> -<span class="sourceLineNo">557</span> regionLocator);<a name="line.557"></a> -<span class="sourceLineNo">558</span> }<a name="line.558"></a> -<span class="sourceLineNo">559</span> }<a name="line.559"></a> -<span class="sourceLineNo">560</span> } else {<a name="line.560"></a> -<span class="sourceLineNo">561</span> if (!admin.tableExists(tableName)) {<a name="line.561"></a> -<span class="sourceLineNo">562</span> String errorMsg = format("Table '%s' does not exist.", tableName);<a name="line.562"></a> -<span class="sourceLineNo">563</span> LOG.error(errorMsg);<a name="line.563"></a> -<span class="sourceLineNo">564</span> throw new TableNotFoundException(errorMsg);<a name="line.564"></a> -<span class="sourceLineNo">565</span> }<a name="line.565"></a> -<span class="sourceLineNo">566</span> if (mapperClass.equals(TsvImporterTextMapper.class)) {<a name="line.566"></a> -<span class="sourceLineNo">567</span> usage(TsvImporterTextMapper.class.toString()<a name="line.567"></a> -<span class="sourceLineNo">568</span> + " should not be used for non bulkloading case. use "<a name="line.568"></a> -<span class="sourceLineNo">569</span> + TsvImporterMapper.class.toString()<a name="line.569"></a> -<span class="sourceLineNo">570</span> + " or custom mapper whose value type is Put.");<a name="line.570"></a> -<span class="sourceLineNo">571</span> System.exit(-1);<a name="line.571"></a> -<span class="sourceLineNo">572</span> }<a name="line.572"></a> -<span class="sourceLineNo">573</span> if (!isDryRun) {<a name="line.573"></a> -<span class="sourceLineNo">574</span> // No reducers. Just write straight to table. Call initTableReducerJob<a name="line.574"></a> -<span class="sourceLineNo">575</span> // to set up the TableOutputFormat.<a name="line.575"></a> -<span class="sourceLineNo">576</span> TableMapReduceUtil.initTableReducerJob(tableName.getNameAsString(), null, job);<a name="line.576"></a> -<span class="sourceLineNo">577</span> }<a name="line.577"></a> -<span class="sourceLineNo">578</span> job.setNumReduceTasks(0);<a name="line.578"></a> -<span class="sourceLineNo">579</span> }<a name="line.579"></a> -<span class="sourceLineNo">580</span> if (isDryRun) {<a name="line.580"></a> -<span class="sourceLineNo">581</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.581"></a> -<span class="sourceLineNo">582</span> job.getConfiguration().setStrings("io.serializations",<a name="line.582"></a> -<span class="sourceLineNo">583</span> job.getConfiguration().get("io.serializations"),<a name="line.583"></a> -<span class="sourceLineNo">584</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.584"></a> -<span class="sourceLineNo">585</span> KeyValueSerialization.class.getName());<a name="line.585"></a> -<span class="sourceLineNo">586</span> }<a name="line.586"></a> -<span class="sourceLineNo">587</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.587"></a> -<span class="sourceLineNo">588</span> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),<a name="line.588"></a> -<span class="sourceLineNo">589</span> com.google.common.base.Function.class /* Guava used by TsvParser */);<a name="line.589"></a> -<span class="sourceLineNo">590</span> }<a name="line.590"></a> -<span class="sourceLineNo">591</span> }<a name="line.591"></a> -<span class="sourceLineNo">592</span> return job;<a name="line.592"></a> -<span class="sourceLineNo">593</span> }<a name="line.593"></a> -<span class="sourceLineNo">594</span><a name="line.594"></a> -<span class="sourceLineNo">595</span> private static void createTable(Admin admin, TableName tableName, String[] columns)<a name="line.595"></a> -<span class="sourceLineNo">596</span> throws IOException {<a name="line.596"></a> -<span class="sourceLineNo">597</span> HTableDescriptor htd = new HTableDescriptor(tableName);<a name="line.597"></a> -<span class="sourceLineNo">598</span> Set<String> cfSet = getColumnFamilies(columns);<a name="line.598"></a> -<span class="sourceLineNo">599</span> for (String cf : cfSet) {<a name="line.599"></a> -<span class="sourceLineNo">600</span> HColumnDescriptor hcd = new HColumnDescriptor(Bytes.toBytes(cf));<a name="line.600"></a> -<span class="sourceLineNo">601</span> htd.addFamily(hcd);<a name="line.601"></a> -<span class="sourceLineNo">602</span> }<a name="line.602"></a> -<span class="sourceLineNo">603</span> LOG.warn(format("Creating table '%s' with '%s' columns and default descriptors.",<a name="line.603"></a> -<span class="sourceLineNo">604</span> tableName, cfSet));<a name="line.604"></a> -<span class="sourceLineNo">605</span> admin.createTable(htd);<a name="line.605"></a> -<span class="sourceLineNo">606</span> }<a name="line.606"></a> -<span class="sourceLineNo">607</span><a name="line.607"></a> -<span class="sourceLineNo">608</span> private static void deleteTable(Configuration conf, String[] args) {<a name="line.608"></a> -<span class="sourceLineNo">609</span> TableName tableName = TableName.valueOf(args[0]);<a name="line.609"></a> -<span class="sourceLineNo">610</span> try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.610"></a> -<span class="sourceLineNo">611</span> Admin admin = connection.getAdmin()) {<a name="line.611"></a> -<span class="sourceLineNo">612</span> try {<a name="line.612"></a> -<span class="sourceLineNo">613</span> admin.disableTable(tableName);<a name="line.613"></a> -<span class="sourceLineNo">614</span> } catch (TableNotEnabledException e) {<a name="line.614"></a> -<span class="sourceLineNo">615</span> LOG.debug("Dry mode: Table: " + tableName + " already disabled, so just deleting it.");<a name="line.615"></a> -<span class="sourceLineNo">616</span> }<a name="line.616"></a> -<span class="sourceLineNo">617</span> admin.deleteTable(tableName);<a name="line.617"></a> -<span class="sourceLineNo">618</span> } catch (IOException e) {<a name="line.618"></a> -<span class="sourceLineNo">619</span> LOG.error(format("***Dry run: Failed to delete table '%s'.***%n%s", tableName,<a name="line.619"></a> -<span class="sourceLineNo">620</span> e.toString()));<a name="line.620"></a> -<span class="sourceLineNo">621</span> return;<a name="line.621"></a> -<span class="sourceLineNo">622</span> }<a name="line.622"></a> -<span class="sourceLineNo">623</span> LOG.info(format("Dry run: Deleted table '%s'.", tableName));<a name="line.623"></a> -<span class="sourceLineNo">624</span> }<a name="line.624"></a> -<span class="sourceLineNo">625</span><a name="line.625"></a> -<span class="sourceLineNo">626</span> private static Set<String> getColumnFamilies(String[] columns) {<a name="line.626"></a> -<span class="sourceLineNo">627</span> Set<String> cfSet = new HashSet<>();<a name="line.627"></a> -<span class="sourceLineNo">628</span> for (String aColumn : columns) {<a name="line.628"></a> -<span class="sourceLineNo">629</span> if (TsvParser.ROWKEY_COLUMN_SPEC.equals(aColumn)<a name="line.629"></a> -<span class="sourceLineNo">630</span> || TsvParser.TIMESTAMPKEY_COLUMN_SPEC.equals(aColumn)<a name="line.630"></a> -<span class="sourceLineNo">631</span> || TsvParser.CELL_VISIBILITY_COLUMN_SPEC.equals(aColumn)<a name="line.631"></a> -<span class="sourceLineNo">632</span> || TsvParser.CELL_TTL_COLUMN_SPEC.equals(aColumn)<a name="line.632"></a> -<span class="sourceLineNo">633</span> || TsvParser.ATTRIBUTES_COLUMN_SPEC.equals(aColumn))<a name="line.633"></a> -<span class="sourceLineNo">634</span> continue;<a name="line.634"></a> -<span class="sourceLineNo">635</span> // we are only concerned with the first one (in case this is a cf:cq)<a name="line.635"></a> -<span class="sourceLineNo">636</span> cfSet.add(aColumn.split(":", 2)[0]);<a name="line.636"></a> -<span class="sourceLineNo">637</span> }<a name="line.637"></a> -<span class="sourceLineNo">638</span> return cfSet;<a name="line.638"></a> -<span class="sourceLineNo">639</span> }<a name="line.639"></a> -<span class="sourceLineNo">640</span><a name="line.640"></a> -<span class="sourceLineNo">641</span> /*<a name="line.641"></a> -<span class="sourceLineNo">642</span> * @param errorMsg Error message. Can be null.<a name="line.642"></a> -<span class="sourceLineNo">643</span> */<a name="line.643"></a> -<span class="sourceLineNo">644</span> private static void usage(final String errorMsg) {<a name="line.644"></a> -<span class="sourceLineNo">645</span> if (errorMsg != null && errorMsg.length() > 0) {<a name="line.645"></a> -<span class="sourceLineNo">646</span> System.err.println("ERROR: " + errorMsg);<a name="line.646"></a> -<span class="sourceLineNo">647</span> }<a name="line.647"></a> -<span class="sourceLineNo">648</span> String usage =<a name="line.648"></a> -<span class="sourceLineNo">649</span> "Usage: " + NAME + " -D"+ COLUMNS_CONF_KEY + "=a,b,c <tablename> <inputdir>\n" +<a name="line.649"></a> -<span class="sourceLineNo">650</span> "\n" +<a name="line.650"></a> -<span class="sourceLineNo">651</span> "Imports the given input directory of TSV data into the specified table.\n" +<a name="line.651"></a> +<span class="sourceLineNo">262</span> tabOffsets.add(length);<a name="line.262"></a> +<span class="sourceLineNo">263</span><a name="line.263"></a> +<span class="sourceLineNo">264</span> if (tabOffsets.size() > maxColumnCount) {<a name="line.264"></a> +<span class="sourceLineNo">265</span> throw new BadTsvLineException("Excessive columns");<a name="line.265"></a> +<span class="sourceLineNo">266</span> } else if (tabOffsets.size() <= getRowKeyColumnIndex()) {<a name="line.266"></a> +<span class="sourceLineNo">267</span> throw new BadTsvLineException("No row key");<a name="line.267"></a> +<span class="sourceLineNo">268</span> } else if (hasTimestamp()<a name="line.268"></a> +<span class="sourceLineNo">269</span> && tabOffsets.size() <= getTimestampKeyColumnIndex()) {<a name="line.269"></a> +<span class="sourceLineNo">270</span> throw new BadTsvLineException("No timestamp");<a name="line.270"></a> +<span class="sourceLineNo">271</span> } else if (hasAttributes() && tabOffsets.size() <= getAttributesKeyColumnIndex()) {<a name="line.271"></a> +<span class="sourceLineNo">272</span> throw new BadTsvLineException("No attributes specified");<a name="line.272"></a> +<span class="sourceLineNo">273</span> } else if (hasCellVisibility() && tabOffsets.size() <= getCellVisibilityColumnIndex()) {<a name="line.273"></a> +<span class="sourceLineNo">274</span> throw new BadTsvLineException("No cell visibility specified");<a name="line.274"></a> +<span class="sourceLineNo">275</span> } else if (hasCellTTL() && tabOffsets.size() <= getCellTTLColumnIndex()) {<a name="line.275"></a> +<span class="sourceLineNo">276</span> throw new BadTsvLineException("No cell TTL specified");<a name="line.276"></a> +<span class="sourceLineNo">277</span> }<a name="line.277"></a> +<span class="sourceLineNo">278</span> return new ParsedLine(tabOffsets, lineBytes);<a name="line.278"></a> +<span class="sourceLineNo">279</span> }<a name="line.279"></a> +<span class="sourceLineNo">280</span><a name="line.280"></a> +<span class="sourceLineNo">281</span> class ParsedLine {<a name="line.281"></a> +<span class="sourceLineNo">282</span> private final ArrayList<Integer> tabOffsets;<a name="line.282"></a> +<span class="sourceLineNo">283</span> private byte[] lineBytes;<a name="line.283"></a> +<span class="sourceLineNo">284</span><a name="line.284"></a> +<span class="sourceLineNo">285</span> ParsedLine(ArrayList<Integer> tabOffsets, byte[] lineBytes) {<a name="line.285"></a> +<span class="sourceLineNo">286</span> this.tabOffsets = tabOffsets;<a name="line.286"></a> +<span class="sourceLineNo">287</span> this.lineBytes = lineBytes;<a name="line.287"></a> +<span class="sourceLineNo">288</span> }<a name="line.288"></a> +<span class="sourceLineNo">289</span><a name="line.289"></a> +<span class="sourceLineNo">290</span> public int getRowKeyOffset() {<a name="line.290"></a> +<span class="sourceLineNo">291</span> return getColumnOffset(rowKeyColumnIndex);<a name="line.291"></a> +<span class="sourceLineNo">292</span> }<a name="line.292"></a> +<span class="sourceLineNo">293</span> public int getRowKeyLength() {<a name="line.293"></a> +<span class="sourceLineNo">294</span> return getColumnLength(rowKeyColumnIndex);<a name="line.294"></a> +<span class="sourceLineNo">295</span> }<a name="line.295"></a> +<span class="sourceLineNo">296</span><a name="line.296"></a> +<span class="sourceLineNo">297</span> public long getTimestamp(long ts) throws BadTsvLineException {<a name="line.297"></a> +<span class="sourceLineNo">298</span> // Return ts if HBASE_TS_KEY is not configured in column spec<a name="line.298"></a> +<span class="sourceLineNo">299</span> if (!hasTimestamp()) {<a name="line.299"></a> +<span class="sourceLineNo">300</span> return ts;<a name="line.300"></a> +<span class="sourceLineNo">301</span> }<a name="line.301"></a> +<span class="sourceLineNo">302</span><a name="line.302"></a> +<span class="sourceLineNo">303</span> String timeStampStr = Bytes.toString(lineBytes,<a name="line.303"></a> +<span class="sourceLineNo">304</span> getColumnOffset(timestampKeyColumnIndex),<a name="line.304"></a> +<span class="sourceLineNo">305</span> getColumnLength(timestampKeyColumnIndex));<a name="line.305"></a> +<span class="sourceLineNo">306</span> try {<a name="line.306"></a> +<span class="sourceLineNo">307</span> return Long.parseLong(timeStampStr);<a name="line.307"></a> +<span class="sourceLineNo">308</span> } catch (NumberFormatException nfe) {<a name="line.308"></a> +<span class="sourceLineNo">309</span> // treat this record as bad record<a name="line.309"></a> +<span class="sourceLineNo">310</span> throw new BadTsvLineException("Invalid timestamp " + timeStampStr);<a name="line.310"></a> +<span class="sourceLineNo">311</span> }<a name="line.311"></a> +<span class="sourceLineNo">312</span> }<a name="line.312"></a> +<span class="sourceLineNo">313</span><a name="line.313"></a> +<span class="sourceLineNo">314</span> private String getAttributes() {<a name="line.314"></a> +<span class="sourceLineNo">315</span> if (!hasAttributes()) {<a name="line.315"></a> +<span class="sourceLineNo">316</span> return null;<a name="line.316"></a> +<span class="sourceLineNo">317</span> } else {<a name="line.317"></a> +<span class="sourceLineNo">318</span> return Bytes.toString(lineBytes, getColumnOffset(attrKeyColumnIndex),<a name="line.318"></a> +<span class="sourceLineNo">319</span> getColumnLength(attrKeyColumnIndex));<a name="line.319"></a> +<span class="sourceLineNo">320</span> }<a name="line.320"></a> +<span class="sourceLineNo">321</span> }<a name="line.321"></a> +<span class="sourceLineNo">322</span><a name="line.322"></a> +<span class="sourceLineNo">323</span> public String[] getIndividualAttributes() {<a name="line.323"></a> +<span class="sourceLineNo">324</span> String attributes = getAttributes();<a name="line.324"></a> +<span class="sourceLineNo">325</span> if (attributes != null) {<a name="line.325"></a> +<span class="sourceLineNo">326</span> return attributes.split(DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR);<a name="line.326"></a> +<span class="sourceLineNo">327</span> } else {<a name="line.327"></a> +<span class="sourceLineNo">328</span> return null;<a name="line.328"></a> +<span class="sourceLineNo">329</span> }<a name="line.329"></a> +<span class="sourceLineNo">330</span> }<a name="line.330"></a> +<span class="sourceLineNo">331</span><a name="line.331"></a> +<span class="sourceLineNo">332</span> public int getAttributeKeyOffset() {<a name="line.332"></a> +<span class="sourceLineNo">333</span> if (hasAttributes()) {<a name="line.333"></a> +<span class="sourceLineNo">334</span> return getColumnOffset(attrKeyColumnIndex);<a name="line.334"></a> +<span class="sourceLineNo">335</span> } else {<a name="line.335"></a> +<span class="sourceLineNo">336</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.336"></a> +<span class="sourceLineNo">337</span> }<a name="line.337"></a> +<span class="sourceLineNo">338</span> }<a name="line.338"></a> +<span class="sourceLineNo">339</span><a name="line.339"></a> +<span class="sourceLineNo">340</span> public int getAttributeKeyLength() {<a name="line.340"></a> +<span class="sourceLineNo">341</span> if (hasAttributes()) {<a name="line.341"></a> +<span class="sourceLineNo">342</span> return getColumnLength(attrKeyColumnIndex);<a name="line.342"></a> +<span class="sourceLineNo">343</span> } else {<a name="line.343"></a> +<span class="sourceLineNo">344</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.344"></a> +<span class="sourceLineNo">345</span> }<a name="line.345"></a> +<span class="sourceLineNo">346</span> }<a name="line.346"></a> +<span class="sourceLineNo">347</span><a name="line.347"></a> +<span class="sourceLineNo">348</span> public int getCellVisibilityColumnOffset() {<a name="line.348"></a> +<span class="sourceLineNo">349</span> if (hasCellVisibility()) {<a name="line.349"></a> +<span class="sourceLineNo">350</span> return getColumnOffset(cellVisibilityColumnIndex);<a name="line.350"></a> +<span class="sourceLineNo">351</span> } else {<a name="line.351"></a> +<span class="sourceLineNo">352</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.352"></a> +<span class="sourceLineNo">353</span> }<a name="line.353"></a> +<span class="sourceLineNo">354</span> }<a name="line.354"></a> +<span class="sourceLineNo">355</span><a name="line.355"></a> +<span class="sourceLineNo">356</span> public int getCellVisibilityColumnLength() {<a name="line.356"></a> +<span class="sourceLineNo">357</span> if (hasCellVisibility()) {<a name="line.357"></a> +<span class="sourceLineNo">358</span> return getColumnLength(cellVisibilityColumnIndex);<a name="line.358"></a> +<span class="sourceLineNo">359</span> } else {<a name="line.359"></a> +<span class="sourceLineNo">360</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.360"></a> +<span class="sourceLineNo">361</span> }<a name="line.361"></a> +<span class="sourceLineNo">362</span> }<a name="line.362"></a> +<span class="sourceLineNo">363</span><a name="line.363"></a> +<span class="sourceLineNo">364</span> public String getCellVisibility() {<a name="line.364"></a> +<span class="sourceLineNo">365</span> if (!hasCellVisibility()) {<a name="line.365"></a> +<span class="sourceLineNo">366</span> return null;<a name="line.366"></a> +<span class="sourceLineNo">367</span> } else {<a name="line.367"></a> +<span class="sourceLineNo">368</span> return Bytes.toString(lineBytes, getColumnOffset(cellVisibilityColumnIndex),<a name="line.368"></a> +<span class="sourceLineNo">369</span> getColumnLength(cellVisibilityColumnIndex));<a name="line.369"></a> +<span class="sourceLineNo">370</span> }<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 getCellTTLColumnOffset() {<a name="line.373"></a> +<span class="sourceLineNo">374</span> if (hasCellTTL()) {<a name="line.374"></a> +<span class="sourceLineNo">375</span> return getColumnOffset(cellTTLColumnIndex);<a name="line.375"></a> +<span class="sourceLineNo">376</span> } else {<a name="line.376"></a> +<span class="sourceLineNo">377</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.377"></a> +<span class="sourceLineNo">378</span> }<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 getCellTTLColumnLength() {<a name="line.381"></a> +<span class="sourceLineNo">382</span> if (hasCellTTL()) {<a name="line.382"></a> +<span class="sourceLineNo">383</span> return getColumnLength(cellTTLColumnIndex);<a name="line.383"></a> +<span class="sourceLineNo">384</span> } else {<a name="line.384"></a> +<span class="sourceLineNo">385</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.385"></a> +<span class="sourceLineNo">386</span> }<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 long getCellTTL() {<a name="line.389"></a> +<span class="sourceLineNo">390</span> if (!hasCellTTL()) {<a name="line.390"></a> +<span class="sourceLineNo">391</span> return 0;<a name="line.391"></a> +<span class="sourceLineNo">392</span> } else {<a name="line.392"></a> +<span class="sourceLineNo">393</span> return Bytes.toLong(lineBytes, getColumnOffset(cellTTLColumnIndex),<a name="line.393"></a> +<span class="sourceLineNo">394</span> getColumnLength(cellTTLColumnIndex));<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><a name="line.397"></a> +<span class="sourceLineNo">398</span> public int getColumnOffset(int idx) {<a name="line.398"></a> +<span class="sourceLineNo">399</span> if (idx > 0)<a name="line.399"></a> +<span class="sourceLineNo">400</span> return tabOffsets.get(idx - 1) + 1;<a name="line.400"></a> +<span class="sourceLineNo">401</span> else<a name="line.401"></a> +<span class="sourceLineNo">402</span> return 0;<a name=
<TRUNCATED>