Differences between revisions 7 and 8
Revision 7 as of 2009-05-15 05:30:17
Size: 3034
Editor: vpn-nat
Revision 8 as of 2009-09-20 23:38:15
Size: 3036
Editor: localhost
Comment: converted to 1.6 markup
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
[[Anchor(Embedding_Pig_In_Java_Programs)]] <<Anchor(Embedding_Pig_In_Java_Programs)>>
Line 9: Line 9:
 * Create an instance of `PigServer`. See [http://incubator.apache.org/pig/javadoc/docs Javadoc] for more details.  * Create an instance of `PigServer`. See [[http://incubator.apache.org/pig/javadoc/docs|Javadoc]] for more details.
Line 14: Line 14:
[[Anchor(Example)]] <<Anchor(Example)>>

Embedding Pig In Java Programs

Sometimes you want more control than Pig scripts can give you. If so, you can embed Pig Latin in Java (just like SQL can be embedded in programs using JDBC).

The following steps need to be carried out:

  • Make sure pig.jar is on your classpath.

  • Create an instance of PigServer. See Javadoc for more details.

  • Issue commands through that PigServer by calling PigServer.registerQuery().

  • To retrieve results, either call PigServer.openIterator() or PigServer.store().

  • If you have user defined functions, register them by calling PigServer.registerJar().


Let's assume you need to count the number of occurrences of each word in a document. Let's also assume that you have EvalFunction Tokenize that parses a line of text and returns all the words for that line. The function is located in /mylocation/tokenize.jar.

The PigLatin script for the computation will look like this:

register /mylocation/tokenize.jar
A = load 'mytext' using TextLoader();
B = foreach A generate flatten(tokenize($0));
C = group B by $1;
D = foreach C generate flatten(group), COUNT(B.$0);
store D into 'myoutput';

The same computation can be performed with this Java program:

import java.io.IOException;
import org.apache.pig.PigServer;

public class WordCount {
   public static void main(String[] args) {
      PigServer pigServer = new PigServer();
      try {
         runMyQuery(pigServer, "myinput.txt";
      catch (IOException e) {
   public static void runMyQuery(PigServer pigServer, String inputFile) throws IOException {        
       pigServer.registerQuery("A = load '" + inputFile + "' using TextLoader();");
       pigServer.registerQuery("B = foreach A generate flatten(tokenize($0));");
       pigServer.registerQuery("C = group B by $1;");
       pigServer.registerQuery("D = foreach C generate flatten(group), COUNT(B.$0);");
       pigServer.store("D", "myoutput");


  • The jar which contains your functions must be registered.
  • The four calls to pigServer.registerQuery() simply cause the query to be parsed and enquired. The query is not actually executed until pigServer.store() is called.

  • The input data referred to on the load statement, must be on HDFS in the specified location.
  • The final result is placed into myoutput file in the your current working directory on HDFS. (By default this is your home directory on HDFS.)

To run your program, you need to first compile it by using the following command:

javac -cp <path>pig.jar WordCount.java

If the compilation is successful, you can then run your program:

java -cp <path>pig.jar WordCount

EmbeddedPig (last edited 2009-09-20 23:38:15 by localhost)