学步园

Hive提供了User Defined Functions的开发接口以给使用者扩展。上手比较简单，下面是开发使用一个MD5的例子。

JAVA代码:

package org.nalang.hive.udf;
import java.io.UnsupportedEncodingException;
import java.security.MessageDigest;
import java.security.NoSuchAlgorithmException;

import org.apache.hadoop.hive.ql.exec.UDF;
import org.apache.hadoop.io.Text;

public class Md5 extends UDF {

 public Text evaluate(Object... args) {
 if(args.length == 1){
 return new Text(rawMd5(args[0],16));
 }else if(args.length == 2 && args[1].equals("16")){
 return new Text(rawMd5(args[0],16));
 }else{
 return new Text(rawMd5(args[0],32));
 }
 }

 private String rawMd5(String str) {
 MessageDigest messageDigest = null;
 try {
 messageDigest = MessageDigest.getInstance("MD5");
 messageDigest.reset();
 messageDigest.update(str.getBytes("UTF-8"));
 } catch (NoSuchAlgorithmException e) {
 System.out.println("NoSuchAlgorithmException caught!");
 System.exit(-1);
 } catch (UnsupportedEncodingException e) {
 e.printStackTrace();
 }

 byte[] byteArray = messageDigest.digest();

 StringBuffer md5StrBuff = new StringBuffer();

 for (int i = 0; i < byteArray.length; i++) {
 if (Integer.toHexString(0xFF & byteArray[i]).length() == 1)
 md5StrBuff.append("0").append(
 Integer.toHexString(0xFF & byteArray[i]));
 else
 md5StrBuff.append(Integer.toHexString(0xFF & byteArray[i]));
 }
 return md5StrBuff.toString().toUpperCase();
 }

 private String rawMd5(Object str,int length){
 if(length == 16){
 return rawMd5(StringUtils.StringFilter(str)).substring(8,24);
 }else{
 return rawMd5(StringUtils.StringFilter(str));
 }
 }
}

Hive应用:

hive>add jar /home/work/local/hive/udf/udf.jar; create temporary function md5 as 'com.baifendian.hive.udf.Md5'; select md5('test') from taba limit 1; 

参考:

https://cwiki.apache.org/confluence/display/Hive/LanguageManual+UDF