[PIG-738] Regexp passed from pigscript fails in UDF

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.3.0
• Fix Version/s: 0.6.0
• Component/s: grunt
• Hadoop Flags: Reviewed
• Assignee: Pradeep Kamath
• Reporter: Viraj Bhat
• Created: 28/Mar/09 09:12
• Updated: 24/Mar/10 22:15
• Resolved: 22/Sep/09 17:33

Description

Consider a pig script which parses and counts regular expressions from a text file.
The regular expression supplied in the Pig script needs to escape the "." (dot) character.

register myregexp.jar;

-- pattern not picked up

define minelogs ci_pig_udfs.RegexGroupCount('www\\.yahoo\\.com/sports');

A = load '/user/viraj/regexpinput.txt'  using PigStorage() as (source : chararray);

B = foreach A generate minelogs(source) as sportslogs;

dump B;

Snippet of UDF RegexGroupCount.java

public class RegexGroupCount extends EvalFunc<Integer> {

    private final Pattern pattern_;

    public RegexGroupCount(String patternStr) {

       System.out.println("My pattern supplied is "+patternStr);

       System.out.println("Equality test "+patternStr.equals("www\\.yahoo\\.com/sports"));

       pattern_ = Pattern.compile(patternStr, Pattern.DOTALL|Pattern.CASE_INSENSITIVE);

   }
  public Integer exec(Tuple input)  throws IOException {
   }
}

Running the above script on the following dataset :
====================================================================================================
dshfdskfwww.yahoo.com/sportsjoadfjdslpdshfdskfwww.yahoo.com/sportsjoadfjdsl
kas;dka;sd
jsjsjwww.yahoo.com/sports
jsdLSJDcom/sports
wwwJyahooMcom/sports
====================================================================================================

Results in the following:

My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
Userfunc: (Name: UserFunc viraj-Sat Mar 28 02:06:31 PDT 2009-14 function: ci_pig_udfs.RegexGroupCount('www\\.yahoo
.com/sports') Operator Key: viraj-Sat Mar 28 02:06:31 PDT 2009-14)
Userfunc fs: int
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false
My pattern supplied is www\\.yahoo
.com/sports
Equality test false

2009-03-28 02:06:43,923 [main] INFO org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher - 100% complete
2009-03-28 02:06:43,923 [main] INFO org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher - Success!
(0)
(0)
(0)
(0)
(0)
====================================================================================================

In essence there seems to be no way of passing this type of constructor argument through the Pig script. The only workaround seems to be hard coding the values in the UDF!!