IIRC Java won't let you update the classpath on the fly (for security reasons I think). But giving a better error message would definitely be good.

Alan.

On Mar 1, 2011, at 10:04 AM, Dmitriy Ryaboy wrote:

patches accepted :-)

D

On Tue, Mar 1, 2011 at 10:02 AM, Dan Brickley <[email protected]> wrote:

On 1 March 2011 17:56, Dmitriy Ryaboy <[email protected]> wrote:
Hi Dan,
iirc, registering a jar does not put it on the Pig client classpath, it
just
tells Pig to ship the jar. You want to put it on the PIG_CLASSPATH before
you invoke pig.

Perfect, that was exactly it. It's running now :)

Would it make sense for REGISTER to augment the classpath? Or maybe
better, for the error message to mention the role of PIG_CLASSPATH?

cheers,

Dan

On Tue, Mar 1, 2011 at 5:57 AM, Dan Brickley <[email protected]> wrote:

I'm trying to use InvokeForString to call a simple static method that
wraps http://mzsanford.github.com/twitter-text-java/docs/api/index.html
https://github.com/twitter/twitter-text-java ... specifically the
Extractor class extractURLs method. In fact since the logical result
is a list of URLs perhaps I should be writing proper Pig-centric
wrapper that returns a tuple, but for now I thought a stringified list would be ok for my immediate purposes. That purpose being pulling out all the URLs from a corpus of tweets, so we can expand the bit.ly and
other short urls...

So - I built the extra class (src below) and packaged it inside the
twitter-text jar, and verify it's in there and usable as follows:

danbri$ java -cp
twitter-text-1.3.1-plus-tv.notube.TwitterExtractor.jar
tv.notube.TwitterExtractor "hello http://example.com/
http://example.org/ world"
URLs: [http://example.com/, http://example.org/]

Then from the same directory, I try run this as a Pig job:

tw06 = load '/user/danbri/twitter/tweets2009-06.tab.txt.lzo' AS (
when: chararray, who: chararray, msg: chararray);
REGISTER twitter-text-1.3.1-plus-tv.notube.TwitterExtractor.jar;
DEFINE ExtractURLs InvokeForString('tv.notube.TwitterExtractor.urls',
'String');
urls = FOREACH tw06 GENERATE ExtractURLs(msg);
x = SAMPLE urls 0.001;
dump x;

...but we don't get past InvokeForString,

2011-03-01 14:50:31,033 [main] ERROR org.apache.pig.tools.grunt.Grunt
- ERROR 1000: Error during parsing. could not instantiate
'InvokeForString' with arguments '[tv.notube.TwitterExtractor.urls,
String]'
Details at logfile: /home/danbri/twitter/pig_1298987430385.log
...->
Caused by: java.lang.reflect.InvocationTargetException
Caused by: java.lang.ClassNotFoundException: tv.notube.TwitterExtractor

I checked that Pig is finding the jar by mis-spelling the filename in the "REGISTER" line (which as expected causes things to fail earlier).
Also double-check that the class is in the jar,
danbri$ jar -tvf
twitter-text-1.3.1-plus-tv.notube.TwitterExtractor.jar | grep tv
   0 Tue Mar 01 12:03:04 CET 2011 tv/
   0 Tue Mar 01 12:03:04 CET 2011 tv/notube/
1114 Tue Mar 01 13:40:30 CET 2011 tv/notube/TwitterExtractor.class

...so I'm finding myself stuck. I'm sure the answer is staring me in the face, but I can't see it. Perhaps I should just do things properly
with "extends EvalFunc<String>" and return the tuples separately
anyway...

Thanks for any pointers,

Dan


package tv.notube;
import com.twitter.Extractor;
import java.util.List;
class TwitterExtractor {

public static void main (String[] args) {
  String in = args[0];
      System.out.println("URLs: " + urls(in));
}

public static String urls(String tweet) {
  Extractor ex = new Extractor();
  List urls = ex.extractURLs(tweet);
  String o = urls.toString();
  return o;
}
}




Reply via email to