Running AE from the CAS Editor view

[Fabien POULARD <gr...@gmail.com>]

Hi folks,

I've been trying to run the WhitespaceTokenizer AE (for example) on
the CAS documents of a Corpus folder within the CAS Editor for some
time now, with no success. I run into a class not found exception each
time, indicating that the WhitespaceTokenizer class cannot be found.

I don't know I'm overlooking on something here or if I'm just dumb...

I'm using Apache UIMA 2.3, the CAS Editor plugin from the repository
and Eclipse 3.3.2

Here is what I've tried :

1- The simple way : create the CAS Editor project, import type system,
create a corpus directory, import text files, create processing
directory and import the WhitespaceTokenizer (let's says WT from now)
descriptor in it. Finally click on a document in the corpus directory
: Annotator > WT.xml => it fails

2- The improved way : do the same as the simple way, then open the
properties > UIMA CDE Property page and specify the absolute path to
the WT jar. => it fails the same way

3- The hacky way : create a Java project and specify the Build path so
that the WT is loaded, edit the .project file, add the NLPProject
nature, open the CAS Editor perspective and redo the simple way => it
still fails

Well, I'm out of ideas... does anybody have a solution to my problem ?
Has anybody already been able to launch an AE on a corpus from the CAS
Editor ?

I'd be grateful to any hint !

--
Fabien Poulard

Re: Running AE from the CAS Editor view

[Fabien POULARD <gr...@gmail.com>]

On Fri, Mar 5, 2010 at 1:52 PM, Fabien POULARD <gr...@gmail.com> wrote:
>> A second way you can take is, create a pear bundle, install it
>> and copy the pear bundle descriptor over to the processing
>> folder.
>
> Thanks for the information, I will try these out over the week-end...

I tried with a PEAR package, it effectively worked, but I needed to
create an aggregate calling the pear descriptor.

I wrote a tutorial on the CAS Editor :

http://www.fabienpoulard.info/index.php?post/2010/03/04/Utilisation-du-CAS-Editor

It's in French but if it can be usefull I may translate it to english.
At least the video parts are barely understandable in any language.

--
Fabien

Re: Running AE from the CAS Editor view

[Fabien POULARD <gr...@gmail.com>]

On Fri, Mar 5, 2010 at 1:27 PM, Jörn Kottmann <ko...@gmail.com> wrote:
> the WhitespcaceTokenizer or any other annotator must
> be in the classpath of the UIMA runtime plugin. Right
> now one way to achieve that is to make an OSGi bundle
> out of your annotator and deploy it to your eclipse installation.
> The bundle must use Eclipse-RegisterBuddy otherwise the
> runtime plugin cannot see the classes.
>
> A second way you can take is, create a pear bundle, install it
> and copy the pear bundle descriptor over to the processing
> folder.

Thanks for the information, I will try these out over the week-end...

> Hopefully we will improve the situation with the next release, I would love
> to have some kind of an eclipse plugin which supports the development,
> running and debugging of AEs.

A way to specify the Jar to be loaded at runtime like the Build Path
entry in Java projects would be ideal.

--
Fabien Poulard

Re: Running AE from the CAS Editor view

[Jörn Kottmann <ko...@gmail.com>]

Hello,

the WhitespcaceTokenizer or any other annotator must
be in the classpath of the UIMA runtime plugin. Right
now one way to achieve that is to make an OSGi bundle
out of your annotator and deploy it to your eclipse installation.
The bundle must use Eclipse-RegisterBuddy otherwise the
runtime plugin cannot see the classes.

A second way you can take is, create a pear bundle, install it
and copy the pear bundle descriptor over to the processing
folder.

Hopefully we will improve the situation with the next release, I would love
to have some kind of an eclipse plugin which supports the development, 
running
and debugging of AEs.

Jörn

Fabien POULARD wrote:
> Hi folks,
>
> I've been trying to run the WhitespaceTokenizer AE (for example) on
> the CAS documents of a Corpus folder within the CAS Editor for some
> time now, with no success. I run into a class not found exception each
> time, indicating that the WhitespaceTokenizer class cannot be found.
>
> I don't know I'm overlooking on something here or if I'm just dumb...
>
> I'm using Apache UIMA 2.3, the CAS Editor plugin from the repository
> and Eclipse 3.3.2
>
> Here is what I've tried :
>
> 1- The simple way : create the CAS Editor project, import type system,
> create a corpus directory, import text files, create processing
> directory and import the WhitespaceTokenizer (let's says WT from now)
> descriptor in it. Finally click on a document in the corpus directory
> : Annotator > WT.xml => it fails
>
> 2- The improved way : do the same as the simple way, then open the
> properties > UIMA CDE Property page and specify the absolute path to
> the WT jar. => it fails the same way
>
> 3- The hacky way : create a Java project and specify the Build path so
> that the WT is loaded, edit the .project file, add the NLPProject
> nature, open the CAS Editor perspective and redo the simple way => it
> still fails
>
> Well, I'm out of ideas... does anybody have a solution to my problem ?
> Has anybody already been able to launch an AE on a corpus from the CAS
> Editor ?
>
> I'd be grateful to any hint !
>
> --
> Fabien Poulard
>