Hi Bob
On 17/05/16 02:37, Bob Paulin wrote:
Sergey,
Great to hear the code works well with the new modules! And I do agree
that Tika has a number of application specific usecases that can be
explored. I think the other goal is making the upgrade paths easier so
developers don't have to drag "JAR Hell" with them into their projects.
It was good to see in your commit you got to remove the maven exclusions
as well. I think you can also remove the explicit tika-core entry as
that should be a transitive dependency of any of the modules.
The reason I has to list it explicitly is that a tika-core has a
provided scope in a CXF module where Tika extensions are shipped,
indeed, I was surprised yesterday when I saw:
Caused by: java.lang.ClassNotFoundException:
org.apache.tika.exception.AccessPermissionException
and then it was gone after I added a tika-core dependency, and it took
me awhile to realize why I had to do it. I guess we might have a
dedicated CXF Tika module introduced later on with the strong
dependencies...
This type
of working is a huge help in moving towards the 2.0 release.
Very minor effort on my part given all the work which has already been
done in Tika 2.0 :-)
I think the next steps should be determining if there are any other
breaking features we'd like to include in 2.0 and perhaps we can get Tim
to run the 2.x branch through his massive regression test :).
Cheers, Sergey
- Bob
On 5/16/2016 10:21 AM, Sergey Beryozkin wrote:
Hi All
Hope this message will be more relevant compared to the one I posted
after a social event at Apache Con NA 2016 :-). I had a chance to talk
to Nick and Bob the next day and we agreed it would be good to have
Tika 2.0-SNAPSHOT tested a bit more. Specifically I committed to
updating a Tika-based demo we ship in Apache CXF to use 2.0-SNAPSHOT
module dependencies - no pressure is expected on CXF master in the
short term given that the master release won't happen in the next few
months for sure.
FYI, in CXF we ship this demo:
https://github.com/apache/cxf/tree/master/distribution/src/main/release/samples/jax_rs/search
IMHO it is a very cool demo written by my CXF colleague Andriy Redko.
This demo was part of his NA 2015 presentation:
http://events.linuxfoundation.org/sites/events/files/slides/Apache%20CXF%2C%20Tika%20and%20Lucene.pdf
Here is a demo description: a user can upload PDF or ODT files to a
JAX-RS service using an HTML form. The uploaded files are submitted to
a CXF Tika extensions:
https://github.com/apache/cxf/tree/master/rt/rs/extensions/search/src/main/java/org/apache/cxf/jaxrs/ext/search/tika
with this code:
https://github.com/apache/cxf/blob/master/distribution/src/main/release/samples/jax_rs/search/src/main/java/demo/jaxrs/search/server/Catalog.java#L115
where the Tika reported content/metadata is saved with Lucene.
Next a user enter a search phrase and finds matching documents, with
the links to them being reported so that a user can download it.
IMHO it is an interesting demo because it shows how Tika can help in
some application specific situations...
Finally, to the actual experiment I did today. Updating the demo to
use individual parser modules was easy:
http://git-wip-us.apache.org/repos/asf/cxf/commit/c2ccecb2
All works well, better modularization in 2.0 will be welcomed
Thanks, Sergey
