Though we can exclude some of the dependencies directly in our pom.xml,
that can work too.
We'll experiment a bit with which dependencies are absolutely needed for
tika-parsers and which may be excluded.
Any feedback will be apreciated
Thanks, Sergey
On 17/06/14 21:16, Sergey Beryozkin wrote:
Hi
A CXF colleague of mine started experimenting with providing a
light-weight search utility code as part CXF Search API which would use
a Lucene handler (shipped with CXF for a while) which can translate FIQL
or ODATA queries into a composite Lucene Query and use it against Tika
provided metadata and content. The idea is not new, I believe SOLR does
some very advanced Tika based search. In CXF users would use it as part
of their regular JAX-RS applications.
The problem seems to be that Tika Parsers module contains many
dependencies that may not be needed by a specific custom JAX-RS
application.
For example, we'd expect a given application dealing with PDF only, or a
certain set of image formats only, or word docs only, etc.
I'm not sure how many Tika-parsers dependencies are strongly required
for any Tika application and which can be made optional.
If Tika Parsers does have some possibly optional dependencies then would
it make sense to make them as such for external Tika consumers having
not to download all the deps ? It would make a difference IMHO
Thanks, Sergey
