You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stanbol.apache.org by florent andré <fl...@4sengines.com> on 2012/01/23 18:02:14 UTC
Test Camel jobManager
Hi there !
Here comes a first implementation of Camel/Stanbol integration.
There is plenty rooms for improvments, but gave a first idea.
You will find the branch here :
https://svn.apache.org/repos/asf/incubator/stanbol/branches/cameltrial/
Changes are :
* in /enhancer : modify the engine endpoint to take care of route/chain
& add the cameljobmanager
* in /launchers : add a camellauncher
°°°°°° build and start °°°°°°
To try do "mvn3 clean install" in /enhancer and /launchers/camellauncher
start as usual.
REMARK : as configured this will create 2 folders in your /tmp folder :
- chaininput : continuously scanned folder for text file that have to be
enhance
- chainoutput : results of processing in an rdf file.
°°°°°° use it °°°°°°
- A default route/chain is define. This default chain do the same like
the weightedjobmanager, it's call by the web-interface.
- Others routes are defined here [1].
- They can be fired with classical engine's REST api, just need to add
the chainName at the end of the url :
http://localhost:8080/engines/{chainName} [2]
So you have this "can be call" chains :
1) metaxa : just call metaxa engine send output to curl
2) metaxa2 : call metaxa then langidEngine sent output to curl *and*
create an rdf file or the result
3) chainlink : call metaxa then another defined route (here the default
one, weighted chain)
And this "pool" chain :
It's a chain that scan files in /tmp/chaininput, process this files and
put the rdf output in /tmp/chainoutput.
Files are delete from chaininput folder, but that's a choice, can be
configure to keep them in place (add a noop=true parameter to Camel
config url).
For now only plain text are accepted but it's just a matter of adding a
Tika mimetype detector to get it enabled for any mimeType (but still
rely on metaxa for extraction).
++
[1] :
https://svn.apache.org/repos/asf/incubator/stanbol/branches/cameltrial/enhancer/jobmanager/cameljobmanager/filepoolchain/src/main/java/org/apache/stanbol/enhancer/jobmanager/defaultRoute/FileRoute.java
[2] exemples :
curl -X POST -H "Accept: text/turtle" -H "Content-type: text/plain"
--data "Here comes a little test with Paris as content and also Berlin
but why not detect city as Boston." http://localhost:8080/engines/metaxa
curl -X POST -H "Accept: text/turtle" -H "Content-type: text/plain"
--data "Here comes a little test with Paris as content and also Berlin
but why not detect city as Boston." http://localhost:8080/engines/chainlink