Jérôme wrote:
Le 13/09/11 16:59, Paolo Castagna a écrit :
Jérôme wrote:
Le 12/09/11 17:52, Paolo Castagna a écrit :
Jérôme wrote:
Le 12/09/11 16:13, Paolo Castagna a écrit :
Jérôme wrote:
Le 12/09/11 15:18, Paolo Castagna a écrit :
Jérôme wrote:
Le 12/09/11 12:24, Paolo Castagna a écrit :
Hi Jérôme,
you are lucky, I've just exactly the same need as you and I've
something about it recently.
Unfortunately, the new LARQ (as a separate module) still did
not make it into Fuseki on trunk.
We have an open JIRA for it which you can
watch|vote|contribute to:
https://issues.apache.org/jira/browse/JENA-63
In the meantime, if you want to use LARQ with Fuseki this is
what you need to do:
cd /tmp
svn co
https://svn.apache.org/repos/asf/incubator/jena/Jena2/Fuseki/trunk/
fuseki
cd /tmp/fuseki
wget
https://issues.apache.org/jira/secure/attachment/12482758/JENA-63_Fuseki_r1136050.patch
patch -p0< JENA-63_Fuseki_r1136050.patch
mvn package
Now, you can simply use the Fuseki config.ttl file as
explained here:
http://openjena.org/wiki/Fuseki#Fuseki_Configuration_File
and use the ja:textIndex property on a dataset to specify an
non existing directory.
Is it possible to have a fuseki configuration example with a
ja:textIndex property? I am trying to
add it on the book service (books.ttl) with no results...
Use tdbloader to load some RDF data into /tmp/tdb, then change
<#dataset>
on the example config.ttl file you have in Fuseki:
http://svn.apache.org/repos/asf/incubator/jena/Jena2/Fuseki/trunk/config.ttl
I've never used the TDB loader - How does it work? Is there an
on-line documentation?
Fortunately, TDB is included in Fuseki uber jar (since it includes
Fuseki
binaries as well as all the jar dependencies, including TDB). So,
in this
case, for an end-users it's quite useful.
Here is what I do:
cd /tmp/fuseki
java -cp target/fuseki-0.2.1-SNAPSHOT-sys.jar tdb.tdbloader
--loc=/tmp/tdb books.ttl
Thank you! It's ok for that!
Good.
So, are you able to query your RDF data using the pf:textMatch property
function? For example:
PREFIX pf: <http://jena.hpl.hp.com/ARQ/property#>
PREFIX dc: <http://purl.org/dc/elements/1.1/>
DESCRIBE ?doc {
?title pf:textMatch 'potter' .
?doc dc:title ?title .
} LIMIT 10
I thought, but it doesn't work...
There is no error, but my resultSet is empty.
It's ok for Sparql queries, not for LARQ ones.
How to be sure that my document is well indexed?
I've ran:
java -cp target/fuseki-0.2.1-SNAPSHOT-sys.jar tdb.tdbloader
--loc=/tmp/tdb books.ttl
My end config.ttl file contains:
<#dataset> rdf:type tdb:DatasetTDB ;
tdb:location "/tmp/tdb" ;
ja:textIndex "/tmp/lucene" ;
# Query timeout on this dataset (milliseconds)
ja:context [ ja:cxtName "arq:queryTimeout" ; ja:cxtValue "1000" ] ;
## tdb:unionDefaultGraph true ;
Can you try to delete the /tmp/lucene directory and restart Fuseki?
Ok...fine play, you're right! The Larq query works now!
This is because LARQ will create a Lucene index, indexing content in your
(TDB) dataset if and only if the directory you point it at does not exist.
In an empty index is there already, it will not overwrite it.
Thank you very much.
I am glad it's working.
Paolo
Jérôme
Let me know,
Paolo
I run Fuseki with this command line:
./fuseki-server --config=config.ttl
Now i would like to add modifications in the larq module.
LARQ is open source and you are free and welcome to do so if you
want/need:
cd /tmp
svn co
https://svn.apache.org/repos/asf/incubator/jena/Jena2/LARQ/trunk/ larq
cd /tmp/larq
... make your changes ...
mvn install
Using mvn install Maven will install LARQ artifacts in your local
Maven repository in your home directory.
However, it would be good if you could share what are your
modifications,
why you need them and your use case. Your changes might be useful to
others.
If your changes do not get contributed back, you will need to
maintain them
and they will represent a cost for you. Every time we release a new
version
of LARQ with features you might want, you will need to re-apply your
changes.
Yes i know.
So, I encourage you to share them and maybe open a new JIRA issue
(with a
patch attached to it). Not all changes are general and useful enough
to get
committed, but let's see.
Ok - it could be a good solution for everybody.
I've downloaded and built it. How can i re compile my Fuseki maven
project using my own larq jar?
Once you have published your modified version of LARQ in your local
Maven
repository, it is available to other projects on your machine.
This is how you recompile Fuseki using your modified LARQ jar:
cd /tmp
svn co
https://svn.apache.org/repos/asf/incubator/jena/Jena2/Fuseki/trunk/
fuseki
cd /tmp/fuseki
wget
https://issues.apache.org/jira/secure/attachment/12482758/JENA-63_Fuseki_r1136050.patch
patch -p0< JENA-63_Fuseki_r1136050.patch
mvn package
Make sure LARQ versions in your LARQ pom.xml and Fuseki pom.xml
correspond.
Once again, if you get your changes adopted and committed to trunk
you would
not need to do all this.
Is it explain() the feature you desperately need?
Not really - finally, explain is not what i'm expecting.
Could you share more on why you need it and what is your use case?
The aim is to querying (from an html form) an RDF graph where not are
documents (text documents).
We would like to obtain matching documents (availables from lucene
Hits) + a list of matching String + a list of offset.
Offsets/Positions will be sent to another module that needs the
positions to perform its tasks.
Example of query (case of a LARQ query, queries could be SPARQL only):
Query: w*
doc1: when I was a child I was a Jedi
Expected result:
Doc id: 1
Matching strings: when, was, was
Offsets: 0-3; 7-9;21-23
Do you think it would be interesting to share it?
Jérôme
explain() can be expensive and it is Lucene specific, if could cause
problems if in the future we want to support/move/change and use Solr
and/or ElasticSearch: https://issues.apache.org/jira/browse/JENA-17.
Paolo
Thanks.
This will load the data in books.ttl and build the TDB indexes in
/tmp/tdb
You can also use the -h option for help:
java -cp target/fuseki-0.2.1-SNAPSHOT-sys.jar tdb.tdbloader -h
tdbloader [--desc DATASET | -loc DIR] FILE ...
Location
--loc=DIR Location (a directory)
--tdb= Assembler description file
Symbol definition
--set Set a configuration symbol to a value
--strict Operate in strict SPARQL mode (no
extensions of any kind)
--graph=IRI Act on a named graph
--desc= Assembler description file
General
-v --verbose Verbose
-q --quiet Run with minimal output
--debug Output information for debugging
--help
--version Version information
Paolo
Thanks
[...]
<#dataset> rdf:type tdb:DatasetTDB ;
tdb:location "/tmp/tdb" ;
ja:textIndex "/tmp/lucene" ;
.
If the /tmp/lucene directory does not exist, LARQ will index
what you have in
/tmp/tdb creating the appropriate Lucene indexes.
Paolo
Thanks
LARQ when you point it at a non existing directory will
perform the indexing for you.
This is particularly useful when you have multiple datasets
configured in Fuseki.
WARNING: it might take a while to index large datasets, so be
patient.
See also: http://markmail.org/thread/tmptip55ru5wxrrj
LARQ snapshots are here:
https://repository.apache.org/content/repositories/snapshots/org/apache/jena/larq/0.2.2-incubating-SNAPSHOT/
and I can quickly fix/improve things if you have problems or
good suggestions.
I hope this helps, let me know how it goes.
Paolo
Jérôme wrote:
Hi,
i'm trying to use LARQ with my Fuseki server.
I would like to programmaticaly indexing(with lucene)
documents when the
server starts.
Something like that:
Model model = ModelFactory.createDefaultModel();
IndexBuilderString larqBuilder = new IndexBuilderString();
model.register(larqBuilder);
FileManager.get().readModel(model, "Data/books.ttl");
larqBuilder.closeWriter();
model.unregister(larqBuilder);
index = larqBuilder.getIndex();
LARQ.setDefaultIndex(index);
Is it possible? In which class it would be the best?
Thanks
Jerome
