Add support for named graphs

[flamingofugang]

I just modified the 'fuseki_example.ttl' a little bit to make it point to two hdt files I have generated. But I got the following error message when I tried to start the fuseki with config file:

[fug2@virtuosodev11 hdt-fuseki]$ bin/hdtEndpoint.sh --config=fuseki_example.ttl
com.hp.hpl.jena.assembler.exceptions.AssemblerException: caught: Adjacency list bitmap and array should have the same size
doing:
root: file:///home/fug2/hdt-java/hdt-fuseki/fuseki_example.ttl#graph1 with type: http://www.rdfhdt.org/fuseki#HDTGraph assembler class: class org.rdfhdt.hdtjena.HDTGraphAssembler
root: file:///home/fug2/hdt-java/hdt-fuseki/fuseki_example.ttl#dataset with type: http://jena.hpl.hp.com/2005/11/Assembler#RDFDataset assembler class: class com.hp.hpl.jena.sparql.core.assembler.DatasetAssembler

---

the changed fuseki_example.ttl file is as follow:

<#graph1> rdfs:label "RDF Graph1 from HDT file" ;
        rdf:type hdt:HDTGraph ;
        hdt:fileName "/export/home/SSD/BIGDATA/hdt/pc_compound_0.hdt" ;
    .
<#graph2> rdfs:label "RDF Graph2 from HDT file" ;
        rdf:type hdt:HDTGraph ;
        hdt:fileName "/export/home/SSD/BIGDATA/hdt/pc_compound_1.hdt" ;
    .

[mielvds]

How did you generate the HDT files?

[flamingofugang]

I built 'hdt-cpp' with 'serd' and used the tool 'rdf2hdt'. I first concat multiple gz files and gunzip them and did the convertion:
cat /path-to-my-files/*.gz | gunzip | /<path>/hdt-cpp/hdt-lib/tools/rdf2hdt -f turtle /dev/stdin output.hdt

[flamingofugang]

I have tried the other way: convert turtle format to nt format first using serdi, then convert nt file into hdt. The fuseki server can start now.

I am starting fuseki server with three hdt files. The server can start and when I run
select distinct ?g where {graph ?g {?s ?p ?o}}
It can correctly return three graphs. But when I try to query one graph, nothing comes back, for instance:
select *
from
where {?s ?p ?o}
limit 100

no records returned...

Any comment?

[flamingofugang]

To generalize my question, how to configure hdt-fuseki to allow sparql query against named graph.

[flamingofugang]

here is part of my config file:

[] rdf:type fuseki:Server ;
   ja:loadClass "org.rdfhdt.hdtjena.HDTGraphAssembler" ;

   fuseki:services (
     <#service1>
   ) .
<#service1> rdf:type fuseki:Service ;
    fuseki:name                     "hdt" ;
    fuseki:serviceQuery             "query" ;
    fuseki:serviceReadGraphStore    "get" ;
    fuseki:dataset                   <#dataset> ;

<#dataset> rdf:type ja:RDFDataset ;
    rdfs:label "Dataset" ;
    ja:namedGraph
       [ ja:graphName  <http://rdf.ncbi.nlm.nih.gov/pubchem/compound7> ;
          ja:graph      <#graph7> ] 
   .


<#graph7> rdfs:label "RDF Graph7 from HDT file" ;
        rdf:type hdt:HDTGraph ;
        hdt:fileName "/export/home/SSD/BIGDATA/hdt/pc_compound_7.hdt" ;
    .

[flamingofugang]

It seems to me that 'hdt-fuseki' can only work with default graph instead of named graph, am I missing something?

[flamingofugang]

It seems FROM and FROM NAMED clauses do not work, but if I do
select * where {graph ?g {?s ?p ?o}} limit 100
it does work

[artob]

Retitled the ticket to reflect the remaining issue at hand.

[bryaakov]

👍
we would love to see this enhancement in HDT, as we use named graphs in production.

[D063520]

@QuentinJanuel: hola, this is the issue and here is an example of data, just take a subset for it ....

https://data.dws.informatik.uni-mannheim.de/structureddata/2022-12/quads/

Also Antoine just pushed code to parse quads. So this part is fine : )