Multiple values in single CSV column

[milvld]

Hi

I'm trying to convert a CSV in which a single column/cell can contain multiple values separated by a dedicated delimiter (in this case a semicolon). Instead of resulting in a single property with the complete column's content as the value, I want two properties with two separate values.

I'm currently using Apache Jena's strsplit function, as has previously been discussed in the issues, cf. #197 . However, this results in no output. When I take the facade-x output and then try to transform that using regular SPARQL via Jena, everything works as expected.

Any pointers as to what is going on?

Sample CSV

terms,values
term1,val1
term2,val2;val3
term3,val4

SPARQL Anything query
With the file being given as input via command: java -jar ../../sparql_anything/sparql-anything-0.8.1.jar -q ./test_strsplit2.rq -v file=./test_strsplit.csv

PREFIX apf: <http://jena.hpl.hp.com/ARQ/property#>
PREFIX fx: <http://sparql.xyz/facade-x/ns/>
PREFIX myns: <http://example.org/myns/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xyz: <http://sparql.xyz/facade-x/data/>

CONSTRUCT {
    ?term myns:value    ?value .
}
WHERE {
    SERVICE <x-sparql-anything:csv.headers=true>
    {
        fx:properties   fx:location       ?_file ;
                        fx:media-type "text/csv" ;
                        fx:csv.null-string          "" .

        ?c  xyz:terms ?term_temp;
            xyz:values  ?value_temp.
        ?value apf:strSplit    (?value_temp ";") .
        BIND (IRI(CONCAT("http://example.org/myns/", ?term_temp)) AS ?term)
    }
}

Expected output

@prefix apf:  <http://jena.hpl.hp.com/ARQ/property#> .
@prefix fx:   <http://sparql.xyz/facade-x/ns/> .
@prefix myns: <http://example.org/myns/> .
@prefix rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xyz:  <http://sparql.xyz/facade-x/data/> .

myns:term2  myns:value  "val2","val3" .

myns:term1  myns:value  "val1" .

myns:term3  myns:value  "val4" .

Current output

@prefix apf:  <http://jena.hpl.hp.com/ARQ/property#> .
@prefix fx:   <http://sparql.xyz/facade-x/ns/> .
@prefix myns: <http://example.org/myns/> .
@prefix rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xyz:  <http://sparql.xyz/facade-x/data/> .

[enridaga]

This seems to work:

PREFIX apf: <http://jena.hpl.hp.com/ARQ/property#>
PREFIX fx: <http://sparql.xyz/facade-x/ns/>
PREFIX myns: <http://example.org/myns/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xyz: <http://sparql.xyz/facade-x/data/>

CONSTRUCT {
    ?term myns:value    ?value .
}
WHERE {
    SERVICE <x-sparql-anything:csv.headers=true>
    {
        fx:properties   fx:location       ?_file ;
                        fx:media-type "text/csv" ;
                        fx:csv.null-string          "" .

        ?c  xyz:terms ?term_temp ;
            xyz:values  ?value_temp .
        BIND (IRI(CONCAT("http://example.org/myns/", ?term_temp)) AS ?term) .
    }
    ?value apf:strSplit    (?value_temp ";") .
}

Curious, it may be possible that the apf magic property is not properly processed within the SERVICE clause; this needs double-checking.

[luigi-asprino]

Curious, it may be possible that the apf magic property is not properly processed within the SERVICE clause; this needs double-checking.

It is possible as the SERVICE clause is supposed to be executed by another endpoint which couldn't implement the the apfs properties. I faced the same issue while implementing fx:anySlot.

[milvld]

Hi @enridaga , thanks for the quick reply, that works indeed! I'm closing the issue.