CODEWITHSUNDEEP
sparqlrdfmarklogic
uniqueR
uniqueR

Reputation: 51

Numbers returned by MarkLogic do not use scientific notation

Write the following RDF data into MarkLogic:

<http://John> <http://have> 2.1123519E7 .

Then execute the following query:

SELECT *
WHERE { ?s ?p ?o. }

The query result is:

?s ?p ?o
<http://John> <http://have> "21123519"^^xsd:double

Expected query result:

The expected query result is:

?s ?p ?o
<http://John> <http://have> "2.1123519E7"^^xsd:double

Alternatives

The written data is expressed in scientific notation, but the data returned in query result is not in scientific notation. Will there be some inconsistencies?

Executing the SparQL query with Java Client API will get the unexpected query result. But executing it in Query Console will get the expected query result.

This query can also return the expected query result in Apache Jena and RDF4j.

Can someone give me an answer or a hint about it?

Upvotes: 2

Views: 55

Answers (1)

David Ennis -CleverLlamas.com
David Ennis -CleverLlamas.com

Reputation: 7770

As Mads points points out: the numbers are the same.

However, this is an interesting one.

MarkLogic does keep and understand the scientific notation. The REST API also handles this for many return types. I tested your query against the /v1/sparql endpoint:

Accept Header: application/sparql-results+xml

    ...
    <result>
        <binding name="s">
            <uri>http://John</uri>
        </binding>
        <binding name="p">
            <uri>http://have</uri>
        </binding>
        <binding name="o">
            <literal datatype="http://www.w3.org/2001/XMLSchema#double">2.1123519E7</literal>
        </binding>
    </result>
   ...

** Accept Header: text/csv **

s,p,o
http://John,http://have,2.1123519E7

Same for HTML ETC..

However, for JSON, things are different:

            {
            "s": {
                "type": "uri",
                "value": "http://John"
            },
            "p": {
                "type": "uri",
                "value": "http://have"
            },
            "o": {
                "datatype": "http://www.w3.org/2001/XMLSchema#double",
                "type": "literal",
                "value": "21123519"
            }
        }

This matches the fact that the scientific notation appears to be lost as double for JSON:

JSON.parse('{ "foo" : 2.1123519E7}')

//return:
{
    "foo": 21123519
}

So, it all comes down to how you are requesting your results in your call to MarkLogic. Some response types return what you expect. At least one (JSON) does not. At this point, I suggest opening a ticket under the Java API project: https://github.com/marklogic/java-client-api/issues

Upvotes: 3

Related Questions