On 25/11/14 23:49, Bruno P. Kinoshita wrote:
Hi all,
Yesterday I was lurking in #jena's channel as I often do, when a user posted
that his query
PREFIX dbpedia-owl: <http://dbpedia.org/ontology/>PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>PREFIX rdf:
<http://www.w3.org/1999/02/22-rdf-syntax-ns#>PREFIX res: <http://www.w3.org/2005/sparql-results#>describe * where
{?movie rdf:type dbpedia-owl:Film.?movie rdfs:label ?label.?movie rdfs:label ?movie_title.FILTER langMatches(lang(?movie_title),
"EN") .filter regex( ?label, "Inglourious Basterds", "i").}
against DBpedia SPARQL endpoint was failing with
Exception in thread "main" com.hp.hpl.jena.sparql.resultset.ResultSetException:
Expected only object keys [type, value, xml:lang, datatype] but encountered 'lang'
I got the same Exception in my Eclipse workspace. While debugging with
jena-2.12.0, I found that the Exception occurred in this code in jena-arq
JSONInput [1]
Now, what intrigued me, was that in JSONResultsKW [2], there are constants that are used to filter
the keys returned in JSON, IIUC. When the error is raised, these possible values include
"xml:lang", as in the W3C documentation. However, DBpedia Virtuoso seems to be using
"lang" as the key, without the xml prefix [3].
Is it a bug in arq, is DBpedia endpoint not following the specification, or is
the error somewhere else? Reading the W3C page it doesn't mention describe
queries, so not sure if it doesn't apply for this kind o query. Replacing
describe by select the query seems to works (didn't look what's happening with
his filter).
Here's the code he used [4]
Thanks!
ps: sent it hours ago, but I couldn't find it in the archives so re-sending it
now.
[1]
https://github.com/apache/jena/blob/6ed1c766e74861db3a2219f9e7eb84ecc1ea8ef0/jena-arq/src/main/java/com/hp/hpl/jena/sparql/resultset/JSONInput.java#L266
[2]
https://github.com/apache/jena/blob/6ed1c766e74861db3a2219f9e7eb84ecc1ea8ef0/jena-arq/src/main/java/com/hp/hpl/jena/sparql/resultset/JSONResultsKW.java#L23
[3]
http://dbpedia.org/sparql?default-graph-uri=http%3A%2F%2Fdbpedia.org&query=PREFIX+dbpedia-owl%3A+%3Chttp%3A%2F%2Fdbpedia.org%2Fontology%2F%3E%0D%0APREFIX+rdfs%3A+%3Chttp%3A%2F%2Fwww.w3.org%2F2000%2F01%2Frdf-schema%23%3E%0D%0APREFIX+rdf%3A+%3Chttp%3A%2F%2Fwww.w3.org%2F1999%2F02%2F22-rdf-syntax-ns%23%3E%0D%0APREFIX+res%3A+%3Chttp%3A%2F%2Fwww.w3.org%2F2005%2Fsparql-results%23%3E%0D%0Adescribe+*+where++%7B%0D%0A%3Fmovie+rdf%3Atype+dbpedia-owl%3AFilm.%0D%0A%3Fmovie+rdfs%3Alabel+%3Flabel.%0D%0A%3Fmovie+rdfs%3Alabel+%3Fmovie_title.%0D%0AFILTER+langMatches%28lang%28%3Fmovie_title%29%2C+%22EN%22%29+.%0D%0Afilter+regex%28+%3Flabel%2C+%22Inglourious+Basterds%22%2C+%22i%22%29.%0D%0A%7D&format=application%2Fsparql-results%2Bjson&timeout=30000&debug=on
[4] http://rdf.pastebay.net/1531054
(can't see that code)
[3] is a SPARQL Result set, for SELECT queries. Format
application/sparq-results+json.
The request is a DESCRIBE query, not a SELECT
The Virtuoso request has "format=application/sparql-results+json"
Setting the Accept header has the same effect.
A DESCRIBE query is an RDF graph and conneg should be for any RDF graph
format. JSON-LD or RDF/JSON [A] being JSON possibilities for encoding
an RDF graph in JSON.
ARQ expects a graph to a graph request (DESCRIBE, CONSTRUCT). A result
set is not a graph.
Virtuoso sends a result set of (s,p,o) rows not a graph but you don't
know it's (s,p,o) from the response. (Somewhat tricky because you have
to open the results first.) Looks like a virtuoso special.
Content negotiation should have generated a "406 Not Acceptable" - the
request is not compatible with the resource. 418 is also acceptable.
Andy
[A] Non standard but reasonably common
We have a copy of the documentation as the Talis site isn't available
any more.
http://jena.apache.org/documentation/io/rdf-json.html
