SPARQL Update insert into default graph does not work for Dataset

[leng-lim]

Using SPARQL Update INSERT or INSERT DATA statements with a rdflib.Dataset does not work as expected when inserting into the default graph. This works as expected when using a rdflib.Graph. The following example code shows the problem:

# The SPARQL Update INSERT statement.
sparql_update = """
    PREFIX book: <http://example.org/book/>
    PREFIX dc: <http://purl.org/dc/elements/1.1/>
    
    INSERT DATA {
           book:book1 dc:title "War of the Worlds" .
    }"""

# Initialise a Dataset and Graph.
ds: rdflib.Dataset = rdflib.Dataset()
g: rdflib.Graph = rdflib.Graph()

# Translate the update.
update = translateUpdate(parseUpdate(sparql_update))

# Apply the update.
ds.update(update)
g.update(update)

# Check inserted triples.
print(f"Dataset default graph triples - {len(list(ds.graph('urn:x-rdflib:default').triples((None, None, None))))}")
print(f"Graph triples - {len(list(g.triples((None, None, None))))}")

This gives me the following output:

Dataset default graph triples - 0
Graph triples - 1

I would expect that the default graph with the identifier urn:x-rdflib:default in the rdflib.Dataset would contain 1 triple. Furthermore, if I try and serialise the Dataset to trig using print(ds.serialize(format="trig")), I get the following error:

self = <rdflib.plugins.serializers.trig.TrigSerializer object at 0x7f8bdb920850>
spo = (rdflib.term.URIRef('http://example.org/book/book1'), rdflib.term.URIRef('http://purl.org/dc/elements/1.1/title'), rdflib.term.Literal('War of the Worlds'), rdflib.term.BNode('N60d1b8ccffe6435fbc96e39e79713cf7'))

    def preprocessTriple(self, spo):
>       s, p, o = spo
E       ValueError: too many values to unpack (expected 3)

../../textfsmgui/.venv/lib64/python3.7/site-packages/rdflib/plugins/serializers/turtle.py:105: ValueError

I don't have these problems if I use a SPARQL Update INSERT command with a named graph e.g:

# The SPARQL Update INSERT statement with a named graph.
sparql_update = """
    PREFIX book: <http://example.org/book/>
    PREFIX dc: <http://purl.org/dc/elements/1.1/>
    
    INSERT DATA {
          GRAPH <http://example.org/library> {
              book:book1 dc:title "War of the Worlds" .
          }
    }"""

This is happening with rdflib 6.1.1

Thanks for the report, the problem is related to RDFLib's “needs-work” implementation of Dataset and is related to a long-ago legacy approach to context identifiers.

Happily this issue has been addressed in the (still needs finalizing) Dataset re-work PR but, because it's an API-breaking change, the Dataset re-work won't be released until RDFLib 7.0 (scheduled for October-or-so).

[leng-lim]

Thanks for the update Graham.