turtle serialization breaks on import

[satra]

i'm serializing a graph with about 16000 statements. the graph serializes to turtle fine, but doesn't read the output back in properly

In [3]: rdfgraph.serialize('outfile.ttl', format='turtle')

In [5]: g = rdflib.Graph().parse('outfile.ttl', format='turtle')
  File "<string>", line unknown
BadSyntax

no error happens if i serialize to rdf+xml or nt.

[joernhees]

sorry for the late reply, but you could've realized that it's hard to debug this with more info... please provide the files of the graph serialized as turtle and xml and your rdflib version

[theoryno3]

What are the contents of your graph? Could you provide more verbose information about its construction? Cheers.

[satra]

the graph is an automated construction from a workflow program, so it's really large. here is an example:

https://dl.dropbox.com/s/an6qyhf9yfvn634/outfile.ttl

but the key here is the roundtrip. i use rdflib to create this graph, but it won't read it back in.

[ghost]

Thanks for that - it was absolutely vital.

Confirmed as a bug in the RDFLib notation3 parser. For a workaround, switch temporarily to using the rdf/xml format for transport.

Despite the misleading error message, it's an issue with the notation3 parser's handling of the namespace localpart, to wit.:

import rdflib

problemlocalname = '''\
@prefix fs: <http://freesurfer.net/fswiki/terms/> .
@prefix prov: <http://www.w3.org/ns/prov#> .

<http://nidm.nidash.org/iri/82b79326488911e3b2fb14109fcf6ae7> a fs:stat_header,
        prov:Entity ;
    fs:mrisurf.c-cvs_version "$Id: mrisurf.c,v .. abridged .. Exp $" .
'''

FS = rdflib.Namespace('http://freesurfer.net/fswiki/terms/')
g = rdflib.Graph()
g.bind('fs', str(FS))
g.add((
    rdflib.URIRef('http://example.org'),
    FS['mrisurf.c-cvs_version'],
    rdflib.Literal("irrelevant")))
turtledump = g.serialize(format="turtle").decode('utf-8')
g1 = rdflib.Graph()
g1.parse(data=turtledump, format="turtle")

Looks like the problem could be in the code for notation3.SinkParser.path

https://github.com/RDFLib/rdflib/blob/master/rdflib/plugins/parsers/notation3.py#L719

Now, if I only knew what a "path production" was ....