Fix a KeyError in N-Quads normalization

[rspeer]

I wanted to try using this package with my new JSON-LD API, with URLs such as http://api.conceptnet.io/c/en/example .

It seems to be able to do simple transformations of the data, but it fails when I ask it to convert to N-Quads format:

>>> from pyld import jsonld
>>> jsonld.to_rdf('http://api.conceptnet.io/c/en/example', options={'format': 'application/nquads'})
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 295, in to_rdf
    return JsonLdProcessor().to_rdf(input_, options)
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1147, in to_rdf
    return self.to_nquads(dataset)
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1562, in to_nquads
    quads.append(JsonLdProcessor.to_nquad(triple, graph_name))
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1617, in to_nquad
    if o['language']:
KeyError: 'language'

Changing this to o.get('language') seems to fix the problem, so that's what this patch does.

This patch passes the tests in normalization/tests. It doesn't pass json-ld.org/test-suite, but it seems the master branch doesn't either.

[davidlehn]

How about if 'language' in o:?

[rspeer]

I don't know what this check is actually for, but o['language'] is assuming the key exists and testing whether its value is falsy, which is why I changed it to o.get('language'), which is also falsy (None) if the key doesn't exist.

'language' in o would test only whether the key exists, which is orthogonal to what it was doing before. That doesn't necessarily mean it's wrong, though.

[anatoly-scherbakov]

From the discussion, it seems the PR should detect an incorrect form of the context. Currently, it will just ignore the issue.

[davidlehn]

There should be a test that runs this code path to avoid regressions. This was probably converted from js and there's never been a test run that didn't have the "language" key set. Do you have minimal JSON-LD and expected output that causes this issue?

[rspeer]

Aha! I've noticed that this is a bug in my context file -- I declared a value as a langString when it's actually just a string. But the library should tell me that there's a problem with my context, instead of just crashing with a KeyError.

Minimal example:

>>> example = {'@context': {'rdf': 'http://www.w3.org/1999/02/22-rdf-syntax-ns#', 'text': {'@id': 'http://example.com/text', '@type': 'rdf:langString'}}, 'text': 'foo'}
>>> jsonld.to_rdf(example, options={'format': 'application/nquads'})
Traceback (most recent call last):
  File "<ipython-input-23-f64152bf1496>", line 1, in <module>
    jsonld.to_rdf(example, options={'format': 'application/nquads'})
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 295, in to_rdf
    return JsonLdProcessor().to_rdf(input_, options)
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1147, in to_rdf
    return self.to_nquads(dataset)
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1562, in to_nquads
    quads.append(JsonLdProcessor.to_nquad(triple, graph_name))
  File "/home/rspeer/.virtualenvs/lum/lib/python3.5/site-packages/pyld/jsonld.py", line 1617, in to_nquad
    if o['language']:
KeyError: 'language'

[anatoly-scherbakov]

From the discussion, it seems the PR should detect an incorrect form of the context. Currently, it will just ignore the issue.