Raise ValueError for empty string namespace prefix

[ghost]

Hiya, I'm getting an invalid namespace prefix error. While I understand the error entirely, it is being raised at the point of serialisation, and in the circumstance that I have defined several namespaces programmatically, I do not know which one isn't being filled properly.

Would it be better/possible to check for namespace validity within the add_namespace() method, to follow the principle of failing fast/early?

[trungdong]

Hello, could you provide an example where it failed? I'm not sure what issue you are having at the moment.

[NathanCummings]

I'm filling a ProvDocument programmatically by reading the metadata. In the event that someone hasn't filled a field, I am trying to define a namespace with an empty string. When this happens, I don't get an exception thrown until I come to serialise the document. Now of course, I have encountered the problem and built a check into my code, but in a more complex pipeline, the error I received and the point in the code that it was thrown did not make it clear where the problem was. An earlier check on the call to add_namespace() would have told me which namespace was invalid.

Traceback (most recent call last):
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/bin/imasprov", line 11, in <module>
    load_entry_point('fusionprov==0.1.4', 'console_scripts', 'imasprov')()
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/fusionprov/imasprov.py", line 225, in main
    ids_prov.prov_from_data_ids(graph=args.graph)
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/fusionprov/imasprov.py", line 197, in prov_from_data_ids
    self.serialize_provenance(graph=graph)
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/fusionprov/imasprov.py", line 139, in serialize_provenance
    self.prov_doc.serialize(xml_path, format="xml")
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/prov/model.py", line 2494, in serialize
    serializer.serialize(stream, **args)
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/prov/serializers/provxml.py", line 52, in serialize
    xml_root = self.serialize_bundle(bundle=self.document, force_types=force_types)
  File "/afs/eufus.eu/g2itmdev/user/g2ncummi/fusionprov/venv/lib/python3.6/site-packages/prov/serializers/provxml.py", line 110, in serialize_bundle
    xml_bundle_root = etree.Element(_ns_prov("document"), nsmap=nsmap)
  File "src/lxml/etree.pyx", line 3022, in lxml.etree.Element
  File "src/lxml/apihelpers.pxi", line 131, in lxml.etree._makeElement
  File "src/lxml/apihelpers.pxi", line 118, in lxml.etree._makeElement
  File "src/lxml/apihelpers.pxi", line 219, in lxml.etree._setNodeNamespaces
  File "src/lxml/apihelpers.pxi", line 1749, in lxml.etree._prefixValidOrRaise
ValueError: Invalid namespace prefix ''

In a script with multiple namespaces defined, the above traceback does not indicate where the problem occurred, as the error isn't caught until much later.

[trungdong]

Thanks for the information, Nathan. I'll pop this test in the code next time.