Partially fix prefix injection when writing to OBO. #1270
Conversation
When writing to OBO, prefixes added with `--add-prefix` are ignored. This is because until recently, the OBODocumentFormat was not a prefix-supporting format, and so there is nothing in the OWLAPI to automatically take care of providing the prefixes the document may contain into the OWL2OBO translator. This commit fixes the issue at least for the case where the `--clean-obo` option is used. In that case, we are instantiating the OWL2OBO translator ourselves, so we can give it access to the prefixes from the document. For the case where `--clean-obo` is not used, the fix needs to happen in the OWLAPI. In that case, we are relying on the ontology manager's save methods, and we have no way to reach the OWL2OBO translator that is, in fine, used by those methods when writing to OBO. (It would be possible to always bypass the manager when writing to OBO, regardless of the `--clean-obo` option, but that would be a more invasive fix).
matentzn
left a comment
matentzn
left a comment
There was a problem hiding this comment.
Thank you for the great explanation in particular why the fix must be partial. Even if a full solution would be attempted in the future, this change here seems good. Thank you!
Just in case it is not completely clear: the “partial fix” here will still be needed anyway even after the issue will have been fixed in the OWLAPI. We have two paths to export ontologies to OBO: one that is fully managed by the OWLAPI, and one that is mostly managed by ROBOT (when What I am trying to say here is that this PR is not a “temporary fix” or a workaround until a better solution is implemented in the OWLAPI. It is a part of the definitive solution (the part that is within ROBOT’s hands). |
|
Thanks! |
3 participants
Partially resolves [#1268]
docs/have been added/updatedmvn verifysays all tests passmvn sitesays all JavaDocs correctCHANGELOG.mdhas been updatedWhen writing to OBO, prefixes added with
--add-prefixare ignored. This is because until recently, the OBODocumentFormat was not a prefix-supporting format, and so there is nothing in the OWLAPI to automatically take care of providing the prefixes the document may contain into the OWL2OBO translator.This PR fixes the issue at least for the case where the
--clean-obooption is used. In that case, we are instantiating the OWL2OBO translator ourselves, so we can give it access to the prefixes from the document.For the case where
--clean-obois not used, the fix needs to happen in the OWLAPI. In that case, we are relying on the ontology manager's save methods, and we have no way to reach the OWL2OBO translator that is, in fine, used by those methods when writing to OBO. (It would be possible to always bypass the manager when writing to OBO, regardless of the--clean-obooption, but that would be a more invasive fix).