search

SynBioDex / pySBOL3

Native python implementation of SBOL 3.0 specification
MIT License
37 stars 16 forks source link

Support parse_namespace from a toplevel object #205

Closed tramyn closed 3 years ago

tramyn commented 3 years ago

I want to get a namespace from toplevel object. @bbartley mentioned that pysbol2 supports this feature by peforming the following:

for obj in doc.objects:
    ns = sbol2.utils.parse_namespace(obj.identity)

I would like to request this feature supported in pysbol3.

tcmitchell commented 3 years ago

pysbol2 does not have an exported parse_namespace or parseNamespace function. There is also no utils submodule in pysbol2. There is a parseNamespace function internal to pysbol2, however. The reason it is not exported is that it is a heuristic that might or might not properly determine the namespace of a URI. As such, its use is limited internally to pysbol2, and is not exported from the pysbol2 module.

If you want a function that heuristically guesses a namespace from an object's URI, you'll want to define that function yourself. You could use one of the two definitions in pysbol2 as a basis for yours or do it a different way. The fact that pysbol2 has two different parseNamespace functions internally demonstrates that this is a heuristic that just isn't a good idea. The two definitions are:

I have no plans to support a namespace parsing function in pysbol3. Section 5.1 of the SBOL3 specification says that TopLevel URLs use the following pattern: [namespace]/[local]/[displayId]. There is no way to determine when the namespace ends and the "local" portion of the URL/URI begins, or if the "local" portion even exists.

Following the URL/URI example in Section 5.1, which is https://synbiohub.org/public/igem/BBa_J23070, the two parseNamespace functions in pysbol2 would probably say that the namespace is https://synbiohub.org/public/igem and they would be wrong. The spec says the namespace is https://synbiohub.org. The namespace could just as easily be https://synbiohub.org/public too. This confusion is why it is a bad idea for pysbol2 and pysbol3 to export a parse_namespace function.

This might get better with the new Namespace definition in the upcoming SBOL3 3.0.1 specification. But until that is published we won't really know.

jakebeal commented 3 years ago

I believe that with 3.0.1, we'll end up with a namespace field on each object, which will make it just a field access. We should formalize this by the end of HARMONY

tcmitchell commented 3 years ago

Hearing no reply I will close this issue. If you have any questions about this, please reopen and add a comment.