check out the `curies` package

[cthoyt]

the curies python package has a lot of the utilities for compressing URIs down to "CURIEs" for semantic web applications. it might help reduce some of the code you're writing

[miselico]

Cool! I probably won't need more than what I did now, as the next tool will convert all identifiers to numeric ones anyway. This was quite specific for this dataset to get the filesizes down a bit (it also generates some invalid n-triples). But, if I encounter a similar problem, I'll use the library instead.