Bug(graphene): extract_graphene_path doesn't parse URI segment prefixes

[nkemnitz]

At Zetta, we have a number of ChunkedGraphs under the same subdomain, but with some additional prefixes, e.g graphene://https://sub.domain.xyz/some_prefix/segmentation/table/dataset_name which results in a UnsupportedFormatError: Unable to parse Graphene URL: https://sub.domain.xyz/some_prefix/segmentation/table/dataset_name error.

I think the current regex is stumbling over the additional /some_prefix in the URL:

cloud-volume/cloudvolume/datasource/graphene/metadata.py

Line 433 in 297c34a

LATEST_API_EXTRACTION_RE = re.compile(r'/?(\w+)/(table)/([\w\d\.\_\-]+)/?')

It only scans for a single segment before /table, but I am not sure if that's intentional, or a bug? At least I don't see why the limitation would be necessary.

[william-silversmith]

The reason the regexp is failing is because it's being invoked as re.match instead of re.search. match only matches from the start of a string. The regexp could be augmented to match more elements. I'm a bit wary of allowing arbitrary matches anywhere in the string as that could unlock all kinds of weird edge cases. I can accept a PR with changes in that direction.