Linting process names with multiple underlines is broken

[fabianegli]

When linting the quantms pipeline tools complains about a non-standard process label process_very but the label is actually process_very_low.

Is there a naming convention which says a label can only use one underscore? I guess tools somewhere splits the string and then only uses the two first segments. Or has some other way to truncate labels. Either way this need asks for a fix.

[awgymer]

try:
    process_label = re.search("process_[A-Za-z]+", process_label[0]).group(0)
except AttributeError:
    process_label = re.search("'([A-Za-z_-]+)'", process_label[0]).group(0)

I don't know if the first regex is just an oversight but it clearly wasn't expected to have one.
I suppose the assumption was that if you are using process_ you would be using (or trying to use) one of the standard labels, which all have just one underscore.

The fact that a non-standard label breaks isn't really an issue imo the real problem here is that process_low_nonsense would get matched as process_low and there'd be no warning!

[awgymer]

Looking at the code for this lint check I am wondering whether it needs more fixes. It currently makes several assumptions (e.g. if there is more than 1 label it only check the first one found) and doesn't necessarily perform much sense-checking