set regex pattern for CURIE and add URL, fixes #400

[sierra-moxon]

• regexes are fairly permissive taking into account that CURIE syntax in the "wild" might be less conformant to prefix best practices.
• the ticket called for a URL type, if we want a semantic URI type (where the URI can be the subject of a triple in RDF), then we should be much more restrictive. When we have a specific use case for this, I'd be happy to refine it.

[edeutsch]

okay with me in principle, but I don't think the regular expression seems helpful. Is there a "standard" regexp for URLs out there on stack overflow or somewhere?

[edeutsch]

Isn't this regular expression anything without a space? Is that really helpful? Testing this regexp:

#!/bin/env python3
import re

inputs = [ 'http://arax.ncats.io', 'foo', '@*$&@#', 'PMID:123', 'http://peptideatlas.org/tmp/hello world.txt' ]

for input in inputs:
    match = re.search(r'^(http(s)?:\/\/.)?(www\.)?\S+$', input)
    if match:
        print(f"MATCHES {input}")
    else:
        print(f"    x   {input}")

yields

MATCHES http://arax.ncats.io
MATCHES foo
MATCHES @*$&@#
MATCHES PMID:123
    x   http://peptideatlas.org/tmp/hello world.txt

only the last one fails.

Yet if you paste that into your browser, it works!

Maybe this is useful with a more restrictive regular expression?

[sierra-moxon]

My first thought too was to find a more comprehensive one. The one I found and tested that was more restrictive did fail the line length linting on this repo and before I tried to break it into many lines, I started asking around a bit for best practice on this. The feedback I got was that a very restrictive regex will mean constant tweaking with "in the wild" implementations of URLs and CURIEs. However, if we want to have a URI (not L) type, that is much more restrictive, we can do that.

[vdancik]

I think that specifying regex in specification is not necessary