[Python-Dev] Re: pre-PEP: Complete, Structured Regular Expression Group Matching

[Mike Coleman]

Paul Moore <pf_moore at yahoo.co.uk> writes:
> I was interested in your message as a break from @decorators, but I
> couldn't muster the energy to think about the proposal because I've
> been reading too many @decorator messages. I guess you won one way and
> lost the other :-)

@:-)

> Maybe a better name would be re.parse? The function has a "feel" of
> parsing a string according to a pattern.

This is a nice suggestion.  The main objection I can think of is that we might
be stealing that name from some future idea that would more deserve it.

> The only other comment I have is that the semantics seem pretty
> complex - I think that in practice, they do more or less what you want
> them to, but the description is pretty obscure.

Yeah, probably the doc will need to be carefully written, with good examples,
to make this go down easily.  I hope that for practical cases it'll be
relatively straightforward.

> And although I can see that the error return has some value, I suspect that
> it might actually complicate real use.

Someone else suggested throwing an exception instead, which I'm now leaning in
favor of, too.

> A suggestion - would it be possible to implement re.structmatch as a
> pure Python prototype, to thrash out some of the usability questions?
> If the function appears valuable in such a form, arguing for
> incorporation into the re module would be a lot easier.

I suspect that in this case a pure Python implementation might actually be a
lot harder than implementing it as a C patch.  This is because so much of the
work is already done in the current re module; the C patch would probably not
be that big.  In order to do it in Python I'd end up reimplementing re.match
in Python, which I wouldn't look forward to.

It's because of this that I'm trying to get as much preliminary comment as I
can.

Cheers,
Mike