Re: [code-review] URI::Find::Iterator

--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Wed, Dec 03, 2003 at 02:29:56PM +0000, Simon Wistow wrote:

> It occured to me that the underlying code could be spun out into         =
                                                =20
> Regex::Iterator to provide a generic way to iterate over regexes with    =
                                                =20
> U:F:I being a thin shim on top.                                          =
                                                =20

That sounds like a very useful idea.

> I finished the code last night when I found it lying around about 6
> months after doing the initial stab . I'm using a slightly odd mechnism
> for iterating and I should probably try and remember why I did that.
> Again comments welcome.=20

I think you should use \G and pos on the $self->{_remain} value,
instead of $` et al. They are notorious for slowing down regexes.

Read about \G under perlre if you're not familiar. It is a zero width
assertion which matches point in the string indexed by the value pos
(perldoc -f pos) is set to. Every time a global match is made, the pos
value is set to the end of the match.

This is with //g matches in scalar context.


When you're generating regexes, try to set the last variable to be

     $var =3D qr/$var whatever .*/;

This way the regex doesn't have to be recompiled each time. You then
just match against the variable:

     $_ =3D~ $var;

And including it in a match string isn't painful either, AFAIK:

    /$var and something else/;


About the interface:

I think it would be useful if the input could be more arbitrary. I
have a spamtrap, and I use

     wget -q -b -o /dev/null -i - -F -O /dev/null --limit-rate=3D3K
     --wait=3D3 --random-wait -Q 256K

to hopefully activate all the web bugs. You're module could be useful,
and better purposed, to do this with some more fine grained control.

I think this should also be able to work on a FH reference passed to
it, with remain being the pending input, and match & result behaving
as they do now.

I've never tried matching generic regexes on arbitrary sized
input. With URLs it shouldn't be tough, because they don't have line
breaks in them - you can just match lines.

This will be harder for a generic Regexp::Iterator.


Some feedback about the distribution:

Many people still don't have Module::Build. I think it's nicer to use

    create_makefile_pl =3D> "traditional"

instead of "passthrough".

The tests all passed, BTW.

--=20
 ()  Yuval Kogman <nothingmuch-JLKJG/[email protected]> 0xEBD27418  perl hacker &
 /\  kung foo master: /me beats up some cheese: neeyah!!!!!!!!!!!!!!!!!


--qMm9M+Fa2AknHoGS
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (Darwin)

iD8DBQE/1yMLVCwRwOvSdBgRAk6CAJ0WqlhBrKAJplsN3G3jbowC4fwxGQCgsmdb
lQe/5q8splVJCQUVDuEZzG4=
=GLNM
-----END PGP SIGNATURE-----

--qMm9M+Fa2AknHoGS--