PCRE and regex question...

[Bill Stoddard <bi...@wstoddard.com>]

Is there a quick way I can check a string to see if it is a regex or not?

I am looking at extending SetEnvIf to accept a regex on the 'header' argument (value may
already be a regex).  My first cut replaces this call in mod_setenvif:

    val = apr_table_get(r->headers_in, b->name);

    where b->name is a nonregex header name

with code that iterated through r->headers_in doing an ap_regex on each under the
assumption that b->name -may- be a regex.  The problem with this is the unnecessary
performance hit on all requests when the b->name is NOT a regex (and where apr_table_get()
will do the right thing).

pcre_infofull() seems a likely candidate but it is not clear to me which field to check


Bill

Re: PCRE and regex question...

[Joshua Slive <sl...@commerce.ubc.ca>]

On Thu, 7 Jun 2001, Greg Stein wrote:

> On Thu, Jun 07, 2001 at 05:01:16PM -0400, Greg Marr wrote:
> > At 04:17 PM 06/07/2001, Bill Stoddard wrote:
> > >Is there a quick way I can check a string to see if it is a regex or
> > >not?  I am looking at extending SetEnvIf to accept a regex on the
> > >'header' argument (value may already be a regex).
> >
> > The absolute easiest would be to require that the regex start and end
> > with a /.
> > SetEnvIf nonregexheader
> > SetEnvIf /regexheader/
> >
> > You could also use a regex on it:
> > /^[-A-Za-z0-9_]*$/
> > This assumes that only letters, numbers, -, and _ are allowed in
> > header names.
>
> Strictly speaking, every string is a regex. It is just that "abc" doesn't
> say a whole lot :-)
>
> I like Greg Marr's suggestion of requiring /regex/. That also helps to
> clarify things for the user who is reading .conf.
>

I'll chime in here with my usual consistency arguments:

There are MANY Apache directives that take a regex and none of them
use the /regex/ notation.  Please don't invent a new notation unless
it is really necessary.

I prefer Greg's second suggestion.  Just evaluate the input at startup to
determine whether or not it is a regex and store the answer so that you
don't waste request-time processing.

Joshua.

Re: PCRE and regex question...

[Greg Stein <gs...@lyra.org>]

On Thu, Jun 07, 2001 at 05:01:16PM -0400, Greg Marr wrote:
> At 04:17 PM 06/07/2001, Bill Stoddard wrote:
> >Is there a quick way I can check a string to see if it is a regex or 
> >not?  I am looking at extending SetEnvIf to accept a regex on the 
> >'header' argument (value may already be a regex).
> 
> The absolute easiest would be to require that the regex start and end 
> with a /.
> SetEnvIf nonregexheader
> SetEnvIf /regexheader/
> 
> You could also use a regex on it:
> /^[-A-Za-z0-9_]*$/
> This assumes that only letters, numbers, -, and _ are allowed in 
> header names.

Strictly speaking, every string is a regex. It is just that "abc" doesn't
say a whole lot :-)

I like Greg Marr's suggestion of requiring /regex/. That also helps to
clarify things for the user who is reading .conf.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/

Re: PCRE and regex question...

[Bill Stoddard <bi...@wstoddard.com>]

> At 04:17 PM 06/07/2001, Bill Stoddard wrote:
> >Is there a quick way I can check a string to see if it is a regex or 
> >not?  I am looking at extending SetEnvIf to accept a regex on the 
> >'header' argument (value may already be a regex).
> 
> The absolute easiest would be to require that the regex start and end 
> with a /.
> SetEnvIf nonregexheader
> SetEnvIf /regexheader/
> 
> You could also use a regex on it:
> /^[-A-Za-z0-9_]*$/
> This assumes that only letters, numbers, -, and _ are allowed in 
> header names.
> 

For the sake of consistency, think I'll pursue option 2.  Thanks.

Bill

Re: PCRE and regex question...

[Greg Marr <gr...@alum.wpi.edu>]

At 04:17 PM 06/07/2001, Bill Stoddard wrote:
>Is there a quick way I can check a string to see if it is a regex or 
>not?  I am looking at extending SetEnvIf to accept a regex on the 
>'header' argument (value may already be a regex).

The absolute easiest would be to require that the regex start and end 
with a /.
SetEnvIf nonregexheader
SetEnvIf /regexheader/

You could also use a regex on it:
/^[-A-Za-z0-9_]*$/
This assumes that only letters, numbers, -, and _ are allowed in 
header names.

-- 
Greg Marr
gregm@alum.wpi.edu
"We thought you were dead."
"I was, but I'm better now." - Sheridan, "The Summoning"