Redirecting URLs using variables

[JJoe]

(Feb. 17, 2013 06:31 AM)zoltan Wrote:  I'm assuming the "([^?]+)\1" means look for a run of any character that's not a question mark and place it into the variable?

Correct.

(Feb. 17, 2013 06:31 AM)zoltan Wrote:  On the other hand, why doesn't it interpret the "?" as the Prox meta character meaning "any character" -- which would seem to prevent it from capturing anything into the variable.

I think \ and ^ are the only special characters that are not escaped when inside []. So you could try [^\?]+, if it helps.

(Feb. 17, 2013 06:31 AM)zoltan Wrote:  

Quote:It is as Scott Lemmon wanted or left it. The spaces are not needed, btw.

OK. It's just not evident to me what the "&" symbol's meaning/function is here. The description of it on the matching language page doesn't seem to cover an instance like this.

It really is about Scott's choices and the time he had to spare. I'll guess:
The & indicates that the matching has not been completed. If there is a \1 in front of the &, there is a good chance that the data available to \1 will be needed later on. So "\1&" causes \1 to fill. Of course, what follows the & doesn't actually have to match code; it has to be "true". This "choice" also keeps us from typing (?+)\1.

(Feb. 17, 2013 06:31 AM)zoltan Wrote:  

Quote:A trailing \1 used in the matching expression of a web page filter captures code. A trailing \1 in a URL match or list entry does not.

Very important to know. I never realized this. So, at least something in parentheses will always be necessary to signal that something should go into the variable (for URLs and lists)?

IIRC, where the expression is used is the deciding factor. A trailing \1 in the filter's URL Match field or in a list that is called from the URL Match field needs help. I think, the same for the "Bypass any URLs that match this expression" field.

HTH