Re: [Phplib-users] Patch to page.inc

[Marko K. <M.K...@os...>]

Hi,

> $HTTP_SERVER_VARS["QUERY_STRING"] = ereg_replace("([?&])" .
> $this->urlaction . "=(.)*&*", "\\1",
> $HTTP_SERVER_VARS["QUERY_STRING"]);
Well, here I don't understand what is actually done. Looks like it
searches for GET parameters (called urlaction) and tries to locate them in
the GET string embraced by [?&] and [&]. But I cannot forsee what happens
if there are more than one '&' for instance. Could be the (.)* construct
would eat that as well. I'd expect at least stg like (.*) to get the full
string, or even better ([^&]*) to avoid any ampersands in the param
content. On the other hand I don't understand the \\1 parameter, since it
would just return the ([?&]) character. \\2 wouldn't be used anyway.

Looking at this is now it looks like it should simply remove the urlaction
parameter from the QUERY_STRING. Is this true???

But then, look at this:

---snip---


 ***script start***

  <?php
     $a='url?test=testen&&test2=test2en&';
     echo "Rewrote:".ereg_replace("([?&])test=(.)*&*", "\\1",$a)."\n";
  ?>

 ***script end***


 ***script output start***

  X-Powered-By: PHP/4.1.1
  Content-type: text/html

  Rewrote:url?

 ***script output end***


---snip---

So everything is eaten by the RE, just "url?" ist left!
Looks like this regexp does not what it should.

Maybe I misunderstood something... What do you really want to do with this
rewrite of the url, looks like I skipped that part of the discussion
somehow, sorry?

> $HTTP_SERVER_VARS["QUERY_STRING"] = ereg_replace("[&][&]+", "&",
> $HTTP_SERVER_VARS["QUERY_STRING"]);
The RE could be rewritten as "[&]{2,}" to delete only multiple ampersands.

Marko