# Rewrite regex with percent signs

**URL:** <https://www.ruby-forum.com/t/rewrite-regex-with-percent-signs/244657>\
**Category:** NGINX\
**Created:** [May 22, 2016, 1:17pm UTC](https://www.ruby-forum.com/t/rewrite-regex-with-percent-signs/244657 "2016-05-22T13:17:20Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Rafa\_F](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/958977/32.png) [@Rafa\_F](https://www.ruby-forum.com/u/Rafa_F)\
**Post date:** [May 22, 2016, 1:17pm UTC](https://www.ruby-forum.com/t/rewrite-regex-with-percent-signs/244657/1 "2016-05-22T13:17:20Z")

</div>

Hello, I am helping an admin sort out some 404 issues by using some  
rewrite  
which have generally been successful. However we have a couple of cases  
that  
are a bit mysterious and hope you can help explain. This is from a  
vbulletin  
forum that used to use the vbseo extension to make the url’s prettier  
but  
that extension has been dropped now so posts with those pretty url’s  
don’t  
point the correct places.

For example, we have a url of the following:  
/members/redrobes-albums-2d%20vs%203d%20?-picture12345-mt-pub01.jpg

it needs to go to

/attachment.php?attachmentid=12345

we have:

location /members/ {  
rewrite ^/members/.±albums-.±picture(\d+)-.\*  
/attachment.php?attachmentid=$1? redirect;  
}

and this particular one is not working. It works with many others where  
the  
original url did not have the %20’s in them. So there is something about  
those %20’s that are causing these to fail.

I can write a perl script and run that url through its regex and it does  
change them.

So what does the nginx regex do different from perl regex with regard to  
%  
signs.

Thanks.

Posted at Nginx Forum:

> **[Rewrite regex with percent signs](https://forum.nginx.org/read.php?2,267039,267039#msg-267039)**
>
> Hello, I am helping an admin sort out some 404 issues by using some rewrite which have generally been successful. However we have a couple of cases that are a bit mysterious and hope you can help explain. This is from a vbulletin forum that used to...

---

<div class="post-metadata">

**Author:** ![redrobes](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/e19adc/32.png) [@redrobes](https://www.ruby-forum.com/u/redrobes)\
**Post date:** [May 22, 2016, 3:42pm UTC](https://www.ruby-forum.com/t/rewrite-regex-with-percent-signs/244657/2 "2016-05-22T15:42:48Z")

</div>

On Sun, May 22, 2016 at 07:16:35AM -0400, redrobes wrote:

Hi there,

> For example, we have a url of the following:  
> /members/redrobes-albums-2d%20vs%203d%20?-picture12345-mt-pub01.jpg

The %20 pieces in there are url-encoded spaces. In a “location” or a  
“rewrite”, you would have to match a single space character each.

However, there is also a ? in the url; that marks the start of the query  
string. A “location” or “rewrite” in nginx will _not_ consider that part  
of the url.

> it needs to go to
> 
> /attachment.php?attachmentid=12345

It is not immediately clear to me which parts of the original url are  
important in deciding whether the request should be redirected or not.

> we have:
> 
> location /members/ {  
> rewrite ^/members/.±albums-.±picture(\d+)-.\*  
> /attachment.php?attachmentid=$1? redirect;  
> }

That suggests that just those three words matter. You might be able  
to put something together involving “$args” matching “-picture(\d+)-”  
if the request matches “^/members/.\*-albums-”, perhaps?

Alternatively, perhaps the thing that created the url in the first  
place,  
incorrectly did not url-encode the ? to %3F.

> and this particular one is not working. It works with many others where the  
> original url did not have the %20’s in them. So there is something about  
> those %20’s that are causing these to fail.

I suspect that it is the ? rather than the %20, from the one example  
you have given.

> I can write a perl script and run that url through its regex and it does  
> change them.
> 
> So what does the nginx regex do different from perl regex with regard to %  
> signs.

With regard to % signs, nginx regex uses the %-unencoded version. With  
regard to ?, some nginx parts do not consider anything after the ? when  
matching.

Good luck with it,

## f

Francis D. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)

---

<div class="post-metadata">

**Author:** ![redrobes](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/e19adc/32.png) [@redrobes](https://www.ruby-forum.com/u/redrobes)\
**Post date:** [May 22, 2016, 7:18pm UTC](https://www.ruby-forum.com/t/rewrite-regex-with-percent-signs/244657/3 "2016-05-22T19:18:56Z")

</div>

## Francis D. Wrote:

> However, there is also a ? in the url; that marks the start of the  
> query  
> string. A “location” or “rewrite” in nginx will _not_ consider that  
> part  
> of the url.

Ah ! Thanks. I didn’t spot that one amongst all the other odd chars  
there.  
Yes nginx does indeed treat the ? as a different character to perl and  
the  
hex codes convert to “2d vs 3d ?” I.e. the ? was in the title of the  
post  
and is not the start of args. I think thats it and it makes sense now. I  
can  
understand what is going on.

It could have been mighty hard to fix this case since we are not going  
to  
know in advance whether the ? was part of the url or the start of args  
but I  
think in our case we know were going to dump all the args anyway and  
substitute our own in. So I think there may be the possibility of  
appending  
the args to the rewrite before we do the match. Not sure at this point.

But thanks Francis - I think you have solved it.

> > location /members/ {  
> > incorrectly did not url-encode the ? to %3F.  
> > I can write a perl script and run that url through its regex and it  
> > matching.  
> > [mailman.nginx.org archive](http://mailman.nginx.org/mailman/listinfo/nginx)  
> > Posted at Nginx Forum:  
> > [https://forum.nginx.org/read.php?2,267039,267044#msg-267044](https://forum.nginx.org/read.php?2,267039,267044#msg-267044)