# Re: question about addition filter **URL:** \ **Category:** NGINX\ **Created:** [January 28, 2009, 10:45pm UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869 "2009-01-28T22:45:05Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![Mohammad\_K](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/8e8cbc/32.png) [@Mohammad\_K](https://www.ruby-forum.com/u/Mohammad_K)\ **Post date:** [January 28, 2009, 10:45pm UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869/1 "2009-01-28T22:45:05Z")
Igor, I tried your config, with 0.7.32 and 0.7.31, with/without debug, with/without optimization, but I constantly get garbage before and after the content of /y/index.html. Maybe it is returning some uninitialized memory? I might be totally wrong, but Evan’s module development tutorial mentions something about NGX\_AGAIN and that we should return NGX\_AGAIN if ngx\_http\_subrequest returns NGX\_AGAIN. But looking in to your code, ngx\_http\_addition\_filter\_module.c, line numbers 154 and 183, you are always returning NGX\_ERROR if ngx\_http\_subrequest does not return NGX\_OK. Could that be the reason? Thanks, -M ps, I doubt it matters, but I have CentOS 5.2 64 bit. * * * From: Igor S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) To: removed\_email\_address@domain.invalid Sent: Wednesday, January 28, 2009 12:04:35 PM Subject: Re: question about addition filter On Wed, Jan 28, 2009 at 11:21:35AM -0800, Mohammad K. wrote: > which would supposedly add the content of the [www.yahoo.com](http://www.yahoo.com) and [www.google.com](http://www.google.com) before and after the content of the page at /y/index.html? The addition filter module states that the texts after add\_before\_body and add\_after\_body are URI, but the above config doesn’t seem to be doing it! This should work, but it’s better to use 0.7.32 for this: ``` location = /x { proxy_pass http://www.yahoo.com/; } location = /g{ proxy_pass http://www.google.com/; } location /y { add_before_body /x; add_after_body /q; } ``` ---
**Author:** ![Mohammad\_K](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/8e8cbc/32.png) [@Mohammad\_K](https://www.ruby-forum.com/u/Mohammad_K)\ **Post date:** [January 28, 2009, 10:49pm UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869/2 "2009-01-28T22:49:47Z")
On Wed, Jan 28, 2009 at 01:28:50PM -0800, Mohammad K. wrote: > ps, I doubt it matters, but I have CentOS 5.2 64 bit. > This may be gzipped content returned by external sites: ``` location = /x { proxy_pass http://www.yahoo.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } location = /g{ proxy_pass http://www.google.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } ``` ---
**Author:** ![Mohammad\_K](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/8e8cbc/32.png) [@Mohammad\_K](https://www.ruby-forum.com/u/Mohammad_K)\ **Post date:** [January 28, 2009, 11:19pm UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869/3 "2009-01-28T23:19:43Z")
That was the problem, the content was zipped, everything is working now. Thanks, -M * * * From: Igor S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) To: removed\_email\_address@domain.invalid Sent: Wednesday, January 28, 2009 1:39:50 PM Subject: Re: question about addition filter On Wed, Jan 28, 2009 at 01:28:50PM -0800, Mohammad K. wrote: > ps, I doubt it matters, but I have CentOS 5.2 64 bit. > This may be gzipped content returned by external sites: ``` location = /x { proxy_pass http://www.yahoo.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } location = /g{ proxy_pass http://www.google.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } ``` ---
**Author:** ![Mohammad\_K](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/8e8cbc/32.png) [@Mohammad\_K](https://www.ruby-forum.com/u/Mohammad_K)\ **Post date:** [January 28, 2009, 11:44pm UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869/4 "2009-01-28T23:44:57Z")
What if I wanted to write a module that does something like this: if input request is something like [http://localhost/y?a=foo&b=bar](http://localhost/y?a=foo&b=bar), then have the before\_body fetch [yahoo.com/foo](http://yahoo.com/foo) and after\_body get the content of [google.com/bar](http://google.com/bar) (i.e., the URLs can bo modified on the fly). Is this doable, possibly by modifying the content of uri that is passed to ngx\_http\_subrequest, or is there another alternative? Thanks, -M * * * From: Igor S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) To: removed\_email\_address@domain.invalid Sent: Wednesday, January 28, 2009 1:39:50 PM Subject: Re: question about addition filter On Wed, Jan 28, 2009 at 01:28:50PM -0800, Mohammad K. wrote: > ps, I doubt it matters, but I have CentOS 5.2 64 bit. > This may be gzipped content returned by external sites: ``` location = /x { proxy_pass http://www.yahoo.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } location = /g{ proxy_pass http://www.google.com/; ``` - ``` proxy_set_header Accept-Encoding ""; } ``` ---
**Author:** ![Mohammad\_K](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/8e8cbc/32.png) [@Mohammad\_K](https://www.ruby-forum.com/u/Mohammad_K)\ **Post date:** [January 29, 2009, 9:03am UTC](https://www.ruby-forum.com/t/re-question-about-addition-filter/158869/5 "2009-01-29T09:03:17Z")
On Wed, Jan 28, 2009 at 02:34:54PM -0800, Mohammad K. wrote: > What if I wanted to write a module that does something like this: if input request is something like [http://localhost/y?a=foo&b=bar](http://localhost/y?a=foo&b=bar), then have the before\_body fetch [yahoo.com/foo](http://yahoo.com/foo) and after\_body get the content of [google.com/bar](http://google.com/bar) (i.e., the URLs can bo modified on the fly). Is this doable, possibly by modifying the content of uri that is passed to ngx\_http\_subrequest, or is there another alternative? Currently add\_xxxx\_body does not support variables. If it could, then it would be possible: ``` location /x/ { proxy_pass http://www.yahoo.com/; } location /g/ { proxy_pass http://www.google.com/; } location /y { add_before_body /x/$arg_a; add_after_body /q/$arg_b; ```