forum.excito.org

mod_proxy_html

Suppose I want to expose a website that is running on a non-standard port internally. I could use mod_proxy, if the website only used relative or path-only absolute URLS. However, the website happens to know what hostname and port it is supposed to be running on, and has some urls which are targetted directly at the hostname and port. mod_proxy will break in this case, because URLs in the webpages aren't rewritten, and I can't access the device directly from the external network.

mod_proxy_html solves this, by rewriting parts of the source code on the fly. It's implemented as a filter in apache2, which means all webpages in the appropriate block pass through it, and can be rewritten on the fly.

Here is an example config for exposing the CUPS web interface, which runs on port 631, via an ApacheReverseProxy:

First, get the mod_proxy_html module:

Code: Select all

# apt-get install libapache2-mod-proxy-html

You may or may not have to enable the module with

Code: Select all

# a2enmod proxy_html

Then edit the <IfModule mod_proxy.c> section of /etc/apache2/mods-enabled/proxy.conf

Code: Select all

        <IfModule mod_proxy.c>
                ProxyRequests Off
                ProxyPass /cups http://cups.internal.org:631/
                ProxyHTMLURLMap http://cups.internal.org:631/ /cups
                <Location /cups>
                        ProxyPassReverse http://cups.internal.org:631/
                        SetOutputFilter proxy-html
                        ProxyHTMLURLMap /               /cups/
                        ProxyHTMLURLMap /cups/          /cups/
                </Location>
        </IfModule mod_proxy.c>

Then reload the config

Code: Select all

/etc/init.d/apache2 force-reload

New directives that are added include the ProxyHTMLURLMap directive, and the SetOutputFilter directive.

SetOutputFilter tells apache to pass the proxied content through the proxy-html filter, which mod_proxy_html provides. This does all the rewriting discussed below:

The first ProxyHTMLURLMap directive tells mod_proxy_html to rewrite any instance of "http://cups.internal.org:631/" to "/cups". This means any absolute URLs will be rewritten to point under the /cups/ Location, which will then get proxied appropriately.

The remaining ProxyHTMLURLMap directives handle URLs with just a set path, eg "/printers/printer1/". This will get rewritten to "/cups/printers/printer1/" and then proxied correctly. The last directive is a no-rewrite rule, intended to prevent infinite looping.