Hi! 2017-02-06 17:25 GMT+01:00 <uwe.pol...@amann.com>:
> Hi, > > I am trying a reverse proxy server based on apache httpd v2.4 on the most > recent release of CentOS: > > # httpd -version > Server version: Apache/2.4.6 (CentOS) > Server built: Nov 14 2016 18:04:44 > > # uname -a > Linux hostname.domain.tld 3.10.0-514.6.1.el7.x86_64 #1 SMP Wed Jan 18 > 13:06:36 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux > > # cat /etc/centos-release > CentOS Linux release 7.3.1611 (Core) > > Within this configuration I have to use mod_substitute to rewrite URLs > from some applications. > For this I am using mod_filter with the SUBSTITUTE Filter as follows: > > ProxyRequests Off > ProxyPass /my-location https://my-server.domain.tld/ > > <Location /my-location/> > ProxyPassReverse /my-location > > FilterDeclare AGFILTER > > FilterProvider AGFILTER SUBSTITUTE "%{resp:Content-Type} =~ > m#^text/html#" > FilterProvider AGFILTER SUBSTITUTE "%{resp:Content-Type} =~ > m#.*/css#" > FilterProvider AGFILTER SUBSTITUTE "%{resp:Content-Type} =~ > m#.*/json#" > FilterProvider AGFILTER SUBSTITUTE "%{resp:Content-Type} =~ > m#.*/javascript#" > > FilterChain AGFILTER > > Substitute "s#/(css|js|images|management| > system|help)/(.*)#/my-location/$1/$2#fi" > </Location> > > It works fine if there is only one occurrence of the search pattern in a > line in the html code. This occurrence will be replaced properly. > However, if there are two or more occurrences of the search pattern in one > html line, only the first one is replaced. It looks like this example: > > <tr><th colspan=3 nowrap></th><th colspan=3 nowrap><a > href="overview.epl?hide=1"><img border=0 src="/my-location/images/hide.gif" > alt=" Spalte ausblenden"></a> <a href="overview.epl?cl=1&pl=1"><img > src="/images/right.gif" border=0 alt=" Spalte nach rechts > schieben"></a></th><th colspan=3 nowrap><a href="overview.epl?cl=2&pl=-1"><img > src="/images/left.gif" alt=" Spalte nach links schieben" border=0 ></a> <a > .... > > Here you see: The first one is replaced, the second image URL is the same > as before. > > Is this works-as-designed? > I think that the issue is in the (.*) of your regex. In your example it will match the first occurrence of the pattern (like "images/") and will end up eating all the rest of the chars (greedy behavior as far as I can see). The following matches more than on occurrences in your example string, because it checks for the .something extension: (css|js|images|management|system|help)\/(\w+\.\w) So mod_substitute seems to be working fine, the regex would needs a bit of tuning imo. The documentation might need to mention the greedy behavior, but I need to triple check that what I just said makes sense :) Hope that helps! Luca
