Apache mod_rewrite encode query string parameters

I have a URL coming into apache 2.4:

http://localhost/index.html?q=asdf&b=a|c|e&c=4&d=dsjklkjhd

I need mod_rewrite to URLencode the | so that I get:

http://localhost/index.html?q=asdf&b=a%7Cc%7Ce&c=4&d=dsjklkjh

I cannot figure out how to do it, looked here:

https://stackoverflow.com/questions/15938598/rewrite-to-append-to-query-string#15938642

Here

https://httpd.apache.org/docs/current/mod/mod_rewrite.html

I'm stumped, tried all kinds of stuff. How to do this?

Edit

If I try this:

RewriteCond %{QUERY_STRING} \|
RewriteRule ^index\.html$ /processing/%{QUERY_STRING} [QSD]
RewriteRule ^processing/(.+) /index.html?$1 [R=302,L]

I get this in the logs:

access.log

"-" 172.17.0.1 - - [16/Mar/2019:08:37:00 -0400] "GET /index.html?q=asdf&b=a|c|e&c=4&d=dsjklkjhd HTTP/1.1" 200 520

error.log

[Sat Mar 16 08:37:00.723831 2019] [rewrite:trace3] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] applying pattern '.*' to uri '/index.html'

[Sat Mar 16 08:37:00.723838 2019] [rewrite:trace4] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] RewriteCond: input='GET' pattern='^TRACE' => not-matched

[Sat Mar 16 08:37:00.723843 2019] [rewrite:trace4] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] RewriteCond: input='GET' pattern='^OPTIONS' => not-matched

[Sat Mar 16 08:37:00.723847 2019] [rewrite:trace4] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] RewriteCond: input='GET' pattern='^DELETE' => not-matched

[Sat Mar 16 08:37:00.723852 2019] [rewrite:trace4] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] RewriteCond: input='GET' pattern='^PUT' => not-matched

[Sat Mar 16 08:37:00.723855 2019] [rewrite:trace3] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] applying pattern '^index\.html$' to uri '/index.html'

[Sat Mar 16 08:37:00.723867 2019] [rewrite:trace3] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] applying pattern '^processing/(.+)' to uri '/index.html'

[Sat Mar 16 08:37:00.723874 2019] [rewrite:trace1] [pid 13] mod_rewrite.c(470): [client 172.17.0.1:45194] 172.17.0.1 - - [localhost/sid#558a781fd350][rid#558a782e2e90/initial] pass through /index.html

[Sat Mar 16 08:37:00.724892 2019] [authz_core:debug] [pid 13] mod_authz_core.c(809): [client 172.17.0.1:45194] AH01626: authorization result of Require all granted: granted

[Sat Mar 16 08:37:00.724914 2019] [authz_core:debug] [pid 13] mod_authz_core.c(809): [client 172.17.0.1:45194] AH01626: authorization result of : granted

Solution 1:

Apache only appears to re-encode the substitution string if it changes in some way. So, one way to do this (to URL encode the pipe symbol | in the requested query string) is to temporarily rewrite the request (internally) to something entirely different and then redirect it back again. The redirect then URL encodes the complete substitution string (avoiding a redirect/rewrite loop):

For example (UPDATED for a server/vHost context):

RewriteCond %{QUERY_STRING} \|
RewriteRule ^/index\.html$ /processing/%{QUERY_STRING} [QSD]
RewriteRule ^/processing/(.+) /index.html?$1 [R=302,L]

The first rule internally rewrites the request (that contains a | symbol in the query string) to /processing/ and passes the QUERY_STRING as pathname information. The query string is discarded.

The very next rule immediately redirects it back to the required URL, but this time it should be re-encoded, URL encoding | as %7c in the query string.

Note that this triggers a 302 (temporary) redirect. Only change it to a 301 (permanent) redirect - if that is the intention - once you have confirmed that it works as intended, to avoid caching issues.