Hidden features of mod_rewrite

Question

There seem to be a decent number of mod_rewrite threads floating around lately with a bit of confusion over how certain aspects of it work. As a result I've compiled a few notes on common functionality, and perhaps a few annoying nuances.

What other features / common issues have you run across using mod_rewrite?

Answer 1

Equation can be done with following example:

RewriteCond %{REQUEST_URI} ^/(server0|server1).*$ [NC]
# %1 is the string that was found above
# %1<>%{HTTP_COOKIE} concatenates first macht with mod_rewrite variable -> "test0<>foo=bar;"
#RewriteCond search for a (.*) in the second part -> \1 is a reference to (.*)
# <> is used as an string separator/indicator, can be replaced by any other character
RewriteCond %1<>%{HTTP_COOKIE} !^(.*)<>.*stickysession=\1.*$ [NC]
RewriteRule ^(.*)$ https://notmatch.domain.com/ [R=301,L]

Dynamic Load Balancing:

If you use the mod_proxy to balance your system, it's possible to add a dynamic range of worker server.

RewriteCond %{HTTP_COOKIE} ^.*stickysession=route\.server([0-9]{1,2}).*$ [NC]
RewriteRule (.*) https://worker%1.internal.com/$1 [P,L]

Answer 2

Another great feature are rewrite-map-expansions. They're especially useful if you have a massive amout of hosts / rewrites to handle:

They are like a key-value-replacement:

RewriteMap examplemap txt:/path/to/file/map.txt

Then you can use a mapping in your rules like:

RewriteRule ^/ex/(.*) ${examplemap:$1}

More information on this topic can be found here:

http://httpd.apache.org/docs/2.0/mod/mod_rewrite.html#mapfunc

Answer 3

A better understanding of the [L] flag is in order. The [L] flag is last, you just have to understand what will cause your request to be routed through the URL parsing engine again. From the docs (http://httpd.apache.org/docs/2.2/rewrite/flags.html#flag_l) (emphasis mine):

The [L] flag causes mod_rewrite to stop processing the rule set. In most contexts, this means that if the rule matches, no further rules will be processed. This corresponds to the last command in Perl, or the break command in C. Use this flag to indicate that the current rule should be applied immediately without considering further rules.

If you are using RewriteRule in either .htaccess files or in <Directory> sections, it is important to have some understanding of how the rules are processed. The simplified form of this is that once the rules have been processed, the rewritten request is handed back to the URL parsing engine to do what it may with it. It is possible that as the rewritten request is handled, the .htaccess file or <Directory> section may be encountered again, and thus the ruleset may be run again from the start. Most commonly this will happen if one of the rules causes a redirect - either internal or external - causing the request process to start over.

So the [L] flag does stop processing any further rewrite rules for that pass through the rule set. However, if your rule marked with [L] modified the request, and you're in the .htaccess context or the <Directory> section, then your modifed request is going to be passed back through the URL parsing engine again. And on the next pass, it may match a different rule this time. If you don't understand what happened, it looks like your first rewrite rule with the [L] flag had no effect.

The best way around this is to use the [END] flag (http://httpd.apache.org/docs/current/rewrite/flags.html#flag_end) instead of the [L] flag, if you truly want to stop all further processing of rules (and subsequent reparsing). However, the [END] flag is only available for Apache v2.3.9+, so if you have v2.2 or lower, you're stuck with just the [L] flag. In this case, you must rely on RewriteCond statements to prevent matching of rules on subsequent passes of the URL parsing engine. Or you must ensure that your RewriteRule's are in a context (i.e. httpd.conf) that will not cause your request to be re-parsed.

Answer 4

Where to place mod_rewrite rules

mod_rewrite rules may be placed within the httpd.conf file, or within the .htaccess file. if you have access to httpd.conf, placing rules here will offer a performance benefit (as the rules are processed once, as opposed to each time the .htaccess file is called).

Logging mod_rewrite requests

Logging may be enabled from within the httpd.conf file (including <Virtual Host>):

# logs can't be enabled from .htaccess
# loglevel > 2 is really spammy!
RewriteLog /path/to/rewrite.log
RewriteLogLevel 2

Common use cases

1. To funnel all requests to a single point:

   RewriteEngine on
   # ignore existing files
   RewriteCond %{REQUEST_FILENAME} !-f   
   # ignore existing directories
   RewriteCond %{REQUEST_FILENAME} !-d   
   # map requests to index.php and append as a query string
   RewriteRule ^(.*)$ index.php?query=$1 
   

   Since Apache 2.2.16 you can also use FallbackResource.

2. Handling 301/302 redirects:

   RewriteEngine on
   # 302 Temporary Redirect (302 is the default, but can be specified for clarity)
   RewriteRule ^oldpage\.html$ /newpage.html [R=302]  
   # 301 Permanent Redirect
   RewriteRule ^oldpage2\.html$ /newpage.html [R=301] 
   

   Note: external redirects are implicitly 302 redirects:

   # this rule:
   RewriteRule ^somepage\.html$ http://google.com
   # is equivalent to:
   RewriteRule ^somepage\.html$ http://google.com [R]
   # and:
   RewriteRule ^somepage\.html$ http://google.com [R=302]
   

3. Forcing SSL

   RewriteEngine on
   RewriteCond %{HTTPS} off
   RewriteRule ^(.*)$ https://example.com/$1 [R,L]
   

4. Common flags:

   • [R] or [redirect] - force a redirect (defaults to a 302 temporary redirect)
   • [R=301] or [redirect=301] - force a 301 permanent redirect
   • [L] or [last] - stop rewriting process (see note below in common pitfalls)
   • [NC] or [nocase] - specify that matching should be case insensitive

   
   Using the long-form of flags is often more readable and will help others who come to read your code later.

   You can separate multiple flags with a comma:

   RewriteRule ^olddir(.*)$ /newdir$1 [L,NC]
   

Common pitfalls

1. Mixing mod_alias style redirects with mod_rewrite

   # Bad
   Redirect 302 /somepage.html http://example.com/otherpage.html
   RewriteEngine on
   RewriteRule ^(.*)$ index.php?query=$1
   
   # Good (use mod_rewrite for both)
   RewriteEngine on
   # 302 redirect and stop processing
   RewriteRule ^somepage.html$ /otherpage.html [R=302,L] 
   RewriteCond %{REQUEST_FILENAME} !-f
   RewriteCond %{REQUEST_FILENAME} !-d
   # handle other redirects
   RewriteRule ^(.*)$ index.php?query=$1                 
   

   Note: you can mix mod_alias with mod_rewrite, but it involves more work than just handling basic redirects as above.

2. Context affects syntax

   Within .htaccess files, a leading slash is not used in the RewriteRule pattern:

   # given: GET /directory/file.html
   
   # .htaccess
   # result: /newdirectory/file.html
   RewriteRule ^directory(.*)$ /newdirectory$1
   
   # .htaccess
   # result: no match!
   RewriteRule ^/directory(.*)$ /newdirectory$1
   
   # httpd.conf
   # result: /newdirectory/file.html
   RewriteRule ^/directory(.*)$ /newdirectory$1
   
   # Putting a "?" after the slash will allow it to work in both contexts:
   RewriteRule ^/?directory(.*)$ /newdirectory$1
   

3. [L] is not last! (sometimes)

   The [L] flag stops processing any further rewrite rules for that pass through the rule set. However, if the URL was modified in that pass and you're in the .htaccess context or the <Directory> section, then your modified request is going to be passed back through the URL parsing engine again. And on the next pass, it may match a different rule this time. If you don't understand this, it often looks like your [L] flag had no effect.

   # processing does not stop here
   RewriteRule ^dirA$ /dirB [L] 
   # /dirC will be the final result
   RewriteRule ^dirB$ /dirC     
   

   Our rewrite log shows that the rules are run twice and the URL is updated twice:

   rewrite 'dirA' -> '/dirB'
   internal redirect with /dirB [INTERNAL REDIRECT]
   rewrite 'dirB' -> '/dirC'
   

   The best way around this is to use the [END] flag (see Apache docs) instead of the [L] flag, if you truly want to stop all further processing of rules (and subsequent passes). However, the [END] flag is only available for Apache v2.3.9+, so if you have v2.2 or lower, you're stuck with just the [L] flag.

   For earlier versions, you must rely on RewriteCond statements to prevent matching of rules on subsequent passes of the URL parsing engine.

   # Only process the following RewriteRule if on the first pass
   RewriteCond %{ENV:REDIRECT_STATUS} ^$
   RewriteRule ...
   

   Or you must ensure that your RewriteRule's are in a context (i.e. httpd.conf) that will not cause your request to be re-parsed.

Answer 5

The deal with RewriteBase:

You almost always need to set RewriteBase. If you don't, apache guesses that your base is the physical disk path to your directory. So start with this:

RewriteBase /

Answer 6

mod_rewrite can modify aspects of request handling without altering the URL, e.g. setting environment variables, setting cookies, etc. This is incredibly useful.

Conditionally set an environment variable:

RewriteCond %{HTTP_COOKIE} myCookie=(a|b) [NC]
RewriteRule .* - [E=MY_ENV_VAR:%b]

Return a 503 response: RewriteRule's [R] flag can take a non-3xx value and return a non-redirecting response, e.g. for managed downtime/maintenance:

RewriteRule .* - [R=503,L]

will return a 503 response (not a redirect per se).

Also, mod_rewrite can act like a super-powered interface to mod_proxy, so you can do this instead of writing ProxyPass directives:

RewriteRule ^/(.*)$ balancer://cluster%{REQUEST_URI} [P,QSA,L]

Opinion: Using RewriteRules and RewriteConds to route requests to different applications or load balancers based on virtually any conceivable aspect of the request is just immensely powerful. Controlling requests on their way to the backend, and being able to modify the responses on their way back out, makes mod_rewrite the ideal place to centralize all routing-related config.

Take the time to learn it, it's well worth it! :)