I'm having some difficulty with a super simple htaccess redirect.
All I want to do is rewrite absolutely everything, except a couple files.
htaccess looks like this:
RewriteEngine On
RewriteCond %{REQUEST_URI} !sitemap
RewriteCond %{REQUEST_URI} !robots
RewriteRule ^(.*)$ http://example.com/$1 [L,R=301]
The part that works is that everything gets redirected to new domain as it should be. And I can also access robots.txt without being forwarded, but not with sitemap.xml. If I try to go to sitemap.xml, the domain forwards along anyway and opens the sitemap file on the new domain.
I have this exact same issue when trying to "ignore" index.html. I can ignore robots, I can ignore alternate html or php files, but if I want to ignore index.html, the regex fails.
Since I can't actually SEE what is in the REQUEST_URI variable, my guess is that somehow index.html and sitemap.xml are some kind of "special" files that don't end up in REQUEST_URI? I know this because of a stupid test. If I choose to ignore index.html like this:
RewriteCond %{REQUEST_URI} !index.html
Then if I type example.com/index.html I will be forwarded. But if I just type example.com/ the ignore actually works and it shows the content of index.html without forwarding!
How is it that when I choose to ignore the regex "index.html", it only works when "index.html" is not actually typed in the address bar!?!
And it gets even weirder! Should I type something like example.com/index.html?option=value, then the ignore rule works and I do NOT get forwarded when there are attributes like this. But index.html by itself doesn't work, and then just having the slash root, the rule works again.
I'm completely confused! Why does it seem like REQUEST_URI is not able to see some filenames like index.html and sitemap.xml? I've been Googling for 2 days and not only can I not find out if this is true, but I can't seem to find any websites which actually give examples of what these htaccess server variables actually contain!
Thanks!
my guess is that somehow index.html and sitemap.xml are some kind of "special" files that don't end up in REQUEST_URI?
This is not true. There is no such special treatment of any requested URL. The REQUEST_URI server variable contains the URL-path (only) of the request. This notably excludes the scheme + hostname and any query string (which are available in their own variables).
However, if there are any other mod_rewrite directives that precede this (including the server config) that rewrite the URL then the REQUEST_URI server variable is also updated to reflect the rewritten URL.
index.html (Directory Index)
index.html is possibly a special case. Although, if you are explicitly requesting index.html as part of the URL itself (as you appear to be doing) then this does not apply.
If, on the other hand, you are requesting a directory, eg. http://example.com/subdir/ and relying on mod_dir issuing an internal subrequest for the directory index (ie. index.html), then the REQUEST_URI variable may or may not contain index.html - depending on the version of Apache (2.2 vs 2.4) you are on. On Apache 2.2 mod_dir executes first, so you would need to check for /subdir/index.html. However, on Apache 2.4, mod_rewrite executes first, so you simply check for the requested URL: /subdir/. It's safer to check for both, particularly if you have other rewrites and there is possibility of a second pass through the rewrite engine.
Caching problems
However, the most probable cause in this scenario is simply a caching issue. If the 301 redirect has previously been in place without these exceptions then it's possible these redirections have been cached by the browser. 301 (permanent) redirects are cached persistently by the browser and can cause issues with testing (as well as your users that also have these redirects cached - there is little you can do about that unfortunately).
RewriteCond %{REQUEST_URI} !(sitemap|index|alternate|alt) [NC]
RewriteRule .* alternate.html [R,L]
The example you presented in comments further suggests a caching issue, since you are now getting different results for sitemap than those posted in your question. (It appears to be working as intended in your second example).
Examining Apache server variables
#zzzaaabbb mentioned one method to examine the value of the Apache server variable. (Note that the Apache server variable REQUEST_URI is different to the PHP variable of the same name.) You can also assign the value of an Apache server variable to an environment variable, which is then readable in your application code.
For example:
RewriteRule ^ - [E=APACHE_REQUEST_URI:%{REQUEST_URI}]
You can then examine the value of the APACHE_REQUEST_URI environment variable in your server-side code. Note that if you have any other rewrites that result in the rewritting process to start over then you could get multiple env vars, each prefixed with REDIRECT_.
With the index.html problem, you probably just need to escape the dot (index\.html). You are in the regex pattern-matching area on the right-hand side of RewriteCond. With the un-escaped dot in there, there would need to be a character at that spot in the request, to match, and there isn't, so you're not matching and are getting the unwanted forward.
For the sitemap not matching problem, you could check to see what REQUEST_URI actually contains, by just creating an empty dummy file (to avoid 404 throwing) and then do a redirect at top of .htaccess. Then, in browser URL, type in anything you want to see the REQUEST_URI for -- it will show in address bar.
RewriteCond %{QUERY_STRING} ^$
RewriteRule ^ /test.php?var=%{REQUEST_URI} [NE,R,L]
Credit MrWhite with that easy test method.
Hopefully that will show that sitemap in URL ends up as something else, so will at least partially explain why it's not pattern-matching and preventing redirect, when it should be pattern-matching and preventing redirect.
I would also test by being sure that the server isn't stepping in front of things with custom 301 directive that for whatever reason makes sitemap behave unexpectedly. Put this at the top of your .htaccess for that test.
ErrorDocument 301 default
Many posters have problems debugging their RewriteRule and RewriteCond statements within their .htaccess files. Most of these are using a shared hosting service and therefore don't have access to the root server configuration. They cannot avoid using .htaccess files for rewriting and cannot enable a RewriteLogLevel" as many respondents suggest. Also there are many .htaccess-specific pitfalls and constraints are aren't covered well. Setting up a local test LAMP stack involves too much of a learning curve for most.
So my Q here is how would we recommend that they debug their rules themselves. I provide a few suggestions below. Other suggestions would be appreciated.
Understand that the mod_rewrite engine cycles through .htaccess files. The engine runs this loop:
do
execute server and vhost rewrites (in the Apache Virtual Host Config)
find the lowest "Per Dir" .htaccess file on the file path with rewrites enabled
if found(.htaccess)
execute .htaccess rewrites (in the user's directory)
while rewrite occurred
So your rules will get executed repeatedly and if you change the URI path then it may end up executing other .htaccessfiles if they exist. So make sure that you terminate this loop, if necessary by adding extra RewriteCond to stop rules firing. Also delete any lower level .htaccess rewrite rulesets unless explicitly intent to use multi-level rulesets.
Make sure that the syntax of each Regexp is correct by testing against a set of test patterns to make sure that is a valid syntax and does what you intend with a fully range of test URIs. See answer below for more details.
Build up your rules incrementally in a test directory. You can make use of the "execute the deepest .htaccess file on the path feature" to set up a separate test directory (tree) and debug rulesets here without screwing up your main rules and stopping your site working. You have to add them one at a time because this is the only way to localise failures to individual rules.
Use a dummy script stub to dump out server and environment variables. (See Listing 2)If your app uses, say, blog/index.php then you can copy this into test/blog/index.php and use it to test out your blog rules in the test subdirectory. You can also use environment variables to make sure that the rewrite engine in interpreting substitution strings correctly, e.g.
RewriteRule ^(.*) - [E=TEST0:%{DOCUMENT_ROOT}/blog/html_cache/$1.html]
and look for these REDIRECT_* variables in the phpinfo dump. BTW, I used this one and discovered on my site that I had to use %{ENV:DOCUMENT_ROOT_REAL} instead. In the case of redirector looping REDIRECT_REDIRECT_* variables list the previous pass. Etc..
Make sure that you don't get bitten by your browser caching incorrect 301 redirects. See answer below. My thanks to Ulrich Palha for this.
The rewrite engine seems sensitive to cascaded rules within an .htaccess context, (that is where a RewriteRule results in a substitution and this falls though to further rules), as I found bugs with internal sub-requests (1), and incorrect PATH_INFO processing which can often be prevents by use of the [NS], [L] and [PT] flags.
Any more comment or suggestions?
Listing 1 -- phpinfo
<?php phpinfo(INFO_ENVIRONMENT|INFO_VARIABLES);
Here are a few additional tips on testing rules that may ease the debugging for users on shared hosting
1. Use a Fake-user agent
When testing a new rule, add a condition to only execute it with a fake user-agent that you will use for your requests. This way it will not affect anyone else on your site.
e.g
#protect with a fake user agent
RewriteCond %{HTTP_USER_AGENT} ^my-fake-user-agent$
#Here is the actual rule I am testing
RewriteCond %{HTTP_HOST} !^www\.domain\.com$ [NC]
RewriteRule ^ http://www.domain.com%{REQUEST_URI} [L,R=302]
If you are using Firefox, you can use the User Agent Switcher to create the fake user agent string and test.
2. Do not use 301 until you are done testing
I have seen so many posts where people are still testing their rules and they are using 301's. DON'T.
If you are not using suggestion 1 on your site, not only you, but anyone visiting your site at the time will be affected by the 301.
Remember that they are permanent, and aggressively cached by your browser.
Use a 302 instead till you are sure, then change it to a 301.
3. Remember that 301's are aggressively cached in your browser
If your rule does not work and it looks right to you, and you were not using suggestions 1 and 2, then re-test after clearing your browser cache or while in private browsing.
4. Use a HTTP Capture tool
Use a HTTP capture tool like Fiddler to see the actual HTTP traffic between your browser and the server.
While others might say that your site does not look right, you could instead see and report that all of the images, css and js are returning 404 errors, quickly narrowing down the problem.
While others will report that you started at URL A and ended at URL C, you will be able to see that they started at URL A, were 302 redirected to URL B and 301 redirected to URL C. Even if URL C was the ultimate goal, you will know that this is bad for SEO and needs to be fixed.
You will be able to see cache headers that were set on the server side, replay requests, modify request headers to test ....
Online .htaccess rewrite testing
I found this Googling for RegEx help, it saved me a lot of time from having to upload new .htaccess files every time I make a small modification.
from the site:
htaccess tester
To test your htaccess rewrite rules, simply fill in the url that you're applying the rules to, place the contents of your htaccess on the larger input area and press "Check Now" button.
Don't forget that in .htaccess files it is a relative URL that is matched.
In a .htaccess file the following RewriteRule will never match:
RewriteRule ^/(.*) /something/$s
Set environment variables and use headers to receive them:
You can create new environment variables with RewriteRule lines, as mentioned by OP:
RewriteRule ^(.*) - [E=TEST0:%{DOCUMENT_ROOT}/blog/html_cache/$1.html]
But if you can't get a server-side script to work, how can you then read this environment variable? One solution is to set a header:
Header set TEST_FOOBAR "%{REDIRECT_TEST0}e"
The value accepts format specifiers, including the %{NAME}e specifier for environment variables (don't forget the lowercase e). Sometimes, you'll need to add the REDIRECT_ prefix, but I haven't worked out when the prefix gets added and when it doesn't.
Make sure that the syntax of each Regexp is correct
by testing against a set of test patterns to make sure that is a valid syntax and does what you intend with a fully range of test URIs.
See regexpCheck.php below for a simple script that you can add to a private/test directory in your site to help you do this. I've kept this brief rather than pretty. Just past this into a file regexpCheck.php in a test directory to use it on your website. This will help you build up any regexp and test it against a list of test cases as you do so. I am using the PHP PCRE engine here, but having had a look at the Apache source, this is basically identical to the one used in Apache. There are many HowTos and tutorials which provide templates and can help you build your regexp skills.
Listing 1 -- regexpCheck.php
<html><head><title>Regexp checker</title></head><body>
<?php
$a_pattern= isset($_POST['pattern']) ? $_POST['pattern'] : "";
$a_ntests = isset($_POST['ntests']) ? $_POST['ntests'] : 1;
$a_test = isset($_POST['test']) ? $_POST['test'] : array();
$res = array(); $maxM=-1;
foreach($a_test as $t ){
$rtn = #preg_match('#'.$a_pattern.'#',$t,$m);
if($rtn == 1){
$maxM=max($maxM,count($m));
$res[]=array_merge( array('matched'), $m );
} else {
$res[]=array(($rtn === FALSE ? 'invalid' : 'non-matched'));
}
}
?> <p> </p>
<form method="post" action="<?php echo $_SERVER['SCRIPT_NAME'];?>">
<label for="pl">Regexp Pattern: </label>
<input id="p" name="pattern" size="50" value="<?php echo htmlentities($a_pattern,ENT_QUOTES,"UTF-8");;?>" />
<label for="n"> Number of test vectors: </label>
<input id="n" name="ntests" size="3" value="<?php echo $a_ntests;?>"/>
<input type="submit" name="go" value="OK"/><hr/><p> </p>
<table><thead><tr><td><b>Test Vector</b></td><td> <b>Result</b></td>
<?php
for ( $i=0; $i<$maxM; $i++ ) echo "<td> <b>\$$i</b></td>";
echo "</tr><tbody>\n";
for( $i=0; $i<$a_ntests; $i++ ){
echo '<tr><td> <input name="test[]" value="',
htmlentities($a_test[$i], ENT_QUOTES,"UTF-8"),'" /></td>';
foreach ($res[$i] as $v) { echo '<td> ',htmlentities($v, ENT_QUOTES,"UTF-8"),' </td>';}
echo "</tr>\n";
}
?> </table></form></body></html>
One from a couple of hours that I wasted:
If you've applied all these tips and are only going on 500 errors because you don't have access to the server error log, maybe the problem isn't in the .htaccess but in the files it redirects to.
After I had fixed my .htaccess-problem I spent two more hours trying to fix it some more, even though I simply had forgotten about some permissions.
Make sure you use the percent sign in front of variables, not the dollar sign.
It's %{HTTP_HOST}, not ${HTTP_HOST}. There will be nothing in the error_log, there will be no Internal Server Errors, your regexp is still correct, the rule will just not match. This is really hideous if you work with django / genshi templates a lot and have ${} for variable substitution in muscle memory.
If you're creating redirections, test with curl to avoid browser caching issues.
Use -I to fetch http headers only.
Use -L to follow all redirections.
Regarding 4., you still need to ensure that your "dummy script stub" is actually the target URL after all the rewriting is done, or you won't see anything!
A similar/related trick (see this question) is to insert a temporary rule such as:
RewriteRule (.*) /show.php?url=$1 [END]
Where show.php is some very simple script that just displays its $_GET parameters (you can display environment variables too, if you want).
This will stop the rewriting at the point you insert it into the ruleset, rather like a breakpoint in a debugger.
If you're using Apache <2.3.9, you'll need to use [L] rather than [END], and you may then need to add:
RewriteRule ^show.php$ - [L]
At the very top of your ruleset, if the URL /show.php is itself being rewritten.
Some mistakes I observed happens when writing .htaccess
Using of ^(.*)$ repetitively in multiple rules, using ^(.*)$ causes other rules to be impotent in most cases, because it matches all of the url in single hit.
So, if we are using rule for this url sapmle/url it will also consume this url sapmle/url/string.
[L] flag should be used to ensure our rule has done processing.
Should know about:
Difference in %n and $n
%n is matched during %{RewriteCond} part and $n is matches on %{RewriteRule} part.
Working of RewriteBase
The RewriteBase directive specifies the URL prefix to be used for
per-directory (htaccess) RewriteRule directives that substitute a
relative path.
This directive is required when you use a relative path in a
substitution in per-directory (htaccess) context unless any of the
following conditions are true:
The original request, and the substitution, are underneath the
DocumentRoot (as opposed to reachable by other means, such as Alias).
The filesystem path to the directory containing the RewriteRule,
suffixed by the relative substitution is also valid as a URL path on
the server (this is rare). In Apache HTTP Server 2.4.16 and later,
this directive may be omitted when the request is mapped via Alias or
mod_userdir.
I found this question while trying to debug my mod_rewrite issues, and it definitely has some helpful advice. But in the end the most important thing is to make sure you have your regex syntax correct. Due to problems with my own RE syntax, installing the regexpCheck.php script was not a viable option.
But since Apache uses Perl-Compatible Regular Expressions (PCRE)s, any tool which helps writing PCREs should help. I've used RegexPlanet's tool with Java and Javascript REs in the past, and was happy to find that they support Perl as well.
Just type in your regular expression and one or more example URLs, and it will tell you if the regex matches (a "1" in the "~=" column) and if applicable, any matching groups (the numbers in the "split" column will correspond to the numbers Apache expects, e.g. $1, $2 etc.) for each URL. They claim PCRE support is "in beta", but it was just what I needed to solve my syntax problems.
http://www.regexplanet.com/advanced/perl/index.html
I'd have simply added a comment to an existing answer but my reputation isn't yet at that level. Hope this helps someone.
In case you are not working in an standard shared hosting environment, but in one to which you have administration access (maybe your local test environment), make sure that use of .htaccess and mod_rewrite are enabled. They are disabled in a default Apache installation. And in that case, no action configured in your .htaccess file works, even if the regexes are perfectly valid.
To enable the use of .htaccess:
Find file apache2.conf, on Debian/Ubuntu this is in /etc/apache2, and within the file the section
<Directory /var/www/>
Options Indexes FollowSymLinks
AllowOverride None
Require all granted
</Directory>
and change the line AllowOverride None to AllowOverride All.
To enable module mod_rewrite:
On Debian/Ubuntu, execute
sudo a2enmod rewrite
By the way, to disable a module, you would use a2dismode instead of a2enmode.
After you did the above configuration changes, restart Apache for them to take effect:
sudo systemctl restart apache2
If you're planning on writing more than just one line of rules in .htacesss,
don't even think about trying one of those hot-fix methods to debug it.
I have wasted days setting multiple rules, without feedback from LOGs, only to finally give up.
I got Apache on my PC, copied the whole site to its HDD, and got the whole rule-set sorted out, using the logs, real fast.
Then I reviewed my old rules, which been working. I saw they were not really doing what was desired. A time bomb, given a slightly different address.
There are so many pit falls in rewrite rules, it's not a straight logic thing at all.
You can get Apache up and running in ten minutes, it's 10MB, good license, *NIX/WIN/MAC ready, even without install.
Also, check the header lines of your server and get the same version of Apache from their archive if it's old. My OP is still on 2.0; many things are not supported.
I'll leave this here, maybe obvious detail, but got me banging my head for hours:
be careful using %{REQUEST_URI} because what #Krist van Besien say in his answer is totally right, but not for the REQUEST_URI string, because the out put of this TestString starts with a /. So take care:
RewriteCond %{REQUEST_URI} ^/assets/$
^
| check this pesky fella right here if missing
Best way to debug it!
Add LogLevel notice rewrite:trace8 to the httpd.conf of apache to log all notices of mod_rewrite. If you are at shared hosting and don't have access to httpd.conf then test it locally and upload to the live site. Once enabled this generate a very large log in very short time, it means it can't be tested on live server anyway.
(Similar to Doin idea)
To show what is being matched, I use this code
$keys = array_keys($_GET);
foreach($keys as $i=>$key){
echo "$i => $key <br>";
}
Save it to r.php on the server root and then do some tests in .htaccess
For example, i want to match urls that do not start with a language prefix
RewriteRule ^(?!(en|de)/)(.*)$ /r.php?$1&$2 [L] #$1&$2&...
RewriteRule ^(.*)$ /r.php?nomatch [L] #report nomatch and exit
as pointed out by #JCastell, the online tester does a good job of testing individual redirects against an .htaccess file. However, more interesting is the api exposed which can be used to batch test a list of urls using a json object. However, to make it more useful, I have written a small bash script file which makes use of curl and jq to submit a list of urls and parse the json response into a CSV formated output with the line number and rule matched in the htaccess file along with the redirected url, making it quite handy to compare a list of urls in a spreadsheet and quickly determine which rules are not working.
Perhaps the best way to debug rewrite rules is not to use rewrite rules at all, but to defer URL processing from the htaccess file to a PHP file (let's call it router.php). Then, you can use PHP to do any manipulating you like, with proper error detection and the usual ways to do debugging. This even runs faster, too, since you don't have to use the rewriting module.
To transfer control immediately from .htaccess to router.php for any URL that is not found in the file system, just put the following line in .htaccess:
FallbackResource router.php
Yes, it's really that easy. And yes, it really works. Give it a try.
Note: You may need an ErrorDocument directive in your .htacess file to transfer control explicitly for certain URLs to your router.php file on HTTP status 404, especially if you inherit from a parent htaccess file that handles status 404. So that would make it a total of two lines to transfer control to a router file.
If you are working with url, You might want to check if you "Enable Mod Rewrite"
I have the following situation:
On my webserver I have an instance of websvn running, where specific repositories and revisions can be accessed by a URL like
http://www.myhost.com/listing.php?repname=repository1&path=%2Ftrunk%2Fbackend
Somehow, out there in the wild, a wrong URL is being used to access this
http://www.myhost.com/listing.php/?repname=repository1&path=%2Ftrunk%2Fbackend
(Notice the slash after listing.php)
Now, although the URL works and websvn still shows the webpage, images and stylesheets do not get loaded correctly, since they are referenced relative.
I tried to add an .htaccess file to the webroot to redirect people accessing the file as directory to the correct URL.
I have tried multiple variations and ended up with this file:
RewriteEngine on
RewriteRule ^/listing.php/ listing.php [R=301,QSA]
But, since I am writing here, you already guessed it: It doesn't work.
I also tried
RewriteEngine on
RewriteRule ^/listing.php(.*) listing.php$1 [R=301,QSA]
What am I doing wrong?
Perhaps among other things, a RewriteRule within .htaccess that starts with “^/” will never match anything at all. (Examples that include a leading slash are for the global configuration file.) Remove the leading forward slash and see if that helps.
Also, I recommend changing the 301 to a 307 until you get it working. Otherwise, your browser will cache the 301 result, redirecting on subsequent references without consulting your server at all and likely giving you very confusing results.