Setting up a page that will intercept all pages that do not already exist?

.oO(Andrew)

>Using the ErrorDocument 404 syntax to redirect everything not found to
>a PHP script will issue a 404 header to the browser,

Not necessarily. In case of an error the server will call the error
document. If that is a static HTML document it will be sent with an
error code back to the user agent. If it's a script you can do whatever
you want. If you want to redirect to another address the user agent will
get the redirection code, but not the error code (there can be only one
status code in the response header).

>but none of the
>browsers I have tried, Opera, Firefox or IE show an error page provided
>the header is followed by html.

They don't show an error because they don't receive an error code.

Micha

>Using the ErrorDocument 404 syntax to redirect everything not found to
>a PHP script will issue a 404 header to the browser, but none of the
>browsers I have tried, Opera, Firefox or IE show an error page provided
>the header is followed by html.
>
>If it works don't knock it, and you can always combine the method with
>some static html caching.
>
>The only other alternative I can think of is to use mod_rewrite in
>apache somehow.

According to the Apache manual:
ErrorDocument 404 /missing.php
sends the result of running missing.php back to the browser
as though it were from the nonexistent page.
ErrorDocument 404 http://www.site.domain/missing.php
sends a REDIRECT to the browser, even if the two statements
above refer to the same script on the same host.

The two are NOT equivalent. The first leaves the bad link in the
browser address bar; the second updates it. The first gives the
REDIRECT_* variables if the error document is a CGI or PHP script;
the second does not. The first refers to a local script; the
second might refer to a different server.

It is worth nothing that if you use ErrorDocument 404 (with Apache)
to redirect to a local PHP script, using the first form of the
ErrorDocument above (no http:) you get some useful variables
handed to you which can be useful for logging for perusal by
the webmaster or handling the error (e.g. try to match the request
to the URL closest in spelling to it.).

$_SERVER['REDIRECT_URL'] is the path/filename of the bad request on
the local server relative to the document root.
$_SERVER['REDIRECT_STATUS'] is the error code, e.g. 404
$_SERVER['REDIRECT_ERROR_NOTE'] gives some indication of the
error (e.g. what file was missing)
$_SERVER['HTTP_REFERER'] (no REDIRECT_ in front of this one)
is the URL of the page with the bad link on it.
$_SERVER['REDIRECT_REQUEST_METHOD'] GET, POST, or whatever to the bad link.

A CGI or PHP script should be able to give a Status: header to
override any error code in the response. It's my opinion that the
status code coming back from an error document is the status code
of the error document (e.g. not 404 unless the document goes out
of its way to return that with a Status: header) for the local
script case, and a redirect code when ErrorDocument is given a full
URL with http:. My browser is not cooperative enough to let me
verify this information, though.

Gordon L. Burditt

Joshua Beall wrote:
> Hi All,
>
> I would like to setup my site so that when a page is request that does not
> exist, transfer is instead passed to a certain page that I specify. For
> instance, if the user requests http://mysite.com/pages/contact and that is
> not a valid page or directory, then the request is handed to
> http://mysite.com/pages/display.php or something like that.
>
> I am wondering what the best way to do this is - I know that I could use an
> .htaccess file to redirect 404 errors, but I would rather not give a 404
> error in the first place. Is there a better/other way? I seem to remember
> having read about another way... but I can't seem to remember...
>
> This is in a LAMP environment, and does not need to be portable to Windows
> or any other webserver.
>
> Any help would be greatly appreciated!

Have you seen the thread I started, "Include" security?, I think you're
looking for the same thing I am. The script in the reply I replied to
lets you set a 404 page, or you could set it to any page.

I have experienced a few web hosts that will not allow you to redirect
404's to a PHP script, you receive a message saying something like "A
404 error occured additionally an error document was not found the
handle the request"

Also remember, any redirection you do will lose the POST data, but you
can retreive the GET data using a piece of code like this:

if (isset($_SERVER['REDIRECT_QUERY_STRING']))
{
$redirected_queries = explode('&', $_SERVER['REDIRECT_QUERY_STRING']);

foreach ($redirected_queries as $redirected_query)
{
$redirected_query = explode('=', $redirected_query, 2);
$_GET[urldecode($redirected_query[0])] =
urldecode($redirected_query[1]);
}

unset($redirected_queries, $redirected_query);
}

This will import all the redirected GET variables into the $_GET global
(there shouldn't be any in there to overwrite though). Oh and the
double urldecode is because & and = may be present in the data itself,
so you can't really do the urldecoding before exploding.

Michael Fesser <netizen@gmx.net> wrote:
>>but none of the
>>browsers I have tried, Opera, Firefox or IE show an error page provided
>>the header is followed by html.
>
> They don't show an error because they don't receive an error code.

But they _will_ receive the error code. IMHO only IE is so silly to
display it's own error page under certain conditions (less than x bytes
IIRC).

.oO(Daniel Tryba)

>Michael Fesser <netizen@gmx.net> wrote:
>>>but none of the
>>>browsers I have tried, Opera, Firefox or IE show an error page provided
>>>the header is followed by html.
>>
>> They don't show an error because they don't receive an error code.
>
>But they _will_ receive the error code.

OK, my statement above was misleading (I should read the posts more
carefully).

Of course browsers will receive the error code, if the error document is
delivered as-is or doesn't send its own status header. But if the error
document is a script and redirects to another address the browser will
get only the redirection code.

Micha

Joshua Beall wrote:

> Hi All,
>
> I would like to setup my site so that when a page is request that does not
> exist, transfer is instead passed to a certain page that I specify. For
> instance, if the user requests http://mysite.com/pages/contact and that is
> not a valid page or directory, then the request is handed to
> http://mysite.com/pages/display.php or something like that.
>
> I am wondering what the best way to do this is - I know that I could use an
> .htaccess file to redirect 404 errors, but I would rather not give a 404
> error in the first place. Is there a better/other way? I seem to remember
> having read about another way... but I can't seem to remember...
>
> This is in a LAMP environment, and does not need to be portable to Windows
> or any other webserver.
>
> Any help would be greatly appreciated!
>
> His,
> -Josh
>
>

You would just setup the 404 redirect for any arbitrary page, it doesn't
have to be an error page.

It's good to know in your logs that people are receiving 404s even if
they never realize it on their end, so you can be aware that you may
have bad links somewhere.

Hope this helps.

Adam

There is a good reason to not use the 404 handler.
Try using W3C's validator service (http://validator.w3.org/) on a page
using this method.
It will throw you an error saying the page could not be validated
because it was not found.

My point is, you can't judge how services like this or search engines
will react.

.oO(Andrew)

>There is a good reason to not use the 404 handler.
>Try using W3C's validator service (http://validator.w3.org/) on a page
>using this method.

www.php.net uses error handlers to allow URL shortcuts and it works.

>It will throw you an error saying the page could not be validated
>because it was not found.

I would guess that there's something wrong with the server then.
If done properly the requesting user agent won't notice that the
delivered content is the output of an error handler.

>My point is, you can't judge how services like this or search engines
>will react.

Sure, but there's not much a difference to any other user agents.
Request -> response, request -> response. It's always the same, even if
some UAs or servers might send some strange headers.

Micha