Non existing url on russian lang redirected to main page

[akudryav]

If you write some blabla url using russian like this: site.ext\проверка no 404 error happens.
Tested on Invo and Vokuro applications.

[andresgutierrez]

Do you have namespaces, controllers or actions with names in Russian?

[sergeyklay]

@andresgutierrez
Russia uses IPv6. This allows you to use a URLs with national symbols encoding. eg http://президент.рф

[stibiumz]

Also occurs with Latin-1 chars
https://invo.phalconphp.com/eñe -> Found
https://invo.phalconphp.com/ene -> Not found (unauthorized)

@sergeyklay this is IDN (not the same as an url path), and it is also supported by IPv4 addresses

[andresgutierrez]

@stibiumz you have controllers in russian?

[stibiumz]

I have not. Why does that matter?

Edit:

In Phalcon 1.3.4 works (with annotations router), I tried it now

[roman-kulish]

@sergeyklay IPv4 vs IPv6 does not matter.

For Russian it is either "%D0%BF%D1%80%D0%BE%D0%B2%D0%B5%D1%80%D0%BA%D0%B0" is not decoded back to "проверка" or preg_match in Route::compilePattern() should use Unicode flag.

For "eñe" vs "ene" they are two different words and I think some kind of a http://php.net/manual/en/class.normalizer.php should be used to process is or a special route to cover both :)

[akudryav]

There is no any controllers or actions with russian names, seems that this is routing bug.

[stibiumz]

@roman-kulish I know, the example was for illustrating, what I mean is that 'ene' route is recognized as not found as when with 'eñe' simply goes to index page, it's strange.

About matching eñe as if were ene, can be done with Transliterator, Normalizer is for character de-/composition

[andresgutierrez]

I think you can create the router using the #u operator:

$router = new \Phalcon\Mvc\Router(false);

$router->add("#^/([\w\\_\\-]+)[/]{0,1}$#u", [
    "controller" => 1
]);

$router->add("#^/([\w0-9\\_\\-]+)/([\w0-9\\.\\_]+)(/.*)*$#u", [
      "controller" => 1,
      "action"  =>  2,
      "params"  =>  3
]);

[hakimio]

I can see this bug too. I think by default Phalcon should just give 404 page in case non-latin words are used and no route can be matched instead of just totally disregarding the non-latin path.

[andresgutierrez]

Fixed in the 2.0.x branch