Wrong Rule pattern matchs

Hi,

In the Route66Rule class, where the regex pattern is generated, it is currently generated in this way:

$this->regex = '/' . addcslashes(preg_replace('/{(.*?)}/', '(.*?)', $this->pattern), '/') . '$/';

So, a K2 rule such us:

"{categoryAlias}/{itemDate}/{itemId}-{itemAlias}"

generates this pattern:

"/(.*?)\/(.*?)\/(.*?)-(.*?)$/"

Please, note that "." matches any character except newline (by default) (https://www.php.net/manual/en/regexp.reference.meta.php), including /.

We are finding cases where itemId is matching "item/99999", so itemId (a number) has "item/" as prefix. To give you a case, the current K2 URL is:

"/prensa/2019-05-02/14284-dia-del-periodista-comienza-la-recepcion-de-trabajos-para-el-concurso-anual"

But, the site has to handle the old URL:

"/secciones/prensa/item/14284-dia-del-periodista-comienza-la-recepcion-de-trabajos-para-el-concurso-anual"

This case is tricky since the site has a category (menu) with alias "prensa" and a username "prensa".

In any case, a better Rule expression generator is:

$this->regex = '/' . addcslashes(preg_replace('/{(.*?)}/', '([a-z0-9\-]*?)', $this->pattern), '/') . '$/';

I'm sure that this restricted pattern is going to be better in many aspects (though it doesn't take into account Unicode Aliases) ... please, confirm if it's safe to apply this workaround to solve the case, or if there is a better solution for the case.

Thanks

Second case, "/component/k2/item/19468-cuidemos-el-agua-en-estas-fiestas", when there is no K2 menu item.

The pattern "/(.*?)\/(.*?)\/(.*?)-(.*?)$/" produces this array:

"Array
(
[0] => Array
(
[0] => component/k2/item/19468-cuidemos-el-agua-en-estas-fiestas
[1] => component
[2] => k2
[3] => item/19468
[4] => cuidemos-el-agua-en-estas-fiestas
)

)
"

If I may suggest an improvement, a better generation of the K2 pattern:

"{categoryAlias}/{itemDate}/{itemId}-{itemAlias}"

It should have:

itemDate as [0-9][0-9][0-9][0-9]-[0-9][0-9]-[0-9][0-9]
itemId as [0-9]+

The last update, this would be the best pattern for our case:

$regex = '/([\w[:punct:]\-]*?)\/([\d\-]*?)\/(\d*?)-([\w[:punct:]\-]*?)$/u';

It is strict, avoid false matches and supports Unicode Aliases such us:

"osse,-cuidémos/2019-12-24/19468-en-estas-fiestas,-cuidémos-el-agua"

Thanks

One more, the usage of *? looks unnecessary and it can generate false matches:

$regex = '/([\w[:punct:]\-]+)\/([\d\-]+)\/(\d+)-([\w[:punct:]\-]+)$/u';

Hi,

About "/component/k2/item/19468-cuidemos-el-agua-en-estas-fiestas", when there is no matching menu item, this is the kind of URLs that you get from Joomla by default.

This week I have already hard coded the restricted pattern and it has been running without issues.

We can't grant you access to the site.

Best regards,

Hi,

When you enable Route 66, the site must also support the incoming links with URLs such as "/component/k2/item/19468-cuidemos-el-agua-en-estas-fiestas". They were generated in the past, so they must reach the right content items.

You can find more examples of thes type os Joomla URLs here: https://getk2.org

/component/user/reset
/component/socialconnect/login/facebookOauth

On your site, it is also used on an Ajax script:

<script type="application/json" class="joomla-script-options new">{"csrf.token":"393cfa9563e820e33e9db1ad69eec883","system.paths":{"root":"","base":""},"system.keepalive":{"interval":3540000,"uri":"\/component\/ajax\/?format=json"}}</script>

So, they are not hardcode on my site, and they frequently found on Joomla sites. If you test only with a perfect site, with only the most common links; then, of course, you are not going to find these additional cases to support. I've detailed here what I've found on the migration of a high-traffic news site with 20K articles created in 3 years. After the implementation of these improvements, it is working Ok and I haven't received more reports in these days.

From my view, you can keep the current rule engine as it is working today. However, it is too error-prone and it can easily misroute URLs. As I understand the extension should extend the router with the new rules and keep the previous routing consistent, reaching the same content items and returning the same HTTP responses.

Regards