404 Errors from RTL characters appended to URL

[AndyBeard]

This is an interesting bug I found via webmaster tools

Example URL

​http://andybeard.eu/2210/google-stopbadware.html%20-%20%D8%A8%D8%B1%DB%8C%D8%AA%D8%A7%D9%86%DB%8C

The URL has some arabic characters added to it I assume by mistake, but is constructed to theoretically handle quite a lot of errors with URL formation by using the post ID at the start.

This results in a 404 error so can't even be handled by canonicalization tags.

[AndyBeard]

Another example showing it isn't just RTL
​http://andybeard.eu/2007/02/ultimate-list-of-dofollow-plugins-banish-nofollow-from-comments-and-trackbacks.html%3Cbr%20/%3E

Without the trailing characters it should (and does) redirect to
​http://andybeard.eu/434/ultimate-list-of-dofollow-plugins-banish-nofollow-from-comments-and-trackbacks.html

[markjaquith]

The slug is wrong... we can't really determine where the legit slug stops and the extra characters begin. I think this is wontfix for now.

[AndyBeard]

The slug shouldn't be needed on the first example because you have the page ID

If you chop off the characters it works

​http://andybeard.eu/2210/

[SergeyBiryukov]

If a permalink structure contains post ID, we can probably use it as the essential part and cut everything else. However this seems to be an edge case which can be solved by a custom rewrite rule.