HHVM failing to parse some valid data: URI

[dunglas]

With PHP:

php > echo file_get_contents('data:text/plain;charset=utf-8;base64,SGVsbG8gV29ybGQh');
Hello World!

With HHVM:

hphpd> echo file_get_contents('data:text/plain;charset=utf-8;base64,SGVsbG8gV29ybGQh');
echo file_get_contents('data:text/plain;charset=utf-8;base64,SGVsbG8gV29ybGQh');

Warning: rfc2396: invalid parameter

[r3wt]

Isn't that a bit ridiculous? Just because PHP lets you do comepletely redundant things doesn't mean hhvm should as well.

[dunglas]

This is not related to PHP. Similar URIs are used in the wild (and it's how I found this problem first).

[r3wt]

Does it choke with base64_decode as well?

[SiebelsTim]

It chokes on the charset. When you remove the charset it works fine.

[dunglas]

Yes the problem is only the charset.

[jwatzman]

https://3v4l.org/VmPKS

[lexidor]

The rfc to cite is 2397. The hhvm error message threw me to 2396. Adding a charset this way is spec compliant.

The data Uri parser in hhvm is (maybe was) not following the spec properly.