QByteArray::fromPercentEncoding incorrectly decodes strings which have percentages (%) followed by 2 characters where either character is not a hex digit.

I.e.

QByteArray::fromPercentEncoding("%$&")

returns the character 'f', while it should leave the entire sequence untouched.

This bug happens because the code decodes any sequence following a percentage, while RFC 3986 only says to decode percent-encoded octets (specified to be a percentage followed by two hex digits in section 2.1)