I've noticed that some manpages have non-printing characters in them; for instance, std::rand has this line (should be 0):

According to grep, there's around 200 files that have the characters, which, comparing to the website versions, seem to almost always be characters in <span> tags, such as 0, a ;, a (, or whitespace, like the below:

1. whitespace (I think)
   

   stdman/man/std::accumulate.3

   Line 42 in 478d353

   value of type Ret.ý¿¿½£¬

2. ;
   

   stdman/man/std::condition_variable::wait_until.3

   Line 66 in 478d353

   bool pred()ý¿¿½£­

3. (
   

   stdman/man/std::iterator_traits.3

   Line 71 in 478d353

   pointer decltypeý¿¿½£¬std::declval<Iter&>().operator->()) if valid,

Some unicode characters also don't seem to be handled correctly:

It'd be nice if all of these could be parsed and rendered properly.