a www. b

Yields:

a www. b

It seems that when www. is found, it is then parsed again as a domain, of which the dot makes it valid, and then it gets truncated to exclude the dot.
Instead, I think a domain (with a dot) should be parsed after www..

a www.example b (probably not this one either)
a www.example.com b (this should of course work)

a www.example b (probably not this one either)
a www.example.com b (this should of course work)

Aside, it seems like dots aren’t needed at all:

https://example.com

https://example

http://example.com

http://example.com

www.example.com

www.example

https://example.com

https://example

http://example.com

http://example.com

www.example.com

www.example