I always found the "use int(abs(z))" part of that message odd,
as well.  As Georg points out, there are many possible ways that
one might want to convert complex to int;  it seems strange to
give advice for one particular one when that may well not match
what the user wanted.

I'd suggest just dropping the "use int(abs(z))" from the error message.
I think it's more likely to be confusing than helpful.

The same applies to float(complex).