Say I have a problem which is of the form, where $W(t)$ is a Brownian Motion and $f(a,b)$ is just some function where we will plug in terms $u, W(u)$. For this example, let's say that $f(a,b) = b-a^2b$. If I have the following form:
$$ \int^{t}_0 W(t)f\big(u,W(u)\big)dW(u), $$
would I be able to take out the $W(t)$ and treat it as if it were a constant so that we have the form
$$ W(t)\int^{t}_0 f\big(u,W(u)\big)dW(u)? $$
If not how would I go about solving such a problem?
The problem that I see with the first integral is that it might not be well-defined. Following Øksendal, Definition 3.1.4, the integrand, let us call it $F(s,\omega)$ with $s:=u$, must satisfy,
Clearly, this does not seem to be the case for your integral, as it depends on $W(t)$ with $t\ge s$ for all $s$.