The two equations you cite:
$$
C_P-C_V=-T\left( \frac{\partial V}{\partial T} \right)^2_P \left( \frac{\partial P}{\partial V} \right)_T
$$
and
$$
\frac{C_P}{C_V}= \frac{\left( \frac{\partial P}{\partial V} \right)_S}{\left( \frac{\partial P}{\partial V} \right)_T}
$$
can only provide ine $C_P$ and $C_V$ in terms of the thermal expansion coefficient
$$
\alpha = \frac{1}{V}\left( \frac{\partial V}{\partial T} \right)_P
$$
and the adiabatic and isothermal compressibilities $\chi_S$ and $\chi_T$ defined as
$$
\chi_S = -\frac{1}{V}\left( \frac{\partial V}{\partial P} \right)_S; ~~~~~~~~~~~~~~ \chi_T = -\frac{1}{V}\left( \frac{\partial V}{\partial P} \right)_T.
$$
The problem is how we can get the adiabatic compressibility if by van der Waals gas we intend a system of atoms such that the equation of state $p=p(N,T,V)$ coincides with van der Waals' equation of state. It should be clear that the equation of state is not enough to reconstruct any thermodynamic potential. Therefore, some information is missing and should be added to the equation of state.