The way the class number formula for Dirichlet L-functions is always proved in modern textbooks and notes is
You prove the general class number formula for $\zeta_K(s)$.
You prove that for quadratic fields, $\zeta_K(s)=\zeta(s)L(s,\chi)$.
Use 1. and 2. to evaluate $L(1,\chi)$.
Dirichlet proved the same formula for $L(1,\chi)$ before Dedekind proved the formula for $\zeta_K(s)$. I hear the he used the theory of quadratic forms.
Is there a proof of the formula for $L(1,\chi)$ in the language of ideals which doesn't use the class number formula for $\zeta_K(1)$ or the theory of quadratic forms?