I am reading the proof of the estimates of Whittaker functions from Jacquet, Hervé, Ilja Iosifovitch Piatetski-Shapiro, and Joseph Shalika. "Automorphic forms on GL (3) I." Annals of Mathematics 109.1 (1979): 169-212. I am referring to Proposition 2.2 (page 181). I don't understand a step in the proof of Lemma 2.2.1 - how from the equality $$\begin{equation} \phi \left( a_1, \dots , a_{n-1}, bc \right) = \sum \eta\left(b\right) \left( A_\eta \phi \right) \left( a_1, a_2, \dots, a_{n-1}, c \right) \end{equation}$$ for a fixed $ c $ and all $b$ with $ \left| b \right| \le 1$ they conclude that $$\begin{equation} \phi \left( a_1, \dots , a_n \right) = \sum_{j, \eta} \eta\left( a_n \right) \phi_j \left( a_n \right) \phi \left( a_1, a_2, \dots, a_{n-1}, b_j \right) \end{equation}$$ for all $ a_1, \dots, a_n \in F^{\times} $, where $ b_j \in F^{\times}$ are constants depending on $ \phi $ only, and $ \phi_j $ are Schwartz functions.
I understand why this is true for $ a_n $ with $ \left| a_n \right| \le \left| c \right| $ if one extends the set $ Y $ of finite functions such that it spans the translation spaces of its elements, but I don't understand how to deduce this equality in the case that some $ \left| a_i \right| $ $(i \ne n)$ is small and $\left| a_n \right| > \left| c \right|$.
My original thought was that the difference between the functions can be thought as a Schwartz function on $ F^n $, but I'm not sure what happens when $ a_i $ is close to zero ($i \ne n$).
Thanks