Your strategy would work, but if possible you might be better off using XmaI rather than SmaI, as the latter is a blunt-end cutting enzyme.
Furthermore, your strategy would not replace the entire _gpdA_ promoter, if that's what you're after. You might wanna refer to their paper, where it's clear from Fig. 1 that the promoter region is much longer than the region between SalI and SmaI:
![Fig 1. from Crespo-Sempere et al. \(2011\)](