Heavy-duty vehicles (HDVs) are a major source of road transportation emissions, yet their electrification remains challenging due to operational constraints. Opportunity charging, which enables electric vehicles to recharge without interrupting regular operations, offers a practical solution for electrifying certain HDV sectors. Motivated by a real-world electrification planning problem in seaport container drayage, this paper addresses the large-scale electrification planning problem for HDV fleets through opportunity charging, optimizing both fleet composition and charging infrastructure deployment. We formulate the problem as a mixed-integer linear program and propose an exact solution algorithm based on Benders decomposition. The algorithm splits the problem into a master problem for fleet composition and charging infrastructure deployment and a large number of subproblems for en-route charging scheduling. In view of the computational burden from solving the master and subproblems, we develop novel acceleration strategies, including problem-specific valid inequalities and subproblem reduction techniques. The algorithm was tested using instances generated from data of a real-world electrification project. The results demonstrate that the strategies significantly improve the performance of the algorithm and that our approach solves realistic-scale instances (e.g., 50 vehicles and over 50 candidate charging stations) to near-optimality within practical computational times. The results also reveal that opportunity charging can achieve cost reductions of over 30% and emissions reductions of up to 94% compared to conventional fleets, while maintaining operational feasibility.