In this paper, we consider a multi-hop energy harvesting (EH) communication system in a fullduplex mode, where arrival data and harvested energy curves in the source and the relays are modeled as general functions. This model includes the EH system with discrete arrival processes as a special case. We investigate the throughput maximization problem considering minimum utilized energy in the source and relays and find the optimal offline algorithm. We show that the optimal solution of the twohop transmission problem have three main steps: (i) Solving a point-to-point throughput maximization problem at the source; (ii) Solving a point-to-point throughput maximization problem at the relay (after applying the solution of first step as the input of this second problem); (iii) Minimizing utilized energy in the source. In addition, we show that how the optimal algorithm for the completion time minimization problem can be derived from the proposed algorithm for throughput maximization problem. Also, for the throughput maximization problem, we propose an online algorithm and show that it is more efficient than the benchmark one (which is a direct application of an existing point-to-point online algorithm to the multi-hop system).