With the growing popularity of electric vehicles (EVs), the number of public charging stations is increasing rapidly, allowing drivers to charge their cars while parked away from home or en-route to their destination. However, as a full charge can take a significant amount of time, drivers may face queues and uncertainty over availability of charging facilities at different stations and times. In this paper, we address this problem by proposing a novel, two-sided market for advance reservations, in which agents, representing EV owners, report their preferences for time slots and charging locations, while charging stations report their availability and costs. In our model, both parties are rational, profit-maximising entities, and buyers enter the market dynamically over time. Given this, we apply techniques from online mechanism design to develop a pricing mechanism which is truthful on the buyer side (i.e., drivers have no incentive to misreport their preferences or to delay their reservations). For the seller side, we adapt three well-known pricing mechanisms and compare them both theoretically and empirically. Using realistic simulations, we demonstrate that two of our proposed mechanisms consistently achieve a high efficiency (90-95% of optimal), while offering a trade-off between stability and budget balance. Surprisingly, the third mechanism, a common payment mechanism that is truthful in simpler settings, achieves a significantly lower efficiency and runs a high deficit.