Ride-sourcing platforms like Uber and Didi Chuxing have revolutionized urban mobility by providing reserved rides, whose fare must be quoted and committed before departure, and real-time rides, whose fare can respond to evolving market conditions. However, current pricing strategies often fail to adequately differentiate between these service types, neglecting the specific operational and economic impacts of reservations. This oversight can result in inefficiencies and perceived pricing unfairness, thereby eroding customer trust. To tackle these challenges, we propose an offline–online combined optimization framework that jointly determines pre-booked prices, real-time prices, and vehicle relocation decisions under demand uncertainty and a shared capacity constraint. In the offline stage, aggregate demand forecasts are used to determine pre-booked prices. During online operations, real-time prices and relocation decisions are updated through a rolling-horizon procedure, supported by two correction mechanisms that exploit recent forecast errors and the demand composition revealed by advance bookings. Under identical willingness-to-pay distributions and shared vehicle capacity, the optimal pre-booked and real-time prices coincide in the offline problem, allowing advance pricing to rely on aggregate rather than type-specific demand forecasts. We further derive an optimality-gap bound that separates the effects of forecast error and finite lookahead. Numerical experiments using six months of Didi Chuxing data from Haikou show that the proposed framework raises platform profit by 3.46% over a benchmark without online adjustment and closes 48.9% of the gap between that benchmark and the offline optimum. The results further show that advance reservation information is particularly valuable when initial forecasts are inaccurate, while excessive pre-booking reduces the flexibility of subsequent online decisions. This information–flexibility trade-off provides crucial insights into the design and operation of reservation services in ride-sourcing markets.