To address the gate assignment problem in large airports, which is characterized by massive scale and frequent dynamic disruptions, an end-to-end gate assignment algorithm considering dynamic gate constraints is proposed. First, to construct a scheduling state space that contains global topological information, a gate-flight heterogeneous graph model is constructed by comprehensively considering the problem constraints and optimization objectives. This establishes topological associations between gate resources and flight tasks, and dynamically represents—in a graph format—the temporal relationships among flights, as well as the constraint matching and preference relationships between gates and flights. Second, following the chronological sequence of flights, the assignment process is formulated as a Markov Decision Process (MDP), wherein Graph Neural Networks (GNN) and Convolutional Neural Networks (CNN) are utilized to extract the features of the graph structure and gate occupation status. This enables the model to fully perceive the necessary information for decision-making, thereby enhancing its decision-making performance. Finally, a feature-based dynamic action space decoder is designed and combined with the Proximal Policy Optimization (PPO) algorithm for end-to-end training, endowing the overall decision-making framework with a problem-scale-invariant property. Experimental results based on real-world data from Guangzhou Baiyun International Airport indicate that, compared with similar algorithms, the proposed algorithm achieves faster convergence, superior optimization results, and stronger generalization capabilities. Furthermore, compared with the traditional Genetic Algorithm (GA), it drastically reduces the computation time to the scale of seconds while maintaining comparable optimization results, which verifies the effectiveness of the algorithm design. This demonstrates that the proposed algorithm can provide decision support for rapid gate reallocation in large airports under disruption scenarios such as flight delays.