Many of the real-world graphs are edge-labeled or node-labeled. A foundational operation on these labeled graphs is how to answer reachability queries fast. For the label-constraint reachability computation problem, a computation method called RP-Hop based on recursive partition was proposed. RP-Hop first utilized the hierarchical structure and independent set property to partition the origin large graph recursively, while keeping reachability and labels on paths between node pairs simultaneously. Combined with greedy and recursive labeling strategies, RP-Hop produced a compressed index for label-constraint reachability queries. Experiments on synthetic and real-world graph data sets demonstrate that RP-Hop can reduce index size and construction time, and guarantee the query efficiency.