In recent years, shale gas development has expanded rapidly. For a shale reservoir with low porosity and ultra-low permeability, multi-cluster fracturing with a high pump rate, large fracturing fluid volume and a low fluid viscosity is the key technology to increasing shale gas production. Due to the strong heterogeneity, relatively developed weak planes such as natural fractures and bedding fractures, and the large fracturing fluid leak-off, hydraulic fractures usually propagate non-uniformly and non-continuously and form into complex fractures network, making the pressure curve of shale gas fracturing much more complex than that of the conventional fracturing. Consequently, the conventional pressure curve diagnosis approach cannot be applied to shale gas fracturing. This paper established six criteria for different propagation patterns of fractures network and proposed an adaptive diagnostic model for shale gas fracturing pressure curves. Then, the diagnostic model was applied to the typical fracturing wells, and the fracturing curves were analyzed through bottomhole net pressure conversion, dynamic segmentation, fracture propagation pattern recognition and fracture network complexity evaluation. Finally, the paper discussed the positive correlation between fracture network complexity evaluated by our diagnostic model and the fracture network complexity interpreted by microseismic monitoring. This research provides a practical method to optimize the shale gas fracturing design and operation, and enhance the potential of shale reservoir development.