In this paper, we present a pure-Python open-source library, called PyPop7, for black-box optimization (BBO). It provides a unified and modular interface for more than 60 versions and variants of different black-box optimization algorithms, particularly population-based optimizers, which can be classified into 12 popular families: Evolution Strategies (ES), Natural Evolution Strategies (NES), Estimation of Distribution Algorithms (EDA), Cross-Entropy Method (CEM), Differential Evolution (DE), Particle Swarm Optimizer (PSO), Cooperative Coevolution (CC), Simulated Annealing (SA), Genetic Algorithms (GA), Evolutionary Programming (EP), Pattern Search (PS), and Random Search (RS). It also provides many examples, interesting tutorials, and full-fledged API documentations. Through this new library, we expect to provide a well-designed platform for benchmarking of optimizers and promote their real-world applications, especially for large-scale BBO. Its source code and documentations are available at https://github.com/Evolutionary-Intelligence/pypop and https://pypop.readthedocs.io/en/latest, respectively.