Safety Evaluation and Enhancement of DeepSeek Models in Chinese Contexts