Face generation and editing have emerged as a critical research direction in computer vision, yet existing StyleGAN-based inversion methods often exhibit limitations in stylization quality and controllability. To address these challenges, we propose StyleToonEdit, an efficient framework for high-fidelity cartoon-style face generation and editing. Our approach introduces a dual-branch encoder architecture that synergistically integrates feature encoding and latent encoding pathways. The feature space preserves identity-critical facial attributes, mitigating distortion and detail degradation caused by excessive stylization deviations. Concurrently, the latent space facilitates precise attribute manipulation (e.g., age, hairstyle, expressions) and stylistic intensity adjustment while ensuring directional consistency in edits. We evaluate our method through extensive experiments across diverse cartoon styles. Results demonstrate that StyleToonEdit achieves superior content reconstruction fidelity and style consistency compared to existing approaches, while offering flexible control over both stylistic and semantic attributes, bridging the gap between photorealistic generation and expressive stylization.