Deciphering the Chaos: Enhancing Jailbreak Attacks via Adversarial Prompt Translation