Clinical information extraction for Low-resource languages with Few-shot learning using Pre-trained language models and Prompting