When Large Language Models Confront Repository-Level Automatic Program Repair: How Well They Done? | Synapse