| import json | |
| input_files = [ | |
| "/home/shuhan/blobdata_sd/AURAD_dataset/mimic-pacemaker/train_prompt_layout2image.json", | |
| "/home/shuhan/blobdata_sd/AURAD_dataset/chexlocalize/train_prompt_layout2image_pacemaker.json", | |
| "/home/shuhan/blobdata_sd/AURAD_dataset/mimic-health/train_prompt_layout2image_multi.json" | |
| ] | |
| output_file = "train_prompt_layout2image_pacemaker.json" | |
| count = 0 | |
| with open(output_file, "w", encoding="utf-8") as fout: | |
| for file in input_files: | |
| with open(file, "r", encoding="utf-8") as fin: | |
| for line in fin: | |
| line = line.strip() | |
| if not line: | |
| continue | |
| # 可选：检查是否是合法 json | |
| obj = json.loads(line) | |
| fout.write(json.dumps(obj, ensure_ascii=False) + "\n") | |
| count += 1 | |
| print(f"Done. Merged {count} lines into {output_file}") |