dsProject/dsLightRag/Tools/T4_Doc.py

import ijson
from Tools.KG_Config import TOPIC

# 文件路径
file_path = rf"{TOPIC}\kv_store_doc_status.json"

# 使用ijson流式读取JSON文件
with open(file_path, 'r', encoding='utf-8') as f:
    # 流式迭代文档ID和状态信息
    for doc_id, status_info in ijson.kvitems(f, ''):
        print(f"文档ID: {doc_id}")
        print(f"状态: {status_info['status']}")
        print(f"分块数量: {status_info['chunks_count']}")
        print(f"内容摘要: {status_info['content_summary'][:100]}...")  # 打印前100字符
        print(f"内容长度: {status_info['content_length']}字符")
        #print(f"创建时间: {status_info['created_at']}")
        #print(f"更新时间: {status_info['updated_at']}")
        #print(f"文件路径: {status_info['file_path']}")
        print("---")