22 lines
708 B
Python
22 lines
708 B
Python
|
import ijson
|
||
|
|
import os
|
||
|
|
|
||
|
|
from Tools.KG_Config import TOPIC
|
||
|
|
|
||
|
|
# 文件路径
|
||
|
|
file_path = rf"d:\dsWork\dsProject\dsLightRag\Topic\{TOPIC}\vdb_entities.json"
|
||
|
|
|
||
|
|
# 检查文件是否存在
|
||
|
|
if not os.path.exists(file_path):
|
||
|
|
raise FileNotFoundError(f"文件不存在: {file_path}")
|
||
|
|
|
||
|
|
# 使用ijson流式读取JSON文件
|
||
|
|
with open(file_path, 'r', encoding='utf-8') as f:
|
||
|
|
# 流式迭代所有实体名称
|
||
|
|
print("正在读取实体...")
|
||
|
|
entity_names = ijson.items(f, 'data.item.entity_name')
|
||
|
|
entity_list = list(entity_names)
|
||
|
|
print(f"共找到 {len(entity_list)} 个实体：")
|
||
|
|
print("文件中的实体列表：")
|
||
|
|
for entity in entity_list:
|
||
|
|
print(f"- {entity}")
|