本地 Milvus

from xparse_client import MilvusDestination

destination = MilvusDestination(
    db_path='./milvus_pipeline.db',      # 本地数据库文件路径
    collection_name='my_collection',      # 集合名称
    dimension=1024                        # 向量维度，必须与 embed 模型一致
)

from xparse_client import MilvusDestination, Pipeline

destination = MilvusDestination(
    db_path='./milvus_pipeline.db',
    collection_name='documents',
    dimension=1024
)

# ... 其他配置

pipeline = Pipeline(
    source=source,
    destination=destination,
    # ...其他配置
)
pipeline.run()

from pymilvus import MilvusClient

client = MilvusClient(uri='./milvus_pipeline.db')

# 将查询文本向量化（需调用相同的 embed API）
query_vector = [0.1, 0.2, 0.3, ...]  # 1024 维向量

results = client.search(
    collection_name='documents',
    data=[query_vector],
    limit=5,
    search_params={"metric_type": "COSINE", "params": {"nprobe": 10}},
    output_fields=['text', 'metadata']
)

for result in results[0]:
    print(f"相似度: {1 - result['distance']:.4f}")
    print(f"文本: {result['entity']['text'][:100]}...")