milvus的collection操作

milvus的collection操作

创建collection

import uuid
from pymilvus import (
    connections,
    FieldSchema, CollectionSchema, DataType,
    Collection,
)
collection_name = "hello_milvus"
host = "192.168.230.71"
port = 19530
username = ""
password = ""
num_entities, dim = 5000, 3
print("start connecting to Milvus")
connections.connect("default", host=host, port=port,user=username,password=password)
fields = [
    FieldSchema(name="pk", dtype=DataType.INT64, is_primary=True, auto_id=False),
    FieldSchema(name="random", dtype=DataType.DOUBLE),
    FieldSchema(name="comment", dtype=DataType.VARCHAR, max_length=200),
    FieldSchema(name="embeddings", dtype=DataType.FLOAT16_VECTOR, dim=dim)
]
schema = CollectionSchema(fields, "hello_milvus is the simplest demo to introduce the APIs")
print("Create collection `hello_world`")
coll = Collection(collection_name, schema, consistency_level="Bounded",shards_num=1)
print("done")

Collection()是一个构造函数，定义如下:

class Collection:
    def __init__(
        self,
        name: str,
        schema: Optional[CollectionSchema] = None,
        using: str = "default",
        **kwargs,
    ) -> None:

name – the name of collection

schema – the schema of collection, defaults to None.

using – Milvus connection alias name, defaults to ‘default’.

**kwargs - 告诉python接受任意数量的关键字参数到这个字典中。

那milvus这里有哪些key?

看注释。

删除collection

coll.drop()

函数定义:

def drop(self, timeout: Optional[float] = None, **kwargs):

有一个timeout参数。

加载collection

功能:加载collection到内存。

coll.load()

函数定义:

def load(
        self,
        partition_names: Optional[list] = None,
        replica_number: int = 1,
        timeout: Optional[float] = None,
        **kwargs,
    )

参数说明:

coll.load() #阻塞
coll.load(_async=True) #非阻塞
utility.wait_for_loading_complete("hello_iterator")

释放collection

从内存中卸载。

coll.release()

描述collection

获取collection的信息

infos = coll.describe()
for key, value in infos.items():
    print(key, value)