Milvus 使用 java方式

---

Milvus 基本概念

---

一、Maven 依赖【注意版本 防止依赖冲突】

 <dependency>
            <groupId>io.milvus</groupId>
            <artifactId>milvus-sdk-java</artifactId>
            <version>2.2.9</version>
            <exclusions>
                <exclusion>
                    <groupId>org.slf4j</groupId>
                    <artifactId>slf4j-api</artifactId>
                </exclusion>
                <exclusion>
                    <groupId>org.apache.logging.log4j</groupId>
                    <artifactId>log4j-slf4j-impl</artifactId>
                </exclusion>
            </exclusions>
        </dependency>

二、MilvusServiceClient 引入

需要跟milvus交互都需要调用MilvusServiceClient，我这里的做法是把它定义成一个Bean，需要用到的地方依赖注入

1.引入配置类

代码如下（示例）：
首先yml

milvus:
  config:
    ipAddr: *****
    port: ****

import io.milvus.client.MilvusServiceClient;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.context.annotation.Scope;

@Configuration
public class MilvusConfiguration {

    /**
     *  milvus ip addr
     */
    @Value("${milvus.config.ipAddr}")
    private String ipAddr;

    /**
     * milvus   port
     */
    @Value("${milvus.config.port}")
    private Integer port;

    @Bean
    @Scope("singleton")
    public MilvusServiceClient getMilvusClient() {
        return getMilvusFactory().getMilvusClient();
    }

    @Bean(initMethod = "init", destroyMethod = "close")
    public MilvusRestClientFactory getMilvusFactory() {
        return  MilvusRestClientFactory.build(ipAddr, port);
    }
}

或 注意 @Value(“${milvus.port}”) 原理一致

@Configuration
public class MilvusConfig {

    @Value("${milvus.host}")
    private String host; //milvus所在服务器地址
    @Value("${milvus.port}")
    private Integer port; //milvus端口

    @Bean
    public MilvusServiceClient milvusServiceClient() {
        ConnectParam connectParam = ConnectParam.newBuilder()
                .withHost(host)
                .withPort(port)
                .build();
        return new MilvusServiceClient(connectParam);
    }
}

2.常用方法

1.判断集合是否已经存在
代码如下（示例）：

R<Boolean> response = milvusServiceClient.hasCollection(
        HasCollectionParam.newBuilder()
            .withCollectionName("表名")
            .build());

---

2.创建集合 + 创建索引 + 把集合加载到内存中
代码如下（示例）：

 FieldType id = FieldType.newBuilder()
                    .withName("id")
                    .withDataType(DataType.Int64)
                    .withPrimaryKey(true)
                    .withAutoID(false)
                    .withDescription("id")
                    .build();
            FieldType user_id  = FieldType.newBuilder()
                    .withName("user_id")
                    .withDataType(DataType.VarChar)
                    .withMaxLength(1000)
                    .withDescription("user_id")
                    .build();
            FieldType question  = FieldType.newBuilder()
                    .withName("question")
                    .withDataType(DataType.VarChar)
                    .withMaxLength(10000)
                    .withDescription("question")
                    .build();
            FieldType answer  = FieldType.newBuilder()
                    .withName("answer")
                    .withDataType(DataType.VarChar)
                    .withMaxLength(10000)
                    .withDescription("answer")
                    .build();
            FieldType question_vector = FieldType.newBuilder()
                    .withName("question_vector")
                    .withDescription("question 向量")
                    .withDataType(DataType.FloatVector)
                    .withDimension(1536)
                    .build();
            CreateCollectionParam createCollectionReq = CreateCollectionParam.newBuilder()
                    .withCollectionName(FaceArchive.COLLECTION_NAME)
                    .addFieldType(id)
                    .addFieldType(user_id)
                    .addFieldType(question)
                    .addFieldType(answer)
                    .addFieldType(question_vector)
                    .build();
            milvusServiceClient.createCollection(createCollectionReq);    
            milvusServiceClient.createIndex(CreateIndexParam.newBuilder()
                .withFieldName("question_vector")
                .withCollectionName(FaceArchive.COLLECTION_NAME)
                .withIndexType(IndexType.HNSW)
                .withMetricType(MetricType.L2)
                //nlist 建议值为 4 × sqrt(n)，其中 n 指 segment 最多包含的 entity 条数。
                .withExtraParam("{\"M\":8,\"efConstruction\":64}")
                .build());
/**
注意 load（） 把数据加的内存中 影响 搜索
*/
             R<RpcStatus> response1 = milvusServiceClient.loadCollection(LoadCollectionParam.newBuilder()
                //集合名称
                .withCollectionName("qalog")
                .build());                    

---

3.数据插入
代码如下（示例）：

LambdaQueryWrapper<QaLog> lambdaQueryWrapper = new LambdaQueryWrapper<>();
        lambdaQueryWrapper.eq(QaLog::getStatus,"未审核");
        //todo 正则 优化
        List<QaLog> qaLogList =  qaLogService.list(lambdaQueryWrapper);
        int total = qaLogList.size();
        for (int i = 0; i < total/100; i ++) {
            List<InsertParam.Field> fields = new ArrayList<>();
            fields.add(new InsertParam.Field("id", Collections.singletonList(Long.valueOf(qaLogList.get(i).getLogId()))));
            fields.add(new InsertParam.Field("user_id", Collections.singletonList(qaLogList.get(i).getQuestion())));
            fields.add(new InsertParam.Field("question", Collections.singletonList(qaLogList.get(i).getQuestion())));
            fields.add(new InsertParam.Field("answer", Collections.singletonList(qaLogList.get(i).getAnswer())));
            fields.add(new InsertParam.Field("question_vector", embeddingsUtils.getEmbeddings(qaLogList.get(i).getQuestion())));
            InsertParam insertParam = InsertParam.newBuilder()
                    .withCollectionName("qalog")
                    .withFields(fields)
                    .build();
            R<MutationResult> insert = milvusServiceClient.insert(insertParam);