OpenSearch:Build a RAG system based on DeepSeek-R1

Java

Step 1: Configure the Java environment

java -version

<dependency>
    <groupId>com.aliyun.opensearch</groupId>
    <artifactId>aliyun-sdk-opensearch</artifactId>
    <version>6.0.0</version>
</dependency>

Step 2: Obtain the parameters

Callling API operations by using SDKs depends on the following key parameters:

• AppName: the name of the application.

• host: the endpoint of the application API.

Step 3: Call the API operations

Run the following sample code:

package com.aliyun.opensearch;

import com.aliyun.opensearch.OpenSearchClient;
import com.aliyun.opensearch.sdk.generated.OpenSearch;
import com.aliyun.opensearch.sdk.generated.commons.OpenSearchClientException;
import com.aliyun.opensearch.sdk.generated.commons.OpenSearchException;
import com.aliyun.opensearch.sdk.generated.commons.OpenSearchResult;

import java.util.HashMap;
import java.util.Map;

public class LLMSearch {

    private static String appName = "Your application name";
    private static String host = "Your API";
    private static String accesskey = "Your AccessKey ID";
    private static String secret = "Your AccessKey Secret";
    private static String path = "/apps/%s/actions/knowledge-search";

    public static void main(String[] args) {
      
        String appPath = String.format(path, appName);

        //ApiReadTimeOut
        OpenSearch openSearch = new OpenSearch(accesskey, secret, host);
        openSearch.setTimeout(90000);

        OpenSearchClient openSearchClient = new OpenSearchClient(openSearch);

        Map<String, String> params = new HashMap<String, String>() {{
            put("format", "full_json");
            put("_POST_BODY", "{\"question\":{\"text\":\"怎么充电\",\"type\":\"TEXT\",\"session\":\"\"},\"options\":{\"retrieve\":{\"doc\":{\"filter\":\"\",\"top_n\":5,\"sf\":\"\",\"dense_weight\":\"0.7\",\"formula\":\"\",\"operator\":\"AND\"},\"entry\":{\"sf\":\"\"},\"image\":{\"sf\":\"\",\"dense_weight\":\"0.7\"},\"qp\":{\"query_extend\":false,\"query_extend_num\":5},\"return_hits\":false,\"rerank\":{\"enable\":true,\"model\":\"ops-bge-reranker-larger\"}},\"chat\":{\"stream\":true,\"prompt_config\":{\"attitude\":\"normal\",\"rule\":\"detailed\",\"noanswer\":\"sorry\",\"language\":\"Chinese\",\"role\":false,\"role_name\":\"AI小助手\",\"out_format\":\"text\"},\"agent\":{\"tools\":[]},\"csi_level\":\"strict\",\"history_max\":\"\",\"link\":\"false\",\"model\":\"deepseek-r1\",\"model_generation\":\"\"}}}");

        }};
          try {
            OpenSearchResult openSearchResult = openSearchClient
            .callAndDecodeResult(appPath, params, "POST");
            System.out.println("RequestID=" + openSearchResult.getTraceInfo().getRequestId());
            System.out.println(openSearchResult.getResult());
        } catch (
            OpenSearchException e) {
            System.out.println("RequestID=" + e.getRequestId());
            System.out.println("ErrorCode=" + e.getCode());
            System.out.println("ErrorMessage=" + e.getMessage());
        } catch (
            OpenSearchClientException e) {
            System.out.println("ErrorMessage=" + e.getMessage());
        }
    }
}

Python

Step 1: Configure the Python environment

# If the execution fails, replace python with python3 and re-run the command.
python -V

pip install alibabacloud_tea_util 
pip install alibabacloud_opensearch_util
pip install alibabacloud_credentials

Step 2: Obtain the parameters

Calling API operations by using SDKs depends on the following key parameters:

• app_name: the name of the application.

• endpoint: the endpoint of the application API.

Step 3: Call the API operations

For more information about BaseRequest, see Sample code for the Python client.

# -*- coding: utf-8 -*-

import time, os
from typing import Dict, Any

from Tea.exceptions import TeaException
from Tea.request import TeaRequest
from alibabacloud_tea_util import models as util_models
from BaseRequest import Config, Client


class LLMSearch:
    def __init__(self, config: Config):
        self.Clients = Client(config=config)
        self.runtime = util_models.RuntimeOptions(
            connect_timeout=10000,
            read_timeout=90000,
            autoretry=False,
            ignore_ssl=False,
            max_idle_conns=50,
            max_attempts=3
        )
        self.header = {}


    def searchDoc(self, app_name: str,body:Dict, query_params: dict={}) -> Dict[str, Any]:
        try:
            response = self.Clients._request(method="POST", pathname=f'/v3/openapi/apps/{app_name}/actions/knowledge-search',
                                             query=query_params, headers=self.header, body=body, runtime=self.runtime)
            return response
        except TeaException as e:
            print(e)


if __name__ == "__main__":
    # Specify the endpoint of the application API. Remove the http:// prefix from the value.
    endpoint = "<endpoint>"

    # Specify the request protocol. Valid values: HTTPS and HTTP.
    endpoint_protocol = "HTTP"

    # Specify your AccessKey pair.
    # Obtain the AccessKey ID and AccessKey secret from environment variables. 
    # You must configure environment variables before you run this code. For more information, see the "Configure an AccessKey pair as environment variables" section of this topic.
    access_key_id = os.environ.get("ALIBABA_CLOUD_ACCESS_KEY_ID")
    access_key_secret = os.environ.get("ALIBABA_CLOUD_ACCESS_KEY_SECRET")

    # Specify the authentication method. Default value: access_key. The value sts indicates authentication based on RAM and Security Token Service (STS).
    # Valid values: sts and access_key.
    auth_type = "access_key"

    # If you use authentication based on RAM and STS, you must configure the security_token parameter. You can call the AssumeRole operation of Alibaba Cloud RAM to obtain an STS token.
    security_token =  "<security_token>"

    # Configure common request parameters.
    # The type and security_token parameters are required only if you use the SDK as a RAM user.
    Configs = Config(endpoint=endpoint, access_key_id=access_key_id, access_key_secret=access_key_secret,
                     security_token=security_token, type=auth_type, protocol=endpoint_protocol)

    # Create an OpenSearch LLM-Based Conversational Search Edition instance.
    # Replace <Application name> with the name of your OpenSearch LLM-Based Conversational Search Edition instance.
    ops = LLMSearch(Configs)
    app_name = "<Application name>"

    # --------------- Search for documents ---------------

    docQuery = {"question": {"text": "Search", "type": "TEXT"}, "options": {"chat": {"model": "deepseek-r1"}}}

    res1 = ops.searchDoc(app_name=app_name, body=docQuery)
    print(res1)