CONVA AI Copilots的Rust SDK

这是用于使用CONVA AI Co-pilots的Rust Crate

用法

将以下内容添加到您的Cargo.toml中

[dependencies]
conva_ai = "0.0.3"
futures-util = "0.3"
tokio = { version = "1.37.0", features = ["full"] }

注意：以下是在部署此包时包的版本，并不代表最新支持的包号。请查阅 futures-util 和 tokio 文档以获取最新包版本。

示例

1. 使用CONVA Co-pilot生成响应的简单示例

use conva_ai::base::{AsyncConvaAI, BaseClient};
use futures_util::stream::StreamExt;

#[tokio::main]
async fn main() {

    const COPILOT_ID: &str = "your-copilot-id";
    const COPILOT_VERSION: &str = "your-copilot-version";
    const API_KEY: &str = "your-copilot-apikey";

    let mut client: BaseClient = AsyncConvaAI::init(
        &String::from(COPILOT_ID), 
        &String::from(COPILOT_VERSION), 
        &String::from(API_KEY)
    );

    let result = client.invoke_capability("how are you?".to_string(), false, "default".to_string()).await;

    match result {
        Ok(mut out) => {
            while let Some(val) = &out.next().await {
                match val {
                    Ok(val) => println!("Response {:?}", val),
                    Err(e) => println!("{:?}", e)
                }
            }
        },
        Err(e) => println!("{:?}", e)
    }

    ()
}

2. 如何清除历史记录

CONVA AI客户端默认会跟踪您的对话历史，并使用它作为智能响应的上下文

您可以通过执行以下代码来清除对话历史

use conva_ai::base::{AsyncConvaAI, BaseClient};
use futures_util::stream::StreamExt;

#[tokio::main]
async fn main() {

    const COPILOT_ID: &str = "your-copilot-id";
    const COPILOT_VERSION: &str = "your-copilot-version";
    const API_KEY: &str = "your-copilot-apikey";

    let mut client: BaseClient = AsyncConvaAI::init(
        &String::from(COPILOT_ID), 
        &String::from(COPILOT_VERSION), 
        &String::from(API_KEY)
    );

    client.clear_history();

    ()
}

如果您正在构建一个不需要跟踪对话历史的应用程序，您可以禁用历史记录跟踪

client.use_history(false)

您可以通过以下方式启用历史记录：

client.use_history(True)

3. 调试响应

Conva AI使用生成式AI为您提供查询的响应。为了帮助您理解响应背后的推理，我们还提供了AI的推理

use conva_ai::base::{AsyncConvaAI, BaseClient};
use futures_util::stream::StreamExt;

#[tokio::main]
async fn main() {

    const COPILOT_ID: &str = "your-copilot-id";
    const COPILOT_VERSION: &str = "your-copilot-version";
    const API_KEY: &str = "your-copilot-apikey";

    let mut client: BaseClient = AsyncConvaAI::init(
        &String::from(COPILOT_ID), 
        &String::from(COPILOT_VERSION), 
        &String::from(API_KEY)
    );

    let result = client.invoke_capability("how are you?".to_string(), false, "default".to_string()).await;

    match result {
        Ok(mut out) => {
            while let Some(val) = &out.next().await {
                match val {
                    Ok(val) => {
                        println!("{:?}", val.reason)
                    },
                    Err(e) => println!("{:?}", e)
                }
            }
        },
        Err(e) => println!("{:?}", e)
    }

    ()
}

4. 如何使用能力组

能力组用于控制Co-pilot将访问的能力列表。您可以在使用 invoke_capability 方法时使用能力组

    let result = client.invoke_capability("how are you?".to_string(), false, "<CAPABILITY_GROUP>".to_string()).await;

    match result {
        Ok(mut out) => {
            while let Some(val) = &out.next().await {
                match val {
                    Ok(val) => {
                        println!("{:?}", val.reason)
                    },
                    Err(e) => println!("{:?}", e)
                }
            }
        },
        Err(e) => println!("{:?}", e)
    }

    ()