Polyv AI Q&A Assistant API Documentation

Table of Contents

• API Overview
• Call Flow
• API Details
  • Get Token API
  • AI Chat Q&A API
• Signature Generation Rules
• Code Examples
• FAQ

API Overview

Call Flow

1. Get Token

   • Call the Get Token API using application information (appId, secretKey)
   • Generate signature and submit the request
   • Retrieve the returned token

2. Use Token for AI Chat

   • Call the AI Chat API using the token obtained in the previous step
   • Process the streamed data
   • Concatenate the returned content to get the complete answer

API Details

Get Token API

Basic Information

• API Endpoint: https://api.polyv.net/live/v3/common/token/get-ai-token
• Request Method: POST
• Data Format: form-data

Request Parameters

Response Parameters

Response Example

{
  "code": 200,
  "message": "success",
  "status": "success",
  "data": {
    "token": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...",
    "userId": "viewer123456",
    "validTime": 3600
  }
}

AI Chat Q&A API

Basic Information

• API Endpoint: https://api.polyv.net/ai/v1/chat/question
• Request Method: GET
• Response Format: EventStream (streaming response)

Request Parameters

Response Format

The API returns data in EventStream (Server-Sent Events) format, requiring client support for this format:

event: message
data: {"content":"我可以回答您关于", "done":false}

Response Data Description

Data Field Description

Signature Generation Rules

Steps to generate the sign parameter:

1. Sort all parameters except sign in ascending order by parameter name ASCII code
2. Concatenate the sorted parameters in key=value format, joined by &
3. Prepend and append the secret key (secretKey) to the concatenated string
4. Perform MD5 encryption on the concatenated string and convert to uppercase

Signature formula:

MD5(secretKey + 排序并拼接后的参数字符串 + secretKey).toUpperCase()

Code Examples

Python Complete Call Flow

import time
import hashlib
import json
import requests
import sseclient  # 需要安装: pip install sseclient-py

POLYV_CONFIG = {
    "app_id": '保利威平台appId',
    "app_secret": '保利威平台appSecret'
}

# 第一步：获取Token
def get_ai_token(viewer_id):
 """获取聊天Token的接口"""

    # 构建请求参数
    if viewerId is None:
        print("viewerId is None")
        viewerId = f"polyvWebscriptViewerId-{uuid.uuid4()}"

    print("最后的viewerId", viewerId)
    form_data = {
        "appId": POLYV_CONFIG["app_id"],
        "timestamp": int(time.time() * 1000),
        "viewerId": viewerId
    }

    # 添加签名
    form_data["sign"] = create_api_sign(POLYV_CONFIG["app_secret"], form_data)

    # 发送请求到保利威API
    response = requests.post(
        "https://api.polyv.net/live/v3/common/token/get-ai-token",
        data=form_data
    )

    print(response.json())
    token = response.json().get('data')
    return token
# 第二步：使用token调用AI聊天接口
def chat_with_ai(token, ai_id, question):
    """向AI助手发送问题并获取回答"""
    # 构建请求URL
    url = f"https://api.polyv.net/ai/v1/chat/question?question={question}&aiId={ai_id}&token={token}"

    # 发送请求
    headers = {
        "Accept": "text/event-stream",
        "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36"
    }

    response = requests.get(url, headers=headers)
    print(response)
    client = sseclient.SSEClient(response)

    # 处理EventStream响应
    answer = ""
    try:
        for event in client.events():
            data = json.loads(event.data)
            print(data)
            content = data.get("content", "")
            done = data.get("done")
            if not done:
                print(f'过程数据：===>{content}')
                answer += content
                # 处理回答片段
            else:
                answer += content
                print("回答完成")
                break
        return answer
    except Exception as e:
        print(f"处理响应出错: {e}")
        raise

# 使用示例
if __name__ == "__main__":
    # 配置参数
    APP_ID = "你的appId"
    SECRET_KEY = "你的secretKey"
    VIEWER_ID = "用户唯一标识"
    AI_ID = "1159"
    QUESTION = "你好，请介绍一下自己"
    
    try:
        # 注意：实际应用中token获取应在后端进行
        token = get_ai_token(APP_ID, SECRET_KEY, VIEWER_ID)
        print(f"获取token成功: {token}")
        
        # 使用token进行聊天
        answer = chat_with_ai(token, AI_ID, QUESTION)
        print(f"完整回答: {answer}")
    except Exception as e:
        print(f"错误: {e}")

FAQ

1. Token Security

   • The Get Token API must be called on the server side to avoid exposing the secretKey to the frontend
   • You can encapsulate a Get Token API in your own backend service for frontend calls

2. Token Validity

   • Token validity is typically 3600 seconds (1 hour)
   • It is recommended to automatically refresh the token before it expires, e.g., set a scheduled task to refresh the token 5-10 minutes before expiration

3. EventStream Handling

   • Different frontend frameworks may handle EventStream differently
   • All content fields from message events must be correctly concatenated to obtain the complete answer
   • When an event of type done is received, it indicates the current answer has ended

4. Error Handling

   • It is recommended to add a timeout mechanism to prevent requests from hanging indefinitely due to network issues
   • Add an error retry mechanism to automatically reconnect during network fluctuations

5. Request Limits

   • API call frequency limit: Maximum 500 requests per second per appId
   • Ensure reasonable control of request frequency to avoid triggering rate limiting