ChatGPT在Mac电脑上的高效开发实践：从API调用到本地优化

1次阅读

共计 2722 个字符，预计需要花费 7 分钟才能阅读完成。

开发者在 Mac 上使用 ChatGPT API 时，常遇到以下典型问题：

长尾延迟：由于网络波动或 API 限流，某些请求响应时间远超平均值
token 计算误差：手动统计 token 容易出错，影响计费和模型输入限制
上下文丢失：在多轮对话中，因网络中断或程序崩溃导致对话历史丢失

原生 REST 调用
优点：灵活，不依赖特定库
缺点：需要手动处理 HTTP 细节，缺乏高级功能
官方 SDK
优点：官方维护，功能完整
缺点：更新可能滞后于 API 变化
自定义封装层
优点：可针对项目定制，集成优化功能
缺点：需要额外开发时间

同步模式：简单直接，适合简单应用
异步模式：性能更好，适合高并发场景

import Foundation

class ChatGPTClient {
    private let apiKey: String
    private let session: URLSession
    private let cache: NSCache<NSString, NSData>

    init(apiKey: String) {
        self.apiKey = apiKey
        self.session = URLSession(configuration: .default)
        self.cache = NSCache()}

    func sendRequest(prompt: String, 
                    temperature: Double = 0.7, 
                    maxTokens: Int = 1000) async throws -> String {
        // 检查缓存
        if let cachedResponse = cache.object(forKey: prompt as NSString) {return String(data: cachedResponse as Data, encoding: .utf8) ?? ""
        }

        // 构造请求
        var request = URLRequest(url: URL(string: "https://api.openai.com/v1/chat/completions")!)
        request.httpMethod = "POST"
        request.setValue("Bearer \(apiKey)", forHTTPHeaderField: "Authorization")
        request.setValue("application/json", forHTTPHeaderField: "Content-Type")

        let messages = [["role": "user", "content": prompt]
        ]

        let parameters: [String: Any] = [
            "model": "gpt-3.5-turbo",
            "messages": messages,
            "temperature": temperature,
            "max_tokens": maxTokens
        ]

        request.httpBody = try JSONSerialization.data(withJSONObject: parameters)

        // 发送请求（带重试机制）var retryCount = 0
        let maxRetries = 3

        while retryCount < maxRetries {
            do {let (data, _) = try await session.data(for: request)
                // 缓存响应
                cache.setObject(data as NSData, forKey: prompt as NSString)
                let response = try JSONDecoder().decode(ChatGPTResponse.self, from: data)
                return response.choices.first?.message.content ?? ""
            } catch {
                retryCount += 1
                if retryCount == maxRetries {throw error}
                try await Task.sleep(nanoseconds: 1_000_000_000) // 等待 1 秒
            }
        }

        return ""
    }
}

import openai
import json
from tenacity import retry, stop_after_attempt, wait_exponential
from functools import lru_cache

openai.api_key = "your-api-key"

@lru_cache(maxsize=100)
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=4, max=10))
def chat_with_gpt(prompt, temperature=0.7, max_tokens=1000):
    try:
        response = openai.ChatCompletion.create(
            model="gpt-3.5-turbo",
            messages=[{"role": "user", "content": prompt}],
            temperature=temperature,
            max_tokens=max_tokens
        )
        return response.choices[0].message.content
    except Exception as e:
        print(f"Error: {str(e)}")
        raise