iPhone 上快速集成 ChatGPT API 的实战指南：从零到生产环境部署

1次阅读

没有评论

共计 3024 个字符，预计需要花费 8 分钟才能阅读完成。

在移动应用中集成 AI 服务正成为提升用户体验的重要方式，但开发者常面临几个核心挑战：

网络延迟敏感度 ：移动网络的不稳定性可能加剧 API 响应延迟，影响用户体验
认证复杂性 ：API 密钥管理、请求签名等安全措施容易在初期配置出错
响应解析成本 ：AI 服务返回的复杂 JSON 结构需要谨慎处理，特别是多轮对话场景
资源消耗控制 ：大语言模型响应可能占用大量内存，需优化处理流程

直接调用 API 和使用官方 SDK 各有优劣：

裸 API 优势 ：
避免 SDK 依赖和版本锁定
更精细控制请求 / 响应生命周期
适合需要深度定制的场景
SDK 优势 ：
快速实现基础功能
内置重试和错误处理机制
官方维护的接口稳定性

本文选择裸 API 方案，因其更适合展示底层原理和性能优化技巧，这些知识同样适用于其他 RESTful 服务集成。

import Foundation

struct ChatGPTService {
    private let apiKey: String
    private let session: URLSession

    init(apiKey: String, session: URLSession = .shared) {
        self.apiKey = apiKey
        self.session = session
    }

    func sendRequest(prompt: String) async throws -> String {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 requestBody: [String: Any] = [
            "model": "gpt-3.5-turbo",
            "messages": [["role": "user", "content": prompt]
            ],
            "temperature": 0.7
        ]

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

        let (data, _) = try await session.data(for: request)
        let response = try JSONDecoder().decode(ChatGPTResponse.self, from: data)
        return response.choices.first?.message.content ?? ""
    }
}

struct ChatGPTResponse: Codable {
    struct Choice: Codable {
        struct Message: Codable {let content: String}
        let message: Message
    }
    let choices: [Choice]
}

func streamResponse(prompt: String) async throws -> AsyncThrowingStream<String, Error> {var request = buildBaseRequest()
    // 添加流式参数
    var body = buildRequestBody(prompt: prompt)
    body["stream"] = true
    request.httpBody = try JSONSerialization.data(withJSONObject: body)

    let (bytes, _) = try await session.bytes(for: request)

    return AsyncThrowingStream { continuation in
        Task {
            do {
                for try await line in bytes.lines {guard line.hasPrefix("data:"), 
                          let jsonData = line.dropFirst(5).data(using: .utf8) else {continue}
                    let response = try JSONDecoder().decode(StreamResponse.self, from: jsonData)
                    if let content = response.choices.first?.delta.content {continuation.yield(content)
                    }
                }
                continuation.finish()} catch {continuation.finish(throwing: error)
            }
        }
    }
}

func sendRequestWithRetry(
    prompt: String,
    maxRetries: Int = 3,
    retryDelay: TimeInterval = 1
) async throws -> String {
    var lastError: Error?

    for _ in 0..<maxRetries {
        do {return try await sendRequest(prompt: prompt)
        } catch {
            lastError = error
            // 指数退避
            let delay = retryDelay * pow(2, Double(retryCount))
            try await Task.sleep(nanoseconds: UInt64(delay * 1_000_000_000))
        }
    }

    throw lastError ?? APIError.maxRetriesReached
}

实现令牌桶算法控制请求频率
响应 429 状态码时自动降级
关键业务请求设置优先级队列

使用 iOS Keychain 存储 API 密钥
实现请求签名防止中间人攻击
敏感日志过滤

import Network

class NetworkMonitor {static let shared = NetworkMonitor()
    private let monitor = NWPathMonitor()

    func startMonitoring() {
        monitor.pathUpdateHandler = { path in
            if path.status == .satisfied {// 网络恢复时重试队列中的请求}
        }
        monitor.start(queue: DispatchQueue.global(qos: .background))
    }
}