Python 调用大模型 API 构建本地 AI 对话机器人

关键点说明：

• api_key：必须妥善保管，不要硬编码在公开仓库中。
• model：根据平台文档填写正确的模型标识符。
• stream：开启流式输出可以让用户更快看到首字响应，提升体验。

3. 构建图形化界面 (GUI)

为了更直观地交互，我们可以使用 PyQt6 封装一个桌面聊天窗口。这个示例实现了消息气泡显示、思维链独立展示以及发送状态管理。

import sys
from PyQt6.QtWidgets import QApplication, QMainWindow, QWidget, QVBoxLayout, QTextEdit, QPushButton, QHBoxLayout, QLabel, QFrame, QProgressBar
from PyQt6.QtCore import Qt, QThread, pyqtSignal, QSize, QTimer
from PyQt6.QtGui import QClipboard, QMovie
from openai import OpenAI

class ChatWorker(QThread):
    response_received = pyqtSignal(str)
    reasoning_received = pyqtSignal(str)
    error_occurred = pyqtSignal(str)

    def __init__(self, api_key, base_url, messages):
        super().__init__()
        self.api_key = api_key
        self.base_url = base_url
        self.messages = messages

    def run(self):
        try:
            client = OpenAI(
                api_key=self.api_key,
                base_url=self.base_url
            )
            chat_completion = client.chat.completions.create(
                model="deepseek-chat",
                messages=self.messages,
                stream=True
            )
            for chunk in chat_completion:
                if hasattr(chunk.choices[0].delta, 'reasoning_content'):
                    self.reasoning_received.emit(chunk.choices[0].delta.reasoning_content)
                if hasattr(chunk.choices[0].delta, 'content'):
                    if chunk.choices[0].delta.content:
                        self.response_received.emit(chunk.choices[0].delta.content)
        except Exception as e:
            self.error_occurred.emit(str(e))

class ChatWindow(QMainWindow):
    def __init__(self):
        super().__init__()
        self.is_sending = False
        self.initUI()

    def initUI(self):
        self.setWindowTitle('AI 对话助手')
        self.setGeometry(100, 100, 800, 600)
        self.setStyleSheet("QMainWindow { background-color: #2b2b2b; }")

        central_widget = QWidget()
        self.setCentralWidget(central_widget)
        main_layout = QVBoxLayout(central_widget)
        main_layout.setSpacing(15)
        main_layout.setContentsMargins(20, 20, 20, 20)

        # 对话历史区域
        self.chat_history = QTextEdit()
        self.chat_history.setReadOnly(True)
        self.chat_history.setStyleSheet("""
            QTextEdit {
                background-color: #363636;
                border: 1px solid #404040;
                border-radius: 10px;
                padding: 15px;
                color: #ffffff;
            }
        """)
        main_layout.addWidget(self.chat_history, stretch=2)

        # 思维链显示区域
        self.reasoning_display = QTextEdit()
        self.reasoning_display.setReadOnly(True)
        self.reasoning_display.setStyleSheet("""
            QTextEdit {
                background-color: #323232;
                border: 1px solid #404040;
                border-radius: 10px;
                padding: 15px;
                color: #b4b4b4;
                font-size: 13px;
            }
        """)
        main_layout.addWidget(self.reasoning_display, stretch=1)

        # 输入区域
        input_layout = QHBoxLayout()
        input_layout.setSpacing(10)
        self.input_field = QTextEdit()
        self.input_field.setMaximumHeight(100)
        self.input_field.setStyleSheet("""
            QTextEdit {
                background-color: #363636;
                border: 1px solid #404040;
                border-radius: 8px;
                padding: 8px;
                color: #ffffff;
            }
        """)
        input_layout.addWidget(self.input_field)

        # 发送按钮
        send_status_layout = QHBoxLayout()
        send_status_layout.setSpacing(5)
        self.send_button = QPushButton('发送')
        self.send_button.setStyleSheet("""
            QPushButton {
                background-color: #4a9eff;
                border: none;
                border-radius: 8px;
                padding: 8px 20px;
                color: white;
                font-weight: bold;
            }
            QPushButton:hover { background-color: #3b8be6; }
            QPushButton:pressed { background-color: #3178cc; }
            QPushButton:disabled { background-color: #666666; }
        """)
        self.send_button.clicked.connect(self.send_message)
        send_status_layout.addWidget(self.send_button)

        # 加载动画
        self.loading_label = QLabel()
        self.loading_movie = QMovie("loading.gif")
        self.loading_movie.setScaledSize(QSize(20, 20))
        self.loading_label.setMovie(self.loading_movie)
        self.loading_label.hide()
        send_status_layout.addWidget(self.loading_label)
        input_layout.addLayout(send_status_layout)
        main_layout.addLayout(input_layout)

        # 配置信息
        self.api_key = "YOUR_API_KEY"
        self.base_url = "https://your-provider-api.com/v1"

    def send_message(self):
        if self.is_sending:
            return
        user_message = self.input_field.toPlainText().strip()
        if not user_message:
            return

        self.is_sending = True
        self.send_button.setEnabled(False)
        self.loading_label.show()
        self.loading_movie.start()

        self.input_field.clear()
        self.reasoning_display.clear()

        message_html = self.create_message_bubble(user_message, is_user=True)
        cursor = self.chat_history.textCursor()
        cursor.movePosition(cursor.MoveOperation.End)
        cursor.insertHtml(message_html)

        messages = [{"role": "user", "content": user_message}]
        self.worker = ChatWorker(self.api_key, self.base_url, messages)
        self.worker.response_received.connect(self.handle_response)
        self.worker.reasoning_received.connect(self.handle_reasoning)
        self.worker.error_occurred.connect(self.handle_error)
        self.worker.start()

    def create_message_bubble(self, text, is_user=True):
        background_color = '#404040' if is_user else '#1a5fb4'
        alignment = 'right' if is_user else 'left'
        html = f'''<div style="text-align:{alignment};"><div style="background-color:{background_color};padding:10px;border-radius:8px;">{text}</div></div>'''
        return html

    def handle_response(self, response):
        if not hasattr(self, 'accumulated_response'):
            self.accumulated_response = ''
            self.response_timer = QTimer()
            self.response_timer.setSingleShot(True)
            self.response_timer.timeout.connect(self.display_accumulated_response)
        
        self.accumulated_response += response
        self.response_timer.start(1000)

    def display_accumulated_response(self):
        if not self.accumulated_response.strip():
            return
        message_html = self.create_message_bubble(self.accumulated_response, is_user=False)
        cursor = self.chat_history.textCursor()
        cursor.movePosition(cursor.MoveOperation.End)
        cursor.insertHtml(message_html)
        self.chat_history.verticalScrollBar().setValue(self.chat_history.verticalScrollBar().maximum())
        self.accumulated_response = ''
        self.is_sending = False
        self.send_button.setEnabled(True)
        self.loading_label.hide()
        self.loading_movie.stop()

    def handle_reasoning(self, reasoning):
        self.reasoning_display.insertPlainText(reasoning)

    def handle_error(self, error_message):
        error_html = f'<div style="color:red;">错误：{error_message}</div>'
        self.chat_history.append(error_html)
        self.is_sending = False
        self.send_button.setEnabled(True)
        self.loading_label.hide()
        self.loading_movie.stop()

def main():
    app = QApplication(sys.argv)
    window = ChatWindow()
    window.show()
    sys.exit(app.exec())

if __name__ == '__main__':
    main()

4. 总结

通过上述步骤，我们成功实现了一个具备流式响应和思维链展示的本地 AI 对话机器人。相较于传统方案，这种基于 MaaS 接口的开发方式更加轻量：

1. 低门槛：无需训练模型，仅需配置 API 参数即可调用强大的底层能力。
2. 跨平台：Python 生态成熟，支持 Windows、Linux、macOS 等多种操作系统。
3. 易扩展：代码结构清晰，后续可轻松添加多模型切换、历史记录保存等功能。

开发者只需关注业务逻辑，无需深入钻研复杂的底层算法细节，即可利用平台技术加速创新。