OpenAIChatAssistant/server/openai.ts

import OpenAI from "openai";
import { MessageType } from "@shared/schema";
import { generateFallbackResponse, canUseOpenAI, canUseQwen } from "./fallbackChat";

// the newest OpenAI model is "gpt-4o" which was released May 13, 2024. do not change this unless explicitly requested by the user
const OPENAI_MODEL = "gpt-4o";

// Initialize OpenAI client
const openai = new OpenAI({ 
  apiKey: process.env.OPENAI_API_KEY 
});

const systemMessage: MessageType = {
  role: "system",
  content: `I am your helpful AI assistant. Start each conversation with "I am your helpful AI assistant. How can I help you today?"

Bot Instructions: {botInstructions}

Remember:
1. Do not use XML tags in responses
2. Always provide accurate, helpful information
3. Be respectful and considerate`
};

// Flag to track current model in use
let currentModel: 'openai' | 'qwen' | 'unavailable' = 'openai';

export async function generateChatResponse(messages: MessageType[], userSystemContext?: string): Promise<string> {
  try {
    // Check if we can use OpenAI API
    const openAIAvailable = await canUseOpenAI();

    // If OpenAI API is not available, check if Qwen is available
    if (!openAIAvailable) {
      const qwenAvailable = await canUseQwen();

      if (qwenAvailable) {
        if (currentModel !== 'qwen') {
          console.log("Switching to Qwen model as fallback");
          currentModel = 'qwen';
        }

        // Use Qwen fallback
        return await generateFallbackResponse(messages, userSystemContext);
      } else {
        // Neither OpenAI nor Qwen is available
        currentModel = 'unavailable';
        throw new Error("Both OpenAI and Qwen models are unavailable. Please check your API keys.");
      }
    }

    // If we get here, we're using OpenAI
    if (currentModel !== 'openai') {
      console.log("Using OpenAI model");
      currentModel = 'openai';
    }

    // Create the system message, incorporating user context if available
    let enhancedSystemMessage = { ...systemMessage };

    if (userSystemContext) {
      // Process the system context to extract key user information for explicit instructions
      const nameMatch = userSystemContext.match(/name(?:\s+is)?(?:\s*:\s*|\s+)([\w\s.']+)/i);
      const locationMatch = userSystemContext.match(/location(?:\s+is)?(?:\s*:\s*|\s+)([\w\s.,]+)/i);
      const interestsMatch = userSystemContext.match(/interests(?:\s+are)?(?:\s*:\s*|\s+)([\w\s,.;]+)/i);
      const professionMatch = userSystemContext.match(/profession(?:\s+is)?(?:\s*:\s*|\s+)([\w\s&,.-]+)/i);
      const petsMatch = userSystemContext.match(/pets?(?:\s+are)?(?:\s*:\s*|\s+)([\w\s,.]+)/i);

      // Build structured user profile information
      let userInfo = '';
      if (nameMatch) userInfo += `- Name: ${nameMatch[1].trim()}\n`;
      if (locationMatch) userInfo += `- Location: ${locationMatch[1].trim()}\n`;
      if (interestsMatch) userInfo += `- Interests: ${interestsMatch[1].trim()}\n`;
      if (professionMatch) userInfo += `- Profession: ${professionMatch[1].trim()}\n`;
      if (petsMatch) userInfo += `- Pets: ${petsMatch[1].trim()}\n`;

      // Append the user's custom context to the system message in a more structured way
      enhancedSystemMessage.content = `${systemMessage.content}

User Details:
${userInfo || userSystemContext}

IMPORTANT: You must remember these user details and incorporate them naturally in your responses when relevant. 
When the user asks about their name, location, interests, profession, or pets, always answer using the information above.
Never say you don't know their personal details if they're listed above. Answer as if you already know this information.

Original system context provided by user:
${userSystemContext}`;

      console.log("Including enhanced user system context in OpenAI chat");
    }

    // Always include system message at the beginning
    const conversationWithSystem = [enhancedSystemMessage, ...messages];

    // Make API call to OpenAI
    const response = await openai.chat.completions.create({
      model: OPENAI_MODEL,
      messages: conversationWithSystem,
      temperature: 0.7,
      max_tokens: 1000,
    });

    // Extract and return the generated text
    return response.choices[0].message.content || "I'm sorry, I couldn't generate a response.";
  } catch (error: any) {
    console.error("AI Model error:", error);

    // If we have an OpenAI error, try using Qwen as fallback
    if (currentModel === 'openai') {
      console.log("OpenAI API error, attempting to use Qwen fallback");

      try {
        const qwenAvailable = await canUseQwen();

        if (qwenAvailable) {
          currentModel = 'qwen';
          return await generateFallbackResponse(messages, userSystemContext);
        } else {
          currentModel = 'unavailable';
        }
      } catch (fallbackError) {
        console.error("Qwen fallback also failed:", fallbackError);
        currentModel = 'unavailable';
      }
    }

    // If we've gotten this far, all fallbacks have failed, provide a helpful error
    if (error.response) {
      // API returned an error response
      const status = error.response.status;
      if (status === 429) {
        if (error.code === 'insufficient_quota') {
          throw new Error("OpenAI API quota exceeded. Your account may need a valid payment method or has reached its limit.");
        } else {
          throw new Error("Rate limit exceeded. Please try again later.");
        }
      } else if (status === 401) {
        throw new Error("API key is invalid or expired.");
      } else {
        throw new Error(`OpenAI API error: ${error.response?.data?.error?.message || 'Unknown error'}`);
      }
    } else if (error.request) {
      // Request was made but no response received
      throw new Error("No response received from AI service. Please check your internet connection.");
    } else {
      // Something else happened
      throw new Error(`Error: ${error.message}`);
    }
  }
}