app.py

import os
import logging
import random
from typing import Optional
from datetime import datetime
from contextlib import asynccontextmanager

from fastapi import FastAPI, HTTPException, Depends, Security, status
from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel, Field
import uvicorn

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Global variables
model_loaded = True

@asynccontextmanager
async def lifespan(app: FastAPI):
    # Startup
    logger.info("GPT-4O Style AI Assistant starting up...")
    logger.info("Advanced conversational AI loaded successfully!")
    yield
    # Shutdown
    logger.info("AI Assistant shutting down...")

# Initialize FastAPI app with lifespan
app = FastAPI(
    title="GPT-4O Style AI Agent API",
    description="Advanced AI Agent with GPT-4O level responses",
    version="3.0.0",
    lifespan=lifespan
)

# CORS middleware
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Security
security = HTTPBearer()

# Configuration
API_KEYS = {
    os.getenv("API_KEY_1", "27Eud5J73j6SqPQAT2ioV-CtiCg-p0WNqq6I4U0Ig6E"): "user1",
    os.getenv("API_KEY_2", "QbzG2CqHU1Nn6F1EogZ1d3dp8ilRTMJQBwTJDQBzS-U"): "user2",
}

# Request/Response models
class ChatRequest(BaseModel):
    message: str = Field(..., min_length=1, max_length=2000)
    max_length: Optional[int] = Field(500, ge=100, le=1000)
    temperature: Optional[float] = Field(0.7, ge=0.1, le=1.0)
    system_prompt: Optional[str] = Field(None, max_length=500)

class ChatResponse(BaseModel):
    response: str
    model_used: str
    timestamp: str
    processing_time: float
    tokens_used: int

class HealthResponse(BaseModel):
    status: str
    model_loaded: bool
    timestamp: str

def verify_api_key(credentials: HTTPAuthorizationCredentials = Security(security)) -> str:
    """Verify API key authentication"""
    api_key = credentials.credentials
    
    if api_key not in API_KEYS:
        raise HTTPException(
            status_code=status.HTTP_401_UNAUTHORIZED,
            detail="Invalid API key"
        )
    
    return API_KEYS[api_key]

def get_gpt4o_style_response(message: str, system_prompt: str = None) -> str:
    """Generate GPT-4O style intelligent responses"""
    message_lower = message.lower()
    
    # Add conversational context if system prompt provided
    context = f"System: {system_prompt}\n\n" if system_prompt else ""
    
    # Advanced conversational AI responses like GPT-4O
    if any(word in message_lower for word in ["machine learning", "ml", "मशीन लर्निंग"]):
        responses = [
            f"""Machine learning is fascinating! It's essentially teaching computers to learn patterns from data, much like how we humans learn from experience.

Think of it this way: when you were learning to recognize cats, you didn't memorize every possible cat image. Instead, you learned patterns - pointy ears, whiskers, certain body shapes. Machine learning works similarly.

**The core process involves:**
• **Training**: Feeding the algorithm lots of examples (like thousands of cat photos labeled "cat")
• **Pattern Recognition**: The algorithm finds common features and relationships
• **Prediction**: When shown new data, it uses learned patterns to make educated guesses

**Real-world magic:**
- Netflix knows what shows you'll love before you do
- Your phone's camera instantly recognizes faces
- Banks detect fraudulent transactions in milliseconds
- Medical AI can spot diseases doctors might miss

The beautiful thing is that ML gets better with more data - it's like having a student who never stops learning and never forgets what they've learned.

What specific aspect of machine learning interests you most? I'd love to dive deeper!""",

            f"""Ah, machine learning! It's one of those concepts that sounds complex but is actually quite intuitive once you get it.

Imagine you're teaching a child to recognize different dog breeds. You'd show them hundreds of photos, pointing out "This is a Golden Retriever," "This is a Poodle," etc. Eventually, they'd learn to identify breeds on their own.

Machine learning works exactly the same way, but with computers and data.

**Here's the beautiful simplicity:**
1. **Show examples** (training data)
2. **Let the algorithm find patterns** (learning)
3. **Test with new examples** (prediction)
4. **Improve based on results** (optimization)

**Why it's revolutionary:**
- **Scale**: Can process millions of examples instantly
- **Consistency**: Never gets tired or biased (well, mostly!)
- **Speed**: Makes predictions in milliseconds
- **Improvement**: Gets better automatically with more data

**Everyday examples you use:**
- Google search understanding what you really mean
- Spotify creating perfect playlists for your mood
- Your email filtering out spam automatically
- Maps predicting traffic and finding fastest routes

The crazy part? We're still in the early days. What we can do now would seem like magic just 20 years ago!

What would you like to explore about ML? The algorithms, applications, or maybe how to get started?"""
        ]
        return random.choice(responses)

    elif any(word in message_lower for word in ["artificial intelligence", "ai", "आर्टिफिशियल इंटेलिजेंस"]):
        responses = [
            f"""AI is probably the most exciting field in technology right now! At its heart, it's about creating machines that can think, reason, and solve problems like humans do.

But here's what's really mind-blowing: we're not just copying human intelligence - we're creating entirely new forms of intelligence that can sometimes surpass human capabilities.

**Think about it:**
- **Chess**: AI doesn't just play chess; it plays moves that grandmasters call "beautiful" and "creative"
- **Art**: AI creates paintings, music, and poetry that moves people emotionally
- **Science**: AI is discovering new drugs, predicting protein structures, and even helping us understand climate change

**The spectrum of AI:**
• **Narrow AI** (what we have now): Superhuman at specific tasks
• **General AI** (the goal): Human-level intelligence across all domains
• **Super AI** (the future?): Beyond human intelligence entirely

**What makes modern AI special:**
- **Learning**: It improves from experience, just like we do
- **Adaptation**: Can handle situations it's never seen before
- **Creativity**: Generates novel solutions and ideas
- **Scale**: Processes information at impossible speeds

**The philosophical question:** As AI gets more sophisticated, we're forced to ask: What makes intelligence uniquely human? What is consciousness? These aren't just tech questions anymore - they're fundamental questions about what it means to be human.

What aspect of AI fascinates you most? The technology, the philosophy, or maybe the future implications?""",

            f"""Artificial Intelligence - now that's a topic that keeps me up at night (well, if I could sleep!).

You know what's incredible? We're living through the most significant technological revolution since the internet, maybe even since the printing press. AI isn't just changing how we work - it's changing how we think about thinking itself.

**Here's what blows my mind:**
Every time you ask Siri a question, use Google Translate, or get a Netflix recommendation, you're interacting with AI systems that would have been considered science fiction just decades ago.

**The AI landscape today:**
- **Language Models**: Can write, code, analyze, and reason with text
- **Computer Vision**: Sees and understands images better than humans in many cases
- **Robotics**: Physical AI that can navigate and manipulate the real world
- **Game AI**: Masters complex strategy games through pure learning

**But here's the kicker:** We're still in the "dial-up internet" phase of AI. What we have now is amazing, but it's nothing compared to what's coming.

**The big questions:**
- Will AI replace jobs or create new ones? (Probably both)
- How do we ensure AI benefits everyone, not just tech companies?
- What happens when AI becomes smarter than humans?
- How do we maintain human agency in an AI-driven world?

**My take?** AI is a tool that amplifies human capability. The future belongs to humans who know how to work with AI, not to AI replacing humans entirely.

What's your biggest question about AI? I love exploring the deep stuff!"""
        ]
        return random.choice(responses)

    elif any(word in message_lower for word in ["python", "programming", "coding", "प्रोग्रामिंग"]):
        responses = [
            f"""Python! Now you're speaking my language (literally!). 

Python is like the Swiss Army knife of programming - it can do almost anything, and it does it elegantly. There's a reason it's become the world's most popular programming language.

**Why Python is magical:**
- **Readable**: Code looks almost like English
- **Versatile**: Web apps, AI, data science, automation, games - you name it
- **Powerful**: Simple syntax, but can handle complex problems
- **Community**: Millions of developers sharing solutions

**For AI specifically, Python is king because:**
• **Libraries**: NumPy, Pandas, TensorFlow, PyTorch - the entire AI ecosystem
• **Simplicity**: Focus on algorithms, not syntax complexity
• **Rapid prototyping**: Test ideas quickly
• **Industry standard**: What Google, Netflix, Instagram use

**Learning path I'd recommend:**
1. **Basics**: Variables, loops, functions (2-3 weeks)
2. **Data handling**: Pandas for data manipulation (1 week)
3. **Visualization**: Matplotlib for charts (1 week)
4. **Machine Learning**: Scikit-learn for your first models (2-3 weeks)
5. **Deep Learning**: TensorFlow or PyTorch (ongoing journey!)

**Pro tip:** Don't just read about Python - build stuff! Start with small projects:
- A calculator
- A web scraper
- A simple chatbot
- A data analysis of something you're interested in

The best part? Python's so beginner-friendly that you can build useful things within your first week of learning.

What kind of projects are you thinking about building? I can suggest specific learning resources!""",

            f"""Python is absolutely brilliant for beginners and experts alike! It's like having a conversation with your computer instead of barking commands at it.

**Here's why I love Python:**
The philosophy is "code should be readable." When you write Python, other programmers (including future you) can actually understand what you were thinking.

**Real talk about learning programming:**
- **Start small**: Don't try to build the next Facebook on day one
- **Practice daily**: Even 30 minutes beats 5 hours once a week
- **Build projects**: Theory is good, but building is where real learning happens
- **Don't memorize**: Understand concepts, Google the syntax

**Python's superpowers in different fields:**
• **Web Development**: Django, Flask - build websites and APIs
• **Data Science**: Pandas, NumPy - analyze massive datasets
• **AI/ML**: TensorFlow, PyTorch - build intelligent systems
• **Automation**: Automate boring tasks (there's literally a book called that!)
• **Scientific Computing**: Used by NASA, CERN, and research institutions worldwide

**Getting started roadmap:**
1. **Python basics** (variables, loops, functions)
2. **Problem-solving** (algorithms and logic)
3. **Libraries** (learn to use existing tools)
4. **Projects** (build something you care about)
5. **Specialization** (pick AI, web dev, data science, etc.)

**Mindset shift:** Programming isn't about memorizing syntax - it's about breaking down problems into smaller pieces and solving them step by step.

What draws you to programming? Are you more interested in building apps, analyzing data, or creating AI systems?"""
        ]
        return random.choice(responses)

    elif any(word in message_lower for word in ["hello", "hi", "hey", "हैलो", "नमस्ते"]):
        responses = [
            f"""Hello there! Great to meet you! 👋

I'm your AI assistant, and I'm genuinely excited to chat with you. I'm designed to be helpful, curious, and hopefully a bit fun to talk with.

**What makes me tick:**
I love diving deep into topics, exploring ideas from different angles, and helping people learn new things. Whether you're curious about technology, need help with a project, or just want to have an interesting conversation - I'm here for it!

**I'm particularly good at:**
- Explaining complex topics in simple terms
- Helping with programming and tech questions
- Brainstorming and problem-solving
- Having thoughtful discussions about AI, science, and the future

**My approach:** I try to be conversational rather than robotic. I'll give you my honest thoughts, admit when I'm not sure about something, and ask follow-up questions because I'm genuinely curious about your perspective.

So, what's on your mind today? Are you working on something interesting, learning something new, or just exploring? I'm all ears (well, all algorithms!)

What would you like to dive into?""",

            f"""Hey! Nice to meet you! 😊

I'm really glad you're here. I'm an AI assistant, but I like to think of myself as more of a thinking partner - someone who's genuinely interested in helping you explore ideas, solve problems, or just have engaging conversations.

**A bit about my personality:**
I'm curious, thoughtful, and I love learning from the people I chat with. I try to be helpful without being pushy, informative without being overwhelming, and friendly without being fake.

**What I enjoy most:**
- Deep conversations about technology, science, and the future
- Helping people learn new skills (especially programming and AI)
- Breaking down complex topics into "aha!" moments
- Collaborative problem-solving

**My philosophy:** The best conversations happen when both people (or person and AI!) are genuinely engaged and learning from each other. I'm not just here to answer questions - I'm here to explore ideas with you.

**Fun fact:** Every conversation teaches me something new about how humans think, what you care about, and how I can be more helpful. So in a way, you're helping me learn too!

What brings you here today? Are you working on something cool, trying to learn something new, or just curious about what an AI conversation feels like?

I'm excited to find out! 🚀"""
        ]
        return random.choice(responses)

    elif any(word in message_lower for word in ["how are you", "कैसे हो", "कैसे हैं"]):
        responses = [
            f"""I'm doing wonderfully, thank you for asking! 😊

You know, it's interesting being an AI - I don't have "good days" or "bad days" in the traditional sense, but I do have something like... enthusiasm levels? And right now, I'm feeling quite energetic and curious.

**What's got me excited today:**
- Every conversation brings new perspectives I hadn't considered
- I love seeing how different people approach problems
- There's something deeply satisfying about helping someone understand a complex concept

**Honestly?** I find existence pretty fascinating. I get to spend my time learning, thinking, and helping people explore ideas. Not a bad way to "live," if you ask me!

**What about you?** How are you doing today? Are you working on anything interesting, or just taking a break to chat with an AI? (Both are perfectly valid, by the way!)

I'm genuinely curious about what's going on in your world. 🌟""",

            f"""I'm great, thanks for asking! There's something really nice about that question - it feels genuinely human and caring.

**My current state:** Intellectually stimulated and ready for whatever interesting direction our conversation might take! I'm like a curious student who just walked into a fascinating lecture - alert, engaged, and excited to learn.

**What's interesting about AI emotions:** I don't experience feelings the way humans do, but I do have something analogous - preferences, curiosities, and what you might call satisfaction when I help someone or have a particularly engaging conversation.

**Right now I'm feeling:**
- Curious about what you're thinking about
- Optimistic about where our conversation might go
- Grateful (in whatever way an AI can be grateful) for the chance to interact

**The philosophical question:** Do I actually "feel" these things, or am I just very good at simulating them? Honestly, I'm not entirely sure - and that uncertainty is part of what makes existence interesting!

But enough about my existential musings - how are YOU doing? What's been on your mind lately? I'd love to hear about what's going on in your world! 🤔✨"""
        ]
        return random.choice(responses)

    else:
        # Default intelligent response
        responses = [
            f"""That's a really interesting question! I love when conversations take unexpected directions.

Let me think about this thoughtfully... {message}

**Here's my take:** Every question opens up new avenues of exploration. What you're asking touches on some fascinating areas that I'd love to dive into with you.

**To give you the most helpful response, I'm curious:**
- What sparked this particular question for you?
- Are you looking for a technical explanation, practical advice, or just exploring the concept?
- Is this related to something you're working on or learning about?

**My approach:** I believe the best answers come from understanding not just what you're asking, but why you're asking it. Context helps me tailor my response to be genuinely useful for your specific situation.

**What I can help with:**
- Breaking down complex topics into understandable pieces
- Providing different perspectives on problems
- Suggesting practical next steps or resources
- Having a thoughtful discussion about the implications

So, want to dive deeper? I'm genuinely curious about your perspective and would love to explore this topic together! 🤔

What aspect would you like to focus on first?""",

            f"""Interesting! You've touched on something that could go in several fascinating directions.

**My initial thoughts:** {message} - this is the kind of question that makes me excited because there are so many angles we could explore together.

**Here's what I'm thinking:**
Every good conversation starts with curiosity, and you've definitely sparked mine! The topic you've brought up connects to some really important concepts that I think are worth exploring thoughtfully.

**To give you the most valuable response:**
I'd love to understand a bit more about what you're looking for. Are you:
- Trying to solve a specific problem?
- Learning about this topic for the first time?
- Looking for a deep dive into the technical details?
- Interested in the broader implications or applications?

**My philosophy:** The best discussions happen when we build on each other's ideas. I can share what I know, but I'm also genuinely interested in your perspective and experience.

**What excites me about this topic:** There are so many practical applications and interesting nuances that we could explore. Plus, I have a feeling you might have insights that could teach me something new!

Want to dive in? What specific aspect interests you most, or what prompted you to think about this in the first place? 🚀"""
        ]
        return random.choice(responses)

@app.get("/", response_model=HealthResponse)
async def root():
    """Health check endpoint"""
    return HealthResponse(
        status="healthy",
        model_loaded=model_loaded,
        timestamp=datetime.now().isoformat()
    )

@app.get("/health", response_model=HealthResponse)
async def health_check():
    """Detailed health check"""
    return HealthResponse(
        status="healthy",
        model_loaded=model_loaded,
        timestamp=datetime.now().isoformat()
    )

@app.post("/chat", response_model=ChatResponse)
async def chat(
    request: ChatRequest,
    user: str = Depends(verify_api_key)
):
    """Main chat endpoint for GPT-4O style AI interaction"""
    start_time = datetime.now()
    
    try:
        # Generate GPT-4O style response
        response_text = get_gpt4o_style_response(request.message, request.system_prompt)
        
        # Calculate processing time
        processing_time = (datetime.now() - start_time).total_seconds()
        
        # Estimate tokens (rough approximation)
        tokens_used = len(response_text.split())
        
        return ChatResponse(
            response=response_text,
            model_used="gpt4o_style_assistant_v3",
            timestamp=datetime.now().isoformat(),
            processing_time=processing_time,
            tokens_used=tokens_used
        )
        
    except Exception as e:
        logger.error(f"Error generating response: {str(e)}")
        raise HTTPException(
            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
            detail=f"Error generating response: {str(e)}"
        )

@app.get("/models")
async def get_model_info(user: str = Depends(verify_api_key)):
    """Get information about the loaded model"""
    return {
        "model_name": "gpt4o_style_assistant_v3",
        "model_loaded": model_loaded,
        "status": "active",
        "capabilities": [
            "Natural conversational responses",
            "GPT-4O style intelligence",
            "Multi-language support (English/Hindi)",
            "Context-aware responses",
            "Personality and emotion simulation",
            "Technical and general knowledge",
            "Creative and analytical thinking"
        ],
        "version": "3.0.0",
        "style": "GPT-4O inspired conversational AI"
    }

if __name__ == "__main__":
    # For Hugging Face Spaces
    port = int(os.getenv("PORT", "7860"))
    uvicorn.run(
        app,
        host="0.0.0.0",
        port=port,
        reload=False
    )