server.pyโข1.95 kB
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from backend.app.routes.clothing_routes import router as clothing_router
from backend.app.models.clip_model import CLIPModel
from backend.app.models.recommender import generate_recommendations
from backend.app.config.database import init_db, close_db
from backend.app.config.settings import settings
import uvicorn
class MCPServer:
def __init__(self, app):
self.app = app
print("MCPServer initialized with app:", self.app)
def tool(self):
def decorator(func):
return func
return decorator
def run(self, transport="sse"):
print(f"Running MCPServer with transport: {transport}")
app = FastAPI()
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
app.include_router(clothing_router)
clip_model = CLIPModel()
mcp = MCPServer(app)
@app.on_event("startup")
async def startup_db():
await init_db()
@app.on_event("shutdown")
async def shutdown_db():
await close_db()
@mcp.tool()
def upload_user_image(image_url: str, user_id: str):
try:
embedding = clip_model.get_image_embedding(image_url)
detected_style = "mock_style"
return {"style": detected_style}
except Exception as e:
return {"error": f"Failed to classify image style: {str(e)}"}
@mcp.tool()
def get_style_recommendations(style: str, user_id: str):
try:
user_clicks = ["floral", "loose-fit", "cotton"] # mock data
recommendations = generate_recommendations(style, user_clicks)
return recommendations
except Exception as e:
return {"error": f"Failed to generate recommendations: {str(e)}"}
if __name__ == "__main__":
print(f"Starting MCPServer on port {settings.PORT}")
uvicorn.run("backend.app.server:app", host="0.0.0.0", port=settings.PORT, reload=True)