Update main.py

main.py

@@ -6,39 +6,48 @@ from sentence_transformers import SentenceTransformer
 os.environ["OMP_NUM_THREADS"] = "2"
 os.environ["MKL_NUM_THREADS"] = "2"
 
-app = FastAPI(title="HF Inference Compatible API")
-
-# Initialize the model at startup
-MODEL_ID = "Snowflake/snowflake-arctic-embed-l"
-model = SentenceTransformer(MODEL_ID, device="cpu")
+app = FastAPI(title="OpenAI Compatible API")
+# Loads the production-grade text embedding model
+model = SentenceTransformer("Snowflake/snowflake-arctic-embed-l", device="cpu")
 
 @app.get("/health")
 def health_check():
     return {"status": "healthy"}
 
-@app.post("/")
-@app.post("/embed")
-async def generate_embeddings(request: Request):
+@app.post("/v1/embeddings")
+async def create_embeddings(request: Request):
     """
-    Mimics the official Hugging Face Inference API for n8n LangChain nodes.
+    Mimics the official OpenAI Embeddings API structure.
     """
     try:
         data = await request.json()
     except Exception:
-        raise HTTPException(status_code=400, detail="Invalid JSON body")
-    
-    # n8n's Hugging Face Embeddings node sends data under the 'inputs' key
-    inputs = data.get("inputs")
+        raise HTTPException(status_code=400, detail="Invalid JSON format")
     
+    # OpenAI sends the target text inside the "input" parameter
+    inputs = data.get("input")
     if not inputs:
-        raise HTTPException(status_code=400, detail="Missing 'inputs' field")
-
-    # Ensure inputs is a list
+        raise HTTPException(status_code=400, detail="Missing 'input' field")
+        
     if isinstance(inputs, str):
         inputs = [inputs]
-        
-    # Generate the embeddings
-    embeddings = model.encode(inputs, normalize_embeddings=True)
+
+    # Generate vectors via CPU
+    embeddings = model.encode(inputs, normalize_embeddings=True).tolist()
+    
+    # Format the array exactly how n8n's OpenAI LangChain node expects it
+    response_data = [
+        {
+            "object": "embedding",
+            "embedding": emb,
+            "index": i
+        }
+        for i, emb in enumerate(embeddings)
+    ]
     
-    # n8n requires a raw list of lists returned directly, NOT wrapped in a dict
-    return embeddings.tolist()
+    return {
+        "object": "list",
+        "data": response_data,
+        "model": "snowflake-arctic-embed-l",
+        "usage": {"prompt_tokens": 0, "total_tokens": 0}
+    }