Update main.py

main.py

@@ -2,13 +2,14 @@ import os
 from fastapi import FastAPI, Request, HTTPException
 from sentence_transformers import SentenceTransformer
 
-# Optimize CPU threads for Hugging Face free tier
 os.environ["OMP_NUM_THREADS"] = "2"
 os.environ["MKL_NUM_THREADS"] = "2"
 
 app = FastAPI(title="OpenAI Compatible API")
-# Loads the production-grade text embedding model
-model = SentenceTransformer("Snowflake/snowflake-arctic-embed-l", device="cpu")
+
+# Keep the model you specifically chose to host
+MODEL_ID = "Snowflake/snowflake-arctic-embed-l"
+model = SentenceTransformer(MODEL_ID, device="cpu")
 
 @app.get("/health")
 def health_check():
@@ -16,26 +17,24 @@ def health_check():
 
 @app.post("/v1/embeddings")
 async def create_embeddings(request: Request):
-    """
-    Mimics the official OpenAI Embeddings API structure.
-    """
     try:
         data = await request.json()
     except Exception:
         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 'input' field")
+    
+    # 🛡️ THE SAFEGUARD: If n8n sends an empty string, force a valid text 
+    # so the model generates real numbers instead of an array of zeros.
+    if not inputs or inputs == [""] or inputs == "":
+        inputs = ["dummy text to prevent zero vector database crash"]
         
     if isinstance(inputs, str):
         inputs = [inputs]
 
-    # Generate vectors via CPU
+    # Generate vectors
     embeddings = model.encode(inputs, normalize_embeddings=True).tolist()
     
-    # Format the array exactly how n8n's OpenAI LangChain node expects it
     response_data = [
         {
             "object": "embedding",
@@ -48,6 +47,6 @@ async def create_embeddings(request: Request):
     return {
         "object": "list",
         "data": response_data,
-        "model": "snowflake-arctic-embed-l",
+        "model": MODEL_ID,
         "usage": {"prompt_tokens": 0, "total_tokens": 0}
     }