Add exception handler to make flat error responses.

gliner_inference_server/main.py

@@ -1,4 +1,6 @@
 from fastapi import FastAPI, HTTPException, status
+from fastapi.responses import JSONResponse
+from starlette.requests import Request
 from typing import List, Optional, Union, Dict, Any, Tuple
 from contextlib import asynccontextmanager
 import os
@@ -31,6 +33,41 @@ app = FastAPI(
     lifespan=lifespan
 )
 
+@app.exception_handler(HTTPException)
+async def http_exception_handler(request: Request, exc: HTTPException):
+    """
+    Handles FastAPI's HTTPException to enforce a consistent, flat error
+    response structure (removing the default 'detail' wrapping).
+
+    The handler expects 'exc.detail' to be a dictionary containing the
+    desired error structure with the key 'error' (e.g., {"error": {...}}).
+
+    :param request: The incoming request object.
+    :param exc: The HTTPException instance raised (e.g., in validate_single_or_batch).
+    :return: A JSONResponse object with the custom error body and the correct status code.
+    """
+
+    # Determine the content to be returned in the response body
+    if isinstance(exc.detail, dict) and "error" in exc.detail:
+        # If the detail matches the expected custom format (e.g., {"error": {...}}),
+        # use the entire dictionary as the response content.
+        response_body = exc.detail
+    else:
+        # Fallback for unexpected or standard FastAPI error formats (e.g., detail is a string).
+        # Ensures the response always adheres to the expected structure.
+        response_body = {
+            "error": {
+                "code": "GENERIC_ERROR",
+                "message": str(exc.detail),
+                "details": {}
+            }
+        }
+
+    return JSONResponse(
+        status_code=exc.status_code,
+        content=response_body
+    )
+
 # ==================== Endpoints ====================
 @app.get("/health")
 async def health_check():