perf: optional orjson JSON codec behind ENABLE_ORJSON (#27583)

Swap the JSON encoder/decoder used across the backend from stdlib json to
orjson when ENABLE_ORJSON is set — HTTP request bodies, JSONResponse
bodies, upstream provider responses, SSE chunks, and socket.io/Redis
payloads.

The flag defaults to off, in which case the app uses stdlib json and
engineio's codec verbatim, so default behaviour is unchanged.

- json_codec exports JSONCodec (stdlib json or the orjson codec) and
  SOCKETIO_JSON (engineio's codec or the orjson codec); call sites import
  JSONCodec and stay implementation-agnostic
- apply_orjson_http_json() is a no-op when the flag is off, leaving
  starlette's Request.json / JSONResponse.render untouched
- the orjson codec falls back to the stdlib for inputs orjson rejects
  (non-str dict keys, ints beyond 64 bits, NaN literals)
- orjson is imported only when the flag is on
- FastAPI(default_response_class=...) is deliberately not used: an
  explicit default disables the Pydantic direct-to-bytes fast path for
  response_model routes

backend/open_webui/routers/ollama.py

@@ -42,6 +42,7 @@ from open_webui.utils.access_control import check_model_access
 from open_webui.utils.auth import get_admin_user, get_verified_user
 from open_webui.utils.headers import get_custom_headers, include_user_info_headers
 from open_webui.utils.model_ids import strip_provider_model_prefix
+from open_webui.utils.json_codec import JSONCodec
 from open_webui.utils.misc import calculate_sha256
 from open_webui.utils.payload import (
     apply_model_params_to_body_ollama,
@@ -86,7 +87,7 @@ async def send_get_request(
             ssl=AIOHTTP_CLIENT_SESSION_SSL,
             timeout=aiohttp.ClientTimeout(total=AIOHTTP_CLIENT_TIMEOUT_MODEL_LIST),
         ) as r:
-            return await r.json()
+            return await r.json(loads=JSONCodec.loads)
     except Exception as exc:
         log.error(f'Connection error: {exc}')
         return None
@@ -137,7 +138,7 @@ async def send_request(
 
         if not r.ok:
             try:
-                res = await r.json()
+                res = await r.json(loads=JSONCodec.loads)
                 await publish_model_provider_request_failed(
                     request,
                     actor=user,
@@ -179,7 +180,7 @@ async def send_request(
             )
         else:
             try:
-                return await r.json()
+                return await r.json(loads=JSONCodec.loads)
             except Exception:
                 return None
 
@@ -270,12 +271,12 @@ async def verify_connection(
         ) as r:
             if r.status != 200:
                 detail = f'HTTP Error: {r.status}'
-                res = await r.json()
+                res = await r.json(loads=JSONCodec.loads)
                 if 'error' in res:
                     detail = f'External Error: {res["error"]}'
                 raise Exception(detail)
 
-            return await r.json()
+            return await r.json(loads=JSONCodec.loads)
     except aiohttp.ClientError as exc:
         log.exception(f'Client error: {exc}')
         raise HTTPException(status_code=500, detail=ERROR_MESSAGES.SERVER_CONNECTION_ERROR)

backend/open_webui/routers/openai.py

@@ -44,6 +44,7 @@ from open_webui.utils.access_control import check_model_access, has_connection_a
 from open_webui.utils.anthropic import get_anthropic_models, is_anthropic_url
 from open_webui.utils.auth import get_admin_user, get_verified_user
 from open_webui.utils.headers import get_custom_headers, include_user_info_headers
+from open_webui.utils.json_codec import JSONCodec
 from open_webui.utils.model_ids import strip_provider_model_prefix
 from open_webui.utils.misc import (
     convert_logit_bias_input_to_json,
@@ -112,7 +113,7 @@ async def send_get_request(
                 cookies=cookies,
                 ssl=AIOHTTP_CLIENT_SESSION_SSL,
             ) as response:
-                return await response.json()
+                return await response.json(loads=JSONCodec.loads)
     except Exception as e:
         # Handle connection error here
         log.error(f'Connection error: {e}')
@@ -357,7 +358,7 @@ async def count_anthropic_tokens(request: Request, form_data: dict, user: UserMo
         )
 
         try:
-            response_data = await response.json()
+            response_data = await response.json(loads=JSONCodec.loads)
         except Exception:
             response_data = await response.text()
 
@@ -509,7 +510,7 @@ async def speech(request: Request, user=Depends(get_verified_user)):
             detail = None
             if r is not None:
                 try:
-                    res = await r.json()
+                    res = await r.json(loads=JSONCodec.loads)
                     if 'error' in res:
                         detail = f'External: {res["error"]}'
                 except Exception:
@@ -763,14 +764,14 @@ async def get_models(request: Request, url_idx: int | None = None, user=Depends(
                         if r.status != 200:
                             error_detail = f'HTTP Error: {r.status}'
                             try:
-                                res = await r.json()
+                                res = await r.json(loads=JSONCodec.loads)
                                 if 'error' in res:
                                     error_detail = f'External Error: {res["error"]}'
                             except Exception:
                                 pass
                             raise Exception(error_detail)
 
-                        response_data = await r.json()
+                        response_data = await r.json(loads=JSONCodec.loads)
 
                         if 'api.openai.com' in url:
                             response_data['data'] = [
@@ -853,7 +854,7 @@ async def verify_connection(
                     ssl=AIOHTTP_CLIENT_SESSION_SSL,
                 ) as r:
                     try:
-                        response_data = await r.json()
+                        response_data = await r.json(loads=JSONCodec.loads)
                     except Exception:
                         response_data = await r.text()
 
@@ -879,7 +880,7 @@ async def verify_connection(
                     ssl=AIOHTTP_CLIENT_SESSION_SSL,
                 ) as r:
                     try:
-                        response_data = await r.json()
+                        response_data = await r.json(loads=JSONCodec.loads)
                     except Exception:
                         response_data = await r.text()
 
@@ -1417,7 +1418,7 @@ async def generate_chat_completion(
             )
         else:
             try:
-                response = await r.json()
+                response = await r.json(loads=JSONCodec.loads)
             except Exception as e:
                 log.error(e)
                 response = await r.text()
@@ -1529,7 +1530,7 @@ async def embeddings(request: Request, form_data: dict, user):
             )
         else:
             try:
-                response_data = await r.json()
+                response_data = await r.json(loads=JSONCodec.loads)
             except Exception:
                 response_data = await r.text()
 
@@ -1656,7 +1657,7 @@ async def responses(
             )
         else:
             try:
-                response_data = await r.json()
+                response_data = await r.json(loads=JSONCodec.loads)
             except Exception:
                 response_data = await r.text()
 
@@ -1778,7 +1779,7 @@ async def proxy(path: str, request: Request, user=Depends(get_verified_user)):
             )
         else:
             try:
-                response_data = await r.json()
+                response_data = await r.json(loads=JSONCodec.loads)
             except Exception:
                 response_data = await r.text()