Pass request and OAuthToken via get_current_context

backend/fastrtc/utils.py

@@ -18,8 +18,10 @@
 import numpy as np
 from fastapi import WebSocket
 from gradio.data_classes import GradioModel, GradioRootModel
+from gradio.oauth import OAuthToken
 from numpy.typing import NDArray
 from pydub import AudioSegment
+from starlette.requests import Request
 
 logger = logging.getLogger(__name__)
 
@@ -81,6 +83,8 @@ def create_message(
 class Context:
     webrtc_id: str
     websocket: WebSocket | None = None
+    request: Request | None = None
+    oauth_token: OAuthToken | None = None
 
 
 current_context: ContextVar[Context | None] = ContextVar(

backend/fastrtc/webrtc.py

@@ -14,7 +14,7 @@
     cast,
 )
 
-from gradio import wasm_utils
+from gradio import Request, wasm_utils
 from gradio.components.base import Component, server
 from gradio_client import handle_file
 
@@ -403,9 +403,36 @@ async def turn(self, _):
             return {"error": str(e)}
 
     @server
-    async def offer(self, body):
+    async def offer(
+        self,
+        body,
+        request: Request | None = None,
+    ):
+        from gradio import oauth
+
+        oauth_token = None
+
+        if request is not None:
+            try:
+                session = request.session
+                oauth_info = session.get("oauth_info", None)
+                oauth_token = (
+                    oauth.OAuthToken(
+                        token=oauth_info["access_token"],
+                        scope=oauth_info["scope"],
+                        expires_at=oauth_info["expires_at"],
+                    )
+                    if oauth_info is not None
+                    else None
+                )
+            except Exception:
+                import traceback
+
+                traceback.print_exc()
+                oauth_token = None
+
         return await self.handle_offer(
-            body, self.set_additional_outputs(body["webrtc_id"])
+            body, self.set_additional_outputs(body["webrtc_id"]), request, oauth_token
         )
 
     @server

backend/fastrtc/webrtc_connection_mixin.py

@@ -221,7 +221,7 @@ async def _trigger_response(self, webrtc_id: str, args: list[Any] | None = None)
         else:
             return {"status": "failed", "meta": {"error": "not_a_reply_on_pause"}}
 
-    async def handle_offer(self, body, set_outputs):
+    async def handle_offer(self, body, set_outputs, request=None, oauth_token=None):
         logger.debug("Starting to handle offer")
         logger.debug("Offer body %s", body)
 
@@ -372,7 +372,9 @@ async def _():
         def _(track):
             relay = MediaRelay()
             handler = self.handlers[body["webrtc_id"]]
-            context = Context(webrtc_id=body["webrtc_id"])
+            context = Context(
+                webrtc_id=body["webrtc_id"], oauth_token=oauth_token, request=request
+            )
             if self.modality == "video" and track.kind == "video":
                 args = {}
                 handler_ = handler
@@ -426,7 +428,9 @@ def _(track):
             elif self.mode == "send":
                 asyncio.create_task(cast(AudioCallback | VideoCallback, cb).start())
 
-        context = Context(webrtc_id=body["webrtc_id"])
+        context = Context(
+            webrtc_id=body["webrtc_id"], oauth_token=oauth_token, request=request
+        )
         if self.mode == "receive":
             if self.modality == "video":
                 if isinstance(self.event_handler, VideoStreamHandler):

demo/integrated_textbox/app.py

@@ -14,6 +14,7 @@
     WebRTC,
     WebRTCData,
     WebRTCError,
+    get_current_context,
     get_hf_turn_credentials,
     get_stt_model,
 )
@@ -28,13 +29,13 @@
 def response(
     data: WebRTCData,
     conversation: list[dict],
-    token: str | None = None,
     model: str = "meta-llama/Llama-3.2-3B-Instruct",
     provider: str = "sambanova",
 ):
+    context = get_current_context()
     print("conversation before", conversation)
-    if not provider.startswith("http") and not token:
-        raise WebRTCError("Please add your HF token.")
+    if not provider.startswith("http") and not context.oauth_token:
+        raise WebRTCError("Please Sign in to use this demo.")
 
     if data.audio is not None and data.audio[1].size > 0:
         user_audio_text = stt_model.stt(data.audio)
@@ -48,7 +49,7 @@ def response(
         client = OpenAI(base_url=provider, api_key="ollama")
     else:
         client = huggingface_hub.InferenceClient(
-            api_key=token,
+            api_key=context.oauth_token.access_token,  # type: ignore
             provider=provider,  # type: ignore
         )
 
@@ -103,9 +104,6 @@ def hide_token(provider: str):
         """
     )
     with gr.Sidebar():
-        token = gr.Textbox(
-            placeholder="Place your HF token here", type="password", label="HF Token"
-        )
         model = gr.Dropdown(
             choices=["meta-llama/Llama-3.2-3B-Instruct"],
             allow_custom_value=True,
@@ -114,11 +112,10 @@ def hide_token(provider: str):
         provider = gr.Dropdown(
             label="Provider",
             choices=providers,
-            value="sambanova",
+            value="auto",
             info="Select a hf-compatible provider or type the url of your server, e.g. http://127.0.0.1:11434/v1 for ollama",
             allow_custom_value=True,
         )
-    provider.change(hide_token, inputs=[provider], outputs=[token])
     cb = gr.Chatbot(type="messages", height=600)
     webrtc = WebRTC(
         modality="audio",
@@ -131,7 +128,7 @@ def hide_token(provider: str):
     )
     webrtc.stream(
         ReplyOnPause(response),  # type: ignore
-        inputs=[webrtc, cb, token, model, provider],
+        inputs=[webrtc, cb, model, provider],
         outputs=[cb],
         concurrency_limit=100,
     )

frontend/shared/InteractiveAudio.svelte

@@ -345,7 +345,8 @@
 {/if}
 <div
   class="audio-container"
-  class:full-screen={full_screen || full_screen === null}
+  class:full-screen={(full_screen || full_screen === null) &&
+    variant !== "textbox"}
 >
   <audio
     class="standard-player"