feat: smart task cancellation on client disconnect

app.py

@@ -15,7 +15,7 @@ from pathlib import Path
 from typing import Optional, Dict, Any, List
 from collections import deque
 
-from fastapi import FastAPI, File, UploadFile, HTTPException, Form
+from fastapi import FastAPI, UploadFile, File, Form, HTTPException, Request
 from fastapi.responses import Response
 from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel
@@ -299,6 +299,7 @@ HALLUCINATION_BLACKLIST = {
 
 @app.post("/api/transcribe", response_model=TranscriptionResponse)
 async def transcribe_audio(
+    request: Request,
     file: UploadFile = File(...),
     language: Optional[str] = Form(None),
     initial_prompt: Optional[str] = Form(None),
@@ -313,6 +314,11 @@ async def transcribe_audio(
             temp_file.write(content)
             temp_path = temp_file.name
 
+        # --- CANCELLATION CHECK 1: Before loading model ---
+        if await request.is_disconnected():
+            logger.info(f"Client disconnected before transcription (Seq: {seq_id})")
+            return TranscriptionResponse(text="", language=language or "en", seq_id=seq_id)
+
         model = load_whisper_model()
         logger.info(f"Transcribing {file.filename} | Lang: {language} | Seq: {seq_id}")
         
@@ -348,6 +354,11 @@ async def transcribe_audio(
                 repetition_penalty=1.6 # Aggressive repetition control
             )
 
+        # --- CANCELLATION CHECK 2: Before running Whisper ---
+        if await request.is_disconnected():
+            logger.info(f"Client disconnected before Whisper (Seq: {seq_id})")
+            return TranscriptionResponse(text="", language=language or "en", seq_id=seq_id)
+
         segments, info = run_transcription(temp_path)
         segments_list = list(segments)