Update app.py

app.py

@@ -21,10 +21,13 @@ class TranslationJob(BaseModel):
     caption_style: str
     resolution: str
     user_token: str
-    # Null (boş) değerlerin 422 hatası vermemesi için Optional yapıldı
     custom_font: Optional[str] = None
     custom_size: Optional[int] = None
     custom_color: Optional[str] = None
+    supabase_url: Optional[str] = None
+    supabase_anon_key: Optional[str] = None
+    gemini_api_key: Optional[str] = None
+    elevenlabs_api_key: Optional[str] = None
 
 def load_env_local():
     """Load environment variables from local .env.local file if it exists."""
@@ -119,6 +122,7 @@ def setup_wav2lip():
     print("Wav2Lip environment setup completed successfully.")
     return True
 
+
 def get_supabase_config():
     supabase_url = os.environ.get("NEXT_PUBLIC_SUPABASE_URL", "")
     supabase_anon_key = os.environ.get("NEXT_PUBLIC_SUPABASE_ANON_KEY", "")
@@ -134,13 +138,22 @@ def get_api_keys():
     
     return gemini_key, elevenlabs_key, (is_gemini_mock or is_eleven_mock)
 
-def update_db_status(video_id: str, user_token: str, status: str, translated_url: str = None, error_message: str = None):
-    supabase_url, supabase_anon_key = get_supabase_config()
+def update_db_status(video_id: str, user_token: str, status: str, translated_url: str = None, error_message: str = None, supabase_url: str = None, supabase_anon_key: str = None):
+    if not supabase_url or not supabase_anon_key:
+        env_url, env_key = get_supabase_config()
+        supabase_url = supabase_url or env_url
+        supabase_anon_key = supabase_anon_key or env_key
+
     if not supabase_url or not supabase_anon_key:
         print("Error: Supabase config is missing.")
         return
 
-    url = f"{supabase_url}/rest/v1/videos?id=eq.{video_id}"
+    # Normalize Supabase base URL (remove trailing slash and rest/v1 path if present)
+    base_url = supabase_url.rstrip("/")
+    if base_url.endswith("/rest/v1"):
+        base_url = base_url[:-8].rstrip("/")
+
+    url = f"{base_url}/rest/v1/videos?id=eq.{video_id}"
     headers = {
         "apikey": supabase_anon_key,
         "Authorization": f"Bearer {user_token}",
@@ -226,6 +239,7 @@ def get_ffmpeg_style(caption_style: str, custom_font: str = None, custom_size: i
     if custom_size:
         style["Fontsize"] = str(custom_size)
     if custom_color:
+        # BGR (Blue-Green-Red) hex format for ASS style colors
         color_map = {
             "yellow": "&H0000FFFF",
             "green": "&H0000FF00",
@@ -258,57 +272,86 @@ def process_video_task(job: TranslationJob):
     user_token = job.user_token
     
     print(f"Starting background process for video ID: {video_id}")
-    gemini_key, elevenlabs_key, is_mock_mode = get_api_keys()
+    
+    # Resolve Supabase config (prioritize request body, fallback to env)
+    supabase_url = job.supabase_url or os.environ.get("NEXT_PUBLIC_SUPABASE_URL", "")
+    supabase_anon_key = job.supabase_anon_key or os.environ.get("NEXT_PUBLIC_SUPABASE_ANON_KEY", "")
+    
+    # Resolve API keys (prioritize request body, fallback to env)
+    gemini_key = job.gemini_api_key or os.environ.get("GEMINI_API_KEY", "")
+    elevenlabs_key = job.elevenlabs_api_key or os.environ.get("ELEVENLABS_API_KEY", "")
+    
+    # Check if they are valid or placeholders
+    is_gemini_mock = not gemini_key or "your-gemini" in gemini_key or "here" in gemini_key
+    is_eleven_mock = not elevenlabs_key or "your-eleven" in elevenlabs_key or "here" in elevenlabs_key
+    is_mock_mode = is_gemini_mock or is_eleven_mock
     
     if is_mock_mode:
-        run_simulation_pipeline(job)
+        run_simulation_pipeline(job, supabase_url, supabase_anon_key)
     else:
-        run_production_pipeline(job, gemini_key, elevenlabs_key)
+        run_production_pipeline(job, gemini_key, elevenlabs_key, supabase_url, supabase_anon_key)
 
-def run_simulation_pipeline(job: TranslationJob):
+def run_simulation_pipeline(job: TranslationJob, supabase_url: str, supabase_anon_key: str):
     video_id = job.video_id
     user_token = job.user_token
     
     print(f"[{video_id}] Running in SIMULATION MODE...")
+    
     try:
-        time.sleep(2)
+        # Step 1: Simulate Audio Extraction (Status is already processing)
         print(f"[{video_id}] Step 1: Simulating audio extraction...")
-        time.sleep(3)
-        print(f"[{video_id}] Step 2: Simulating Gemini translation...")
-        update_db_status(video_id, user_token, "processing")
         time.sleep(2)
-        print(f"[{video_id}] Step 3: Simulating ElevenLabs Voice synthesis...")
+        
+        # Step 2: Simulate Gemini Translation
+        print(f"[{video_id}] Step 2: Simulating Gemini 2.0 Flash transcription & translation...")
+        update_db_status(video_id, user_token, "processing", supabase_url=supabase_url, supabase_anon_key=supabase_anon_key) # Just refresh database connection
+        time.sleep(3)
+        
+        # Step 3: Simulate ElevenLabs Voice Clone Synthesis
+        print(f"[{video_id}] Step 3: Simulating ElevenLabs voice clone synthesis with '{job.voice_tone}' tone...")
         time.sleep(2)
         
+        # Step 4: Simulate Subtitle Burning and Audio-Video Merge
         if job.has_captions:
+            override_msg = ""
+            if job.custom_font or job.custom_size or job.custom_color:
+                override_msg = f" (Overrides: font={job.custom_font}, size={job.custom_size}, color={job.custom_color})"
+            print(f"[{video_id}] Step 4: Simulating FFmpeg subtitle burning in '{job.caption_style}' style{override_msg}...")
             time.sleep(2)
+            
         if job.has_lip_sync:
+            print(f"[{video_id}] Step 5: Simulating Wav2Lip alignment (LipSync active)...")
             time.sleep(2)
+        else:
+            print(f"[{video_id}] Step 5: Simulating FFmpeg video & audio track merge...")
+            time.sleep(1)
             
+        # Target Lang Mock URL selector
+        # To make it feel premium, we can point to a high quality public MP4 file
         mock_output_video = "https://commondatastorage.googleapis.com/gtv-videos-bucket/sample/ForBiggerEscapes.mp4"
-        print(f"[{video_id}] Simulation complete. Updating DB to completed.")
-        update_db_status(video_id, user_token, "completed", translated_url=mock_output_video)
+        
+        print(f"[{video_id}] Simulation complete. Updating database to completed.")
+        update_db_status(video_id, user_token, "completed", translated_url=mock_output_video, supabase_url=supabase_url, supabase_anon_key=supabase_anon_key)
         
     except Exception as e:
         print(f"[{video_id}] Simulation Error: {e}")
-        update_db_status(video_id, user_token, "failed", error_message=f"Simülasyon Hatası: {str(e)}")
+        update_db_status(video_id, user_token, "failed", error_message=f"Simülasyon Hatası: {str(e)}", supabase_url=supabase_url, supabase_anon_key=supabase_anon_key)
 
-def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key: str):
+def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key: str, supabase_url: str, supabase_anon_key: str):
     video_id = job.video_id
     user_token = job.user_token
-    supabase_url, supabase_anon_key = get_supabase_config()
     
     print(f"[{video_id}] Running in PRODUCTION MODE...")
     
-    # Gelen URL'den uzantıyı tespit edip kaydediyoruz (.mp4, .mov, .webm vb.)
+    # Detect extension from URL or default to mp4
     url_without_params = job.original_video_url.split('?')[0]
     file_ext = url_without_params.split('.')[-1].lower() if '.' in url_without_params else 'mp4'
     if file_ext not in ['mp4', 'mov', 'webm']:
         file_ext = 'mp4'
         
-    # Geçici çalışma dosyaları
+    # Temporary filenames
     downloaded_video_path = f"temp_{video_id}_downloaded.{file_ext}"
-    input_video_path = f"temp_{video_id}_input.mp4"  # Standartlaştırılmış MP4 dosyası
+    input_video_path = f"temp_{video_id}_input.mp4"  # This will hold the standardized version
     extracted_audio_path = f"temp_{video_id}_audio.mp3"
     synthesized_audio_path = f"temp_{video_id}_tts.mp3"
     srt_file_path = f"temp_{video_id}.srt"
@@ -316,7 +359,7 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
     lipsync_output_path = f"temp_{video_id}_lipsync.mp4"
     
     try:
-        # 1. Orijinal dosyayı indir
+        # 1. Download original video
         print(f"[{video_id}] Downloading original video from: {job.original_video_url}")
         res = requests.get(job.original_video_url, stream=True)
         res.raise_for_status()
@@ -324,8 +367,7 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             for chunk in res.iter_content(chunk_size=8192):
                 f.write(chunk)
                 
-        # 1.5. Video standardizasyon (H.264/AAC, 25 FPS, YUV420p)
-        # Webm ve Mov formatlarını OpenCV ve Wav2Lip ile %100 uyumlu hale getirmek için standart MP4 formatına dönüştürüyoruz.
+        # 1.5. Standardize video (H.264/AAC, 25 FPS, YUV420p) for OpenCV / Wav2Lip stability
         print(f"[{video_id}] Standardizing downloaded video to MP4 format...")
         cmd_standardize = [
             "ffmpeg", "-y", "-i", downloaded_video_path,
@@ -335,7 +377,7 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
         ]
         subprocess.run(cmd_standardize, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
         
-        # 2. Ses Ayıklama
+        # 2. Extract audio from video using FFmpeg
         print(f"[{video_id}] Extracting audio using FFmpeg...")
         cmd_extract = [
             "ffmpeg", "-y", "-i", input_video_path,
@@ -344,8 +386,8 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
         ]
         subprocess.run(cmd_extract, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
         
-        # 3. Gemini 2.0 ile Transkript & Çeviri
-        print(f"[{video_id}] Calling Gemini 2.0 Flash...")
+        # 3. Request Transcription & Translation from Gemini 2.0 Flash
+        print(f"[{video_id}] Calling Gemini 2.0 Flash for transcription/translation to '{job.target_lang}'...")
         with open(extracted_audio_path, "rb") as audio_file:
             audio_data = base64.b64encode(audio_file.read()).decode("utf-8")
             
@@ -360,8 +402,15 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             "contents": [
                 {
                     "parts": [
-                        {"inline_data": {"mime_type": "audio/mp3", "data": audio_data}},
-                        {"text": gemini_prompt}
+                        {
+                            "inline_data": {
+                                "mime_type": "audio/mp3",
+                                "data": audio_data
+                            }
+                        },
+                        {
+                            "text": gemini_prompt
+                        }
                     ]
                 }
             ]
@@ -377,11 +426,14 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             with open(srt_file_path, "w", encoding="utf-8") as srt_file:
                 srt_file.write(srt_content)
             translated_text = srt_to_plain_text(srt_content)
+            print(f"[{video_id}] Gemini SRT translation generated. Plain text length: {len(translated_text)}")
         else:
             translated_text = gemini_output
+            print(f"[{video_id}] Gemini Plain text translation: '{translated_text[:100]}...'")
         
-        # 4. ElevenLabs Yapay Zeka Ses Sentezleme
-        print(f"[{video_id}] Calling ElevenLabs TTS Voice Cloning...")
+        # 4. Synthesize voice cloning using ElevenLabs API
+        print(f"[{video_id}] Calling ElevenLabs TTS Voice Cloning with tone '{job.voice_tone}'...")
+        # We can use the default standard voice (e.g. Rachel: 21m00Tcm4TlvDq8ikWAM)
         voice_id = "21m00Tcm4TlvDq8ikWAM" 
         elevenlabs_url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
         
@@ -390,6 +442,7 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             "Content-Type": "application/json"
         }
         
+        # Optional stability configurations based on voice tone
         stability = 0.5
         similarity_boost = 0.75
         if job.voice_tone == 'excited':
@@ -414,8 +467,11 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
         with open(synthesized_audio_path, "wb") as f:
             f.write(tts_res.content)
             
-        # 5. Lip-Sync & Video Birleştirme
+        # 5. Merge synthesized audio back with the original video (or run Lip-Sync if selected)
+        lipsync_output_path = f"temp_{video_id}_lipsync.mp4"
+        
         if job.has_lip_sync:
+            print(f"[{video_id}] Setting up Wav2Lip model checkpoint files...")
             setup_success = setup_wav2lip()
             if not setup_success:
                 raise Exception("Wav2Lip model files setup failed. Please check server logs.")
@@ -428,15 +484,21 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
                 "--audio", f"../{synthesized_audio_path}",
                 "--outfile", f"../{lipsync_output_path}"
             ]
+            
+            # Execute in the Wav2Lip directory to resolve relative imports
             result = subprocess.run(cmd_lipsync, cwd="Wav2Lip", capture_output=True, text=True)
             if result.returncode != 0:
-                raise Exception(f"Wav2Lip inference failed with exit code {result.returncode}")
+                print(f"Wav2Lip stdout: {result.stdout}")
+                print(f"Wav2Lip stderr: {result.stderr}")
+                raise Exception(f"Wav2Lip inference failed with exit code {result.returncode}: {result.stderr}")
+                
+            print(f"[{video_id}] Wav2Lip alignment completed successfully.")
             video_src_for_subtitles = lipsync_output_path
         else:
             video_src_for_subtitles = input_video_path
             
         if job.has_captions:
-            print(f"[{video_id}] Burning subtitles using FFmpeg...")
+            print(f"[{video_id}] Burning subtitles with style '{job.caption_style}' using FFmpeg...")
             style_str = get_ffmpeg_style(
                 job.caption_style,
                 custom_font=job.custom_font,
@@ -445,6 +507,8 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             )
             
             if job.has_lip_sync:
+                # Lip-synced video already has the synthesized audio merged inside it.
+                # So we map audio track from the same file (0:a)
                 cmd_merge = [
                     "ffmpeg", "-y", "-i", video_src_for_subtitles,
                     "-vf", f"subtitles={srt_file_path}:force_style='{style_str}'",
@@ -453,6 +517,7 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
                     output_video_path
                 ]
             else:
+                # Map audio from input 1 (synthesized audio path)
                 cmd_merge = [
                     "ffmpeg", "-y", "-i", video_src_for_subtitles, "-i", synthesized_audio_path,
                     "-vf", f"subtitles={srt_file_path}:force_style='{style_str}'",
@@ -463,8 +528,10 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             subprocess.run(cmd_merge, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
         else:
             if job.has_lip_sync:
+                print(f"[{video_id}] Copying lip-synced video to final output path...")
                 shutil.copy2(lipsync_output_path, output_video_path)
             else:
+                print(f"[{video_id}] Merging audio track into original video using FFmpeg...")
                 cmd_merge = [
                     "ffmpeg", "-y", "-i", video_src_for_subtitles, "-i", synthesized_audio_path,
                     "-map", "0:v", "-map", "1:a", "-c:v", "copy", "-c:a", "aac", "-shortest",
@@ -472,9 +539,14 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
                 ]
                 subprocess.run(cmd_merge, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
                 
-        # 6. Sonuç Videosunu Supabase Storage'a Yükle
-        print(f"[{video_id}] Uploading output to Supabase Storage...")
-        storage_upload_url = f"{supabase_url}/storage/v1/object/videos/{video_id}_translated.mp4"
+        # Normalize Supabase base URL (remove trailing slash and rest/v1 path if present)
+        base_url = supabase_url.rstrip("/")
+        if base_url.endswith("/rest/v1"):
+            base_url = base_url[:-8].rstrip("/")
+
+        # 6. Upload output video to Supabase Storage
+        print(f"[{video_id}] Uploading output video to Supabase Storage...")
+        storage_upload_url = f"{base_url}/storage/v1/object/videos/{video_id}_translated.mp4"
         
         with open(output_video_path, "rb") as out_file:
             upload_headers = {
@@ -483,20 +555,25 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
                 "Content-Type": "video/mp4"
             }
             upload_res = requests.post(storage_upload_url, headers=upload_headers, data=out_file)
+            # If exists, we can try to PUT (overwrite)
             if upload_res.status_code == 400 and "AlreadyExists" in upload_res.text:
                 upload_res = requests.put(storage_upload_url, headers=upload_headers, data=out_file)
+            
             upload_res.raise_for_status()
             
-        public_video_url = f"{supabase_url}/storage/v1/object/public/videos/{video_id}_translated.mp4"
+        # Get public url
+        public_video_url = f"{base_url}/storage/v1/object/public/videos/{video_id}_translated.mp4"
         print(f"[{video_id}] Video successfully uploaded. Public URL: {public_video_url}")
-        update_db_status(video_id, user_token, "completed", translated_url=public_video_url)
+        
+        # 7. Update database to completed
+        update_db_status(video_id, user_token, "completed", translated_url=public_video_url, supabase_url=supabase_url, supabase_anon_key=supabase_anon_key)
         
     except Exception as e:
         print(f"[{video_id}] Production Pipeline Error: {e}")
-        update_db_status(video_id, user_token, "failed", error_message=f"Hata: {str(e)}")
+        update_db_status(video_id, user_token, "failed", error_message=f"Hata: {str(e)}", supabase_url=supabase_url, supabase_anon_key=supabase_anon_key)
         
     finally:
-        # Geçici dosyaların temizlenmesi
+        # Cleanup temporary files
         for temp_file in [downloaded_video_path, input_video_path, extracted_audio_path, synthesized_audio_path, srt_file_path, output_video_path, lipsync_output_path]:
             if os.path.exists(temp_file):
                 try:
@@ -511,5 +588,6 @@ def process_video(job: TranslationJob, background_tasks: BackgroundTasks):
 
 if __name__ == "__main__":
     import uvicorn
+    # Make sure we load local variables before starting
     load_env_local()
     uvicorn.run(app, host="0.0.0.0", port=7860)