Update app.py

app.py

@@ -13,13 +13,11 @@ import uvicorn
 import subprocess
 import shutil
 from pathlib import Path
-from gradio.routes import mount_gradio_app
 from fastapi import FastAPI, UploadFile, File, Form, HTTPException
 from fastapi.responses import JSONResponse
 from fastapi.middleware.cors import CORSMiddleware
 
 
-
 def wrap_text(text, max_line_length=29):
     words = text.split()
     lines = []
@@ -393,6 +391,89 @@ def process_alignment(media_file, text_file, language, progress=gr.Progress()):
             print(f"Debug: Error cleaning up temporary files: {cleanup_error}")
 
 
+# Create FastAPI instance
+fastapi_app = FastAPI()
+
+fastapi_app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+@fastapi_app.post("/align")
+async def align_api(
+    media_file: UploadFile = File(...),
+    text_file: UploadFile = File(...),
+    language: str = Form(default="en")
+):
+    try:
+        # Validate text file
+        if not text_file.filename.endswith(".txt"):
+            raise HTTPException(
+                status_code=400,
+                detail="Text file must be a .txt file"
+            )
+        
+        # Check if media file is supported
+        media_filename = media_file.filename.lower()
+        audio_extensions = {'.wav', '.mp3', '.flac', '.aac', '.ogg', '.wma', '.m4a', '.opus'}
+        video_extensions = {'.mp4', '.avi', '.mkv', '.mov', '.wmv', '.flv', '.webm', '.m4v', '.3gp', '.mpg', '.mpeg'}
+        
+        file_ext = Path(media_filename).suffix.lower()
+        if file_ext not in audio_extensions and file_ext not in video_extensions:
+            raise HTTPException(
+                status_code=400,
+                detail=f"Unsupported media file format: {file_ext}. Supported formats: {', '.join(sorted(audio_extensions | video_extensions))}"
+            )
+        
+        # Save uploaded files temporarily
+        with tempfile.NamedTemporaryFile(delete=False, suffix=file_ext) as temp_media:
+            shutil.copyfileobj(media_file.file, temp_media)
+            media_path = temp_media.name
+
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".txt", mode='w+', encoding='utf-8') as temp_text:
+            content = (await text_file.read()).decode('utf-8', errors='ignore')
+            temp_text.write(content)
+            temp_text.flush()
+            text_path = temp_text.name
+
+        # Process alignment
+        status, df, json_path, summary, srt_path, vtt_path = process_alignment(media_path, text_path, language)
+
+        # Clean up uploaded files
+        try:
+            os.unlink(media_path)
+            os.unlink(text_path)
+        except Exception as cleanup_error:
+            print(f"Warning: Error cleaning up uploaded files: {cleanup_error}")
+
+        if "Error" in status or status.startswith("❌"):
+            raise HTTPException(status_code=500, detail=status)
+
+        response = {
+            "status": status,
+            "summary": summary,
+            "segments": df.to_dict(orient="records") if df is not None else [],
+            "download_links": {
+                "alignment_json": json_path,
+                "srt": srt_path,
+                "vtt": vtt_path
+            }
+        }
+
+        return JSONResponse(status_code=200, content=response)
+
+    except HTTPException:
+        raise
+    except Exception as e:
+        raise HTTPException(
+            status_code=500,
+            detail=f"Unexpected server error: {str(e)}"
+        )
+
+
 def create_interface():
     
     with gr.Blocks(title="Aeneas Forced Alignment Tool", theme=gr.themes.Soft()) as interface:
@@ -405,6 +486,8 @@ def create_interface():
         **Supported formats:**
         - **Audio:** WAV, MP3, FLAC, AAC, OGG, WMA, M4A, OPUS
         - **Video:** MP4, AVI, MKV, MOV, WMV, FLV, WebM, M4V, 3GP, MPG, MPEG
+        
+        **API Access:** Use `/align` endpoint for programmatic access
         """)
         
         with gr.Row():
@@ -497,90 +580,30 @@ def create_interface():
     
     return interface
 
-interface = create_interface()
-
-fastapi_app = FastAPI()
-
-mount_gradio_app(fastapi_app, interface, path="/")
-
-# Add CORS middleware
-fastapi_app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"],
-)
 
-@fastapi_app.post("/align")
-async def align_api(
-    media_file: UploadFile = File(...),
-    text_file: UploadFile = File(...),
-    language: str = Form(default="en")
-):
+def main():
     try:
-        # Validate text file
-        if not text_file.filename.endswith(".txt"):
-            raise HTTPException(
-                status_code=400,
-                detail="Text file must be a .txt file"
-            )
-        
-        # Check if media file is supported
-        media_filename = media_file.filename.lower()
-        audio_extensions = {'.wav', '.mp3', '.flac', '.aac', '.ogg', '.wma', '.m4a', '.opus'}
-        video_extensions = {'.mp4', '.avi', '.mkv', '.mov', '.wmv', '.flv', '.webm', '.m4v', '.3gp', '.mpg', '.mpeg'}
+        # Create Gradio interface
+        interface = create_interface()
         
-        file_ext = Path(media_filename).suffix.lower()
-        if file_ext not in audio_extensions and file_ext not in video_extensions:
-            raise HTTPException(
-                status_code=400,
-                detail=f"Unsupported media file format: {file_ext}. Supported formats: {', '.join(sorted(audio_extensions | video_extensions))}"
-            )
+        # Mount FastAPI on Gradio
+        interface.mount_gradio_app(fastapi_app, path="/api")
         
-        # Save uploaded files temporarily
-        with tempfile.NamedTemporaryFile(delete=False, suffix=file_ext) as temp_media:
-            shutil.copyfileobj(media_file.file, temp_media)
-            media_path = temp_media.name
-
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".txt", mode='w+', encoding='utf-8') as temp_text:
-            content = (await text_file.read()).decode('utf-8', errors='ignore')
-            temp_text.write(content)
-            temp_text.flush()
-            text_path = temp_text.name
-
-        # Process alignment
-        status, df, json_path, summary, srt_path, vtt_path = process_alignment(media_path, text_path, language)
-
-        # Clean up uploaded files
-        try:
-            os.unlink(media_path)
-            os.unlink(text_path)
-        except Exception as cleanup_error:
-            print(f"Warning: Error cleaning up uploaded files: {cleanup_error}")
-
-        if "Error" in status or status.startswith("❌"):
-            raise HTTPException(status_code=500, detail=status)
+        print("🚀 Starting Gradio UI on http://localhost:7860")
+        print("🧠 FastAPI API endpoint available at http://localhost:7860/api/align")
 
-        response = {
-            "status": status,
-            "summary": summary,
-            "segments": df.to_dict(orient="records") if df is not None else [],
-            "download_links": {
-                "alignment_json": json_path,
-                "srt": srt_path,
-                "vtt": vtt_path
-            }
-        }
-
-        return JSONResponse(status_code=200, content=response)
+        interface.launch(
+            server_name="0.0.0.0",
+            server_port=7860,
+            share=False,
+            debug=False
+        )
 
-    except HTTPException:
-        raise
+    except ImportError as e:
+        print("❌ Missing dependency:", e)
     except Exception as e:
-        raise HTTPException(
-            status_code=500,
-            detail=f"Unexpected server error: {str(e)}"
-        )
+        print("❌ Error launching application:", e)
+
 
-app = fastapi_app
+if __name__ == "__main__":
+    main()