Update app.py

app.py

@@ -1,19 +1,249 @@
 import gradio as gr
+import os
+import tempfile
+import shutil
+from typing import Optional, Union
+from pathlib import Path
+from huggingface_hub import InferenceClient
 
-# Create the Gradio interface
-with gr.Blocks(theme=gr.themes.Soft(), title="Sora 2") as demo:
-    gr.Markdown(
-        """
-        # 🚧 Under Maintenance
-        
-        Sora 2 app is currently under maintenance. Please check back later.
-        
-        We apologize for any inconvenience.
-
-        Check https://huggingface.co/spaces/akhaliq/sora-2/discussions/32 for latest updates
-        """,
+# -------------------------
+# Utilities
+# -------------------------
+
+def cleanup_temp_files():
+    try:
+        temp_dir = tempfile.gettempdir()
+        for file_path in Path(temp_dir).glob("*.mp4"):
+            try:
+                import time
+                if file_path.stat().st_mtime < (time.time() - 300):
+                    file_path.unlink(missing_ok=True)
+            except Exception:
+                pass
+    except Exception as e:
+        print(f"Cleanup error: {e}")
+
+def _client_from_token(token: Optional[str]) -> InferenceClient:
+    if not token:
+        raise gr.Error("Please sign in first. This app requires your Hugging Face login.")
+    # IMPORTANT: do not set bill_to when using user OAuth tokens
+    return InferenceClient(
+        provider="fal-ai",
+        api_key=token,
     )
 
-# Launch the app
+def _save_bytes_as_temp_mp4(data: bytes) -> str:
+    temp_file = tempfile.NamedTemporaryFile(suffix=".mp4", delete=False)
+    try:
+        temp_file.write(data)
+        temp_file.flush()
+        return temp_file.name
+    finally:
+        temp_file.close()
+
+# -------------------------
+# Inference wrappers (no env fallback; always require LoginButton)
+# -------------------------
+
+def generate_video(
+    prompt: str,
+    token: gr.OAuthToken | None,
+    duration: int = 8,          # kept for future use
+    size: str = "1280x720",     # kept for future use
+    *_                          # tolerate extra event payloads
+) -> Optional[str]:
+    if token is None or not getattr(token, "token", None):
+        raise gr.Error("Sign in with Hugging Face to continue. This app uses your inference provider credits.")
+    if not prompt or not prompt.strip():
+        return None
+
+    cleanup_temp_files()
+    try:
+        client = _client_from_token(token.token)
+        # Ensure model id matches what users can access. Change if you intend provider repo.
+        model_id = "akhaliq/sora-2"
+        try:
+            video_bytes = client.text_to_video(prompt, model=model_id)
+        except Exception as e:
+            # Provide a clearer message if this is an HTTP 403 from requests
+            import requests
+            if isinstance(e, requests.HTTPError) and getattr(e.response, "status_code", None) == 403:
+                raise gr.Error(
+                    "Access denied by provider (403). Make sure your HF account has credits/permission "
+                    f"for provider 'fal-ai' and model '{model_id}'."
+                )
+            raise
+        return _save_bytes_as_temp_mp4(video_bytes)
+    except gr.Error:
+        raise
+    except Exception:
+        raise gr.Error("Generation failed. If this keeps happening, check your provider quota or try again later.")
+
+def generate_video_from_image(
+    image: Union[str, bytes, None],
+    prompt: str,
+    token: gr.OAuthToken | None,
+    *_
+) -> Optional[str]:
+    if token is None or not getattr(token, "token", None):
+        raise gr.Error("Sign in with Hugging Face to continue. This app uses your inference provider credits.")
+    if not image or not prompt or not prompt.strip():
+        return None
+
+    cleanup_temp_files()
+    try:
+        # Load image bytes
+        if isinstance(image, str):
+            with open(image, "rb") as f:
+                input_image = f.read()
+        elif isinstance(image, (bytes, bytearray)):
+            input_image = image
+        else:
+            return None
+
+        client = _client_from_token(token.token)
+        model_id = "akhaliq/sora-2-image-to-video"
+        try:
+            video_bytes = client.image_to_video(
+                input_image,
+                prompt=prompt,
+                model=model_id,
+            )
+        except Exception as e:
+            import requests
+            if isinstance(e, requests.HTTPError) and getattr(e.response, "status_code", None) == 403:
+                raise gr.Error(
+                    "Access denied by provider (403). Make sure your HF account has credits/permission "
+                    f"for provider 'fal-ai' and model '{model_id}'."
+                )
+            raise
+        return _save_bytes_as_temp_mp4(video_bytes)
+    except gr.Error:
+        raise
+    except Exception:
+        raise gr.Error("Generation failed. If this keeps happening, check your provider quota or try again later.")
+
+# -------------------------
+# UI
+# -------------------------
+
+def create_ui():
+    css = '''
+    .logo-dark{display: none}
+    .dark .logo-dark{display: block !important}
+    .dark .logo-light{display: none}
+    #sub_title{margin-top: -20px !important}
+    .notice {
+        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+        color: white;
+        padding: 14px 16px;
+        border-radius: 12px;
+        margin: 18px auto 6px;
+        max-width: 860px;
+        text-align: center;
+        font-size: 0.98rem;
+    }
+    '''
+
+    with gr.Blocks(title="Sora-2 (uses your provider credits)", theme=gr.themes.Soft(), css=css) as demo:
+        gr.HTML("""
+            <div style="text-align:center; max-width:900px; margin:0 auto;">
+                <h1 style="font-size:2.2em; margin-bottom:6px;">🎬 Sora-2</h1>
+                <p style="color:#777; margin:0 0 8px;">Generate videos via the Hugging Face Inference API (provider: fal-ai)</p>
+                <div class="notice">
+                    <b>Heads up:</b> This is a paid app that uses <b>your</b> inference provider credits when you run generations.
+                    Free users get <b>$0.10 in included credits</b>. <b>PRO users</b> get <b>$2 in included credits</b> 
+                    and can continue using beyond that (with billing). 
+                    <a href='http://huggingface.co/subscribe/pro?source=sora_2' target='_blank' style='color:#fff; text-decoration:underline; font-weight:bold;'>Subscribe to PRO</a> 
+                    for more credits. Please sign in with your Hugging Face account to continue.
+                </div>
+                <p style="font-size: 0.9em; color: #999; margin-top: 10px;">
+                    Built with <a href="https://huggingface.co/spaces/akhaliq/anycoder" target="_blank" style="color:#fff; text-decoration:underline;">anycoder</a>
+                </p>
+            </div>
+        """)
+
+        login_btn = gr.LoginButton("Sign in with Hugging Face")
+
+        # Text -> Video
+        with gr.Row():
+            with gr.Column(scale=1):
+                prompt_input = gr.Textbox(
+                    label="Enter your prompt",
+                    placeholder="Describe the video you want to create…",
+                    lines=4,
+                    elem_id="prompt-text-input"
+                )
+                generate_btn = gr.Button("🎥 Generate Video", variant="primary")
+            with gr.Column(scale=1):
+                video_output = gr.Video(
+                    label="Generated Video",
+                    height=400,
+                    interactive=False,
+                    show_download_button=True,
+                    elem_id="text-to-video"
+                )
+
+        # Order of inputs: prompt, token
+        generate_btn.click(
+            fn=generate_video,
+            inputs=[prompt_input, login_btn],
+            outputs=[video_output],
+        )
+
+        # Image -> Video
+        gr.HTML("""
+            <div style="text-align:center; margin: 34px 0 10px;">
+                <h3 style="margin-bottom:6px;">🖼️ ➜ 🎬 Image → Video (beta)</h3>
+                <p style="color:#666; margin:0;">Turn a single image into a short video with a guiding prompt.</p>
+            </div>
+        """)
+        with gr.Row():
+            with gr.Column(scale=1):
+                image_input = gr.Image(label="Upload an image", type="filepath")
+                img_prompt_input = gr.Textbox(
+                    label="Describe how the scene should evolve",
+                    placeholder="e.g., The cat starts to dance and spins playfully",
+                    lines=3,
+                    elem_id="img-prompt-text-input"
+                )
+                generate_img_btn = gr.Button("🎥 Generate from Image", variant="primary")
+            with gr.Column(scale=1):
+                video_output_img = gr.Video(
+                    label="Generated Video (from Image)",
+                    height=400,
+                    interactive=False,
+                    show_download_button=True,
+                    elem_id="image-to-video"
+                )
+
+        # Order of inputs: image, prompt, token
+        generate_img_btn.click(
+            fn=generate_video_from_image,
+            inputs=[image_input, img_prompt_input, login_btn],
+            outputs=[video_output_img],
+        )
+
+        # Examples: keep UI-only (no automatic inference to avoid accidental charges)
+        gr.Examples(
+            examples=[["A majestic golden eagle soaring through a vibrant sunset sky"]],
+            inputs=prompt_input
+        )
+
+    return demo
+
+# -------------------------
+# Entrypoint
+# -------------------------
+
 if __name__ == "__main__":
-    demo.launch()
+    try:
+        cleanup_temp_files()
+        if os.path.exists("gradio_cached_examples"):
+            shutil.rmtree("gradio_cached_examples", ignore_errors=True)
+    except Exception as e:
+        print(f"Initial cleanup error: {e}")
+
+    app = create_ui()
+    app.queue(status_update_rate="auto", api_open=False, default_concurrency_limit=None)
+    app.launch(show_api=False, enable_monitoring=False, quiet=True, ssr_mode=True)