update

app.py

@@ -30,23 +30,18 @@ class DatasetManager:
         """Authenticate with Google Drive"""
         try:
             gauth = GoogleAuth()
-            # Use the settings from yaml file
             gauth.settings['client_config_file'] = settings['client_secrets_file']
             
             # Try to load saved credentials
             gauth.LoadCredentialsFile("credentials.txt")
             
             if gauth.credentials is None:
-                # Authenticate if no credentials found
                 gauth.LocalWebserverAuth()
             elif gauth.access_token_expired:
-                # Refresh them if expired
                 gauth.Refresh()
             else:
-                # Initialize the saved credentials
                 gauth.Authorize()
                 
-            # Save the credentials for future use
             gauth.SaveCredentialsFile("credentials.txt")
             
             self.drive = GoogleDrive(gauth)
@@ -60,12 +55,10 @@ class DatasetManager:
             return False, "Google Drive not authenticated", []
         
         try:
-            # List files in the folder
             query = f"'{drive_folder_id}' in parents and trashed=false"
             file_list = self.drive.ListFile({'q': query}).GetList()
             
             if not file_list:
-                # Try to get single file if folder is empty
                 file = self.drive.CreateFile({'id': drive_folder_id})
                 if file:
                     file_list = [file]
@@ -73,7 +66,6 @@ class DatasetManager:
                     return False, "No files found with the specified ID", []
             
             renamed_files = []
-            existing_dataset = None
             try:
                 existing_dataset = load_dataset(self.dataset_name)
                 logger.info(f"Loaded existing dataset: {self.dataset_name}")
@@ -87,10 +79,8 @@ class DatasetManager:
                     new_filename = f"{naming_convention}_{start_index + i + 1}.jpg"
                     file_path = os.path.join(self.local_images_dir, new_filename)
                     
-                    # Download file
                     file.GetContentFile(file_path)
                     
-                    # Verify the image can be opened
                     try:
                         with Image.open(file_path) as img:
                             img.verify()
@@ -112,22 +102,16 @@ class DatasetManager:
     def update_huggingface_dataset(self, renamed_files):
         """Update the sports-cards dataset with new images"""
         try:
-            # Create a DataFrame with the file information
             df = pd.DataFrame(renamed_files)
-            
-            # Create a Hugging Face Dataset
             new_dataset = Dataset.from_pandas(df)
             
             try:
-                # Try to load existing dataset
                 existing_dataset = load_dataset(self.dataset_name)
-                # Concatenate with existing dataset if it exists
                 if 'train' in existing_dataset:
                     new_dataset = concatenate_datasets([existing_dataset['train'], new_dataset])
             except Exception:
                 logger.info("Creating new dataset")
             
-            # Push to Hugging Face Hub
             new_dataset.push_to_hub(self.dataset_name, split="train")
             
             return True, f"Successfully updated dataset '{self.dataset_name}' with {len(renamed_files)} new images"
@@ -138,39 +122,84 @@ def process_pipeline(folder_id, naming_convention):
     """Main pipeline to process images and update dataset"""
     manager = DatasetManager()
     
-    # Step 1: Authenticate
     auth_success, auth_message = manager.authenticate_drive()
     if not auth_success:
         return auth_message
     
-    # Step 2: Download and rename files
     success, message, renamed_files = manager.download_and_rename_files(folder_id, naming_convention)
     if not success:
         return message
     
-    # Step 3: Update Hugging Face dataset
     success, hf_message = manager.update_huggingface_dataset(renamed_files)
     return f"{message}\n{hf_message}"
 
-# Gradio interface
-demo = gr.Interface(
-    fn=process_pipeline,
-    inputs=[
-        gr.Textbox(
-            label="Google Drive File/Folder ID",
-            placeholder="Enter the ID from your Google Drive URL",
-            value="151VOxPO91mg0C3ORiioGUd4hogzP1ujm"
-        ),
-        gr.Textbox(
-            label="Naming Convention",
-            placeholder="e.g., sports_card",
-            value="sports_card"
+# Custom CSS for web-safe fonts and clean styling
+custom_css = """
+.gradio-container {
+    font-family: Arial, sans-serif !important;
+}
+
+h1, h2, h3 {
+    font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif !important;
+    font-weight: 600 !important;
+}
+
+.gr-button {
+    font-family: Arial, sans-serif !important;
+}
+
+.gr-input {
+    font-family: 'Courier New', Courier, monospace !important;
+}
+
+.gr-box {
+    border-radius: 8px !important;
+    border: 1px solid #e5e5e5 !important;
+}
+
+.gr-padded {
+    padding: 16px !important;
+}
+"""
+
+# Gradio interface with custom theme
+with gr.Blocks(css=custom_css) as demo:
+    gr.Markdown("# Sports Cards Dataset Processor")
+    
+    with gr.Box():
+        gr.Markdown("""
+        ### Instructions
+        1. Enter the Google Drive folder/file ID
+        2. Choose a naming convention for your cards
+        3. Click Process to start
+        """)
+    
+    with gr.Row():
+        with gr.Column():
+            folder_id = gr.Textbox(
+                label="Google Drive File/Folder ID",
+                placeholder="Enter the ID from your Google Drive URL",
+                value="151VOxPO91mg0C3ORiioGUd4hogzP1ujm"
+            )
+            naming = gr.Textbox(
+                label="Naming Convention",
+                placeholder="e.g., sports_card",
+                value="sports_card"
+            )
+            process_btn = gr.Button("Process Images", variant="primary")
+    
+    with gr.Box():
+        output = gr.Textbox(
+            label="Processing Status",
+            show_label=True,
+            lines=5
         )
-    ],
-    outputs=gr.Textbox(label="Status"),
-    title="Sports Cards Dataset Processor",
-    description="Download card images from Google Drive and add them to the sports-cards dataset"
-)
+    
+    process_btn.click(
+        fn=process_pipeline,
+        inputs=[folder_id, naming],
+        outputs=output
+    )
 
 if __name__ == "__main__":
     demo.launch()