private

app.py

@@ -68,12 +68,13 @@ def clean(text):
     return text
 
 
-def pdf2dataset(pathes, user_id, dataset_id, token, progress=gr.Progress()):
+def pdf2dataset(pathes, user_id, dataset_id, token, private, progress=gr.Progress()):
     if any([user_id, dataset_id, token]) and not all([user_id, dataset_id, token]):
         raise gr.Error("Please provide all three: User ID, Dataset ID, and API token.")
 
     if user_id == "":
         user_id = "pdf2dataset"
+        private = False
     if dataset_id == "":
         dataset_id = f"{random.getrandbits(128):x}"
     if token == "":
@@ -104,7 +105,7 @@ def pdf2dataset(pathes, user_id, dataset_id, token, progress=gr.Progress()):
     # Upload the dataset to Hugging Face
     progress(0, desc="Uploading to Hugging Face...")
     dataset = Dataset.from_dict({"text": page_texts, "source": page_filenames})
-    dataset.push_to_hub(f"{user_id}/{dataset_id}", token=token)
+    dataset.push_to_hub(f"{user_id}/{dataset_id}", token=token, private=private)
     progress(1, desc="Done!")
 
     instructions = instructions_template.substitute(user_id=user_id, dataset_id=dataset_id)
@@ -164,7 +165,7 @@ with gr.Blocks() as demo:
     gr.Markdown("## 1️⃣ Upload PDFs")
     file = gr.File(file_types=["pdf"], file_count="multiple")
     gr.Markdown(caution_text)
-    with gr.Accordion("🔒 Pushing to my personal Hugging Face account", open=False):
+    with gr.Accordion("🔒 Pushing to my personal Hugging Face namespace", open=False):
         gr.Markdown(
             """Recommended for API token
 - Go to https://huggingface.co/settings/tokens?new_token=true 
@@ -176,6 +177,7 @@ with gr.Blocks() as demo:
             user_id = gr.Textbox(label="User ID", placeholder="Enter your Hugging Face user ID")
             dataset_id = gr.Textbox(label="Dataset ID", placeholder="Enter the desired dataset ID")
             token = gr.Textbox(label="API token", placeholder="Enter a Hugging Face API token")
+            private = gr.Checkbox(label="Private", default=False)
 
     gr.Markdown("## 2️⃣ Convert the PDFs and upload")
     convert_button = gr.Button("🔄 Convert and upload")
@@ -189,7 +191,9 @@ with gr.Blocks() as demo:
     delete_button = gr.Button("🗑️ Delete dataset")
 
     # Define the actions
-    convert_button.click(pdf2dataset, inputs=[file, user_id, dataset_id, token], outputs=[instructions, preview, dataset_id_to_delete])
+    convert_button.click(
+        pdf2dataset, inputs=[file, user_id, dataset_id, token, private], outputs=[instructions, preview, dataset_id_to_delete]
+    )
     delete_button.click(delete_dataset, inputs=[dataset_id_to_delete], outputs=[delete_button])
     dataset_id_to_delete.input(lambda: "🗑️ Delete dataset", outputs=[delete_button])