allow api token

app.py

@@ -68,13 +68,24 @@ def clean(text):
     return text
 
 
-def pdf2dataset(pathes, progress=gr.Progress()):
+def pdf2dataset(pathes, user_id, dataset_id, token, 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"
+    if dataset_id == "":
+        dataset_id = f"{random.getrandbits(128):x}"
+    if token == "":
+        token = os.getenv("HF_TOKEN")
+
     progress(0, desc="Starting...")
     readers = []
     for path in pathes:
-        if not path.endswith(".pdf"):
+        try:
+            readers.append(PdfReader(path))
+        except Exception as e:
             raise gr.Error(f"Failed to read {path.split('/')[-1]}.")
-    readers = [PdfReader(path) for path in pathes]
     num_pages = sum(len(reader.pages) for reader in readers)
     filenames = [path.split("/")[-1] for path in pathes]
 
@@ -93,48 +104,57 @@ def pdf2dataset(pathes, 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_name = f"{random.getrandbits(128):x}"
-    dataset.push_to_hub(f"pdf2dataset/{dataset_name}", token=os.getenv("HF_TOKEN"))
+    dataset.push_to_hub(f"{user_id}/{dataset_id}", token=token)
     progress(1, desc="Done!")
 
-    instructions = instructions_template.substitute(dataset_name=dataset_name)
+    instructions = instructions_template.substitute(user_id=user_id, dataset_id=dataset_id)
     preview = pd.DataFrame(dataset[:10])
-    print(f"Dataset {dataset_name} uploaded successfully.")
-    return instructions, preview, dataset_name
+    print(f"Dataset {dataset_id} uploaded successfully.")
+    delete_dataset_id = dataset_id if user_id == "pdf2dataset" else ""
+    return instructions, preview, delete_dataset_id
 
 
-def delete_dataset(dataset_name):
-    api = HfApi()
-    if "/" in dataset_name:
-        user_id, dataset_name = dataset_name.split("/")
+def delete_dataset(repo_id_or_dataset_id):
+    # Get the user_id,  dataset_id
+    if "/" in repo_id_or_dataset_id:
+        user_id, dataset_id = repo_id_or_dataset_id.split("/")
+        repo_id = repo_id_or_dataset_id
     else:
         user_id = "pdf2dataset"
+        dataset_id = repo_id_or_dataset_id
+        repo_id = f"{user_id}/{dataset_id}"
+
+    # Only allow the deletion of datasets in the pdf2dataset namespace
     if not user_id == "pdf2dataset":
-        print(f"Invalid namespace deteced in {dataset_name}.")
-        return f"❌ Invalid namespace deteced: {user_id}"
-    repo_id = f"{user_id}/{dataset_name}"
+        print(f"Deleting datasets in the {user_id} namespace is not allowed.")
+        return f"❌ Deleting datasets in the {user_id} namespace is not allowed."
+
+    # Delete the dataset
+    api = HfApi()
     try:
         api.delete_repo(repo_id, repo_type="dataset")
-        print(f"Dataset {dataset_name} deleted successfully.")
+        print(f"Dataset {repo_id} deleted successfully.")
         return "✅ Dataset deleted successfully."
     except Exception as e:
-        print(f"Error deleting dataset{dataset_name}: {e}")
+        print(f"Error deleting dataset{repo_id}: {e}")
         return f"❌ Error deleting dataset: {e}"
 
 
 caution_text = """⚠️ Caution:
 - This process will upload your data to a public Hugging Face repository. Do not upload sensitive information.
 - Anyone (including you) will be able to delete the dataset once it is uploaded.
+
+To avoid this, you can push the dataset to your personal Hugging Face account ⬇️
 """
 
 instructions_template = Template(
     """
-🔗: https://huggingface.co/datasets/pdf2dataset/$dataset_name.
+🔗: https://huggingface.co/datasets/$user_id/$dataset_id.
 
 ```python
 from datasets import load_dataset
 
-dataset = load_dataset("pdf2dataset/$dataset_name")
+dataset = load_dataset("$user_id/$dataset_id")
 ```
     """
 )
@@ -144,20 +164,33 @@ 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):
+        gr.Markdown(
+            """Recommended for API token
+- Go to https://huggingface.co/settings/tokens?new_token=true 
+- Choose "Fine-grained"
+- Check only _**Repos**/Write access to contents/settings of all repos under your personal namespace_
+- Revoke the token after use"""
+        )
+        with gr.Row():
+            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")
+
     gr.Markdown("## 2️⃣ Convert the PDFs and upload")
     convert_button = gr.Button("🔄 Convert and upload")
     preview = gr.Dataframe(
         label="Preview (first 10 rows)", headers=["text", "source"], datatype=["str", "str"], row_count=10, wrap=True, height=200
     )
     gr.Markdown("## 3️⃣ Use the dataset in your code")
-    instructions = gr.Markdown(instructions_template.substitute(dataset_name="generated_dataset_name"))
+    instructions = gr.Markdown(instructions_template.substitute(user_id="pdf2dataset", dataset_id="generated_dataset_id"))
     gr.Markdown("## 4️⃣ Delete the dataset (optional)")
-    dataset_name_to_delete = gr.Textbox("", placeholder="Enter dataset name to delete", label="Dataset to delete")
+    dataset_id_to_delete = gr.Textbox("", placeholder="Enter dataset name to delete", label="Dataset to delete")
     delete_button = gr.Button("🗑️ Delete dataset")
 
     # Define the actions
-    convert_button.click(pdf2dataset, inputs=[file], outputs=[instructions, preview, dataset_name_to_delete])
-    delete_button.click(delete_dataset, inputs=[dataset_name_to_delete], outputs=[delete_button])
-    dataset_name_to_delete.input(lambda: "🗑️ Delete dataset", outputs=[delete_button])
+    convert_button.click(pdf2dataset, inputs=[file, user_id, dataset_id, token], 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])
 
 demo.launch()