UnboundLocalError When Using Commit Context Manager

[hahunavth]

Describe the bug

When using the commit context manager to clone, create a commit with a file, and push, an UnboundLocalError is raised in the Hugging Face library code. This error occurs when manually copying a file to the local repository directory and then rerunning the process.

Reproduction

1. Use the commit context manager to clone, create a commit with a file, and push.
2. Manually copy a file to the local repo directory.
3. Rerun step 1.

Logs

Token is valid (permission: write).
Your token has been saved in your configured git credential helpers (store).
Your token has been saved to /home/hahunavth/.cache/huggingface/token
Login successful
/home/hahunavth/PycharmProjects/fastApiProject/./test-repo is already a clone of https://huggingface.co/hahunavth/test-repo. Make sure you pull the latest changes with `repo.git_pull()`.
Traceback (most recent call last):
  File "/home/hahunavth/PycharmProjects/fastApiProject/./utils/hf_utils.py", line 20, in <module>
    with repo.commit(commit_message="My cool model :)", blocking=False):
  File "/home/hahunavth/miniconda3/envs/py3.9/lib/python3.9/contextlib.py", line 119, in __enter__
    return next(self.gen)
  File "/home/hahunavth/miniconda3/envs/py3.9/lib/python3.9/site-packages/huggingface_hub/repository.py", line 1375, in commit
    f" committed: {files_in_msg}. This may lead to errors if checking out"
UnboundLocalError: local variable 'files_in_msg' referenced before assignment

System info

- huggingface_hub version: 0.17.3
- Platform: Linux-6.2.0-33-generic-x86_64-with-glibc2.35
- Python version: 3.9.18
- Running in iPython ?: No
- Running in notebook ?: No
- Running in Google Colab ?: No
- Token path ?: /home/hahunavth/.cache/huggingface/token
- Has saved token ?: True
- Who am I ?: hahunavth
- Configured git credential helpers: store
- FastAI: N/A
- Tensorflow: N/A
- Torch: 2.0.1
- Jinja2: 3.1.2
- Graphviz: N/A
- Pydot: N/A
- Pillow: 10.0.1
- hf_transfer: N/A
- gradio: 3.47.0
- tensorboard: N/A
- numpy: 1.25.2
- pydantic: 2.4.2
- aiohttp: N/A
- ENDPOINT: https://huggingface.co
- HUGGINGFACE_HUB_CACHE: /home/hahunavth/.cache/huggingface/hub
- HUGGINGFACE_ASSETS_CACHE: /home/hahunavth/.cache/huggingface/assets
- HF_TOKEN_PATH: /home/hahunavth/.cache/huggingface/token
- HF_HUB_OFFLINE: False
- HF_HUB_DISABLE_TELEMETRY: False
- HF_HUB_DISABLE_PROGRESS_BARS: None
- HF_HUB_DISABLE_SYMLINKS_WARNING: False
- HF_HUB_DISABLE_EXPERIMENTAL_WARNING: False
- HF_HUB_DISABLE_IMPLICIT_TOKEN: False
- HF_HUB_ENABLE_HF_TRANSFER: False

[hahunavth]

Is that an indentation error? https://github.com/huggingface/huggingface_hub/blob/33b0405a61ad2ff0de732521b79aca226ea787aa/src/huggingface_hub/repository.py#L1369-L1378

[Wauplin]

Hi @hahunavth, thanks for reporting. I think it's due to a forgotten case where less than 5 files are to be staged.

If we replace

     if len(files_to_stage) > 5: 
         files_in_msg = str(files_to_stage[:5])[:-1] + ", ...]" 

by

     files_in_msg = str(files_to_stage[:5])[:-1] + ", ...]" if len(files_to_stage) > 5 else str(files_to_stage)

in the snippet you've shared, it should fix it.

Would you like to open a PR for that? :)

[Wauplin]

Also, Repository is a bit of legacy code now. It is still supported and maintained but we are slowly moving out from it in favor of more practical http-based methods. For more details, check out this explanation guide.

[hahunavth]

Hi @Wauplin, thanks for the quick response and for pinpointing the issue. I've submitted a pull request with the proposed code change. You can review it by following this #1722 . I'm a new to Hugging Face, the information and the link to the explanation guide you provided have been very helpful to me. Thank you for your assistance!

[Wauplin]

Great, thank you! I'll merge the PR once the CI is happy :) Have a great continuation with HF ecosystem :hugs: