bugfix huggingface validator #330
Conversation
|
|
||
| REPO_ID_ILLEGAL_CHARACTERS = re.compile(r"[^0-9a-zA-Z-_./]+") | ||
| MSG_PREFIX = "The platform_resource_identifier for HuggingFace should be a valid repo_id. " | ||
| from huggingface_hub.utils import validate_repo_id | ||
|
|
||
|
|
||
| def throw_error_on_invalid_identifier(platform_resource_identifier: str): |
There was a problem hiding this comment.
This is kind of outside the scope of this PR, but it would be nice to have this docstring updated to avoid misunderstanding.
[a-zA-Z0-9] or ”-”, ”_”, ”.”
”—” and ”..” are forbiddenNotice that "-" in the first line has a different encoding than the second one. As I understand, this validation comes from HF, and the second dash is a double dash instead "--" that was probably replaced by the code editor.
In [1]: "-".encode()
Out [1]: b'-'
In [2]: "—".encode()
Out[2]: b'\xe2\x80\x94'
There was a problem hiding this comment.
I agree this would be better. And maybe even to wrap the error hugging face throws since it has similar hard to read glyphs and a confusing sentence structure (https://github.com/huggingface/huggingface_hub/blob/4c7aa33bac0b4fb59796e76fcc9cd7f5ff0fd426/src/huggingface_hub/utils/_validators.py#L161).
There was a problem hiding this comment.
I created a PR to replace the character. It simply does the replacement, but I agree that wrapping it would be better to avoid confusion. #331
There was a problem hiding this comment.
Thanks, I agree. I am wrapping the error HF throws.
There was a problem hiding this comment.
Merging this PR now. I will raise a new PR for better handling error messages,
…ator_docstring Correct docstring of Huggingface validator
|
Is it a distinct error? Then yes, I think we should tackle it in a separate PR. |
I can create an issue and also an initial PR with the changes. We might identify and fix more issues in the following hours :) . |
Bugs in the validator implementation. It is now replaced with the
huggingface_hub.utils.validate_repo_idas suggested in the docs.