Small fixes for utils/check_docstrings.py #38915
Conversation
manueldeprada
force-pushed
the
check_docstrings
branch
from
5f6f622 to
b51c9d5
Compare
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
…d use `ast.Constant` More verbose exceptions in `fix_docstring` on docstring formatting issues.
manueldeprada
force-pushed
the
check_docstrings
branch
from
b51c9d5 to
42669f6
Compare
|
Hi @manueldeprada Thanks a lot! This part is old and I am not the author. I think it makes sense but just have one question. Looks like our current code is passing this modified check. |
| @@ -947,7 +948,17 @@ def fix_docstring(obj: Any, old_doc_args: str, new_doc_args: str): | |||
|
|
|||
| if "".join(source[start_idx:idx])[:-1] != old_doc_args: | |||
| # Args are not fully defined in the docstring of this object | |||
| return | |||
| obj_file = find_source_file(obj) | |||
There was a problem hiding this comment.
Logically it looks good. But I have a bit hard time to understand why we might have this situation.
In match_docstring_with_signature, we have
old_doc_arg = "\n".join(obj_doc_lines[start_idx:idx])
and here old_doc_args is the original content of obj.
If you happen to know why we might have this situation "".join(source[start_idx:idx])[:-1] != old_doc_args (and even better if you have an example) , that would be very nice 🙏
There was a problem hiding this comment.
But it's not because of you PR confusing me, the block is already there , just return silently.
node.nbynode.value(deprecation notice)ast.Numbyast.Constant(deprecation notice)--fix_and_overwriteis passed but fixing fails, exit with a detailed exception instead of failing silently.