change create_symlink with type hints and error handling

[marvin1099]

Refactor create_symlink function to add type hints and enhance error handling for existing symlinks and files.

This is how i would change create_symlink.

[gemini-code-assist]

Code Review

This pull request refactors the create_symlink function in src/mainutils.py to use pathlib for more robust path handling and adds logic to manage existing destination files or directories. The review feedback identifies a missing pathlib import that would lead to a NameError and highlights that resolving the source path to an absolute path may break relative symlinks, suggesting the use of the original source path to ensure portability.

[marvin1099]

gemini-code-assist was good this time.
Fixed missing import pathlib
and using a resolved path only for checks not for making the link to maintain relative paths.
Force pushed to keep commits clean.

[shymega]

Yes, it was effective on this one. The only thing I'm a little bit against is the nested conditionals and excessive on that front. Do we need so many? Could we use AND/OR logic?

…

On 30.04.2026 04:38, Marvin1099 wrote: marvin1099 left a comment (DeckCheatz/wemod-launcher#257) [gemini-code-assist](https://github.com/apps/gemini-code-assist) was good this time. Fixed missing `import pathlib` and using a resolved path only for checks not for making the link to maintain relative paths. Force pushed to keep commits clean. -- Reply to this email directly or view it on GitHub: #257 (comment) You are receiving this because your review was requested. Message ID: ***@***.***>

[marvin1099]

Yes, it was effective on this one. The only thing I'm a little bit against is the nested conditionals and excessive on that front. Do we need so many? Could we use AND/OR logic?

Flattend code as to your request.

[shymega]

Thanks. Looks like you need to rebase again, and then we can do a onceover of this PR. Might be worth triggering Gemini again.

…

On 01.05.2026 01:00, Marvin1099 wrote: marvin1099 left a comment (DeckCheatz/wemod-launcher#257) > Yes, it was effective on this one. The only thing I'm a little bit against is the nested conditionals and excessive on that front. Do we need so many? Could we use AND/OR logic? Flattend code as to your request. -- Reply to this email directly or view it on GitHub: #257 (comment) You are receiving this because your review was requested. Message ID: ***@***.***>

[marvin1099]

Now its rebased

[shymega]

/gemini review

[shymega]

Just prodding Gemini for another review, and then I'll test.

[gemini-code-assist]

Code Review

This pull request refactors src/mainutils.py by moving pathlib imports to the top level and rewriting the create_symlink function with type hints and enhanced replacement logic. However, the refactored function introduces critical bugs: it incorrectly resolves relative paths relative to the current working directory instead of the destination's parent, contains a logic error that prevents creating new symlinks when replace is False, and risks crashing by calling unlink on non-existent paths.

[shymega]

@marvin1099 I'm not sure what you think of the bot's recent review, but I'd like to make sure the launcher doesn't crash.

OOI, what version of Python are you on? I'm on 3.13.x, but if people use varying levels of Python (before we transition to a compiled binary), it could have unexpected behaviour between systems.

[marvin1099]

Fixed. I know what a long comment.

[shymega]

@marvin1099 I've rebased this PR, would like to get it into my PR soon 👍

[shymega]

/gemini review

[gemini-code-assist]

Code Review

This pull request removes the requests dependency by implementing a custom http_get utility and a SimpleResponse class based on urllib.request. It also refactors create_symlink to use pathlib for more robust path management. Feedback highlights issues in SimpleResponse where the text property consumes the entire stream and iter_content is not reusable. Additionally, the default destructive behavior of create_symlink when encountering directories was flagged as a safety concern.

I am having trouble creating individual review comments. Click here to see my feedback.

src/coreutils.py (91-95)

The text property has a side effect of consuming the entire response stream when _stream is True. This exhausts the underlying socket, which means any subsequent attempts to use iter_content() will return no data. Additionally, it loads the entire content into memory, defeating the purpose of streaming for large files. It is recommended to avoid consuming the stream in a property getter or to cache the content in a way that doesn't break the streaming interface.

src/coreutils.py (108-111)

The iter_content method uses an instance variable self._content_pos to track iteration progress for non-streamed content. This makes the response object's content iterable only once. Subsequent calls to iter_content() will yield nothing because self._content_pos remains at the end of the content. It is better to use a local variable for the iteration to allow multiple passes over the cached data.

        else:
            for i in range(0, len(self.content), chunk_size):
                yield self.content[i:i + chunk_size]

src/mainutils.py (721-724)

The create_symlink function defaults to replace=True, which triggers shutil.rmtree(dst_path) if the destination is a directory. This is a highly destructive operation for a utility function that users might expect to only manage symlinks or files. A safer approach would be to default replace to None or False, or to require an explicit confirmation/flag before recursively deleting a directory tree to prevent accidental data loss.

[shymega]

That was a old push, I've fixed it now. I'll re-run the review with Gemini.

[shymega]

/gemini review

[gemini-code-assist]

Code Review

This pull request refactors src/mainutils.py by promoting the pathlib import to the top level and rewriting the create_symlink function to include type hints and more robust path resolution. Review feedback suggests simplifying the path resolution logic and type hints. Additionally, a high-severity safety concern was raised regarding the default replace=True parameter, which could lead to unintended recursive directory deletion via shutil.rmtree.

[gemini-code-assist]

The type hint Union[bool, None] can be simplified to Optional[bool] since Optional is already imported. More importantly, defaulting replace to True is dangerous because it allows shutil.rmtree to recursively delete entire directory trees by default if they exist at the destination. This is a significant departure from the previous implementation's safety. It is highly recommended to change the default to None (replace only files and symlinks) or False for safety.

Suggested change

	def create_symlink(src: Union[str, pathlib.Path], dst: Union[str, pathlib.Path], replace: Union[bool, None] = True) -> None:
	def create_symlink(src: Union[str, pathlib.Path], dst: Union[str, pathlib.Path], replace: Optional[bool] = None) -> None:

[gemini-code-assist]

The conditional check if not src_path.is_absolute() is redundant when using pathlib. Joining a path with an absolute path (e.g., parent / absolute_src) automatically returns the absolute path. This can be simplified to a single expression.

Suggested change

	check_src = (dst_path.parent / src_path).resolve() if not src_path.is_absolute() else src_path.resolve()
	check_src = (dst_path.parent / src_path).resolve()