Description
Hello,
I am writing to report technical inconsistencies between the ScraperFC repository as cloned/used on a Linux environment and the official GitHub source. These discrepancies are causing significant issues for reproducibility and integration in automated pipelines.
Technical Issues Observed:
API Surface Differences:
The local repo (installed via pip or from source on Linux) exposes different methods and attributes compared to the GitHub version.
For example, the documentation and some code samples reference scraperfc.fbref.leagues(), but in the current version (3.3.0), this method does not exist. Instead, FBref.comps is the correct attribute.
Similarly, some methods are present in the codebase but not actually exposed after installation, or vice versa.
Module Structure and Import Paths:
There are inconsistencies in how modules are structured and imported.
Imports like import scraperfc.fbref or from scraperfc import fbref sometimes fail, depending on whether the package is installed via pip or used directly from a cloned repo.
Relative imports inside the package can break when running scripts outside the package root.
Documentation Out-of-Sync:
The README and in-code docstrings do not always match the actual available API.
Example: The usage of scraperfc.fbref.leagues() is shown, but in practice, only FBref().comps works.
Some method signatures and expected arguments differ between the docs and the real code.
Cross-Platform/Environment Issues:
The behavior and available methods differ between Linux environments and what’s shown on GitHub or in the published package.
This leads to AttributeError, ImportError, and confusion when porting scripts between environments.
Concrete Examples:
AttributeError: module 'scraperfc.fbref' has no attribute 'leagues' when following the documentation.
Import errors when using from scraperfc import fbref in a venv on Linux, but not on Windows.
The need to patch scripts to use FBref().comps and other attributes not mentioned in the docs.
Suggestions for Resolution:
Ensure the published PyPI package, the GitHub repo, and the documentation are always in sync.
Add CI tests for importability and API surface on both Linux and Windows.
Update the documentation to reflect the actual, current API (e.g., use FBref().comps instead of fbref.leagues()).
Consider using absolute imports throughout the package to avoid context-dependent import errors.
If you need more details, error logs, or my environment specs, I am happy to provide them.
Thank you for your attention to these technical details. Resolving these issues would greatly improve developer experience and reliability for all users.
Best regards,