DOC: repurpose `Styler` as a general display tool and not an HTML-CSS tool

attack68 commented 2 years ago

Pandas version checks

[X] I have checked that the issue still exists on the latest versions of the docs on main here

Location of the documentation

https://pandas.pydata.org/pandas-docs/dev/user_guide/style.html

Documentation problem

When Styler was originally designed and released, its aim was to display HTML in a Jupyter Notebook, that could be manipulated and conditionally formatted. The user guide and documentation over the years has therefore focused on the HTML side of things, and the more recent display and general formatting features are seen as extras to suit that environment, and patchwork included in the documentation.

This issue probably targets pandas 2.0 where I am suggesting that the Table Visualization part of the user guide pushes Styler as a generalist DataFrame formatter, describing the generic functionality it has, and also expanding the conditional aspects which can now be output to HTML, LaTeX and Excel.

My philosophy with the development I put into Styler was for it to be;

First and foremost a formatting tool to display data in any way the user wants, without changing the underlying data, with outputs to String (i.e. CSV), HTML, LaTeX, Excel, and (ultimately) JSON.
Provide extra conditional formatting features that suit the environment, e.g. HTML-CSS, LaTeX commands, Excel colors, etc.

Any comments welcome, but I think moving forward having a variety of easily accessible, configurable display representations is a great tool for data science.

Comments very welcome.

Suggested fix for documentation

.

datapythonista commented 2 years ago

Makes sense to me. I'd also consider moving the styler to a separate project. Moving to another repo should be trivial, and I think it'd allow the new project to evolve faster, not being constrained by pandas development. And also, it'd open the door to have other competing or related projects, making this part of pandas an extension.

mroeschke commented 2 years ago

+1 for generalizing and spinning off into a dedicated pandas extension (within the pandas-dev Github project).

attack68 commented 2 years ago

Yes this might make sense, possible as part of pandas 2.0. I'm afraid I have little experience in the separating package, new repo management side of things, as regards to setting up tests, and workflows etc, so would need dedicated resources as per the pandas dev team here to do all this - might make this a non-starter? Not a lot of people actively involved in Styler.

datapythonista commented 2 years ago

Once there is agreement, I'm happy to help with this. I think it should be quite quick. In Ibis I moved several parts to third-party repos, and the only tricky part is defining the API among the components. Which I think in this case it's trivial, as I guess the styler is mostly using the pandas public API, and pandas itself doesn't know anything about the styler.

tehunter commented 2 years ago

My two cents as a Styler.to_excel "power user", I like the idea of separating the Styler into a separate package within the pandas environment. I currently have an internal package to export DataFrames to Excel using Styler as a backend, following an extensive set of formatting rules for a consistent design in line with our style guide regardless of DataFrame. I see one benefit of separating the Styler package to be allowing the possible growth a more vibrant Theming and Plugins community.

Also, I've had to do some overriding of Styler class functions to get certain behavior I need. Some of these bug fixes and enhancements I've submitted as pull requests, but since they're subject to Pandas' release tempo, I've had to maintain a duplicate set of overrides to access these features in between Pandas releases. Having a separate package would allow the features of Styler to advance at its own pace.

pandas-dev / pandas