acow

I highly recommend nix-direnv and the emacs direnv package. For this example, I quickly threw together a flake using a minimal template,

{
  description = "Python environment for plotting with Seaborn";
  inputs = {
    nixpkgs.url = github:nixos/nixpkgs/nixpkgs-23.05-darwin;
    flake-utils.url = github:numtide/flake-utils;
  };

  outputs = { self, nixpkgs, flake-utils }: 
    flake-utils.lib.eachDefaultSystem (system: 
      let pkgs = import nixpkgs { inherit system; };
          python = pkgs.python3.withPackages (ps: [
            ps.pandas
            ps.seaborn
          ]);
      in {
        devShell = pkgs.mkShell {
          buildInputs = [ python ];
        };
      }
    );
}

Then created a file .envrc with the contents use flake in the same directory as the flake.nix file, and ran direnv allow to allow use of it. I then used this org mode file to test,

Example of plotting from this [[https://andykuszyk.github.io/2023-11-18-using-emacs-org-mode-as-a-jupyter-notebook.html][blog post]].

Some data to work with,

#+begin_src javascript :tangle data.njson
{"name": "Spock", "editor": "Emacs"}
{"name": "James Kirk", "editor": "Vim"}
{"name": "Dr McCoy", "editor": "Vim"}
{"name": "Scotty", "editor": "Emacs"}
{"name": "Worf", "editor": "ed"}
{"name": "Geordi LaForge", "editor": "Emacs"}
{"name": "Data", "editor": "Emacs"}
{"name": "Jean-luc Picard", "editor": "VS Code"}
{"name": "Wesley Crusher", "editor": "VS Code"}
{"name": "William Riker", "editor": "Vim"}
#+end_src

And now we plot,

#+begin_src python :results output file :file usage.png
import pandas as pd
import seaborn as sns
import sys

df = pd.read_json("data.njson", lines=True)
axes = sns.histplot(df, x="editor")
axes.get_figure().savefig(sys.stdout.buffer)
#+end_src

#+RESULTS:
[[file:usage.png]]

I highly recommend nix-direnv along with the direnv emacs package. I do everything with flakes, so I'd have a flake.nix that defines a shell with the inputs I need, then I'd have a .envrc with the contents use flake in the same directory. With those in place, you may need to run direnv allow in that directory, and then you can edit a .org file in the same directory as the flake and the python environment will have what you need.

Here's a flake I quickly made using a minimal template.

{
  description = "Python environment for plotting with Seaborn";
  inputs = {
    nixpkgs.url = github:nixos/nixpkgs/nixpkgs-23.05-darwin;
    flake-utils.url = github:numtide/flake-utils;
  };

  outputs = { self, nixpkgs, flake-utils }: 
    flake-utils.lib.eachDefaultSystem (system: 
      let pkgs = import nixpkgs { inherit system; };
          python = pkgs.python3.withPackages (ps: [
            ps.pandas
            ps.seaborn
          ]);
      in {
        devShell = pkgs.mkShell {
          buildInputs = [ python ];
        };
      }
    );
}

And then here's the .org file I tested with,

Example of plotting from this [[https://andykuszyk.github.io/2023-11-18-using-emacs-org-mode-as-a-jupyter-notebook.html][blog post]].

Some data to work with,

#+begin_src javascript :tangle data.njson
{"name": "Spock", "editor": "Emacs"}
{"name": "James Kirk", "editor": "Vim"}
{"name": "Dr McCoy", "editor": "Vim"}
{"name": "Scotty", "editor": "Emacs"}
{"name": "Worf", "editor": "ed"}
{"name": "Geordi LaForge", "editor": "Emacs"}
{"name": "Data", "editor": "Emacs"}
{"name": "Jean-luc Picard", "editor": "VS Code"}
{"name": "Wesley Crusher", "editor": "VS Code"}
{"name": "William Riker", "editor": "Vim"}
#+end_src

And now we plot,

#+begin_src python :results output file :file usage.png
import pandas as pd
import seaborn as sns
import sys

df = pd.read_json("data.njson", lines=True)
axes = sns.histplot(df, x="editor")
axes.get_figure().savefig(sys.stdout.buffer)
#+end_src

#+RESULTS:
[[file:usage.png]]