If it helps there’s a standalone version of the logic.

At a rough scan, it looks like it tries to find a best guess “main content” node by stripping unlikely nodes and then scoring each node. Element type and content contribute to the score

/**
       * Loop through all paragraphs, and assign a score to them based on how content-y they look.
       * Then add their score to their parent node.
       *
       * A score is determined by things like number of commas, class names, etc. Maybe eventually link density.
       **/

Super is for my window manager.

Which I guess is kind of where copy paste live so I’m on board, barring semantic nitpicks

What’s the second crank for?