Fix plot_layout.__radd__ crash with plotnine's native composition operators

[Copilot]

The docs build was failing with AttributeError: 'plot_layout' object has no attribute 'setup' when the composition-and-export.ipynb notebook executed (left | right) + plot_layout(widths=[1.6, 1]).

Root cause

In plotnine 0.15.x, ggplot.__or__ creates a native plotnine.composition.Beside. The native Compose.__add__ doesn't recognise plot_layout as a composition modifier, so it delegates it to the last ggplot:

native_Beside + plot_layout(...)
  → ggplot.__iadd__(plot_layout(...))
    → plot_layout.__radd__(ggplot_obj)   # ← called with a ggplot, not a Compose
      → ggplot.layout = self             # ← overwrites plotnine's internal Layout

When ggplot._build() subsequently called layout.setup(layers, self), it found plot_layout (which has no setup) instead of plotnine's internal Layout object.

Fix

Add a type guard in plot_layout.__radd__: if cmp is a ggplot instance (i.e. we're being added via native composition plumbing rather than directly to a plotnine_extra Compose), return cmp unchanged instead of corrupting its internal layout.

def __radd__(self, cmp: Compose) -> Compose:
    from plotnine import ggplot as _ggplot
    if isinstance(cmp, _ggplot):
        # Cannot apply plot_layout to a ggplot directly; leave it unchanged
        # so native composition can still render without crashing.
        return cmp
    cmp.layout = self
    return cmp

plot_layout applied via a plotnine_extra Compose (e.g. Beside([p1, p2]) + plot_layout(...)) continues to work as before. The full custom-width/height feature is gated on plotnine ≥ 0.16.0 (which adds Compose.layout), consistent with the existing skip condition in test_plot_composition.py.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.