@gcu/plot

Matplotlib-style plotting for the browser. Canvas-based, zero runtime dependencies. Thin familiar API: subplots, plot, scatter, imshow, hist, bar.

Part of Auditable.

Pre-1.0 — APIs may break on minor version bumps.

Install

npm install @gcu/plot

Usage

import { subplots, scatter } from '@gcu/plot';

const { fig, ax } = subplots(1, 1);
ax.scatter([1, 2, 3, 4], [1, 4, 9, 16]);
ax.xlabel('x');
ax.ylabel('y²');
document.body.append(fig.canvas);

Quick helpers (plot, scatter, imshow, hist, bar) accept an options object with ax to target an existing axes, otherwise create a new figure.

subplots(R, C) with both R and C greater than 1 returns a nested 2D axes array (matplotlib shape, for [[a,b],[c,d]] = ... destructuring) with a non-enumerable .flat property carrying the row-major flat axes list — mirrors matplotlib's ndarray.flat, so axes.flat[k] works without manual row/col math.

Style palette

plt.style.use(name) swaps the module-level palette read by Axes / Figure / legend fallbacks (per-Axes _bgcolor / _textColor / _gridColor / _frameColor overrides still win). Default is auditable-native dark (#1a1a1a plot bg, #ccc text, transparent figure facecolor). 'default' / 'classic' / 'matplotlib' / 'seaborn' / 'ggplot' / 'bmh' flip to a light matplotlib-shape palette (white plot bg, dark text, opaque #fafafa figure card). 'dark_background' restores dark. Unknown names are silently ignored.

Figure layout extras

• gap / wspace / hspace (px, default 10) on subplots(..., { gap }) or Figure — inter-subplot spacing.
• Figure-level fig._rowLabels / fig._colLabels (arrays of strings) render in 28px-left / 18px-bottom gutters reserved outside the subplot grid. Used by sadpan.plotting.scatter_matrix to live the column names on the figure edges instead of cluttering each cell with axis labels.
• Per-Axes ax._margins = {left, right, top, bottom} overrides the label/tick-driven margin computation — lets a grid of cells pin identical plot rects.
• aspect='equal' for imshow-dominated axes shrinks the plot rect to match data aspect (was: expanded the data range). colorbar is pinned to the cell's right edge using the pre-aspect plot height so it doesn't shrink along.

Matplotlib compatibility

• scatter default marker size is 20 (matplotlib parity; was 4).
• scatter accepts plural matplotlib forms edgecolors / linewidths as aliases for edgecolor / linewidth.
• plot / scatter / bar / barh coerce x/y/heights/widths through _toArr at trace ingress, so sadpan Series / numpy-shaped wrappers without a .length plot correctly.

License

MIT.