matplotlib to Plotly and back

from plotlyhep import convert

pfig = convert.from_mpl(mpl_fig)       # walk the drawn artists into a go.Figure
mfig = convert.to_mpl(pfig)            # and back

from_mpl reads what matplotlib actually drew: lines (with dash styles), step patches from mplhep.histplot, error-bar containers (into error_y), patches, collections, images, texts placed by their window extent, arrow annotations (standoff and shrink converted), the legend with its order kept through legendrank, and the axes' ranges, tick positions and labels, including log scales and minor ticks. Sizes convert at the figure's dpi, so a 10×10 inch figure at 100 dpi becomes a 1000×1000 px figure whose text is the same size on screen.

to_mpl does the reverse for the trace types Plotly produces from the helpers here: scatter lines and markers, hv steps, bars, error bars, heatmaps, annotations as texts or arrows.

The round trip is tested by pixel diff: mplhep → from_mpl → to_mpl renders within 0.05 grey levels of the original on average. mathtext_to_html and html_to_mathtext bridge the two text markups for the common sub- and superscript cases.

Use it when a plot already exists as matplotlib code and you want the live, hoverable version on a slide; write the plot with plotlyhep directly when you want per-bin hover, because a converted line carries no bin contents.