Build a deck from Python¶

Generate slides from data that you already compute, so that nobody retypes a number.

from mkdeck import Deck, Embed, Slide

runs = [
    ("run3", "Run 3", 5, 22),
    ("run4", "Run 4", 4, 20),
]  # name, label, crossings, cap

deck = Deck(title="Vault runs", date="2026-09-18")

for name, label, crossings, cap in runs:
    deck.slides.append(
        Slide(
            id=name,
            sentence=f"{crossings} of 5 seeds cross at {cap} N m.",
            embeds=[Embed(f"assets/{name}.html", label=label)],
        )
    )

deck.build("site/")

The loop adds one slide for each run, and then deck.build("site/") writes the deck to site/.

Build the deck¶

Deck.build writes the deck to a folder and returns the path of the HTML file. It takes these arguments:

Find your figures¶

mkdeck resolves every embed relative to source, so Embed("assets/run3.html") points at ./assets/run3.html. Run the script from the folder that holds assets/, or name another folder:

deck.build("site/", source="slides/")

mkdeck copies the assets/ tree into the build, together with every embed that the slides name.

Build again¶

A second build into the same folder brings it up to date. First, mkdeck copies every file afresh. Then it removes each file that the previous build wrote and this build does not need. It finds those files in a .mkdeck-build.json list in the folder.

Limits¶

Next: The slide model.