material-plus-ui 1.6.0 → 1.8.0

package/dist/types.d.ts

@@ -37,6 +37,91 @@ import type * as React from 'react';
  * the caller wants a custom control instead.
  */
 export type MPSize = 'xs' | 'sm' | 'md' | 'lg' | 'xl';
+/**
+ * How tightly a component that holds other things is packed.
+ *
+ * Material's own scale, and Material's own numbers: `0` is the component at the
+ * size it was asked for, and each step below it takes 4dp out. Nothing here
+ * invents a word for that, because the specification already counts — and a
+ * `'compact'` of this library's own would be a second name for a thing MD3 has
+ * a name for, which is the one rule `MPSize` bends and nothing else should.
+ *
+ * ## It is not a second size ladder
+ *
+ * `size` picks which control this is: the height, the type role, and the padding
+ * that follows from both. `density` takes room out of the one that was picked,
+ * and takes it out of the **spacing only** — the type scale does not move. A
+ * table at `-2` is the same words in less room rather than smaller words, which
+ * is what a reader of a dense screen actually wants: more rows, at the size they
+ * could already read.
+ *
+ * That is also why the two axes cannot be collapsed into one. `size="sm"` on a
+ * list is a small list; `density={-2}` on a list is a normal list with more of
+ * it on the screen. A single ladder would make those the same request.
+ *
+ * ## Where it stops
+ *
+ * A step that would take a control under 24px is not taken. That floor is the
+ * one `MPSize` already names — below it a control stops meeting a touch target —
+ * and clamping is the better of the three answers available: refusing the value
+ * would make `density={-3}` an error on exactly the rung most likely to be given
+ * it, and honouring it would ship a control nobody can hit.
+ *
+ * ## Only containers take it
+ *
+ * A button is one control at one height, and `size` is the axis it has. A list,
+ * a table, a toolbar, a card — anything whose job is to hold a number of things
+ * — changes character with how many of them fit on a screen, and that is the
+ * question this answers.
+ */
+export type MPDensity = 0 | -1 | -2 | -3;
+/**
+ * How far off the page a surface is lifted.
+ *
+ * MD3's own levels, and all five of the raised ones. `0` is flat on the page,
+ * `1` is where an elevated card and a resting elevated button sit, `2` is a menu
+ * or a button under the pointer, `3` is a dialog, and `4` and `5` are there
+ * because the specification defines them and a ladder with holes in it is worse
+ * than no ladder at all.
+ *
+ * ## It moves the tone as well as the shadow
+ *
+ * Which is the whole reason this can exist here, and the reason a naive
+ * `elevation` prop cannot. MD3 does not treat height as a free axis: an elevated
+ * surface is `surface-container-low` **under** a level-1 shadow, and the tone
+ * and the shadow are one decision rather than two. A prop that only cast a
+ * shadow would raise a `filled` box into a surface the specification has no name
+ * for — a raised object that is somehow still the flattest tone in the system.
+ *
+ * So each level names a surface role and the shadow that goes with it, which is
+ * the pairing MD3 publishes:
+ *
+ * | `elevation` | Surface role                | Shadow   |
+ * | ----------- | --------------------------- | -------- |
+ * | `0`         | `surface`                   | none     |
+ * | `1`         | `surface-container-low`     | level 1  |
+ * | `2`         | `surface-container`         | level 2  |
+ * | `3`         | `surface-container-high`    | level 3  |
+ * | `4`         | `surface-container-high`    | level 4  |
+ * | `5`         | `surface-container-highest` | level 5  |
+ *
+ * Four and five share a tone on purpose. The specification runs out of container
+ * roles before it runs out of levels, and inventing a sixth tone to keep the
+ * columns tidy would be inventing a colour role.
+ *
+ * ## Its relationship to `variant`
+ *
+ * `variant="elevated"` **is** `elevation={1}`, named. It stays because the
+ * vocabulary is about emphasis — a caller choosing between `filled` and
+ * `elevated` is choosing how loud a card is, not how many pixels it floats — and
+ * because it is the answer nearly every raised surface wants.
+ *
+ * Given a level, the level decides the surface and `variant` is left holding
+ * only its hairline: `outlined` keeps its border, and nothing else paints.
+ * Anything else would be two props writing one `background-color`, and the
+ * winner would depend on the order two class names happened to be generated in.
+ */
+export type MPElevation = 0 | 1 | 2 | 3 | 4 | 5;
 /**
  * An accent colour role.
  *
@@ -45,6 +130,29 @@ export type MPSize = 'xs' | 'sm' | 'md' | 'lg' | 'xl';
  * token sheet has no way to derive.
  */
 export type MPColor = 'primary' | 'secondary' | 'tertiary' | 'error';
+/**
+ * Where a reading changes colour.
+ *
+ * The one place in this library where an accent family is *computed* rather than
+ * chosen, and it lives here rather than inside whichever component needed it
+ * first because more than one component reports the same reading. A meter and a
+ * gauge are one quantity in two shapes, and a page carrying both must not
+ * disagree about where the amber starts.
+ *
+ * Left to the caller this is a ternary at every call site, and the fourth one
+ * would get it wrong.
+ *
+ * The families are still MD3's four. There is no `warning` in the specification,
+ * so an amber band is `tertiary` under whatever source colour the page is
+ * themed from — which is the honest answer, and the reason a threshold names a
+ * role rather than a colour.
+ */
+export interface MPThreshold {
+    /** The value from which this family applies, in the reading's own units. */
+    from: number;
+    /** What the reading turns at and above that point. */
+    color: MPColor;
+}
 /**
  * How much of a surface a component paints.
  *
@@ -83,6 +191,242 @@ export type MPSide = 'top' | 'right' | 'bottom' | 'left';
  * the reading-end corner, which is the left one under RTL.
  */
 export type MPCorner = 'top-start' | 'top-end' | 'bottom-start' | 'bottom-end';
+/**
+ * Where a point sits along the category axis.
+ *
+ * A string is a name and has no position of its own, so it takes the slot its
+ * index gives it. A number or a `Date` is a place on a number line, which is
+ * what lets a chart of readings taken at uneven times draw the gaps between
+ * them rather than spacing them evenly and reporting a pause as a steady rate.
+ */
+export type MPChartCategory = string | number | Date;
+/**
+ * One value, and everything a chart might want to know about it.
+ *
+ * `y: null` is a **gap** rather than a zero — a sensor that was offline, a month
+ * that has not closed. A line breaks across it, an area breaks with it, and a
+ * bar is not drawn. That distinction is the whole reason a datum may be `null`
+ * at all: a chart that draws missing data as zero reports an outage as a
+ * collapse, and it does it in the one shape a reader trusts without checking.
+ */
+export interface MPChartPoint {
+    /** The value. `null` is a gap. */
+    y: number | null;
+    /**
+     * Its place on the category axis. Without one the point is placed by its
+     * index, against the chart's `categories` if it was given any.
+     */
+    x?: MPChartCategory;
+    /**
+     * A second magnitude, for the marks that carry one — a bubble's radius, a
+     * cell's weight. The charts with no use for it ignore it.
+     */
+    z?: number;
+    /**
+     * Overrides its series' colour for this one point: the slice worth pointing
+     * at, the bar that went over budget. An `MPColor` family, or any CSS colour.
+     */
+    color?: MPColor | (string & {});
+    /** What the tooltip, the table and any value label say in place of `y`. */
+    label?: React.ReactNode;
+}
+/** A number, a gap, or a point with more to say about itself. */
+export type MPChartDatum = number | null | MPChartPoint;
+/**
+ * One line, one band of bars, one ring of slices — the unit identity attaches
+ * to.
+ *
+ * Colour follows the series and never its position in the drawing. Hiding
+ * Europe has to leave Asia the colour it already was, so the palette slot comes
+ * from where a series sits in this array rather than from how many of its
+ * neighbours happen to be visible: a reader who learned that blue is Europe
+ * learned something a re-render is not allowed to take back.
+ */
+export interface MPChartSeries {
+    /**
+     * Its name in the legend, the tooltip and the table. Two series or more
+     * always draw a legend, so a series without a name is one the reader has no
+     * way to tell from the others.
+     */
+    name?: string;
+    /** Its values, in category order. */
+    data: readonly MPChartDatum[];
+    /**
+     * Overrides the palette slot this series would take. An `MPColor` family, or
+     * any CSS colour.
+     *
+     * The one place in this library where a colour is not a semantic role, and
+     * deliberately so: a series is an *entity* — a region, a plan, a competitor —
+     * and nothing about being one means success or danger. Reach for it to match
+     * a brand, or to hold one entity's colour steady across two charts. Not to
+     * say how a number should be felt.
+     */
+    color?: MPColor | (string & {});
+    /**
+     * Starts the series hidden. Only means anything beside an interactive legend,
+     * which is what turns it back on.
+     * @default false
+     */
+    hidden?: boolean;
+}
+/**
+ * One span on a timeline — a stretch with two ends.
+ *
+ * Its own type rather than an `end` bolted onto `MPChartPoint`, because a
+ * second position on the axis is a field the other seven charts would carry and
+ * never read. The trade this file makes is that a name means one thing
+ * everywhere, and a point that sometimes has an end and usually does not is the
+ * opposite of that.
+ *
+ * `start` and `end` rather than `x` and `end`: a span has two places on the
+ * axis, and naming one of them `x` only reads correctly to somebody who already
+ * knows which one it is.
+ */
+export interface MPTimelineSpan {
+    /** When it begins. A `Date`, or a number of milliseconds. */
+    start: MPChartCategory;
+    /** And when it is done. One that ends before it starts is drawn either way round. */
+    end: MPChartCategory;
+    /** What the span is called, in the hover panel and the table. */
+    label?: React.ReactNode;
+    /** Overrides its row's colour for this one span. */
+    color?: MPColor | (string & {});
+}
+/**
+ * One row of a timeline, and everything on it.
+ *
+ * A row is a series — one entity, one name, one colour — but its data are spans
+ * rather than values, so it cannot be an `MPChartSeries`. There is no `hidden`
+ * here and no legend to pair one with: a timeline's rows **are** its category
+ * axis, already named down the side, and a twenty-entry legend restating them
+ * is not a filter anybody wants.
+ */
+export interface MPTimelineSeries {
+    /** Its name on the axis, in the hover panel and in the table. */
+    name?: string;
+    /** The spans on this row. Overlapping ones are drawn over each other. */
+    data: readonly MPTimelineSpan[];
+    /** Overrides the palette slot this row would otherwise take. */
+    color?: MPColor | (string & {});
+}
+/** How a line gets from one point to the next. */
+export type MPChartCurve = 'linear' | 'smooth' | 'step';
+/**
+ * Which values are written onto the marks themselves.
+ *
+ * `none` by default everywhere, which is not timidity: a number beside every
+ * point is the most reliable way to make a chart unreadable. Label the last one
+ * or the two extremes, and let the axis and the hover layer carry the rest.
+ * `all` is there for the eight-bar chart where it genuinely is the answer.
+ */
+export type MPChartValueLabels = 'none' | 'last' | 'extremes' | 'all';
+/**
+ * What the pointer uncovers.
+ *
+ * - `index` — every series at the category under the pointer, with a crosshair
+ *   dropped through it. The default wherever the series share an x, because the
+ *   question a line chart is asked is "what happened in March" and not "what is
+ *   this pixel".
+ * - `item` — the one mark being pointed at.
+ * - `none` — nothing, and then the numbers have to be readable some other way.
+ *   The table under every chart is that way.
+ */
+export type MPChartTooltipMode = 'index' | 'item' | 'none';
+/** One series' answer at the category the pointer is resting on. */
+export interface MPChartTooltipItem {
+    /** Its place in the `series` array — the same index its colour came from. */
+    seriesIndex: number;
+    name?: string;
+    color: string;
+    value: number | null;
+    /** `value`, written the way this chart writes numbers. */
+    formatted: string;
+    /** What the point called itself, if it said. */
+    label?: React.ReactNode;
+}
+/** What a replacement tooltip is handed. */
+export interface MPChartTooltipContext {
+    index: number;
+    category: MPChartCategory;
+    /** Only the series that are visible and have a value here. */
+    items: readonly MPChartTooltipItem[];
+}
+/** The hover layer, where `true` and `false` are not enough. */
+export interface MPChartTooltip {
+    /** @default 'index', or 'item' where the marks are not arranged in columns */
+    mode?: MPChartTooltipMode;
+    /**
+     * The line dropped through the plot at the active category. On in `index`
+     * mode and never drawn in `item` mode: a crosshair says "these numbers all
+     * belong to this column", and where there is no column it is a line through
+     * one dot.
+     * @default true
+     */
+    crosshair?: boolean;
+    /** Draws the panel yourself. The frame still decides when it opens and where. */
+    render?: (context: MPChartTooltipContext) => React.ReactNode;
+}
+/**
+ * One of a chart's two axes.
+ *
+ * `xAxis` is the category axis and `yAxis` is the value axis on every chart and
+ * in both orientations. Turning a bar chart on its side changes the drawing and
+ * not what the caller's data means, so it must not also move their axis options
+ * from one prop to the other.
+ */
+export interface MPChartAxis {
+    /**
+     * Leaves it undrawn — its rule, its ticks and its labels — and gives the room
+     * back to the plot.
+     */
+    hidden?: boolean;
+    /** A name for what it measures, set beside it. */
+    label?: React.ReactNode;
+    /**
+     * The gridlines it casts across the plot. On for the value axis and off for
+     * the category axis, which is the one arrangement that helps a value be read
+     * without turning the plot into graph paper.
+     */
+    grid?: boolean;
+    /**
+     * Where the scale starts and ends. Taken from the data otherwise, and from
+     * zero on the value axis — a bar's length is proportional to its value only
+     * when the baseline is zero. Set `min` where zero genuinely is not it.
+     */
+    min?: number;
+    max?: number;
+    /**
+     * Roughly how many ticks. The scale still rounds them to numbers a reader can
+     * do arithmetic on, so this is a target and not a count.
+     */
+    tickCount?: number;
+    /** How a tick is written, overriding the chart's own `format`. */
+    tickFormat?: (value: MPChartCategory, index: number) => React.ReactNode;
+    /**
+     * How much room it keeps for its ticks and its label, in pixels. Measured off
+     * the ticks themselves otherwise. Set it when a long category name needs
+     * more, or when two charts stacked on a dashboard have to line their plots up.
+     */
+    thickness?: number;
+}
+/** Where the legend sits, and what it does when it is used. */
+export interface MPChartLegend {
+    /** Which edge of the plot. @default 'bottom' */
+    side?: MPSide;
+    /** Where along that edge. @default 'center' */
+    align?: MPAlign;
+    /**
+     * Clicking an entry hides and shows its series, and hovering one dims the
+     * rest.
+     * @default true
+     */
+    interactive?: boolean;
+    /**
+     * Writes each series' value at the active category beside its name.
+     * @default false
+     */
+    showValue?: boolean;
+}
 /**
  * Which day a week is drawn as starting on. `0` is Sunday, the way `Date`
  * counts.
@@ -135,6 +479,63 @@ export type MPWindowClass = 'compact' | 'medium' | 'expanded' | 'large' | 'extra
  * five — anything not named keeps whatever the class below it said.
  */
 export type MPResponsive<T> = T | Partial<Record<MPWindowClass, T>>;
+/**
+ * How wide content is allowed to get.
+ *
+ * Two vocabularies in one prop, because a measure is asked for in two different
+ * ways and neither covers the other.
+ *
+ * A **rung of the size ladder** is the common one, and it is pinned to the
+ * window size classes rather than to the type scale: `maxWidth="md"` is "never
+ * wider than an expanded window", which is a sentence about the specification
+ * rather than a number somebody liked. `xs` is one rung below the ladder, at
+ * 480dp, for the column of a form or a sign-in card.
+ *
+ * A **CSS length** is the other: `'60ch'`, `'42rem'`, `'800px'`, or anything else
+ * `max-width` takes. A measure is often a decision about the text rather than
+ * about the window — the classic answer is around 60 characters, which no ladder
+ * of window widths can spell — and a library that only offered its own five
+ * rungs would be making every one of those callers write a `className` to undo
+ * one it did not want.
+ *
+ * `'none'` is neither, and is what nearly every component means by default: no
+ * limit at all.
+ *
+ * The value is not validated. An unusable length reaches `max-width`, which
+ * ignores what it cannot parse and leaves the element unbounded — the same
+ * answer `'none'` gives, and a better one than a component that renders nothing
+ * because a unit was mistyped.
+ */
+export type MPMeasure = MPSize | 'none' | (string & {});
+/**
+ * A class name per part, for the parts a `className` cannot reach.
+ *
+ * `className` lands on a component's **root**, and for most components that is
+ * the whole story: one element, one class list, merged with the component's own.
+ * A few draw parts that no selector written against the root will ever find,
+ * because they are not under it — a select's popup, a dialog's backdrop, a
+ * tooltip's bubble all render at the end of `<body>`. Those are what this is
+ * for.
+ *
+ * ## There is never a `root` key
+ *
+ * `className` is the root, on every component. A second spelling of it is
+ * exactly the drift the prop conventions exist to prevent, and a caller reading
+ * two names for one thing has to work out which one wins.
+ *
+ * ## The part classes are still there
+ *
+ * Every part this library draws already carries a BEM-style class of its own —
+ * `mp-select__popup`, `mp-dialog__backdrop` — and a stylesheet of your own can
+ * always select those. This prop is for the other path: a class written at the
+ * call site, generated in the same Tailwind pass as the component's, which is
+ * the only way a *utility* can reach a part.
+ *
+ * The names are shared where the parts are: `popup` is the same idea on a
+ * select, a menu and a popover. What each component adds beyond that is on its
+ * own page.
+ */
+export type MPSlots<Slot extends string> = Partial<Record<Slot, string>>;
 /**
  * The axes most components share.
  *
@@ -189,8 +590,8 @@ export interface MPStyleProps {
 export interface MPControlEventProps<Element extends HTMLElement = HTMLElement> {
     /**
      * Every keystroke that reaches the control, modifiers included — which makes
-     * this the one to reach for for a combination. It is also the only one of
-     * these where `preventDefault` still means anything.
+     * this the one to reach for when a combination matters. It is also the only
+     * one of these where `preventDefault` still means anything.
      */
     onKeyDown?: React.KeyboardEventHandler<Element>;
     /** The same keystroke released. */
@@ -273,18 +674,83 @@ export type MPIconGlyph = React.ComponentType<MPIconGlyphProps> | React.ReactEle
  */
 export type MPEasing = 'linear' | 'standard' | 'standard-decelerate' | 'standard-accelerate' | 'emphasized' | 'emphasized-decelerate' | 'emphasized-accelerate';
 /**
- * The six effects the `MPAnimate*` components are built out of.
+ * The seven effects the `MPAnimate*` components are built out of.
  *
  * Named after what a reader sees rather than after the CSS property underneath:
  * `zoom` and `grow` are both a change of scale, and they are two words because
  * they are two *gestures* — one arrives from the middle of where it will end
  * up, the other unfolds from an edge of it.
  *
- * Everything past these six is a component rather than a value. A marquee that
+ * Everything past these seven is a component rather than a value. A marquee that
  * duplicates its children, a headline that swaps between them, a typewriter
  * that counts characters — none of those can be a class name and a few numbers.
  */
-export type MPAnimation = 'fade' | 'grow' | 'slide' | 'zoom' | 'rotate' | 'blink';
+export type MPAnimation = 'fade' | 'grow' | 'slide' | 'zoom' | 'rotate' | 'blink' | 'reveal';
+/**
+ * An entrance, on a component that displays something.
+ *
+ * A bare effect name is the whole of what most callers want —
+ * `transition="fade"` — and the object form is there for the rest: a different
+ * duration, a slide from another edge, a zoom that starts closer in.
+ *
+ * ## It runs on mount, once
+ *
+ * There is no `trigger`, no `repeat` and no scroll timeline here, and their
+ * absence is the design. Those are what the `MPAnimate*` components are, and a
+ * prop offering half of them would be a second, worse spelling of machinery that
+ * already exists. The moment the answer is "when it scrolls into view", the
+ * wrapper is the right tool.
+ *
+ * ## Why the prop exists at all, then
+ *
+ * Because wrapping is the wrong shape twice. A component that lays its children
+ * into boxes of its own — `MPStack` — puts them somewhere nothing outside can
+ * reach, so a per-item entrance is only expressible from inside. And wrapping
+ * every card of a grid costs a second element per card, which a
+ * `display: contents` wrapper cannot avoid: an element that generates no box has
+ * nothing to animate.
+ *
+ * It is on the components that **display** something and on none of the
+ * controls. That is a cost decision as much as a taste one — see
+ * `internal/transition.ts` — and the taste half is that a button arriving with a
+ * flourish is a page's decision about a region, made with a wrapper, rather than
+ * something every button in the library should carry the tables for.
+ */
+export type MPTransition = MPAnimation | MPTransitionOptions;
+/** What a `transition` takes when a bare effect name is not enough. */
+export interface MPTransitionOptions {
+    /** Which of the seven effects runs. */
+    effect: MPAnimation;
+    /** How long it takes, in milliseconds. Unset takes the effect's own token. */
+    duration?: number;
+    /**
+     * How long before it starts, in milliseconds.
+     * @default 0
+     */
+    delay?: number;
+    /** Which Material easing curve it runs on. */
+    easing?: MPEasing;
+    /**
+     * Whether it fades in as it arrives. Every effect does unless told not to,
+     * because an entrance that only moves reads as a jump.
+     * @default true
+     */
+    fade?: boolean;
+    /** How far a `slide` travels: a CSS length, or a percentage of the element. */
+    distance?: number | string;
+    /**
+     * Which edge a `slide` comes from, or a `reveal` is wiped from.
+     * @default 'bottom' for `slide`, 'left' for `reveal`
+     */
+    from?: MPSide;
+    /** Where a `grow` or a `zoom` starts, as a scale. */
+    scale?: number;
+    /**
+     * How far a `rotate` turns from, in degrees.
+     * @default -180
+     */
+    angle?: number;
+}
 /**
  * What makes an animation run.
  *
@@ -365,3 +831,98 @@ export interface MPAnimateProps {
      */
     threshold?: number;
 }
+/**
+ * Spreading one effect across a set rather than playing it on the box.
+ *
+ * Three props rather than an `MPAnimateStagger` component, and that is the
+ * whole design: a list settling in is not a different effect from a fade, it is
+ * the same fade told when to start. A wrapper component would be a second
+ * spelling of something the effects already do, and a caller would have to
+ * choose between "fade" and "fade one at a time" at the import.
+ *
+ * They belong to the effects that are **one `@keyframes` on the element
+ * itself**, which is these six. `MPAnimateMarquee` duplicates its children,
+ * `MPAnimateHeadline` swaps between them and `MPAnimateTyping` counts their
+ * characters; none of the three can hand an arbitrary child an animation, and
+ * `MPAnimateLighting` puts its movement on a pseudo-element that a child has
+ * no equivalent of. `MPAnimateAppear` is this with `stagger` already on and a
+ * default of 80ms, and runs on the same code.
+ */
+/**
+ * What drives an effect: a clock, or the reader's own scrolling.
+ *
+ * `auto` is `document.timeline` — the animation runs for its `duration` once
+ * something starts it, which is what every effect here has always done.
+ *
+ * `view` hands the animation to CSS's `view()` timeline, so its progress *is*
+ * the element's progress through the scrollport. Scrolling back runs it
+ * backwards, and stopping halfway leaves it halfway: it is the same keyframe,
+ * driven by a position instead of by time.
+ */
+export type MPAnimateTimeline = 'auto' | 'view';
+/**
+ * Scroll-driven playback, on the effects that are one `@keyframes` on the
+ * element itself.
+ *
+ * Two props rather than a component, for the reason the stagger ones are: this
+ * is not a different effect, it is the same effect on a different clock. Every
+ * one of the six gets it, and a component that has to know what its children
+ * *are* gets none of it.
+ */
+export interface MPAnimateTimelineProps {
+    /**
+     * What drives it.
+     *
+     * On `view`, three of the props above stop meaning anything: `duration`,
+     * `delay` and `repeat` are the clock's units and there is no clock. So is
+     * `trigger` — the scroll position *is* the trigger — and the animation is
+     * held `running` for it, because a paused scroll-driven animation shows
+     * nothing at all. An explicit `paused` is still honoured; it is the one of
+     * the four that is a caller saying stop rather than a default.
+     *
+     * A browser without `view()` falls back to the clock, and the effect plays
+     * once the way it always did. Degraded rather than blank, which is the whole
+     * reason the declaration is behind an `@supports`.
+     * @default 'auto'
+     */
+    timeline?: MPAnimateTimeline;
+    /**
+     * Which part of the element's travel through the scrollport the animation is
+     * spread over — any CSS `animation-range`. Only read on `timeline="view"`.
+     *
+     * The default runs from the moment the element's leading edge appears to a
+     * little under halfway across, so it has finished arriving by the time it is
+     * somewhere a reader would be looking at it. A range that ended at the far
+     * edge would leave everything on the page permanently mid-animation.
+     * @default 'entry 0% cover 45%'
+     */
+    range?: string;
+}
+export interface MPAnimateStaggerProps {
+    /**
+     * How long after one child the next one starts, in milliseconds.
+     *
+     * This is what turns the effect into a per-child one. At `0` — the default —
+     * the box itself animates, which is what these components have always done
+     * and what is right for a single subject.
+     *
+     * The step is per **child**, so what you pass matters: eight children are
+     * eight steps and one child holding eight things is one step. That is also
+     * how to opt part of a set out — group it.
+     * @default 0
+     */
+    stagger?: number;
+    /**
+     * How much longer each child takes than the one before it, in milliseconds.
+     * Negative shortens instead, clamped at zero. Only read alongside `stagger`.
+     * @default 0
+     */
+    durationStep?: number;
+    /**
+     * Runs the set from the last child to the first. Only the order is reversed —
+     * each child still plays forwards, which is what separates this from
+     * `mode="out"`.
+     * @default false
+     */
+    reverse?: boolean;
+}