API Nuage de points
Utilisez ce graphique quand la question est « ces deux nombres sont-ils liés ? » - les propriétés et le moteur ci-dessous ; la réponse se trouve dans la démo Nuage de points.
Import
ts
import "@michi-vz/wc/scatter-chart";
// <michi-vz-scatter-chart> is now definedts
import { mountScatterChart } from "@michi-vz/core";
const chart = mountScatterChart(el, props);Propriétés
| Prop | Type | Default | Description |
|---|---|---|---|
dataSet* | ScatterDataPoint[] | - | Array of points to plot in the cloud |
title | string | - | Optional chart title rendered above the plot |
xAxisDataType | "date_annual" | "date_monthly" | "number" | "number" | number / date / band (categorical) for x. y is always linear. |
xAxisFormat | (d: number | string) => string | - | Formats an x tick value into its display label |
yAxisFormat | (d: number | string) => string | - | Formats a y tick value into its display label |
xAxisDomain | [number, number] | string[] | - | Fix the x-axis range ([min,max]); for band x, the ordered category labels (string[]). |
yAxisDomain | [number, number] | - | Fix the y-axis range as [min, max] instead of deriving it from the data |
sizeRange | [number, number] | [4, 20] | [minRadius, maxRadius] px for the size scale (default [4, 20]). |
ticks | number | 5 | Approximate number of axis ticks to generate |
tickValues | Array<number | Date> | - | Explicit tick values, overriding the generated ones |
filter | Filter | - | Top-N / sort filter applied to the data before rendering |
timeline | boolean | TimelinePeriodConfig | - | Opt-in "play through years": snapshots one period at a time over the distinct per-row `date` values, with a headless controller (`chart.timeline()`) plus an optional built-in play button + scrubber. Off by default; requires rows with `date`. The user's `filter` still applies within each period. |
tooltipFormatter | (d: ScatterDataPoint) => string | - | Returns custom tooltip HTML for a hovered datum (sanitized before it is inserted) |
isLoading | boolean | - | Show the loading overlay and skip the no-data check (legacy michi-vz parity). |
isNodata | | boolean | ((dataSet: ScatterDataPoint[] | null | undefined) => boolean) | - | No-data override: boolean, or a predicate on the data; default = empty data. |
noDataLabel | string | - | Text for the vanilla default no-data overlay (ignored when suppressed). |
suppressDefaultOverlay | boolean | - | A framework wrapper sets this to render its OWN loading/no-data node instead. |
onHighlightItem | (labels: string[]) => void | - | Called when the hovered/highlighted label(s) change |
showCrosshair | boolean | false | Show a crosshair overlay tracking the hovered point. Default false. |
crosshairLabels | boolean | false | Render axis-badge value readouts at the crosshair intersection. Default false. |
crosshairLineStyle | "solid" | "dashed" | - | "dashed" → both hover+pinned dashed; "solid" → both solid; undefined → hover dashed. |
crosshairSpan | "full" | "half" | "full" | "full" → both crosshair lines span the whole plot; "half" → an L-shaped arm from the bubble to the two axes. Default "full". |
crosshairLabelPlacement | "auto" | "fixed" | - | "auto" → collision-flip badges; "fixed" → anchor to bottom-left. Default "auto". |
dScaleLegend | { title?: string; valueFormatter?: (d: number) => string } | - | Bubble-size reference legend (half-arc trio + domain labels). |
yTicksQty | number | - | Override the y-axis tick count (scale.ticks(n)); falls back to `ticks`. |
showGrid | boolean | { x?: boolean; y?: boolean } | - | Grid lines per axis: boolean → both; { x?, y? } → per-axis (each defaults true). Default both on. |
pinIcon | boolean | - | Pass false to suppress the sticky-tooltip pin icon. Default: shown (no-op until an icon exists). |
pointLabels | boolean | ScatterPointLabelsConfig | - | Per-point text label (the point's `label` by default). Omitted, or `false`, is a byte-for-byte no-op - zero `.mv-point-label` DOM, default off. Placement is a SIMPLIFIED right-of-point + overlap-hide strategy: each label anchors just to the right of its point (at the current z/draw-order position - see `drawOrder`), and a label is skipped entirely if its estimated bounding box would overlap an already-placed label's box (points processed in that same draw order, so the result never depends on unordered iteration). This intentionally diverges from the legacy sdg-trade Scatterplot (`components/Charts/Scatterplot/Chart.js`), which used d3-voronoi to pick a non-colliding label DIRECTION (right/left/top/ bottom, from the point toward its voronoi cell's centroid) and hid a label only when its voronoi cell's polygon area was <= 10000px². No d3-voronoi dependency was added for this port (none allowed) - the parity bar accepts a cosmetically different placement; the requirement is that labels exist and don't visually collide. Painted on the SVG scaffold layer unconditionally (same treatment as the title/axes and ComparableBar's `deltaIndicator`), so labels look identical whichever `renderer` painted the points themselves. |
drawOrder | "sizeDescending" | "sizeAscending" | "sizeDescending" | Draw-order for overlapping bubbles. Two values, no "off" state - the chart always draws bubbles in SOME order, so the default just names the pre-existing one instead of pretending there's no ordering at all. - `"sizeDescending"` (default, omitted resolves here): today's existing, zero-diff behaviour. `buildScatterRenderModel` has, since before this prop existed, unconditionally sorted points largest-first whenever radii vary, so smaller bubbles paint over larger ones (readability: small bubbles don't get hidden behind big ones). - `"sizeAscending"`: a genuine opt-in that flips the comparator so the LARGEST bubble draws LAST / on top instead. This reproduces the actual legacy sdg-trade z-order, confirmed by reading `sdg-trade/.../Scatterplot/Scatterplot.js` (`data.sort((a,b) => a[rValueKey]-b[rValueKey])` - ascending by raw size) and `.../Scatterplot/Chart.js` (draws circles via `data.map` in that array order; later SVG siblings paint over earlier ones) - net effect: the LARGEST bubble is drawn last and ends up on top there, the opposite of this chart's own default. Use `"sizeAscending"` for true legacy parity. |
svgChildren | string | - | Pre-serialised SVG markup injected as direct <svg> children (the React wrapper fills this from `children`). |
Common props - shared by every chart (14)
| Prop | Type | Default | Description |
|---|---|---|---|
width | number | 900 | Chart width in pixels |
height | number | 480 | Chart height in pixels |
margin | Margin | { top: 50, right: 50, bottom: 50, left: 60 } | Inner margins (top/right/bottom/left, in px) reserved for axes, titles, and labels |
colors | string[] | - | Categorical palette for series/labels without an explicit colour or colorsMapping entry |
colorsMapping | Record<string, string> | - | Explicit label -> colour map; takes precedence over the palette and per-item colours |
highlightItems | string[] | - | Labels to emphasise; all other marks dim |
disabledItems | string[] | - | Labels to hide and exclude from scales/stacks |
renderer | Renderer | "svg" | Render as inline SVG (default) or to a canvas (faster for large datasets); getContext() is identical either way |
locale | string | - | BCP-47 locale used for number and date formatting |
skipColorMappingDispatch | boolean | false | External-CSS mode: unmapped labels resolve to transparent and onColorMappingGenerated is not emitted, so mark colours come from your CSS via the data-label-safe contract |
enableTransitions | boolean | true | Animate updates with CSS transitions (default true) |
onColorMappingGenerated | (mapping: Record<string, string>) => void | - | Called with the resolved label -> colour map after the chart assigns colours |
onChartDataProcessed | (context: ChartContext) => void | - | Called with the renderer-agnostic ChartContext whenever the data is (re)processed |
onDataWarning | (warnings: DataWarning[]) => void | - | Called with any non-fatal data warnings (duplicate labels, non-finite values, non-monotonic dates, ...) |
Événements
Le composant web émet ces CustomEvents à bouillonnement (bubbling) (le moteur expose les mêmes via les callbacks on* du tableau ci-dessus) :
| Événement | Détail | Se déclenche quand |
|---|---|---|
michi-vz:highlight | string[] | la surbrillance au survol change |
michi-vz:colormapping | Record<string, string> | un mapping de couleurs est généré |
michi-vz:dataprocessed | ChartContext | les données sont (re)traitées |
michi-vz:datawarning | DataWarning[] | des avertissements sur les données d'entrée sont détectés |
getContext()
mountScatterChart(el, props).getContext() retourne un ScatterChartContext indépendant du moteur de rendu (statistiques structurées + un résumé déterministe en langage naturel + une table d'accessibilité). Voir Contexte LLM.
Source
Les propriétés sont typées comme ScatterChartProps dans @michi-vz/core.