@data-ui/xy-chart v0.0.84

@data-ui/xy-chart

A package that supports making charts with x- and y- cartesian coordinates.

See it live at williaster.github.io/data-ui.

Example usage

The React <XYChart /> container coordinates scales across its children and is composable. You can pass it <XAxis />, <YAxis />, one or more <*Series /> components, and <defs>-based components such as <LinearGradients />s and <PatternLines />.

Note that the order of children passed to <XYChart /> determines their rendering order, for example the a <LineSeries /> passed after a <BarSeries /> will overlay the line on the bars. The same applies to axes.

import { XYChart, BarSeries, CrossHair, XAxis, YAxis, LinearGradient } from '@data-ui/xy-chart';

/// ...
<XYChart
  ariaLabel="Bar chart showing ..."
  width={width}
  height={height}
  margin={{ top, right, bottom, left }}
  xScale={{ type: 'time' }}
  yScale={{ type: 'linear' }}
  renderTooltip={({ event, datum, data, color }) => (
    <div>
      <strong style={{ color }}>{datum.label}</strong>
      <div>
        <strong>x </strong>
        {datum.x}
      </div>
      <div>
        <strong>y </strong>
        {datum.y}
      </div>
    </div>
  )}
  snapTooltipToDataX
>
  <LinearGradient id="my_fancy_gradient" from={startColor} to={endColor} />
  <XAxis label="X-axis Label" />
  <YAxis label="Y-axis Label" />
  <BarSeries data={timeSeriesData} fill="url('#my_fancy_gradient')" />
  <CrossHair showHorizontalLine={false} fullHeight stroke="pink" />
</XYChart>;

Components

Check out the example source code and PropTable tabs in the Storybook williaster.github.io/data-ui for more!

<XYChart />

The XYChart renders an <svg /> and coordinates scales across all of its child series and axes. It takes the following props

Scale config

X and y-scales are configured using xScale and yScale config props which essentially configure d3/vx scales:

const scaleConfigShape = PropTypes.shape({
  type: PropTypes.oneOf(['time', 'timeUtc', 'linear', 'band', 'ordinal']).isRequired,
  includeZero: PropTypes.bool,

  // these would override any computation done by XYChart, allowing specific ranges or colors
  // see storybook for more examples
  range: PropTypes.arrayOf(PropTypes.oneOfType([PropTypes.number, PropTypes.string])),
  rangeRound: PropTypes.arrayOf(PropTypes.oneOfType([PropTypes.number, PropTypes.string])),
  domain: PropTypes.arrayOf(PropTypes.oneOfType([PropTypes.number, PropTypes.string])),
});

Entries in scale objects are shallow checked so new objects don't trigger re-renders.

<XAxis /> and <YAxis />

Series

Several types of series types are exported by the package, and can be used in combination. See the storybook source for more proptables for your series of interest. Here is an overview of scale support and data shapes:

* The y boundaries of the <AreaSeries/> may be specified by either

• defined y0 and y1 values or
• a single y value, in which case its lower bound is set to 0 (a "closed" area series)

Series labels

The <PointSeries /> and <BarSeries /> components support rendering labels per-datum via the renderLabel and defaultLabelProps props.

• by default, if a datum has a label property, it will have a label rendered out of the box using the @vx/text <Text /> component (which wraps svg text, etc.). labels are always rendered on top of the Bars and Points themeselves.
• The label has "smart" default aesthetics (taking from the @data-ui theme), text anchors, and wrapping behavior, but you can override them by setting defaultLabelProps to your own object. By default these props are passed to the underlying <Text /> label component, and d.label is rendered as the child
• to support full label customization, you may define a renderLabel function with the signature ({ datum, index, labelProps }) => node. labelProps includes all values from defaultLabelProps as well as "smart" default values for width, x, y, dx, dy, verticalAnchor, and textAnchor based on Bar and Point position, size, and orientation (horizontal vs vertical).
• Example usage:

<BarSeries
  {...restProps}
  renderLabel={({ datum, labelProps, index: i }) =>
    datum.label ? (
      <Text {...labelProps} fill={datum.selected ? COLOR_2 : COLOR_1}>
        {datum.label}
      </Text>
    ) : null
  }
/>

<CirclePackSeries />

This series implements the Circle packing algorithm described by Wang et al. Visualization of large hierarchical data by circle packing, but attempts to preserve datum x values (although they may be modified slightly). It is useful for visualizing e.g., atomic events where x values may partially overlap, and provides an alternative to an atomic histogram without a requirement for binning x values. Alternatively, users can pass their own layout algorithm as the value of prop layout (one example is included in the demo package.)

Note that only x values are needed for CirclePackSeries, y values are computed based on x and size (if specified). Similar to PointSeries, size, fill, and fillOpacity may be set on datum themseleves or passed as props to the CirclePackSeries component.

<AreaDifferenceSeries />

This series has a different API from other series in that it wraps two AreaSeries (see the storybook example for more details):

<AreaDifferenceSeries>
  <AreaSeries data={data1} {...moreAreaSeriesProps} />
  <AreaSeries data={data2} {...moreAreaSeriesProps} />
</AreaDifferenceSeries>

The result will show the difference between the two AreaSeries, with a fill that matches the AreaSeries with the greater y-value.

Reference Lines

<HorizontalReferenceLine /> and <VerticalReferenceLine /> are available for chart annotations with the following usage pattern:

<XYChart ...>
  <LineSeries data={[ { x: new Date('2018-01-01'), y: 10 }, ... ]} />
  {/* Wraps text within width of 100px */}
  <HorizontalReferenceLine
    reference={25}
    stroke="magenta"
    label="My y-threshold"
    labelProps={{ width: 100, verticalAnchor: 'middle' }}
  />
  <VerticalReferenceLine
    reference={new Date('2018-01-05')}
    label="My birthday"
    labelProps={{ width: 100, textAnchor: 'start', dx: '0.5em' }}
  />
</XYChart>

The both take the following props

Tooltips, Mouse Events, and Triggers

Tooltips

Tooltips are supported for all series types, but how you trigger and configure them triggers you want, will likely depend on which series combinations you're using and how much customization you need. The easiest way to use tooltips out of the box is by passing a renderTooltip function to <XYChart /> as shown in the above example. This function takes an object with the shape { datum, event [, coords [, data, [, color [, series [, seriesKey]]]]] } (see function signatures section below for more) as input and should return the inner contents of the tooltip (not the tooltip container!) as shown above. You may snap tooltips to data x and y values by setting snapTooltipToDataX and/or snapTooltipToDataY to true on XYChart.

Under the covers this will wrap the <XYChart /> component in the exported <WithTooltip /> HOC, which wraps the <svg /> in a <div /> and handles the positioning and rendering of an HTML-based tooltip with the contents returned by renderTooltip(). This tooltip is aware of the bounds of its container and should position itself "smartly".

If you'd like more customizability over tooltip rendering you can do either of the following:

1. Roll your own tooltip positioning logic and pass onMouseMove and onMouseLeave functions to XYChart. These functions are triggered according to the eventTrigger prop and are called with the signature described below upon appropriate trigger. Note that you must also pass tooltipData to XYChart if you are using the CrossHair component, which has an expected shape of { datum [, series] } containing the datum(s) to emphasize.

2. Wrap <XYChart /> with <WithTooltip /> yourself, which accepts props for additional customization:

Note that to correctly position a tooltip, the <WithTooltip /> onMouseMove function minimally requires an event or coords object of the form { x: Number, y: Number }. If coords is specified it takes precedent over any position computed from the event. See function signatures below for more.

Accessibility

Note that unless disableMouseEvents=true, most series currently invoke onMouseMove and onMouseLeave when focused and blured, respectively, so that tooltips are accessible for keyboard-only users. Support for these events is reflected in the SeriesComponent table above.

<CrossHair />

The <CrossHair /> component may be used in combination with tooltips for additional visual feedback (see the storybook for many examples!). Simply pass the component as a child of <XYChart /> and it will automatically position itself upon tooltip trigger. Compared to a tooltip, this component snaps to actual data points for improved precision. It accepts the following props:

Mouse Events & Triggers

XYChart has hooks for mousemove, mouseleave, and click events that can be triggered at different levels as specified by the eventTrigger prop:

eventTrigger='series'

For the series event trigger, XYChart will pass along event handlers to its child series unless a series has disableMouseEvents set to true, and any event handlers defined at the series level will override those defined at the XYChart level. Series-level events are triggered by interactions with the series DOM elements themselves.

eventTrigger='container'

For the container event trigger, the XYChart container will intercept all mouse events and event handlers will be called with all datums nearest the hovered x value. This type of event trigger is useful if you want to implement a shared tooltip. Note that data passed to series should be sorted by x-value for this to work correctly.

eventTrigger='voronoi'

For series components that have "small" mouse areas, such as PointSeries and LineSeries, you may opt to use an invisible Voronoi overlay on top of the visualization to increase the target area of interaction sites and improve user experience. To view or debug a voronoi you may set the convenience prop showVoronoi to true. Note that this will compute a voronoi layout for all data points across all series.

Note ‼️

It is worth noting that voronoi overlays require a defined y attribute, so use of voronoi with only y0 and y1 values will not work (this is reflected in the compatibility table above).

Additionally, because of the polygonal shapes generated by the voronoi layout, you probably don't want to use this option if you are e.g., only rendering a BarSeries because the bar points represent the tops of the bars and thus polygons for one bar may overlap the rect of another bar (again, you may use showVoronoi to debug this).

Functions and Function Signatures

XYChart and all series support onMouseMove, onMouseLeave, and onClick event handlers with the following signatures:

onMouseMove({ datum, event [, coords [, data, [, color [, series [, seriesKey]]]]] })
onClick({ datum, event [, coords [, data, [, color [, series [, seriesKey]]]]] })
onMouseLeave()

A seriesKey is passed when eventTrigger=series for <StackedAreaSeries />, <StackedBarSeries />, or <GroupedBarSeries />. It corresponds to the relevant stackKey or groupKey that triggered the event.

series is passed when eventTrigger=container and represents an object of datums across all series components nearest the current mouse x. The closest datum across all series components is passed as datum in the function signature. Within the series object, datums are keyed on the seriesKey prop set on the series component itself. similar to React, if seriesKey is not set its index as a child of XYChart will be used which is more error prone

coords is an object of the form { x: Number, y: Number }. XYChart passes x and y only if snapTooltipToDataX or snapTooltipToDataY are true, respectively.

Programmatically triggering tooltips

XYChart exposes hooks to manually trigger any of these handlers with the eventTriggerRefs prop. Similar to React refs, this prop is a callback function that is called by XYChart after mounting. The callback receives an object as input, with keys corresponding to the event type names and respective handlers as values: eventTriggerRefs({ click, mousemove, mouseleave }). The ref handlers have the same signatures as defined above.

Note that snapTooltipToData* props will still have an effect when events are triggered this way.

Theme

A theme object with the following shape can be passed to <XYChart /> to style the chart, axes, and series. See @data-ui/theme for an example.

export const themeShape = PropTypes.shape({
  gridStyles: PropTypes.shape({
    stroke: PropTypes.string,
    strokeWidth: PropTypes.number,
  }),
  xAxisStyles: PropTypes.shape({
    stroke: PropTypes.string,
    strokeWidth: PropTypes.number,
    label: PropTypes.shape({
      bottom: PropTypes.object,
      top: PropTypes.object,
    }),
  }),
  yAxisStyles: PropTypes.shape({
    stroke: PropTypes.string,
    strokeWidth: PropTypes.number,
    label: PropTypes.shape({
      left: PropTypes.object,
      right: PropTypes.object,
    }),
  })
  xTickStyles: PropTypes.shape({
    stroke: PropTypes.string,
    tickLength: PropTypes.number,
    label: PropTypes.shape({
      bottom: PropTypes.object,
      top: PropTypes.object,
    }),
  }),
  yTickStyles: PropTypes.shape({
    stroke: PropTypes.string,
    tickLength: PropTypes.number,
    label: PropTypes.shape({
      left: PropTypes.object,
      right: PropTypes.object,
    }),
  }),
});

More on the way.

Other

• <PatternLines />
• <PatternCircles />
• <PatternWaves />
• <PatternHexagons />
• <LinearGradient />

These vx gradients and patterns are exported in @data-ui/xy-chart to customize the style of series. These components create <defs> elements in the chart SVG with ids that you can reference in another component. See the storybook for example usage!

Development

npm install
yarn run dev # or 'build'