Using Annotations - nteract/semiotic GitHub Wiki

Annotations provide enormous value to charts and they're less of a challenge now that solutions like react-annotation are available. All frames have some shared annotation handling capabilities that let you deploy annotations easily and dynamically and even do some rudimentary label adjustment in crowded charts.

Further Reading

Making Annotations First-Class Citizens in Data Visualization

Built-in SVG Annotation Processing

Each frame can take an array of objects with a type and other values passed to annotations.

Each kind of frame has a few built-in supported annotation types and also each frame will process a generic react-annotation type or a function passed as a type, which is assumed to be a particular type of react-annotation (such as AnnotationCallout or AnnotationBadge).

All Frames

  • react-annotation: Expects a data object that matches the structure of the data you've sent to the frame that is also structured according to the react-annotation spec. The x and y values will be overwritten with x & y values for the data values of the object e.g. if all of the data you send depends on a date property to determine the x position, the annotation's object should also have a date property if you want it scaled with the rest of the data.
  • function: Any function you send as a type of the annotation will be expected to be one of the react-annotation types.

TODO: add in eventListener functionality

XYFrame

  • xy: Has the data values corresponding to the x and y accessors of the frame, and will put a labeled circle with the label determined by the label value of the object.
  • x: Has a data value corresponding to the x accessor of the frame, and will put a labeled line with the label determined by the label value of the object.
  • y: Has a data value corresponding to the y accessor of the frame, and will put a labeled line with the label determined by the label value of the object.
  • bounds: Has a bounds property of format [{left-x, top-y},{right-x, bottom-y}] and will draw a rectangle around that bounding box.
  • line: Has a coordinates array with two objects in the data format of the frame and will draw a line from the first point to the second.
  • area: Has a coordinates array with objects in the data format of the frame and will draw an area with the coordinates in that array.
  • enclose: Has a coordinates array with objects in the data format of the frame and will draw a minimum bounding circle around the points in that array.

Example:

import { AnnotationCalloutElbow } from 'react-annotation'

const variousAnnotations = [
    { type: 'react-annotation', x: 4, y: 300, dx: -30, dy: 0, note: { title: 'Note at 4,300' }, subject: { text: 'A', radius: 12 } },
    { type: AnnotationCalloutElbow, x: 7, id: 'linedata-1', dx: 30, dy: -50, note: { title: 'linedata-1 at 7' }, subject: { text: 'C', radius: 12 } },
    { type: 'xy', x: 2, y: 2, label: 'Simply XY Annotation' },
    { type: 'x', x: 2, label: 'Simply X Threshold' },
    { type: 'y', y: 2, label: 'Simply Y Threshold' },
    { type: 'enclose', coordinates: [ {x: 3, y: 4}, {x: 5, y: 3} ], label: 'An enclosure' }
]

<XYFrame
   annotations={variousAnnotations}
/>

ORFrame

TODO

NetworkFrame

  • node: Has an id value that corresponds to the idAccessor value of a node, and will draw a circle around that node with the label determined by the label value of the annotation object.
  • enclose: Has an ids array with strings corresponding to values returned by the idAccessor value of nodes in this graph and will draw a minimum bounding circle around the nodes found with a label specified in the label value of the annotation object.

Custom Annotation Rules

You can send objects with any type value to the annotations array but if they don't correspond to any built-in types, they won't display.

You can write custom rules to handle new types and the built-in types to override the default behavior.

Use svgAnnotationRules or htmlAnnotationRules depending on the type of JSX elements you want to create. Each of these functions is called on every item in the annotations array sent to the Frame. If hoverAnnotations is true semiotic will append hover annotations such as frame-hover or column-hover to the array of annotations each function is run against.

Passed values

Because it's hard to tell all the different things you might want to do with an annotation, the custom rules expose not only the datapoint hovered, but also the scales and other properties of the frame.

  • NetworkFrame: d, i, networkFrameProps, networkFrameState, nodes, edges
  • XYFrame: d, i, screenCoordinates, xScale, yScale, xAccessor, yAccessor, xyFrameProps, xyFrameState, areas, points, lines
  • ORFrame: d, i, oScale, rScale, oAccessor, rAccessor, orFrameProps, orFrameState

svgAnnotationRules { function }

This function is run on every item in the annotation array, in addition to any hover annotations added by the Frame, it should return:

  • an SVG element in JSX
  • null if you want it to process all default types in addition to your custom rules
  • return false if you only want to process your custom rules

TODO: start with a simpler example, you're introducing 'parentLine' and 'coincidentPoints' all in this one example without the context they need. For example parentLine only shows up on the annotations that are hover annotations. Something that isn't clear as a first time reader of the docs.

The example below shows how to create a parent line based on all lines that cross through the hovered point. Notice it checks to see if the annotation has a parentLine prop--you can use any test you want, whether it's of the type or like this. It returns null otherwise so that it doesn't eat other built-in annotations. It also uses the coincidentPoints property of a hovered element, which is an array of all points being hovered over, which is useful for charts that have overlapping lines on the same point.

import { line } from 'd3-shape'
<XYFrame
  svgAnnotationRules ={({ d, xScale, yScale }) => {
    if (!d.parentLine) {
        return null
    }
    const lineRenderer = line().x(d => xScale(d._xyfX)).y(d => yScale(d._xyfY)).curve(curveMonotoneX)

    return d.coincidentPoints.map((p,q) => {
        const lineD = lineRenderer(p.parentLine.data)
        return <path key={`hover-line-${q}`} d={lineD} style={{ fill: "none", stroke: "black", strokeWidth: 3 }} 
    />
    })
  }
/>

In contrast this is a custom svg rule for drawing points on all lines at the hovered x position. The _xyfX value is the calculated XYFrame x value of the point (but not the scaled value, it still needs to be passed to a scale to get the pixel position). Again, it returns null otherwise so it doesn't eat other annotations that don't match. The lines array is a projected array of lines data, which makes it easy to get the datapoints that correspond to the x position of the hovered datapoint. It returns an array of <circle> elements.

<XYFrame
  svgAnnotationRules ={({ d, lines, xScale, yScale }) => {
    if (d && d._xyfX) {
        const thesePoints = lines.map(line => {
            return line.data.find(p => p._xyfX === d._xyfX)
        })

        return thesePoints.map(p => {
            return <circle r={2} style={{ fill: "red" }} cx={xScale(p._xyfX)} cy={yScale(p._xyfY)} />
        })
    }
    return null
  }
/>

htmlAnnotationRules { function }

This function is run on every item in the annotation array, in addition to any hover annotations added by the Frame, it should return:

  • an HTML element in JSX
  • null if you want it to process all default types in addition to your custom rules
  • return false if you only want to process your custom rules

TODO: this statement needs more context If you want to place it over the hovered element on the chart, the x and y position corresponds to an absolute positioned HTML element with left equal to x and top equal to y.

TODO: add example using frame-hover-column-hover

⚠️ **GitHub.com Fallback** ⚠️