flutter-chart（Flutter chart 命令行）

Flutter Chart with Graphic

A skill for helping users build data visualizations in Flutter using the Graphic library — a Grammar of Graphics-based charting library.

When to Use

• User wants to create any type of chart in Flutter (bar, line, area, pie, scatter, heatmap, etc.)
• User needs help configuring chart appearance, interactivity, or animations
• User asks about Graphic library APIs or usage patterns
• User wants to convert data into visual representations
• User wants to create custom, bespoke visualizations — Graphic's customization system is its most powerful feature and a key strength of AI-assisted development

Why Custom Charts?

Standard chart types (bar, line, pie) only cover a fraction of data visualization needs. Graphic is built on Grammar of Graphics theory, which means every visual layer is independently customizable:

• Custom Shapes — Render any geometry (triangles, lollipops, arrows, gauges, bullet charts, sparklines, etc.)
• Custom Tooltips — Fully custom interactive overlays with any layout
• Custom Annotations — Draw any graphics at data or absolute positions
• Custom Encoders — Map data to visuals using arbitrary logic
• Custom Modifiers — Define custom collision/arrangement behavior

This makes AI-assisted development especially valuable: the AI can write custom shape renderers, coordinate math, and drawing code that would be tedious to implement manually.

Always consider customization when the user's requirements don't perfectly match a standard chart type. See references/customization.md for the comprehensive customization guide.

Quick Start

Installation

Add to pubspec.yaml:

dependencies:
  graphic: ^latest

Then run flutter pub get.

Import

import 'package:graphic/graphic.dart';

Minimal Example

Chart(
  data: [
    {'category': 'A', 'value': 10},
    {'category': 'B', 'value': 20},
    {'category': 'C', 'value': 15},
  ],
  variables: {
    'category': Variable(
      accessor: (Map map) => map['category'] as String,
    ),
    'value': Variable(
      accessor: (Map map) => map['value'] as num,
    ),
  },
  marks: [IntervalMark()],
  axes: [Defaults.horizontalAxis, Defaults.verticalAxis],
)

Core Concepts

Graphic follows the Grammar of Graphics theory. A chart is composed of independent, declarative layers:

data → Variable → Scale → Encode → Mark → Shape → Render

Everything is configured through the single Chart<D> widget constructor. There are no imperative APIs — all configuration is declarative.

The Chart Widget

Chart<D> is the only public widget. The type parameter D is the data item type. Key parameters:

See references/chart-widget.md for full parameter details.

Variables

Variables define how raw data maps to abstract values:

variables: {
  'date': Variable(
    accessor: (MyData d) => d.date,
    scale: TimeScale(formatter: (t) => DateFormat.MMMd().format(t)),
  ),
  'value': Variable(
    accessor: (MyData d) => d.value,
    scale: LinearScale(min: 0),
  ),
}

Each variable can have a Scale that controls domain-to-range mapping. See references/scales.md.

Marks

Marks are the geometric elements that represent data:

See references/marks.md for parameters and references/shapes.md for shape options.

Encodes (Aesthetic Mappings)

Encodes map data values to visual properties. Every encode supports three modes:

1. Fixed value: ColorEncode(value: Colors.blue)
2. Variable mapping: ColorEncode(variable: 'type', values: Defaults.colors10)
3. Custom function: ColorEncode(encoder: (tuple) => myColorLogic(tuple))

Available encodes: ColorEncode, SizeEncode, ShapeEncode, LabelEncode, GradientEncode, ElevationEncode.

See references/encodes.md for details.

Position Algebra (Varset)

Position is specified using Varset algebra with three operators:

// x=date, y=value, grouped by type
position: Varset('date') * Varset('value') / Varset('type')

See references/algebra.md for details.

Coordinates

• RectCoord — Cartesian coordinates (default). Supports transposed, horizontalRange, verticalRange.
• PolarCoord — Polar/radial coordinates for pie charts, radar charts, rose charts. Supports startAngle, endAngle, startRadius, endRadius.

See references/coordinates.md for details.

Interaction

Define named selections and use them in encode updaters:

selections: {
  'tap': PointSelection(dim: Dim.x),
},
marks: [
  IntervalMark(
    color: ColorEncode(
      value: Colors.blue,
      updaters: {
        'tap': {
          true: (color) => color.withAlpha(255),   // selected
          false: (color) => color.withAlpha(100),  // not selected
        },
      },
    ),
  ),
],

See references/selections.md for selection types and gesture configuration.

Modifiers

Modifiers handle geometry collision/arrangement:

• StackModifier() — Stack elements (stacked bar, stacked area)
• DodgeModifier() — Place side by side (grouped bar)
• JitterModifier() — Random scatter (strip plot)
• SymmetricModifier() — Center symmetrically (stream graph)

See references/modifiers.md.

Animation

IntervalMark(
  transition: Transition(duration: Duration(seconds: 1), curve: Curves.easeOut),
  entrance: {MarkEntrance.y},  // Animate from y=0
  tag: (tuple) => tuple['id'].toString(),  // Element matching for transitions
)

See references/animation.md.

Common Chart Recipes

Bar Chart

marks: [IntervalMark()]
coord: RectCoord()  // default

Horizontal Bar Chart

marks: [IntervalMark()]
coord: RectCoord(transposed: true)

Grouped Bar Chart

marks: [
  IntervalMark(
    position: Varset('x') * Varset('y') / Varset('group'),
    color: ColorEncode(variable: 'group', values: Defaults.colors10),
    modifiers: [DodgeModifier()],
  ),
]

Stacked Bar Chart

marks: [
  IntervalMark(
    position: Varset('x') * Varset('y') / Varset('group'),
    color: ColorEncode(variable: 'group', values: Defaults.colors10),
    modifiers: [StackModifier()],
  ),
]

Line Chart

marks: [LineMark()]

Smooth Line Chart

marks: [
  LineMark(shape: ShapeEncode(value: BasicLineShape(smooth: true))),
]

Area Chart

marks: [AreaMark()]

Pie Chart

transforms: [Proportion(variable: 'value', as: 'percent')],
marks: [
  IntervalMark(
    position: Varset('percent') / Varset('category'),
    color: ColorEncode(variable: 'category', values: Defaults.colors10),
    modifiers: [StackModifier()],
  ),
],
coord: PolarCoord(transposed: true, dimCount: 1),

Scatter Plot

marks: [
  PointMark(
    size: SizeEncode(variable: 'magnitude', values: [2, 20]),
    color: ColorEncode(variable: 'type', values: Defaults.colors10),
  ),
]

Rose Chart

marks: [IntervalMark(color: ColorEncode(variable: 'name', values: Defaults.colors10))],
coord: PolarCoord(startRadius: 0.15),

See references/examples.md for more complete examples.

Custom Chart Development

Graphic's greatest strength is its fully customizable rendering pipeline. When standard chart types don't meet requirements, create custom visualizations by implementing your own shapes, tooltips, annotations, and encoders.

Custom Shapes — The Core Extension Point

Create entirely new chart geometries by extending a Shape base class and implementing drawGroupPrimitives():

class LollipopShape extends IntervalShape {
  LollipopShape({this.radius = 6});
  final double radius;

  @override
  List<MarkElement> drawGroupPrimitives(
    List<Attributes> group, CoordConv coord, Offset origin,
  ) {
    final rst = <MarkElement>[];
    for (var item in group) {
      if (item.position.any((p) => !p.dy.isFinite)) continue;
      final style = getPaintStyle(item, false, 0, null, null);
      final base = coord.convert(item.position[0]);
      final tip = coord.convert(item.position[1]);

      // Stem line
      rst.add(PolylineElement(
        points: [base, tip],
        style: PaintStyle(strokeColor: style.fillColor, strokeWidth: 2),
        tag: item.tag,
      ));
      // Circle head
      rst.add(CircleElement(
        center: tip, radius: radius, style: style, tag: item.tag,
      ));
    }
    return rst;
  }

  @override
  bool equalTo(Object other) =>
    other is LollipopShape && radius == other.radius;
}

Available base classes: IntervalShape, LineShape, AreaShape, PointShape, PolygonShape

Available drawing primitives: RectElement, CircleElement, PolygonElement, PolylineElement, ArcElement, SectorElement, SplineElement, PathElement, LabelElement, GroupElement

Custom Tooltip Renderer

tooltip: TooltipGuide(
  renderer: (Size size, Offset anchor, Map<int, Tuple> selected) {
    final t = selected.values.first;
    return [
      RectElement(
        rect: Rect.fromCenter(center: anchor, width: 100, height: 36),
        borderRadius: BorderRadius.circular(6),
        style: PaintStyle(fillColor: Colors.black87, elevation: 4),
      ),
      LabelElement(
        text: '${t['name']}: ${t['value']}',
        anchor: anchor,
        style: LabelStyle(
          textStyle: TextStyle(color: Colors.white, fontSize: 12),
          align: Alignment.center,
        ),
      ),
    ];
  },
)

Custom Encoder Functions

Every encode supports arbitrary logic via encoder:

color: ColorEncode(encoder: (tuple) {
  final v = tuple['value'] as num;
  return v > 100 ? Colors.red : v > 50 ? Colors.orange : Colors.green;
}),

label: LabelEncode(encoder: (tuple) => Label(
  '${tuple['value']}%',
  LabelStyle(textStyle: TextStyle(
    fontSize: (tuple['value'] as num) > 50 ? 14 : 10,
    fontWeight: FontWeight.bold,
  )),
)),

Key Classes for Custom Development

See references/customization.md for the comprehensive guide including coordinate handling, all drawing primitives, custom annotations, custom modifiers, and best practices.

Guides & Annotations

• Axes: Use Defaults.horizontalAxis, Defaults.verticalAxis for quick setup, or customize with AxisGuide. See references/guides.md.
• Tooltip: TooltipGuide() with optional custom renderer. See references/guides.md.
• Crosshair: CrosshairGuide(). See references/guides.md.
• Annotations: LineAnnotation, RegionAnnotation, TagAnnotation, CustomAnnotation. See references/annotations.md.

Styling

• PaintStyle — Fill/stroke colors, gradients, dash patterns, shadows
• LabelStyle — Text style, alignment, rotation, offset
• Defaults — Built-in color palettes (colors10, colors20), preset axes, default styles

See references/styling.md.

Event Streams

Charts expose StreamController parameters for external event coupling:

• gestureStream — Send/receive gesture events
• resizeStream — React to resize events
• changeDataStream — React to data change events
• selectionStream (on Mark) — Programmatically set selections

Dynamic Data

Simply update the data parameter via setState():

setState(() {
  data = newData;
});

The chart automatically re-renders with transition animations when tag is set.

Important Notes

• The Chart widget is the only public widget — all configuration is via its constructor
• color and gradient on encodes are mutually exclusive
• Pie charts require Proportion transform + PolarCoord(transposed: true, dimCount: 1)
• Use tag on marks for smooth transition animations between data states
• Varset / (nest) operator is required for grouping (multi-series, stacked, dodged)
• Function-typed properties are always treated as "unchanged" in equality comparisons

Reference Files

When answering user questions, read the source code for the most accurate and up-to-date API details. The library uses extensive code comments as documentation. Key source directories:

• lib/src/chart/ — Chart widget
• lib/src/mark/ — Mark types
• lib/src/encode/ — Encode types
• lib/src/shape/ — Shape implementations
• lib/src/scale/ — Scale types
• lib/src/coord/ — Coordinate systems
• lib/src/guide/ — Axis, Tooltip, Crosshair, Annotations
• lib/src/interaction/ — Selection, gestures
• lib/src/variable/ — Variable and transforms
• lib/src/algebra/ — Varset algebra
• lib/src/common/ — Shared types (Label, PaintStyle, Defaults)