| // Copyright 2013 The Flutter Authors. All rights reserved. | |
| // Use of this source code is governed by a BSD-style license that can be | |
| // found in the LICENSE file. | |
| part of dart.ui; | |
| /// Base class for [Size] and [Offset], which are both ways to describe | |
| /// a distance as a two-dimensional axis-aligned vector. | |
| abstract class OffsetBase { | |
| /// Abstract const constructor. This constructor enables subclasses to provide | |
| /// const constructors so that they can be used in const expressions. | |
| /// | |
| /// The first argument sets the horizontal component, and the second the | |
| /// vertical component. | |
| const OffsetBase(this._dx, this._dy); | |
| final double _dx; | |
| final double _dy; | |
| /// Returns true if either component is [double.infinity], and false if both | |
| /// are finite (or negative infinity, or NaN). | |
| /// | |
| /// This is different than comparing for equality with an instance that has | |
| /// _both_ components set to [double.infinity]. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [isFinite], which is true if both components are finite (and not NaN). | |
| bool get isInfinite => _dx >= double.infinity || _dy >= double.infinity; | |
| /// Whether both components are finite (neither infinite nor NaN). | |
| /// | |
| /// See also: | |
| /// | |
| /// * [isInfinite], which returns true if either component is equal to | |
| /// positive infinity. | |
| bool get isFinite => _dx.isFinite && _dy.isFinite; | |
| /// Less-than operator. Compares an [Offset] or [Size] to another [Offset] or | |
| /// [Size], and returns true if both the horizontal and vertical values of the | |
| /// left-hand-side operand are smaller than the horizontal and vertical values | |
| /// of the right-hand-side operand respectively. Returns false otherwise. | |
| /// | |
| /// This is a partial ordering. It is possible for two values to be neither | |
| /// less, nor greater than, nor equal to, another. | |
| bool operator <(OffsetBase other) => _dx < other._dx && _dy < other._dy; | |
| /// Less-than-or-equal-to operator. Compares an [Offset] or [Size] to another | |
| /// [Offset] or [Size], and returns true if both the horizontal and vertical | |
| /// values of the left-hand-side operand are smaller than or equal to the | |
| /// horizontal and vertical values of the right-hand-side operand | |
| /// respectively. Returns false otherwise. | |
| /// | |
| /// This is a partial ordering. It is possible for two values to be neither | |
| /// less, nor greater than, nor equal to, another. | |
| bool operator <=(OffsetBase other) => _dx <= other._dx && _dy <= other._dy; | |
| /// Greater-than operator. Compares an [Offset] or [Size] to another [Offset] | |
| /// or [Size], and returns true if both the horizontal and vertical values of | |
| /// the left-hand-side operand are bigger than the horizontal and vertical | |
| /// values of the right-hand-side operand respectively. Returns false | |
| /// otherwise. | |
| /// | |
| /// This is a partial ordering. It is possible for two values to be neither | |
| /// less, nor greater than, nor equal to, another. | |
| bool operator >(OffsetBase other) => _dx > other._dx && _dy > other._dy; | |
| /// Greater-than-or-equal-to operator. Compares an [Offset] or [Size] to | |
| /// another [Offset] or [Size], and returns true if both the horizontal and | |
| /// vertical values of the left-hand-side operand are bigger than or equal to | |
| /// the horizontal and vertical values of the right-hand-side operand | |
| /// respectively. Returns false otherwise. | |
| /// | |
| /// This is a partial ordering. It is possible for two values to be neither | |
| /// less, nor greater than, nor equal to, another. | |
| bool operator >=(OffsetBase other) => _dx >= other._dx && _dy >= other._dy; | |
| /// Equality operator. Compares an [Offset] or [Size] to another [Offset] or | |
| /// [Size], and returns true if the horizontal and vertical values of the | |
| /// left-hand-side operand are equal to the horizontal and vertical values of | |
| /// the right-hand-side operand respectively. Returns false otherwise. | |
| @override | |
| bool operator ==(Object other) { | |
| return other is OffsetBase | |
| && other._dx == _dx | |
| && other._dy == _dy; | |
| } | |
| @override | |
| int get hashCode => Object.hash(_dx, _dy); | |
| @override | |
| String toString() => 'OffsetBase(${_dx.toStringAsFixed(1)}, ${_dy.toStringAsFixed(1)})'; | |
| } | |
| /// An immutable 2D floating-point offset. | |
| /// | |
| /// Generally speaking, Offsets can be interpreted in two ways: | |
| /// | |
| /// 1. As representing a point in Cartesian space a specified distance from a | |
| /// separately-maintained origin. For example, the top-left position of | |
| /// children in the [RenderBox] protocol is typically represented as an | |
| /// [Offset] from the top left of the parent box. | |
| /// | |
| /// 2. As a vector that can be applied to coordinates. For example, when | |
| /// painting a [RenderObject], the parent is passed an [Offset] from the | |
| /// screen's origin which it can add to the offsets of its children to find | |
| /// the [Offset] from the screen's origin to each of the children. | |
| /// | |
| /// Because a particular [Offset] can be interpreted as one sense at one time | |
| /// then as the other sense at a later time, the same class is used for both | |
| /// senses. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [Size], which represents a vector describing the size of a rectangle. | |
| class Offset extends OffsetBase { | |
| /// Creates an offset. The first argument sets [dx], the horizontal component, | |
| /// and the second sets [dy], the vertical component. | |
| const Offset(super.dx, super.dy); | |
| /// Creates an offset from its [direction] and [distance]. | |
| /// | |
| /// The direction is in radians clockwise from the positive x-axis. | |
| /// | |
| /// The distance can be omitted, to create a unit vector (distance = 1.0). | |
| factory Offset.fromDirection(double direction, [ double distance = 1.0 ]) { | |
| return Offset(distance * math.cos(direction), distance * math.sin(direction)); | |
| } | |
| /// The x component of the offset. | |
| /// | |
| /// The y component is given by [dy]. | |
| double get dx => _dx; | |
| /// The y component of the offset. | |
| /// | |
| /// The x component is given by [dx]. | |
| double get dy => _dy; | |
| /// The magnitude of the offset. | |
| /// | |
| /// If you need this value to compare it to another [Offset]'s distance, | |
| /// consider using [distanceSquared] instead, since it is cheaper to compute. | |
| double get distance => math.sqrt(dx * dx + dy * dy); | |
| /// The square of the magnitude of the offset. | |
| /// | |
| /// This is cheaper than computing the [distance] itself. | |
| double get distanceSquared => dx * dx + dy * dy; | |
| /// The angle of this offset as radians clockwise from the positive x-axis, in | |
| /// the range -[pi] to [pi], assuming positive values of the x-axis go to the | |
| /// right and positive values of the y-axis go down. | |
| /// | |
| /// Zero means that [dy] is zero and [dx] is zero or positive. | |
| /// | |
| /// Values from zero to [pi]/2 indicate positive values of [dx] and [dy], the | |
| /// bottom-right quadrant. | |
| /// | |
| /// Values from [pi]/2 to [pi] indicate negative values of [dx] and positive | |
| /// values of [dy], the bottom-left quadrant. | |
| /// | |
| /// Values from zero to -[pi]/2 indicate positive values of [dx] and negative | |
| /// values of [dy], the top-right quadrant. | |
| /// | |
| /// Values from -[pi]/2 to -[pi] indicate negative values of [dx] and [dy], | |
| /// the top-left quadrant. | |
| /// | |
| /// When [dy] is zero and [dx] is negative, the [direction] is [pi]. | |
| /// | |
| /// When [dx] is zero, [direction] is [pi]/2 if [dy] is positive and -[pi]/2 | |
| /// if [dy] is negative. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [distance], to compute the magnitude of the vector. | |
| /// * [Canvas.rotate], which uses the same convention for its angle. | |
| double get direction => math.atan2(dy, dx); | |
| /// An offset with zero magnitude. | |
| /// | |
| /// This can be used to represent the origin of a coordinate space. | |
| static const Offset zero = Offset(0.0, 0.0); | |
| /// An offset with infinite x and y components. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [isInfinite], which checks whether either component is infinite. | |
| /// * [isFinite], which checks whether both components are finite. | |
| // This is included for completeness, because [Size.infinite] exists. | |
| static const Offset infinite = Offset(double.infinity, double.infinity); | |
| /// Returns a new offset with the x component scaled by `scaleX` and the y | |
| /// component scaled by `scaleY`. | |
| /// | |
| /// If the two scale arguments are the same, consider using the `*` operator | |
| /// instead: | |
| /// | |
| /// ```dart | |
| /// Offset a = const Offset(10.0, 10.0); | |
| /// Offset b = a * 2.0; // same as: a.scale(2.0, 2.0) | |
| /// ``` | |
| /// | |
| /// If the two arguments are -1, consider using the unary `-` operator | |
| /// instead: | |
| /// | |
| /// ```dart | |
| /// Offset a = const Offset(10.0, 10.0); | |
| /// Offset b = -a; // same as: a.scale(-1.0, -1.0) | |
| /// ``` | |
| Offset scale(double scaleX, double scaleY) => Offset(dx * scaleX, dy * scaleY); | |
| /// Returns a new offset with translateX added to the x component and | |
| /// translateY added to the y component. | |
| /// | |
| /// If the arguments come from another [Offset], consider using the `+` or `-` | |
| /// operators instead: | |
| /// | |
| /// ```dart | |
| /// Offset a = const Offset(10.0, 10.0); | |
| /// Offset b = const Offset(10.0, 10.0); | |
| /// Offset c = a + b; // same as: a.translate(b.dx, b.dy) | |
| /// Offset d = a - b; // same as: a.translate(-b.dx, -b.dy) | |
| /// ``` | |
| Offset translate(double translateX, double translateY) => Offset(dx + translateX, dy + translateY); | |
| /// Unary negation operator. | |
| /// | |
| /// Returns an offset with the coordinates negated. | |
| /// | |
| /// If the [Offset] represents an arrow on a plane, this operator returns the | |
| /// same arrow but pointing in the reverse direction. | |
| Offset operator -() => Offset(-dx, -dy); | |
| /// Binary subtraction operator. | |
| /// | |
| /// Returns an offset whose [dx] value is the left-hand-side operand's [dx] | |
| /// minus the right-hand-side operand's [dx] and whose [dy] value is the | |
| /// left-hand-side operand's [dy] minus the right-hand-side operand's [dy]. | |
| /// | |
| /// See also [translate]. | |
| Offset operator -(Offset other) => Offset(dx - other.dx, dy - other.dy); | |
| /// Binary addition operator. | |
| /// | |
| /// Returns an offset whose [dx] value is the sum of the [dx] values of the | |
| /// two operands, and whose [dy] value is the sum of the [dy] values of the | |
| /// two operands. | |
| /// | |
| /// See also [translate]. | |
| Offset operator +(Offset other) => Offset(dx + other.dx, dy + other.dy); | |
| /// Multiplication operator. | |
| /// | |
| /// Returns an offset whose coordinates are the coordinates of the | |
| /// left-hand-side operand (an Offset) multiplied by the scalar | |
| /// right-hand-side operand (a double). | |
| /// | |
| /// See also [scale]. | |
| Offset operator *(double operand) => Offset(dx * operand, dy * operand); | |
| /// Division operator. | |
| /// | |
| /// Returns an offset whose coordinates are the coordinates of the | |
| /// left-hand-side operand (an Offset) divided by the scalar right-hand-side | |
| /// operand (a double). | |
| /// | |
| /// See also [scale]. | |
| Offset operator /(double operand) => Offset(dx / operand, dy / operand); | |
| /// Integer (truncating) division operator. | |
| /// | |
| /// Returns an offset whose coordinates are the coordinates of the | |
| /// left-hand-side operand (an Offset) divided by the scalar right-hand-side | |
| /// operand (a double), rounded towards zero. | |
| Offset operator ~/(double operand) => Offset((dx ~/ operand).toDouble(), (dy ~/ operand).toDouble()); | |
| /// Modulo (remainder) operator. | |
| /// | |
| /// Returns an offset whose coordinates are the remainder of dividing the | |
| /// coordinates of the left-hand-side operand (an Offset) by the scalar | |
| /// right-hand-side operand (a double). | |
| Offset operator %(double operand) => Offset(dx % operand, dy % operand); | |
| /// Rectangle constructor operator. | |
| /// | |
| /// Combines an [Offset] and a [Size] to form a [Rect] whose top-left | |
| /// coordinate is the point given by adding this offset, the left-hand-side | |
| /// operand, to the origin, and whose size is the right-hand-side operand. | |
| /// | |
| /// ```dart | |
| /// Rect myRect = Offset.zero & const Size(100.0, 100.0); | |
| /// // same as: Rect.fromLTWH(0.0, 0.0, 100.0, 100.0) | |
| /// ``` | |
| Rect operator &(Size other) => Rect.fromLTWH(dx, dy, other.width, other.height); | |
| /// Linearly interpolate between two offsets. | |
| /// | |
| /// If either offset is null, this function interpolates from [Offset.zero]. | |
| /// | |
| /// The `t` argument represents position on the timeline, with 0.0 meaning | |
| /// that the interpolation has not started, returning `a` (or something | |
| /// equivalent to `a`), 1.0 meaning that the interpolation has finished, | |
| /// returning `b` (or something equivalent to `b`), and values in between | |
| /// meaning that the interpolation is at the relevant point on the timeline | |
| /// between `a` and `b`. The interpolation can be extrapolated beyond 0.0 and | |
| /// 1.0, so negative values and values greater than 1.0 are valid (and can | |
| /// easily be generated by curves such as [Curves.elasticInOut]). | |
| /// | |
| /// Values for `t` are usually obtained from an [Animation<double>], such as | |
| /// an [AnimationController]. | |
| static Offset? lerp(Offset? a, Offset? b, double t) { | |
| if (b == null) { | |
| if (a == null) { | |
| return null; | |
| } else { | |
| return a * (1.0 - t); | |
| } | |
| } else { | |
| if (a == null) { | |
| return b * t; | |
| } else { | |
| return Offset(_lerpDouble(a.dx, b.dx, t), _lerpDouble(a.dy, b.dy, t)); | |
| } | |
| } | |
| } | |
| /// Compares two Offsets for equality. | |
| @override | |
| bool operator ==(Object other) { | |
| return other is Offset | |
| && other.dx == dx | |
| && other.dy == dy; | |
| } | |
| @override | |
| int get hashCode => Object.hash(dx, dy); | |
| @override | |
| String toString() => 'Offset(${dx.toStringAsFixed(1)}, ${dy.toStringAsFixed(1)})'; | |
| } | |
| /// Holds a 2D floating-point size. | |
| /// | |
| /// You can think of this as an [Offset] from the origin. | |
| class Size extends OffsetBase { | |
| /// Creates a [Size] with the given [width] and [height]. | |
| const Size(super.width, super.height); | |
| /// Creates an instance of [Size] that has the same values as another. | |
| // Used by the rendering library's _DebugSize hack. | |
| Size.copy(Size source) : super(source.width, source.height); | |
| /// Creates a square [Size] whose [width] and [height] are the given dimension. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [Size.fromRadius], which is more convenient when the available size | |
| /// is the radius of a circle. | |
| const Size.square(double dimension) : super(dimension, dimension); // ignore: use_super_parameters | |
| /// Creates a [Size] with the given [width] and an infinite [height]. | |
| const Size.fromWidth(double width) : super(width, double.infinity); | |
| /// Creates a [Size] with the given [height] and an infinite [width]. | |
| const Size.fromHeight(double height) : super(double.infinity, height); | |
| /// Creates a square [Size] whose [width] and [height] are twice the given | |
| /// dimension. | |
| /// | |
| /// This is a square that contains a circle with the given radius. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [Size.square], which creates a square with the given dimension. | |
| const Size.fromRadius(double radius) : super(radius * 2.0, radius * 2.0); | |
| /// The horizontal extent of this size. | |
| double get width => _dx; | |
| /// The vertical extent of this size. | |
| double get height => _dy; | |
| /// The aspect ratio of this size. | |
| /// | |
| /// This returns the [width] divided by the [height]. | |
| /// | |
| /// If the [width] is zero, the result will be zero. If the [height] is zero | |
| /// (and the [width] is not), the result will be [double.infinity] or | |
| /// [double.negativeInfinity] as determined by the sign of [width]. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [AspectRatio], a widget for giving a child widget a specific aspect | |
| /// ratio. | |
| /// * [FittedBox], a widget that (in most modes) attempts to maintain a | |
| /// child widget's aspect ratio while changing its size. | |
| double get aspectRatio { | |
| if (height != 0.0) { | |
| return width / height; | |
| } | |
| if (width > 0.0) { | |
| return double.infinity; | |
| } | |
| if (width < 0.0) { | |
| return double.negativeInfinity; | |
| } | |
| return 0.0; | |
| } | |
| /// An empty size, one with a zero width and a zero height. | |
| static const Size zero = Size(0.0, 0.0); | |
| /// A size whose [width] and [height] are infinite. | |
| /// | |
| /// See also: | |
| /// | |
| /// * [isInfinite], which checks whether either dimension is infinite. | |
| /// * [isFinite], which checks whether both dimensions are finite. | |
| static const Size infinite = Size(double.infinity, double.infinity); | |
| /// Whether this size encloses a non-zero area. | |
| /// | |
| /// Negative areas are considered empty. | |
| bool get isEmpty => width <= 0.0 || height <= 0.0; | |
| /// Binary subtraction operator for [Size]. | |
| /// | |
| /// Subtracting a [Size] from a [Size] returns the [Offset] that describes how | |
| /// much bigger the left-hand-side operand is than the right-hand-side | |
| /// operand. Adding that resulting [Offset] to the [Size] that was the | |
| /// right-hand-side operand would return a [Size] equal to the [Size] that was | |
| /// the left-hand-side operand. (i.e. if `sizeA - sizeB -> offsetA`, then | |
| /// `offsetA + sizeB -> sizeA`) | |
| /// | |
| /// Subtracting an [Offset] from a [Size] returns the [Size] that is smaller than | |
| /// the [Size] operand by the difference given by the [Offset] operand. In other | |
| /// words, the returned [Size] has a [width] consisting of the [width] of the | |
| /// left-hand-side operand minus the [Offset.dx] dimension of the | |
| /// right-hand-side operand, and a [height] consisting of the [height] of the | |
| /// left-hand-side operand minus the [Offset.dy] dimension of the | |
| /// right-hand-side operand. | |
| OffsetBase operator -(OffsetBase other) { | |
| if (other is Size) { | |
| return Offset(width - other.width, height - other.height); | |
| } | |
| if (other is Offset) { | |
| return Size(width - other.dx, height - other.dy); | |
| } | |
| throw ArgumentError(other); | |
| } | |
| /// Binary addition operator for adding an [Offset] to a [Size]. | |
| /// | |
| /// Returns a [Size] whose [width] is the sum of the [width] of the | |
| /// left-hand-side operand, a [Size], and the [Offset.dx] dimension of the | |
| /// right-hand-side operand, an [Offset], and whose [height] is the sum of the | |
| /// [height] of the left-hand-side operand and the [Offset.dy] dimension of | |
| /// the right-hand-side operand. | |
| Size operator +(Offset other) => Size(width + other.dx, height + other.dy); | |
| /// Multiplication operator. | |
| /// | |
| /// Returns a [Size] whose dimensions are the dimensions of the left-hand-side | |
| /// operand (a [Size]) multiplied by the scalar right-hand-side operand (a | |
| /// [double]). | |
| Size operator *(double operand) => Size(width * operand, height * operand); | |
| /// Division operator. | |
| /// | |
| /// Returns a [Size] whose dimensions are the dimensions of the left-hand-side | |
| /// operand (a [Size]) divided by the scalar right-hand-side operand (a | |
| /// [double]). | |
| Size operator /(double operand) => Size(width / operand, height / operand); | |
| /// Integer (truncating) division operator. | |
| /// | |
| /// Returns a [Size] whose dimensions are the dimensions of the left-hand-side | |
| /// operand (a [Size]) divided by the scalar right-hand-side operand (a | |
| /// [double]), rounded towards zero. | |
| Size operator ~/(double operand) => Size((width ~/ operand).toDouble(), (height ~/ operand).toDouble()); | |
| /// Modulo (remainder) operator. | |
| /// | |
| /// Returns a [Size] whose dimensions are the remainder of dividing the | |
| /// left-hand-side operand (a [Size]) by the scalar right-hand-side operand (a | |
| /// [double]). | |
| Size operator %(double operand) => Size(width % operand, height % operand); | |
| /// The lesser of the magnitudes of the [width] and the [height]. | |
| double get shortestSide => math.min(width.abs(), height.abs()); | |
| /// The greater of the magnitudes of the [width] and the [height]. | |
| double get longestSide => math.max(width.abs(), height.abs()); | |
| // Convenience methods that do the equivalent of calling the similarly named | |
| // methods on a Rect constructed from the given origin and this size. | |
| /// The offset to the intersection of the top and left edges of the rectangle | |
| /// described by the given [Offset] (which is interpreted as the top-left corner) | |
| /// and this [Size]. | |
| /// | |
| /// See also [Rect.topLeft]. | |
| Offset topLeft(Offset origin) => origin; | |
| /// The offset to the center of the top edge of the rectangle described by the | |
| /// given offset (which is interpreted as the top-left corner) and this size. | |
| /// | |
| /// See also [Rect.topCenter]. | |
| Offset topCenter(Offset origin) => Offset(origin.dx + width / 2.0, origin.dy); | |
| /// The offset to the intersection of the top and right edges of the rectangle | |
| /// described by the given offset (which is interpreted as the top-left corner) | |
| /// and this size. | |
| /// | |
| /// See also [Rect.topRight]. | |
| Offset topRight(Offset origin) => Offset(origin.dx + width, origin.dy); | |
| /// The offset to the center of the left edge of the rectangle described by the | |
| /// given offset (which is interpreted as the top-left corner) and this size. | |
| /// | |
| /// See also [Rect.centerLeft]. | |
| Offset centerLeft(Offset origin) => Offset(origin.dx, origin.dy + height / 2.0); | |
| /// The offset to the point halfway between the left and right and the top and | |
| /// bottom edges of the rectangle described by the given offset (which is | |
| /// interpreted as the top-left corner) and this size. | |
| /// | |
| /// See also [Rect.center]. | |
| Offset center(Offset origin) => Offset(origin.dx + width / 2.0, origin.dy + height / 2.0); | |
| /// The offset to the center of the right edge of the rectangle described by the | |
| /// given offset (which is interpreted as the top-left corner) and this size. | |
| /// | |
| /// See also [Rect.centerLeft]. | |
| Offset centerRight(Offset origin) => Offset(origin.dx + width, origin.dy + height / 2.0); | |
| /// The offset to the intersection of the bottom and left edges of the | |
| /// rectangle described by the given offset (which is interpreted as the | |
| /// top-left corner) and this size. | |
| /// | |
| /// See also [Rect.bottomLeft]. | |
| Offset bottomLeft(Offset origin) => Offset(origin.dx, origin.dy + height); | |
| /// The offset to the center of the bottom edge of the rectangle described by | |
| /// the given offset (which is interpreted as the top-left corner) and this | |
| /// size. | |
| /// | |
| /// See also [Rect.bottomLeft]. | |
| Offset bottomCenter(Offset origin) => Offset(origin.dx + width / 2.0, origin.dy + height); | |
| /// The offset to the intersection of the bottom and right edges of the | |
| /// rectangle described by the given offset (which is interpreted as the | |
| /// top-left corner) and this size. | |
| /// | |
| /// See also [Rect.bottomRight]. | |
| Offset bottomRight(Offset origin) => Offset(origin.dx + width, origin.dy + height); | |
| /// Whether the point specified by the given offset (which is assumed to be | |
| /// relative to the top left of the size) lies between the left and right and | |
| /// the top and bottom edges of a rectangle of this size. | |
| /// | |
| /// Rectangles include their top and left edges but exclude their bottom and | |
| /// right edges. | |
| bool contains(Offset offset) { | |
| return offset.dx >= 0.0 && offset.dx < width && offset.dy >= 0.0 && offset.dy < height; | |
| } | |
| /// A [Size] with the [width] and [height] swapped. | |
| Size get flipped => Size(height, width); | |
| /// Linearly interpolate between two sizes | |
| /// | |
| /// If either size is null, this function interpolates from [Size.zero]. | |
| /// | |
| /// The `t` argument represents position on the timeline, with 0.0 meaning | |
| /// that the interpolation has not started, returning `a` (or something | |
| /// equivalent to `a`), 1.0 meaning that the interpolation has finished, | |
| /// returning `b` (or something equivalent to `b`), and values in between | |
| /// meaning that the interpolation is at the relevant point on the timeline | |
| /// between `a` and `b`. The interpolation can be extrapolated beyond 0.0 and | |
| /// 1.0, so negative values and values greater than 1.0 are valid (and can | |
| /// easily be generated by curves such as [Curves.elasticInOut]). | |
| /// | |
| /// Values for `t` are usually obtained from an [Animation<double>], such as | |
| /// an [AnimationController]. | |
| static Size? lerp(Size? a, Size? b, double t) { | |
| if (b == null) { | |
| if (a == null) { | |
| return null; | |
| } else { | |
| return a * (1.0 - t); | |
| } | |
| } else { | |
| if (a == null) { | |
| return b * t; | |
| } else { | |
| return Size(_lerpDouble(a.width, b.width, t), _lerpDouble(a.height, b.height, t)); | |
| } | |
| } | |
| } | |
| /// Compares two Sizes for equality. | |
| // We don't compare the runtimeType because of _DebugSize in the framework. | |
| @override | |
| bool operator ==(Object other) { | |
| return other is Size | |
| && other._dx == _dx | |
| && other._dy == _dy; | |
| } | |
| @override | |
| int get hashCode => Object.hash(_dx, _dy); | |
| @override | |
| String toString() => 'Size(${width.toStringAsFixed(1)}, ${height.toStringAsFixed(1)})'; | |
| } | |
| /// An immutable, 2D, axis-aligned, floating-point rectangle whose coordinates | |
| /// are relative to a given origin. | |
| /// | |
| /// A Rect can be created with one of its constructors or from an [Offset] and a | |
| /// [Size] using the `&` operator: | |
| /// | |
| /// ```dart | |
| /// Rect myRect = const Offset(1.0, 2.0) & const Size(3.0, 4.0); | |
| /// ``` | |
| class Rect { | |
| /// Construct a rectangle from its left, top, right, and bottom edges. | |
| /// | |
| ///  | |
| ///  | |
| const Rect.fromLTRB(this.left, this.top, this.right, this.bottom); | |
| /// Construct a rectangle from its left and top edges, its width, and its | |
| /// height. | |
| /// | |
| /// To construct a [Rect] from an [Offset] and a [Size], you can use the | |
| /// rectangle constructor operator `&`. See [Offset.&]. | |
| /// | |
| ///  | |
| ///  | |
| const Rect.fromLTWH(double left, double top, double width, double height) : this.fromLTRB(left, top, left + width, top + height); | |
| /// Construct a rectangle that bounds the given circle. | |
| /// | |
| /// The `center` argument is assumed to be an offset from the origin. | |
| /// | |
| ///  | |
| ///  | |
| Rect.fromCircle({ required Offset center, required double radius }) : this.fromCenter( | |
| center: center, | |
| width: radius * 2, | |
| height: radius * 2, | |
| ); | |
| /// Constructs a rectangle from its center point, width, and height. | |
| /// | |
| /// The `center` argument is assumed to be an offset from the origin. | |
| /// | |
| ///  | |
| ///  | |
| Rect.fromCenter({ required Offset center, required double width, required double height }) : this.fromLTRB( | |
| center.dx - width / 2, | |
| center.dy - height / 2, | |
| center.dx + width / 2, | |
| center.dy + height / 2, | |
| ); | |
| /// Construct the smallest rectangle that encloses the given offsets, treating | |
| /// them as vectors from the origin. | |
| /// | |
| ///  | |
| ///  | |
| Rect.fromPoints(Offset a, Offset b) : this.fromLTRB( | |
| math.min(a.dx, b.dx), | |
| math.min(a.dy, b.dy), | |
| math.max(a.dx, b.dx), | |
| math.max(a.dy, b.dy), | |
| ); | |
| Float32List _getValue32() { | |
| final Float32List result = Float32List(4); | |
| result[0] = left; | |
| result[1] = top; | |
| result[2] = right; | |
| result[3] = bottom; | |
| return result; | |
| } | |
| /// The offset of the left edge of this rectangle from the x axis. | |
| final double left; | |
| /// The offset of the top edge of this rectangle from the y axis. | |
| final double top; | |
| /// The offset of the right edge of this rectangle from the x axis. | |
| final double right; | |
| /// The offset of the bottom edge of this rectangle from the y axis. | |
| final double bottom; | |
| /// The distance between the left and right edges of this rectangle. | |
| double get width => right - left; | |
| /// The distance between the top and bottom edges of this rectangle. | |
| double get height => bottom - top; | |
| /// The distance between the upper-left corner and the lower-right corner of | |
| /// this rectangle. | |
| Size get size => Size(width, height); | |
| /// Whether any of the dimensions are `NaN`. | |
| bool get hasNaN => left.isNaN || top.isNaN || right.isNaN || bottom.isNaN; | |
| /// A rectangle with left, top, right, and bottom edges all at zero. | |
| static const Rect zero = Rect.fromLTRB(0.0, 0.0, 0.0, 0.0); | |
| static const double _giantScalar = 1.0E+9; // matches kGiantRect from layer.h | |
| /// A rectangle that covers the entire coordinate space. | |
| /// | |
| /// This covers the space from -1e9,-1e9 to 1e9,1e9. | |
| /// This is the space over which graphics operations are valid. | |
| static const Rect largest = Rect.fromLTRB(-_giantScalar, -_giantScalar, _giantScalar, _giantScalar); | |
| /// Whether any of the coordinates of this rectangle are equal to positive infinity. | |
| // included for consistency with Offset and Size | |
| bool get isInfinite { | |
| return left >= double.infinity | |
| || top >= double.infinity | |
| || right >= double.infinity | |
| || bottom >= double.infinity; | |
| } | |
| /// Whether all coordinates of this rectangle are finite. | |
| bool get isFinite => left.isFinite && top.isFinite && right.isFinite && bottom.isFinite; | |
| /// Whether this rectangle encloses a non-zero area. Negative areas are | |
| /// considered empty. | |
| bool get isEmpty => left >= right || top >= bottom; | |
| /// Returns a new rectangle translated by the given offset. | |
| /// | |
| /// To translate a rectangle by separate x and y components rather than by an | |
| /// [Offset], consider [translate]. | |
| Rect shift(Offset offset) { | |
| return Rect.fromLTRB(left + offset.dx, top + offset.dy, right + offset.dx, bottom + offset.dy); | |
| } | |
| /// Returns a new rectangle with translateX added to the x components and | |
| /// translateY added to the y components. | |
| /// | |
| /// To translate a rectangle by an [Offset] rather than by separate x and y | |
| /// components, consider [shift]. | |
| Rect translate(double translateX, double translateY) { | |
| return Rect.fromLTRB(left + translateX, top + translateY, right + translateX, bottom + translateY); | |
| } | |
| /// Returns a new rectangle with edges moved outwards by the given delta. | |
| Rect inflate(double delta) { | |
| return Rect.fromLTRB(left - delta, top - delta, right + delta, bottom + delta); | |
| } | |
| /// Returns a new rectangle with edges moved inwards by the given delta. | |
| Rect deflate(double delta) => inflate(-delta); | |
| /// Returns a new rectangle that is the intersection of the given | |
| /// rectangle and this rectangle. The two rectangles must overlap | |
| /// for this to be meaningful. If the two rectangles do not overlap, | |
| /// then the resulting Rect will have a negative width or height. | |
| Rect intersect(Rect other) { | |
| return Rect.fromLTRB( | |
| math.max(left, other.left), | |
| math.max(top, other.top), | |
| math.min(right, other.right), | |
| math.min(bottom, other.bottom) | |
| ); | |
| } | |
| /// Returns a new rectangle which is the bounding box containing this | |
| /// rectangle and the given rectangle. | |
| Rect expandToInclude(Rect other) { | |
| return Rect.fromLTRB( | |
| math.min(left, other.left), | |
| math.min(top, other.top), | |
| math.max(right, other.right), | |
| math.max(bottom, other.bottom), | |
| ); | |
| } | |
| /// Whether `other` has a nonzero area of overlap with this rectangle. | |
| bool overlaps(Rect other) { | |
| if (right <= other.left || other.right <= left) { | |
| return false; | |
| } | |
| if (bottom <= other.top || other.bottom <= top) { | |
| return false; | |
| } | |
| return true; | |
| } | |
| /// The lesser of the magnitudes of the [width] and the [height] of this | |
| /// rectangle. | |
| double get shortestSide => math.min(width.abs(), height.abs()); | |
| /// The greater of the magnitudes of the [width] and the [height] of this | |
| /// rectangle. | |
| double get longestSide => math.max(width.abs(), height.abs()); | |
| /// The offset to the intersection of the top and left edges of this rectangle. | |
| /// | |
| /// See also [Size.topLeft]. | |
| Offset get topLeft => Offset(left, top); | |
| /// The offset to the center of the top edge of this rectangle. | |
| /// | |
| /// See also [Size.topCenter]. | |
| Offset get topCenter => Offset(left + width / 2.0, top); | |
| /// The offset to the intersection of the top and right edges of this rectangle. | |
| /// | |
| /// See also [Size.topRight]. | |
| Offset get topRight => Offset(right, top); | |
| /// The offset to the center of the left edge of this rectangle. | |
| /// | |
| /// See also [Size.centerLeft]. | |
| Offset get centerLeft => Offset(left, top + height / 2.0); | |
| /// The offset to the point halfway between the left and right and the top and | |
| /// bottom edges of this rectangle. | |
| /// | |
| /// See also [Size.center]. | |
| Offset get center => Offset(left + width / 2.0, top + height / 2.0); | |
| /// The offset to the center of the right edge of this rectangle. | |
| /// | |
| /// See also [Size.centerLeft]. | |
| Offset get centerRight => Offset(right, top + height / 2.0); | |
| /// The offset to the intersection of the bottom and left edges of this rectangle. | |
| /// | |
| /// See also [Size.bottomLeft]. | |
| Offset get bottomLeft => Offset(left, bottom); | |
| /// The offset to the center of the bottom edge of this rectangle. | |
| /// | |
| /// See also [Size.bottomLeft]. | |
| Offset get bottomCenter => Offset(left + width / 2.0, bottom); | |
| /// The offset to the intersection of the bottom and right edges of this rectangle. | |
| /// | |
| /// See also [Size.bottomRight]. | |
| Offset get bottomRight => Offset(right, bottom); | |
| /// Whether the point specified by the given offset (which is assumed to be | |
| /// relative to the origin) lies between the left and right and the top and | |
| /// bottom edges of this rectangle. | |
| /// | |
| /// Rectangles include their top and left edges but exclude their bottom and | |
| /// right edges. | |
| bool contains(Offset offset) { | |
| return offset.dx >= left && offset.dx < right && offset.dy >= top && offset.dy < bottom; | |
| } | |
| /// Linearly interpolate between two rectangles. | |
| /// | |
| /// If either rect is null, [Rect.zero] is used as a substitute. | |
| /// | |
| /// The `t` argument represents position on the timeline, with 0.0 meaning | |
| /// that the interpolation has not started, returning `a` (or something | |
| /// equivalent to `a`), 1.0 meaning that the interpolation has finished, | |
| /// returning `b` (or something equivalent to `b`), and values in between | |
| /// meaning that the interpolation is at the relevant point on the timeline | |
| /// between `a` and `b`. The interpolation can be extrapolated beyond 0.0 and | |
| /// 1.0, so negative values and values greater than 1.0 are valid (and can | |
| /// easily be generated by curves such as [Curves.elasticInOut]). | |
| /// | |
| /// Values for `t` are usually obtained from an [Animation<double>], such as | |
| /// an [AnimationController]. | |
| static Rect? lerp(Rect? a, Rect? b, double t) { | |
| if (b == null) { | |
| if (a == null) { | |
| return null; | |
| } else { | |
| final double k = 1.0 - t; | |
| return Rect.fromLTRB(a.left * k, a.top * k, a.right * k, a.bottom * k); | |
| } | |
| } else { | |
| if (a == null) { | |
| return Rect.fromLTRB(b.left * t, b.top * t, b.right * t, b.bottom * t); | |
| } else { | |
| return Rect.fromLTRB( | |
| _lerpDouble(a.left, b.left, t), | |
| _lerpDouble(a.top, b.top, t), | |
| _lerpDouble(a.right, b.right, t), | |
| _lerpDouble(a.bottom, b.bottom, t), | |
| ); | |
| } | |
| } | |
| } | |
| @override | |
| bool operator ==(Object other) { | |
| if (identical(this, other)) { | |
| return true; | |
| } | |
| if (runtimeType != other.runtimeType) { | |
| return false; | |
| } | |
| return other is Rect | |
| && other.left == left | |
| && other.top == top | |
| && other.right == right | |
| && other.bottom == bottom; | |
| } | |
| @override | |
| int get hashCode => Object.hash(left, top, right, bottom); | |
| @override | |
| String toString() => 'Rect.fromLTRB(${left.toStringAsFixed(1)}, ${top.toStringAsFixed(1)}, ${right.toStringAsFixed(1)}, ${bottom.toStringAsFixed(1)})'; | |
| } | |
| /// A radius for either circular or elliptical shapes. | |
| class Radius { | |
| /// Constructs a circular radius. [x] and [y] will have the same radius value. | |
| /// | |
| ///  | |
| ///  | |
| const Radius.circular(double radius) : this.elliptical(radius, radius); | |
| /// Constructs an elliptical radius with the given radii. | |
| /// | |
| ///  | |
| ///  | |
| const Radius.elliptical(this.x, this.y); | |
| /// The radius value on the horizontal axis. | |
| final double x; | |
| /// The radius value on the vertical axis. | |
| final double y; | |
| /// A radius with [x] and [y] values set to zero. | |
| /// | |
| /// You can use [Radius.zero] with [RRect] to have right-angle corners. | |
| static const Radius zero = Radius.circular(0.0); | |
| /// Returns this [Radius], with values clamped to the given min and max | |
| /// [Radius] values. | |
| /// | |
| /// The `min` value defaults to `Radius.circular(-double.infinity)`, and | |
| /// the `max` value defaults to `Radius.circular(double.infinity)`. | |
| Radius clamp({Radius? minimum, Radius? maximum}) { | |
| minimum ??= const Radius.circular(-double.infinity); | |
| maximum ??= const Radius.circular(double.infinity); | |
| return Radius.elliptical( | |
| clampDouble(x, minimum.x, maximum.x), | |
| clampDouble(y, minimum.y, maximum.y), | |
| ); | |
| } | |
| /// Returns this [Radius], with values clamped to the given min and max | |
| /// values in each dimension | |
| /// | |
| /// The `minimumX` and `minimumY` values default to `-double.infinity`, and | |
| /// the `maximumX` and `maximumY` values default to `double.infinity`. | |
| Radius clampValues({ | |
| double? minimumX, | |
| double? minimumY, | |
| double? maximumX, | |
| double? maximumY, | |
| }) { | |
| return Radius.elliptical( | |
| clampDouble(x, minimumX ?? -double.infinity, maximumX ?? double.infinity), | |
| clampDouble(y, minimumY ?? -double.infinity, maximumY ?? double.infinity), | |
| ); | |
| } | |
| /// Unary negation operator. | |
| /// | |
| /// Returns a Radius with the distances negated. | |
| /// | |
| /// Radiuses with negative values aren't geometrically meaningful, but could | |
| /// occur as part of expressions. For example, negating a radius of one pixel | |
| /// and then adding the result to another radius is equivalent to subtracting | |
| /// a radius of one pixel from the other. | |
| Radius operator -() => Radius.elliptical(-x, -y); | |
| /// Binary subtraction operator. | |
| /// | |
| /// Returns a radius whose [x] value is the left-hand-side operand's [x] | |
| /// minus the right-hand-side operand's [x] and whose [y] value is the | |
| /// left-hand-side operand's [y] minus the right-hand-side operand's [y]. | |
| Radius operator -(Radius other) => Radius.elliptical(x - other.x, y - other.y); | |
| /// Binary addition operator. | |
| /// | |
| /// Returns a radius whose [x] value is the sum of the [x] values of the | |
| /// two operands, and whose [y] value is the sum of the [y] values of the | |
| /// two operands. | |
| Radius operator +(Radius other) => Radius.elliptical(x + other.x, y + other.y); | |
| /// Multiplication operator. | |
| /// | |
| /// Returns a radius whose coordinates are the coordinates of the | |
| /// left-hand-side operand (a radius) multiplied by the scalar | |
| /// right-hand-side operand (a double). | |
| Radius operator *(double operand) => Radius.elliptical(x * operand, y * operand); | |
| /// Division operator. | |
| /// | |
| /// Returns a radius whose coordinates are the coordinates of the | |
| /// left-hand-side operand (a radius) divided by the scalar right-hand-side | |
| /// operand (a double). | |
| Radius operator /(double operand) => Radius.elliptical(x / operand, y / operand); | |
| /// Integer (truncating) division operator. | |
| /// | |
| /// Returns a radius whose coordinates are the coordinates of the | |
| /// left-hand-side operand (a radius) divided by the scalar right-hand-side | |
| /// operand (a double), rounded towards zero. | |
| Radius operator ~/(double operand) => Radius.elliptical((x ~/ operand).toDouble(), (y ~/ operand).toDouble()); | |
| /// Modulo (remainder) operator. | |
| /// | |
| /// Returns a radius whose coordinates are the remainder of dividing the | |
| /// coordinates of the left-hand-side operand (a radius) by the scalar | |
| /// right-hand-side operand (a double). | |
| Radius operator %(double operand) => Radius.elliptical(x % operand, y % operand); | |
| /// Linearly interpolate between two radii. | |
| /// | |
| /// If either is null, this function substitutes [Radius.zero] instead. | |
| /// | |
| /// The `t` argument represents position on the timeline, with 0.0 meaning | |
| /// that the interpolation has not started, returning `a` (or something | |
| /// equivalent to `a`), 1.0 meaning that the interpolation has finished, | |
| /// returning `b` (or something equivalent to `b`), and values in between | |
| /// meaning that the interpolation is at the relevant point on the timeline | |
| /// between `a` and `b`. The interpolation can be extrapolated beyond 0.0 and | |
| /// 1.0, so negative values and values greater than 1.0 are valid (and can | |
| /// easily be generated by curves such as [Curves.elasticInOut]). | |
| /// | |
| /// Values for `t` are usually obtained from an [Animation<double>], such as | |
| /// an [AnimationController]. | |
| static Radius? lerp(Radius? a, Radius? b, double t) { | |
| if (b == null) { | |
| if (a == null) { | |
| return null; | |
| } else { | |
| final double k = 1.0 - t; | |
| return Radius.elliptical(a.x * k, a.y * k); | |
| } | |
| } else { | |
| if (a == null) { | |
| return Radius.elliptical(b.x * t, b.y * t); | |
| } else { | |
| return Radius.elliptical( | |
| _lerpDouble(a.x, b.x, t), | |
| _lerpDouble(a.y, b.y, t), | |
| ); | |
| } | |
| } | |
| } | |
| @override | |
| bool operator ==(Object other) { | |
| if (identical(this, other)) { | |
| return true; | |
| } | |
| if (runtimeType != other.runtimeType) { | |
| return false; | |
| } | |
| return other is Radius | |
| && other.x == x | |
| && other.y == y; | |
| } | |
| @override | |
| int get hashCode => Object.hash(x, y); | |
| @override | |
| String toString() { | |
| return x == y ? 'Radius.circular(${x.toStringAsFixed(1)})' : | |
| 'Radius.elliptical(${x.toStringAsFixed(1)}, ' | |
| '${y.toStringAsFixed(1)})'; | |
| } | |
| } | |
| /// An immutable rounded rectangle with the custom radii for all four corners. | |
| class RRect { | |
| /// Construct a rounded rectangle from its left, top, right, and bottom edges, | |
| /// and the same radii along its horizontal axis and its vertical axis. | |
| /// | |
| /// Will assert in debug mode if `radiusX` or `radiusY` are negative. | |
| const RRect.fromLTRBXY( | |
| double left, | |
| double top, | |
| double right, | |
| double bottom, | |
| double radiusX, | |
| double radiusY, | |
| ) : this._raw( | |
| top: top, | |
| left: left, | |
| right: right, | |
| bottom: bottom, | |
| tlRadiusX: radiusX, | |
| tlRadiusY: radiusY, | |
| trRadiusX: radiusX, | |
| trRadiusY: radiusY, | |
| blRadiusX: radiusX, | |
| blRadiusY: radiusY, | |
| brRadiusX: radiusX, | |
| brRadiusY: radiusY, | |
| ); | |
| /// Construct a rounded rectangle from its left, top, right, and bottom edges, | |
| /// and the same radius in each corner. | |
| /// | |
| /// Will assert in debug mode if the `radius` is negative in either x or y. | |
| RRect.fromLTRBR( | |
| double left, | |
| double top, | |
| double right, | |
| double bottom, | |
| Radius radius, | |
| ) | |
| : this._raw( | |
| top: top, | |
| left: left, | |
| right: right, | |
| bottom: bottom, | |
| tlRadiusX: radius.x, | |
| tlRadiusY: radius.y, | |
| trRadiusX: radius.x, | |
| trRadiusY: radius.y, | |
| blRadiusX: radius.x, | |
| blRadiusY: radius.y, | |
| brRadiusX: radius.x, | |
| brRadiusY: radius.y, | |
| ); | |
| /// Construct a rounded rectangle from its bounding box and the same radii | |
| /// along its horizontal axis and its vertical axis. | |
| /// | |
| /// Will assert in debug mode if `radiusX` or `radiusY` are negative. | |
| RRect.fromRectXY(Rect rect, double radiusX, double radiusY) | |
| : this._raw( | |
| top: rect.top, | |
| left: rect.left, | |
| right: rect.right, | |
| bottom: rect.bottom, | |
| tlRadiusX: radiusX, | |
| tlRadiusY: radiusY, | |
| trRadiusX: radiusX, | |
| trRadiusY: radiusY, | |
| blRadiusX: radiusX, | |
| blRadiusY: radiusY, | |
| brRadiusX: radiusX, | |
| brRadiusY: radiusY, | |
| ); | |
| /// Construct a rounded rectangle from its bounding box and a radius that is | |
| /// the same in each corner. | |
| /// | |
| /// Will assert in debug mode if the `radius` is negative in either x or y. | |
| RRect.fromRectAndRadius(Rect rect, Radius radius) | |
| : this._raw( | |
| top: rect.top, | |
| left: rect.left, | |
| right: rect.right, | |
| bottom: rect.bottom, | |
| tlRadiusX: radius.x, | |
| tlRadiusY: radius.y, | |
| trRadiusX: radius.x, | |
| trRadiusY: radius.y, | |
| blRadiusX: radius.x, | |
| blRadiusY: radius.y, | |
| brRadiusX: radius.x, | |
| brRadiusY: radius.y, | |
| ); | |
| /// Construct a rounded rectangle from its left, top, right, and bottom edges, | |
| /// and topLeft, topRight, bottomRight, and bottomLeft radii. | |
| /// | |
| /// The corner radii default to [Radius.zero], i.e. right-angled corners. Will | |
| /// assert in debug mode if any of the radii are negative in either x or y. | |
| RRect.fromLTRBAndCorners( | |
| double left, | |
| double top, | |
| double right, | |
| double bottom, { | |
| Radius topLeft = Radius.zero, | |
| Radius topRight = Radius.zero, | |
| Radius bottomRight = Radius.zero, | |
| Radius bottomLeft = Radius.zero, | |
| }) : this._raw( | |
| top: top, | |
| left: left, | |
| right: right, | |
| bottom: bottom, | |
| tlRadiusX: topLeft.x, | |
| tlRadiusY: topLeft.y, | |
| trRadiusX: topRight.x, | |
| trRadiusY: topRight.y, | |
| blRadiusX: bottomLeft.x, | |
| blRadiusY: bottomLeft.y, | |
| brRadiusX: bottomRight.x, | |
| brRadiusY: bottomRight.y, | |
| ); | |
| /// Construct a rounded rectangle from its bounding box and topLeft, | |
| /// topRight, bottomRight, and bottomLeft radii. | |
| /// | |
| /// The corner radii default to [Radius.zero], i.e. right-angled corners. Will | |
| /// assert in debug mode if any of the radii are negative in either x or y. | |
| RRect.fromRectAndCorners( | |
| Rect rect, | |
| { | |
| Radius topLeft = Radius.zero, | |
| Radius topRight = Radius.zero, | |
| Radius bottomRight = Radius.zero, | |
| Radius bottomLeft = Radius.zero | |
| } | |
| ) : this._raw( | |
| top: rect.top, | |
| left: rect.left, | |
| right: rect.right, | |
| bottom: rect.bottom, | |
| tlRadiusX: topLeft.x, | |
| tlRadiusY: topLeft.y, | |
| trRadiusX: topRight.x, | |
| trRadiusY: topRight.y, | |
| blRadiusX: bottomLeft.x, | |
| blRadiusY: bottomLeft.y, | |
| brRadiusX: bottomRight.x, | |
| brRadiusY: bottomRight.y, | |
| ); | |
| const RRect._raw({ | |
| this.left = 0.0, | |
| this.top = 0.0, | |
| this.right = 0.0, | |
| this.bottom = 0.0, | |
| this.tlRadiusX = 0.0, | |
| this.tlRadiusY = 0.0, | |
| this.trRadiusX = 0.0, | |
| this.trRadiusY = 0.0, | |
| this.brRadiusX = 0.0, | |
| this.brRadiusY = 0.0, | |
| this.blRadiusX = 0.0, | |
| this.blRadiusY = 0.0, | |
| }) : assert(tlRadiusX >= 0), | |
| assert(tlRadiusY >= 0), | |
| assert(trRadiusX >= 0), | |
| assert(trRadiusY >= 0), | |
| assert(brRadiusX >= 0), | |
| assert(brRadiusY >= 0), | |
| assert(blRadiusX >= 0), | |
| assert(blRadiusY >= 0); | |
| Float32List _getValue32() { | |
| final Float32List result = Float32List(12); | |
| result[0] = left; | |
| result[1] = top; | |
| result[2] = right; | |
| result[3] = bottom; | |
| result[4] = tlRadiusX; | |
| result[5] = tlRadiusY; | |
| result[6] = trRadiusX; | |
| result[7] = trRadiusY; | |
| result[8] = brRadiusX; | |
| result[9] = brRadiusY; | |
| result[10] = blRadiusX; | |
| result[11] = blRadiusY; | |
| return result; | |
| } | |
| /// The offset of the left edge of this rectangle from the x axis. | |
| final double left; | |
| /// The offset of the top edge of this rectangle from the y axis. | |
| final double top; | |
| /// The offset of the right edge of this rectangle from the x axis. | |
| final double right; | |
| /// The offset of the bottom edge of this rectangle from the y axis. | |
| final double bottom; | |
| /// The top-left horizontal radius. | |
| final double tlRadiusX; | |
| /// The top-left vertical radius. | |
| final double tlRadiusY; | |
| /// The top-left [Radius]. | |
| Radius get tlRadius => Radius.elliptical(tlRadiusX, tlRadiusY); | |
| /// The top-right horizontal radius. | |
| final double trRadiusX; | |
| /// The top-right vertical radius. | |
| final double trRadiusY; | |
| /// The top-right [Radius]. | |
| Radius get trRadius => Radius.elliptical(trRadiusX, trRadiusY); | |
| /// The bottom-right horizontal radius. | |
| final double brRadiusX; | |
| /// The bottom-right vertical radius. | |
| final double brRadiusY; | |
| /// The bottom-right [Radius]. | |
| Radius get brRadius => Radius.elliptical(brRadiusX, brRadiusY); | |
| /// The bottom-left horizontal radius. | |
| final double blRadiusX; | |
| /// The bottom-left vertical radius. | |
| final double blRadiusY; | |
| /// The bottom-left [Radius]. | |
| Radius get blRadius => Radius.elliptical(blRadiusX, blRadiusY); | |
| /// A rounded rectangle with all the values set to zero. | |
| static const RRect zero = RRect._raw(); | |
| /// Returns a new [RRect] translated by the given offset. | |
| RRect shift(Offset offset) { | |
| return RRect._raw( | |
| left: left + offset.dx, | |
| top: top + offset.dy, | |
| right: right + offset.dx, | |
| bottom: bottom + offset.dy, | |
| tlRadiusX: tlRadiusX, | |
| tlRadiusY: tlRadiusY, | |
| trRadiusX: trRadiusX, | |
| trRadiusY: trRadiusY, | |
| blRadiusX: blRadiusX, | |
| blRadiusY: blRadiusY, | |
| brRadiusX: brRadiusX, | |
| brRadiusY: brRadiusY, | |
| ); | |
| } | |
| /// Returns a new [RRect] with edges and radii moved outwards by the given | |
| /// delta. | |
| RRect inflate(double delta) { | |
| return RRect._raw( | |
| left: left - delta, | |
| top: top - delta, | |
| right: right + delta, | |
| bottom: bottom + delta, | |
| tlRadiusX: math.max(0, tlRadiusX + delta), | |
| tlRadiusY: math.max(0, tlRadiusY + delta), | |
| trRadiusX: math.max(0, trRadiusX + delta), | |
| trRadiusY: math.max(0, trRadiusY + delta), | |
| blRadiusX: math.max(0, blRadiusX + delta), | |
| blRadiusY: math.max(0, blRadiusY + delta), | |
| brRadiusX: math.max(0, brRadiusX + delta), | |
| brRadiusY: math.max(0, brRadiusY + delta), | |
| ); | |
| } | |
| /// Returns a new [RRect] with edges and radii moved inwards by the given delta. | |
| RRect deflate(double delta) => inflate(-delta); | |
| /// The distance between the left and right edges of this rectangle. | |
| double get width => right - left; | |
| /// The distance between the top and bottom edges of this rectangle. | |
| double get height => bottom - top; | |
| /// The bounding box of this rounded rectangle (the rectangle with no rounded corners). | |
| Rect get outerRect => Rect.fromLTRB(left, top, right, bottom); | |
| /// The non-rounded rectangle that is constrained by the smaller of the two | |
| /// diagonals, with each diagonal traveling through the middle of the curve | |
| /// corners. The middle of a corner is the intersection of the curve with its | |
| /// respective quadrant bisector. | |
| Rect get safeInnerRect { | |
| const double kInsetFactor = 0.29289321881; // 1-cos(pi/4) | |
| final double leftRadius = math.max(blRadiusX, tlRadiusX); | |
| final double topRadius = math.max(tlRadiusY, trRadiusY); | |
| final double rightRadius = math.max(trRadiusX, brRadiusX); | |
| final double bottomRadius = math.max(brRadiusY, blRadiusY); | |
| return Rect.fromLTRB( | |
| left + leftRadius * kInsetFactor, | |
| top + topRadius * kInsetFactor, | |
| right - rightRadius * kInsetFactor, | |
| bottom - bottomRadius * kInsetFactor | |
| ); | |
| } | |
| /// The rectangle that would be formed using the axis-aligned intersection of | |
| /// the sides of the rectangle, i.e., the rectangle formed from the | |
| /// inner-most centers of the ellipses that form the corners. This is the | |
| /// intersection of the [wideMiddleRect] and the [tallMiddleRect]. If any of | |
| /// the intersections are void, the resulting [Rect] will have negative width | |
| /// or height. | |
| Rect get middleRect { | |
| final double leftRadius = math.max(blRadiusX, tlRadiusX); | |
| final double topRadius = math.max(tlRadiusY, trRadiusY); | |
| final double rightRadius = math.max(trRadiusX, brRadiusX); | |
| final double bottomRadius = math.max(brRadiusY, blRadiusY); | |
| return Rect.fromLTRB( | |
| left + leftRadius, | |
| top + topRadius, | |
| right - rightRadius, | |
| bottom - bottomRadius | |
| ); | |
| } | |
| /// The biggest rectangle that is entirely inside the rounded rectangle and | |
| /// has the full width of the rounded rectangle. If the rounded rectangle does | |
| /// not have an axis-aligned intersection of its left and right side, the | |
| /// resulting [Rect] will have negative width or height. | |
| Rect get wideMiddleRect { | |
| final double topRadius = math.max(tlRadiusY, trRadiusY); | |
| final double bottomRadius = math.max(brRadiusY, blRadiusY); | |
| return Rect.fromLTRB( | |
| left, | |
| top + topRadius, | |
| right, | |
| bottom - bottomRadius | |
| ); | |
| } | |
| /// The biggest rectangle that is entirely inside the rounded rectangle and | |
| /// has the full height of the rounded rectangle. If the rounded rectangle | |
| /// does not have an axis-aligned intersection of its top and bottom side, the | |
| /// resulting [Rect] will have negative width or height. | |
| Rect get tallMiddleRect { | |
| final double leftRadius = math.max(blRadiusX, tlRadiusX); | |
| final double rightRadius = math.max(trRadiusX, brRadiusX); | |
| return Rect.fromLTRB( | |
| left + leftRadius, | |
| top, | |
| right - rightRadius, | |
| bottom | |
| ); | |
| } | |
| /// Whether this rounded rectangle encloses a non-zero area. | |
| /// Negative areas are considered empty. | |
| bool get isEmpty => left >= right || top >= bottom; | |
| /// Whether all coordinates of this rounded rectangle are finite. | |
| bool get isFinite => left.isFinite && top.isFinite && right.isFinite && bottom.isFinite; | |
| /// Whether this rounded rectangle is a simple rectangle with zero | |
| /// corner radii. | |
| bool get isRect { | |
| return (tlRadiusX == 0.0 || tlRadiusY == 0.0) && | |
| (trRadiusX == 0.0 || trRadiusY == 0.0) && | |
| (blRadiusX == 0.0 || blRadiusY == 0.0) && | |
| (brRadiusX == 0.0 || brRadiusY == 0.0); | |
| } | |
| /// Whether this rounded rectangle has a side with no straight section. | |
| bool get isStadium { | |
| return tlRadius == trRadius | |
| && trRadius == brRadius | |
| && brRadius == blRadius | |
| && (width <= 2.0 * tlRadiusX || height <= 2.0 * tlRadiusY); | |
| } | |
| /// Whether this rounded rectangle has no side with a straight section. | |
| bool get isEllipse { | |
| return tlRadius == trRadius | |
| && trRadius == brRadius | |
| && brRadius == blRadius | |
| && width <= 2.0 * tlRadiusX | |
| && height <= 2.0 * tlRadiusY; | |
| } | |
| /// Whether this rounded rectangle would draw as a circle. | |
| bool get isCircle => width == height && isEllipse; | |
| /// The lesser of the magnitudes of the [width] and the [height] of this | |
| /// rounded rectangle. | |
| double get shortestSide => math.min(width.abs(), height.abs()); | |
| /// The greater of the magnitudes of the [width] and the [height] of this | |
| /// rounded rectangle. | |
| double get longestSide => math.max(width.abs(), height.abs()); | |
| /// Whether any of the dimensions are `NaN`. | |
| bool get hasNaN => left.isNaN || top.isNaN || right.isNaN || bottom.isNaN || | |
| trRadiusX.isNaN || trRadiusY.isNaN || tlRadiusX.isNaN || tlRadiusY.isNaN || | |
| brRadiusX.isNaN || brRadiusY.isNaN || blRadiusX.isNaN || blRadiusY.isNaN; | |
| /// The offset to the point halfway between the left and right and the top and | |
| /// bottom edges of this rectangle. | |
| Offset get center => Offset(left + width / 2.0, top + height / 2.0); | |
| // Returns the minimum between min and scale to which radius1 and radius2 | |
| // should be scaled with in order not to exceed the limit. | |
| double _getMin(double min, double radius1, double radius2, double limit) { | |
| final double sum = radius1 + radius2; | |
| if (sum > limit && sum != 0.0) { | |
| return math.min(min, limit / sum); | |
| } | |
| return min; | |
| } | |
| /// Scales all radii so that on each side their sum will not exceed the size | |
| /// of the width/height. | |
| /// | |
| /// Skia already handles RRects with radii that are too large in this way. | |
| /// Therefore, this method is only needed for RRect use cases that require | |
| /// the appropriately scaled radii values. | |
| /// | |
| /// See the [Skia scaling implementation](https://github.com/google/skia/blob/main/src/core/SkRRect.cpp) | |
| /// for more details. | |
| RRect scaleRadii() { | |
| double scale = 1.0; | |
| scale = _getMin(scale, blRadiusY, tlRadiusY, height); | |
| scale = _getMin(scale, tlRadiusX, trRadiusX, width); | |
| scale = _getMin(scale, trRadiusY, brRadiusY, height); | |
| scale = _getMin(scale, brRadiusX, blRadiusX, width); | |
| assert(scale >= 0); | |
| if (scale < 1.0) { | |
| return RRect._raw( | |
| top: top, | |
| left: left, | |
| right: right, | |
| bottom: bottom, | |
| tlRadiusX: tlRadiusX * scale, | |
| tlRadiusY: tlRadiusY * scale, | |
| trRadiusX: trRadiusX * scale, | |
| trRadiusY: trRadiusY * scale, | |
| blRadiusX: blRadiusX * scale, | |
| blRadiusY: blRadiusY * scale, | |
| brRadiusX: brRadiusX * scale, | |
| brRadiusY: brRadiusY * scale, | |
| ); | |
| } | |
| return RRect._raw( | |
| top: top, | |
| left: left, | |
| right: right, | |
| bottom: bottom, | |
| tlRadiusX: tlRadiusX, | |
| tlRadiusY: tlRadiusY, | |
| trRadiusX: trRadiusX, | |
| trRadiusY: trRadiusY, | |
| blRadiusX: blRadiusX, | |
| blRadiusY: blRadiusY, | |
| brRadiusX: brRadiusX, | |
| brRadiusY: brRadiusY, | |
| ); | |
| } | |
| /// Whether the point specified by the given offset (which is assumed to be | |
| /// relative to the origin) lies inside the rounded rectangle. | |
| /// | |
| /// This method may allocate (and cache) a copy of the object with normalized | |
| /// radii the first time it is called on a particular [RRect] instance. When | |
| /// using this method, prefer to reuse existing [RRect]s rather than | |
| /// recreating the object each time. | |
| bool contains(Offset point) { | |
| if (point.dx < left || point.dx >= right || point.dy < top || point.dy >= bottom) { | |
| return false; | |
| } // outside bounding box | |
| final RRect scaled = scaleRadii(); | |
| double x; | |
| double y; | |
| double radiusX; | |
| double radiusY; | |
| // check whether point is in one of the rounded corner areas | |
| // x, y -> translate to ellipse center | |
| if (point.dx < left + scaled.tlRadiusX && | |
| point.dy < top + scaled.tlRadiusY) { | |
| x = point.dx - left - scaled.tlRadiusX; | |
| y = point.dy - top - scaled.tlRadiusY; | |
| radiusX = scaled.tlRadiusX; | |
| radiusY = scaled.tlRadiusY; | |
| } else if (point.dx > right - scaled.trRadiusX && | |
| point.dy < top + scaled.trRadiusY) { | |
| x = point.dx - right + scaled.trRadiusX; | |
| y = point.dy - top - scaled.trRadiusY; | |
| radiusX = scaled.trRadiusX; | |
| radiusY = scaled.trRadiusY; | |
| } else if (point.dx > right - scaled.brRadiusX && | |
| point.dy > bottom - scaled.brRadiusY) { | |
| x = point.dx - right + scaled.brRadiusX; | |
| y = point.dy - bottom + scaled.brRadiusY; | |
| radiusX = scaled.brRadiusX; | |
| radiusY = scaled.brRadiusY; | |
| } else if (point.dx < left + scaled.blRadiusX && | |
| point.dy > bottom - scaled.blRadiusY) { | |
| x = point.dx - left - scaled.blRadiusX; | |
| y = point.dy - bottom + scaled.blRadiusY; | |
| radiusX = scaled.blRadiusX; | |
| radiusY = scaled.blRadiusY; | |
| } else { | |
| return true; // inside and not within the rounded corner area | |
| } | |
| x = x / radiusX; | |
| y = y / radiusY; | |
| // check if the point is outside the unit circle | |
| if (x * x + y * y > 1.0) { | |
| return false; | |
| } | |
| return true; | |
| } | |
| /// Linearly interpolate between two rounded rectangles. | |
| /// | |
| /// If either is null, this function substitutes [RRect.zero] instead. | |
| /// | |
| /// The `t` argument represents position on the timeline, with 0.0 meaning | |
| /// that the interpolation has not started, returning `a` (or something | |
| /// equivalent to `a`), 1.0 meaning that the interpolation has finished, | |
| /// returning `b` (or something equivalent to `b`), and values in between | |
| /// meaning that the interpolation is at the relevant point on the timeline | |
| /// between `a` and `b`. The interpolation can be extrapolated beyond 0.0 and | |
| /// 1.0, so negative values and values greater than 1.0 are valid (and can | |
| /// easily be generated by curves such as [Curves.elasticInOut]). | |
| /// | |
| /// Values for `t` are usually obtained from an [Animation<double>], such as | |
| /// an [AnimationController]. | |
| static RRect? lerp(RRect? a, RRect? b, double t) { | |
| if (b == null) { | |
| if (a == null) { | |
| return null; | |
| } else { | |
| final double k = 1.0 - t; | |
| return RRect._raw( | |
| left: a.left * k, | |
| top: a.top * k, | |
| right: a.right * k, | |
| bottom: a.bottom * k, | |
| tlRadiusX: math.max(0, a.tlRadiusX * k), | |
| tlRadiusY: math.max(0, a.tlRadiusY * k), | |
| trRadiusX: math.max(0, a.trRadiusX * k), | |
| trRadiusY: math.max(0, a.trRadiusY * k), | |
| brRadiusX: math.max(0, a.brRadiusX * k), | |
| brRadiusY: math.max(0, a.brRadiusY * k), | |
| blRadiusX: math.max(0, a.blRadiusX * k), | |
| blRadiusY: math.max(0, a.blRadiusY * k), | |
| ); | |
| } | |
| } else { | |
| if (a == null) { | |
| return RRect._raw( | |
| left: b.left * t, | |
| top: b.top * t, | |
| right: b.right * t, | |
| bottom: b.bottom * t, | |
| tlRadiusX: math.max(0, b.tlRadiusX * t), | |
| tlRadiusY: math.max(0, b.tlRadiusY * t), | |
| trRadiusX: math.max(0, b.trRadiusX * t), | |
| trRadiusY: math.max(0, b.trRadiusY * t), | |
| brRadiusX: math.max(0, b.brRadiusX * t), | |
| brRadiusY: math.max(0, b.brRadiusY * t), | |
| blRadiusX: math.max(0, b.blRadiusX * t), | |
| blRadiusY: math.max(0, b.blRadiusY * t), | |
| ); | |
| } else { | |
| return RRect._raw( | |
| left: _lerpDouble(a.left, b.left, t), | |
| top: _lerpDouble(a.top, b.top, t), | |
| right: _lerpDouble(a.right, b.right, t), | |
| bottom: _lerpDouble(a.bottom, b.bottom, t), | |
| tlRadiusX: math.max(0, _lerpDouble(a.tlRadiusX, b.tlRadiusX, t)), | |
| tlRadiusY: math.max(0, _lerpDouble(a.tlRadiusY, b.tlRadiusY, t)), | |
| trRadiusX: math.max(0, _lerpDouble(a.trRadiusX, b.trRadiusX, t)), | |
| trRadiusY: math.max(0, _lerpDouble(a.trRadiusY, b.trRadiusY, t)), | |
| brRadiusX: math.max(0, _lerpDouble(a.brRadiusX, b.brRadiusX, t)), | |
| brRadiusY: math.max(0, _lerpDouble(a.brRadiusY, b.brRadiusY, t)), | |
| blRadiusX: math.max(0, _lerpDouble(a.blRadiusX, b.blRadiusX, t)), | |
| blRadiusY: math.max(0, _lerpDouble(a.blRadiusY, b.blRadiusY, t)), | |
| ); | |
| } | |
| } | |
| } | |
| @override | |
| bool operator ==(Object other) { | |
| if (identical(this, other)) { | |
| return true; | |
| } | |
| if (runtimeType != other.runtimeType) { | |
| return false; | |
| } | |
| return other is RRect | |
| && other.left == left | |
| && other.top == top | |
| && other.right == right | |
| && other.bottom == bottom | |
| && other.tlRadiusX == tlRadiusX | |
| && other.tlRadiusY == tlRadiusY | |
| && other.trRadiusX == trRadiusX | |
| && other.trRadiusY == trRadiusY | |
| && other.blRadiusX == blRadiusX | |
| && other.blRadiusY == blRadiusY | |
| && other.brRadiusX == brRadiusX | |
| && other.brRadiusY == brRadiusY; | |
| } | |
| @override | |
| int get hashCode => Object.hash(left, top, right, bottom, | |
| tlRadiusX, tlRadiusY, trRadiusX, trRadiusY, | |
| blRadiusX, blRadiusY, brRadiusX, brRadiusY); | |
| @override | |
| String toString() { | |
| final String rect = '${left.toStringAsFixed(1)}, ' | |
| '${top.toStringAsFixed(1)}, ' | |
| '${right.toStringAsFixed(1)}, ' | |
| '${bottom.toStringAsFixed(1)}'; | |
| if (tlRadius == trRadius && | |
| trRadius == brRadius && | |
| brRadius == blRadius) { | |
| if (tlRadius.x == tlRadius.y) { | |
| return 'RRect.fromLTRBR($rect, ${tlRadius.x.toStringAsFixed(1)})'; | |
| } | |
| return 'RRect.fromLTRBXY($rect, ${tlRadius.x.toStringAsFixed(1)}, ${tlRadius.y.toStringAsFixed(1)})'; | |
| } | |
| return 'RRect.fromLTRBAndCorners(' | |
| '$rect, ' | |
| 'topLeft: $tlRadius, ' | |
| 'topRight: $trRadius, ' | |
| 'bottomRight: $brRadius, ' | |
| 'bottomLeft: $blRadius' | |
| ')'; | |
| } | |
| } | |
| /// A transform consisting of a translation, a rotation, and a uniform scale. | |
| /// | |
| /// Used by [Canvas.drawAtlas]. This is a more efficient way to represent these | |
| /// simple transformations than a full matrix. | |
| // Modeled after Skia's SkRSXform. | |
| class RSTransform { | |
| /// Creates an RSTransform. | |
| /// | |
| /// An [RSTransform] expresses the combination of a translation, a rotation | |
| /// around a particular point, and a scale factor. | |
| /// | |
| /// The first argument, `scos`, is the cosine of the rotation, multiplied by | |
| /// the scale factor. | |
| /// | |
| /// The second argument, `ssin`, is the sine of the rotation, multiplied by | |
| /// that same scale factor. | |
| /// | |
| /// The third argument is the x coordinate of the translation, minus the | |
| /// `scos` argument multiplied by the x-coordinate of the rotation point, plus | |
| /// the `ssin` argument multiplied by the y-coordinate of the rotation point. | |
| /// | |
| /// The fourth argument is the y coordinate of the translation, minus the `ssin` | |
| /// argument multiplied by the x-coordinate of the rotation point, minus the | |
| /// `scos` argument multiplied by the y-coordinate of the rotation point. | |
| /// | |
| /// The [RSTransform.fromComponents] method may be a simpler way to | |
| /// construct these values. However, if there is a way to factor out the | |
| /// computations of the sine and cosine of the rotation so that they can be | |
| /// reused over multiple calls to this constructor, it may be more efficient | |
| /// to directly use this constructor instead. | |
| RSTransform(double scos, double ssin, double tx, double ty) { | |
| _value | |
| ..[0] = scos | |
| ..[1] = ssin | |
| ..[2] = tx | |
| ..[3] = ty; | |
| } | |
| /// Creates an RSTransform from its individual components. | |
| /// | |
| /// The `rotation` parameter gives the rotation in radians. | |
| /// | |
| /// The `scale` parameter describes the uniform scale factor. | |
| /// | |
| /// The `anchorX` and `anchorY` parameters give the coordinate of the point | |
| /// around which to rotate. | |
| /// | |
| /// The `translateX` and `translateY` parameters give the coordinate of the | |
| /// offset by which to translate. | |
| /// | |
| /// This constructor computes the arguments of the [RSTransform.new] | |
| /// constructor and then defers to that constructor to actually create the | |
| /// object. If many [RSTransform] objects are being created and there is a way | |
| /// to factor out the computations of the sine and cosine of the rotation | |
| /// (which are computed each time this constructor is called) and reuse them | |
| /// over multiple [RSTransform] objects, it may be more efficient to directly | |
| /// use the more direct [RSTransform.new] constructor instead. | |
| factory RSTransform.fromComponents({ | |
| required double rotation, | |
| required double scale, | |
| required double anchorX, | |
| required double anchorY, | |
| required double translateX, | |
| required double translateY | |
| }) { | |
| final double scos = math.cos(rotation) * scale; | |
| final double ssin = math.sin(rotation) * scale; | |
| final double tx = translateX + -scos * anchorX + ssin * anchorY; | |
| final double ty = translateY + -ssin * anchorX - scos * anchorY; | |
| return RSTransform(scos, ssin, tx, ty); | |
| } | |
| final Float32List _value = Float32List(4); | |
| /// The cosine of the rotation multiplied by the scale factor. | |
| double get scos => _value[0]; | |
| /// The sine of the rotation multiplied by that same scale factor. | |
| double get ssin => _value[1]; | |
| /// The x coordinate of the translation, minus [scos] multiplied by the | |
| /// x-coordinate of the rotation point, plus [ssin] multiplied by the | |
| /// y-coordinate of the rotation point. | |
| double get tx => _value[2]; | |
| /// The y coordinate of the translation, minus [ssin] multiplied by the | |
| /// x-coordinate of the rotation point, minus [scos] multiplied by the | |
| /// y-coordinate of the rotation point. | |
| double get ty => _value[3]; | |
| } |