diff options
Diffstat (limited to 'layout/base/UnitTransforms.h')
-rw-r--r-- | layout/base/UnitTransforms.h | 294 |
1 files changed, 294 insertions, 0 deletions
diff --git a/layout/base/UnitTransforms.h b/layout/base/UnitTransforms.h new file mode 100644 index 000000000..e3a5af2d2 --- /dev/null +++ b/layout/base/UnitTransforms.h @@ -0,0 +1,294 @@ +/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ +/* vim: set ts=8 sts=2 et sw=2 tw=80: */ +/* This Source Code Form is subject to the terms of the Mozilla Public + * License, v. 2.0. If a copy of the MPL was not distributed with this + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */ + +#ifndef MOZ_UNIT_TRANSFORMS_H_ +#define MOZ_UNIT_TRANSFORMS_H_ + +#include "Units.h" +#include "mozilla/gfx/Matrix.h" +#include "mozilla/Maybe.h" +#include "nsRegion.h" + +namespace mozilla { + +// Convenience functions for converting an entity from one strongly-typed +// coordinate system to another without changing the values it stores (this +// can be thought of as a cast). +// To use these functions, you must provide a justification for each use! +// Feel free to add more justifications to PixelCastJustification, along with +// a comment that explains under what circumstances it is appropriate to use. + +enum class PixelCastJustification : uint8_t { + // For the root layer, Screen Pixel = Parent Layer Pixel. + ScreenIsParentLayerForRoot, + // On the layout side, Screen Pixel = LayoutDevice at the outer-window level. + LayoutDeviceIsScreenForBounds, + // For the root layer, Render Target Pixel = Parent Layer Pixel. + RenderTargetIsParentLayerForRoot, + // For the root composition size we want to view it as layer pixels in any layer + ParentLayerToLayerForRootComposition, + // The Layer coordinate space for one layer is the ParentLayer coordinate + // space for its children + MovingDownToChildren, + // The transform that is usually used to convert between two coordinate + // systems is not available (for example, because the object that stores it + // is being destroyed), so fall back to the identity. + TransformNotAvailable, + // When an OS event is initially constructed, its reference point is + // technically in screen pixels, as it has not yet accounted for any + // asynchronous transforms. This justification is for viewing the initial + // reference point as a screen point. The reverse is useful when synthetically + // created WidgetEvents need to be converted back to InputData. + LayoutDeviceIsScreenForUntransformedEvent, + // Similar to LayoutDeviceIsScreenForUntransformedEvent, PBrowser handles + // some widget/tab dimension information as the OS does -- in screen units. + LayoutDeviceIsScreenForTabDims, + // A combination of LayoutDeviceIsScreenForBounds and + // ScreenIsParentLayerForRoot, which is how we're using it. + LayoutDeviceIsParentLayerForRCDRSF, + // Used to treat the product of AsyncTransformComponentMatrix objects + // as an AsyncTransformMatrix. See the definitions of these matrices in + // LayersTypes.h for details. + MultipleAsyncTransforms, + // We have reason to believe a layer doesn't have a local transform. + // Should only be used if we've already checked or asserted this. + NoTransformOnLayer +}; + +template <class TargetUnits, class SourceUnits> +gfx::CoordTyped<TargetUnits> ViewAs(const gfx::CoordTyped<SourceUnits>& aCoord, PixelCastJustification) { + return gfx::CoordTyped<TargetUnits>(aCoord.value); +} +template <class TargetUnits, class SourceUnits> +gfx::SizeTyped<TargetUnits> ViewAs(const gfx::SizeTyped<SourceUnits>& aSize, PixelCastJustification) { + return gfx::SizeTyped<TargetUnits>(aSize.width, aSize.height); +} +template <class TargetUnits, class SourceUnits> +gfx::IntSizeTyped<TargetUnits> ViewAs(const gfx::IntSizeTyped<SourceUnits>& aSize, PixelCastJustification) { + return gfx::IntSizeTyped<TargetUnits>(aSize.width, aSize.height); +} +template <class TargetUnits, class SourceUnits> +gfx::PointTyped<TargetUnits> ViewAs(const gfx::PointTyped<SourceUnits>& aPoint, PixelCastJustification) { + return gfx::PointTyped<TargetUnits>(aPoint.x, aPoint.y); +} +template <class TargetUnits, class SourceUnits> +gfx::IntPointTyped<TargetUnits> ViewAs(const gfx::IntPointTyped<SourceUnits>& aPoint, PixelCastJustification) { + return gfx::IntPointTyped<TargetUnits>(aPoint.x, aPoint.y); +} +template <class TargetUnits, class SourceUnits> +gfx::RectTyped<TargetUnits> ViewAs(const gfx::RectTyped<SourceUnits>& aRect, PixelCastJustification) { + return gfx::RectTyped<TargetUnits>(aRect.x, aRect.y, aRect.width, aRect.height); +} +template <class TargetUnits, class SourceUnits> +gfx::IntRectTyped<TargetUnits> ViewAs(const gfx::IntRectTyped<SourceUnits>& aRect, PixelCastJustification) { + return gfx::IntRectTyped<TargetUnits>(aRect.x, aRect.y, aRect.width, aRect.height); +} +template <class TargetUnits, class SourceUnits> +gfx::MarginTyped<TargetUnits> ViewAs(const gfx::MarginTyped<SourceUnits>& aMargin, PixelCastJustification) { + return gfx::MarginTyped<TargetUnits>(aMargin.top, aMargin.right, aMargin.bottom, aMargin.left); +} +template <class TargetUnits, class SourceUnits> +gfx::IntMarginTyped<TargetUnits> ViewAs(const gfx::IntMarginTyped<SourceUnits>& aMargin, PixelCastJustification) { + return gfx::IntMarginTyped<TargetUnits>(aMargin.top, aMargin.right, aMargin.bottom, aMargin.left); +} +template <class TargetUnits, class SourceUnits> +gfx::IntRegionTyped<TargetUnits> ViewAs(const gfx::IntRegionTyped<SourceUnits>& aRegion, PixelCastJustification) { + return gfx::IntRegionTyped<TargetUnits>::FromUnknownRegion(aRegion.ToUnknownRegion()); +} +template <class NewTargetUnits, class OldTargetUnits, class SourceUnits> +gfx::ScaleFactor<SourceUnits, NewTargetUnits> ViewTargetAs( + const gfx::ScaleFactor<SourceUnits, OldTargetUnits>& aScaleFactor, + PixelCastJustification) { + return gfx::ScaleFactor<SourceUnits, NewTargetUnits>(aScaleFactor.scale); +} +// Unlike the other functions in this category, this function takes the +// target matrix type, rather than its source and target unit types, as +// the explicit template argument, so an example invocation is: +// ViewAs<ScreenToLayerMatrix4x4>(otherTypedMatrix, justification) +// The reason is that if it took the source and target unit types as two +// template arguments, there may be some confusion as to which is the +// source and which is the target. +template <class TargetMatrix, class SourceMatrixSourceUnits, class SourceMatrixTargetUnits> +TargetMatrix ViewAs( + const gfx::Matrix4x4Typed<SourceMatrixSourceUnits, SourceMatrixTargetUnits>& aMatrix, + PixelCastJustification) { + return TargetMatrix::FromUnknownMatrix(aMatrix.ToUnknownMatrix()); +} + +// Convenience functions for casting untyped entities to typed entities. +// Using these functions does not require a justification, but once we convert +// all code to use strongly typed units they should not be needed any longer. +template <class TargetUnits> +gfx::PointTyped<TargetUnits> ViewAs(const gfxPoint& aPoint) { + return gfx::PointTyped<TargetUnits>(aPoint.x, aPoint.y); +} +template <class TargetUnits> +gfx::PointTyped<TargetUnits> ViewAs(const gfx::Point& aPoint) { + return gfx::PointTyped<TargetUnits>(aPoint.x, aPoint.y); +} +template <class TargetUnits> +gfx::RectTyped<TargetUnits> ViewAs(const gfx::Rect& aRect) { + return gfx::RectTyped<TargetUnits>(aRect.x, aRect.y, aRect.width, aRect.height); +} +template <class TargetUnits> +gfx::IntSizeTyped<TargetUnits> ViewAs(const nsIntSize& aSize) { + return gfx::IntSizeTyped<TargetUnits>(aSize.width, aSize.height); +} +template <class TargetUnits> +gfx::IntPointTyped<TargetUnits> ViewAs(const nsIntPoint& aPoint) { + return gfx::IntPointTyped<TargetUnits>(aPoint.x, aPoint.y); +} +template <class TargetUnits> +gfx::IntRectTyped<TargetUnits> ViewAs(const nsIntRect& aRect) { + return gfx::IntRectTyped<TargetUnits>(aRect.x, aRect.y, aRect.width, aRect.height); +} +template <class TargetUnits> +gfx::IntRegionTyped<TargetUnits> ViewAs(const nsIntRegion& aRegion) { + return gfx::IntRegionTyped<TargetUnits>::FromUnknownRegion(aRegion); +} +// Unlike the other functions in this category, this function takes the +// target matrix type, rather than its source and target unit types, as +// the template argument, so an example invocation is: +// ViewAs<ScreenToLayerMatrix4x4>(untypedMatrix) +// The reason is that if it took the source and target unit types as two +// template arguments, there may be some confusion as to which is the +// source and which is the target. +template <class TypedMatrix> +TypedMatrix ViewAs(const gfx::Matrix4x4& aMatrix) { + return TypedMatrix::FromUnknownMatrix(aMatrix); +} + +// Convenience functions for transforming an entity from one strongly-typed +// coordinate system to another using the provided transformation matrix. +template <typename TargetUnits, typename SourceUnits> +static gfx::PointTyped<TargetUnits> +TransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::PointTyped<SourceUnits>& aPoint) +{ + return aTransform.TransformPoint(aPoint); +} +template <typename TargetUnits, typename SourceUnits> +static gfx::IntPointTyped<TargetUnits> +TransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::IntPointTyped<SourceUnits>& aPoint) +{ + return RoundedToInt(TransformBy(aTransform, gfx::PointTyped<SourceUnits>(aPoint))); +} +template <typename TargetUnits, typename SourceUnits> +static gfx::RectTyped<TargetUnits> +TransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::RectTyped<SourceUnits>& aRect) +{ + return aTransform.TransformBounds(aRect); +} +template <typename TargetUnits, typename SourceUnits> +static gfx::IntRectTyped<TargetUnits> +TransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::IntRectTyped<SourceUnits>& aRect) +{ + return RoundedToInt(TransformBy(aTransform, gfx::RectTyped<SourceUnits>(aRect))); +} +template <typename TargetUnits, typename SourceUnits> +static gfx::IntRegionTyped<TargetUnits> +TransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::IntRegionTyped<SourceUnits>& aRegion) +{ + return ViewAs<TargetUnits>(aRegion.ToUnknownRegion().Transform( + aTransform.ToUnknownMatrix())); +} + +// Transform |aVector|, which is anchored at |aAnchor|, by the given transform +// matrix, yielding a point in |TargetUnits|. +// The anchor is necessary because with 3D tranforms, the location of the +// vector can affect the result of the transform. +template <typename TargetUnits, typename SourceUnits> +static gfx::PointTyped<TargetUnits> +TransformVector(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::PointTyped<SourceUnits>& aVector, + const gfx::PointTyped<SourceUnits>& aAnchor) +{ + gfx::PointTyped<TargetUnits> transformedStart = TransformBy(aTransform, aAnchor); + gfx::PointTyped<TargetUnits> transformedEnd = TransformBy(aTransform, aAnchor + aVector); + return transformedEnd - transformedStart; +} + +// UntransformBy() and UntransformVector() are like TransformBy() and +// TransformVector(), respectively, but are intended for cases where +// the transformation matrix is the inverse of a 3D projection. When +// using such transforms, the resulting Point4D is only meaningful +// if it has a positive w-coordinate. To handle this, these functions +// return a Maybe object which contains a value if and only if the +// result is meaningful +template <typename TargetUnits, typename SourceUnits> +static Maybe<gfx::PointTyped<TargetUnits>> +UntransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::PointTyped<SourceUnits>& aPoint) +{ + gfx::Point4DTyped<TargetUnits> point = aTransform.ProjectPoint(aPoint); + if (!point.HasPositiveWCoord()) { + return Nothing(); + } + return Some(point.As2DPoint()); +} +template <typename TargetUnits, typename SourceUnits> +static Maybe<gfx::IntPointTyped<TargetUnits>> +UntransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::IntPointTyped<SourceUnits>& aPoint) +{ + gfx::PointTyped<SourceUnits> p = aPoint; + gfx::Point4DTyped<TargetUnits> point = aTransform.ProjectPoint(p); + if (!point.HasPositiveWCoord()) { + return Nothing(); + } + return Some(RoundedToInt(point.As2DPoint())); +} + +// The versions of UntransformBy() that take a rectangle also take a clip, +// which represents the bounds within which the target must fall. The +// result of the transform is intersected with this clip, and is considered +// meaningful if the intersection is not empty. +template <typename TargetUnits, typename SourceUnits> +static Maybe<gfx::RectTyped<TargetUnits>> +UntransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::RectTyped<SourceUnits>& aRect, + const gfx::RectTyped<TargetUnits>& aClip) +{ + gfx::RectTyped<TargetUnits> rect = aTransform.ProjectRectBounds(aRect, aClip); + if (rect.IsEmpty()) { + return Nothing(); + } + return Some(rect); +} +template <typename TargetUnits, typename SourceUnits> +static Maybe<gfx::IntRectTyped<TargetUnits>> +UntransformBy(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::IntRectTyped<SourceUnits>& aRect, + const gfx::IntRectTyped<TargetUnits>& aClip) +{ + gfx::RectTyped<TargetUnits> rect = aTransform.ProjectRectBounds(aRect, aClip); + if (rect.IsEmpty()) { + return Nothing(); + } + return Some(RoundedToInt(rect)); +} + +template <typename TargetUnits, typename SourceUnits> +static Maybe<gfx::PointTyped<TargetUnits>> +UntransformVector(const gfx::Matrix4x4Typed<SourceUnits, TargetUnits>& aTransform, + const gfx::PointTyped<SourceUnits>& aVector, + const gfx::PointTyped<SourceUnits>& aAnchor) +{ + gfx::Point4DTyped<TargetUnits> projectedAnchor = aTransform.ProjectPoint(aAnchor); + gfx::Point4DTyped<TargetUnits> projectedTarget = aTransform.ProjectPoint(aAnchor + aVector); + if (!projectedAnchor.HasPositiveWCoord() || !projectedTarget.HasPositiveWCoord()){ + return Nothing(); + } + return Some(projectedTarget.As2DPoint() - projectedAnchor.As2DPoint()); +} + +} // namespace mozilla + +#endif |