diff options
Diffstat (limited to 'image/ClippedImage.cpp')
-rw-r--r-- | image/ClippedImage.cpp | 516 |
1 files changed, 516 insertions, 0 deletions
diff --git a/image/ClippedImage.cpp b/image/ClippedImage.cpp new file mode 100644 index 0000000000..12e38bdde8 --- /dev/null +++ b/image/ClippedImage.cpp @@ -0,0 +1,516 @@ +/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ +/* 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/. */ + +#include "ClippedImage.h" + +#include <algorithm> +#include <cmath> +#include <new> // Workaround for bug in VS10; see bug 981264. +#include <utility> + +#include "ImageRegion.h" +#include "Orientation.h" +#include "gfxContext.h" +#include "gfxDrawable.h" +#include "gfxPlatform.h" +#include "gfxUtils.h" +#include "mozilla/RefPtr.h" +#include "mozilla/SVGImageContext.h" +#include "mozilla/Tuple.h" +#include "mozilla/gfx/2D.h" + +namespace mozilla { + +using namespace gfx; +using std::max; + +namespace image { + +class ClippedImageCachedSurface { + public: + ClippedImageCachedSurface(already_AddRefed<SourceSurface> aSurface, + const nsIntSize& aSize, + const SVGImageContext& aSVGContext, float aFrame, + uint32_t aFlags, ImgDrawResult aDrawResult) + : mSurface(aSurface), + mSize(aSize), + mSVGContext(aSVGContext), + mFrame(aFrame), + mFlags(aFlags), + mDrawResult(aDrawResult) { + MOZ_ASSERT(mSurface, "Must have a valid surface"); + } + + bool Matches(const nsIntSize& aSize, const SVGImageContext& aSVGContext, + float aFrame, uint32_t aFlags) const { + return mSize == aSize && mSVGContext == aSVGContext && mFrame == aFrame && + mFlags == aFlags; + } + + already_AddRefed<SourceSurface> Surface() const { + RefPtr<SourceSurface> surf(mSurface); + return surf.forget(); + } + + ImgDrawResult GetDrawResult() const { return mDrawResult; } + + bool NeedsRedraw() const { + return mDrawResult != ImgDrawResult::SUCCESS && + mDrawResult != ImgDrawResult::BAD_IMAGE; + } + + private: + RefPtr<SourceSurface> mSurface; + const nsIntSize mSize; + SVGImageContext mSVGContext; + const float mFrame; + const uint32_t mFlags; + const ImgDrawResult mDrawResult; +}; + +class DrawSingleTileCallback : public gfxDrawingCallback { + public: + DrawSingleTileCallback(ClippedImage* aImage, const nsIntSize& aSize, + const SVGImageContext& aSVGContext, + uint32_t aWhichFrame, uint32_t aFlags, float aOpacity) + : mImage(aImage), + mSize(aSize), + mSVGContext(aSVGContext), + mWhichFrame(aWhichFrame), + mFlags(aFlags), + mDrawResult(ImgDrawResult::NOT_READY), + mOpacity(aOpacity) { + MOZ_ASSERT(mImage, "Must have an image to clip"); + } + + virtual bool operator()(gfxContext* aContext, const gfxRect& aFillRect, + const SamplingFilter aSamplingFilter, + const gfxMatrix& aTransform) override { + MOZ_ASSERT(aTransform.IsIdentity(), + "Caller is probably CreateSamplingRestrictedDrawable, " + "which should not happen"); + + // Draw the image. |gfxCallbackDrawable| always calls this function with + // arguments that guarantee we never tile. + mDrawResult = mImage->DrawSingleTile( + aContext, mSize, ImageRegion::Create(aFillRect), mWhichFrame, + aSamplingFilter, mSVGContext, mFlags, mOpacity); + + return true; + } + + ImgDrawResult GetDrawResult() { return mDrawResult; } + + private: + RefPtr<ClippedImage> mImage; + const nsIntSize mSize; + const SVGImageContext& mSVGContext; + const uint32_t mWhichFrame; + const uint32_t mFlags; + ImgDrawResult mDrawResult; + float mOpacity; +}; + +ClippedImage::ClippedImage(Image* aImage, nsIntRect aClip, + const Maybe<nsSize>& aSVGViewportSize) + : ImageWrapper(aImage), mClip(aClip) { + MOZ_ASSERT(aImage != nullptr, "ClippedImage requires an existing Image"); + MOZ_ASSERT_IF(aSVGViewportSize, + aImage->GetType() == imgIContainer::TYPE_VECTOR); + if (aSVGViewportSize) { + mSVGViewportSize = + Some(aSVGViewportSize->ToNearestPixels(AppUnitsPerCSSPixel())); + } +} + +ClippedImage::~ClippedImage() {} + +bool ClippedImage::ShouldClip() { + // We need to evaluate the clipping region against the image's width and + // height once they're available to determine if it's valid and whether we + // actually need to do any work. We may fail if the image's width and height + // aren't available yet, in which case we'll try again later. + if (mShouldClip.isNothing()) { + int32_t width, height; + RefPtr<ProgressTracker> progressTracker = + InnerImage()->GetProgressTracker(); + if (InnerImage()->HasError()) { + // If there's a problem with the inner image we'll let it handle + // everything. + mShouldClip.emplace(false); + } else if (mSVGViewportSize && !mSVGViewportSize->IsEmpty()) { + // Clamp the clipping region to the size of the SVG viewport. + nsIntRect svgViewportRect(nsIntPoint(0, 0), *mSVGViewportSize); + + mClip = mClip.Intersect(svgViewportRect); + + // If the clipping region is the same size as the SVG viewport size + // we don't have to do anything. + mShouldClip.emplace(!mClip.IsEqualInterior(svgViewportRect)); + } else if (NS_SUCCEEDED(InnerImage()->GetWidth(&width)) && width > 0 && + NS_SUCCEEDED(InnerImage()->GetHeight(&height)) && height > 0) { + // Clamp the clipping region to the size of the underlying image. + mClip = mClip.Intersect(nsIntRect(0, 0, width, height)); + + // If the clipping region is the same size as the underlying image we + // don't have to do anything. + mShouldClip.emplace( + !mClip.IsEqualInterior(nsIntRect(0, 0, width, height))); + } else if (progressTracker && + !(progressTracker->GetProgress() & FLAG_LOAD_COMPLETE)) { + // The image just hasn't finished loading yet. We don't yet know whether + // clipping with be needed or not for now. Just return without memorizing + // anything. + return false; + } else { + // We have a fully loaded image without a clearly defined width and + // height. This can happen with SVG images. + mShouldClip.emplace(false); + } + } + + MOZ_ASSERT(mShouldClip.isSome(), "Should have computed a result"); + return *mShouldClip; +} + +NS_IMETHODIMP +ClippedImage::GetWidth(int32_t* aWidth) { + if (!ShouldClip()) { + return InnerImage()->GetWidth(aWidth); + } + + *aWidth = mClip.Width(); + return NS_OK; +} + +NS_IMETHODIMP +ClippedImage::GetHeight(int32_t* aHeight) { + if (!ShouldClip()) { + return InnerImage()->GetHeight(aHeight); + } + + *aHeight = mClip.Height(); + return NS_OK; +} + +NS_IMETHODIMP +ClippedImage::GetIntrinsicSize(nsSize* aSize) { + if (!ShouldClip()) { + return InnerImage()->GetIntrinsicSize(aSize); + } + + *aSize = nsSize(mClip.Width(), mClip.Height()); + return NS_OK; +} + +Maybe<AspectRatio> ClippedImage::GetIntrinsicRatio() { + if (!ShouldClip()) { + return InnerImage()->GetIntrinsicRatio(); + } + + return Some(AspectRatio::FromSize(mClip.Width(), mClip.Height())); +} + +NS_IMETHODIMP_(already_AddRefed<SourceSurface>) +ClippedImage::GetFrame(uint32_t aWhichFrame, uint32_t aFlags) { + ImgDrawResult result; + RefPtr<SourceSurface> surface; + Tie(result, surface) = GetFrameInternal(mClip.Size(), SVGImageContext(), + Nothing(), aWhichFrame, aFlags, 1.0); + return surface.forget(); +} + +NS_IMETHODIMP_(already_AddRefed<SourceSurface>) +ClippedImage::GetFrameAtSize(const IntSize& aSize, uint32_t aWhichFrame, + uint32_t aFlags) { + // XXX(seth): It'd be nice to support downscale-during-decode for this case, + // but right now we just fall back to the intrinsic size. + return GetFrame(aWhichFrame, aFlags); +} + +std::pair<ImgDrawResult, RefPtr<SourceSurface>> ClippedImage::GetFrameInternal( + const nsIntSize& aSize, const SVGImageContext& aSVGContext, + const Maybe<ImageIntRegion>& aRegion, uint32_t aWhichFrame, uint32_t aFlags, + float aOpacity) { + if (!ShouldClip()) { + RefPtr<SourceSurface> surface = InnerImage()->GetFrame(aWhichFrame, aFlags); + return std::make_pair( + surface ? ImgDrawResult::SUCCESS : ImgDrawResult::NOT_READY, + std::move(surface)); + } + + float frameToDraw = InnerImage()->GetFrameIndex(aWhichFrame); + if (!mCachedSurface || + !mCachedSurface->Matches(aSize, aSVGContext, frameToDraw, aFlags) || + mCachedSurface->NeedsRedraw()) { + // Create a surface to draw into. + RefPtr<DrawTarget> target = + gfxPlatform::GetPlatform()->CreateOffscreenContentDrawTarget( + IntSize(aSize.width, aSize.height), SurfaceFormat::OS_RGBA); + if (!target || !target->IsValid()) { + NS_ERROR("Could not create a DrawTarget"); + return std::make_pair(ImgDrawResult::TEMPORARY_ERROR, + RefPtr<SourceSurface>()); + } + + RefPtr<gfxContext> ctx = gfxContext::CreateOrNull(target); + MOZ_ASSERT(ctx); // already checked the draw target above + + // Create our callback. + RefPtr<DrawSingleTileCallback> drawTileCallback = + new DrawSingleTileCallback(this, aSize, aSVGContext, aWhichFrame, + aFlags, aOpacity); + RefPtr<gfxDrawable> drawable = + new gfxCallbackDrawable(drawTileCallback, aSize); + + // Actually draw. The callback will end up invoking DrawSingleTile. + gfxUtils::DrawPixelSnapped(ctx, drawable, SizeDouble(aSize), + ImageRegion::Create(aSize), + SurfaceFormat::OS_RGBA, SamplingFilter::LINEAR, + imgIContainer::FLAG_CLAMP); + + // Cache the resulting surface. + mCachedSurface = MakeUnique<ClippedImageCachedSurface>( + target->Snapshot(), aSize, aSVGContext, frameToDraw, aFlags, + drawTileCallback->GetDrawResult()); + } + + MOZ_ASSERT(mCachedSurface, "Should have a cached surface now"); + RefPtr<SourceSurface> surface = mCachedSurface->Surface(); + return std::make_pair(mCachedSurface->GetDrawResult(), std::move(surface)); +} + +NS_IMETHODIMP_(bool) +ClippedImage::IsImageContainerAvailable(WindowRenderer* aRenderer, + uint32_t aFlags) { + if (!ShouldClip()) { + return InnerImage()->IsImageContainerAvailable(aRenderer, aFlags); + } + return false; +} + +NS_IMETHODIMP_(ImgDrawResult) +ClippedImage::GetImageProvider(WindowRenderer* aRenderer, + const gfx::IntSize& aSize, + const SVGImageContext& aSVGContext, + const Maybe<ImageIntRegion>& aRegion, + uint32_t aFlags, + WebRenderImageProvider** aProvider) { + // XXX(seth): We currently don't have a way of clipping the result of + // GetImageContainer. We work around this by always returning null, but if it + // ever turns out that ClippedImage is widely used on codepaths that can + // actually benefit from GetImageContainer, it would be a good idea to fix + // that method for performance reasons. + + if (!ShouldClip()) { + return InnerImage()->GetImageProvider(aRenderer, aSize, aSVGContext, + aRegion, aFlags, aProvider); + } + + return ImgDrawResult::NOT_SUPPORTED; +} + +static bool MustCreateSurface(gfxContext* aContext, const nsIntSize& aSize, + const ImageRegion& aRegion, + const uint32_t aFlags) { + gfxRect imageRect(0, 0, aSize.width, aSize.height); + bool willTile = !imageRect.Contains(aRegion.Rect()) && + !(aFlags & imgIContainer::FLAG_CLAMP); + bool willResample = aContext->CurrentMatrix().HasNonIntegerTranslation() && + (willTile || !aRegion.RestrictionContains(imageRect)); + return willTile || willResample; +} + +NS_IMETHODIMP_(ImgDrawResult) +ClippedImage::Draw(gfxContext* aContext, const nsIntSize& aSize, + const ImageRegion& aRegion, uint32_t aWhichFrame, + SamplingFilter aSamplingFilter, + const SVGImageContext& aSVGContext, uint32_t aFlags, + float aOpacity) { + if (!ShouldClip()) { + return InnerImage()->Draw(aContext, aSize, aRegion, aWhichFrame, + aSamplingFilter, aSVGContext, aFlags, aOpacity); + } + + // Check for tiling. If we need to tile then we need to create a + // gfxCallbackDrawable to handle drawing for us. + if (MustCreateSurface(aContext, aSize, aRegion, aFlags)) { + // Create a temporary surface containing a single tile of this image. + // GetFrame will call DrawSingleTile internally. + ImgDrawResult result; + RefPtr<SourceSurface> surface; + Tie(result, surface) = GetFrameInternal(aSize, aSVGContext, Nothing(), + aWhichFrame, aFlags, aOpacity); + if (!surface) { + MOZ_ASSERT(result != ImgDrawResult::SUCCESS); + return result; + } + + // Create a drawable from that surface. + RefPtr<gfxSurfaceDrawable> drawable = + new gfxSurfaceDrawable(surface, aSize); + + // Draw. + gfxUtils::DrawPixelSnapped(aContext, drawable, SizeDouble(aSize), aRegion, + SurfaceFormat::OS_RGBA, aSamplingFilter, + aOpacity); + + return result; + } + + return DrawSingleTile(aContext, aSize, aRegion, aWhichFrame, aSamplingFilter, + aSVGContext, aFlags, aOpacity); +} + +ImgDrawResult ClippedImage::DrawSingleTile( + gfxContext* aContext, const nsIntSize& aSize, const ImageRegion& aRegion, + uint32_t aWhichFrame, SamplingFilter aSamplingFilter, + const SVGImageContext& aSVGContext, uint32_t aFlags, float aOpacity) { + MOZ_ASSERT(!MustCreateSurface(aContext, aSize, aRegion, aFlags), + "Shouldn't need to create a surface"); + + gfxRect clip(mClip.X(), mClip.Y(), mClip.Width(), mClip.Height()); + nsIntSize size(aSize), innerSize(aSize); + bool needScale = false; + if (mSVGViewportSize && !mSVGViewportSize->IsEmpty()) { + innerSize = *mSVGViewportSize; + needScale = true; + } else if (NS_SUCCEEDED(InnerImage()->GetWidth(&innerSize.width)) && + NS_SUCCEEDED(InnerImage()->GetHeight(&innerSize.height))) { + needScale = true; + } else { + MOZ_ASSERT_UNREACHABLE( + "If ShouldClip() led us to draw then we should never get here"); + } + + if (needScale) { + double scaleX = aSize.width / clip.Width(); + double scaleY = aSize.height / clip.Height(); + + // Map the clip and size to the scale requested by the caller. + clip.Scale(scaleX, scaleY); + size = innerSize; + size.Scale(scaleX, scaleY); + } + + // We restrict our drawing to only the clipping region, and translate so that + // the clipping region is placed at the position the caller expects. + ImageRegion region(aRegion); + region.MoveBy(clip.X(), clip.Y()); + region = region.Intersect(clip); + + gfxContextMatrixAutoSaveRestore saveMatrix(aContext); + aContext->Multiply(gfxMatrix::Translation(-clip.X(), -clip.Y())); + + auto unclipViewport = [&](const SVGImageContext& aOldContext) { + // Map the viewport to the inner image. Note that we don't take the aSize + // parameter of imgIContainer::Draw into account, just the clipping region. + // The size in pixels at which the output will ultimately be drawn is + // irrelevant here since the purpose of the SVG viewport size is to + // determine what *region* of the SVG document will be drawn. + SVGImageContext context(aOldContext); + auto oldViewport = aOldContext.GetViewportSize(); + if (oldViewport) { + CSSIntSize newViewport; + newViewport.width = + ceil(oldViewport->width * double(innerSize.width) / mClip.Width()); + newViewport.height = + ceil(oldViewport->height * double(innerSize.height) / mClip.Height()); + context.SetViewportSize(Some(newViewport)); + } + return context; + }; + + return InnerImage()->Draw(aContext, size, region, aWhichFrame, + aSamplingFilter, unclipViewport(aSVGContext), + aFlags, aOpacity); +} + +NS_IMETHODIMP +ClippedImage::RequestDiscard() { + // We're very aggressive about discarding. + mCachedSurface = nullptr; + + return InnerImage()->RequestDiscard(); +} + +NS_IMETHODIMP_(Orientation) +ClippedImage::GetOrientation() { + // XXX(seth): This should not actually be here; this is just to work around a + // what appears to be a bug in MSVC's linker. + return InnerImage()->GetOrientation(); +} + +nsIntSize ClippedImage::OptimalImageSizeForDest(const gfxSize& aDest, + uint32_t aWhichFrame, + SamplingFilter aSamplingFilter, + uint32_t aFlags) { + if (!ShouldClip()) { + return InnerImage()->OptimalImageSizeForDest(aDest, aWhichFrame, + aSamplingFilter, aFlags); + } + + int32_t imgWidth, imgHeight; + bool needScale = false; + bool forceUniformScaling = false; + if (mSVGViewportSize && !mSVGViewportSize->IsEmpty()) { + imgWidth = mSVGViewportSize->width; + imgHeight = mSVGViewportSize->height; + needScale = true; + forceUniformScaling = (aFlags & imgIContainer::FLAG_FORCE_UNIFORM_SCALING); + } else if (NS_SUCCEEDED(InnerImage()->GetWidth(&imgWidth)) && + NS_SUCCEEDED(InnerImage()->GetHeight(&imgHeight))) { + needScale = true; + } + + if (needScale) { + // To avoid ugly sampling artifacts, ClippedImage needs the image size to + // be chosen such that the clipping region lies on pixel boundaries. + + // First, we select a scale that's good for ClippedImage. An integer + // multiple of the size of the clipping region is always fine. + IntSize scale = IntSize::Ceil(aDest.width / mClip.Width(), + aDest.height / mClip.Height()); + + if (forceUniformScaling) { + scale.width = scale.height = max(scale.height, scale.width); + } + + // Determine the size we'd prefer to render the inner image at, and ask the + // inner image what size we should actually use. + gfxSize desiredSize(double(imgWidth) * scale.width, + double(imgHeight) * scale.height); + nsIntSize innerDesiredSize = InnerImage()->OptimalImageSizeForDest( + desiredSize, aWhichFrame, aSamplingFilter, aFlags); + + // To get our final result, we take the inner image's desired size and + // determine how large the clipped region would be at that scale. (Again, we + // ensure an integer multiple of the size of the clipping region.) + IntSize finalScale = + IntSize::Ceil(double(innerDesiredSize.width) / imgWidth, + double(innerDesiredSize.height) / imgHeight); + return mClip.Size() * finalScale; + } + + MOZ_ASSERT(false, + "If ShouldClip() led us to draw then we should never get here"); + return InnerImage()->OptimalImageSizeForDest(aDest, aWhichFrame, + aSamplingFilter, aFlags); +} + +NS_IMETHODIMP_(nsIntRect) +ClippedImage::GetImageSpaceInvalidationRect(const nsIntRect& aRect) { + if (!ShouldClip()) { + return InnerImage()->GetImageSpaceInvalidationRect(aRect); + } + + nsIntRect rect(InnerImage()->GetImageSpaceInvalidationRect(aRect)); + rect = rect.Intersect(mClip); + rect.MoveBy(-mClip.X(), -mClip.Y()); + return rect; +} + +} // namespace image +} // namespace mozilla |