Source code

Revision control

Other Tools

/* -*- 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 */
/* rendering object for replaced elements with image data */
#ifndef nsImageFrame_h___
#define nsImageFrame_h___
#include "nsAtomicContainerFrame.h"
#include "nsIObserver.h"
#include "imgINotificationObserver.h"
#include "nsDisplayList.h"
#include "imgIContainer.h"
#include "mozilla/Attributes.h"
#include "mozilla/DebugOnly.h"
#include "mozilla/StaticPtr.h"
#include "nsIReflowCallback.h"
#include "nsTObserverArray.h"
class nsFontMetrics;
class nsImageMap;
class nsIURI;
class nsILoadGroup;
class nsDisplayImage;
class nsPresContext;
class nsImageFrame;
class nsTransform2D;
class nsImageLoadingContent;
namespace mozilla {
class PresShell;
namespace layers {
class ImageContainer;
class ImageLayer;
class LayerManager;
} // namespace layers
} // namespace mozilla
class nsImageListener final : public imgINotificationObserver {
virtual ~nsImageListener();
explicit nsImageListener(nsImageFrame* aFrame);
void SetFrame(nsImageFrame* frame) { mFrame = frame; }
nsImageFrame* mFrame;
class nsImageFrame : public nsAtomicContainerFrame, public nsIReflowCallback {
template <typename T>
using Maybe = mozilla::Maybe<T>;
using Nothing = mozilla::Nothing;
using Visibility = mozilla::Visibility;
typedef mozilla::image::ImgDrawResult ImgDrawResult;
typedef mozilla::layers::ImageContainer ImageContainer;
typedef mozilla::layers::ImageLayer ImageLayer;
typedef mozilla::layers::LayerManager LayerManager;
void DestroyFrom(nsIFrame* aDestructRoot, PostDestroyData&) override;
void DidSetComputedStyle(ComputedStyle* aOldStyle) final;
void Init(nsIContent* aContent, nsContainerFrame* aParent,
nsIFrame* aPrevInFlow) override;
void BuildDisplayList(nsDisplayListBuilder*, const nsDisplayListSet&) final;
nscoord GetMinISize(gfxContext* aRenderingContext) final;
nscoord GetPrefISize(gfxContext* aRenderingContext) final;
mozilla::IntrinsicSize GetIntrinsicSize() final { return mIntrinsicSize; }
mozilla::AspectRatio GetIntrinsicRatio() const final {
return mIntrinsicRatio;
void Reflow(nsPresContext*, ReflowOutput&, const ReflowInput&,
nsReflowStatus&) override;
nsresult GetContentForEvent(mozilla::WidgetEvent*,
nsIContent** aContent) final;
nsresult HandleEvent(nsPresContext*, mozilla::WidgetGUIEvent*,
nsEventStatus*) override;
mozilla::Maybe<Cursor> GetCursor(const nsPoint&) override;
nsresult AttributeChanged(int32_t aNameSpaceID, nsAtom* aAttribute,
int32_t aModType) final;
void OnVisibilityChange(
Visibility aNewVisibility,
const Maybe<OnNonvisible>& aNonvisibleAction = Nothing()) final;
void ResponsiveContentDensityChanged();
void SetupForContentURLRequest();
bool ShouldShowBrokenImageIcon() const;
mozilla::a11y::AccType AccessibleType() override;
bool IsFrameOfType(uint32_t aFlags) const final {
return nsAtomicContainerFrame::IsFrameOfType(
aFlags & ~(nsIFrame::eReplaced | nsIFrame::eReplacedSizing));
nsresult GetFrameName(nsAString& aResult) const override;
void List(FILE* out = stderr, const char* aPrefix = "",
ListFlags aFlags = ListFlags()) const final;
LogicalSides GetLogicalSkipSides() const final;
static void ReleaseGlobals() {
if (gIconLoad) {
gIconLoad = nullptr;
nsresult RestartAnimation();
nsresult StopAnimation();
already_AddRefed<imgIRequest> GetCurrentRequest() const;
void Notify(imgIRequest*, int32_t aType, const nsIntRect* aData);
* Function to test whether given an element and its style, that element
* should get an image frame. Note that this method is only used by the
* frame constructor; it's only here because it uses gIconLoad for now.
static bool ShouldCreateImageFrameFor(const mozilla::dom::Element&,
ImgDrawResult DisplayAltFeedback(gfxContext& aRenderingContext,
const nsRect& aDirtyRect, nsPoint aPt,
uint32_t aFlags);
ImgDrawResult DisplayAltFeedbackWithoutLayer(
nsDisplayItem*, mozilla::wr::DisplayListBuilder&,
const mozilla::layers::StackingContextHelper&,
mozilla::layers::RenderRootStateManager*, nsDisplayListBuilder*,
nsPoint aPt, uint32_t aFlags);
nsRect GetInnerArea() const;
* Return a map element associated with this image.
mozilla::dom::Element* GetMapElement() const;
* Return true if the image has associated image map.
bool HasImageMap() const { return mImageMap || GetMapElement(); }
nsImageMap* GetImageMap();
nsImageMap* GetExistingImageMap() const { return mImageMap; }
void AddInlineMinISize(gfxContext* aRenderingContext,
InlineMinISizeData* aData) final;
void DisconnectMap();
// nsIReflowCallback
bool ReflowFinished() final;
void ReflowCallbackCanceled() final;
// The kind of image frame we are.
enum class Kind : uint8_t {
// For an nsImageLoadingContent.
// For css 'content: url(..)' on non-generated content.
// For a child of a ::before / ::after pseudo-element that had an url() item
// for the content property.
// Creates a suitable continuing frame for this frame.
nsImageFrame* CreateContinuingFrame(mozilla::PresShell*,
ComputedStyle*) const;
friend nsIFrame* NS_NewImageFrame(mozilla::PresShell*, ComputedStyle*);
friend nsIFrame* NS_NewImageFrameForContentProperty(mozilla::PresShell*,
friend nsIFrame* NS_NewImageFrameForGeneratedContentIndex(mozilla::PresShell*,
nsImageFrame(ComputedStyle* aStyle, nsPresContext* aPresContext, Kind aKind)
: nsImageFrame(aStyle, aPresContext, kClassID, aKind) {}
nsImageFrame(ComputedStyle*, nsPresContext* aPresContext, ClassID, Kind);
nsImageFrame(ComputedStyle* aStyle, nsPresContext* aPresContext, ClassID aID)
: nsImageFrame(aStyle, aPresContext, aID, Kind::ImageElement) {}
~nsImageFrame() override;
void EnsureIntrinsicSizeAndRatio();
bool GotInitialReflow() const {
return !HasAnyStateBits(NS_FRAME_FIRST_REFLOW);
SizeComputationResult ComputeSize(
gfxContext* aRenderingContext, mozilla::WritingMode aWM,
const mozilla::LogicalSize& aCBSize, nscoord aAvailableISize,
const mozilla::LogicalSize& aMargin,
const mozilla::LogicalSize& aBorderPadding,
const mozilla::StyleSizeOverrides& aSizeOverrides,
mozilla::ComputeSizeFlags aFlags) final;
bool IsServerImageMap();
void TranslateEventCoords(const nsPoint& aPoint, nsIntPoint& aResult);
bool GetAnchorHREFTargetAndNode(nsIURI** aHref, nsString& aTarget,
nsIContent** aNode);
* Computes the width of the string that fits into the available space
* @param in aLength total length of the string in PRUnichars
* @param in aMaxWidth width not to be exceeded
* @param out aMaxFit length of the string that fits within aMaxWidth
* in PRUnichars
* @return width of the string that fits within aMaxWidth
nscoord MeasureString(const char16_t* aString, int32_t aLength,
nscoord aMaxWidth, uint32_t& aMaxFit,
gfxContext& aContext, nsFontMetrics& aFontMetrics);
void DisplayAltText(nsPresContext* aPresContext,
gfxContext& aRenderingContext, const nsString& aAltText,
const nsRect& aRect);
ImgDrawResult PaintImage(gfxContext& aRenderingContext, nsPoint aPt,
const nsRect& aDirtyRect, imgIContainer* aImage,
uint32_t aFlags);
* If we're ready to decode - that is, if our current request's image is
* available and our decoding heuristics are satisfied - then trigger a decode
* for our image at the size we predict it will be drawn next time it's
* painted.
void MaybeDecodeForPredictedSize();
friend class nsImageListener;
friend class nsImageLoadingContent;
friend class mozilla::PresShell;
void OnSizeAvailable(imgIRequest* aRequest, imgIContainer* aImage);
void OnFrameUpdate(imgIRequest* aRequest, const nsIntRect* aRect);
void OnLoadComplete(imgIRequest* aRequest, nsresult aStatus);
* Notification that aRequest will now be the current request.
void NotifyNewCurrentRequest(imgIRequest* aRequest, nsresult aStatus);
/// Always sync decode our image when painting if @aForce is true.
void SetForceSyncDecoding(bool aForce) { mForceSyncDecoding = aForce; }
* Computes the predicted dest rect that we'll draw into, in app units, based
* upon the provided frame content box. (The content box is what
* nsDisplayImage::GetBounds() returns.)
* The result is not necessarily contained in the frame content box.
nsRect PredictedDestRect(const nsRect& aFrameContentBox);
void MaybeRecordContentUrlOnImageTelemetry();
// random helpers
inline void SpecToURI(const nsAString& aSpec, nsIURI** aURI);
inline void GetLoadGroup(nsPresContext* aPresContext,
nsILoadGroup** aLoadGroup);
nscoord GetContinuationOffset() const;
void GetDocumentCharacterSet(nsACString& aCharset) const;
bool ShouldDisplaySelection();
// Whether the image frame should use the mapped aspect ratio from width=""
// and height="".
bool ShouldUseMappedAspectRatio() const;
// Recalculate mIntrinsicSize from the image.
bool UpdateIntrinsicSize();
// Recalculate mIntrinsicRatio from the image.
bool UpdateIntrinsicRatio();
* This function calculates the transform for converting between
* source space & destination space. May fail if our image has a
* percent-valued or zero-valued height or width.
* @param aTransform The transform object to populate.
* @return whether we succeeded in creating the transform.
bool GetSourceToDestTransform(nsTransform2D& aTransform);
* Helper function to check whether the request corresponds to a load we don't
* care about. Most of the decoder observer methods will bail early if this
* returns true.
bool IsPendingLoad(imgIRequest*) const;
* Updates mImage based on the current image request (cannot be null), and the
* image passed in (can be null), and invalidate layout and paint as needed.
void UpdateImage(imgIRequest*, imgIContainer*);
* Function to convert a dirty rect in the source image to a dirty
* rect for the image frame.
nsRect SourceRectToDest(const nsIntRect& aRect);
* Triggers invalidation for both our image display item and, if appropriate,
* our alt-feedback display item.
* @param aLayerInvalidRect The area to invalidate in layer space. If null,
* the entire layer will be invalidated.
* @param aFrameInvalidRect The area to invalidate in frame space. If null,
* the entire frame will be invalidated.
void InvalidateSelf(const nsIntRect* aLayerInvalidRect,
const nsRect* aFrameInvalidRect);
RefPtr<nsImageMap> mImageMap;
RefPtr<nsImageListener> mListener;
// An image request created for content: url(..).
RefPtr<imgRequestProxy> mContentURLRequest;
nsCOMPtr<imgIContainer> mImage;
nsCOMPtr<imgIContainer> mPrevImage;
nsSize mComputedSize;
mozilla::IntrinsicSize mIntrinsicSize;
// Stores mImage's intrinsic ratio, or a default AspectRatio if there's no
// intrinsic ratio.
mozilla::AspectRatio mIntrinsicRatio;
const Kind mKind;
bool mContentURLRequestRegistered;
bool mDisplayingIcon;
bool mFirstFrameComplete;
bool mReflowCallbackPosted;
bool mForceSyncDecoding;
/* loading / broken image icon support */
// XXXbz this should be handled by the prescontext, I think; that
// way we would have a single iconload per mozilla session instead
// of one per document...
// LoadIcons: initiate the loading of the static icons used to show
// loading / broken images
nsresult LoadIcons(nsPresContext* aPresContext);
nsresult LoadIcon(const nsAString& aSpec, nsPresContext* aPresContext,
imgRequestProxy** aRequest);
class IconLoad final : public nsIObserver, public imgINotificationObserver {
// private class that wraps the data and logic needed for
// broken image and loading image icons
void Shutdown();
void AddIconObserver(nsImageFrame* frame) {
"Observer shouldn't aleady be in array");
void RemoveIconObserver(nsImageFrame* frame) {
mozilla::DebugOnly<bool> didRemove = mIconObservers.RemoveElement(frame);
MOZ_ASSERT(didRemove, "Observer not in array");
~IconLoad() = default;
void GetPrefs();
nsTObserverArray<nsImageFrame*> mIconObservers;
RefPtr<imgRequestProxy> mLoadingImage;
RefPtr<imgRequestProxy> mBrokenImage;
bool mPrefForceInlineAltText;
bool mPrefShowPlaceholders;
bool mPrefShowLoadingPlaceholder;
// singleton pattern: one LoadIcons instance is used
static mozilla::StaticRefPtr<IconLoad> gIconLoad;
friend class nsDisplayImage;
* Note that nsDisplayImage does not receive events. However, an image element
* is replaced content so its background will be z-adjacent to the
* image itself, and hence receive events just as if the image itself
* received events.
class nsDisplayImage final : public nsDisplayImageContainer {
typedef mozilla::layers::LayerManager LayerManager;
nsDisplayImage(nsDisplayListBuilder* aBuilder, nsImageFrame* aFrame,
imgIContainer* aImage, imgIContainer* aPrevImage)
: nsDisplayImageContainer(aBuilder, aFrame),
mPrevImage(aPrevImage) {
~nsDisplayImage() final { MOZ_COUNT_DTOR(nsDisplayImage); }
nsDisplayItemGeometry* AllocateGeometry(nsDisplayListBuilder*) final;
void ComputeInvalidationRegion(nsDisplayListBuilder*,
const nsDisplayItemGeometry*,
nsRegion* aInvalidRegion) const final;
void Paint(nsDisplayListBuilder*, gfxContext* aCtx) final;
already_AddRefed<imgIContainer> GetImage() final;
* @return The dest rect we'll use when drawing this image, in app units.
* Not necessarily contained in this item's bounds.
nsRect GetDestRect() const final;
void UpdateDrawResult(mozilla::image::ImgDrawResult aResult) final {
nsDisplayItemGenericImageGeometry::UpdateDrawResult(this, aResult);
LayerState GetLayerState(nsDisplayListBuilder*, LayerManager*,
const ContainerLayerParameters&) final;
nsRect GetBounds(bool* aSnap) const {
*aSnap = true;
nsImageFrame* imageFrame = static_cast<nsImageFrame*>(mFrame);
return imageFrame->GetInnerArea() + ToReferenceFrame();
nsRect GetBounds(nsDisplayListBuilder*, bool* aSnap) const final {
return GetBounds(aSnap);
nsRegion GetOpaqueRegion(nsDisplayListBuilder*, bool* aSnap) const final;
already_AddRefed<Layer> BuildLayer(nsDisplayListBuilder*, LayerManager*,
const ContainerLayerParameters&) final;
bool CreateWebRenderCommands(mozilla::wr::DisplayListBuilder&,
const StackingContextHelper&,
nsDisplayListBuilder*) final;
nsCOMPtr<imgIContainer> mImage;
nsCOMPtr<imgIContainer> mPrevImage;
#endif /* nsImageFrame_h___ */