summaryrefslogtreecommitdiffstats
path: root/layout/base/nsLayoutUtils.cpp
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-19 01:47:29 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-19 01:47:29 +0000
commit0ebf5bdf043a27fd3dfb7f92e0cb63d88954c44d (patch)
treea31f07c9bcca9d56ce61e9a1ffd30ef350d513aa /layout/base/nsLayoutUtils.cpp
parentInitial commit. (diff)
downloadfirefox-esr-0ebf5bdf043a27fd3dfb7f92e0cb63d88954c44d.tar.xz
firefox-esr-0ebf5bdf043a27fd3dfb7f92e0cb63d88954c44d.zip
Adding upstream version 115.8.0esr.upstream/115.8.0esr
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'layout/base/nsLayoutUtils.cpp')
-rw-r--r--layout/base/nsLayoutUtils.cpp9978
1 files changed, 9978 insertions, 0 deletions
diff --git a/layout/base/nsLayoutUtils.cpp b/layout/base/nsLayoutUtils.cpp
new file mode 100644
index 0000000000..5963a74ab2
--- /dev/null
+++ b/layout/base/nsLayoutUtils.cpp
@@ -0,0 +1,9978 @@
+/* -*- 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/. */
+
+#include "nsLayoutUtils.h"
+
+#include <algorithm>
+#include <limits>
+
+#include "ActiveLayerTracker.h"
+#include "DisplayItemClip.h"
+#include "gfx2DGlue.h"
+#include "gfxContext.h"
+#include "gfxDrawable.h"
+#include "gfxEnv.h"
+#include "gfxMatrix.h"
+#include "gfxPlatform.h"
+#include "gfxRect.h"
+#include "gfxTypes.h"
+#include "gfxUtils.h"
+#include "ImageContainer.h"
+#include "ImageOps.h"
+#include "ImageRegion.h"
+#include "imgIContainer.h"
+#include "imgIRequest.h"
+#include "LayoutLogging.h"
+#include "MobileViewportManager.h"
+#include "mozilla/AccessibleCaretEventHub.h"
+#include "mozilla/ArrayUtils.h"
+#include "mozilla/Baseline.h"
+#include "mozilla/BasicEvents.h"
+#include "mozilla/ClearOnShutdown.h"
+#include "mozilla/DisplayPortUtils.h"
+#include "mozilla/glean/GleanMetrics.h"
+#include "mozilla/dom/AnonymousContent.h"
+#include "mozilla/dom/BrowserChild.h"
+#include "mozilla/dom/CanvasUtils.h"
+#include "mozilla/dom/Document.h"
+#include "mozilla/dom/DocumentInlines.h"
+#include "mozilla/dom/DOMRect.h"
+#include "mozilla/dom/DOMStringList.h"
+#include "mozilla/dom/Element.h"
+#include "mozilla/dom/HTMLBodyElement.h"
+#include "mozilla/dom/HTMLCanvasElement.h"
+#include "mozilla/dom/HTMLImageElement.h"
+#include "mozilla/dom/HTMLMediaElementBinding.h"
+#include "mozilla/dom/HTMLVideoElement.h"
+#include "mozilla/dom/InspectorFontFace.h"
+#include "mozilla/dom/ImageBitmap.h"
+#include "mozilla/dom/KeyframeEffect.h"
+#include "mozilla/dom/SVGViewportElement.h"
+#include "mozilla/dom/UIEvent.h"
+#include "mozilla/intl/BidiEmbeddingLevel.h"
+#include "mozilla/EffectCompositor.h"
+#include "mozilla/EffectSet.h"
+#include "mozilla/EventDispatcher.h"
+#include "mozilla/EventStateManager.h"
+#include "mozilla/FloatingPoint.h"
+#include "mozilla/gfx/2D.h"
+#include "mozilla/gfx/gfxVars.h"
+#include "mozilla/gfx/PathHelpers.h"
+#include "mozilla/IntegerRange.h"
+#include "mozilla/layers/APZCCallbackHelper.h"
+#include "mozilla/layers/APZPublicUtils.h" // for apz::CalculatePendingDisplayPort
+#include "mozilla/layers/CompositorBridgeChild.h"
+#include "mozilla/layers/PAPZ.h"
+#include "mozilla/layers/StackingContextHelper.h"
+#include "mozilla/layers/WebRenderLayerManager.h"
+#include "mozilla/Likely.h"
+#include "mozilla/LookAndFeel.h"
+#include "mozilla/Maybe.h"
+#include "mozilla/MemoryReporting.h"
+#include "mozilla/PerfStats.h"
+#include "mozilla/Preferences.h"
+#include "mozilla/PresShell.h"
+#include "mozilla/ProfilerLabels.h"
+#include "mozilla/ProfilerMarkers.h"
+#include "mozilla/RestyleManager.h"
+#include "mozilla/ScopeExit.h"
+#include "mozilla/ScrollOrigin.h"
+#include "mozilla/ServoStyleSet.h"
+#include "mozilla/ServoStyleSetInlines.h"
+#include "mozilla/StaticPrefs_apz.h"
+#include "mozilla/StaticPrefs_dom.h"
+#include "mozilla/StaticPrefs_font.h"
+#include "mozilla/StaticPrefs_general.h"
+#include "mozilla/StaticPrefs_gfx.h"
+#include "mozilla/StaticPrefs_image.h"
+#include "mozilla/StaticPrefs_layers.h"
+#include "mozilla/StaticPrefs_layout.h"
+#include "mozilla/StaticPtr.h"
+#include "mozilla/StyleAnimationValue.h"
+#include "mozilla/SVGImageContext.h"
+#include "mozilla/SVGIntegrationUtils.h"
+#include "mozilla/SVGTextFrame.h"
+#include "mozilla/SVGUtils.h"
+#include "mozilla/Telemetry.h"
+#include "mozilla/ToString.h"
+#include "mozilla/Unused.h"
+#include "mozilla/ViewportFrame.h"
+#include "mozilla/ViewportUtils.h"
+#include "mozilla/WheelHandlingHelper.h" // for WheelHandlingUtils
+#include "nsAnimationManager.h"
+#include "nsAtom.h"
+#include "nsBidiPresUtils.h"
+#include "nsBlockFrame.h"
+#include "nsCanvasFrame.h"
+#include "nsCaret.h"
+#include "nsCharTraits.h"
+#include "nsCOMPtr.h"
+#include "nsComputedDOMStyle.h"
+#include "nsContentUtils.h"
+#include "nsCSSAnonBoxes.h"
+#include "nsCSSColorUtils.h"
+#include "nsCSSFrameConstructor.h"
+#include "nsCSSProps.h"
+#include "nsCSSPseudoElements.h"
+#include "nsCSSRendering.h"
+#include "nsDisplayList.h"
+#include "nsFieldSetFrame.h"
+#include "nsFlexContainerFrame.h"
+#include "nsFontInflationData.h"
+#include "nsFontMetrics.h"
+#include "nsFrameList.h"
+#include "nsFrameSelection.h"
+#include "nsGenericHTMLElement.h"
+#include "nsGkAtoms.h"
+#include "nsICanvasRenderingContextInternal.h"
+#include "nsIContent.h"
+#include "nsIContentInlines.h"
+#include "nsIContentViewer.h"
+#include "nsIDocShell.h"
+#include "nsIFrameInlines.h"
+#include "nsIImageLoadingContent.h"
+#include "nsIInterfaceRequestorUtils.h"
+#include "nsIScrollableFrame.h"
+#include "nsIWidget.h"
+#include "nsListControlFrame.h"
+#include "nsPIDOMWindow.h"
+#include "nsPlaceholderFrame.h"
+#include "nsPresContext.h"
+#include "nsPresContextInlines.h"
+#include "nsRefreshDriver.h"
+#include "nsRegion.h"
+#include "nsStyleConsts.h"
+#include "nsStyleStructInlines.h"
+#include "nsStyleTransformMatrix.h"
+#include "nsSubDocumentFrame.h"
+#include "nsTableWrapperFrame.h"
+#include "nsTArray.h"
+#include "nsTextFragment.h"
+#include "nsTextFrame.h"
+#include "nsTHashMap.h"
+#include "nsTransitionManager.h"
+#include "nsView.h"
+#include "nsViewManager.h"
+#include "prenv.h"
+#include "RegionBuilder.h"
+#include "RetainedDisplayListBuilder.h"
+#include "TextDrawTarget.h"
+#include "UnitTransforms.h"
+#include "ViewportFrame.h"
+
+#include "nsXULPopupManager.h"
+
+// Make sure getpid() works.
+#ifdef XP_WIN
+# include <process.h>
+# define getpid _getpid
+#else
+# include <unistd.h>
+#endif
+
+using namespace mozilla;
+using namespace mozilla::dom;
+using namespace mozilla::image;
+using namespace mozilla::layers;
+using namespace mozilla::layout;
+using namespace mozilla::gfx;
+using mozilla::dom::HTMLMediaElement_Binding::HAVE_METADATA;
+using mozilla::dom::HTMLMediaElement_Binding::HAVE_NOTHING;
+
+#ifdef DEBUG
+// TODO: remove, see bug 598468.
+bool nsLayoutUtils::gPreventAssertInCompareTreePosition = false;
+#endif // DEBUG
+
+typedef ScrollableLayerGuid::ViewID ViewID;
+typedef nsStyleTransformMatrix::TransformReferenceBox TransformReferenceBox;
+
+static ViewID sScrollIdCounter = ScrollableLayerGuid::START_SCROLL_ID;
+
+typedef nsTHashMap<nsUint64HashKey, nsIContent*> ContentMap;
+static StaticAutoPtr<ContentMap> sContentMap;
+
+static ContentMap& GetContentMap() {
+ if (!sContentMap) {
+ sContentMap = new ContentMap();
+ }
+ return *sContentMap;
+}
+
+template <typename TestType>
+static bool HasMatchingAnimations(EffectSet& aEffects, TestType&& aTest) {
+ for (KeyframeEffect* effect : aEffects) {
+ if (!effect->GetAnimation() || !effect->GetAnimation()->IsRelevant()) {
+ continue;
+ }
+
+ if (aTest(*effect, aEffects)) {
+ return true;
+ }
+ }
+
+ return false;
+}
+
+template <typename TestType>
+static bool HasMatchingAnimations(const nsIFrame* aFrame,
+ const nsCSSPropertyIDSet& aPropertySet,
+ TestType&& aTest) {
+ MOZ_ASSERT(aFrame);
+
+ if (aPropertySet.IsSubsetOf(nsCSSPropertyIDSet::OpacityProperties()) &&
+ !aFrame->MayHaveOpacityAnimation()) {
+ return false;
+ }
+
+ if (aPropertySet.IsSubsetOf(nsCSSPropertyIDSet::TransformLikeProperties()) &&
+ !aFrame->MayHaveTransformAnimation()) {
+ return false;
+ }
+
+ EffectSet* effectSet = EffectSet::GetForFrame(aFrame, aPropertySet);
+ if (!effectSet) {
+ return false;
+ }
+
+ return HasMatchingAnimations(*effectSet, aTest);
+}
+
+/* static */
+bool nsLayoutUtils::HasAnimationOfPropertySet(
+ const nsIFrame* aFrame, const nsCSSPropertyIDSet& aPropertySet) {
+ return HasMatchingAnimations(
+ aFrame, aPropertySet,
+ [&aPropertySet](KeyframeEffect& aEffect, const EffectSet&) {
+ return aEffect.HasAnimationOfPropertySet(aPropertySet);
+ });
+}
+
+/* static */
+bool nsLayoutUtils::HasAnimationOfPropertySet(
+ const nsIFrame* aFrame, const nsCSSPropertyIDSet& aPropertySet,
+ EffectSet* aEffectSet) {
+ MOZ_ASSERT(
+ !aEffectSet || EffectSet::GetForFrame(aFrame, aPropertySet) == aEffectSet,
+ "The EffectSet, if supplied, should match what we would otherwise fetch");
+
+ if (!aEffectSet) {
+ return nsLayoutUtils::HasAnimationOfPropertySet(aFrame, aPropertySet);
+ }
+
+ if (aPropertySet.IsSubsetOf(nsCSSPropertyIDSet::TransformLikeProperties()) &&
+ !aEffectSet->MayHaveTransformAnimation()) {
+ return false;
+ }
+
+ if (aPropertySet.IsSubsetOf(nsCSSPropertyIDSet::OpacityProperties()) &&
+ !aEffectSet->MayHaveOpacityAnimation()) {
+ return false;
+ }
+
+ return HasMatchingAnimations(
+ *aEffectSet,
+ [&aPropertySet](KeyframeEffect& aEffect, const EffectSet& aEffectSet) {
+ return aEffect.HasAnimationOfPropertySet(aPropertySet);
+ });
+}
+
+/* static */
+bool nsLayoutUtils::HasAnimationOfTransformAndMotionPath(
+ const nsIFrame* aFrame) {
+ return nsLayoutUtils::HasAnimationOfPropertySet(
+ aFrame,
+ nsCSSPropertyIDSet{eCSSProperty_transform, eCSSProperty_translate,
+ eCSSProperty_rotate, eCSSProperty_scale,
+ eCSSProperty_offset_path}) ||
+ (!aFrame->StyleDisplay()->mOffsetPath.IsNone() &&
+ nsLayoutUtils::HasAnimationOfPropertySet(
+ aFrame, nsCSSPropertyIDSet::MotionPathProperties()));
+}
+
+/* static */
+bool nsLayoutUtils::HasEffectiveAnimation(
+ const nsIFrame* aFrame, const nsCSSPropertyIDSet& aPropertySet) {
+ return HasMatchingAnimations(
+ aFrame, aPropertySet,
+ [&aPropertySet](KeyframeEffect& aEffect, const EffectSet& aEffectSet) {
+ return aEffect.HasEffectiveAnimationOfPropertySet(aPropertySet,
+ aEffectSet);
+ });
+}
+
+/* static */
+nsCSSPropertyIDSet nsLayoutUtils::GetAnimationPropertiesForCompositor(
+ const nsIFrame* aStyleFrame) {
+ nsCSSPropertyIDSet properties;
+
+ // We fetch the effects for the style frame here since this method is called
+ // by RestyleManager::AddLayerChangesForAnimation which takes care to apply
+ // the relevant hints to the primary frame as needed.
+ EffectSet* effects = EffectSet::GetForStyleFrame(aStyleFrame);
+ if (!effects) {
+ return properties;
+ }
+
+ AnimationPerformanceWarning::Type warning;
+ if (!EffectCompositor::AllowCompositorAnimationsOnFrame(aStyleFrame,
+ warning)) {
+ return properties;
+ }
+
+ for (const KeyframeEffect* effect : *effects) {
+ properties |= effect->GetPropertiesForCompositor(*effects, aStyleFrame);
+ }
+
+ // If properties only have motion-path properties, we have to make sure they
+ // have effects. i.e. offset-path is not none or we have offset-path
+ // animations.
+ if (properties.IsSubsetOf(nsCSSPropertyIDSet::MotionPathProperties()) &&
+ !properties.HasProperty(eCSSProperty_offset_path) &&
+ aStyleFrame->StyleDisplay()->mOffsetPath.IsNone()) {
+ properties.Empty();
+ }
+
+ return properties;
+}
+
+static float GetSuitableScale(float aMaxScale, float aMinScale,
+ nscoord aVisibleDimension,
+ nscoord aDisplayDimension) {
+ float displayVisibleRatio =
+ float(aDisplayDimension) / float(aVisibleDimension);
+ // We want to rasterize based on the largest scale used during the
+ // transform animation, unless that would make us rasterize something
+ // larger than the screen. But we never want to go smaller than the
+ // minimum scale over the animation.
+ if (FuzzyEqualsMultiplicative(displayVisibleRatio, aMaxScale, .01f)) {
+ // Using aMaxScale may make us rasterize something a fraction larger than
+ // the screen. However, if aMaxScale happens to be the final scale of a
+ // transform animation it is better to use aMaxScale so that for the
+ // fraction of a second before we delayerize the composited texture it has
+ // a better chance of being pixel aligned and composited without resampling
+ // (avoiding visually clunky delayerization).
+ return aMaxScale;
+ }
+ return std::max(std::min(aMaxScale, displayVisibleRatio), aMinScale);
+}
+
+// The first value in this pair is the min scale, and the second one is the max
+// scale.
+using MinAndMaxScale = std::pair<MatrixScales, MatrixScales>;
+
+static inline void UpdateMinMaxScale(const nsIFrame* aFrame,
+ const AnimationValue& aValue,
+ MinAndMaxScale& aMinAndMaxScale) {
+ MatrixScales size = aValue.GetScaleValue(aFrame);
+ MatrixScales& minScale = aMinAndMaxScale.first;
+ MatrixScales& maxScale = aMinAndMaxScale.second;
+
+ minScale = Min(minScale, size);
+ maxScale = Max(maxScale, size);
+}
+
+// The final transform matrix is calculated by merging the final results of each
+// transform-like properties, so do the scale factors. In other words, the
+// potential min/max scales could be gotten by multiplying the max/min scales of
+// each properties.
+//
+// For example, there is an animation:
+// from { "transform: scale(1, 1)", "scale: 3, 3" };
+// to { "transform: scale(2, 2)", "scale: 1, 1" };
+//
+// the min scale is (1, 1) * (1, 1) = (1, 1), and
+// The max scale is (2, 2) * (3, 3) = (6, 6).
+// This means we multiply the min/max scale factor of transform property and the
+// min/max scale factor of scale property to get the final max/min scale factor.
+static Array<MinAndMaxScale, 2> GetMinAndMaxScaleForAnimationProperty(
+ const nsIFrame* aFrame,
+ const nsTArray<RefPtr<dom::Animation>>& aAnimations) {
+ // We use a fixed array to store the min/max scales for each property.
+ // The first element in the array is for eCSSProperty_transform, and the
+ // second one is for eCSSProperty_scale.
+ const MinAndMaxScale defaultValue =
+ std::make_pair(MatrixScales(std::numeric_limits<float>::max(),
+ std::numeric_limits<float>::max()),
+ MatrixScales(std::numeric_limits<float>::min(),
+ std::numeric_limits<float>::min()));
+ Array<MinAndMaxScale, 2> minAndMaxScales(defaultValue, defaultValue);
+
+ for (dom::Animation* anim : aAnimations) {
+ // This method is only expected to be passed animations that are running on
+ // the compositor and we only pass playing animations to the compositor,
+ // which are, by definition, "relevant" animations (animations that are
+ // not yet finished or which are filling forwards).
+ MOZ_ASSERT(anim->IsRelevant());
+
+ const dom::KeyframeEffect* effect =
+ anim->GetEffect() ? anim->GetEffect()->AsKeyframeEffect() : nullptr;
+ MOZ_ASSERT(effect, "A playing animation should have a keyframe effect");
+ for (const AnimationProperty& prop : effect->Properties()) {
+ if (prop.mProperty != eCSSProperty_transform &&
+ prop.mProperty != eCSSProperty_scale) {
+ continue;
+ }
+
+ // 0: eCSSProperty_transform.
+ // 1: eCSSProperty_scale.
+ MinAndMaxScale& scales =
+ minAndMaxScales[prop.mProperty == eCSSProperty_transform ? 0 : 1];
+
+ // We need to factor in the scale of the base style if the base style
+ // will be used on the compositor.
+ const AnimationValue& baseStyle = effect->BaseStyle(prop.mProperty);
+ if (!baseStyle.IsNull()) {
+ UpdateMinMaxScale(aFrame, baseStyle, scales);
+ }
+
+ for (const AnimationPropertySegment& segment : prop.mSegments) {
+ // In case of add or accumulate composite, StyleAnimationValue does
+ // not have a valid value.
+ if (segment.HasReplaceableFromValue()) {
+ UpdateMinMaxScale(aFrame, segment.mFromValue, scales);
+ }
+
+ if (segment.HasReplaceableToValue()) {
+ UpdateMinMaxScale(aFrame, segment.mToValue, scales);
+ }
+ }
+ }
+ }
+
+ return minAndMaxScales;
+}
+
+MatrixScales nsLayoutUtils::ComputeSuitableScaleForAnimation(
+ const nsIFrame* aFrame, const nsSize& aVisibleSize,
+ const nsSize& aDisplaySize) {
+ const nsTArray<RefPtr<dom::Animation>> compositorAnimations =
+ EffectCompositor::GetAnimationsForCompositor(
+ aFrame,
+ nsCSSPropertyIDSet{eCSSProperty_transform, eCSSProperty_scale});
+
+ if (compositorAnimations.IsEmpty()) {
+ return MatrixScales();
+ }
+
+ const Array<MinAndMaxScale, 2> minAndMaxScales =
+ GetMinAndMaxScaleForAnimationProperty(aFrame, compositorAnimations);
+
+ // This might cause an issue if users use std::numeric_limits<float>::min()
+ // (or max()) as the scale value. However, in this case, we may render an
+ // extreme small (or large) element, so this may not be a problem. If so,
+ // please fix this.
+ MatrixScales maxScale(std::numeric_limits<float>::min(),
+ std::numeric_limits<float>::min());
+ MatrixScales minScale(std::numeric_limits<float>::max(),
+ std::numeric_limits<float>::max());
+
+ auto isUnset = [](const MatrixScales& aMax, const MatrixScales& aMin) {
+ return aMax.xScale == std::numeric_limits<float>::min() &&
+ aMax.yScale == std::numeric_limits<float>::min() &&
+ aMin.xScale == std::numeric_limits<float>::max() &&
+ aMin.yScale == std::numeric_limits<float>::max();
+ };
+
+ // Iterate the slots to get the final scale value.
+ for (const auto& pair : minAndMaxScales) {
+ const MatrixScales& currMinScale = pair.first;
+ const MatrixScales& currMaxScale = pair.second;
+
+ if (isUnset(currMaxScale, currMinScale)) {
+ // We don't have this animation property, so skip.
+ continue;
+ }
+
+ if (isUnset(maxScale, minScale)) {
+ // Initialize maxScale and minScale.
+ maxScale = currMaxScale;
+ minScale = currMinScale;
+ } else {
+ // The scale factors of each transform-like property should be multiplied
+ // by others because we merge their sampled values as a final matrix by
+ // matrix multiplication, so here we multiply the scale factors by the
+ // previous one to get the possible max and min scale factors.
+ maxScale = maxScale * currMaxScale;
+ minScale = minScale * currMinScale;
+ }
+ }
+
+ if (isUnset(maxScale, minScale)) {
+ // We didn't encounter any transform-like property.
+ return MatrixScales();
+ }
+
+ return MatrixScales(
+ GetSuitableScale(maxScale.xScale, minScale.xScale, aVisibleSize.width,
+ aDisplaySize.width),
+ GetSuitableScale(maxScale.yScale, minScale.yScale, aVisibleSize.height,
+ aDisplaySize.height));
+}
+
+bool nsLayoutUtils::AreAsyncAnimationsEnabled() {
+ return StaticPrefs::layers_offmainthreadcomposition_async_animations() &&
+ gfxPlatform::OffMainThreadCompositingEnabled();
+}
+
+bool nsLayoutUtils::AreRetainedDisplayListsEnabled() {
+#ifdef MOZ_WIDGET_ANDROID
+ return StaticPrefs::layout_display_list_retain();
+#else
+ if (XRE_IsContentProcess()) {
+ return StaticPrefs::layout_display_list_retain();
+ }
+
+ if (XRE_IsE10sParentProcess()) {
+ return StaticPrefs::layout_display_list_retain_chrome();
+ }
+
+ // Retained display lists require e10s.
+ return false;
+#endif
+}
+
+bool nsLayoutUtils::DisplayRootHasRetainedDisplayListBuilder(nsIFrame* aFrame) {
+ return GetRetainedDisplayListBuilder(aFrame) != nullptr;
+}
+
+RetainedDisplayListBuilder* nsLayoutUtils::GetRetainedDisplayListBuilder(
+ nsIFrame* aFrame) {
+ MOZ_ASSERT(aFrame);
+ MOZ_ASSERT(aFrame->PresShell());
+
+ // Use the pres shell root frame to get the display root frame. This skips
+ // the early exit in |nsLayoutUtils::GetDisplayRootFrame()| for popup frames.
+ const nsIFrame* rootFrame = aFrame->PresShell()->GetRootFrame();
+ if (!rootFrame) {
+ return nullptr;
+ }
+
+ const nsIFrame* displayRootFrame = GetDisplayRootFrame(rootFrame);
+ MOZ_ASSERT(displayRootFrame);
+
+ return displayRootFrame->GetProperty(RetainedDisplayListBuilder::Cached());
+}
+
+bool nsLayoutUtils::GPUImageScalingEnabled() {
+ static bool sGPUImageScalingEnabled;
+ static bool sGPUImageScalingPrefInitialised = false;
+
+ if (!sGPUImageScalingPrefInitialised) {
+ sGPUImageScalingPrefInitialised = true;
+ sGPUImageScalingEnabled =
+ Preferences::GetBool("layout.gpu-image-scaling.enabled", false);
+ }
+
+ return sGPUImageScalingEnabled;
+}
+
+void nsLayoutUtils::UnionChildOverflow(nsIFrame* aFrame,
+ OverflowAreas& aOverflowAreas,
+ FrameChildListIDs aSkipChildLists) {
+ // Iterate over all children except pop-ups.
+ FrameChildListIDs skip(aSkipChildLists);
+ skip += FrameChildListID::Popup;
+
+ for (const auto& [list, listID] : aFrame->ChildLists()) {
+ if (skip.contains(listID)) {
+ continue;
+ }
+ for (nsIFrame* child : list) {
+ aOverflowAreas.UnionWith(
+ child->GetActualAndNormalOverflowAreasRelativeToParent());
+ }
+ }
+}
+
+static void DestroyViewID(void* aObject, nsAtom* aPropertyName,
+ void* aPropertyValue, void* aData) {
+ ViewID* id = static_cast<ViewID*>(aPropertyValue);
+ GetContentMap().Remove(*id);
+ delete id;
+}
+
+/**
+ * A namespace class for static layout utilities.
+ */
+
+bool nsLayoutUtils::FindIDFor(const nsIContent* aContent, ViewID* aOutViewId) {
+ void* scrollIdProperty = aContent->GetProperty(nsGkAtoms::RemoteId);
+ if (scrollIdProperty) {
+ *aOutViewId = *static_cast<ViewID*>(scrollIdProperty);
+ return true;
+ }
+ return false;
+}
+
+ViewID nsLayoutUtils::FindOrCreateIDFor(nsIContent* aContent) {
+ ViewID scrollId;
+
+ if (!FindIDFor(aContent, &scrollId)) {
+ scrollId = sScrollIdCounter++;
+ aContent->SetProperty(nsGkAtoms::RemoteId, new ViewID(scrollId),
+ DestroyViewID);
+ GetContentMap().InsertOrUpdate(scrollId, aContent);
+ }
+
+ return scrollId;
+}
+
+nsIContent* nsLayoutUtils::FindContentFor(ViewID aId) {
+ MOZ_ASSERT(aId != ScrollableLayerGuid::NULL_SCROLL_ID,
+ "Cannot find a content element in map for null IDs.");
+ nsIContent* content;
+ bool exists = GetContentMap().Get(aId, &content);
+
+ if (exists) {
+ return content;
+ } else {
+ return nullptr;
+ }
+}
+
+nsIFrame* nsLayoutUtils::GetScrollFrameFromContent(nsIContent* aContent) {
+ nsIFrame* frame = aContent->GetPrimaryFrame();
+ if (aContent->OwnerDoc()->GetRootElement() == aContent) {
+ PresShell* presShell = frame ? frame->PresShell() : nullptr;
+ if (!presShell) {
+ presShell = aContent->OwnerDoc()->GetPresShell();
+ }
+ // We want the scroll frame, the root scroll frame differs from all
+ // others in that the primary frame is not the scroll frame.
+ nsIFrame* rootScrollFrame =
+ presShell ? presShell->GetRootScrollFrame() : nullptr;
+ if (rootScrollFrame) {
+ frame = rootScrollFrame;
+ }
+ }
+ return frame;
+}
+
+nsIScrollableFrame* nsLayoutUtils::FindScrollableFrameFor(
+ nsIContent* aContent) {
+ nsIFrame* scrollFrame = GetScrollFrameFromContent(aContent);
+ return scrollFrame ? scrollFrame->GetScrollTargetFrame() : nullptr;
+}
+
+nsIScrollableFrame* nsLayoutUtils::FindScrollableFrameFor(ViewID aId) {
+ nsIContent* content = FindContentFor(aId);
+ if (!content) {
+ return nullptr;
+ }
+
+ return FindScrollableFrameFor(content);
+}
+
+ViewID nsLayoutUtils::FindIDForScrollableFrame(
+ nsIScrollableFrame* aScrollable) {
+ if (!aScrollable) {
+ return ScrollableLayerGuid::NULL_SCROLL_ID;
+ }
+
+ nsIFrame* scrollFrame = do_QueryFrame(aScrollable);
+ nsIContent* scrollContent = scrollFrame->GetContent();
+
+ ScrollableLayerGuid::ViewID scrollId;
+ if (scrollContent && nsLayoutUtils::FindIDFor(scrollContent, &scrollId)) {
+ return scrollId;
+ }
+
+ return ScrollableLayerGuid::NULL_SCROLL_ID;
+}
+
+bool nsLayoutUtils::UsesAsyncScrolling(nsIFrame* aFrame) {
+#ifdef MOZ_WIDGET_ANDROID
+ // We always have async scrolling for android
+ return true;
+#endif
+
+ return AsyncPanZoomEnabled(aFrame);
+}
+
+bool nsLayoutUtils::AsyncPanZoomEnabled(const nsIFrame* aFrame) {
+ // We use this as a shortcut, since if the compositor will never use APZ,
+ // no widget will either.
+ if (!gfxPlatform::AsyncPanZoomEnabled()) {
+ return false;
+ }
+
+ const nsIFrame* frame = nsLayoutUtils::GetDisplayRootFrame(aFrame);
+ nsIWidget* widget = frame->GetNearestWidget();
+ if (!widget) {
+ return false;
+ }
+ return widget->AsyncPanZoomEnabled();
+}
+
+bool nsLayoutUtils::AllowZoomingForDocument(
+ const mozilla::dom::Document* aDocument) {
+ if (aDocument->GetPresShell() &&
+ !aDocument->GetPresShell()->AsyncPanZoomEnabled()) {
+ return false;
+ }
+ // True if we allow zooming for all documents on this platform, or if we are
+ // in RDM and handling meta viewports, which force zoom under some
+ // circumstances.
+ BrowsingContext* bc = aDocument ? aDocument->GetBrowsingContext() : nullptr;
+ return StaticPrefs::apz_allow_zooming() ||
+ (bc && bc->InRDMPane() &&
+ nsLayoutUtils::ShouldHandleMetaViewport(aDocument));
+}
+
+static bool HasVisibleAnonymousContents(Document* aDoc) {
+ for (RefPtr<AnonymousContent>& ac : aDoc->GetAnonymousContents()) {
+ // We check to see if the anonymous content node has a frame. If it doesn't,
+ // that means that's not visible to the user because e.g. it's display:none.
+ // For now we assume that if it has a frame, it is visible. We might be able
+ // to refine this further by adding complexity if it turns out this
+ // condition results in a lot of false positives.
+ if (ac->ContentNode().GetPrimaryFrame()) {
+ return true;
+ }
+ }
+ return false;
+}
+
+bool nsLayoutUtils::ShouldDisableApzForElement(nsIContent* aContent) {
+ if (!aContent) {
+ return false;
+ }
+
+ if (aContent->GetProperty(nsGkAtoms::apzDisabled)) {
+ return true;
+ }
+
+ Document* doc = aContent->GetComposedDoc();
+ if (PresShell* rootPresShell =
+ APZCCallbackHelper::GetRootContentDocumentPresShellForContent(
+ aContent)) {
+ if (Document* rootDoc = rootPresShell->GetDocument()) {
+ nsIContent* rootContent =
+ rootPresShell->GetRootScrollFrame()
+ ? rootPresShell->GetRootScrollFrame()->GetContent()
+ : rootDoc->GetDocumentElement();
+ // For the AccessibleCaret and other anonymous contents: disable APZ on
+ // any scrollable subframes that are not the root scrollframe of a
+ // document, if the document has any visible anonymous contents.
+ //
+ // If we find this is triggering in too many scenarios then we might
+ // want to tighten this check further. The main use cases for which we
+ // want to disable APZ as of this writing are listed in bug 1316318.
+ if (aContent != rootContent && HasVisibleAnonymousContents(rootDoc)) {
+ return true;
+ }
+ }
+ }
+
+ if (!doc) {
+ return false;
+ }
+
+ if (PresShell* presShell = doc->GetPresShell()) {
+ if (RefPtr<AccessibleCaretEventHub> eventHub =
+ presShell->GetAccessibleCaretEventHub()) {
+ // Disable APZ for all elements if AccessibleCaret tells us to do so.
+ if (eventHub->ShouldDisableApz()) {
+ return true;
+ }
+ }
+ }
+
+ return StaticPrefs::apz_disable_for_scroll_linked_effects() &&
+ doc->HasScrollLinkedEffect();
+}
+
+void nsLayoutUtils::NotifyPaintSkipTransaction(ViewID aScrollId) {
+ if (nsIScrollableFrame* scrollFrame =
+ nsLayoutUtils::FindScrollableFrameFor(aScrollId)) {
+#ifdef DEBUG
+ nsIFrame* f = do_QueryFrame(scrollFrame);
+ MOZ_ASSERT(f && f->PresShell() && !f->PresShell()->IsResolutionUpdated());
+#endif
+ scrollFrame->NotifyApzTransaction();
+ }
+}
+
+nsContainerFrame* nsLayoutUtils::LastContinuationWithChild(
+ nsContainerFrame* aFrame) {
+ MOZ_ASSERT(aFrame, "NULL frame pointer");
+ for (auto f = aFrame->LastContinuation(); f; f = f->GetPrevContinuation()) {
+ for (const auto& childList : f->ChildLists()) {
+ if (MOZ_LIKELY(!childList.mList.IsEmpty())) {
+ return static_cast<nsContainerFrame*>(f);
+ }
+ }
+ }
+ return aFrame;
+}
+
+// static
+FrameChildListID nsLayoutUtils::GetChildListNameFor(nsIFrame* aChildFrame) {
+ FrameChildListID id = FrameChildListID::Principal;
+
+ MOZ_DIAGNOSTIC_ASSERT(!aChildFrame->HasAnyStateBits(NS_FRAME_OUT_OF_FLOW));
+
+ if (aChildFrame->HasAnyStateBits(NS_FRAME_IS_OVERFLOW_CONTAINER)) {
+ nsIFrame* pif = aChildFrame->GetPrevInFlow();
+ if (pif->GetParent() == aChildFrame->GetParent()) {
+ id = FrameChildListID::ExcessOverflowContainers;
+ } else {
+ id = FrameChildListID::OverflowContainers;
+ }
+ } else {
+ LayoutFrameType childType = aChildFrame->Type();
+ if (LayoutFrameType::TableColGroup == childType) {
+ id = FrameChildListID::ColGroup;
+ } else if (aChildFrame->IsTableCaption()) {
+ id = FrameChildListID::Caption;
+ } else {
+ id = FrameChildListID::Principal;
+ }
+ }
+
+#ifdef DEBUG
+ // Verify that the frame is actually in that child list or in the
+ // corresponding overflow list.
+ nsContainerFrame* parent = aChildFrame->GetParent();
+ bool found = parent->GetChildList(id).ContainsFrame(aChildFrame);
+ if (!found) {
+ found = parent->GetChildList(FrameChildListID::Overflow)
+ .ContainsFrame(aChildFrame);
+ MOZ_ASSERT(found, "not in child list");
+ }
+#endif
+
+ return id;
+}
+
+static Element* GetPseudo(const nsIContent* aContent, nsAtom* aPseudoProperty) {
+ MOZ_ASSERT(aPseudoProperty == nsGkAtoms::beforePseudoProperty ||
+ aPseudoProperty == nsGkAtoms::afterPseudoProperty ||
+ aPseudoProperty == nsGkAtoms::markerPseudoProperty);
+ if (!aContent->MayHaveAnonymousChildren()) {
+ return nullptr;
+ }
+ return static_cast<Element*>(aContent->GetProperty(aPseudoProperty));
+}
+
+/*static*/
+Element* nsLayoutUtils::GetBeforePseudo(const nsIContent* aContent) {
+ return GetPseudo(aContent, nsGkAtoms::beforePseudoProperty);
+}
+
+/*static*/
+nsIFrame* nsLayoutUtils::GetBeforeFrame(const nsIContent* aContent) {
+ Element* pseudo = GetBeforePseudo(aContent);
+ return pseudo ? pseudo->GetPrimaryFrame() : nullptr;
+}
+
+/*static*/
+Element* nsLayoutUtils::GetAfterPseudo(const nsIContent* aContent) {
+ return GetPseudo(aContent, nsGkAtoms::afterPseudoProperty);
+}
+
+/*static*/
+nsIFrame* nsLayoutUtils::GetAfterFrame(const nsIContent* aContent) {
+ Element* pseudo = GetAfterPseudo(aContent);
+ return pseudo ? pseudo->GetPrimaryFrame() : nullptr;
+}
+
+/*static*/
+Element* nsLayoutUtils::GetMarkerPseudo(const nsIContent* aContent) {
+ return GetPseudo(aContent, nsGkAtoms::markerPseudoProperty);
+}
+
+/*static*/
+nsIFrame* nsLayoutUtils::GetMarkerFrame(const nsIContent* aContent) {
+ Element* pseudo = GetMarkerPseudo(aContent);
+ return pseudo ? pseudo->GetPrimaryFrame() : nullptr;
+}
+
+#ifdef ACCESSIBILITY
+void nsLayoutUtils::GetMarkerSpokenText(const nsIContent* aContent,
+ nsAString& aText) {
+ MOZ_ASSERT(aContent && aContent->IsGeneratedContentContainerForMarker());
+
+ aText.Truncate();
+
+ nsIFrame* frame = aContent->GetPrimaryFrame();
+ if (!frame) {
+ return;
+ }
+
+ if (frame->StyleContent()->ContentCount() > 0) {
+ for (nsIFrame* child : frame->PrincipalChildList()) {
+ nsIFrame::RenderedText text = child->GetRenderedText();
+ aText += text.mString;
+ }
+ return;
+ }
+
+ if (!frame->StyleList()->mListStyleImage.IsNone()) {
+ // ::marker is an image, so use default bullet character.
+ static const char16_t kDiscMarkerString[] = {0x2022, ' ', 0};
+ aText.AssignLiteral(kDiscMarkerString);
+ return;
+ }
+
+ frame->PresContext()
+ ->FrameConstructor()
+ ->GetContainStyleScopeManager()
+ .GetSpokenCounterText(frame, aText);
+}
+#endif
+
+// static
+nsIFrame* nsLayoutUtils::GetClosestFrameOfType(nsIFrame* aFrame,
+ LayoutFrameType aFrameType,
+ nsIFrame* aStopAt) {
+ for (nsIFrame* frame = aFrame; frame; frame = frame->GetParent()) {
+ if (frame->Type() == aFrameType) {
+ return frame;
+ }
+ if (frame == aStopAt) {
+ break;
+ }
+ }
+ return nullptr;
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetPageFrame(nsIFrame* aFrame) {
+ return GetClosestFrameOfType(aFrame, LayoutFrameType::Page);
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetStyleFrame(nsIFrame* aPrimaryFrame) {
+ MOZ_ASSERT(aPrimaryFrame);
+ if (aPrimaryFrame->IsTableWrapperFrame()) {
+ nsIFrame* inner = aPrimaryFrame->PrincipalChildList().FirstChild();
+ // inner may be null, if aPrimaryFrame is mid-destruction
+ return inner;
+ }
+
+ return aPrimaryFrame;
+}
+
+const nsIFrame* nsLayoutUtils::GetStyleFrame(const nsIFrame* aPrimaryFrame) {
+ return nsLayoutUtils::GetStyleFrame(const_cast<nsIFrame*>(aPrimaryFrame));
+}
+
+nsIFrame* nsLayoutUtils::GetStyleFrame(const nsIContent* aContent) {
+ nsIFrame* frame = aContent->GetPrimaryFrame();
+ if (!frame) {
+ return nullptr;
+ }
+
+ return nsLayoutUtils::GetStyleFrame(frame);
+}
+
+CSSIntCoord nsLayoutUtils::UnthemedScrollbarSize(StyleScrollbarWidth aWidth) {
+ switch (aWidth) {
+ case StyleScrollbarWidth::Auto:
+ return 12;
+ case StyleScrollbarWidth::Thin:
+ return 6;
+ case StyleScrollbarWidth::None:
+ return 0;
+ }
+ return 0;
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetPrimaryFrameFromStyleFrame(nsIFrame* aStyleFrame) {
+ nsIFrame* parent = aStyleFrame->GetParent();
+ return parent && parent->IsTableWrapperFrame() ? parent : aStyleFrame;
+}
+
+/* static */
+const nsIFrame* nsLayoutUtils::GetPrimaryFrameFromStyleFrame(
+ const nsIFrame* aStyleFrame) {
+ return nsLayoutUtils::GetPrimaryFrameFromStyleFrame(
+ const_cast<nsIFrame*>(aStyleFrame));
+}
+
+/*static*/
+bool nsLayoutUtils::IsPrimaryStyleFrame(const nsIFrame* aFrame) {
+ if (aFrame->IsTableWrapperFrame()) {
+ return false;
+ }
+
+ const nsIFrame* parent = aFrame->GetParent();
+ if (parent && parent->IsTableWrapperFrame()) {
+ return parent->PrincipalChildList().FirstChild() == aFrame;
+ }
+
+ return aFrame->IsPrimaryFrame();
+}
+
+nsIFrame* nsLayoutUtils::GetFloatFromPlaceholder(nsIFrame* aFrame) {
+ NS_ASSERTION(aFrame->IsPlaceholderFrame(), "Must have a placeholder here");
+ if (aFrame->HasAnyStateBits(PLACEHOLDER_FOR_FLOAT)) {
+ nsIFrame* outOfFlowFrame =
+ nsPlaceholderFrame::GetRealFrameForPlaceholder(aFrame);
+ NS_ASSERTION(outOfFlowFrame && outOfFlowFrame->IsFloating(),
+ "How did that happen?");
+ return outOfFlowFrame;
+ }
+
+ return nullptr;
+}
+
+// static
+nsIFrame* nsLayoutUtils::GetCrossDocParentFrameInProcess(
+ const nsIFrame* aFrame, nsPoint* aCrossDocOffset) {
+ nsIFrame* p = aFrame->GetParent();
+ if (p) {
+ return p;
+ }
+
+ nsView* v = aFrame->GetView();
+ if (!v) {
+ return nullptr;
+ }
+ v = v->GetParent(); // anonymous inner view
+ if (!v) {
+ return nullptr;
+ }
+ v = v->GetParent(); // subdocumentframe's view
+ if (!v) {
+ return nullptr;
+ }
+
+ p = v->GetFrame();
+ if (p && aCrossDocOffset) {
+ nsSubDocumentFrame* subdocumentFrame = do_QueryFrame(p);
+ MOZ_ASSERT(subdocumentFrame);
+ *aCrossDocOffset += subdocumentFrame->GetExtraOffset();
+ }
+
+ return p;
+}
+
+// static
+nsIFrame* nsLayoutUtils::GetCrossDocParentFrame(const nsIFrame* aFrame,
+ nsPoint* aCrossDocOffset) {
+ return GetCrossDocParentFrameInProcess(aFrame, aCrossDocOffset);
+}
+
+// static
+bool nsLayoutUtils::IsProperAncestorFrameCrossDoc(
+ const nsIFrame* aAncestorFrame, const nsIFrame* aFrame,
+ const nsIFrame* aCommonAncestor) {
+ if (aFrame == aAncestorFrame) return false;
+ return IsAncestorFrameCrossDoc(aAncestorFrame, aFrame, aCommonAncestor);
+}
+
+// static
+bool nsLayoutUtils::IsProperAncestorFrameCrossDocInProcess(
+ const nsIFrame* aAncestorFrame, const nsIFrame* aFrame,
+ const nsIFrame* aCommonAncestor) {
+ if (aFrame == aAncestorFrame) return false;
+ return IsAncestorFrameCrossDocInProcess(aAncestorFrame, aFrame,
+ aCommonAncestor);
+}
+
+// static
+bool nsLayoutUtils::IsAncestorFrameCrossDoc(const nsIFrame* aAncestorFrame,
+ const nsIFrame* aFrame,
+ const nsIFrame* aCommonAncestor) {
+ for (const nsIFrame* f = aFrame; f != aCommonAncestor;
+ f = GetCrossDocParentFrameInProcess(f)) {
+ if (f == aAncestorFrame) return true;
+ }
+ return aCommonAncestor == aAncestorFrame;
+}
+
+// static
+bool nsLayoutUtils::IsAncestorFrameCrossDocInProcess(
+ const nsIFrame* aAncestorFrame, const nsIFrame* aFrame,
+ const nsIFrame* aCommonAncestor) {
+ for (const nsIFrame* f = aFrame; f != aCommonAncestor;
+ f = GetCrossDocParentFrameInProcess(f)) {
+ if (f == aAncestorFrame) return true;
+ }
+ return aCommonAncestor == aAncestorFrame;
+}
+
+// static
+bool nsLayoutUtils::IsProperAncestorFrame(const nsIFrame* aAncestorFrame,
+ const nsIFrame* aFrame,
+ const nsIFrame* aCommonAncestor) {
+ if (aFrame == aAncestorFrame) return false;
+ for (const nsIFrame* f = aFrame; f != aCommonAncestor; f = f->GetParent()) {
+ if (f == aAncestorFrame) return true;
+ }
+ return aCommonAncestor == aAncestorFrame;
+}
+
+// static
+int32_t nsLayoutUtils::DoCompareTreePosition(
+ nsIContent* aContent1, nsIContent* aContent2, int32_t aIf1Ancestor,
+ int32_t aIf2Ancestor, const nsIContent* aCommonAncestor) {
+ MOZ_ASSERT(aIf1Ancestor == -1 || aIf1Ancestor == 0 || aIf1Ancestor == 1);
+ MOZ_ASSERT(aIf2Ancestor == -1 || aIf2Ancestor == 0 || aIf2Ancestor == 1);
+ MOZ_ASSERT(aContent1, "aContent1 must not be null");
+ MOZ_ASSERT(aContent2, "aContent2 must not be null");
+
+ AutoTArray<nsIContent*, 32> content1Ancestors;
+ nsIContent* c1;
+ for (c1 = aContent1; c1 && c1 != aCommonAncestor;
+ c1 = c1->GetFlattenedTreeParent()) {
+ content1Ancestors.AppendElement(c1);
+ }
+ if (!c1 && aCommonAncestor) {
+ // So, it turns out aCommonAncestor was not an ancestor of c1. Oops.
+ // Never mind. We can continue as if aCommonAncestor was null.
+ aCommonAncestor = nullptr;
+ }
+
+ AutoTArray<nsIContent*, 32> content2Ancestors;
+ nsIContent* c2;
+ for (c2 = aContent2; c2 && c2 != aCommonAncestor;
+ c2 = c2->GetFlattenedTreeParent()) {
+ content2Ancestors.AppendElement(c2);
+ }
+ if (!c2 && aCommonAncestor) {
+ // So, it turns out aCommonAncestor was not an ancestor of c2.
+ // We need to retry with no common ancestor hint.
+ return DoCompareTreePosition(aContent1, aContent2, aIf1Ancestor,
+ aIf2Ancestor, nullptr);
+ }
+
+ int last1 = content1Ancestors.Length() - 1;
+ int last2 = content2Ancestors.Length() - 1;
+ nsIContent* content1Ancestor = nullptr;
+ nsIContent* content2Ancestor = nullptr;
+ while (last1 >= 0 && last2 >= 0 &&
+ ((content1Ancestor = content1Ancestors.ElementAt(last1)) ==
+ (content2Ancestor = content2Ancestors.ElementAt(last2)))) {
+ last1--;
+ last2--;
+ }
+
+ if (last1 < 0) {
+ if (last2 < 0) {
+ NS_ASSERTION(aContent1 == aContent2, "internal error?");
+ return 0;
+ }
+ // aContent1 is an ancestor of aContent2
+ return aIf1Ancestor;
+ }
+
+ if (last2 < 0) {
+ // aContent2 is an ancestor of aContent1
+ return aIf2Ancestor;
+ }
+
+ // content1Ancestor != content2Ancestor, so they must be siblings with the
+ // same parent
+ nsIContent* parent = content1Ancestor->GetFlattenedTreeParent();
+#ifdef DEBUG
+ // TODO: remove the uglyness, see bug 598468.
+ NS_ASSERTION(gPreventAssertInCompareTreePosition || parent,
+ "no common ancestor at all???");
+#endif // DEBUG
+ if (!parent) { // different documents??
+ return 0;
+ }
+
+ const Maybe<uint32_t> index1 =
+ parent->ComputeFlatTreeIndexOf(content1Ancestor);
+ const Maybe<uint32_t> index2 =
+ parent->ComputeFlatTreeIndexOf(content2Ancestor);
+
+ // None of the nodes are anonymous, just do a regular comparison.
+ if (index1.isSome() && index2.isSome()) {
+ return static_cast<int32_t>(static_cast<int64_t>(*index1) - *index2);
+ }
+
+ // Otherwise handle pseudo-element and anonymous content ordering.
+ //
+ // ::marker -> ::before -> anon siblings -> regular siblings -> ::after
+ auto PseudoIndex = [](const nsINode* aNode,
+ const Maybe<uint32_t>& aNodeIndex) -> int32_t {
+ if (aNodeIndex.isSome()) {
+ return 1; // Not a pseudo.
+ }
+ if (aNode->IsContent()) {
+ if (aNode->AsContent()->IsGeneratedContentContainerForMarker()) {
+ return -2;
+ }
+ if (aNode->AsContent()->IsGeneratedContentContainerForBefore()) {
+ return -1;
+ }
+ if (aNode->AsContent()->IsGeneratedContentContainerForAfter()) {
+ return 2;
+ }
+ }
+ return 0;
+ };
+
+ return PseudoIndex(content1Ancestor, index1) -
+ PseudoIndex(content2Ancestor, index2);
+}
+
+// static
+nsIFrame* nsLayoutUtils::FillAncestors(nsIFrame* aFrame,
+ nsIFrame* aStopAtAncestor,
+ nsTArray<nsIFrame*>* aAncestors) {
+ while (aFrame && aFrame != aStopAtAncestor) {
+ aAncestors->AppendElement(aFrame);
+ aFrame = nsLayoutUtils::GetParentOrPlaceholderFor(aFrame);
+ }
+ return aFrame;
+}
+
+// Return true if aFrame1 is after aFrame2
+static bool IsFrameAfter(nsIFrame* aFrame1, nsIFrame* aFrame2) {
+ nsIFrame* f = aFrame2;
+ do {
+ f = f->GetNextSibling();
+ if (f == aFrame1) return true;
+ } while (f);
+ return false;
+}
+
+// static
+int32_t nsLayoutUtils::DoCompareTreePosition(nsIFrame* aFrame1,
+ nsIFrame* aFrame2,
+ int32_t aIf1Ancestor,
+ int32_t aIf2Ancestor,
+ nsIFrame* aCommonAncestor) {
+ MOZ_ASSERT(aIf1Ancestor == -1 || aIf1Ancestor == 0 || aIf1Ancestor == 1);
+ MOZ_ASSERT(aIf2Ancestor == -1 || aIf2Ancestor == 0 || aIf2Ancestor == 1);
+ MOZ_ASSERT(aFrame1, "aFrame1 must not be null");
+ MOZ_ASSERT(aFrame2, "aFrame2 must not be null");
+
+ AutoTArray<nsIFrame*, 20> frame2Ancestors;
+ nsIFrame* nonCommonAncestor =
+ FillAncestors(aFrame2, aCommonAncestor, &frame2Ancestors);
+
+ return DoCompareTreePosition(aFrame1, aFrame2, frame2Ancestors, aIf1Ancestor,
+ aIf2Ancestor,
+ nonCommonAncestor ? aCommonAncestor : nullptr);
+}
+
+// static
+int32_t nsLayoutUtils::DoCompareTreePosition(
+ nsIFrame* aFrame1, nsIFrame* aFrame2, nsTArray<nsIFrame*>& aFrame2Ancestors,
+ int32_t aIf1Ancestor, int32_t aIf2Ancestor, nsIFrame* aCommonAncestor) {
+ MOZ_ASSERT(aIf1Ancestor == -1 || aIf1Ancestor == 0 || aIf1Ancestor == 1);
+ MOZ_ASSERT(aIf2Ancestor == -1 || aIf2Ancestor == 0 || aIf2Ancestor == 1);
+ MOZ_ASSERT(aFrame1, "aFrame1 must not be null");
+ MOZ_ASSERT(aFrame2, "aFrame2 must not be null");
+
+ nsPresContext* presContext = aFrame1->PresContext();
+ if (presContext != aFrame2->PresContext()) {
+ NS_ERROR("no common ancestor at all, different documents");
+ return 0;
+ }
+
+ AutoTArray<nsIFrame*, 20> frame1Ancestors;
+ if (aCommonAncestor &&
+ !FillAncestors(aFrame1, aCommonAncestor, &frame1Ancestors)) {
+ // We reached the root of the frame tree ... if aCommonAncestor was set,
+ // it is wrong
+ return DoCompareTreePosition(aFrame1, aFrame2, aIf1Ancestor, aIf2Ancestor,
+ nullptr);
+ }
+
+ int32_t last1 = int32_t(frame1Ancestors.Length()) - 1;
+ int32_t last2 = int32_t(aFrame2Ancestors.Length()) - 1;
+ while (last1 >= 0 && last2 >= 0 &&
+ frame1Ancestors[last1] == aFrame2Ancestors[last2]) {
+ last1--;
+ last2--;
+ }
+
+ if (last1 < 0) {
+ if (last2 < 0) {
+ NS_ASSERTION(aFrame1 == aFrame2, "internal error?");
+ return 0;
+ }
+ // aFrame1 is an ancestor of aFrame2
+ return aIf1Ancestor;
+ }
+
+ if (last2 < 0) {
+ // aFrame2 is an ancestor of aFrame1
+ return aIf2Ancestor;
+ }
+
+ nsIFrame* ancestor1 = frame1Ancestors[last1];
+ nsIFrame* ancestor2 = aFrame2Ancestors[last2];
+ // Now we should be able to walk sibling chains to find which one is first
+ if (IsFrameAfter(ancestor2, ancestor1)) return -1;
+ if (IsFrameAfter(ancestor1, ancestor2)) return 1;
+ NS_WARNING("Frames were in different child lists???");
+ return 0;
+}
+
+// static
+nsIFrame* nsLayoutUtils::GetLastSibling(nsIFrame* aFrame) {
+ if (!aFrame) {
+ return nullptr;
+ }
+
+ nsIFrame* next;
+ while ((next = aFrame->GetNextSibling()) != nullptr) {
+ aFrame = next;
+ }
+ return aFrame;
+}
+
+// static
+nsView* nsLayoutUtils::FindSiblingViewFor(nsView* aParentView,
+ nsIFrame* aFrame) {
+ nsIFrame* parentViewFrame = aParentView->GetFrame();
+ nsIContent* parentViewContent =
+ parentViewFrame ? parentViewFrame->GetContent() : nullptr;
+ for (nsView* insertBefore = aParentView->GetFirstChild(); insertBefore;
+ insertBefore = insertBefore->GetNextSibling()) {
+ nsIFrame* f = insertBefore->GetFrame();
+ if (!f) {
+ // this view could be some anonymous view attached to a meaningful parent
+ for (nsView* searchView = insertBefore->GetParent(); searchView;
+ searchView = searchView->GetParent()) {
+ f = searchView->GetFrame();
+ if (f) {
+ break;
+ }
+ }
+ NS_ASSERTION(f, "Can't find a frame anywhere!");
+ }
+ if (!f || !aFrame->GetContent() || !f->GetContent() ||
+ CompareTreePosition(aFrame->GetContent(), f->GetContent(),
+ parentViewContent) > 0) {
+ // aFrame's content is after f's content (or we just don't know),
+ // so put our view before f's view
+ return insertBefore;
+ }
+ }
+ return nullptr;
+}
+
+// static
+nsIScrollableFrame* nsLayoutUtils::GetScrollableFrameFor(
+ const nsIFrame* aScrolledFrame) {
+ nsIFrame* frame = aScrolledFrame->GetParent();
+ nsIScrollableFrame* sf = do_QueryFrame(frame);
+ return (sf && sf->GetScrolledFrame() == aScrolledFrame) ? sf : nullptr;
+}
+
+/* static */
+SideBits nsLayoutUtils::GetSideBitsForFixedPositionContent(
+ const nsIFrame* aFixedPosFrame) {
+ SideBits sides = SideBits::eNone;
+ if (aFixedPosFrame) {
+ const nsStylePosition* position = aFixedPosFrame->StylePosition();
+ if (!position->mOffset.Get(eSideRight).IsAuto()) {
+ sides |= SideBits::eRight;
+ }
+ if (!position->mOffset.Get(eSideLeft).IsAuto()) {
+ sides |= SideBits::eLeft;
+ }
+ if (!position->mOffset.Get(eSideBottom).IsAuto()) {
+ sides |= SideBits::eBottom;
+ }
+ if (!position->mOffset.Get(eSideTop).IsAuto()) {
+ sides |= SideBits::eTop;
+ }
+ }
+ return sides;
+}
+
+ScrollableLayerGuid::ViewID nsLayoutUtils::ScrollIdForRootScrollFrame(
+ nsPresContext* aPresContext) {
+ ViewID id = ScrollableLayerGuid::NULL_SCROLL_ID;
+ if (nsIFrame* rootScrollFrame =
+ aPresContext->PresShell()->GetRootScrollFrame()) {
+ if (nsIContent* content = rootScrollFrame->GetContent()) {
+ id = FindOrCreateIDFor(content);
+ }
+ }
+ return id;
+}
+
+// static
+nsIScrollableFrame* nsLayoutUtils::GetNearestScrollableFrameForDirection(
+ nsIFrame* aFrame, ScrollDirections aDirections) {
+ NS_ASSERTION(
+ aFrame, "GetNearestScrollableFrameForDirection expects a non-null frame");
+ // FIXME Bug 1714720 : This nearest scroll target is not going to work over
+ // process boundaries, in such cases we need to hand over in APZ side.
+ for (nsIFrame* f = aFrame; f;
+ f = nsLayoutUtils::GetCrossDocParentFrameInProcess(f)) {
+ nsIScrollableFrame* scrollableFrame = do_QueryFrame(f);
+ if (scrollableFrame) {
+ ScrollDirections directions =
+ scrollableFrame->GetAvailableScrollingDirectionsForUserInputEvents();
+ if (aDirections.contains(ScrollDirection::eVertical)) {
+ if (directions.contains(ScrollDirection::eVertical)) {
+ return scrollableFrame;
+ }
+ }
+ if (aDirections.contains(ScrollDirection::eHorizontal)) {
+ if (directions.contains(ScrollDirection::eHorizontal)) {
+ return scrollableFrame;
+ }
+ }
+ }
+ }
+ return nullptr;
+}
+
+static nsIFrame* GetNearestScrollableOrOverflowClipFrame(
+ nsIFrame* aFrame, uint32_t aFlags,
+ const std::function<bool(const nsIFrame* aCurrentFrame)>& aClipFrameCheck =
+ nullptr) {
+ MOZ_ASSERT(
+ aFrame,
+ "GetNearestScrollableOrOverflowClipFrame expects a non-null frame");
+
+ auto GetNextFrame = [aFlags](const nsIFrame* aFrame) -> nsIFrame* {
+ return (aFlags & nsLayoutUtils::SCROLLABLE_SAME_DOC)
+ ? aFrame->GetParent()
+ : nsLayoutUtils::GetCrossDocParentFrameInProcess(aFrame);
+ };
+
+ for (nsIFrame* f = aFrame; f; f = GetNextFrame(f)) {
+ if (aClipFrameCheck && aClipFrameCheck(f)) {
+ return f;
+ }
+
+ if ((aFlags & nsLayoutUtils::SCROLLABLE_STOP_AT_PAGE) && f->IsPageFrame()) {
+ break;
+ }
+ if (nsIScrollableFrame* scrollableFrame = do_QueryFrame(f)) {
+ if (aFlags & nsLayoutUtils::SCROLLABLE_ONLY_ASYNC_SCROLLABLE) {
+ if (scrollableFrame->WantAsyncScroll()) {
+ return f;
+ }
+ } else {
+ ScrollStyles ss = scrollableFrame->GetScrollStyles();
+ if ((aFlags & nsLayoutUtils::SCROLLABLE_INCLUDE_HIDDEN) ||
+ ss.mVertical != StyleOverflow::Hidden ||
+ ss.mHorizontal != StyleOverflow::Hidden) {
+ return f;
+ }
+ }
+ if (aFlags & nsLayoutUtils::SCROLLABLE_ALWAYS_MATCH_ROOT) {
+ PresShell* presShell = f->PresShell();
+ if (presShell->GetRootScrollFrame() == f && presShell->GetDocument() &&
+ presShell->GetDocument()->IsRootDisplayDocument()) {
+ return f;
+ }
+ }
+ }
+ if ((aFlags & nsLayoutUtils::SCROLLABLE_FIXEDPOS_FINDS_ROOT) &&
+ f->StyleDisplay()->mPosition == StylePositionProperty::Fixed &&
+ nsLayoutUtils::IsReallyFixedPos(f)) {
+ return f->PresShell()->GetRootScrollFrame();
+ }
+ }
+ return nullptr;
+}
+
+// static
+nsIScrollableFrame* nsLayoutUtils::GetNearestScrollableFrame(nsIFrame* aFrame,
+ uint32_t aFlags) {
+ nsIFrame* found = GetNearestScrollableOrOverflowClipFrame(aFrame, aFlags);
+ if (!found) {
+ return nullptr;
+ }
+
+ return do_QueryFrame(found);
+}
+
+// static
+nsIFrame* nsLayoutUtils::GetNearestOverflowClipFrame(nsIFrame* aFrame) {
+ return GetNearestScrollableOrOverflowClipFrame(
+ aFrame, SCROLLABLE_SAME_DOC | SCROLLABLE_INCLUDE_HIDDEN,
+ [](const nsIFrame* currentFrame) -> bool {
+ // In cases of SVG Inner/Outer frames it basically clips descendants
+ // unless overflow: visible is explicitly specified.
+ LayoutFrameType type = currentFrame->Type();
+ return ((type == LayoutFrameType::SVGOuterSVG ||
+ type == LayoutFrameType::SVGInnerSVG) &&
+ (currentFrame->StyleDisplay()->mOverflowX !=
+ StyleOverflow::Visible &&
+ currentFrame->StyleDisplay()->mOverflowY !=
+ StyleOverflow::Visible));
+ });
+}
+
+// static
+nsRect nsLayoutUtils::GetScrolledRect(nsIFrame* aScrolledFrame,
+ const nsRect& aScrolledFrameOverflowArea,
+ const nsSize& aScrollPortSize,
+ StyleDirection aDirection) {
+ WritingMode wm = aScrolledFrame->GetWritingMode();
+ // Potentially override the frame's direction to use the direction found
+ // by nsHTMLScrollFrame::GetScrolledFrameDir()
+ wm.SetDirectionFromBidiLevel(aDirection == StyleDirection::Rtl
+ ? mozilla::intl::BidiEmbeddingLevel::RTL()
+ : mozilla::intl::BidiEmbeddingLevel::LTR());
+
+ nscoord x1 = aScrolledFrameOverflowArea.x,
+ x2 = aScrolledFrameOverflowArea.XMost(),
+ y1 = aScrolledFrameOverflowArea.y,
+ y2 = aScrolledFrameOverflowArea.YMost();
+
+ const bool isHorizontalWM = !wm.IsVertical();
+ const bool isVerticalWM = wm.IsVertical();
+ bool isInlineFlowFromTopOrLeft = !wm.IsInlineReversed();
+ bool isBlockFlowFromTopOrLeft = isHorizontalWM || wm.IsVerticalLR();
+
+ if (aScrolledFrame->IsFlexContainerFrame()) {
+ // In a flex container, the children flow (and overflow) along the flex
+ // container's main axis and cross axis. These are analogous to the
+ // inline/block axes, and by default they correspond exactly to those axes;
+ // but the flex container's CSS (e.g. flex-direction: column-reverse) may
+ // have swapped and/or reversed them, and we need to account for that here.
+ FlexboxAxisInfo info(aScrolledFrame);
+ if (info.mIsRowOriented) {
+ // The flex container's inline axis is the main axis.
+ isInlineFlowFromTopOrLeft =
+ isInlineFlowFromTopOrLeft == !info.mIsMainAxisReversed;
+ isBlockFlowFromTopOrLeft =
+ isBlockFlowFromTopOrLeft == !info.mIsCrossAxisReversed;
+ } else {
+ // The flex container's block axis is the main axis.
+ isBlockFlowFromTopOrLeft =
+ isBlockFlowFromTopOrLeft == !info.mIsMainAxisReversed;
+ isInlineFlowFromTopOrLeft =
+ isInlineFlowFromTopOrLeft == !info.mIsCrossAxisReversed;
+ }
+ }
+
+ // Clamp the horizontal start-edge (x1 or x2, depending whether the logical
+ // axis that corresponds to horizontal progresses from left-to-right or
+ // right-to-left).
+ if ((isHorizontalWM && isInlineFlowFromTopOrLeft) ||
+ (isVerticalWM && isBlockFlowFromTopOrLeft)) {
+ if (x1 < 0) {
+ x1 = 0;
+ }
+ } else {
+ if (x2 > aScrollPortSize.width) {
+ x2 = aScrollPortSize.width;
+ }
+ // When the scrolled frame chooses a size larger than its available width
+ // (because its padding alone is larger than the available width), we need
+ // to keep the start-edge of the scroll frame anchored to the start-edge of
+ // the scrollport.
+ // When the scrolled frame is RTL, this means moving it in our left-based
+ // coordinate system, so we need to compensate for its extra width here by
+ // effectively repositioning the frame.
+ nscoord extraWidth =
+ std::max(0, aScrolledFrame->GetSize().width - aScrollPortSize.width);
+ x2 += extraWidth;
+ }
+
+ // Similarly, clamp the vertical start-edge (y1 or y2, depending whether the
+ // logical axis that corresponds to vertical progresses from top-to-bottom or
+ // buttom-to-top).
+ if ((isHorizontalWM && isBlockFlowFromTopOrLeft) ||
+ (isVerticalWM && isInlineFlowFromTopOrLeft)) {
+ if (y1 < 0) {
+ y1 = 0;
+ }
+ } else {
+ if (y2 > aScrollPortSize.height) {
+ y2 = aScrollPortSize.height;
+ }
+ nscoord extraHeight =
+ std::max(0, aScrolledFrame->GetSize().height - aScrollPortSize.height);
+ y2 += extraHeight;
+ }
+
+ return nsRect(x1, y1, x2 - x1, y2 - y1);
+}
+
+// static
+bool nsLayoutUtils::HasPseudoStyle(nsIContent* aContent,
+ ComputedStyle* aComputedStyle,
+ PseudoStyleType aPseudoElement,
+ nsPresContext* aPresContext) {
+ MOZ_ASSERT(aPresContext, "Must have a prescontext");
+
+ RefPtr<ComputedStyle> pseudoContext;
+ if (aContent) {
+ pseudoContext = aPresContext->StyleSet()->ProbePseudoElementStyle(
+ *aContent->AsElement(), aPseudoElement, aComputedStyle);
+ }
+ return pseudoContext != nullptr;
+}
+
+nsPoint nsLayoutUtils::GetDOMEventCoordinatesRelativeTo(Event* aDOMEvent,
+ nsIFrame* aFrame) {
+ if (!aDOMEvent) return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ WidgetEvent* event = aDOMEvent->WidgetEventPtr();
+ if (!event) return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ return GetEventCoordinatesRelativeTo(event, RelativeTo{aFrame});
+}
+
+static bool IsValidCoordinateTypeEvent(const WidgetEvent* aEvent) {
+ if (!aEvent) {
+ return false;
+ }
+ return aEvent->mClass == eMouseEventClass ||
+ aEvent->mClass == eMouseScrollEventClass ||
+ aEvent->mClass == eWheelEventClass ||
+ aEvent->mClass == eDragEventClass ||
+ aEvent->mClass == eSimpleGestureEventClass ||
+ aEvent->mClass == ePointerEventClass ||
+ aEvent->mClass == eGestureNotifyEventClass ||
+ aEvent->mClass == eTouchEventClass ||
+ aEvent->mClass == eQueryContentEventClass;
+}
+
+nsPoint nsLayoutUtils::GetEventCoordinatesRelativeTo(const WidgetEvent* aEvent,
+ RelativeTo aFrame) {
+ if (!IsValidCoordinateTypeEvent(aEvent)) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ return GetEventCoordinatesRelativeTo(aEvent, aEvent->AsGUIEvent()->mRefPoint,
+ aFrame);
+}
+
+nsPoint nsLayoutUtils::GetEventCoordinatesRelativeTo(
+ const WidgetEvent* aEvent, const LayoutDeviceIntPoint& aPoint,
+ RelativeTo aFrame) {
+ if (!aFrame.mFrame) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ nsIWidget* widget = aEvent->AsGUIEvent()->mWidget;
+ if (!widget) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ return GetEventCoordinatesRelativeTo(widget, aPoint, aFrame);
+}
+
+nsPoint GetEventCoordinatesRelativeTo(nsIWidget* aWidget,
+ const LayoutDeviceIntPoint& aPoint,
+ RelativeTo aFrame) {
+ const nsIFrame* frame = aFrame.mFrame;
+ if (!frame || !aWidget) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ nsView* view = frame->GetView();
+ if (view) {
+ nsIWidget* frameWidget = view->GetWidget();
+ if (frameWidget && frameWidget == aWidget) {
+ // Special case this cause it happens a lot.
+ // This also fixes bug 664707, events in the extra-special case of select
+ // dropdown popups that are transformed.
+ nsPresContext* presContext = frame->PresContext();
+ nsPoint pt(presContext->DevPixelsToAppUnits(aPoint.x),
+ presContext->DevPixelsToAppUnits(aPoint.y));
+ return pt - view->ViewToWidgetOffset();
+ }
+ }
+
+ /* If we walk up the frame tree and discover that any of the frames are
+ * transformed, we need to do extra work to convert from the global
+ * space to the local space.
+ */
+ const nsIFrame* rootFrame = frame;
+ bool transformFound = false;
+ for (const nsIFrame* f = frame; f;
+ f = nsLayoutUtils::GetCrossDocParentFrameInProcess(f)) {
+ if (f->IsTransformed() || ViewportUtils::IsZoomedContentRoot(f)) {
+ transformFound = true;
+ }
+
+ rootFrame = f;
+ }
+
+ nsView* rootView = rootFrame->GetView();
+ if (!rootView) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ nsPoint widgetToView = nsLayoutUtils::TranslateWidgetToView(
+ rootFrame->PresContext(), aWidget, aPoint, rootView);
+
+ if (widgetToView == nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE)) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ // Convert from root document app units to app units of the document aFrame
+ // is in.
+ int32_t rootAPD = rootFrame->PresContext()->AppUnitsPerDevPixel();
+ int32_t localAPD = frame->PresContext()->AppUnitsPerDevPixel();
+ widgetToView = widgetToView.ScaleToOtherAppUnits(rootAPD, localAPD);
+
+ /* If we encountered a transform, we can't do simple arithmetic to figure
+ * out how to convert back to aFrame's coordinates and must use the CTM.
+ */
+ if (transformFound || frame->IsInSVGTextSubtree()) {
+ return nsLayoutUtils::TransformRootPointToFrame(ViewportType::Visual,
+ aFrame, widgetToView);
+ }
+
+ /* Otherwise, all coordinate systems are translations of one another,
+ * so we can just subtract out the difference.
+ */
+ return widgetToView - frame->GetOffsetToCrossDoc(rootFrame);
+}
+
+nsPoint nsLayoutUtils::GetEventCoordinatesRelativeTo(
+ nsIWidget* aWidget, const LayoutDeviceIntPoint& aPoint, RelativeTo aFrame) {
+ nsPoint result = ::GetEventCoordinatesRelativeTo(aWidget, aPoint, aFrame);
+ if (aFrame.mViewportType == ViewportType::Layout && aFrame.mFrame &&
+ aFrame.mFrame->Type() == LayoutFrameType::Viewport &&
+ aFrame.mFrame->PresContext()->IsRootContentDocumentCrossProcess()) {
+ result = ViewportUtils::VisualToLayout(result, aFrame.mFrame->PresShell());
+ }
+ return result;
+}
+
+nsIFrame* nsLayoutUtils::GetPopupFrameForEventCoordinates(
+ nsPresContext* aRootPresContext, const WidgetEvent* aEvent) {
+ if (!IsValidCoordinateTypeEvent(aEvent)) {
+ return nullptr;
+ }
+
+ const auto* guiEvent = aEvent->AsGUIEvent();
+ return GetPopupFrameForPoint(aRootPresContext, guiEvent->mWidget,
+ guiEvent->mRefPoint);
+}
+
+nsIFrame* nsLayoutUtils::GetPopupFrameForPoint(
+ nsPresContext* aRootPresContext, nsIWidget* aWidget,
+ const mozilla::LayoutDeviceIntPoint& aPoint,
+ GetPopupFrameForPointFlags aFlags /* = GetPopupFrameForPointFlags(0) */) {
+ nsXULPopupManager* pm = nsXULPopupManager::GetInstance();
+ if (!pm) {
+ return nullptr;
+ }
+ nsTArray<nsIFrame*> popups;
+ pm->GetVisiblePopups(popups);
+ // Search from top to bottom
+ for (nsIFrame* popup : popups) {
+ if (popup->PresContext()->GetRootPresContext() != aRootPresContext) {
+ continue;
+ }
+ if (!popup->ScrollableOverflowRect().Contains(GetEventCoordinatesRelativeTo(
+ aWidget, aPoint, RelativeTo{popup}))) {
+ continue;
+ }
+ if (aFlags & GetPopupFrameForPointFlags::OnlyReturnFramesWithWidgets) {
+ if (!popup->HasView() || !popup->GetView()->HasWidget()) {
+ continue;
+ }
+ }
+ return popup;
+ }
+ return nullptr;
+}
+
+void nsLayoutUtils::GetContainerAndOffsetAtEvent(PresShell* aPresShell,
+ const WidgetEvent* aEvent,
+ nsIContent** aContainer,
+ int32_t* aOffset) {
+ MOZ_ASSERT(aContainer || aOffset);
+
+ if (aContainer) {
+ *aContainer = nullptr;
+ }
+ if (aOffset) {
+ *aOffset = 0;
+ }
+
+ if (!aPresShell) {
+ return;
+ }
+
+ aPresShell->FlushPendingNotifications(FlushType::Layout);
+
+ RefPtr<nsPresContext> presContext = aPresShell->GetPresContext();
+ if (!presContext) {
+ return;
+ }
+
+ nsIFrame* targetFrame = presContext->EventStateManager()->GetEventTarget();
+ if (!targetFrame) {
+ return;
+ }
+
+ WidgetEvent* openingEvent = nullptr;
+ // For popupshowing events, redirect via the original mouse event
+ // that triggered the popup to open.
+ if (aEvent->mMessage == eXULPopupShowing) {
+ if (auto* pm = nsXULPopupManager::GetInstance()) {
+ if (Event* openingPopupEvent = pm->GetOpeningPopupEvent()) {
+ openingEvent = openingPopupEvent->WidgetEventPtr();
+ }
+ }
+ }
+
+ nsPoint point = nsLayoutUtils::GetEventCoordinatesRelativeTo(
+ openingEvent ? openingEvent : aEvent, RelativeTo{targetFrame});
+
+ if (aContainer) {
+ // TODO: This result may be useful to change to Selection. However, this
+ // may return improper node (e.g., native anonymous node) for the
+ // Selection. Perhaps, this should take Selection optionally and
+ // if it's specified, needs to check if it's proper for the
+ // Selection.
+ nsCOMPtr<nsIContent> container =
+ targetFrame->GetContentOffsetsFromPoint(point).content;
+ if (container && (!container->ChromeOnlyAccess() ||
+ nsContentUtils::CanAccessNativeAnon())) {
+ container.forget(aContainer);
+ }
+ }
+ if (aOffset) {
+ *aOffset = targetFrame->GetContentOffsetsFromPoint(point).offset;
+ }
+}
+
+void nsLayoutUtils::ConstrainToCoordValues(float& aStart, float& aSize) {
+ MOZ_ASSERT(aSize >= 0);
+
+ // Here we try to make sure that the resulting nsRect will continue to cover
+ // as much of the area that was covered by the original gfx Rect as possible.
+
+ // We clamp the bounds of the rect to {nscoord_MIN,nscoord_MAX} since
+ // nsRect::X/Y() and nsRect::XMost/YMost() can't return values outwith this
+ // range:
+ float end = aStart + aSize;
+ aStart = clamped(aStart, float(nscoord_MIN), float(nscoord_MAX));
+ end = clamped(end, float(nscoord_MIN), float(nscoord_MAX));
+
+ aSize = end - aStart;
+
+ // We must also clamp aSize to {0,nscoord_MAX} since nsRect::Width/Height()
+ // can't return a value greater than nscoord_MAX. If aSize is greater than
+ // nscoord_MAX then we reduce it to nscoord_MAX while keeping the rect
+ // centered:
+ if (MOZ_UNLIKELY(std::isnan(aSize))) {
+ // Can happen if aStart is -inf and aSize is +inf for example.
+ aStart = 0.0f;
+ aSize = float(nscoord_MAX);
+ } else if (aSize > float(nscoord_MAX)) {
+ float excess = aSize - float(nscoord_MAX);
+ excess /= 2;
+ aStart += excess;
+ aSize = float(nscoord_MAX);
+ }
+}
+
+/**
+ * Given a gfxFloat, constrains its value to be between nscoord_MIN and
+ * nscoord_MAX.
+ *
+ * @param aVal The value to constrain (in/out)
+ */
+static void ConstrainToCoordValues(gfxFloat& aVal) {
+ if (aVal <= nscoord_MIN)
+ aVal = nscoord_MIN;
+ else if (aVal >= nscoord_MAX)
+ aVal = nscoord_MAX;
+}
+
+void nsLayoutUtils::ConstrainToCoordValues(gfxFloat& aStart, gfxFloat& aSize) {
+ gfxFloat max = aStart + aSize;
+
+ // Clamp the end points to within nscoord range
+ ::ConstrainToCoordValues(aStart);
+ ::ConstrainToCoordValues(max);
+
+ aSize = max - aStart;
+ // If the width if still greater than the max nscoord, then bring both
+ // endpoints in by the same amount until it fits.
+ if (MOZ_UNLIKELY(std::isnan(aSize))) {
+ // Can happen if aStart is -inf and aSize is +inf for example.
+ aStart = 0.0f;
+ aSize = nscoord_MAX;
+ } else if (aSize > nscoord_MAX) {
+ gfxFloat excess = aSize - nscoord_MAX;
+ excess /= 2;
+
+ aStart += excess;
+ aSize = nscoord_MAX;
+ } else if (aSize < nscoord_MIN) {
+ gfxFloat excess = aSize - nscoord_MIN;
+ excess /= 2;
+
+ aStart -= excess;
+ aSize = nscoord_MIN;
+ }
+}
+
+nsRegion nsLayoutUtils::RoundedRectIntersectRect(const nsRect& aRoundedRect,
+ const nscoord aRadii[8],
+ const nsRect& aContainedRect) {
+ // rectFullHeight and rectFullWidth together will approximately contain
+ // the total area of the frame minus the rounded corners.
+ nsRect rectFullHeight = aRoundedRect;
+ nscoord xDiff = std::max(aRadii[eCornerTopLeftX], aRadii[eCornerBottomLeftX]);
+ rectFullHeight.x += xDiff;
+ rectFullHeight.width -=
+ std::max(aRadii[eCornerTopRightX], aRadii[eCornerBottomRightX]) + xDiff;
+ nsRect r1;
+ r1.IntersectRect(rectFullHeight, aContainedRect);
+
+ nsRect rectFullWidth = aRoundedRect;
+ nscoord yDiff = std::max(aRadii[eCornerTopLeftY], aRadii[eCornerTopRightY]);
+ rectFullWidth.y += yDiff;
+ rectFullWidth.height -=
+ std::max(aRadii[eCornerBottomLeftY], aRadii[eCornerBottomRightY]) + yDiff;
+ nsRect r2;
+ r2.IntersectRect(rectFullWidth, aContainedRect);
+
+ nsRegion result;
+ result.Or(r1, r2);
+ return result;
+}
+
+nsIntRegion nsLayoutUtils::RoundedRectIntersectIntRect(
+ const nsIntRect& aRoundedRect, const RectCornerRadii& aCornerRadii,
+ const nsIntRect& aContainedRect) {
+ // rectFullHeight and rectFullWidth together will approximately contain
+ // the total area of the frame minus the rounded corners.
+ nsIntRect rectFullHeight = aRoundedRect;
+ uint32_t xDiff =
+ std::max(aCornerRadii.TopLeft().width, aCornerRadii.BottomLeft().width);
+ rectFullHeight.x += xDiff;
+ rectFullHeight.width -= std::max(aCornerRadii.TopRight().width,
+ aCornerRadii.BottomRight().width) +
+ xDiff;
+ nsIntRect r1;
+ r1.IntersectRect(rectFullHeight, aContainedRect);
+
+ nsIntRect rectFullWidth = aRoundedRect;
+ uint32_t yDiff =
+ std::max(aCornerRadii.TopLeft().height, aCornerRadii.TopRight().height);
+ rectFullWidth.y += yDiff;
+ rectFullWidth.height -= std::max(aCornerRadii.BottomLeft().height,
+ aCornerRadii.BottomRight().height) +
+ yDiff;
+ nsIntRect r2;
+ r2.IntersectRect(rectFullWidth, aContainedRect);
+
+ nsIntRegion result;
+ result.Or(r1, r2);
+ return result;
+}
+
+// Helper for RoundedRectIntersectsRect.
+static bool CheckCorner(nscoord aXOffset, nscoord aYOffset, nscoord aXRadius,
+ nscoord aYRadius) {
+ MOZ_ASSERT(aXOffset > 0 && aYOffset > 0,
+ "must not pass nonpositives to CheckCorner");
+ MOZ_ASSERT(aXRadius >= 0 && aYRadius >= 0,
+ "must not pass negatives to CheckCorner");
+
+ // Avoid floating point math unless we're either (1) within the
+ // quarter-ellipse area at the rounded corner or (2) outside the
+ // rounding.
+ if (aXOffset >= aXRadius || aYOffset >= aYRadius) return true;
+
+ // Convert coordinates to a unit circle with (0,0) as the center of
+ // curvature, and see if we're inside the circle or outside.
+ float scaledX = float(aXRadius - aXOffset) / float(aXRadius);
+ float scaledY = float(aYRadius - aYOffset) / float(aYRadius);
+ return scaledX * scaledX + scaledY * scaledY < 1.0f;
+}
+
+bool nsLayoutUtils::RoundedRectIntersectsRect(const nsRect& aRoundedRect,
+ const nscoord aRadii[8],
+ const nsRect& aTestRect) {
+ if (!aTestRect.Intersects(aRoundedRect)) return false;
+
+ // distances from this edge of aRoundedRect to opposite edge of aTestRect,
+ // which we know are positive due to the Intersects check above.
+ nsMargin insets;
+ insets.top = aTestRect.YMost() - aRoundedRect.y;
+ insets.right = aRoundedRect.XMost() - aTestRect.x;
+ insets.bottom = aRoundedRect.YMost() - aTestRect.y;
+ insets.left = aTestRect.XMost() - aRoundedRect.x;
+
+ // Check whether the bottom-right corner of aTestRect is inside the
+ // top left corner of aBounds when rounded by aRadii, etc. If any
+ // corner is not, then fail; otherwise succeed.
+ return CheckCorner(insets.left, insets.top, aRadii[eCornerTopLeftX],
+ aRadii[eCornerTopLeftY]) &&
+ CheckCorner(insets.right, insets.top, aRadii[eCornerTopRightX],
+ aRadii[eCornerTopRightY]) &&
+ CheckCorner(insets.right, insets.bottom, aRadii[eCornerBottomRightX],
+ aRadii[eCornerBottomRightY]) &&
+ CheckCorner(insets.left, insets.bottom, aRadii[eCornerBottomLeftX],
+ aRadii[eCornerBottomLeftY]);
+}
+
+nsRect nsLayoutUtils::MatrixTransformRect(const nsRect& aBounds,
+ const Matrix4x4& aMatrix,
+ float aFactor) {
+ RectDouble image =
+ RectDouble(NSAppUnitsToDoublePixels(aBounds.x, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.y, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.width, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.height, aFactor));
+
+ RectDouble maxBounds = RectDouble(
+ double(nscoord_MIN) / aFactor * 0.5, double(nscoord_MIN) / aFactor * 0.5,
+ double(nscoord_MAX) / aFactor, double(nscoord_MAX) / aFactor);
+
+ image = aMatrix.TransformAndClipBounds(image, maxBounds);
+
+ return RoundGfxRectToAppRect(ThebesRect(image), aFactor);
+}
+
+nsRect nsLayoutUtils::MatrixTransformRect(const nsRect& aBounds,
+ const Matrix4x4Flagged& aMatrix,
+ float aFactor) {
+ RectDouble image =
+ RectDouble(NSAppUnitsToDoublePixels(aBounds.x, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.y, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.width, aFactor),
+ NSAppUnitsToDoublePixels(aBounds.height, aFactor));
+
+ RectDouble maxBounds = RectDouble(
+ double(nscoord_MIN) / aFactor * 0.5, double(nscoord_MIN) / aFactor * 0.5,
+ double(nscoord_MAX) / aFactor, double(nscoord_MAX) / aFactor);
+
+ image = aMatrix.TransformAndClipBounds(image, maxBounds);
+
+ return RoundGfxRectToAppRect(ThebesRect(image), aFactor);
+}
+
+nsPoint nsLayoutUtils::MatrixTransformPoint(const nsPoint& aPoint,
+ const Matrix4x4& aMatrix,
+ float aFactor) {
+ gfxPoint image = gfxPoint(NSAppUnitsToFloatPixels(aPoint.x, aFactor),
+ NSAppUnitsToFloatPixels(aPoint.y, aFactor));
+ image = aMatrix.TransformPoint(image);
+ return nsPoint(NSFloatPixelsToAppUnits(float(image.x), aFactor),
+ NSFloatPixelsToAppUnits(float(image.y), aFactor));
+}
+
+void nsLayoutUtils::PostTranslate(Matrix4x4& aTransform, const nsPoint& aOrigin,
+ float aAppUnitsPerPixel, bool aRounded) {
+ Point3D gfxOrigin =
+ Point3D(NSAppUnitsToFloatPixels(aOrigin.x, aAppUnitsPerPixel),
+ NSAppUnitsToFloatPixels(aOrigin.y, aAppUnitsPerPixel), 0.0f);
+ if (aRounded) {
+ gfxOrigin.x = NS_round(gfxOrigin.x);
+ gfxOrigin.y = NS_round(gfxOrigin.y);
+ }
+ aTransform.PostTranslate(gfxOrigin);
+}
+
+bool nsLayoutUtils::ShouldSnapToGrid(const nsIFrame* aFrame) {
+ return !aFrame || !aFrame->HasAnyStateBits(NS_FRAME_SVG_LAYOUT) ||
+ aFrame->IsSVGOuterSVGAnonChildFrame();
+}
+
+Matrix4x4Flagged nsLayoutUtils::GetTransformToAncestor(
+ RelativeTo aFrame, RelativeTo aAncestor, uint32_t aFlags,
+ nsIFrame** aOutAncestor) {
+ nsIFrame* parent;
+ Matrix4x4Flagged ctm;
+ // Make sure we don't get an invalid combination of source and destination
+ // RelativeTo values.
+ MOZ_ASSERT(!(aFrame.mViewportType == ViewportType::Visual &&
+ aAncestor.mViewportType == ViewportType::Layout));
+ if (aFrame == aAncestor) {
+ return ctm;
+ }
+ ctm = aFrame.mFrame->GetTransformMatrix(aFrame.mViewportType, aAncestor,
+ &parent, aFlags);
+ if (!aFrame.mFrame->Combines3DTransformWithAncestors()) {
+ ctm.ProjectTo2D();
+ }
+ while (parent && parent != aAncestor.mFrame &&
+ (!(aFlags & nsIFrame::STOP_AT_STACKING_CONTEXT_AND_DISPLAY_PORT) ||
+ (!parent->IsStackingContext() &&
+ !DisplayPortUtils::FrameHasDisplayPort(parent)))) {
+ nsIFrame* cur = parent;
+ ctm = ctm * cur->GetTransformMatrix(aFrame.mViewportType, aAncestor,
+ &parent, aFlags);
+ if (!cur->Combines3DTransformWithAncestors()) {
+ ctm.ProjectTo2D();
+ }
+ }
+ if (aOutAncestor) {
+ *aOutAncestor = parent;
+ }
+ return ctm;
+}
+
+MatrixScales nsLayoutUtils::GetTransformToAncestorScale(
+ const nsIFrame* aFrame) {
+ Matrix4x4Flagged transform = GetTransformToAncestor(
+ RelativeTo{aFrame},
+ RelativeTo{nsLayoutUtils::GetDisplayRootFrame(aFrame)});
+ Matrix transform2D;
+ if (transform.CanDraw2D(&transform2D)) {
+ return ThebesMatrix(transform2D).ScaleFactors().ConvertTo<float>();
+ }
+ return MatrixScales();
+}
+
+static Matrix4x4Flagged GetTransformToAncestorExcludingAnimated(
+ nsIFrame* aFrame, const nsIFrame* aAncestor) {
+ nsIFrame* parent;
+ Matrix4x4Flagged ctm;
+ if (aFrame == aAncestor) {
+ return ctm;
+ }
+ if (ActiveLayerTracker::IsScaleSubjectToAnimation(aFrame)) {
+ return ctm;
+ }
+ ctm = aFrame->GetTransformMatrix(ViewportType::Layout, RelativeTo{aAncestor},
+ &parent);
+ while (parent && parent != aAncestor) {
+ if (ActiveLayerTracker::IsScaleSubjectToAnimation(parent)) {
+ return Matrix4x4Flagged();
+ }
+ if (!parent->Extend3DContext()) {
+ ctm.ProjectTo2D();
+ }
+ ctm = ctm * parent->GetTransformMatrix(ViewportType::Layout,
+ RelativeTo{aAncestor}, &parent);
+ }
+ return ctm;
+}
+
+MatrixScales nsLayoutUtils::GetTransformToAncestorScaleExcludingAnimated(
+ nsIFrame* aFrame) {
+ Matrix4x4Flagged transform = GetTransformToAncestorExcludingAnimated(
+ aFrame, nsLayoutUtils::GetDisplayRootFrame(aFrame));
+ Matrix transform2D;
+ if (transform.Is2D(&transform2D)) {
+ return ThebesMatrix(transform2D).ScaleFactors().ConvertTo<float>();
+ }
+ return MatrixScales();
+}
+
+const nsIFrame* nsLayoutUtils::FindNearestCommonAncestorFrame(
+ const nsIFrame* aFrame1, const nsIFrame* aFrame2) {
+ AutoTArray<const nsIFrame*, 100> ancestors1;
+ AutoTArray<const nsIFrame*, 100> ancestors2;
+ const nsIFrame* commonAncestor = nullptr;
+ if (aFrame1->PresContext() == aFrame2->PresContext()) {
+ commonAncestor = aFrame1->PresShell()->GetRootFrame();
+ }
+ for (const nsIFrame* f = aFrame1; f != commonAncestor;
+ f = nsLayoutUtils::GetCrossDocParentFrameInProcess(f)) {
+ ancestors1.AppendElement(f);
+ }
+ for (const nsIFrame* f = aFrame2; f != commonAncestor;
+ f = nsLayoutUtils::GetCrossDocParentFrameInProcess(f)) {
+ ancestors2.AppendElement(f);
+ }
+ uint32_t minLengths = std::min(ancestors1.Length(), ancestors2.Length());
+ for (uint32_t i = 1; i <= minLengths; ++i) {
+ if (ancestors1[ancestors1.Length() - i] ==
+ ancestors2[ancestors2.Length() - i]) {
+ commonAncestor = ancestors1[ancestors1.Length() - i];
+ } else {
+ break;
+ }
+ }
+ return commonAncestor;
+}
+
+const nsIFrame* nsLayoutUtils::FindNearestCommonAncestorFrameWithinBlock(
+ const nsTextFrame* aFrame1, const nsTextFrame* aFrame2) {
+ MOZ_ASSERT(aFrame1);
+ MOZ_ASSERT(aFrame2);
+
+ const nsIFrame* f1 = aFrame1;
+ const nsIFrame* f2 = aFrame2;
+
+ int n1 = 1;
+ int n2 = 1;
+
+ for (auto f = f1->GetParent();;) {
+ NS_ASSERTION(f, "All text frames should have a block ancestor");
+ if (!f) {
+ return nullptr;
+ }
+ if (f->IsBlockFrameOrSubclass()) {
+ break;
+ }
+ ++n1;
+ f = f->GetParent();
+ }
+
+ for (auto f = f2->GetParent();;) {
+ NS_ASSERTION(f, "All text frames should have a block ancestor");
+ if (!f) {
+ return nullptr;
+ }
+ if (f->IsBlockFrameOrSubclass()) {
+ break;
+ }
+ ++n2;
+ f = f->GetParent();
+ }
+
+ if (n1 > n2) {
+ std::swap(n1, n2);
+ std::swap(f1, f2);
+ }
+
+ while (n2 > n1) {
+ f2 = f2->GetParent();
+ --n2;
+ }
+
+ while (n2 >= 0) {
+ if (f1 == f2) {
+ return f1;
+ }
+ f1 = f1->GetParent();
+ f2 = f2->GetParent();
+ --n2;
+ }
+
+ return nullptr;
+}
+
+bool nsLayoutUtils::AuthorSpecifiedBorderBackgroundDisablesTheming(
+ StyleAppearance aAppearance) {
+ return aAppearance == StyleAppearance::NumberInput ||
+ aAppearance == StyleAppearance::Button ||
+ aAppearance == StyleAppearance::Textfield ||
+ aAppearance == StyleAppearance::Textarea ||
+ aAppearance == StyleAppearance::Listbox ||
+ aAppearance == StyleAppearance::Menulist ||
+ aAppearance == StyleAppearance::MenulistButton;
+}
+
+static SVGTextFrame* GetContainingSVGTextFrame(const nsIFrame* aFrame) {
+ if (!aFrame->IsInSVGTextSubtree()) {
+ return nullptr;
+ }
+
+ return static_cast<SVGTextFrame*>(nsLayoutUtils::GetClosestFrameOfType(
+ aFrame->GetParent(), LayoutFrameType::SVGText));
+}
+
+static bool TransformGfxPointFromAncestor(RelativeTo aFrame,
+ const Point& aPoint,
+ RelativeTo aAncestor,
+ Maybe<Matrix4x4Flagged>& aMatrixCache,
+ Point* aOut) {
+ SVGTextFrame* text = GetContainingSVGTextFrame(aFrame.mFrame);
+
+ if (!aMatrixCache) {
+ auto matrix = nsLayoutUtils::GetTransformToAncestor(
+ RelativeTo{text ? text : aFrame.mFrame, aFrame.mViewportType},
+ aAncestor);
+ if (matrix.IsSingular()) {
+ return false;
+ }
+ matrix.Invert();
+ aMatrixCache.emplace(matrix);
+ }
+
+ const Matrix4x4Flagged& ctm = *aMatrixCache;
+ Point4D point = ctm.ProjectPoint(aPoint);
+ if (!point.HasPositiveWCoord()) {
+ return false;
+ }
+
+ *aOut = point.As2DPoint();
+
+ if (text) {
+ *aOut = text->TransformFramePointToTextChild(*aOut, aFrame.mFrame);
+ }
+
+ return true;
+}
+
+static Point TransformGfxPointToAncestor(
+ RelativeTo aFrame, const Point& aPoint, RelativeTo aAncestor,
+ Maybe<Matrix4x4Flagged>& aMatrixCache) {
+ if (SVGTextFrame* text = GetContainingSVGTextFrame(aFrame.mFrame)) {
+ Point result =
+ text->TransformFramePointFromTextChild(aPoint, aFrame.mFrame);
+ return TransformGfxPointToAncestor(RelativeTo{text}, result, aAncestor,
+ aMatrixCache);
+ }
+ if (!aMatrixCache) {
+ aMatrixCache.emplace(
+ nsLayoutUtils::GetTransformToAncestor(aFrame, aAncestor));
+ }
+ return aMatrixCache->ProjectPoint(aPoint).As2DPoint();
+}
+
+static Rect TransformGfxRectToAncestor(
+ RelativeTo aFrame, const Rect& aRect, RelativeTo aAncestor,
+ bool* aPreservesAxisAlignedRectangles = nullptr,
+ Maybe<Matrix4x4Flagged>* aMatrixCache = nullptr,
+ bool aStopAtStackingContextAndDisplayPortAndOOFFrame = false,
+ nsIFrame** aOutAncestor = nullptr) {
+ Rect result;
+ Matrix4x4Flagged ctm;
+ if (SVGTextFrame* text = GetContainingSVGTextFrame(aFrame.mFrame)) {
+ result = text->TransformFrameRectFromTextChild(aRect, aFrame.mFrame);
+
+ result = TransformGfxRectToAncestor(
+ RelativeTo{text}, result, aAncestor, nullptr, aMatrixCache,
+ aStopAtStackingContextAndDisplayPortAndOOFFrame, aOutAncestor);
+ if (aPreservesAxisAlignedRectangles) {
+ // TransformFrameRectFromTextChild could involve any kind of transform, we
+ // could drill down into it to get an answer out of it but we don't yet.
+ *aPreservesAxisAlignedRectangles = false;
+ }
+ return result;
+ }
+ if (aMatrixCache && *aMatrixCache) {
+ // We are given a matrix to use, so use it
+ ctm = aMatrixCache->value();
+ } else {
+ // Else, compute it
+ uint32_t flags = 0;
+ if (aStopAtStackingContextAndDisplayPortAndOOFFrame) {
+ flags |= nsIFrame::STOP_AT_STACKING_CONTEXT_AND_DISPLAY_PORT;
+ }
+ ctm = nsLayoutUtils::GetTransformToAncestor(aFrame, aAncestor, flags,
+ aOutAncestor);
+ if (aMatrixCache) {
+ // and put it in the cache, if provided
+ *aMatrixCache = Some(ctm);
+ }
+ }
+ // Fill out the axis-alignment flag
+ if (aPreservesAxisAlignedRectangles) {
+ // TransformFrameRectFromTextChild could involve any kind of transform, we
+ // could drill down into it to get an answer out of it but we don't yet.
+ Matrix matrix2d;
+ *aPreservesAxisAlignedRectangles =
+ ctm.Is2D(&matrix2d) && matrix2d.PreservesAxisAlignedRectangles();
+ }
+ const nsIFrame* ancestor = aOutAncestor ? *aOutAncestor : aAncestor.mFrame;
+ float factor = ancestor->PresContext()->AppUnitsPerDevPixel();
+ Rect maxBounds =
+ Rect(float(nscoord_MIN) / factor * 0.5, float(nscoord_MIN) / factor * 0.5,
+ float(nscoord_MAX) / factor, float(nscoord_MAX) / factor);
+ return ctm.TransformAndClipBounds(aRect, maxBounds);
+}
+
+nsLayoutUtils::TransformResult nsLayoutUtils::TransformPoints(
+ RelativeTo aFromFrame, RelativeTo aToFrame, uint32_t aPointCount,
+ CSSPoint* aPoints) {
+ // Conceptually, {ViewportFrame, Visual} is an ancestor of
+ // {ViewportFrame, Layout}, so factor that into the nearest ancestor
+ // computation.
+ RelativeTo nearestCommonAncestor{
+ FindNearestCommonAncestorFrame(aFromFrame.mFrame, aToFrame.mFrame),
+ aFromFrame.mViewportType == ViewportType::Visual ||
+ aToFrame.mViewportType == ViewportType::Visual
+ ? ViewportType::Visual
+ : ViewportType::Layout};
+ if (!nearestCommonAncestor.mFrame) {
+ return NO_COMMON_ANCESTOR;
+ }
+ CSSToLayoutDeviceScale devPixelsPerCSSPixelFromFrame =
+ aFromFrame.mFrame->PresContext()->CSSToDevPixelScale();
+ CSSToLayoutDeviceScale devPixelsPerCSSPixelToFrame =
+ aToFrame.mFrame->PresContext()->CSSToDevPixelScale();
+ Maybe<Matrix4x4Flagged> cacheTo;
+ Maybe<Matrix4x4Flagged> cacheFrom;
+ for (uint32_t i = 0; i < aPointCount; ++i) {
+ LayoutDevicePoint devPixels = aPoints[i] * devPixelsPerCSSPixelFromFrame;
+ // What should the behaviour be if some of the points aren't invertible
+ // and others are? Just assume all points are for now.
+ Point toDevPixels =
+ TransformGfxPointToAncestor(aFromFrame, Point(devPixels.x, devPixels.y),
+ nearestCommonAncestor, cacheTo);
+ Point result;
+ if (!TransformGfxPointFromAncestor(
+ aToFrame, toDevPixels, nearestCommonAncestor, cacheFrom, &result)) {
+ return NONINVERTIBLE_TRANSFORM;
+ }
+ // Divide here so that when the devPixelsPerCSSPixels are the same, we get
+ // the correct answer instead of some inaccuracy multiplying a number by its
+ // reciprocal.
+ aPoints[i] =
+ LayoutDevicePoint(result.x, result.y) / devPixelsPerCSSPixelToFrame;
+ }
+ return TRANSFORM_SUCCEEDED;
+}
+
+nsLayoutUtils::TransformResult nsLayoutUtils::TransformPoint(
+ RelativeTo aFromFrame, RelativeTo aToFrame, nsPoint& aPoint) {
+ CSSPoint point = CSSPoint::FromAppUnits(aPoint);
+ auto result = TransformPoints(aFromFrame, aToFrame, 1, &point);
+ if (result == TRANSFORM_SUCCEEDED) {
+ aPoint = CSSPoint::ToAppUnits(point);
+ }
+ return result;
+}
+
+nsLayoutUtils::TransformResult nsLayoutUtils::TransformRect(
+ const nsIFrame* aFromFrame, const nsIFrame* aToFrame, nsRect& aRect) {
+ const nsIFrame* nearestCommonAncestor =
+ FindNearestCommonAncestorFrame(aFromFrame, aToFrame);
+ if (!nearestCommonAncestor) {
+ return NO_COMMON_ANCESTOR;
+ }
+ Matrix4x4Flagged downToDest = GetTransformToAncestor(
+ RelativeTo{aToFrame}, RelativeTo{nearestCommonAncestor});
+ if (downToDest.IsSingular()) {
+ return NONINVERTIBLE_TRANSFORM;
+ }
+ downToDest.Invert();
+ aRect = TransformFrameRectToAncestor(aFromFrame, aRect,
+ RelativeTo{nearestCommonAncestor});
+
+ float devPixelsPerAppUnitFromFrame =
+ 1.0f / nearestCommonAncestor->PresContext()->AppUnitsPerDevPixel();
+ float devPixelsPerAppUnitToFrame =
+ 1.0f / aToFrame->PresContext()->AppUnitsPerDevPixel();
+ gfx::Rect toDevPixels = downToDest.ProjectRectBounds(
+ gfx::Rect(aRect.x * devPixelsPerAppUnitFromFrame,
+ aRect.y * devPixelsPerAppUnitFromFrame,
+ aRect.width * devPixelsPerAppUnitFromFrame,
+ aRect.height * devPixelsPerAppUnitFromFrame),
+ Rect(-std::numeric_limits<Float>::max() * devPixelsPerAppUnitFromFrame *
+ 0.5f,
+ -std::numeric_limits<Float>::max() * devPixelsPerAppUnitFromFrame *
+ 0.5f,
+ std::numeric_limits<Float>::max() * devPixelsPerAppUnitFromFrame,
+ std::numeric_limits<Float>::max() * devPixelsPerAppUnitFromFrame));
+ aRect.x = NSToCoordRoundWithClamp(toDevPixels.x / devPixelsPerAppUnitToFrame);
+ aRect.y = NSToCoordRoundWithClamp(toDevPixels.y / devPixelsPerAppUnitToFrame);
+ aRect.width =
+ NSToCoordRoundWithClamp(toDevPixels.width / devPixelsPerAppUnitToFrame);
+ aRect.height =
+ NSToCoordRoundWithClamp(toDevPixels.height / devPixelsPerAppUnitToFrame);
+ return TRANSFORM_SUCCEEDED;
+}
+
+nsRect nsLayoutUtils::GetRectRelativeToFrame(Element* aElement,
+ nsIFrame* aFrame) {
+ if (!aElement || !aFrame) {
+ return nsRect();
+ }
+
+ nsIFrame* frame = aElement->GetPrimaryFrame();
+ if (!frame) {
+ return nsRect();
+ }
+
+ nsRect rect = frame->GetRectRelativeToSelf();
+ nsLayoutUtils::TransformResult rv =
+ nsLayoutUtils::TransformRect(frame, aFrame, rect);
+ if (rv != nsLayoutUtils::TRANSFORM_SUCCEEDED) {
+ return nsRect();
+ }
+
+ return rect;
+}
+
+bool nsLayoutUtils::ContainsPoint(const nsRect& aRect, const nsPoint& aPoint,
+ nscoord aInflateSize) {
+ nsRect rect = aRect;
+ rect.Inflate(aInflateSize);
+ return rect.Contains(aPoint);
+}
+
+nsRect nsLayoutUtils::ClampRectToScrollFrames(nsIFrame* aFrame,
+ const nsRect& aRect) {
+ nsIFrame* closestScrollFrame =
+ nsLayoutUtils::GetClosestFrameOfType(aFrame, LayoutFrameType::Scroll);
+
+ nsRect resultRect = aRect;
+
+ while (closestScrollFrame) {
+ nsIScrollableFrame* sf = do_QueryFrame(closestScrollFrame);
+
+ nsRect scrollPortRect = sf->GetScrollPortRect();
+ nsLayoutUtils::TransformRect(closestScrollFrame, aFrame, scrollPortRect);
+
+ resultRect = resultRect.Intersect(scrollPortRect);
+
+ // Check whether aRect is visible in the scroll frame or not.
+ if (resultRect.IsEmpty()) {
+ break;
+ }
+
+ // Get next ancestor scroll frame.
+ closestScrollFrame = nsLayoutUtils::GetClosestFrameOfType(
+ closestScrollFrame->GetParent(), LayoutFrameType::Scroll);
+ }
+
+ return resultRect;
+}
+
+nsPoint nsLayoutUtils::TransformAncestorPointToFrame(RelativeTo aFrame,
+ const nsPoint& aPoint,
+ RelativeTo aAncestor) {
+ float factor = aFrame.mFrame->PresContext()->AppUnitsPerDevPixel();
+ Point result(NSAppUnitsToFloatPixels(aPoint.x, factor),
+ NSAppUnitsToFloatPixels(aPoint.y, factor));
+
+ Maybe<Matrix4x4Flagged> matrixCache;
+ if (!TransformGfxPointFromAncestor(aFrame, result, aAncestor, matrixCache,
+ &result)) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ return nsPoint(NSFloatPixelsToAppUnits(float(result.x), factor),
+ NSFloatPixelsToAppUnits(float(result.y), factor));
+}
+
+nsRect nsLayoutUtils::TransformFrameRectToAncestor(
+ const nsIFrame* aFrame, const nsRect& aRect, RelativeTo aAncestor,
+ bool* aPreservesAxisAlignedRectangles /* = nullptr */,
+ Maybe<Matrix4x4Flagged>* aMatrixCache /* = nullptr */,
+ bool aStopAtStackingContextAndDisplayPortAndOOFFrame /* = false */,
+ nsIFrame** aOutAncestor /* = nullptr */) {
+ MOZ_ASSERT(IsAncestorFrameCrossDocInProcess(aAncestor.mFrame, aFrame),
+ "Fix the caller");
+ float srcAppUnitsPerDevPixel = aFrame->PresContext()->AppUnitsPerDevPixel();
+ Rect result(NSAppUnitsToFloatPixels(aRect.x, srcAppUnitsPerDevPixel),
+ NSAppUnitsToFloatPixels(aRect.y, srcAppUnitsPerDevPixel),
+ NSAppUnitsToFloatPixels(aRect.width, srcAppUnitsPerDevPixel),
+ NSAppUnitsToFloatPixels(aRect.height, srcAppUnitsPerDevPixel));
+ result = TransformGfxRectToAncestor(
+ RelativeTo{aFrame}, result, aAncestor, aPreservesAxisAlignedRectangles,
+ aMatrixCache, aStopAtStackingContextAndDisplayPortAndOOFFrame,
+ aOutAncestor);
+
+ float destAppUnitsPerDevPixel =
+ aAncestor.mFrame->PresContext()->AppUnitsPerDevPixel();
+ return nsRect(
+ NSFloatPixelsToAppUnits(float(result.x), destAppUnitsPerDevPixel),
+ NSFloatPixelsToAppUnits(float(result.y), destAppUnitsPerDevPixel),
+ NSFloatPixelsToAppUnits(float(result.width), destAppUnitsPerDevPixel),
+ NSFloatPixelsToAppUnits(float(result.height), destAppUnitsPerDevPixel));
+}
+
+static LayoutDeviceIntPoint GetWidgetOffset(nsIWidget* aWidget,
+ nsIWidget*& aRootWidget) {
+ LayoutDeviceIntPoint offset(0, 0);
+ while (aWidget->GetWindowType() == widget::WindowType::Child) {
+ nsIWidget* parent = aWidget->GetParent();
+ if (!parent) {
+ break;
+ }
+ LayoutDeviceIntRect bounds = aWidget->GetBounds();
+ offset += bounds.TopLeft();
+ aWidget = parent;
+ }
+ aRootWidget = aWidget;
+ return offset;
+}
+
+LayoutDeviceIntPoint nsLayoutUtils::WidgetToWidgetOffset(nsIWidget* aFrom,
+ nsIWidget* aTo) {
+ nsIWidget* fromRoot;
+ LayoutDeviceIntPoint fromOffset = GetWidgetOffset(aFrom, fromRoot);
+ nsIWidget* toRoot;
+ LayoutDeviceIntPoint toOffset = GetWidgetOffset(aTo, toRoot);
+
+ if (fromRoot != toRoot) {
+ fromOffset = aFrom->WidgetToScreenOffset();
+ toOffset = aTo->WidgetToScreenOffset();
+ }
+ return fromOffset - toOffset;
+}
+
+nsPoint nsLayoutUtils::TranslateWidgetToView(nsPresContext* aPresContext,
+ nsIWidget* aWidget,
+ const LayoutDeviceIntPoint& aPt,
+ nsView* aView) {
+ nsPoint viewOffset;
+ nsIWidget* viewWidget = aView->GetNearestWidget(&viewOffset);
+ if (!viewWidget) {
+ return nsPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ LayoutDeviceIntPoint widgetPoint =
+ aPt + WidgetToWidgetOffset(aWidget, viewWidget);
+ nsPoint widgetAppUnits(aPresContext->DevPixelsToAppUnits(widgetPoint.x),
+ aPresContext->DevPixelsToAppUnits(widgetPoint.y));
+ return widgetAppUnits - viewOffset;
+}
+
+LayoutDeviceIntPoint nsLayoutUtils::TranslateViewToWidget(
+ nsPresContext* aPresContext, nsView* aView, nsPoint aPt,
+ ViewportType aViewportType, nsIWidget* aWidget) {
+ nsPoint viewOffset;
+ nsIWidget* viewWidget = aView->GetNearestWidget(&viewOffset);
+ if (!viewWidget) {
+ return LayoutDeviceIntPoint(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
+ }
+
+ nsPoint pt = (aPt + viewOffset);
+ // The target coordinates are visual, so perform a layout-to-visual
+ // conversion if the incoming coordinates are layout.
+ if (aViewportType == ViewportType::Layout && aPresContext->GetPresShell()) {
+ pt = ViewportUtils::LayoutToVisual(pt, aPresContext->GetPresShell());
+ }
+ LayoutDeviceIntPoint relativeToViewWidget(
+ aPresContext->AppUnitsToDevPixels(pt.x),
+ aPresContext->AppUnitsToDevPixels(pt.y));
+ return relativeToViewWidget + WidgetToWidgetOffset(viewWidget, aWidget);
+}
+
+StyleClear nsLayoutUtils::CombineClearType(StyleClear aOrigClearType,
+ StyleClear aNewClearType) {
+ StyleClear clearType = aOrigClearType;
+ switch (clearType) {
+ case StyleClear::Left:
+ if (StyleClear::Right == aNewClearType ||
+ StyleClear::Both == aNewClearType) {
+ clearType = StyleClear::Both;
+ }
+ break;
+ case StyleClear::Right:
+ if (StyleClear::Left == aNewClearType ||
+ StyleClear::Both == aNewClearType) {
+ clearType = StyleClear::Both;
+ }
+ break;
+ case StyleClear::None:
+ if (StyleClear::Left == aNewClearType ||
+ StyleClear::Right == aNewClearType ||
+ StyleClear::Both == aNewClearType) {
+ clearType = aNewClearType;
+ }
+ break;
+ case StyleClear::Both:
+ // Do nothing.
+ break;
+ }
+ return clearType;
+}
+
+#ifdef MOZ_DUMP_PAINTING
+# include <stdio.h>
+
+static bool gDumpEventList = false;
+
+// nsLayoutUtils::PaintFrame() can call itself recursively, so rather than
+// maintaining a single paint count, we need a stack.
+StaticAutoPtr<nsTArray<int>> gPaintCountStack;
+
+struct AutoNestedPaintCount {
+ AutoNestedPaintCount() { gPaintCountStack->AppendElement(0); }
+ ~AutoNestedPaintCount() { gPaintCountStack->RemoveLastElement(); }
+};
+
+#endif
+
+nsIFrame* nsLayoutUtils::GetFrameForPoint(
+ RelativeTo aRelativeTo, nsPoint aPt, const FrameForPointOptions& aOptions) {
+ AUTO_PROFILER_LABEL("nsLayoutUtils::GetFrameForPoint", LAYOUT);
+
+ nsresult rv;
+ AutoTArray<nsIFrame*, 8> outFrames;
+ rv = GetFramesForArea(aRelativeTo, nsRect(aPt, nsSize(1, 1)), outFrames,
+ aOptions);
+ NS_ENSURE_SUCCESS(rv, nullptr);
+ return outFrames.SafeElementAt(0);
+}
+
+nsresult nsLayoutUtils::GetFramesForArea(RelativeTo aRelativeTo,
+ const nsRect& aRect,
+ nsTArray<nsIFrame*>& aOutFrames,
+ const FrameForPointOptions& aOptions) {
+ AUTO_PROFILER_LABEL("nsLayoutUtils::GetFramesForArea", LAYOUT);
+
+ nsIFrame* frame = const_cast<nsIFrame*>(aRelativeTo.mFrame);
+
+ nsDisplayListBuilder builder(frame, nsDisplayListBuilderMode::EventDelivery,
+ false);
+ builder.BeginFrame();
+ nsDisplayList list(&builder);
+
+ if (aOptions.mBits.contains(FrameForPointOption::IgnorePaintSuppression)) {
+ builder.IgnorePaintSuppression();
+ }
+ if (aOptions.mBits.contains(FrameForPointOption::IgnoreRootScrollFrame)) {
+ nsIFrame* rootScrollFrame = frame->PresShell()->GetRootScrollFrame();
+ if (rootScrollFrame) {
+ builder.SetIgnoreScrollFrame(rootScrollFrame);
+ }
+ }
+ if (aRelativeTo.mViewportType == ViewportType::Layout) {
+ builder.SetIsRelativeToLayoutViewport();
+ }
+ if (aOptions.mBits.contains(FrameForPointOption::IgnoreCrossDoc)) {
+ builder.SetDescendIntoSubdocuments(false);
+ }
+
+ if (aOptions.mBits.contains(FrameForPointOption::OnlyVisible)) {
+ builder.SetHitTestIsForVisibility(aOptions.mVisibleThreshold);
+ }
+
+ builder.EnterPresShell(frame);
+
+ builder.SetVisibleRect(aRect);
+ builder.SetDirtyRect(aRect);
+
+ frame->BuildDisplayListForStackingContext(&builder, &list);
+ builder.LeavePresShell(frame, nullptr);
+
+#ifdef MOZ_DUMP_PAINTING
+ if (gDumpEventList) {
+ fprintf_stderr(stderr, "Event handling --- (%d,%d):\n", aRect.x, aRect.y);
+
+ std::stringstream ss;
+ nsIFrame::PrintDisplayList(&builder, list, ss);
+ print_stderr(ss);
+ }
+#endif
+
+ nsDisplayItem::HitTestState hitTestState;
+ list.HitTest(&builder, aRect, &hitTestState, &aOutFrames);
+ list.DeleteAll(&builder);
+ builder.EndFrame();
+ return NS_OK;
+}
+
+mozilla::ParentLayerToScreenScale2D
+nsLayoutUtils::GetTransformToAncestorScaleCrossProcessForFrameMetrics(
+ const nsIFrame* aFrame) {
+ ParentLayerToScreenScale2D transformToAncestorScale =
+ ViewAs<ParentLayerToScreenScale2D>(
+ nsLayoutUtils::GetTransformToAncestorScale(aFrame));
+
+ if (BrowserChild* browserChild = BrowserChild::GetFrom(aFrame->PresShell())) {
+ transformToAncestorScale =
+ ViewTargetAs<ParentLayerPixel>(
+ transformToAncestorScale,
+ PixelCastJustification::PropagatingToChildProcess) *
+ browserChild->GetEffectsInfo().mTransformToAncestorScale;
+ }
+
+ return transformToAncestorScale;
+}
+
+// aScrollFrameAsScrollable must be non-nullptr and queryable to an nsIFrame
+FrameMetrics nsLayoutUtils::CalculateBasicFrameMetrics(
+ nsIScrollableFrame* aScrollFrame) {
+ nsIFrame* frame = do_QueryFrame(aScrollFrame);
+ MOZ_ASSERT(frame);
+
+ // Calculate the metrics necessary for calculating the displayport.
+ // This code has a lot in common with the code in ComputeFrameMetrics();
+ // we may want to refactor this at some point.
+ FrameMetrics metrics;
+ nsPresContext* presContext = frame->PresContext();
+ PresShell* presShell = presContext->PresShell();
+ CSSToLayoutDeviceScale deviceScale = presContext->CSSToDevPixelScale();
+ float resolution = 1.0f;
+ bool isRcdRsf = aScrollFrame->IsRootScrollFrameOfDocument() &&
+ presContext->IsRootContentDocumentCrossProcess();
+ metrics.SetIsRootContent(isRcdRsf);
+ if (isRcdRsf) {
+ // Only the root content document's root scrollable frame should pick up
+ // the presShell's resolution. All the other frames are 1.0.
+ resolution = presShell->GetResolution();
+ }
+ LayoutDeviceToLayerScale cumulativeResolution(
+ LayoutDeviceToLayerScale(presShell->GetCumulativeResolution()));
+
+ LayerToParentLayerScale layerToParentLayerScale(1.0f);
+ metrics.SetDevPixelsPerCSSPixel(deviceScale);
+ metrics.SetPresShellResolution(resolution);
+
+ metrics.SetTransformToAncestorScale(
+ GetTransformToAncestorScaleCrossProcessForFrameMetrics(frame));
+ metrics.SetCumulativeResolution(cumulativeResolution);
+ metrics.SetZoom(deviceScale * cumulativeResolution * layerToParentLayerScale);
+
+ // Only the size of the composition bounds is relevant to the
+ // displayport calculation, not its origin.
+ nsSize compositionSize =
+ nsLayoutUtils::CalculateCompositionSizeForFrame(frame);
+ LayoutDeviceToParentLayerScale compBoundsScale;
+ if (frame == presShell->GetRootScrollFrame() &&
+ presContext->IsRootContentDocumentCrossProcess()) {
+ if (presContext->GetParentPresContext()) {
+ float res = presContext->GetParentPresContext()
+ ->PresShell()
+ ->GetCumulativeResolution();
+ compBoundsScale = LayoutDeviceToParentLayerScale(res);
+ }
+ } else {
+ compBoundsScale = cumulativeResolution * layerToParentLayerScale;
+ }
+ metrics.SetCompositionBounds(
+ LayoutDeviceRect::FromAppUnits(nsRect(nsPoint(0, 0), compositionSize),
+ presContext->AppUnitsPerDevPixel()) *
+ compBoundsScale);
+
+ metrics.SetBoundingCompositionSize(
+ nsLayoutUtils::CalculateBoundingCompositionSize(frame, false, metrics));
+
+ metrics.SetLayoutViewport(
+ CSSRect::FromAppUnits(nsRect(aScrollFrame->GetScrollPosition(),
+ aScrollFrame->GetScrollPortRect().Size())));
+ metrics.SetVisualScrollOffset(
+ isRcdRsf ? CSSPoint::FromAppUnits(presShell->GetVisualViewportOffset())
+ : metrics.GetLayoutViewport().TopLeft());
+
+ metrics.SetScrollableRect(CSSRect::FromAppUnits(
+ nsLayoutUtils::CalculateScrollableRectForFrame(aScrollFrame, nullptr)));
+
+ return metrics;
+}
+
+nsIScrollableFrame* nsLayoutUtils::GetAsyncScrollableAncestorFrame(
+ nsIFrame* aTarget) {
+ uint32_t flags = nsLayoutUtils::SCROLLABLE_ALWAYS_MATCH_ROOT |
+ nsLayoutUtils::SCROLLABLE_ONLY_ASYNC_SCROLLABLE |
+ nsLayoutUtils::SCROLLABLE_FIXEDPOS_FINDS_ROOT;
+ return nsLayoutUtils::GetNearestScrollableFrame(aTarget, flags);
+}
+
+void nsLayoutUtils::AddExtraBackgroundItems(nsDisplayListBuilder* aBuilder,
+ nsDisplayList* aList,
+ nsIFrame* aFrame,
+ const nsRect& aCanvasArea,
+ const nsRegion& aVisibleRegion,
+ nscolor aBackstop) {
+ if (aFrame->IsPageFrame()) {
+ // For printing, this function is first called on an nsPageFrame, which
+ // creates a display list with a PageContent item. The PageContent item's
+ // paint function calls this function on the nsPageFrame's child which is an
+ // nsPageContentFrame. We only want to add the canvas background color item
+ // once, for the nsPageContentFrame.
+ return;
+ }
+ // Add the canvas background color to the bottom of the list. This
+ // happens after we've built the list so that AddCanvasBackgroundColorItem
+ // can monkey with the contents if necessary.
+ nsRect canvasArea = aVisibleRegion.GetBounds();
+ canvasArea.IntersectRect(aCanvasArea, canvasArea);
+ nsDisplayListBuilder::AutoBuildingDisplayList buildingDisplayList(
+ aBuilder, aFrame, canvasArea, canvasArea);
+ aFrame->PresShell()->AddCanvasBackgroundColorItem(aBuilder, aList, aFrame,
+ canvasArea, aBackstop);
+}
+
+// #define PRINT_HITTESTINFO_STATS
+#ifdef PRINT_HITTESTINFO_STATS
+void PrintHitTestInfoStatsInternal(nsDisplayList* aList, int& aTotal,
+ int& aHitTest, int& aVisible,
+ int& aSpecial) {
+ for (nsDisplayItem* i : *aList) {
+ aTotal++;
+
+ if (i->GetChildren()) {
+ PrintHitTestInfoStatsInternal(i->GetChildren(), aTotal, aHitTest,
+ aVisible, aSpecial);
+ }
+
+ if (i->GetType() == DisplayItemType::TYPE_COMPOSITOR_HITTEST_INFO) {
+ aHitTest++;
+
+ const auto& hitTestInfo = static_cast<nsDisplayCompositorHitTestInfo*>(i)
+ ->GetHitTestInfo()
+ .Info();
+
+ if (hitTestInfo.size() > 1) {
+ aSpecial++;
+ continue;
+ }
+
+ if (hitTestInfo == CompositorHitTestFlags::eVisibleToHitTest) {
+ aVisible++;
+ continue;
+ }
+
+ aSpecial++;
+ }
+ }
+}
+
+void PrintHitTestInfoStats(nsDisplayList* aList) {
+ int total = 0;
+ int hitTest = 0;
+ int visible = 0;
+ int special = 0;
+
+ PrintHitTestInfoStatsInternal(aList, total, hitTest, visible, special);
+
+ double ratio = (double)hitTest / (double)total;
+
+ printf(
+ "List %p: total items: %d, hit test items: %d, ratio: %f, visible: %d, "
+ "special: %d\n",
+ aList, total, hitTest, ratio, visible, special);
+}
+#endif
+
+// Apply a batch of effects updates generated during a paint to their
+// respective remote browsers.
+static void ApplyEffectsUpdates(
+ const nsTHashMap<nsPtrHashKey<RemoteBrowser>, EffectsInfo>& aUpdates) {
+ for (const auto& entry : aUpdates) {
+ auto* browser = entry.GetKey();
+ const auto& update = entry.GetData();
+ browser->UpdateEffects(update);
+ }
+}
+
+static void DumpBeforePaintDisplayList(UniquePtr<std::stringstream>& aStream,
+ nsDisplayListBuilder* aBuilder,
+ nsDisplayList* aList,
+ const nsRect& aVisibleRect) {
+#ifdef MOZ_DUMP_PAINTING
+ if (gfxEnv::MOZ_DUMP_PAINT_TO_FILE()) {
+ nsCString string("dump-");
+ // Include the process ID in the dump file name, to make sure that in an
+ // e10s setup different processes don't clobber each other's dump files.
+ string.AppendInt(getpid());
+ for (int paintCount : *gPaintCountStack) {
+ string.AppendLiteral("-");
+ string.AppendInt(paintCount);
+ }
+ string.AppendLiteral(".html");
+ gfxUtils::sDumpPaintFile = fopen(string.BeginReading(), "w");
+ } else {
+ gfxUtils::sDumpPaintFile = stderr;
+ }
+ if (gfxEnv::MOZ_DUMP_PAINT_TO_FILE()) {
+ *aStream << "<html><head><script>\n"
+ "var array = {};\n"
+ "function ViewImage(index) { \n"
+ " var image = document.getElementById(index);\n"
+ " if (image.src) {\n"
+ " image.removeAttribute('src');\n"
+ " } else {\n"
+ " image.src = array[index];\n"
+ " }\n"
+ "}</script></head><body>";
+ }
+#endif
+ *aStream << nsPrintfCString(
+ "Painting --- before optimization (dirty %d,%d,%d,%d):\n",
+ aVisibleRect.x, aVisibleRect.y, aVisibleRect.width,
+ aVisibleRect.height)
+ .get();
+ nsIFrame::PrintDisplayList(aBuilder, *aList, *aStream,
+ gfxEnv::MOZ_DUMP_PAINT_TO_FILE());
+
+ if (gfxEnv::MOZ_DUMP_PAINT() || gfxEnv::MOZ_DUMP_PAINT_ITEMS()) {
+ // Flush stream now to avoid reordering dump output relative to
+ // messages dumped by PaintRoot below.
+ fprint_stderr(gfxUtils::sDumpPaintFile, *aStream);
+ aStream = MakeUnique<std::stringstream>();
+ }
+}
+
+static void DumpAfterPaintDisplayList(UniquePtr<std::stringstream>& aStream,
+ nsDisplayListBuilder* aBuilder,
+ nsDisplayList* aList) {
+ *aStream << "Painting --- after optimization:\n";
+ nsIFrame::PrintDisplayList(aBuilder, *aList, *aStream,
+ gfxEnv::MOZ_DUMP_PAINT_TO_FILE());
+
+ fprint_stderr(gfxUtils::sDumpPaintFile, *aStream);
+
+#ifdef MOZ_DUMP_PAINTING
+ if (gfxEnv::MOZ_DUMP_PAINT_TO_FILE()) {
+ *aStream << "</body></html>";
+ }
+ if (gfxEnv::MOZ_DUMP_PAINT_TO_FILE()) {
+ fclose(gfxUtils::sDumpPaintFile);
+ }
+#endif
+
+ std::stringstream lsStream;
+ nsIFrame::PrintDisplayList(aBuilder, *aList, lsStream);
+}
+
+struct TemporaryDisplayListBuilder {
+ TemporaryDisplayListBuilder(nsIFrame* aFrame,
+ nsDisplayListBuilderMode aBuilderMode,
+ const bool aBuildCaret)
+ : mBuilder(aFrame, aBuilderMode, aBuildCaret), mList(&mBuilder) {}
+
+ ~TemporaryDisplayListBuilder() { mList.DeleteAll(&mBuilder); }
+
+ nsDisplayListBuilder mBuilder;
+ nsDisplayList mList;
+ RetainedDisplayListMetrics mMetrics;
+};
+
+void nsLayoutUtils::PaintFrame(gfxContext* aRenderingContext, nsIFrame* aFrame,
+ const nsRegion& aDirtyRegion, nscolor aBackstop,
+ nsDisplayListBuilderMode aBuilderMode,
+ PaintFrameFlags aFlags) {
+ AUTO_PROFILER_LABEL("nsLayoutUtils::PaintFrame", GRAPHICS);
+
+ // Create a static storage counter that is incremented on eacy entry to
+ // PaintFrame and decremented on exit. We can use this later to determine if
+ // this is a top-level paint.
+ static uint32_t paintFrameDepth = 0;
+ ++paintFrameDepth;
+
+#ifdef MOZ_DUMP_PAINTING
+ if (!gPaintCountStack) {
+ gPaintCountStack = new nsTArray<int>();
+ ClearOnShutdown(&gPaintCountStack);
+
+ gPaintCountStack->AppendElement(0);
+ }
+ ++gPaintCountStack->LastElement();
+ AutoNestedPaintCount nestedPaintCount;
+#endif
+
+ nsIFrame* displayRoot = GetDisplayRootFrame(aFrame);
+
+ if (aFlags & PaintFrameFlags::WidgetLayers) {
+ nsView* view = aFrame->GetView();
+ if (!(view && view->GetWidget() && displayRoot == aFrame)) {
+ aFlags &= ~PaintFrameFlags::WidgetLayers;
+ NS_ASSERTION(aRenderingContext, "need a rendering context");
+ }
+ }
+
+ nsPresContext* presContext = aFrame->PresContext();
+ PresShell* presShell = presContext->PresShell();
+
+ TimeStamp startBuildDisplayList = TimeStamp::Now();
+ auto dlTimerId = mozilla::glean::paint::build_displaylist_time.Start();
+
+ const bool buildCaret = !(aFlags & PaintFrameFlags::HideCaret);
+
+ // Note that isForPainting here does not include the PaintForPrinting builder
+ // mode; that's OK because there is no point in using retained display lists
+ // for a print destination.
+ const bool isForPainting = (aFlags & PaintFrameFlags::WidgetLayers) &&
+ aBuilderMode == nsDisplayListBuilderMode::Painting;
+
+ // Only allow retaining for painting when preffed on, and for root frames
+ // (since the modified frame tracking is per-root-frame).
+ const bool retainDisplayList =
+ isForPainting && AreRetainedDisplayListsEnabled() && !aFrame->GetParent();
+
+ RetainedDisplayListBuilder* retainedBuilder = nullptr;
+ Maybe<TemporaryDisplayListBuilder> temporaryBuilder;
+
+ nsDisplayListBuilder* builder = nullptr;
+ nsDisplayList* list = nullptr;
+ RetainedDisplayListMetrics* metrics = nullptr;
+
+ if (retainDisplayList) {
+ MOZ_ASSERT(aFrame == displayRoot);
+ retainedBuilder = aFrame->GetProperty(RetainedDisplayListBuilder::Cached());
+ if (!retainedBuilder) {
+ retainedBuilder =
+ new RetainedDisplayListBuilder(aFrame, aBuilderMode, buildCaret);
+ aFrame->SetProperty(RetainedDisplayListBuilder::Cached(),
+ retainedBuilder);
+ }
+
+ builder = retainedBuilder->Builder();
+ list = retainedBuilder->List();
+ metrics = retainedBuilder->Metrics();
+ } else {
+ temporaryBuilder.emplace(aFrame, aBuilderMode, buildCaret);
+ builder = &temporaryBuilder->mBuilder;
+ list = &temporaryBuilder->mList;
+ metrics = &temporaryBuilder->mMetrics;
+ }
+
+ MOZ_ASSERT(builder && list && metrics);
+
+ nsAutoString uri;
+ if (MOZ_LOG_TEST(GetLoggerByProcess(), LogLevel::Info) ||
+ MOZ_UNLIKELY(gfxUtils::DumpDisplayList()) ||
+ MOZ_UNLIKELY(gfxEnv::MOZ_DUMP_PAINT())) {
+ if (Document* doc = presContext->Document()) {
+ Unused << doc->GetDocumentURI(uri);
+ }
+ }
+
+ nsAutoString frameName, displayRootName;
+#ifdef DEBUG_FRAME_DUMP
+ if (MOZ_LOG_TEST(GetLoggerByProcess(), LogLevel::Info)) {
+ aFrame->GetFrameName(frameName);
+ displayRoot->GetFrameName(displayRootName);
+ }
+#endif
+
+ DL_LOGI("PaintFrame: %p (%s), DisplayRoot: %p (%s), Builder: %p, URI: %s",
+ aFrame, NS_ConvertUTF16toUTF8(frameName).get(), displayRoot,
+ NS_ConvertUTF16toUTF8(displayRootName).get(), retainedBuilder,
+ NS_ConvertUTF16toUTF8(uri).get());
+
+ metrics->Reset();
+ metrics->StartBuild();
+
+ builder->BeginFrame();
+
+ MOZ_ASSERT(paintFrameDepth >= 1);
+ // If this is a top-level paint, increment the paint sequence number.
+ if (paintFrameDepth == 1) {
+ // Increment the paint sequence number for the display list builder.
+ nsDisplayListBuilder::IncrementPaintSequenceNumber();
+ }
+
+ if (aFlags & PaintFrameFlags::InTransform) {
+ builder->SetInTransform(true);
+ }
+ if (aFlags & PaintFrameFlags::SyncDecodeImages) {
+ builder->SetSyncDecodeImages(true);
+ }
+ if (aFlags & (PaintFrameFlags::WidgetLayers | PaintFrameFlags::ToWindow)) {
+ builder->SetPaintingToWindow(true);
+ }
+ if (aFlags & PaintFrameFlags::UseHighQualityScaling) {
+ builder->SetUseHighQualityScaling(true);
+ }
+ if (aFlags & PaintFrameFlags::ForWebRender) {
+ builder->SetPaintingForWebRender(true);
+ }
+ if (aFlags & PaintFrameFlags::IgnoreSuppression) {
+ builder->IgnorePaintSuppression();
+ }
+
+ if (BrowsingContext* bc = presContext->Document()->GetBrowsingContext()) {
+ builder->SetInActiveDocShell(bc->IsActive());
+ }
+
+ nsRect rootInkOverflow = aFrame->InkOverflowRectRelativeToSelf();
+
+ // If the dynamic toolbar is completely collapsed, the visible rect should
+ // be expanded to include this area.
+ const bool hasDynamicToolbar =
+ presContext->IsRootContentDocumentCrossProcess() &&
+ presContext->HasDynamicToolbar();
+ if (hasDynamicToolbar) {
+ rootInkOverflow.SizeTo(nsLayoutUtils::ExpandHeightForDynamicToolbar(
+ presContext, rootInkOverflow.Size()));
+ }
+
+ // If we are in a remote browser, then apply clipping from ancestor browsers
+ if (BrowserChild* browserChild = BrowserChild::GetFrom(presShell)) {
+ if (!browserChild->IsTopLevel()) {
+ Maybe<nsRect> unscaledVisibleRect = browserChild->GetVisibleRect();
+
+ if (!unscaledVisibleRect) {
+ unscaledVisibleRect = Some(nsRect());
+ }
+
+ rootInkOverflow.IntersectRect(rootInkOverflow, *unscaledVisibleRect);
+ }
+ }
+
+ builder->ClearHaveScrollableDisplayPort();
+ if (builder->IsPaintingToWindow()) {
+ DisplayPortUtils::MaybeCreateDisplayPortInFirstScrollFrameEncountered(
+ aFrame, builder);
+ }
+
+ nsIFrame* rootScrollFrame = presShell->GetRootScrollFrame();
+ if (rootScrollFrame && !aFrame->GetParent()) {
+ nsIScrollableFrame* rootScrollableFrame =
+ presShell->GetRootScrollFrameAsScrollable();
+ MOZ_ASSERT(rootScrollableFrame);
+ nsRect displayPortBase = rootInkOverflow;
+ nsRect temp = displayPortBase;
+ Unused << rootScrollableFrame->DecideScrollableLayer(
+ builder, &displayPortBase, &temp,
+ /* aSetBase = */ true);
+ }
+
+ nsRegion visibleRegion;
+ if (aFlags & PaintFrameFlags::WidgetLayers) {
+ // This layer tree will be reused, so we'll need to calculate it
+ // for the whole "visible" area of the window
+ //
+ // |ignoreViewportScrolling| and |usingDisplayPort| are persistent
+ // document-rendering state. We rely on PresShell to flush
+ // retained layers as needed when that persistent state changes.
+ visibleRegion = rootInkOverflow;
+ } else {
+ visibleRegion = aDirtyRegion;
+ }
+
+ Maybe<nsPoint> originalScrollPosition;
+ auto maybeResetScrollPosition = MakeScopeExit([&]() {
+ if (originalScrollPosition && rootScrollFrame) {
+ nsIScrollableFrame* rootScrollableFrame =
+ presShell->GetRootScrollFrameAsScrollable();
+ MOZ_ASSERT(rootScrollableFrame->GetScrolledFrame()->GetPosition() ==
+ nsPoint());
+ rootScrollableFrame->GetScrolledFrame()->SetPosition(
+ *originalScrollPosition);
+ }
+ });
+
+ nsRect canvasArea(nsPoint(0, 0),
+ aFrame->InkOverflowRectRelativeToSelf().Size());
+ bool ignoreViewportScrolling =
+ !aFrame->GetParent() && presShell->IgnoringViewportScrolling();
+
+ if (!aFrame->GetParent() && hasDynamicToolbar) {
+ canvasArea.SizeTo(nsLayoutUtils::ExpandHeightForDynamicToolbar(
+ presContext, canvasArea.Size()));
+ }
+
+ if (ignoreViewportScrolling && rootScrollFrame) {
+ nsIScrollableFrame* rootScrollableFrame =
+ presShell->GetRootScrollFrameAsScrollable();
+ if (aFlags & PaintFrameFlags::ResetViewportScrolling) {
+ // Temporarily scroll the root scroll frame to 0,0 so that position:fixed
+ // elements will appear fixed to the top-left of the document. We manually
+ // set the position of the scrolled frame instead of using ScrollTo, since
+ // the latter fires scroll listeners, which we don't want.
+ originalScrollPosition.emplace(
+ rootScrollableFrame->GetScrolledFrame()->GetPosition());
+ rootScrollableFrame->GetScrolledFrame()->SetPosition(nsPoint());
+ }
+ if (aFlags & PaintFrameFlags::DocumentRelative) {
+ // Make visibleRegion and aRenderingContext relative to the
+ // scrolled frame instead of the root frame.
+ nsPoint pos = rootScrollableFrame->GetScrollPosition();
+ visibleRegion.MoveBy(-pos);
+ if (aRenderingContext) {
+ gfxPoint devPixelOffset = nsLayoutUtils::PointToGfxPoint(
+ pos, presContext->AppUnitsPerDevPixel());
+ aRenderingContext->SetMatrixDouble(
+ aRenderingContext->CurrentMatrixDouble().PreTranslate(
+ devPixelOffset));
+ }
+ }
+ builder->SetIgnoreScrollFrame(rootScrollFrame);
+
+ nsCanvasFrame* canvasFrame =
+ do_QueryFrame(rootScrollableFrame->GetScrolledFrame());
+ if (canvasFrame) {
+ // Use UnionRect here to ensure that areas where the scrollbars
+ // were are still filled with the background color.
+ canvasArea.UnionRect(
+ canvasArea,
+ canvasFrame->CanvasArea() + builder->ToReferenceFrame(canvasFrame));
+ }
+ }
+
+ nsRect visibleRect = visibleRegion.GetBounds();
+ PartialUpdateResult updateState = PartialUpdateResult::Failed;
+
+ {
+ AUTO_PROFILER_LABEL_CATEGORY_PAIR(GRAPHICS_DisplayListBuilding);
+ AUTO_PROFILER_TRACING_MARKER("Paint", "DisplayList", GRAPHICS);
+ PerfStats::AutoMetricRecording<PerfStats::Metric::DisplayListBuilding>
+ autoRecording;
+
+ ViewID id = ScrollableLayerGuid::NULL_SCROLL_ID;
+ nsDisplayListBuilder::AutoCurrentActiveScrolledRootSetter asrSetter(
+ builder);
+
+ if (presShell->GetDocument() &&
+ presShell->GetDocument()->IsRootDisplayDocument() &&
+ !presShell->GetRootScrollFrame()) {
+ // In cases where the root document is a XUL document, we want to take
+ // the ViewID from the root element, as that will be the ViewID of the
+ // root APZC in the tree. Skip doing this in cases where we know
+ // nsGfxScrollFrame::BuilDisplayList will do it instead.
+ if (dom::Element* element =
+ presShell->GetDocument()->GetDocumentElement()) {
+ id = nsLayoutUtils::FindOrCreateIDFor(element);
+ }
+ // In some cases we get a root document here on an APZ-enabled window
+ // that doesn't have the root displayport initialized yet, even though
+ // the ChromeProcessController is supposed to do it when the widget is
+ // created. This can happen simply because the ChromeProcessController
+ // does it on the next spin of the event loop, and we can trigger a
+ // paint synchronously after window creation but before that runs. In
+ // that case we should initialize the root displayport here before we do
+ // the paint.
+ } else if (XRE_IsParentProcess() && presContext->IsRoot() &&
+ presShell->GetDocument() != nullptr &&
+ presShell->GetRootScrollFrame() != nullptr &&
+ nsLayoutUtils::UsesAsyncScrolling(
+ presShell->GetRootScrollFrame())) {
+ if (dom::Element* element =
+ presShell->GetDocument()->GetDocumentElement()) {
+ if (!DisplayPortUtils::HasNonMinimalDisplayPort(element)) {
+ APZCCallbackHelper::InitializeRootDisplayport(presShell);
+ }
+ }
+ }
+
+ asrSetter.SetCurrentScrollParentId(id);
+
+ builder->SetVisibleRect(visibleRect);
+ builder->SetIsBuilding(true);
+ builder->SetAncestorHasApzAwareEventHandler(
+ gfxPlatform::AsyncPanZoomEnabled() &&
+ nsLayoutUtils::HasDocumentLevelListenersForApzAwareEvents(presShell));
+
+ // If a pref is toggled that adds or removes display list items,
+ // we need to rebuild the display list. The pref may be toggled
+ // manually by the user, or during test setup.
+ if (retainDisplayList &&
+ !builder->ShouldRebuildDisplayListDueToPrefChange()) {
+ // Attempt to do a partial build and merge into the existing list.
+ // This calls BuildDisplayListForStacking context on a subset of the
+ // viewport.
+ updateState = retainedBuilder->AttemptPartialUpdate(aBackstop);
+ metrics->EndPartialBuild(updateState);
+ } else {
+ // Partial updates are disabled.
+ DL_LOGI("Partial updates are disabled");
+ metrics->mPartialUpdateResult = PartialUpdateResult::Failed;
+ metrics->mPartialUpdateFailReason = PartialUpdateFailReason::Disabled;
+ }
+
+ // Rebuild the full display list if the partial display list build failed.
+ bool doFullRebuild = updateState == PartialUpdateResult::Failed;
+
+ if (StaticPrefs::layout_display_list_build_twice()) {
+ // Build display list twice to compare partial and full display list
+ // build times.
+ metrics->StartBuild();
+ doFullRebuild = true;
+ }
+
+ if (doFullRebuild) {
+ if (retainDisplayList) {
+ retainedBuilder->ClearRetainedData();
+#ifdef DEBUG
+ mozilla::RDLUtils::AssertFrameSubtreeUnmodified(
+ builder->RootReferenceFrame());
+#endif
+ }
+
+ list->DeleteAll(builder);
+
+ builder->ClearRetainedWindowRegions();
+ builder->ClearWillChangeBudgets();
+
+ builder->EnterPresShell(aFrame);
+ builder->SetDirtyRect(visibleRect);
+
+ DL_LOGI("Starting full display list build, root frame: %p",
+ builder->RootReferenceFrame());
+
+ aFrame->BuildDisplayListForStackingContext(builder, list);
+ AddExtraBackgroundItems(builder, list, aFrame, canvasArea, visibleRegion,
+ aBackstop);
+
+ builder->LeavePresShell(aFrame, list);
+ metrics->EndFullBuild();
+
+ DL_LOGI("Finished full display list build");
+ updateState = PartialUpdateResult::Updated;
+ }
+
+ builder->SetIsBuilding(false);
+ builder->IncrementPresShellPaintCount(presShell);
+ }
+
+ MOZ_ASSERT(updateState != PartialUpdateResult::Failed);
+ builder->Check();
+
+ const double geckoDLBuildTime =
+ (TimeStamp::Now() - startBuildDisplayList).ToMilliseconds();
+ mozilla::glean::paint::build_displaylist_time.StopAndAccumulate(
+ std::move(dlTimerId));
+
+ bool consoleNeedsDisplayList =
+ (gfxUtils::DumpDisplayList() || gfxEnv::MOZ_DUMP_PAINT()) &&
+ builder->IsInActiveDocShell();
+#ifdef MOZ_DUMP_PAINTING
+ FILE* savedDumpFile = gfxUtils::sDumpPaintFile;
+#endif
+
+ UniquePtr<std::stringstream> ss;
+ if (consoleNeedsDisplayList) {
+ ss = MakeUnique<std::stringstream>();
+ *ss << "Display list for " << uri << "\n";
+ DumpBeforePaintDisplayList(ss, builder, list, visibleRect);
+ }
+
+ uint32_t flags = nsDisplayList::PAINT_DEFAULT;
+ if (aFlags & PaintFrameFlags::WidgetLayers) {
+ flags |= nsDisplayList::PAINT_USE_WIDGET_LAYERS;
+ if (!(aFlags & PaintFrameFlags::DocumentRelative)) {
+ nsIWidget* widget = aFrame->GetNearestWidget();
+ if (widget) {
+ // If we're finished building display list items for painting of the
+ // outermost pres shell, notify the widget about any toolbars we've
+ // encountered.
+ widget->UpdateThemeGeometries(builder->GetThemeGeometries());
+ }
+ }
+ }
+ if (aFlags & PaintFrameFlags::ExistingTransaction) {
+ flags |= nsDisplayList::PAINT_EXISTING_TRANSACTION;
+ }
+ if (updateState == PartialUpdateResult::NoChange && !aRenderingContext) {
+ flags |= nsDisplayList::PAINT_IDENTICAL_DISPLAY_LIST;
+ }
+
+#ifdef PRINT_HITTESTINFO_STATS
+ if (XRE_IsContentProcess()) {
+ PrintHitTestInfoStats(list);
+ }
+#endif
+
+ TimeStamp paintStart = TimeStamp::Now();
+ list->PaintRoot(builder, aRenderingContext, flags, Some(geckoDLBuildTime));
+ Telemetry::AccumulateTimeDelta(Telemetry::PAINT_RASTERIZE_TIME, paintStart);
+
+ if (builder->IsPaintingToWindow()) {
+ presShell->EndPaint();
+ }
+ builder->Check();
+
+ if (consoleNeedsDisplayList) {
+ DumpAfterPaintDisplayList(ss, builder, list);
+ }
+
+#ifdef MOZ_DUMP_PAINTING
+ gfxUtils::sDumpPaintFile = savedDumpFile;
+#endif
+
+ // Update the widget's opaque region information. This sets
+ // glass boundaries on Windows. Also set up the window dragging region.
+ if ((aFlags & PaintFrameFlags::WidgetLayers) &&
+ !(aFlags & PaintFrameFlags::DocumentRelative)) {
+ nsIWidget* widget = aFrame->GetNearestWidget();
+ if (widget) {
+ nsRegion opaqueRegion;
+ opaqueRegion.And(builder->GetWindowExcludeGlassRegion(),
+ builder->GetWindowOpaqueRegion());
+ widget->UpdateOpaqueRegion(LayoutDeviceIntRegion::FromUnknownRegion(
+ opaqueRegion.ToNearestPixels(presContext->AppUnitsPerDevPixel())));
+
+ widget->UpdateWindowDraggingRegion(builder->GetWindowDraggingRegion());
+ }
+ }
+
+ // Apply effects updates if we were actually painting
+ if (isForPainting) {
+ ApplyEffectsUpdates(builder->GetEffectUpdates());
+ }
+
+ builder->Check();
+
+ {
+ AUTO_PROFILER_TRACING_MARKER("Paint", "DisplayListResources", GRAPHICS);
+
+ builder->EndFrame();
+
+ if (temporaryBuilder) {
+ temporaryBuilder.reset();
+ }
+ }
+
+ --paintFrameDepth;
+#if 0
+ if (XRE_IsParentProcess()) {
+ if (metrics->mPartialUpdateResult == PartialUpdateResult::Failed) {
+ printf("DL partial update failed: %s, Frame: %p\n",
+ metrics->FailReasonString(), aFrame);
+ } else {
+ printf(
+ "DL partial build success!"
+ " new: %d, reused: %d, rebuilt: %d, removed: %d, total: %d\n",
+ metrics->mNewItems, metrics->mReusedItems, metrics->mRebuiltItems,
+ metrics->mRemovedItems, metrics->mTotalItems);
+ }
+ }
+#endif
+}
+
+/**
+ * Uses a binary search for find where the cursor falls in the line of text
+ * It also keeps track of the part of the string that has already been measured
+ * so it doesn't have to keep measuring the same text over and over
+ *
+ * @param "aBaseWidth" contains the width in twips of the portion
+ * of the text that has already been measured, and aBaseInx contains
+ * the index of the text that has already been measured.
+ *
+ * @param aTextWidth returns the (in twips) the length of the text that falls
+ * before the cursor aIndex contains the index of the text where the cursor
+ * falls
+ */
+bool nsLayoutUtils::BinarySearchForPosition(
+ DrawTarget* aDrawTarget, nsFontMetrics& aFontMetrics, const char16_t* aText,
+ int32_t aBaseWidth, int32_t aBaseInx, int32_t aStartInx, int32_t aEndInx,
+ int32_t aCursorPos, int32_t& aIndex, int32_t& aTextWidth) {
+ int32_t range = aEndInx - aStartInx;
+ if ((range == 1) || (range == 2 && NS_IS_HIGH_SURROGATE(aText[aStartInx]))) {
+ aIndex = aStartInx + aBaseInx;
+ aTextWidth = nsLayoutUtils::AppUnitWidthOfString(aText, aIndex,
+ aFontMetrics, aDrawTarget);
+ return true;
+ }
+
+ int32_t inx = aStartInx + (range / 2);
+
+ // Make sure we don't leave a dangling low surrogate
+ if (NS_IS_HIGH_SURROGATE(aText[inx - 1])) inx++;
+
+ int32_t textWidth = nsLayoutUtils::AppUnitWidthOfString(
+ aText, inx, aFontMetrics, aDrawTarget);
+
+ int32_t fullWidth = aBaseWidth + textWidth;
+ if (fullWidth == aCursorPos) {
+ aTextWidth = textWidth;
+ aIndex = inx;
+ return true;
+ } else if (aCursorPos < fullWidth) {
+ aTextWidth = aBaseWidth;
+ if (BinarySearchForPosition(aDrawTarget, aFontMetrics, aText, aBaseWidth,
+ aBaseInx, aStartInx, inx, aCursorPos, aIndex,
+ aTextWidth)) {
+ return true;
+ }
+ } else {
+ aTextWidth = fullWidth;
+ if (BinarySearchForPosition(aDrawTarget, aFontMetrics, aText, aBaseWidth,
+ aBaseInx, inx, aEndInx, aCursorPos, aIndex,
+ aTextWidth)) {
+ return true;
+ }
+ }
+ return false;
+}
+
+void nsLayoutUtils::AddBoxesForFrame(nsIFrame* aFrame,
+ nsLayoutUtils::BoxCallback* aCallback) {
+ auto pseudoType = aFrame->Style()->GetPseudoType();
+
+ if (pseudoType == PseudoStyleType::tableWrapper) {
+ AddBoxesForFrame(aFrame->PrincipalChildList().FirstChild(), aCallback);
+ if (aCallback->mIncludeCaptionBoxForTable) {
+ nsIFrame* kid =
+ aFrame->GetChildList(FrameChildListID::Caption).FirstChild();
+ if (kid) {
+ AddBoxesForFrame(kid, aCallback);
+ }
+ }
+ } else if (pseudoType == PseudoStyleType::mozBlockInsideInlineWrapper ||
+ pseudoType == PseudoStyleType::mozMathMLAnonymousBlock) {
+ for (nsIFrame* kid : aFrame->PrincipalChildList()) {
+ AddBoxesForFrame(kid, aCallback);
+ }
+ } else {
+ aCallback->AddBox(aFrame);
+ }
+}
+
+void nsLayoutUtils::GetAllInFlowBoxes(nsIFrame* aFrame,
+ BoxCallback* aCallback) {
+ aCallback->mInTargetContinuation = false;
+ while (aFrame) {
+ AddBoxesForFrame(aFrame, aCallback);
+ aFrame = nsLayoutUtils::GetNextContinuationOrIBSplitSibling(aFrame);
+ aCallback->mInTargetContinuation = true;
+ }
+}
+
+nsIFrame* nsLayoutUtils::GetFirstNonAnonymousFrame(nsIFrame* aFrame) {
+ while (aFrame) {
+ auto pseudoType = aFrame->Style()->GetPseudoType();
+
+ if (pseudoType == PseudoStyleType::tableWrapper) {
+ nsIFrame* f =
+ GetFirstNonAnonymousFrame(aFrame->PrincipalChildList().FirstChild());
+ if (f) {
+ return f;
+ }
+ nsIFrame* kid =
+ aFrame->GetChildList(FrameChildListID::Caption).FirstChild();
+ if (kid) {
+ f = GetFirstNonAnonymousFrame(kid);
+ if (f) {
+ return f;
+ }
+ }
+ } else if (pseudoType == PseudoStyleType::mozBlockInsideInlineWrapper ||
+ pseudoType == PseudoStyleType::mozMathMLAnonymousBlock) {
+ for (nsIFrame* kid : aFrame->PrincipalChildList()) {
+ nsIFrame* f = GetFirstNonAnonymousFrame(kid);
+ if (f) {
+ return f;
+ }
+ }
+ } else {
+ return aFrame;
+ }
+
+ aFrame = nsLayoutUtils::GetNextContinuationOrIBSplitSibling(aFrame);
+ }
+ return nullptr;
+}
+
+struct BoxToRect : public nsLayoutUtils::BoxCallback {
+ const nsIFrame* mRelativeTo;
+ RectCallback* mCallback;
+ uint32_t mFlags;
+ // If the frame we're measuring relative to is the root, we know all frames
+ // are descendants of it, so we don't need to compute the common ancestor
+ // between a frame and mRelativeTo.
+ bool mRelativeToIsRoot;
+ // For the same reason, if the frame we're measuring relative to is the target
+ // (this is useful for IntersectionObserver), we know all frames are
+ // descendants of it except if we're in a continuation or ib-split-sibling of
+ // it.
+ bool mRelativeToIsTarget;
+
+ BoxToRect(const nsIFrame* aTargetFrame, const nsIFrame* aRelativeTo,
+ RectCallback* aCallback, uint32_t aFlags)
+ : mRelativeTo(aRelativeTo),
+ mCallback(aCallback),
+ mFlags(aFlags),
+ mRelativeToIsRoot(!aRelativeTo->GetParent()),
+ mRelativeToIsTarget(aRelativeTo == aTargetFrame) {}
+
+ void AddBox(nsIFrame* aFrame) override {
+ nsRect r;
+ nsIFrame* outer = SVGUtils::GetOuterSVGFrameAndCoveredRegion(aFrame, &r);
+ const bool usingSVGOuterFrame = !!outer;
+ if (!outer) {
+ outer = aFrame;
+ switch (mFlags & nsLayoutUtils::RECTS_WHICH_BOX_MASK) {
+ case nsLayoutUtils::RECTS_USE_CONTENT_BOX:
+ r = aFrame->GetContentRectRelativeToSelf();
+ break;
+ case nsLayoutUtils::RECTS_USE_PADDING_BOX:
+ r = aFrame->GetPaddingRectRelativeToSelf();
+ break;
+ case nsLayoutUtils::RECTS_USE_MARGIN_BOX:
+ r = aFrame->GetMarginRectRelativeToSelf();
+ break;
+ default: // Use the border box
+ r = aFrame->GetRectRelativeToSelf();
+ }
+ }
+ if (mFlags & nsLayoutUtils::RECTS_ACCOUNT_FOR_TRANSFORMS) {
+ const bool isAncestorKnown = [&] {
+ if (mRelativeToIsRoot) {
+ return true;
+ }
+ if (mRelativeToIsTarget && !mInTargetContinuation) {
+ return !usingSVGOuterFrame;
+ }
+ return false;
+ }();
+ if (isAncestorKnown) {
+ r = nsLayoutUtils::TransformFrameRectToAncestor(outer, r, mRelativeTo);
+ } else {
+ nsLayoutUtils::TransformRect(outer, mRelativeTo, r);
+ }
+ } else {
+ if (aFrame->PresContext() != mRelativeTo->PresContext()) {
+ r += outer->GetOffsetToCrossDoc(mRelativeTo);
+ } else {
+ r += outer->GetOffsetTo(mRelativeTo);
+ }
+ }
+ mCallback->AddRect(r);
+ }
+};
+
+struct MOZ_RAII BoxToRectAndText : public BoxToRect {
+ Sequence<nsString>* mTextList;
+
+ BoxToRectAndText(const nsIFrame* aTargetFrame, const nsIFrame* aRelativeTo,
+ RectCallback* aCallback, Sequence<nsString>* aTextList,
+ uint32_t aFlags)
+ : BoxToRect(aTargetFrame, aRelativeTo, aCallback, aFlags),
+ mTextList(aTextList) {}
+
+ static void AccumulateText(nsIFrame* aFrame, nsAString& aResult) {
+ MOZ_ASSERT(aFrame);
+
+ // Get all the text in aFrame and child frames, while respecting
+ // the content offsets in each of the nsTextFrames.
+ if (aFrame->IsTextFrame()) {
+ nsTextFrame* textFrame = static_cast<nsTextFrame*>(aFrame);
+
+ nsIFrame::RenderedText renderedText = textFrame->GetRenderedText(
+ textFrame->GetContentOffset(),
+ textFrame->GetContentOffset() + textFrame->GetContentLength(),
+ nsIFrame::TextOffsetType::OffsetsInContentText,
+ nsIFrame::TrailingWhitespace::DontTrim);
+
+ aResult.Append(renderedText.mString);
+ }
+
+ for (nsIFrame* child = aFrame->PrincipalChildList().FirstChild(); child;
+ child = child->GetNextSibling()) {
+ AccumulateText(child, aResult);
+ }
+ }
+
+ void AddBox(nsIFrame* aFrame) override {
+ BoxToRect::AddBox(aFrame);
+ if (mTextList) {
+ nsString* textForFrame = mTextList->AppendElement(fallible);
+ if (textForFrame) {
+ AccumulateText(aFrame, *textForFrame);
+ }
+ }
+ }
+};
+
+void nsLayoutUtils::GetAllInFlowRects(nsIFrame* aFrame,
+ const nsIFrame* aRelativeTo,
+ RectCallback* aCallback,
+ uint32_t aFlags) {
+ BoxToRect converter(aFrame, aRelativeTo, aCallback, aFlags);
+ GetAllInFlowBoxes(aFrame, &converter);
+}
+
+void nsLayoutUtils::GetAllInFlowRectsAndTexts(nsIFrame* aFrame,
+ const nsIFrame* aRelativeTo,
+ RectCallback* aCallback,
+ Sequence<nsString>* aTextList,
+ uint32_t aFlags) {
+ BoxToRectAndText converter(aFrame, aRelativeTo, aCallback, aTextList, aFlags);
+ GetAllInFlowBoxes(aFrame, &converter);
+}
+
+nsLayoutUtils::RectAccumulator::RectAccumulator() : mSeenFirstRect(false) {}
+
+void nsLayoutUtils::RectAccumulator::AddRect(const nsRect& aRect) {
+ mResultRect.UnionRect(mResultRect, aRect);
+ if (!mSeenFirstRect) {
+ mSeenFirstRect = true;
+ mFirstRect = aRect;
+ }
+}
+
+nsLayoutUtils::RectListBuilder::RectListBuilder(DOMRectList* aList)
+ : mRectList(aList) {}
+
+void nsLayoutUtils::RectListBuilder::AddRect(const nsRect& aRect) {
+ RefPtr<DOMRect> rect = new DOMRect(mRectList);
+
+ rect->SetLayoutRect(aRect);
+ mRectList->Append(rect);
+}
+
+nsIFrame* nsLayoutUtils::GetContainingBlockForClientRect(nsIFrame* aFrame) {
+ return aFrame->PresShell()->GetRootFrame();
+}
+
+nsRect nsLayoutUtils::GetAllInFlowRectsUnion(nsIFrame* aFrame,
+ const nsIFrame* aRelativeTo,
+ uint32_t aFlags) {
+ RectAccumulator accumulator;
+ GetAllInFlowRects(aFrame, aRelativeTo, &accumulator, aFlags);
+ return accumulator.mResultRect.IsEmpty() ? accumulator.mFirstRect
+ : accumulator.mResultRect;
+}
+
+nsRect nsLayoutUtils::GetTextShadowRectsUnion(
+ const nsRect& aTextAndDecorationsRect, nsIFrame* aFrame, uint32_t aFlags) {
+ const nsStyleText* textStyle = aFrame->StyleText();
+ auto shadows = textStyle->mTextShadow.AsSpan();
+ if (shadows.IsEmpty()) {
+ return aTextAndDecorationsRect;
+ }
+
+ nsRect resultRect = aTextAndDecorationsRect;
+ int32_t A2D = aFrame->PresContext()->AppUnitsPerDevPixel();
+ for (auto& shadow : shadows) {
+ nsMargin blur =
+ nsContextBoxBlur::GetBlurRadiusMargin(shadow.blur.ToAppUnits(), A2D);
+ if ((aFlags & EXCLUDE_BLUR_SHADOWS) && blur != nsMargin(0, 0, 0, 0))
+ continue;
+
+ nsRect tmpRect(aTextAndDecorationsRect);
+
+ tmpRect.MoveBy(
+ nsPoint(shadow.horizontal.ToAppUnits(), shadow.vertical.ToAppUnits()));
+ tmpRect.Inflate(blur);
+
+ resultRect.UnionRect(resultRect, tmpRect);
+ }
+ return resultRect;
+}
+
+enum ObjectDimensionType { eWidth, eHeight };
+static nscoord ComputeMissingDimension(
+ const nsSize& aDefaultObjectSize, const AspectRatio& aIntrinsicRatio,
+ const Maybe<nscoord>& aSpecifiedWidth,
+ const Maybe<nscoord>& aSpecifiedHeight,
+ ObjectDimensionType aDimensionToCompute) {
+ // The "default sizing algorithm" computes the missing dimension as follows:
+ // (source: http://dev.w3.org/csswg/css-images-3/#default-sizing )
+
+ // 1. "If the object has an intrinsic aspect ratio, the missing dimension of
+ // the concrete object size is calculated using the intrinsic aspect
+ // ratio and the present dimension."
+ if (aIntrinsicRatio) {
+ // Fill in the missing dimension using the intrinsic aspect ratio.
+ if (aDimensionToCompute == eWidth) {
+ return aIntrinsicRatio.ApplyTo(*aSpecifiedHeight);
+ }
+ return aIntrinsicRatio.Inverted().ApplyTo(*aSpecifiedWidth);
+ }
+
+ // 2. "Otherwise, if the missing dimension is present in the object's
+ // intrinsic dimensions, [...]"
+ // NOTE: *Skipping* this case, because we already know it's not true -- we're
+ // in this function because the missing dimension is *not* present in
+ // the object's intrinsic dimensions.
+
+ // 3. "Otherwise, the missing dimension of the concrete object size is taken
+ // from the default object size. "
+ return (aDimensionToCompute == eWidth) ? aDefaultObjectSize.width
+ : aDefaultObjectSize.height;
+}
+
+/*
+ * This computes & returns the concrete object size of replaced content, if
+ * that content were to be rendered with "object-fit: none". (Or, if the
+ * element has neither an intrinsic height nor width, this method returns an
+ * empty Maybe<> object.)
+ *
+ * As specced...
+ * http://dev.w3.org/csswg/css-images-3/#valdef-object-fit-none
+ * ..we use "the default sizing algorithm with no specified size,
+ * and a default object size equal to the replaced element's used width and
+ * height."
+ *
+ * The default sizing algorithm is described here:
+ * http://dev.w3.org/csswg/css-images-3/#default-sizing
+ * Quotes in the function-impl are taken from that ^ spec-text.
+ *
+ * Per its final bulleted section: since there's no specified size,
+ * we run the default sizing algorithm using the object's intrinsic size in
+ * place of the specified size. But if the object has neither an intrinsic
+ * height nor an intrinsic width, then we instead return without populating our
+ * outparam, and we let the caller figure out the size (using a contain
+ * constraint).
+ */
+static Maybe<nsSize> MaybeComputeObjectFitNoneSize(
+ const nsSize& aDefaultObjectSize, const IntrinsicSize& aIntrinsicSize,
+ const AspectRatio& aIntrinsicRatio) {
+ // "If the object has an intrinsic height or width, its size is resolved as
+ // if its intrinsic dimensions were given as the specified size."
+ //
+ // So, first we check if we have an intrinsic height and/or width:
+ const Maybe<nscoord>& specifiedWidth = aIntrinsicSize.width;
+ const Maybe<nscoord>& specifiedHeight = aIntrinsicSize.height;
+
+ Maybe<nsSize> noneSize; // (the value we'll return)
+ if (specifiedWidth || specifiedHeight) {
+ // We have at least one specified dimension; use whichever dimension is
+ // specified, and compute the other one using our intrinsic ratio, or (if
+ // no valid ratio) using the default object size.
+ noneSize.emplace();
+
+ noneSize->width =
+ specifiedWidth
+ ? *specifiedWidth
+ : ComputeMissingDimension(aDefaultObjectSize, aIntrinsicRatio,
+ specifiedWidth, specifiedHeight, eWidth);
+
+ noneSize->height =
+ specifiedHeight
+ ? *specifiedHeight
+ : ComputeMissingDimension(aDefaultObjectSize, aIntrinsicRatio,
+ specifiedWidth, specifiedHeight, eHeight);
+ }
+ // [else:] "Otherwise [if there's neither an intrinsic height nor width], its
+ // size is resolved as a contain constraint against the default object size."
+ // We'll let our caller do that, to share code & avoid redundant
+ // computations; so, we return w/out populating noneSize.
+ return noneSize;
+}
+
+// Computes the concrete object size to render into, as described at
+// http://dev.w3.org/csswg/css-images-3/#concrete-size-resolution
+static nsSize ComputeConcreteObjectSize(const nsSize& aConstraintSize,
+ const IntrinsicSize& aIntrinsicSize,
+ const AspectRatio& aIntrinsicRatio,
+ StyleObjectFit aObjectFit) {
+ // Handle default behavior (filling the container) w/ fast early return.
+ // (Also: if there's no valid intrinsic ratio, then we have the "fill"
+ // behavior & just use the constraint size.)
+ if (MOZ_LIKELY(aObjectFit == StyleObjectFit::Fill) || !aIntrinsicRatio) {
+ return aConstraintSize;
+ }
+
+ // The type of constraint to compute (cover/contain), if needed:
+ Maybe<nsImageRenderer::FitType> fitType;
+
+ Maybe<nsSize> noneSize;
+ if (aObjectFit == StyleObjectFit::None ||
+ aObjectFit == StyleObjectFit::ScaleDown) {
+ noneSize = MaybeComputeObjectFitNoneSize(aConstraintSize, aIntrinsicSize,
+ aIntrinsicRatio);
+ if (!noneSize || aObjectFit == StyleObjectFit::ScaleDown) {
+ // Need to compute a 'CONTAIN' constraint (either for the 'none' size
+ // itself, or for comparison w/ the 'none' size to resolve 'scale-down'.)
+ fitType.emplace(nsImageRenderer::CONTAIN);
+ }
+ } else if (aObjectFit == StyleObjectFit::Cover) {
+ fitType.emplace(nsImageRenderer::COVER);
+ } else if (aObjectFit == StyleObjectFit::Contain) {
+ fitType.emplace(nsImageRenderer::CONTAIN);
+ }
+
+ Maybe<nsSize> constrainedSize;
+ if (fitType) {
+ constrainedSize.emplace(nsImageRenderer::ComputeConstrainedSize(
+ aConstraintSize, aIntrinsicRatio, *fitType));
+ }
+
+ // Now, we should have all the sizing information that we need.
+ switch (aObjectFit) {
+ // skipping StyleObjectFit::Fill; we handled it w/ early-return.
+ case StyleObjectFit::Contain:
+ case StyleObjectFit::Cover:
+ MOZ_ASSERT(constrainedSize);
+ return *constrainedSize;
+
+ case StyleObjectFit::None:
+ if (noneSize) {
+ return *noneSize;
+ }
+ MOZ_ASSERT(constrainedSize);
+ return *constrainedSize;
+
+ case StyleObjectFit::ScaleDown:
+ MOZ_ASSERT(constrainedSize);
+ if (noneSize) {
+ constrainedSize->width =
+ std::min(constrainedSize->width, noneSize->width);
+ constrainedSize->height =
+ std::min(constrainedSize->height, noneSize->height);
+ }
+ return *constrainedSize;
+
+ default:
+ MOZ_ASSERT_UNREACHABLE("Unexpected enum value for 'object-fit'");
+ return aConstraintSize; // fall back to (default) 'fill' behavior
+ }
+}
+
+// (Helper for HasInitialObjectFitAndPosition, to check
+// each "object-position" coord.)
+static bool IsCoord50Pct(const LengthPercentage& aCoord) {
+ return aCoord.ConvertsToPercentage() && aCoord.ToPercentage() == 0.5f;
+}
+
+// Indicates whether the given nsStylePosition has the initial values
+// for the "object-fit" and "object-position" properties.
+static bool HasInitialObjectFitAndPosition(const nsStylePosition* aStylePos) {
+ const Position& objectPos = aStylePos->mObjectPosition;
+
+ return aStylePos->mObjectFit == StyleObjectFit::Fill &&
+ IsCoord50Pct(objectPos.horizontal) && IsCoord50Pct(objectPos.vertical);
+}
+
+/* static */
+nsRect nsLayoutUtils::ComputeObjectDestRect(const nsRect& aConstraintRect,
+ const IntrinsicSize& aIntrinsicSize,
+ const AspectRatio& aIntrinsicRatio,
+ const nsStylePosition* aStylePos,
+ nsPoint* aAnchorPoint) {
+ // Step 1: Figure out our "concrete object size"
+ // (the size of the region we'll actually draw our image's pixels into).
+ nsSize concreteObjectSize =
+ ComputeConcreteObjectSize(aConstraintRect.Size(), aIntrinsicSize,
+ aIntrinsicRatio, aStylePos->mObjectFit);
+
+ // Step 2: Figure out how to align that region in the element's content-box.
+ nsPoint imageTopLeftPt, imageAnchorPt;
+ nsImageRenderer::ComputeObjectAnchorPoint(
+ aStylePos->mObjectPosition, aConstraintRect.Size(), concreteObjectSize,
+ &imageTopLeftPt, &imageAnchorPt);
+ // Right now, we're with respect to aConstraintRect's top-left point. We add
+ // that point here, to convert to the same broader coordinate space that
+ // aConstraintRect is in.
+ imageTopLeftPt += aConstraintRect.TopLeft();
+ imageAnchorPt += aConstraintRect.TopLeft();
+
+ if (aAnchorPoint) {
+ // Special-case: if our "object-fit" and "object-position" properties have
+ // their default values ("object-fit: fill; object-position:50% 50%"), then
+ // we'll override the calculated imageAnchorPt, and instead use the
+ // object's top-left corner.
+ //
+ // This special case is partly for backwards compatibility (since
+ // traditionally we've pixel-aligned the top-left corner of e.g. <img>
+ // elements), and partly because ComputeSnappedDrawingParameters produces
+ // less error if the anchor point is at the top-left corner. So, all other
+ // things being equal, we prefer that code path with less error.
+ if (HasInitialObjectFitAndPosition(aStylePos)) {
+ *aAnchorPoint = imageTopLeftPt;
+ } else {
+ *aAnchorPoint = imageAnchorPt;
+ }
+ }
+ return nsRect(imageTopLeftPt, concreteObjectSize);
+}
+
+already_AddRefed<nsFontMetrics> nsLayoutUtils::GetFontMetricsForFrame(
+ const nsIFrame* aFrame, float aInflation) {
+ ComputedStyle* computedStyle = aFrame->Style();
+ uint8_t variantWidth = NS_FONT_VARIANT_WIDTH_NORMAL;
+ if (computedStyle->IsTextCombined()) {
+ MOZ_ASSERT(aFrame->IsTextFrame());
+ auto textFrame = static_cast<const nsTextFrame*>(aFrame);
+ auto clusters = textFrame->CountGraphemeClusters();
+ if (clusters == 2) {
+ variantWidth = NS_FONT_VARIANT_WIDTH_HALF;
+ } else if (clusters == 3) {
+ variantWidth = NS_FONT_VARIANT_WIDTH_THIRD;
+ } else if (clusters == 4) {
+ variantWidth = NS_FONT_VARIANT_WIDTH_QUARTER;
+ }
+ }
+ return GetFontMetricsForComputedStyle(computedStyle, aFrame->PresContext(),
+ aInflation, variantWidth);
+}
+
+already_AddRefed<nsFontMetrics> nsLayoutUtils::GetFontMetricsForComputedStyle(
+ const ComputedStyle* aComputedStyle, nsPresContext* aPresContext,
+ float aInflation, uint8_t aVariantWidth) {
+ WritingMode wm(aComputedStyle);
+ const nsStyleFont* styleFont = aComputedStyle->StyleFont();
+ nsFontMetrics::Params params;
+ params.language = styleFont->mLanguage;
+ params.explicitLanguage = styleFont->mExplicitLanguage;
+ params.orientation = wm.IsVertical() && !wm.IsSideways()
+ ? nsFontMetrics::eVertical
+ : nsFontMetrics::eHorizontal;
+ // pass the user font set object into the device context to
+ // pass along to CreateFontGroup
+ params.userFontSet = aPresContext->GetUserFontSet();
+ params.textPerf = aPresContext->GetTextPerfMetrics();
+ params.featureValueLookup = aPresContext->GetFontFeatureValuesLookup();
+
+ // When aInflation is 1.0 and we don't require width variant, avoid
+ // making a local copy of the nsFont.
+ // This also avoids running font.size through floats when it is large,
+ // which would be lossy. Fortunately, in such cases, aInflation is
+ // guaranteed to be 1.0f.
+ if (aInflation == 1.0f && aVariantWidth == NS_FONT_VARIANT_WIDTH_NORMAL) {
+ return aPresContext->GetMetricsFor(styleFont->mFont, params);
+ }
+
+ nsFont font = styleFont->mFont;
+ MOZ_ASSERT(!std::isnan(float(font.size.ToCSSPixels())),
+ "Style font should never be NaN");
+ font.size.ScaleBy(aInflation);
+ if (MOZ_UNLIKELY(std::isnan(float(font.size.ToCSSPixels())))) {
+ font.size = {0};
+ }
+ font.variantWidth = aVariantWidth;
+ return aPresContext->GetMetricsFor(font, params);
+}
+
+nsIFrame* nsLayoutUtils::FindChildContainingDescendant(
+ nsIFrame* aParent, nsIFrame* aDescendantFrame) {
+ nsIFrame* result = aDescendantFrame;
+
+ while (result) {
+ nsIFrame* parent = result->GetParent();
+ if (parent == aParent) {
+ break;
+ }
+
+ // The frame is not an immediate child of aParent so walk up another level
+ result = parent;
+ }
+
+ return result;
+}
+
+nsBlockFrame* nsLayoutUtils::FindNearestBlockAncestor(nsIFrame* aFrame) {
+ nsIFrame* nextAncestor;
+ for (nextAncestor = aFrame->GetParent(); nextAncestor;
+ nextAncestor = nextAncestor->GetParent()) {
+ nsBlockFrame* block = do_QueryFrame(nextAncestor);
+ if (block) return block;
+ }
+ return nullptr;
+}
+
+nsIFrame* nsLayoutUtils::GetNonGeneratedAncestor(nsIFrame* aFrame) {
+ if (!aFrame->HasAnyStateBits(NS_FRAME_GENERATED_CONTENT)) return aFrame;
+
+ nsIFrame* f = aFrame;
+ do {
+ f = GetParentOrPlaceholderFor(f);
+ } while (f->HasAnyStateBits(NS_FRAME_GENERATED_CONTENT));
+ return f;
+}
+
+nsIFrame* nsLayoutUtils::GetParentOrPlaceholderFor(const nsIFrame* aFrame) {
+ // This condition must match the condition in FindContainingBlocks in
+ // RetainedDisplayListBuider.cpp, MarkFrameForDisplayIfVisible and
+ // UnmarkFrameForDisplayIfVisible in nsDisplayList.cpp
+ if (aFrame->HasAnyStateBits(NS_FRAME_OUT_OF_FLOW) &&
+ !aFrame->GetPrevInFlow()) {
+ return aFrame->GetProperty(nsIFrame::PlaceholderFrameProperty());
+ }
+ return aFrame->GetParent();
+}
+
+nsIFrame* nsLayoutUtils::GetParentOrPlaceholderForCrossDoc(
+ const nsIFrame* aFrame) {
+ nsIFrame* f = GetParentOrPlaceholderFor(aFrame);
+ if (f) return f;
+ return GetCrossDocParentFrameInProcess(aFrame);
+}
+
+nsIFrame* nsLayoutUtils::GetDisplayListParent(nsIFrame* aFrame) {
+ if (aFrame->HasAnyStateBits(NS_FRAME_IS_PUSHED_FLOAT)) {
+ return aFrame->GetParent();
+ }
+ return nsLayoutUtils::GetParentOrPlaceholderForCrossDoc(aFrame);
+}
+
+nsIFrame* nsLayoutUtils::GetPrevContinuationOrIBSplitSibling(
+ const nsIFrame* aFrame) {
+ if (nsIFrame* result = aFrame->GetPrevContinuation()) {
+ return result;
+ }
+
+ if (aFrame->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT)) {
+ // We are the first frame in the continuation chain. Get the ib-split prev
+ // sibling property stored in us.
+ return aFrame->GetProperty(nsIFrame::IBSplitPrevSibling());
+ }
+
+ return nullptr;
+}
+
+nsIFrame* nsLayoutUtils::GetNextContinuationOrIBSplitSibling(
+ const nsIFrame* aFrame) {
+ if (nsIFrame* result = aFrame->GetNextContinuation()) {
+ return result;
+ }
+
+ if (aFrame->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT)) {
+ // We only store the ib-split sibling annotation with the first frame in the
+ // continuation chain.
+ return aFrame->FirstContinuation()->GetProperty(nsIFrame::IBSplitSibling());
+ }
+
+ return nullptr;
+}
+
+nsIFrame* nsLayoutUtils::FirstContinuationOrIBSplitSibling(
+ const nsIFrame* aFrame) {
+ nsIFrame* result = aFrame->FirstContinuation();
+
+ if (result->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT)) {
+ while (auto* f = result->GetProperty(nsIFrame::IBSplitPrevSibling())) {
+ result = f;
+ }
+ }
+
+ return result;
+}
+
+nsIFrame* nsLayoutUtils::LastContinuationOrIBSplitSibling(
+ const nsIFrame* aFrame) {
+ nsIFrame* result = aFrame->FirstContinuation();
+
+ if (result->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT)) {
+ while (auto* f = result->GetProperty(nsIFrame::IBSplitSibling())) {
+ result = f;
+ }
+ }
+
+ return result->LastContinuation();
+}
+
+bool nsLayoutUtils::IsFirstContinuationOrIBSplitSibling(
+ const nsIFrame* aFrame) {
+ if (aFrame->GetPrevContinuation()) {
+ return false;
+ }
+ if (aFrame->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT) &&
+ aFrame->GetProperty(nsIFrame::IBSplitPrevSibling())) {
+ return false;
+ }
+
+ return true;
+}
+
+bool nsLayoutUtils::IsViewportScrollbarFrame(nsIFrame* aFrame) {
+ if (!aFrame) return false;
+
+ nsIFrame* rootScrollFrame = aFrame->PresShell()->GetRootScrollFrame();
+ if (!rootScrollFrame) return false;
+
+ nsIScrollableFrame* rootScrollableFrame = do_QueryFrame(rootScrollFrame);
+ NS_ASSERTION(rootScrollableFrame, "The root scorollable frame is null");
+
+ if (!IsProperAncestorFrame(rootScrollFrame, aFrame)) return false;
+
+ nsIFrame* rootScrolledFrame = rootScrollableFrame->GetScrolledFrame();
+ return !(rootScrolledFrame == aFrame ||
+ IsProperAncestorFrame(rootScrolledFrame, aFrame));
+}
+
+/**
+ * Use only for paddings / widths / heights, since it clamps negative calc() to
+ * 0.
+ */
+template <typename LengthPercentageLike>
+static bool GetAbsoluteCoord(const LengthPercentageLike& aStyle,
+ nscoord& aResult) {
+ if (!aStyle.ConvertsToLength()) {
+ return false;
+ }
+ aResult = std::max(0, aStyle.ToLength());
+ return true;
+}
+
+static nscoord GetBSizePercentBasisAdjustment(StyleBoxSizing aBoxSizing,
+ nsIFrame* aFrame,
+ bool aHorizontalAxis,
+ bool aResolvesAgainstPaddingBox);
+
+static bool GetPercentBSize(const LengthPercentage& aStyle, nsIFrame* aFrame,
+ bool aHorizontalAxis, nscoord& aResult);
+
+// Only call on style coords for which GetAbsoluteCoord returned false.
+template <typename SizeOrMaxSize>
+static bool GetPercentBSize(const SizeOrMaxSize& aStyle, nsIFrame* aFrame,
+ bool aHorizontalAxis, nscoord& aResult) {
+ if (!aStyle.IsLengthPercentage()) {
+ return false;
+ }
+ return GetPercentBSize(aStyle.AsLengthPercentage(), aFrame, aHorizontalAxis,
+ aResult);
+}
+
+static bool GetPercentBSize(const LengthPercentage& aStyle, nsIFrame* aFrame,
+ bool aHorizontalAxis, nscoord& aResult) {
+ if (!aStyle.HasPercent()) {
+ return false;
+ }
+
+ MOZ_ASSERT(!aStyle.ConvertsToLength(),
+ "GetAbsoluteCoord should have handled this");
+
+ // During reflow, nsHTMLScrollFrame::ReflowScrolledFrame uses
+ // SetComputedHeight on the reflow input for its child to propagate its
+ // computed height to the scrolled content. So here we skip to the scroll
+ // frame that contains this scrolled content in order to get the same
+ // behavior as layout when computing percentage heights.
+ nsIFrame* f = aFrame->GetContainingBlock(nsIFrame::SKIP_SCROLLED_FRAME);
+ if (!f) {
+ MOZ_ASSERT_UNREACHABLE("top of frame tree not a containing block");
+ return false;
+ }
+
+ WritingMode wm = f->GetWritingMode();
+
+ const nsStylePosition* pos = f->StylePosition();
+ const auto& bSizeCoord = pos->BSize(wm);
+ nscoord h;
+ if (!GetAbsoluteCoord(bSizeCoord, h) &&
+ !GetPercentBSize(bSizeCoord, f, aHorizontalAxis, h)) {
+ LayoutFrameType fType = f->Type();
+ if (fType != LayoutFrameType::Viewport &&
+ fType != LayoutFrameType::Canvas &&
+ fType != LayoutFrameType::PageContent) {
+ // There's no basis for the percentage height, so it acts like auto.
+ // Should we consider a max-height < min-height pair a basis for
+ // percentage heights? The spec is somewhat unclear, and not doing
+ // so is simpler and avoids troubling discontinuities in behavior,
+ // so I'll choose not to. -LDB
+ return false;
+ }
+ // For the viewport, canvas, and page-content kids, the percentage
+ // basis is just the parent block-size.
+ h = f->BSize(wm);
+ if (h == NS_UNCONSTRAINEDSIZE) {
+ // We don't have a percentage basis after all
+ return false;
+ }
+ }
+
+ const auto& maxBSizeCoord = pos->MaxBSize(wm);
+
+ nscoord maxh;
+ if (GetAbsoluteCoord(maxBSizeCoord, maxh) ||
+ GetPercentBSize(maxBSizeCoord, f, aHorizontalAxis, maxh)) {
+ if (maxh < h) h = maxh;
+ }
+
+ const auto& minBSizeCoord = pos->MinBSize(wm);
+
+ nscoord minh;
+ if (GetAbsoluteCoord(minBSizeCoord, minh) ||
+ GetPercentBSize(minBSizeCoord, f, aHorizontalAxis, minh)) {
+ if (minh > h) {
+ h = minh;
+ }
+ }
+
+ // If we're an abspos box, percentages in that case resolve against the
+ // padding box.
+ //
+ // TODO: This could conceivably cause some problems with fieldsets (which are
+ // the other place that wants to ignore padding), but solving that here
+ // without hardcoding a check for f being a fieldset-content frame is a bit of
+ // a pain.
+ const bool resolvesAgainstPaddingBox = aFrame->IsAbsolutelyPositioned();
+ h += GetBSizePercentBasisAdjustment(pos->mBoxSizing, f, aHorizontalAxis,
+ resolvesAgainstPaddingBox);
+
+ aResult = std::max(aStyle.Resolve(std::max(h, 0)), 0);
+ return true;
+}
+
+// Return true if aStyle can be resolved to a definite value and if so
+// return that value in aResult.
+static bool GetDefiniteSize(const LengthPercentage& aStyle, nsIFrame* aFrame,
+ bool aIsInlineAxis,
+ const Maybe<LogicalSize>& aPercentageBasis,
+ nscoord* aResult) {
+ if (aStyle.ConvertsToLength()) {
+ *aResult = aStyle.ToLength();
+ return true;
+ }
+
+ if (!aPercentageBasis) {
+ return false;
+ }
+
+ auto wm = aFrame->GetWritingMode();
+ nscoord pb = aIsInlineAxis ? aPercentageBasis.value().ISize(wm)
+ : aPercentageBasis.value().BSize(wm);
+ if (pb == NS_UNCONSTRAINEDSIZE) {
+ return false;
+ }
+ *aResult = std::max(0, aStyle.Resolve(pb));
+ return true;
+}
+
+// Return true if aStyle can be resolved to a definite value and if so
+// return that value in aResult.
+template <typename SizeOrMaxSize>
+static bool GetDefiniteSize(const SizeOrMaxSize& aStyle, nsIFrame* aFrame,
+ bool aIsInlineAxis,
+ const Maybe<LogicalSize>& aPercentageBasis,
+ nscoord* aResult) {
+ if (!aStyle.IsLengthPercentage()) {
+ return false;
+ }
+ return GetDefiniteSize(aStyle.AsLengthPercentage(), aFrame, aIsInlineAxis,
+ aPercentageBasis, aResult);
+}
+
+// NOTE: this function will be replaced by GetDefiniteSizeTakenByBoxSizing (bug
+// 1363918). Please do not add new uses of this function.
+//
+// Get the amount of space to add or subtract out of aFrame's 'block-size' or
+// property value due its borders and paddings, given the box-sizing value in
+// aBoxSizing.
+//
+// aHorizontalAxis is true if our inline direction is horizontal and our block
+// direction is vertical. aResolvesAgainstPaddingBox is true if padding should
+// be added or not removed.
+static nscoord GetBSizePercentBasisAdjustment(StyleBoxSizing aBoxSizing,
+ nsIFrame* aFrame,
+ bool aHorizontalAxis,
+ bool aResolvesAgainstPaddingBox) {
+ nscoord adjustment = 0;
+ if (aBoxSizing == StyleBoxSizing::Border) {
+ const auto& border = aFrame->StyleBorder()->GetComputedBorder();
+ adjustment -= aHorizontalAxis ? border.TopBottom() : border.LeftRight();
+ }
+ if ((aBoxSizing == StyleBoxSizing::Border) == !aResolvesAgainstPaddingBox) {
+ const auto& stylePadding = aFrame->StylePadding()->mPadding;
+ const LengthPercentage& paddingStart =
+ stylePadding.Get(aHorizontalAxis ? eSideTop : eSideLeft);
+ const LengthPercentage& paddingEnd =
+ stylePadding.Get(aHorizontalAxis ? eSideBottom : eSideRight);
+ nscoord pad;
+ // XXXbz Calling GetPercentBSize on padding values looks bogus, since
+ // percent padding is always a percentage of the inline-size of the
+ // containing block. We should perhaps just treat non-absolute paddings
+ // here as 0 instead, except that in some cases the width may in fact be
+ // known. See bug 1231059.
+ if (GetAbsoluteCoord(paddingStart, pad) ||
+ GetPercentBSize(paddingStart, aFrame, aHorizontalAxis, pad)) {
+ adjustment += aResolvesAgainstPaddingBox ? pad : -pad;
+ }
+ if (GetAbsoluteCoord(paddingEnd, pad) ||
+ GetPercentBSize(paddingEnd, aFrame, aHorizontalAxis, pad)) {
+ adjustment += aResolvesAgainstPaddingBox ? pad : -pad;
+ }
+ }
+ return adjustment;
+}
+
+// Get the amount of space taken out of aFrame's content area due to its
+// borders and paddings given the box-sizing value in aBoxSizing. We don't
+// get aBoxSizing from the frame because some callers want to compute this for
+// specific box-sizing values.
+// aIsInlineAxis is true if we're computing for aFrame's inline axis.
+// aIgnorePadding is true if padding should be ignored.
+static nscoord GetDefiniteSizeTakenByBoxSizing(
+ StyleBoxSizing aBoxSizing, nsIFrame* aFrame, bool aIsInlineAxis,
+ bool aIgnorePadding, const Maybe<LogicalSize>& aPercentageBasis) {
+ nscoord sizeTakenByBoxSizing = 0;
+ if (MOZ_UNLIKELY(aBoxSizing == StyleBoxSizing::Border)) {
+ const bool isHorizontalAxis =
+ aIsInlineAxis == !aFrame->GetWritingMode().IsVertical();
+ const nsStyleBorder* styleBorder = aFrame->StyleBorder();
+ sizeTakenByBoxSizing = isHorizontalAxis
+ ? styleBorder->GetComputedBorder().LeftRight()
+ : styleBorder->GetComputedBorder().TopBottom();
+ if (!aIgnorePadding) {
+ const auto& stylePadding = aFrame->StylePadding()->mPadding;
+ const LengthPercentage& pStart =
+ stylePadding.Get(isHorizontalAxis ? eSideLeft : eSideTop);
+ const LengthPercentage& pEnd =
+ stylePadding.Get(isHorizontalAxis ? eSideRight : eSideBottom);
+ nscoord pad;
+ // XXXbz Calling GetPercentBSize on padding values looks bogus, since
+ // percent padding is always a percentage of the inline-size of the
+ // containing block. We should perhaps just treat non-absolute paddings
+ // here as 0 instead, except that in some cases the width may in fact be
+ // known. See bug 1231059.
+ if (GetDefiniteSize(pStart, aFrame, aIsInlineAxis, aPercentageBasis,
+ &pad) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(pStart, aFrame, isHorizontalAxis, pad))) {
+ sizeTakenByBoxSizing += pad;
+ }
+ if (GetDefiniteSize(pEnd, aFrame, aIsInlineAxis, aPercentageBasis,
+ &pad) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(pEnd, aFrame, isHorizontalAxis, pad))) {
+ sizeTakenByBoxSizing += pad;
+ }
+ }
+ }
+ return sizeTakenByBoxSizing;
+}
+
+/**
+ * Handles only max-content and min-content, and
+ * -moz-fit-content for min-width and max-width, since the others
+ * (-moz-fit-content for width, and -moz-available) have no effect on
+ * intrinsic widths.
+ */
+static bool GetIntrinsicCoord(nsIFrame::ExtremumLength aStyle,
+ gfxContext* aRenderingContext, nsIFrame* aFrame,
+ Maybe<nscoord> aInlineSizeFromAspectRatio,
+ nsIFrame::SizeProperty aProperty,
+ nscoord aContentBoxToBoxSizingDiff,
+ nscoord& aResult) {
+ if (aStyle == nsIFrame::ExtremumLength::MozAvailable) {
+ return false;
+ }
+
+ if (aStyle == nsIFrame::ExtremumLength::FitContentFunction) {
+ // fit-content() should be handled by the caller.
+ return false;
+ }
+
+ if (aStyle == nsIFrame::ExtremumLength::FitContent) {
+ switch (aProperty) {
+ case nsIFrame::SizeProperty::Size:
+ // handle like 'width: auto'
+ return false;
+ case nsIFrame::SizeProperty::MaxSize:
+ // constrain large 'width' values down to max-content
+ aStyle = nsIFrame::ExtremumLength::MaxContent;
+ break;
+ case nsIFrame::SizeProperty::MinSize:
+ // constrain small 'width' or 'max-width' values up to min-content
+ aStyle = nsIFrame::ExtremumLength::MinContent;
+ break;
+ }
+ }
+
+ NS_ASSERTION(aStyle == nsIFrame::ExtremumLength::MinContent ||
+ aStyle == nsIFrame::ExtremumLength::MaxContent,
+ "should have reduced everything remaining to one of these");
+
+ // If aFrame is a container for font size inflation, then shrink
+ // wrapping inside of it should not apply font size inflation.
+ AutoMaybeDisableFontInflation an(aFrame);
+
+ if (aInlineSizeFromAspectRatio) {
+ aResult = *aInlineSizeFromAspectRatio;
+ } else if (aStyle == nsIFrame::ExtremumLength::MaxContent) {
+ aResult = aFrame->GetPrefISize(aRenderingContext);
+ } else {
+ aResult = aFrame->GetMinISize(aRenderingContext);
+ }
+
+ aResult += aContentBoxToBoxSizingDiff;
+
+ return true;
+}
+
+template <typename SizeOrMaxSize>
+static bool GetIntrinsicCoord(const SizeOrMaxSize& aStyle,
+ gfxContext* aRenderingContext, nsIFrame* aFrame,
+ Maybe<nscoord> aInlineSizeFromAspectRatio,
+ nsIFrame::SizeProperty aProperty,
+ nscoord aContentBoxToBoxSizingDiff,
+ nscoord& aResult) {
+ auto length = nsIFrame::ToExtremumLength(aStyle);
+ if (!length) {
+ return false;
+ }
+ return GetIntrinsicCoord(*length, aRenderingContext, aFrame,
+ aInlineSizeFromAspectRatio, aProperty,
+ aContentBoxToBoxSizingDiff, aResult);
+}
+
+#undef DEBUG_INTRINSIC_WIDTH
+
+#ifdef DEBUG_INTRINSIC_WIDTH
+static int32_t gNoiseIndent = 0;
+#endif
+
+static nscoord GetFitContentSizeForMaxOrPreferredSize(
+ const IntrinsicISizeType aType, const nsIFrame::SizeProperty aProperty,
+ const nsIFrame* aFrame, const LengthPercentage& aStyleSize,
+ const nscoord aInitialValue, const nscoord aMinContentSize,
+ const nscoord aMaxContentSize) {
+ MOZ_ASSERT(aProperty != nsIFrame::SizeProperty::MinSize);
+
+ nscoord size = NS_UNCONSTRAINEDSIZE;
+ // 1. Treat fit-content()'s arg as a plain LengthPercentage
+ // However, we have to handle the cyclic percentage contribution first.
+ //
+ // https://drafts.csswg.org/css-sizing-3/#cyclic-percentage-contribution
+ if (aType == IntrinsicISizeType::MinISize &&
+ aFrame->IsPercentageResolvedAgainstZero(aStyleSize, aProperty)) {
+ // Case (c) in the spec.
+ // FIXME: This doesn't follow the spec for calc(). We should fix this in
+ // Bug 1463700.
+ size = 0;
+ } else if (!GetAbsoluteCoord(aStyleSize, size)) {
+ // As initial value. Case (a) and (b) in the spec.
+ size = aInitialValue;
+ }
+
+ // 2. Clamp size by min-content and max-content.
+ return std::max(aMinContentSize, std::min(aMaxContentSize, size));
+}
+
+/**
+ * Add aOffsets which describes what to add on outside of the content box
+ * aContentSize (controlled by 'box-sizing') and apply min/max properties.
+ * We have to account for these properties after getting all the offsets
+ * (margin, border, padding) because percentages do not operate linearly.
+ * Doing this is ok because although percentages aren't handled linearly,
+ * they are handled monotonically.
+ *
+ * @param aContentSize the content size calculated so far
+ (@see IntrinsicForContainer)
+ * @param aContentMinSize ditto min content size
+ * @param aStyleSize a 'width' or 'height' property value
+ * @param aFixedMinSize if aStyleMinSize is a definite size then this points to
+ * the value, otherwise nullptr
+ * @param aStyleMinSize a 'min-width' or 'min-height' property value
+ * @param aFixedMaxSize if aStyleMaxSize is a definite size then this points to
+ * the value, otherwise nullptr
+ * @param aStyleMaxSize a 'max-width' or 'max-height' property value
+ * @param aInlineSizeFromAspectRatio the content-box inline size computed from
+ * aspect-ratio and the definite block size.
+ * We use this value to resolve
+ * {min|max}-content.
+ * @param aFlags same as for IntrinsicForContainer
+ * @param aContainerWM the container's WM
+ */
+static nscoord AddIntrinsicSizeOffset(
+ gfxContext* aRenderingContext, nsIFrame* aFrame,
+ const nsIFrame::IntrinsicSizeOffsetData& aOffsets, IntrinsicISizeType aType,
+ StyleBoxSizing aBoxSizing, nscoord aContentSize, nscoord aContentMinSize,
+ const StyleSize& aStyleSize, const nscoord* aFixedMinSize,
+ const StyleSize& aStyleMinSize, const nscoord* aFixedMaxSize,
+ const StyleMaxSize& aStyleMaxSize,
+ Maybe<nscoord> aInlineSizeFromAspectRatio, uint32_t aFlags,
+ PhysicalAxis aAxis) {
+ nscoord result = aContentSize;
+ nscoord min = aContentMinSize;
+ nscoord coordOutsideSize = 0;
+ nscoord contentBoxToBoxSizingDiff = 0;
+
+ if (!(aFlags & nsLayoutUtils::IGNORE_PADDING)) {
+ coordOutsideSize += aOffsets.padding;
+ }
+
+ coordOutsideSize += aOffsets.border;
+
+ if (aBoxSizing == StyleBoxSizing::Border) {
+ // Store the padding of the box so we can pass it into
+ // functions that works with content box-sizing.
+ contentBoxToBoxSizingDiff = coordOutsideSize;
+
+ min += coordOutsideSize;
+ result = NSCoordSaturatingAdd(result, coordOutsideSize);
+
+ coordOutsideSize = 0;
+ }
+
+ coordOutsideSize += aOffsets.margin;
+
+ min += coordOutsideSize;
+
+ // Compute min-content/max-content for fit-content().
+ nscoord minContent = 0;
+ nscoord maxContent = NS_UNCONSTRAINEDSIZE;
+ if (aStyleSize.IsFitContentFunction() ||
+ aStyleMaxSize.IsFitContentFunction() ||
+ aStyleMinSize.IsFitContentFunction()) {
+ if (aInlineSizeFromAspectRatio) {
+ minContent = maxContent = *aInlineSizeFromAspectRatio;
+ } else {
+ minContent = aFrame->GetMinISize(aRenderingContext);
+ maxContent = aFrame->GetPrefISize(aRenderingContext);
+ }
+ minContent += contentBoxToBoxSizingDiff;
+ maxContent += contentBoxToBoxSizingDiff;
+ }
+
+ // Compute size.
+ nscoord size = NS_UNCONSTRAINEDSIZE;
+ if (aType == IntrinsicISizeType::MinISize &&
+ aFrame->IsPercentageResolvedAgainstZero(aStyleSize, aStyleMaxSize)) {
+ // XXX bug 1463700: this doesn't handle calc() according to spec
+ result = 0; // let |min| handle padding/border/margin
+ } else if (GetAbsoluteCoord(aStyleSize, size) ||
+ GetIntrinsicCoord(aStyleSize, aRenderingContext, aFrame,
+ aInlineSizeFromAspectRatio,
+ nsIFrame::SizeProperty::Size,
+ contentBoxToBoxSizingDiff, size)) {
+ result = size + coordOutsideSize;
+ } else if (aStyleSize.IsFitContentFunction()) {
+ // |result| here is the content size or border size, depends on
+ // StyleBoxSizing. We use it as the initial value when handling the cyclic
+ // percentage.
+ nscoord initial = result;
+ nscoord fitContentFuncSize = GetFitContentSizeForMaxOrPreferredSize(
+ aType, nsIFrame::SizeProperty::Size, aFrame,
+ aStyleSize.AsFitContentFunction(), initial, minContent, maxContent);
+ // Add border and padding.
+ result = NSCoordSaturatingAdd(fitContentFuncSize, coordOutsideSize);
+ } else {
+ result = NSCoordSaturatingAdd(result, coordOutsideSize);
+ }
+
+ // Compute max-size.
+ nscoord maxSize = aFixedMaxSize ? *aFixedMaxSize : 0;
+ if (aFixedMaxSize || GetIntrinsicCoord(aStyleMaxSize, aRenderingContext,
+ aFrame, aInlineSizeFromAspectRatio,
+ nsIFrame::SizeProperty::MaxSize,
+ contentBoxToBoxSizingDiff, maxSize)) {
+ maxSize += coordOutsideSize;
+ if (result > maxSize) {
+ result = maxSize;
+ }
+ } else if (aStyleMaxSize.IsFitContentFunction()) {
+ nscoord fitContentFuncSize = GetFitContentSizeForMaxOrPreferredSize(
+ aType, nsIFrame::SizeProperty::MaxSize, aFrame,
+ aStyleMaxSize.AsFitContentFunction(), NS_UNCONSTRAINEDSIZE, minContent,
+ maxContent);
+ maxSize = NSCoordSaturatingAdd(fitContentFuncSize, coordOutsideSize);
+ if (result > maxSize) {
+ result = maxSize;
+ }
+ }
+
+ // Compute min-size.
+ nscoord minSize = aFixedMinSize ? *aFixedMinSize : 0;
+ if (aFixedMinSize || GetIntrinsicCoord(aStyleMinSize, aRenderingContext,
+ aFrame, aInlineSizeFromAspectRatio,
+ nsIFrame::SizeProperty::MinSize,
+ contentBoxToBoxSizingDiff, minSize)) {
+ minSize += coordOutsideSize;
+ if (result < minSize) {
+ result = minSize;
+ }
+ } else if (aStyleMinSize.IsFitContentFunction()) {
+ if (!GetAbsoluteCoord(aStyleMinSize.AsFitContentFunction(), minSize)) {
+ // FIXME: Bug 1463700, we should resolve only the percentage part to 0
+ // such as min-width: fit-content(calc(50% + 50px)).
+ minSize = 0;
+ }
+ nscoord fitContentFuncSize =
+ std::max(minContent, std::min(maxContent, minSize));
+ minSize = NSCoordSaturatingAdd(fitContentFuncSize, coordOutsideSize);
+ if (result < minSize) {
+ result = minSize;
+ }
+ }
+
+ if (result < min) {
+ result = min;
+ }
+
+ const nsStyleDisplay* disp = aFrame->StyleDisplay();
+ if (aFrame->IsThemed(disp)) {
+ nsPresContext* pc = aFrame->PresContext();
+ LayoutDeviceIntSize devSize = pc->Theme()->GetMinimumWidgetSize(
+ pc, aFrame, disp->EffectiveAppearance());
+ nscoord themeSize = pc->DevPixelsToAppUnits(
+ aAxis == eAxisVertical ? devSize.height : devSize.width);
+ // GetMinimumWidgetSize() returns a border-box width.
+ themeSize += aOffsets.margin;
+ if (themeSize > result) {
+ result = themeSize;
+ }
+ }
+ return result;
+}
+
+static void AddStateBitToAncestors(nsIFrame* aFrame, nsFrameState aBit) {
+ for (nsIFrame* f = aFrame; f; f = f->GetParent()) {
+ if (f->HasAnyStateBits(aBit)) {
+ break;
+ }
+ f->AddStateBits(aBit);
+ }
+}
+
+/* static */
+nscoord nsLayoutUtils::IntrinsicForAxis(
+ PhysicalAxis aAxis, gfxContext* aRenderingContext, nsIFrame* aFrame,
+ IntrinsicISizeType aType, const Maybe<LogicalSize>& aPercentageBasis,
+ uint32_t aFlags, nscoord aMarginBoxMinSizeClamp) {
+ MOZ_ASSERT(aFrame, "null frame");
+ MOZ_ASSERT(aFrame->GetParent(),
+ "IntrinsicForAxis called on frame not in tree");
+ MOZ_ASSERT(aFrame->GetParent()->Type() != LayoutFrameType::GridContainer ||
+ aPercentageBasis.isSome(),
+ "grid layout should always pass a percentage basis");
+
+ const bool horizontalAxis = MOZ_LIKELY(aAxis == eAxisHorizontal);
+#ifdef DEBUG_INTRINSIC_WIDTH
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s %s intrinsic size for container:\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref",
+ horizontalAxis ? "horizontal" : "vertical");
+#endif
+
+ // If aFrame is a container for font size inflation, then shrink
+ // wrapping inside of it should not apply font size inflation.
+ AutoMaybeDisableFontInflation an(aFrame);
+
+ // We want the size this frame will contribute to the parent's inline-size,
+ // so we work in the parent's writing mode; but if aFrame is orthogonal to
+ // its parent, we'll need to look at its BSize instead of min/pref-ISize.
+ const nsStylePosition* stylePos = aFrame->StylePosition();
+ StyleBoxSizing boxSizing = stylePos->mBoxSizing;
+
+ StyleSize styleMinISize =
+ horizontalAxis ? stylePos->mMinWidth : stylePos->mMinHeight;
+ StyleSize styleISize =
+ (aFlags & MIN_INTRINSIC_ISIZE)
+ ? styleMinISize
+ : (horizontalAxis ? stylePos->mWidth : stylePos->mHeight);
+ MOZ_ASSERT(!(aFlags & MIN_INTRINSIC_ISIZE) || styleISize.IsAuto() ||
+ nsIFrame::ToExtremumLength(styleISize),
+ "should only use MIN_INTRINSIC_ISIZE for intrinsic values");
+ StyleMaxSize styleMaxISize =
+ horizontalAxis ? stylePos->mMaxWidth : stylePos->mMaxHeight;
+
+ PhysicalAxis ourInlineAxis =
+ aFrame->GetWritingMode().PhysicalAxis(eLogicalAxisInline);
+ const bool isInlineAxis = aAxis == ourInlineAxis;
+
+ auto resetIfKeywords = [](StyleSize& aSize, StyleSize& aMinSize,
+ StyleMaxSize& aMaxSize) {
+ if (!aSize.IsLengthPercentage()) {
+ aSize = StyleSize::Auto();
+ }
+ if (!aMinSize.IsLengthPercentage()) {
+ aMinSize = StyleSize::Auto();
+ }
+ if (!aMaxSize.IsLengthPercentage()) {
+ aMaxSize = StyleMaxSize::None();
+ }
+ };
+ // According to the spec, max-content and min-content should behave as the
+ // property's initial values in block axis.
+ // It also make senses to use the initial values for -moz-fit-content and
+ // -moz-available for intrinsic size in block axis. Therefore, we reset them
+ // if needed.
+ if (!isInlineAxis) {
+ resetIfKeywords(styleISize, styleMinISize, styleMaxISize);
+ }
+
+ // We build up two values starting with the content box, and then
+ // adding padding, border and margin. The result is normally
+ // |result|. Then, when we handle 'width', 'min-width', and
+ // 'max-width', we use the results we've been building in |min| as a
+ // minimum, overriding 'min-width'. This ensures two things:
+ // * that we don't let a value of 'box-sizing' specifying a width
+ // smaller than the padding/border inside the box-sizing box give
+ // a content width less than zero
+ // * that we prevent tables from becoming smaller than their
+ // intrinsic minimum width
+ nscoord result = 0, min = 0;
+
+ nscoord maxISize;
+ bool haveFixedMaxISize = GetAbsoluteCoord(styleMaxISize, maxISize);
+ nscoord minISize;
+
+ // Treat "min-width: auto" as 0.
+ bool haveFixedMinISize;
+ if (styleMinISize.IsAuto()) {
+ // NOTE: Technically, "auto" is supposed to behave like "min-content" on
+ // flex items. However, we don't need to worry about that here, because
+ // flex items' min-sizes are intentionally ignored until the flex
+ // container explicitly considers them during space distribution.
+ minISize = 0;
+ haveFixedMinISize = true;
+ } else {
+ haveFixedMinISize = GetAbsoluteCoord(styleMinISize, minISize);
+ }
+
+ auto childWM = aFrame->GetWritingMode();
+ nscoord pmPercentageBasis = NS_UNCONSTRAINEDSIZE;
+ if (aPercentageBasis.isSome()) {
+ // The padding/margin percentage basis is the inline-size in the parent's
+ // writing-mode.
+ pmPercentageBasis =
+ aFrame->GetParent()->GetWritingMode().IsOrthogonalTo(childWM)
+ ? aPercentageBasis->BSize(childWM)
+ : aPercentageBasis->ISize(childWM);
+ }
+ nsIFrame::IntrinsicSizeOffsetData offsets =
+ MOZ_LIKELY(isInlineAxis)
+ ? aFrame->IntrinsicISizeOffsets(pmPercentageBasis)
+ : aFrame->IntrinsicBSizeOffsets(pmPercentageBasis);
+
+ auto getContentBoxSizeToBoxSizingAdjust =
+ [childWM, &offsets, &aFrame, isInlineAxis,
+ pmPercentageBasis](const StyleBoxSizing aBoxSizing) {
+ return aBoxSizing == StyleBoxSizing::Border
+ ? LogicalSize(childWM,
+ (isInlineAxis ? offsets
+ : aFrame->IntrinsicISizeOffsets(
+ pmPercentageBasis))
+ .BorderPadding(),
+ (!isInlineAxis ? offsets
+ : aFrame->IntrinsicBSizeOffsets(
+ pmPercentageBasis))
+ .BorderPadding())
+ : LogicalSize(childWM);
+ };
+
+ Maybe<nscoord> inlineSizeFromAspectRatio;
+ Maybe<LogicalSize> contentBoxSizeToBoxSizingAdjust;
+
+ const bool ignorePadding =
+ (aFlags & IGNORE_PADDING) || aFrame->IsAbsolutelyPositioned();
+
+ // If we have a specified width (or a specified 'min-width' greater
+ // than the specified 'max-width', which works out to the same thing),
+ // don't even bother getting the frame's intrinsic width, because in
+ // this case GetAbsoluteCoord(styleISize, w) will always succeed, so
+ // we'll never need the intrinsic dimensions.
+ if (styleISize.IsMaxContent() || styleISize.IsMinContent()) {
+ MOZ_ASSERT(isInlineAxis);
+ // -moz-fit-content and -moz-available enumerated widths compute intrinsic
+ // widths just like auto.
+ // For max-content and min-content, we handle them like
+ // specified widths, but ignore box-sizing.
+ boxSizing = StyleBoxSizing::Content;
+ } else if (!styleISize.ConvertsToLength() &&
+ !(styleISize.IsFitContentFunction() &&
+ styleISize.AsFitContentFunction().ConvertsToLength()) &&
+ !(haveFixedMinISize && haveFixedMaxISize &&
+ maxISize <= minISize)) {
+#ifdef DEBUG_INTRINSIC_WIDTH
+ ++gNoiseIndent;
+#endif
+ if (MOZ_UNLIKELY(!isInlineAxis)) {
+ IntrinsicSize intrinsicSize = aFrame->GetIntrinsicSize();
+ const auto& intrinsicBSize =
+ horizontalAxis ? intrinsicSize.width : intrinsicSize.height;
+ if (intrinsicBSize) {
+ result = *intrinsicBSize;
+ } else {
+ // We don't have an intrinsic bsize and we need aFrame's block-dir size.
+ if (aFlags & BAIL_IF_REFLOW_NEEDED) {
+ return NS_INTRINSIC_ISIZE_UNKNOWN;
+ }
+ // XXX Unfortunately, we probably don't know this yet, so this is
+ // wrong... but it's not clear what we should do. If aFrame's inline
+ // size hasn't been determined yet, we can't necessarily figure out its
+ // block size either. For now, authors who put orthogonal elements into
+ // things like buttons or table cells may have to explicitly provide
+ // sizes rather than expecting intrinsic sizing to work "perfectly" in
+ // underspecified cases.
+ result = aFrame->BSize();
+ }
+ } else {
+ result = aType == IntrinsicISizeType::MinISize
+ ? aFrame->GetMinISize(aRenderingContext)
+ : aFrame->GetPrefISize(aRenderingContext);
+ }
+#ifdef DEBUG_INTRINSIC_WIDTH
+ --gNoiseIndent;
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s %s intrinsic size from frame is %d.\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref",
+ horizontalAxis ? "horizontal" : "vertical", result);
+#endif
+
+ // Handle elements with an intrinsic ratio (or size) and a specified
+ // height, min-height, or max-height.
+ // NOTE:
+ // 1. We treat "min-height:auto" as "0" for the purpose of this code,
+ // since that's what it means in all cases except for on flex items -- and
+ // even there, we're supposed to ignore it (i.e. treat it as 0) until the
+ // flex container explicitly considers it.
+ // 2. The 'B' in |styleBSize|, |styleMinBSize|, and |styleMaxBSize|
+ // represents the ratio-determining axis of |aFrame|. It could be the inline
+ // axis or the block axis of |aFrame|. (So we are calculating the size
+ // along the ratio-dependent axis in this if-branch.)
+ StyleSize styleBSize =
+ horizontalAxis ? stylePos->mHeight : stylePos->mWidth;
+ StyleSize styleMinBSize =
+ horizontalAxis ? stylePos->mMinHeight : stylePos->mMinWidth;
+ StyleMaxSize styleMaxBSize =
+ horizontalAxis ? stylePos->mMaxHeight : stylePos->mMaxWidth;
+
+ // According to the spec, max-content and min-content should behave as the
+ // property's initial values in block axis.
+ // It also make senses to use the initial values for -moz-fit-content and
+ // -moz-available for intrinsic size in block axis. Therefore, we reset them
+ // if needed.
+ if (isInlineAxis) {
+ resetIfKeywords(styleBSize, styleMinBSize, styleMaxBSize);
+ }
+
+ // If our BSize or min/max-BSize properties are set to values that we can
+ // resolve and that will impose a constraint when transferred through our
+ // aspect ratio (if we have one), then compute and apply that constraint.
+ //
+ // (Note: This helper-bool & lambda just let us optimize away the actual
+ // transferring-and-clamping arithmetic, for the common case where we can
+ // tell that none of the block-axis size properties establish a meaningful
+ // transferred constraint.)
+ const bool mightHaveBlockAxisConstraintToTransfer = [&] {
+ if (!styleBSize.BehavesLikeInitialValueOnBlockAxis()) {
+ return true; // BSize property might have a constraint to transfer.
+ }
+ // Check for min-BSize values that would obviously produce zero in the
+ // transferring logic that follows; zero is trivially-ignorable as a
+ // transferred lower-bound. (These include the the property's initial
+ // value, explicit 0, and values that are equivalent to these.)
+ bool minBSizeHasNoConstraintToTransfer =
+ styleMinBSize.BehavesLikeInitialValueOnBlockAxis() ||
+ (styleMinBSize.IsLengthPercentage() &&
+ styleMinBSize.AsLengthPercentage().IsDefinitelyZero());
+ if (!minBSizeHasNoConstraintToTransfer) {
+ return true; // min-BSize property might have a constraint to transfer.
+ }
+ if (!styleMaxBSize.BehavesLikeInitialValueOnBlockAxis()) {
+ return true; // max-BSize property might have a constraint to transfer.
+ }
+ return false;
+ }();
+ if (mightHaveBlockAxisConstraintToTransfer) {
+ if (AspectRatio ratio = aFrame->GetAspectRatio()) {
+ AddStateBitToAncestors(
+ aFrame, NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE);
+
+ nscoord bSizeTakenByBoxSizing = GetDefiniteSizeTakenByBoxSizing(
+ boxSizing, aFrame, !isInlineAxis, ignorePadding, aPercentageBasis);
+ contentBoxSizeToBoxSizingAdjust.emplace(
+ getContentBoxSizeToBoxSizingAdjust(boxSizing));
+ // NOTE: This is only the minContentSize if we've been passed
+ // MIN_INTRINSIC_ISIZE (which is fine, because this should only be used
+ // inside a check for that flag).
+ nscoord minContentSize = result;
+ nscoord h;
+ if (GetDefiniteSize(styleBSize, aFrame, !isInlineAxis, aPercentageBasis,
+ &h) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(styleBSize, aFrame, horizontalAxis, h))) {
+ h = std::max(0, h - bSizeTakenByBoxSizing);
+ // We are computing the size of |aFrame|, so we use the inline & block
+ // dimensions of |aFrame|.
+ result = ratio.ComputeRatioDependentSize(
+ isInlineAxis ? eLogicalAxisInline : eLogicalAxisBlock, childWM, h,
+ *contentBoxSizeToBoxSizingAdjust);
+ // We have get the inlineSizeForAspectRatio value, so we don't have to
+ // recompute this again below.
+ inlineSizeFromAspectRatio.emplace(result);
+ }
+
+ if (GetDefiniteSize(styleMaxBSize, aFrame, !isInlineAxis,
+ aPercentageBasis, &h) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(styleMaxBSize, aFrame, horizontalAxis, h))) {
+ h = std::max(0, h - bSizeTakenByBoxSizing);
+ nscoord maxISize = ratio.ComputeRatioDependentSize(
+ isInlineAxis ? eLogicalAxisInline : eLogicalAxisBlock, childWM, h,
+ *contentBoxSizeToBoxSizingAdjust);
+ if (maxISize < result) {
+ result = maxISize;
+ }
+ if (maxISize < minContentSize) {
+ minContentSize = maxISize;
+ }
+ }
+
+ if (GetDefiniteSize(styleMinBSize, aFrame, !isInlineAxis,
+ aPercentageBasis, &h) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(styleMinBSize, aFrame, horizontalAxis, h))) {
+ h = std::max(0, h - bSizeTakenByBoxSizing);
+ nscoord minISize = ratio.ComputeRatioDependentSize(
+ isInlineAxis ? eLogicalAxisInline : eLogicalAxisBlock, childWM, h,
+ *contentBoxSizeToBoxSizingAdjust);
+ if (minISize > result) {
+ result = minISize;
+ }
+ if (minISize > minContentSize) {
+ minContentSize = minISize;
+ }
+ }
+
+ if (MOZ_UNLIKELY(aFlags & nsLayoutUtils::MIN_INTRINSIC_ISIZE) &&
+ // FIXME: Bug 1715681. Should we use eReplacedSizing instead
+ // because eReplaced is set on some other frames which are
+ // non-replaced elements, e.g. <select>?
+ aFrame->IsFrameOfType(nsIFrame::eReplaced)) {
+ // This is the 'min-width/height:auto' "transferred size" piece of:
+ // https://drafts.csswg.org/css-flexbox-1/#min-size-auto
+ // https://drafts.csswg.org/css-grid/#min-size-auto
+ // Per spec, we handle it only for replaced elements.
+ result = std::min(result, minContentSize);
+ }
+ }
+ }
+ }
+
+ if (aFrame->IsTableFrame()) {
+ // Tables can't shrink smaller than their intrinsic minimum width,
+ // no matter what.
+ min = aFrame->GetMinISize(aRenderingContext);
+ }
+
+ // If we have an aspect-ratio and a definite block size of |aFrame|, we
+ // resolve the {min|max}-content size by the aspect-ratio and the block size.
+ // If |aAxis| is not the inline axis of |aFrame|, {min|max}-content should
+ // behaves as auto, so we don't need this.
+ //
+ // FIXME(emilio): For -moz-available it seems we shouldn't need this.
+ //
+ // https://github.com/w3c/csswg-drafts/issues/5032
+ // FIXME: Bug 1670151: Use GetAspectRatio() to cover replaced elements (and
+ // then we can drop the check of eSupportsAspectRatio).
+ const AspectRatio ar = stylePos->mAspectRatio.ToLayoutRatio();
+ if (isInlineAxis && ar && nsIFrame::ToExtremumLength(styleISize) &&
+ aFrame->IsFrameOfType(nsIFrame::eSupportsAspectRatio) &&
+ !inlineSizeFromAspectRatio) {
+ // This 'B' in |styleBSize| means the block size of |aFrame|. We go into
+ // this branch only if |aAxis| is the inline axis of |aFrame|.
+ const StyleSize& styleBSize =
+ horizontalAxis ? stylePos->mHeight : stylePos->mWidth;
+ nscoord bSize;
+ if (GetDefiniteSize(styleBSize, aFrame, !isInlineAxis, aPercentageBasis,
+ &bSize) ||
+ (aPercentageBasis.isNothing() &&
+ GetPercentBSize(styleBSize, aFrame, horizontalAxis, bSize))) {
+ // We cannot reuse |boxSizing| because it may be updated to content-box
+ // in the above if-branch.
+ const StyleBoxSizing boxSizingForAR = stylePos->mBoxSizing;
+ if (!contentBoxSizeToBoxSizingAdjust) {
+ contentBoxSizeToBoxSizingAdjust.emplace(
+ getContentBoxSizeToBoxSizingAdjust(boxSizingForAR));
+ }
+ nscoord bSizeTakenByBoxSizing =
+ GetDefiniteSizeTakenByBoxSizing(boxSizingForAR, aFrame, !isInlineAxis,
+ ignorePadding, aPercentageBasis);
+ bSize -= bSizeTakenByBoxSizing;
+ inlineSizeFromAspectRatio.emplace(ar.ComputeRatioDependentSize(
+ LogicalAxis::eLogicalAxisInline, childWM, bSize,
+ *contentBoxSizeToBoxSizingAdjust));
+ }
+ }
+
+ nscoord contentBoxSize = result;
+ result = AddIntrinsicSizeOffset(
+ aRenderingContext, aFrame, offsets, aType, boxSizing, result, min,
+ styleISize, haveFixedMinISize ? &minISize : nullptr, styleMinISize,
+ haveFixedMaxISize ? &maxISize : nullptr, styleMaxISize,
+ inlineSizeFromAspectRatio, aFlags, aAxis);
+ nscoord overflow = result - aMarginBoxMinSizeClamp;
+ if (MOZ_UNLIKELY(overflow > 0)) {
+ nscoord newContentBoxSize = std::max(nscoord(0), contentBoxSize - overflow);
+ result -= contentBoxSize - newContentBoxSize;
+ }
+
+#ifdef DEBUG_INTRINSIC_WIDTH
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s %s intrinsic size for container is %d twips.\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref",
+ horizontalAxis ? "horizontal" : "vertical", result);
+#endif
+
+ return result;
+}
+
+/* static */
+nscoord nsLayoutUtils::IntrinsicForContainer(gfxContext* aRenderingContext,
+ nsIFrame* aFrame,
+ IntrinsicISizeType aType,
+ uint32_t aFlags) {
+ MOZ_ASSERT(aFrame && aFrame->GetParent());
+ // We want the size aFrame will contribute to its parent's inline-size.
+ PhysicalAxis axis =
+ aFrame->GetParent()->GetWritingMode().PhysicalAxis(eLogicalAxisInline);
+ return IntrinsicForAxis(axis, aRenderingContext, aFrame, aType, Nothing(),
+ aFlags);
+}
+
+/* static */
+nscoord nsLayoutUtils::MinSizeContributionForAxis(
+ PhysicalAxis aAxis, gfxContext* aRC, nsIFrame* aFrame,
+ IntrinsicISizeType aType, const LogicalSize& aPercentageBasis,
+ uint32_t aFlags) {
+ MOZ_ASSERT(aFrame);
+ MOZ_ASSERT(aFrame->IsFlexOrGridItem(),
+ "only grid/flex items have this behavior currently");
+
+#ifdef DEBUG_INTRINSIC_WIDTH
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s min-isize for %s WM:\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref",
+ aAxis == eAxisVertical ? "vertical" : "horizontal");
+#endif
+
+ // Note: this method is only meant for grid/flex items.
+ const nsStylePosition* const stylePos = aFrame->StylePosition();
+ StyleSize size =
+ aAxis == eAxisHorizontal ? stylePos->mMinWidth : stylePos->mMinHeight;
+ StyleMaxSize maxSize =
+ aAxis == eAxisHorizontal ? stylePos->mMaxWidth : stylePos->mMaxHeight;
+ auto childWM = aFrame->GetWritingMode();
+ PhysicalAxis ourInlineAxis = childWM.PhysicalAxis(eLogicalAxisInline);
+ // According to the spec, max-content and min-content should behave as the
+ // property's initial values in block axis.
+ // It also make senses to use the initial values for -moz-fit-content and
+ // -moz-available for intrinsic size in block axis. Therefore, we reset them
+ // if needed.
+ if (aAxis != ourInlineAxis) {
+ if (size.BehavesLikeInitialValueOnBlockAxis()) {
+ size = StyleSize::Auto();
+ }
+ if (maxSize.BehavesLikeInitialValueOnBlockAxis()) {
+ maxSize = StyleMaxSize::None();
+ }
+ }
+
+ nscoord minSize;
+ nscoord* fixedMinSize = nullptr;
+ if (size.IsAuto()) {
+ if (aFrame->StyleDisplay()->mOverflowX == StyleOverflow::Visible) {
+ size = aAxis == eAxisHorizontal ? stylePos->mWidth : stylePos->mHeight;
+ // This is same as above: keywords should behaves as property's initial
+ // values in block axis.
+ if (aAxis != ourInlineAxis && size.BehavesLikeInitialValueOnBlockAxis()) {
+ size = StyleSize::Auto();
+ }
+
+ if (GetAbsoluteCoord(size, minSize)) {
+ // We have a definite width/height. This is the "specified size" in:
+ // https://drafts.csswg.org/css-grid/#min-size-auto
+ fixedMinSize = &minSize;
+ } else if (aFrame->IsPercentageResolvedAgainstZero(size, maxSize)) {
+ // XXX bug 1463700: this doesn't handle calc() according to spec
+ minSize = 0;
+ fixedMinSize = &minSize;
+ }
+ // fall through - the caller will have to deal with "transferred size"
+ } else {
+ // min-[width|height]:auto with overflow != visible computes to zero.
+ minSize = 0;
+ fixedMinSize = &minSize;
+ }
+ } else if (GetAbsoluteCoord(size, minSize)) {
+ fixedMinSize = &minSize;
+ } else if (size.IsLengthPercentage()) {
+ MOZ_ASSERT(size.HasPercent());
+ minSize = 0;
+ fixedMinSize = &minSize;
+ }
+
+ if (!fixedMinSize) {
+ // Let the caller deal with the "content size" cases.
+#ifdef DEBUG_INTRINSIC_WIDTH
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s min-isize is indefinite.\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref");
+#endif
+ return NS_UNCONSTRAINEDSIZE;
+ }
+
+ // If aFrame is a container for font size inflation, then shrink
+ // wrapping inside of it should not apply font size inflation.
+ AutoMaybeDisableFontInflation an(aFrame);
+
+ // The padding/margin percentage basis is the inline-size in the parent's
+ // writing-mode.
+ nscoord pmPercentageBasis =
+ aFrame->GetParent()->GetWritingMode().IsOrthogonalTo(childWM)
+ ? aPercentageBasis.BSize(childWM)
+ : aPercentageBasis.ISize(childWM);
+ nsIFrame::IntrinsicSizeOffsetData offsets =
+ ourInlineAxis == aAxis ? aFrame->IntrinsicISizeOffsets(pmPercentageBasis)
+ : aFrame->IntrinsicBSizeOffsets(pmPercentageBasis);
+ nscoord result = 0;
+ nscoord min = 0;
+ // Note: aInlineSizeFromAspectRatio is Nothing() here because we don't handle
+ // "content size" cases here (i.e. |fixedMinSize| is false here).
+ result = AddIntrinsicSizeOffset(
+ aRC, aFrame, offsets, aType, stylePos->mBoxSizing, result, min, size,
+ fixedMinSize, size, nullptr, maxSize, Nothing(), aFlags, aAxis);
+
+#ifdef DEBUG_INTRINSIC_WIDTH
+ nsIFrame::IndentBy(stderr, gNoiseIndent);
+ aFrame->ListTag(stderr);
+ printf_stderr(" %s min-isize is %d twips.\n",
+ aType == IntrinsicISizeType::MinISize ? "min" : "pref", result);
+#endif
+
+ return result;
+}
+
+/* static */
+nscoord nsLayoutUtils::ComputeBSizeDependentValue(
+ nscoord aContainingBlockBSize, const LengthPercentageOrAuto& aCoord) {
+ // XXXldb Some callers explicitly check aContainingBlockBSize
+ // against NS_UNCONSTRAINEDSIZE *and* unit against eStyleUnit_Percent or
+ // calc()s containing percents before calling this function.
+ // However, it would be much more likely to catch problems without
+ // the unit conditions.
+ // XXXldb Many callers pass a non-'auto' containing block height when
+ // according to CSS2.1 they should be passing 'auto'.
+ NS_ASSERTION(
+ NS_UNCONSTRAINEDSIZE != aContainingBlockBSize || !aCoord.HasPercent(),
+ "unexpected containing block block-size");
+
+ if (aCoord.IsAuto()) {
+ return 0;
+ }
+
+ return aCoord.AsLengthPercentage().Resolve(aContainingBlockBSize);
+}
+
+/* static */
+void nsLayoutUtils::MarkDescendantsDirty(nsIFrame* aSubtreeRoot) {
+ AutoTArray<nsIFrame*, 4> subtrees;
+ subtrees.AppendElement(aSubtreeRoot);
+
+ // dirty descendants, iterating over subtrees that may include
+ // additional subtrees associated with placeholders
+ do {
+ nsIFrame* subtreeRoot = subtrees.PopLastElement();
+
+ // Mark all descendants dirty (using an nsTArray stack rather than
+ // recursion).
+ // Note that ReflowInput::InitResizeFlags has some similar
+ // code; see comments there for how and why it differs.
+ AutoTArray<nsIFrame*, 32> stack;
+ stack.AppendElement(subtreeRoot);
+
+ do {
+ nsIFrame* f = stack.PopLastElement();
+
+ f->MarkIntrinsicISizesDirty();
+
+ if (f->IsPlaceholderFrame()) {
+ nsIFrame* oof = nsPlaceholderFrame::GetRealFrameForPlaceholder(f);
+ if (!nsLayoutUtils::IsProperAncestorFrame(subtreeRoot, oof)) {
+ // We have another distinct subtree we need to mark.
+ subtrees.AppendElement(oof);
+ }
+ }
+
+ for (const auto& childList : f->ChildLists()) {
+ for (nsIFrame* kid : childList.mList) {
+ stack.AppendElement(kid);
+ }
+ }
+ } while (stack.Length() != 0);
+ } while (subtrees.Length() != 0);
+}
+
+/* static */
+void nsLayoutUtils::MarkIntrinsicISizesDirtyIfDependentOnBSize(
+ nsIFrame* aFrame) {
+ AutoTArray<nsIFrame*, 32> stack;
+ stack.AppendElement(aFrame);
+
+ do {
+ nsIFrame* f = stack.PopLastElement();
+
+ if (!f->HasAnyStateBits(
+ NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE)) {
+ continue;
+ }
+ f->MarkIntrinsicISizesDirty();
+
+ for (const auto& childList : f->ChildLists()) {
+ for (nsIFrame* kid : childList.mList) {
+ stack.AppendElement(kid);
+ }
+ }
+ } while (stack.Length() != 0);
+}
+
+nsSize nsLayoutUtils::ComputeAutoSizeWithIntrinsicDimensions(
+ nscoord minWidth, nscoord minHeight, nscoord maxWidth, nscoord maxHeight,
+ nscoord tentWidth, nscoord tentHeight) {
+ // Now apply min/max-width/height - CSS 2.1 sections 10.4 and 10.7:
+
+ if (minWidth > maxWidth) maxWidth = minWidth;
+ if (minHeight > maxHeight) maxHeight = minHeight;
+
+ nscoord heightAtMaxWidth, heightAtMinWidth, widthAtMaxHeight,
+ widthAtMinHeight;
+
+ if (tentWidth > 0) {
+ heightAtMaxWidth = NSCoordMulDiv(maxWidth, tentHeight, tentWidth);
+ if (heightAtMaxWidth < minHeight) heightAtMaxWidth = minHeight;
+ heightAtMinWidth = NSCoordMulDiv(minWidth, tentHeight, tentWidth);
+ if (heightAtMinWidth > maxHeight) heightAtMinWidth = maxHeight;
+ } else {
+ heightAtMaxWidth = heightAtMinWidth =
+ NS_CSS_MINMAX(tentHeight, minHeight, maxHeight);
+ }
+
+ if (tentHeight > 0) {
+ widthAtMaxHeight = NSCoordMulDiv(maxHeight, tentWidth, tentHeight);
+ if (widthAtMaxHeight < minWidth) widthAtMaxHeight = minWidth;
+ widthAtMinHeight = NSCoordMulDiv(minHeight, tentWidth, tentHeight);
+ if (widthAtMinHeight > maxWidth) widthAtMinHeight = maxWidth;
+ } else {
+ widthAtMaxHeight = widthAtMinHeight =
+ NS_CSS_MINMAX(tentWidth, minWidth, maxWidth);
+ }
+
+ // The table at http://www.w3.org/TR/CSS21/visudet.html#min-max-widths :
+
+ nscoord width, height;
+
+ if (tentWidth > maxWidth) {
+ if (tentHeight > maxHeight) {
+ if (int64_t(maxWidth) * int64_t(tentHeight) <=
+ int64_t(maxHeight) * int64_t(tentWidth)) {
+ width = maxWidth;
+ height = heightAtMaxWidth;
+ } else {
+ width = widthAtMaxHeight;
+ height = maxHeight;
+ }
+ } else {
+ // This also covers "(w > max-width) and (h < min-height)" since in
+ // that case (max-width/w < 1), and with (h < min-height):
+ // max(max-width * h/w, min-height) == min-height
+ width = maxWidth;
+ height = heightAtMaxWidth;
+ }
+ } else if (tentWidth < minWidth) {
+ if (tentHeight < minHeight) {
+ if (int64_t(minWidth) * int64_t(tentHeight) <=
+ int64_t(minHeight) * int64_t(tentWidth)) {
+ width = widthAtMinHeight;
+ height = minHeight;
+ } else {
+ width = minWidth;
+ height = heightAtMinWidth;
+ }
+ } else {
+ // This also covers "(w < min-width) and (h > max-height)" since in
+ // that case (min-width/w > 1), and with (h > max-height):
+ // min(min-width * h/w, max-height) == max-height
+ width = minWidth;
+ height = heightAtMinWidth;
+ }
+ } else {
+ if (tentHeight > maxHeight) {
+ width = widthAtMaxHeight;
+ height = maxHeight;
+ } else if (tentHeight < minHeight) {
+ width = widthAtMinHeight;
+ height = minHeight;
+ } else {
+ width = tentWidth;
+ height = tentHeight;
+ }
+ }
+
+ return nsSize(width, height);
+}
+
+/* static */
+nscoord nsLayoutUtils::MinISizeFromInline(nsIFrame* aFrame,
+ gfxContext* aRenderingContext) {
+ NS_ASSERTION(!aFrame->IsContainerForFontSizeInflation(),
+ "should not be container for font size inflation");
+
+ nsIFrame::InlineMinISizeData data;
+ DISPLAY_MIN_INLINE_SIZE(aFrame, data.mPrevLines);
+ aFrame->AddInlineMinISize(aRenderingContext, &data);
+ data.ForceBreak();
+ return data.mPrevLines;
+}
+
+/* static */
+nscoord nsLayoutUtils::PrefISizeFromInline(nsIFrame* aFrame,
+ gfxContext* aRenderingContext) {
+ NS_ASSERTION(!aFrame->IsContainerForFontSizeInflation(),
+ "should not be container for font size inflation");
+
+ nsIFrame::InlinePrefISizeData data;
+ DISPLAY_PREF_INLINE_SIZE(aFrame, data.mPrevLines);
+ aFrame->AddInlinePrefISize(aRenderingContext, &data);
+ data.ForceBreak();
+ return data.mPrevLines;
+}
+
+static nscolor DarkenColor(nscolor aColor) {
+ uint16_t hue, sat, value;
+ uint8_t alpha;
+
+ // convert the RBG to HSV so we can get the lightness (which is the v)
+ NS_RGB2HSV(aColor, hue, sat, value, alpha);
+
+ // The goal here is to send white to black while letting colored
+ // stuff stay colored... So we adopt the following approach.
+ // Something with sat = 0 should end up with value = 0. Something
+ // with a high sat can end up with a high value and it's ok.... At
+ // the same time, we don't want to make things lighter. Do
+ // something simple, since it seems to work.
+ if (value > sat) {
+ value = sat;
+ // convert this color back into the RGB color space.
+ NS_HSV2RGB(aColor, hue, sat, value, alpha);
+ }
+ return aColor;
+}
+
+// Check whether we should darken text/decoration colors. We need to do this if
+// background images and colors are being suppressed, because that means
+// light text will not be visible against the (presumed light-colored)
+// background.
+static bool ShouldDarkenColors(nsIFrame* aFrame) {
+ nsPresContext* pc = aFrame->PresContext();
+ if (pc->GetBackgroundColorDraw() || pc->GetBackgroundImageDraw()) {
+ return false;
+ }
+ return aFrame->StyleVisibility()->mPrintColorAdjust !=
+ StylePrintColorAdjust::Exact;
+}
+
+nscolor nsLayoutUtils::DarkenColorIfNeeded(nsIFrame* aFrame, nscolor aColor) {
+ return ShouldDarkenColors(aFrame) ? DarkenColor(aColor) : aColor;
+}
+
+gfxFloat nsLayoutUtils::GetSnappedBaselineY(nsIFrame* aFrame,
+ gfxContext* aContext, nscoord aY,
+ nscoord aAscent) {
+ gfxFloat appUnitsPerDevUnit = aFrame->PresContext()->AppUnitsPerDevPixel();
+ gfxFloat baseline = gfxFloat(aY) + aAscent;
+ gfxRect putativeRect(0, baseline / appUnitsPerDevUnit, 1, 1);
+ if (!aContext->UserToDevicePixelSnapped(
+ putativeRect, gfxContext::SnapOption::IgnoreScale)) {
+ return baseline;
+ }
+ return aContext->DeviceToUser(putativeRect.TopLeft()).y * appUnitsPerDevUnit;
+}
+
+gfxFloat nsLayoutUtils::GetSnappedBaselineX(nsIFrame* aFrame,
+ gfxContext* aContext, nscoord aX,
+ nscoord aAscent) {
+ gfxFloat appUnitsPerDevUnit = aFrame->PresContext()->AppUnitsPerDevPixel();
+ gfxFloat baseline = gfxFloat(aX) + aAscent;
+ gfxRect putativeRect(baseline / appUnitsPerDevUnit, 0, 1, 1);
+ if (!aContext->UserToDevicePixelSnapped(
+ putativeRect, gfxContext::SnapOption::IgnoreScale)) {
+ return baseline;
+ }
+ return aContext->DeviceToUser(putativeRect.TopLeft()).x * appUnitsPerDevUnit;
+}
+
+// Hard limit substring lengths to 8000 characters ... this lets us statically
+// size the cluster buffer array in FindSafeLength
+#define MAX_GFX_TEXT_BUF_SIZE 8000
+
+static int32_t FindSafeLength(const char16_t* aString, uint32_t aLength,
+ uint32_t aMaxChunkLength) {
+ if (aLength <= aMaxChunkLength) return aLength;
+
+ int32_t len = aMaxChunkLength;
+
+ // Ensure that we don't break inside a surrogate pair
+ while (len > 0 && NS_IS_LOW_SURROGATE(aString[len])) {
+ len--;
+ }
+ if (len == 0) {
+ // We don't want our caller to go into an infinite loop, so don't
+ // return zero. It's hard to imagine how we could actually get here
+ // unless there are languages that allow clusters of arbitrary size.
+ // If there are and someone feeds us a 500+ character cluster, too
+ // bad.
+ return aMaxChunkLength;
+ }
+ return len;
+}
+
+static int32_t GetMaxChunkLength(nsFontMetrics& aFontMetrics) {
+ return std::min(aFontMetrics.GetMaxStringLength(), MAX_GFX_TEXT_BUF_SIZE);
+}
+
+nscoord nsLayoutUtils::AppUnitWidthOfString(const char16_t* aString,
+ uint32_t aLength,
+ nsFontMetrics& aFontMetrics,
+ DrawTarget* aDrawTarget) {
+ uint32_t maxChunkLength = GetMaxChunkLength(aFontMetrics);
+ nscoord width = 0;
+ while (aLength > 0) {
+ int32_t len = FindSafeLength(aString, aLength, maxChunkLength);
+ width += aFontMetrics.GetWidth(aString, len, aDrawTarget);
+ aLength -= len;
+ aString += len;
+ }
+ return width;
+}
+
+nscoord nsLayoutUtils::AppUnitWidthOfStringBidi(const char16_t* aString,
+ uint32_t aLength,
+ const nsIFrame* aFrame,
+ nsFontMetrics& aFontMetrics,
+ gfxContext& aContext) {
+ nsPresContext* presContext = aFrame->PresContext();
+ if (presContext->BidiEnabled()) {
+ mozilla::intl::BidiEmbeddingLevel level =
+ nsBidiPresUtils::BidiLevelFromStyle(aFrame->Style());
+ return nsBidiPresUtils::MeasureTextWidth(
+ aString, aLength, level, presContext, aContext, aFontMetrics);
+ }
+ aFontMetrics.SetTextRunRTL(false);
+ aFontMetrics.SetVertical(aFrame->GetWritingMode().IsVertical());
+ aFontMetrics.SetTextOrientation(aFrame->StyleVisibility()->mTextOrientation);
+ return nsLayoutUtils::AppUnitWidthOfString(aString, aLength, aFontMetrics,
+ aContext.GetDrawTarget());
+}
+
+bool nsLayoutUtils::StringWidthIsGreaterThan(const nsString& aString,
+ nsFontMetrics& aFontMetrics,
+ DrawTarget* aDrawTarget,
+ nscoord aWidth) {
+ const char16_t* string = aString.get();
+ uint32_t length = aString.Length();
+ uint32_t maxChunkLength = GetMaxChunkLength(aFontMetrics);
+ nscoord width = 0;
+ while (length > 0) {
+ int32_t len = FindSafeLength(string, length, maxChunkLength);
+ width += aFontMetrics.GetWidth(string, len, aDrawTarget);
+ if (width > aWidth) {
+ return true;
+ }
+ length -= len;
+ string += len;
+ }
+ return false;
+}
+
+nsBoundingMetrics nsLayoutUtils::AppUnitBoundsOfString(
+ const char16_t* aString, uint32_t aLength, nsFontMetrics& aFontMetrics,
+ DrawTarget* aDrawTarget) {
+ uint32_t maxChunkLength = GetMaxChunkLength(aFontMetrics);
+ int32_t len = FindSafeLength(aString, aLength, maxChunkLength);
+ // Assign directly in the first iteration. This ensures that
+ // negative ascent/descent can be returned and the left bearing
+ // is properly initialized.
+ nsBoundingMetrics totalMetrics =
+ aFontMetrics.GetBoundingMetrics(aString, len, aDrawTarget);
+ aLength -= len;
+ aString += len;
+
+ while (aLength > 0) {
+ len = FindSafeLength(aString, aLength, maxChunkLength);
+ nsBoundingMetrics metrics =
+ aFontMetrics.GetBoundingMetrics(aString, len, aDrawTarget);
+ totalMetrics += metrics;
+ aLength -= len;
+ aString += len;
+ }
+ return totalMetrics;
+}
+
+void nsLayoutUtils::DrawString(const nsIFrame* aFrame,
+ nsFontMetrics& aFontMetrics,
+ gfxContext* aContext, const char16_t* aString,
+ int32_t aLength, nsPoint aPoint,
+ ComputedStyle* aComputedStyle,
+ DrawStringFlags aFlags) {
+ nsresult rv = NS_ERROR_FAILURE;
+
+ // If caller didn't pass a style, use the frame's.
+ if (!aComputedStyle) {
+ aComputedStyle = aFrame->Style();
+ }
+
+ if (aFlags & DrawStringFlags::ForceHorizontal) {
+ aFontMetrics.SetVertical(false);
+ } else {
+ aFontMetrics.SetVertical(WritingMode(aComputedStyle).IsVertical());
+ }
+
+ aFontMetrics.SetTextOrientation(
+ aComputedStyle->StyleVisibility()->mTextOrientation);
+
+ nsPresContext* presContext = aFrame->PresContext();
+ if (presContext->BidiEnabled()) {
+ mozilla::intl::BidiEmbeddingLevel level =
+ nsBidiPresUtils::BidiLevelFromStyle(aComputedStyle);
+ rv = nsBidiPresUtils::RenderText(aString, aLength, level, presContext,
+ *aContext, aContext->GetDrawTarget(),
+ aFontMetrics, aPoint.x, aPoint.y);
+ }
+ if (NS_FAILED(rv)) {
+ aFontMetrics.SetTextRunRTL(false);
+ DrawUniDirString(aString, aLength, aPoint, aFontMetrics, *aContext);
+ }
+}
+
+void nsLayoutUtils::DrawUniDirString(const char16_t* aString, uint32_t aLength,
+ const nsPoint& aPoint,
+ nsFontMetrics& aFontMetrics,
+ gfxContext& aContext) {
+ nscoord x = aPoint.x;
+ nscoord y = aPoint.y;
+
+ uint32_t maxChunkLength = GetMaxChunkLength(aFontMetrics);
+ if (aLength <= maxChunkLength) {
+ aFontMetrics.DrawString(aString, aLength, x, y, &aContext,
+ aContext.GetDrawTarget());
+ return;
+ }
+
+ bool isRTL = aFontMetrics.GetTextRunRTL();
+
+ // If we're drawing right to left, we must start at the end.
+ if (isRTL) {
+ x += nsLayoutUtils::AppUnitWidthOfString(aString, aLength, aFontMetrics,
+ aContext.GetDrawTarget());
+ }
+
+ while (aLength > 0) {
+ int32_t len = FindSafeLength(aString, aLength, maxChunkLength);
+ nscoord width =
+ aFontMetrics.GetWidth(aString, len, aContext.GetDrawTarget());
+ if (isRTL) {
+ x -= width;
+ }
+ aFontMetrics.DrawString(aString, len, x, y, &aContext,
+ aContext.GetDrawTarget());
+ if (!isRTL) {
+ x += width;
+ }
+ aLength -= len;
+ aString += len;
+ }
+}
+
+/* static */
+void nsLayoutUtils::PaintTextShadow(
+ const nsIFrame* aFrame, gfxContext* aContext, const nsRect& aTextRect,
+ const nsRect& aDirtyRect, const nscolor& aForegroundColor,
+ TextShadowCallback aCallback, void* aCallbackData) {
+ const nsStyleText* textStyle = aFrame->StyleText();
+ auto shadows = textStyle->mTextShadow.AsSpan();
+ if (shadows.IsEmpty()) {
+ return;
+ }
+
+ // Text shadow happens with the last value being painted at the back,
+ // ie. it is painted first.
+ gfxContext* aDestCtx = aContext;
+ for (auto& shadow : Reversed(shadows)) {
+ nsPoint shadowOffset(shadow.horizontal.ToAppUnits(),
+ shadow.vertical.ToAppUnits());
+ nscoord blurRadius = std::max(shadow.blur.ToAppUnits(), 0);
+
+ nsRect shadowRect(aTextRect);
+ shadowRect.MoveBy(shadowOffset);
+
+ nsPresContext* presCtx = aFrame->PresContext();
+ nsContextBoxBlur contextBoxBlur;
+
+ nscolor shadowColor = shadow.color.CalcColor(aForegroundColor);
+
+ // Webrender just needs the shadow details
+ if (auto* textDrawer = aContext->GetTextDrawer()) {
+ wr::Shadow wrShadow;
+
+ wrShadow.offset = {
+ presCtx->AppUnitsToFloatDevPixels(shadow.horizontal.ToAppUnits()),
+ presCtx->AppUnitsToFloatDevPixels(shadow.vertical.ToAppUnits())};
+
+ wrShadow.blur_radius = presCtx->AppUnitsToFloatDevPixels(blurRadius);
+ wrShadow.color = wr::ToColorF(ToDeviceColor(shadowColor));
+
+ // Gecko already inflates the bounding rect of text shadows,
+ // so tell WR not to inflate again.
+ bool inflate = false;
+ textDrawer->AppendShadow(wrShadow, inflate);
+ continue;
+ }
+
+ gfxContext* shadowContext = contextBoxBlur.Init(
+ shadowRect, 0, blurRadius, presCtx->AppUnitsPerDevPixel(), aDestCtx,
+ aDirtyRect, nullptr,
+ nsContextBoxBlur::DISABLE_HARDWARE_ACCELERATION_BLUR);
+ if (!shadowContext) continue;
+
+ aDestCtx->Save();
+ aDestCtx->NewPath();
+ aDestCtx->SetColor(sRGBColor::FromABGR(shadowColor));
+
+ // The callback will draw whatever we want to blur as a shadow.
+ aCallback(shadowContext, shadowOffset, shadowColor, aCallbackData);
+
+ contextBoxBlur.DoPaint();
+ aDestCtx->Restore();
+ }
+}
+
+/* static */
+nscoord nsLayoutUtils::GetCenteredFontBaseline(nsFontMetrics* aFontMetrics,
+ nscoord aLineHeight,
+ bool aIsInverted) {
+ nscoord fontAscent =
+ aIsInverted ? aFontMetrics->MaxDescent() : aFontMetrics->MaxAscent();
+ nscoord fontHeight = aFontMetrics->MaxHeight();
+
+ nscoord leading = aLineHeight - fontHeight;
+ return fontAscent + leading / 2;
+}
+
+/* static */
+bool nsLayoutUtils::GetFirstLineBaseline(WritingMode aWritingMode,
+ const nsIFrame* aFrame,
+ nscoord* aResult) {
+ LinePosition position;
+ if (!GetFirstLinePosition(aWritingMode, aFrame, &position)) return false;
+ *aResult = position.mBaseline;
+ return true;
+}
+
+/* static */
+bool nsLayoutUtils::GetFirstLinePosition(WritingMode aWM,
+ const nsIFrame* aFrame,
+ LinePosition* aResult) {
+ if (aFrame->StyleDisplay()->IsContainLayout()) {
+ return false;
+ }
+ const nsBlockFrame* block = do_QueryFrame(aFrame);
+ if (!block) {
+ // For the first-line baseline we also have to check for a table, and if
+ // so, use the baseline of its first row.
+ LayoutFrameType fType = aFrame->Type();
+ if (fType == LayoutFrameType::TableWrapper ||
+ fType == LayoutFrameType::FlexContainer ||
+ fType == LayoutFrameType::GridContainer) {
+ if ((fType == LayoutFrameType::GridContainer &&
+ aFrame->HasAnyStateBits(NS_STATE_GRID_SYNTHESIZE_BASELINE)) ||
+ (fType == LayoutFrameType::FlexContainer &&
+ aFrame->HasAnyStateBits(NS_STATE_FLEX_SYNTHESIZE_BASELINE)) ||
+ (fType == LayoutFrameType::TableWrapper &&
+ static_cast<const nsTableWrapperFrame*>(aFrame)->GetRowCount() ==
+ 0)) {
+ // empty grid/flex/table container
+ aResult->mBStart = 0;
+ aResult->mBaseline = Baseline::SynthesizeBOffsetFromBorderBox(
+ aFrame, aWM, BaselineSharingGroup::First);
+ aResult->mBEnd = aFrame->BSize(aWM);
+ return true;
+ }
+ if (fType == LayoutFrameType::TableWrapper &&
+ aFrame->GetWritingMode().IsOrthogonalTo(aWM)) {
+ // For tables, the upcoming GetLogicalBaseline call would determine the
+ // table's baseline from its first row that has a baseline. However:
+ // this doesn't make sense for an orthogonal writing mode, so in that
+ // case we report no baseline instead. The table wrapper and its rows
+ // should flow the same way, so we can bail out early, but this logic
+ // wouldn't be correct to transplant into other places in the codebase
+ // (Depending on how bug 1786633 is resolved).
+ return false;
+ }
+ aResult->mBStart = 0;
+ aResult->mBaseline = aFrame->GetLogicalBaseline(aWM);
+ // This is what we want for the list bullet caller; not sure if
+ // other future callers will want the same.
+ aResult->mBEnd = aFrame->BSize(aWM);
+ return true;
+ }
+
+ // For first-line baselines, we have to consider scroll frames.
+ if (nsIScrollableFrame* sFrame =
+ do_QueryFrame(const_cast<nsIFrame*>(aFrame))) {
+ LinePosition kidPosition;
+ if (GetFirstLinePosition(aWM, sFrame->GetScrolledFrame(), &kidPosition)) {
+ // Consider only the border (Padding is ignored, since
+ // `-moz-scrolled-content` inherits and handles the padding) that
+ // contributes to the kid's position, not the scrolling, so we get the
+ // initial position.
+ *aResult = kidPosition + aFrame->GetLogicalUsedBorder(aWM).BStart(aWM);
+ // Don't want to move the line's block positioning, but the baseline
+ // needs to be clamped (See bug 1791069).
+ aResult->mBaseline = std::clamp(aResult->mBaseline, 0,
+ aFrame->GetLogicalSize(aWM).BSize(aWM));
+ return true;
+ }
+ return false;
+ }
+
+ if (fType == LayoutFrameType::FieldSet) {
+ LinePosition kidPosition;
+ // Get the first baseline from the fieldset content, not from the legend.
+ nsIFrame* kid = static_cast<const nsFieldSetFrame*>(aFrame)->GetInner();
+ if (kid && GetFirstLinePosition(aWM, kid, &kidPosition)) {
+ *aResult = kidPosition +
+ kid->GetLogicalNormalPosition(aWM, aFrame->GetSize()).B(aWM);
+ return true;
+ }
+ return false;
+ }
+
+ if (fType == LayoutFrameType::ColumnSet) {
+ // Note(dshin): This is basically the same as
+ // `nsColumnSetFrame::GetNaturalBaselineBOffset`, but with line start and
+ // end, all stored in `LinePosition`. Field value apart from baseline is
+ // used in one other place
+ // (`nsBlockFrame`) - if that goes away, this becomes a duplication that
+ // should be removed.
+ LinePosition kidPosition;
+ for (const auto* kid : aFrame->PrincipalChildList()) {
+ LinePosition position;
+ if (!GetFirstLinePosition(aWM, kid, &position)) {
+ continue;
+ }
+ if (position.mBaseline < kidPosition.mBaseline) {
+ kidPosition = position;
+ }
+ }
+ if (kidPosition.mBaseline != nscoord_MAX) {
+ *aResult = kidPosition;
+ return true;
+ }
+ }
+
+ // No baseline.
+ return false;
+ }
+
+ for (const auto& line : block->Lines()) {
+ if (line.IsBlock()) {
+ const nsIFrame* kid = line.mFirstChild;
+ LinePosition kidPosition;
+ if (GetFirstLinePosition(aWM, kid, &kidPosition)) {
+ // XXX Not sure if this is the correct value to use for container
+ // width here. It will only be used in vertical-rl layout,
+ // which we don't have full support and testing for yet.
+ const auto& containerSize = line.mContainerSize;
+ *aResult = kidPosition +
+ kid->GetLogicalNormalPosition(aWM, containerSize).B(aWM);
+ return true;
+ }
+ } else {
+ // XXX Is this the right test? We have some bogus empty lines
+ // floating around, but IsEmpty is perhaps too weak.
+ if (0 != line.BSize() || !line.IsEmpty()) {
+ nscoord bStart = line.BStart();
+ aResult->mBStart = bStart;
+ aResult->mBaseline = bStart + line.GetLogicalAscent();
+ aResult->mBEnd = bStart + line.BSize();
+ return true;
+ }
+ }
+ }
+ return false;
+}
+
+/* static */
+bool nsLayoutUtils::GetLastLineBaseline(WritingMode aWM, const nsIFrame* aFrame,
+ nscoord* aResult) {
+ if (aFrame->StyleDisplay()->IsContainLayout()) {
+ return false;
+ }
+
+ const nsBlockFrame* block = do_QueryFrame(aFrame);
+ if (!block) {
+ if (nsIScrollableFrame* sFrame = do_QueryFrame(aFrame)) {
+ // Use the baseline position only if the last line's baseline is within
+ // the scrolling frame's box in the initial position.
+ const auto* scrolledFrame = sFrame->GetScrolledFrame();
+ if (!GetLastLineBaseline(aWM, scrolledFrame, aResult)) {
+ return false;
+ }
+ // Go from scrolled frame to scrollable frame position.
+ *aResult += aFrame->GetLogicalUsedBorder(aWM).BStart(aWM);
+ const auto maxBaseline = aFrame->GetLogicalSize(aWM).BSize(aWM);
+ // Clamp the last baseline to border (See bug 1791069).
+ *aResult = std::clamp(*aResult, 0, maxBaseline);
+ return true;
+ }
+
+ // No need to duplicate the baseline logic (Unlike `GetFirstLinePosition`,
+ // we don't need to return any other value apart from baseline), just defer
+ // to `GetNaturalBaselineBOffset`. Technically, we could do this at
+ // `ColumnSetWrapperFrame` level, but this keeps it symmetric to
+ // `GetFirstLinePosition`.
+ if (aFrame->IsColumnSetFrame()) {
+ const auto baseline = aFrame->GetNaturalBaselineBOffset(
+ aWM, BaselineSharingGroup::Last, BaselineExportContext::Other);
+ if (!baseline) {
+ return false;
+ }
+ *aResult = aFrame->BSize(aWM) - *baseline;
+ return true;
+ }
+ // No baseline.
+ return false;
+ }
+
+ for (nsBlockFrame::ConstReverseLineIterator line = block->LinesRBegin(),
+ line_end = block->LinesREnd();
+ line != line_end; ++line) {
+ if (line->IsBlock()) {
+ nsIFrame* kid = line->mFirstChild;
+ nscoord kidBaseline;
+ const nsSize& containerSize = line->mContainerSize;
+ if (GetLastLineBaseline(aWM, kid, &kidBaseline)) {
+ // Ignore relative positioning for baseline calculations
+ *aResult = kidBaseline +
+ kid->GetLogicalNormalPosition(aWM, containerSize).B(aWM);
+ return true;
+ }
+ if (kid->IsScrollFrame()) {
+ // Defer to nsIFrame::GetLogicalBaseline (which synthesizes a baseline
+ // from the margin-box).
+ kidBaseline = kid->GetLogicalBaseline(aWM);
+ *aResult = kidBaseline +
+ kid->GetLogicalNormalPosition(aWM, containerSize).B(aWM);
+ return true;
+ }
+ } else {
+ // XXX Is this the right test? We have some bogus empty lines
+ // floating around, but IsEmpty is perhaps too weak.
+ if (line->BSize() != 0 || !line->IsEmpty()) {
+ *aResult = line->BStart() + line->GetLogicalAscent();
+ return true;
+ }
+ }
+ }
+ return false;
+}
+
+static nscoord CalculateBlockContentBEnd(WritingMode aWM,
+ nsBlockFrame* aFrame) {
+ MOZ_ASSERT(aFrame, "null ptr");
+
+ nscoord contentBEnd = 0;
+
+ for (const auto& line : aFrame->Lines()) {
+ if (line.IsBlock()) {
+ nsIFrame* child = line.mFirstChild;
+ const auto& containerSize = line.mContainerSize;
+ nscoord offset =
+ child->GetLogicalNormalPosition(aWM, containerSize).B(aWM);
+ contentBEnd =
+ std::max(contentBEnd,
+ nsLayoutUtils::CalculateContentBEnd(aWM, child) + offset);
+ } else {
+ contentBEnd = std::max(contentBEnd, line.BEnd());
+ }
+ }
+ return contentBEnd;
+}
+
+/* static */
+nscoord nsLayoutUtils::CalculateContentBEnd(WritingMode aWM, nsIFrame* aFrame) {
+ MOZ_ASSERT(aFrame, "null ptr");
+
+ nscoord contentBEnd = aFrame->BSize(aWM);
+
+ // We want scrollable overflow rather than visual because this
+ // calculation is intended to affect layout.
+ LogicalSize overflowSize(aWM, aFrame->ScrollableOverflowRect().Size());
+ if (overflowSize.BSize(aWM) > contentBEnd) {
+ FrameChildListIDs skip = {FrameChildListID::Overflow,
+ FrameChildListID::ExcessOverflowContainers,
+ FrameChildListID::OverflowOutOfFlow};
+ nsBlockFrame* blockFrame = do_QueryFrame(aFrame);
+ if (blockFrame) {
+ contentBEnd =
+ std::max(contentBEnd, CalculateBlockContentBEnd(aWM, blockFrame));
+ skip += FrameChildListID::Principal;
+ }
+ for (const auto& [list, listID] : aFrame->ChildLists()) {
+ if (!skip.contains(listID)) {
+ for (nsIFrame* child : list) {
+ nscoord offset =
+ child->GetLogicalNormalPosition(aWM, aFrame->GetSize()).B(aWM);
+ contentBEnd =
+ std::max(contentBEnd, CalculateContentBEnd(aWM, child) + offset);
+ }
+ }
+ }
+ }
+ return contentBEnd;
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetClosestLayer(nsIFrame* aFrame) {
+ nsIFrame* layer;
+ for (layer = aFrame; layer; layer = layer->GetParent()) {
+ if (layer->IsAbsPosContainingBlock() ||
+ (layer->GetParent() && layer->GetParent()->IsScrollFrame()))
+ break;
+ }
+ if (layer) return layer;
+ return aFrame->PresShell()->GetRootFrame();
+}
+
+SamplingFilter nsLayoutUtils::GetSamplingFilterForFrame(nsIFrame* aForFrame) {
+ switch (aForFrame->UsedImageRendering()) {
+ case StyleImageRendering::Smooth:
+ case StyleImageRendering::Optimizequality:
+ return SamplingFilter::LINEAR;
+ case StyleImageRendering::CrispEdges:
+ case StyleImageRendering::Optimizespeed:
+ case StyleImageRendering::Pixelated:
+ return SamplingFilter::POINT;
+ case StyleImageRendering::Auto:
+ return SamplingFilter::GOOD;
+ }
+ MOZ_ASSERT_UNREACHABLE("Unknown image-rendering value");
+ return SamplingFilter::GOOD;
+}
+
+/**
+ * Given an image being drawn into an appunit coordinate system, and
+ * a point in that coordinate system, map the point back into image
+ * pixel space.
+ * @param aSize the size of the image, in pixels
+ * @param aDest the rectangle that the image is being mapped into
+ * @param aPt a point in the same coordinate system as the rectangle
+ */
+static gfxPoint MapToFloatImagePixels(const gfxSize& aSize,
+ const gfxRect& aDest,
+ const gfxPoint& aPt) {
+ return gfxPoint(((aPt.x - aDest.X()) * aSize.width) / aDest.Width(),
+ ((aPt.y - aDest.Y()) * aSize.height) / aDest.Height());
+}
+
+/**
+ * Given an image being drawn into an pixel-based coordinate system, and
+ * a point in image space, map the point into the pixel-based coordinate
+ * system.
+ * @param aSize the size of the image, in pixels
+ * @param aDest the rectangle that the image is being mapped into
+ * @param aPt a point in image space
+ */
+static gfxPoint MapToFloatUserPixels(const gfxSize& aSize, const gfxRect& aDest,
+ const gfxPoint& aPt) {
+ return gfxPoint(aPt.x * aDest.Width() / aSize.width + aDest.X(),
+ aPt.y * aDest.Height() / aSize.height + aDest.Y());
+}
+
+/* static */
+gfxRect nsLayoutUtils::RectToGfxRect(const nsRect& aRect,
+ int32_t aAppUnitsPerDevPixel) {
+ return gfxRect(gfxFloat(aRect.x) / aAppUnitsPerDevPixel,
+ gfxFloat(aRect.y) / aAppUnitsPerDevPixel,
+ gfxFloat(aRect.width) / aAppUnitsPerDevPixel,
+ gfxFloat(aRect.height) / aAppUnitsPerDevPixel);
+}
+
+struct SnappedImageDrawingParameters {
+ // A transform from image space to device space.
+ gfxMatrix imageSpaceToDeviceSpace;
+ // The size at which the image should be drawn (which may not be its
+ // intrinsic size due to, for example, HQ scaling).
+ nsIntSize size;
+ // The region in tiled image space which will be drawn, with an associated
+ // region to which sampling should be restricted.
+ ImageRegion region;
+ // The default viewport size for SVG images, which we use unless a different
+ // one has been explicitly specified. This is the same as |size| except that
+ // it does not take into account any transformation on the gfxContext we're
+ // drawing to - for example, CSS transforms are not taken into account.
+ CSSIntSize svgViewportSize;
+ // Whether there's anything to draw at all.
+ bool shouldDraw;
+
+ SnappedImageDrawingParameters()
+ : region(ImageRegion::Empty()), shouldDraw(false) {}
+
+ SnappedImageDrawingParameters(const gfxMatrix& aImageSpaceToDeviceSpace,
+ const nsIntSize& aSize,
+ const ImageRegion& aRegion,
+ const CSSIntSize& aSVGViewportSize)
+ : imageSpaceToDeviceSpace(aImageSpaceToDeviceSpace),
+ size(aSize),
+ region(aRegion),
+ svgViewportSize(aSVGViewportSize),
+ shouldDraw(true) {}
+};
+
+/**
+ * Given two axis-aligned rectangles, returns the transformation that maps the
+ * first onto the second.
+ *
+ * @param aFrom The rect to be transformed.
+ * @param aTo The rect that aFrom should be mapped onto by the transformation.
+ */
+static gfxMatrix TransformBetweenRects(const gfxRect& aFrom,
+ const gfxRect& aTo) {
+ MatrixScalesDouble scale(aTo.width / aFrom.width, aTo.height / aFrom.height);
+ gfxPoint translation(aTo.x - aFrom.x * scale.xScale,
+ aTo.y - aFrom.y * scale.yScale);
+ return gfxMatrix(scale.xScale, 0, 0, scale.yScale, translation.x,
+ translation.y);
+}
+
+static nsRect TileNearRect(const nsRect& aAnyTile, const nsRect& aTargetRect) {
+ nsPoint distance = aTargetRect.TopLeft() - aAnyTile.TopLeft();
+ return aAnyTile + nsPoint(distance.x / aAnyTile.width * aAnyTile.width,
+ distance.y / aAnyTile.height * aAnyTile.height);
+}
+
+static gfxFloat StableRound(gfxFloat aValue) {
+ // Values slightly less than 0.5 should round up like 0.5 would; we're
+ // assuming they were meant to be 0.5.
+ return floor(aValue + 0.5001);
+}
+
+static gfxPoint StableRound(const gfxPoint& aPoint) {
+ return gfxPoint(StableRound(aPoint.x), StableRound(aPoint.y));
+}
+
+/**
+ * Given a set of input parameters, compute certain output parameters
+ * for drawing an image with the image snapping algorithm.
+ * See https://wiki.mozilla.org/Gecko:Image_Snapping_and_Rendering
+ *
+ * @see nsLayoutUtils::DrawImage() for the descriptions of input parameters
+ */
+static SnappedImageDrawingParameters ComputeSnappedImageDrawingParameters(
+ gfxContext* aCtx, int32_t aAppUnitsPerDevPixel, const nsRect aDest,
+ const nsRect aFill, const nsPoint aAnchor, const nsRect aDirty,
+ imgIContainer* aImage, const SamplingFilter aSamplingFilter,
+ uint32_t aImageFlags, ExtendMode aExtendMode) {
+ if (aDest.IsEmpty() || aFill.IsEmpty())
+ return SnappedImageDrawingParameters();
+
+ // Avoid unnecessarily large offsets.
+ bool doTile = !aDest.Contains(aFill);
+ nsRect appUnitDest =
+ doTile ? TileNearRect(aDest, aFill.Intersect(aDirty)) : aDest;
+ nsPoint anchor = aAnchor + (appUnitDest.TopLeft() - aDest.TopLeft());
+
+ gfxRect devPixelDest =
+ nsLayoutUtils::RectToGfxRect(appUnitDest, aAppUnitsPerDevPixel);
+ gfxRect devPixelFill =
+ nsLayoutUtils::RectToGfxRect(aFill, aAppUnitsPerDevPixel);
+ gfxRect devPixelDirty =
+ nsLayoutUtils::RectToGfxRect(aDirty, aAppUnitsPerDevPixel);
+
+ gfxMatrix currentMatrix = aCtx->CurrentMatrixDouble();
+ gfxRect fill = devPixelFill;
+ gfxRect dest = devPixelDest;
+ bool didSnap;
+ // Snap even if we have a scale in the context. But don't snap if
+ // we have something that's not translation+scale, or if the scale flips in
+ // the X or Y direction, because snapped image drawing can't handle that yet.
+ if (!currentMatrix.HasNonAxisAlignedTransform() && currentMatrix._11 > 0.0 &&
+ currentMatrix._22 > 0.0 &&
+ aCtx->UserToDevicePixelSnapped(fill,
+ gfxContext::SnapOption::IgnoreScale) &&
+ aCtx->UserToDevicePixelSnapped(dest,
+ gfxContext::SnapOption::IgnoreScale)) {
+ // We snapped. On this code path, |fill| and |dest| take into account
+ // currentMatrix's transform.
+ didSnap = true;
+ } else {
+ // We didn't snap. On this code path, |fill| and |dest| do not take into
+ // account currentMatrix's transform.
+ didSnap = false;
+ fill = devPixelFill;
+ dest = devPixelDest;
+ }
+
+ // If we snapped above, |dest| already takes into account |currentMatrix|'s
+ // scale and has integer coordinates. If not, we need these properties to
+ // compute the optimal drawn image size, so compute |snappedDestSize| here.
+ gfxSize snappedDestSize = dest.Size();
+ auto scaleFactors = currentMatrix.ScaleFactors();
+ if (!didSnap) {
+ snappedDestSize.Scale(scaleFactors.xScale, scaleFactors.yScale);
+ snappedDestSize.width = NS_round(snappedDestSize.width);
+ snappedDestSize.height = NS_round(snappedDestSize.height);
+ }
+
+ // We need to be sure that this is at least one pixel in width and height,
+ // or we'll end up drawing nothing even if we have a nonempty fill.
+ snappedDestSize.width = std::max(snappedDestSize.width, 1.0);
+ snappedDestSize.height = std::max(snappedDestSize.height, 1.0);
+
+ // Bail if we're not going to end up drawing anything.
+ if (fill.IsEmpty()) {
+ return SnappedImageDrawingParameters();
+ }
+
+ nsIntSize intImageSize = aImage->OptimalImageSizeForDest(
+ snappedDestSize, imgIContainer::FRAME_CURRENT, aSamplingFilter,
+ aImageFlags);
+
+ nsIntSize svgViewportSize;
+ if (scaleFactors.xScale == 1.0 && scaleFactors.yScale == 1.0) {
+ // intImageSize is scaled by currentMatrix. But since there are no scale
+ // factors in currentMatrix, it is safe to assign intImageSize to
+ // svgViewportSize directly.
+ svgViewportSize = intImageSize;
+ } else {
+ // We should not take into account any transformation of currentMatrix
+ // when computing svg viewport size. Since currentMatrix contains scale
+ // factors, we need to recompute SVG viewport by unscaled devPixelDest.
+ svgViewportSize = aImage->OptimalImageSizeForDest(
+ devPixelDest.Size(), imgIContainer::FRAME_CURRENT, aSamplingFilter,
+ aImageFlags);
+ }
+
+ gfxSize imageSize(intImageSize.width, intImageSize.height);
+
+ // Compute the set of pixels that would be sampled by an ideal rendering
+ gfxPoint subimageTopLeft =
+ MapToFloatImagePixels(imageSize, devPixelDest, devPixelFill.TopLeft());
+ gfxPoint subimageBottomRight = MapToFloatImagePixels(
+ imageSize, devPixelDest, devPixelFill.BottomRight());
+ gfxRect subimage;
+ subimage.MoveTo(NSToIntFloor(subimageTopLeft.x),
+ NSToIntFloor(subimageTopLeft.y));
+ subimage.SizeTo(NSToIntCeil(subimageBottomRight.x) - subimage.x,
+ NSToIntCeil(subimageBottomRight.y) - subimage.y);
+
+ if (subimage.IsEmpty()) {
+ // Bail if the subimage is empty (we're not going to be drawing anything).
+ return SnappedImageDrawingParameters();
+ }
+
+ gfxMatrix transform;
+ gfxMatrix invTransform;
+
+ bool anchorAtUpperLeft =
+ anchor.x == appUnitDest.x && anchor.y == appUnitDest.y;
+ bool exactlyOneImageCopy = aFill.IsEqualEdges(appUnitDest);
+ if (anchorAtUpperLeft && exactlyOneImageCopy) {
+ // The simple case: we can ignore the anchor point and compute the
+ // transformation from the sampled region (the subimage) to the fill rect.
+ // This approach is preferable when it works since it tends to produce
+ // less numerical error.
+ transform = TransformBetweenRects(subimage, fill);
+ invTransform = TransformBetweenRects(fill, subimage);
+ } else {
+ // The more complicated case: we compute the transformation from the
+ // image rect positioned at the image space anchor point to the dest rect
+ // positioned at the device space anchor point.
+
+ // Compute the anchor point in both device space and image space. This
+ // code assumes that pixel-based devices have one pixel per device unit!
+ gfxPoint anchorPoint(gfxFloat(anchor.x) / aAppUnitsPerDevPixel,
+ gfxFloat(anchor.y) / aAppUnitsPerDevPixel);
+ gfxPoint imageSpaceAnchorPoint =
+ MapToFloatImagePixels(imageSize, devPixelDest, anchorPoint);
+
+ if (didSnap) {
+ imageSpaceAnchorPoint = StableRound(imageSpaceAnchorPoint);
+ anchorPoint = imageSpaceAnchorPoint;
+ anchorPoint = MapToFloatUserPixels(imageSize, devPixelDest, anchorPoint);
+ anchorPoint = currentMatrix.TransformPoint(anchorPoint);
+ anchorPoint = StableRound(anchorPoint);
+ }
+
+ // Compute an unsnapped version of the dest rect's size. We continue to
+ // follow the pattern that we take |currentMatrix| into account only if
+ // |didSnap| is true.
+ gfxSize unsnappedDestSize =
+ didSnap ? devPixelDest.Size() * currentMatrix.ScaleFactors()
+ : devPixelDest.Size();
+
+ gfxRect anchoredDestRect(anchorPoint, unsnappedDestSize);
+ gfxRect anchoredImageRect(imageSpaceAnchorPoint, imageSize);
+
+ // Calculate anchoredDestRect with snapped fill rect when the devPixelFill
+ // rect corresponds to just a single tile in that direction
+ if (fill.Width() != devPixelFill.Width() &&
+ devPixelDest.x == devPixelFill.x &&
+ devPixelDest.XMost() == devPixelFill.XMost()) {
+ anchoredDestRect.width = fill.width;
+ }
+ if (fill.Height() != devPixelFill.Height() &&
+ devPixelDest.y == devPixelFill.y &&
+ devPixelDest.YMost() == devPixelFill.YMost()) {
+ anchoredDestRect.height = fill.height;
+ }
+
+ transform = TransformBetweenRects(anchoredImageRect, anchoredDestRect);
+ invTransform = TransformBetweenRects(anchoredDestRect, anchoredImageRect);
+ }
+
+ // If the transform is not a straight translation by integers, then
+ // filtering will occur, and restricting the fill rect to the dirty rect
+ // would change the values computed for edge pixels, which we can't allow.
+ // Also, if 'didSnap' is false then rounding out 'devPixelDirty' might not
+ // produce pixel-aligned coordinates, which would also break the values
+ // computed for edge pixels.
+ if (didSnap && !invTransform.HasNonIntegerTranslation()) {
+ // This form of Transform is safe to call since non-axis-aligned
+ // transforms wouldn't be snapped.
+ devPixelDirty = currentMatrix.TransformRect(devPixelDirty);
+ devPixelDirty.RoundOut();
+ fill = fill.Intersect(devPixelDirty);
+ }
+ if (fill.IsEmpty()) return SnappedImageDrawingParameters();
+
+ gfxRect imageSpaceFill(didSnap ? invTransform.TransformRect(fill)
+ : invTransform.TransformBounds(fill));
+
+ // If we didn't snap, we need to post-multiply the matrix on the context to
+ // get the final matrix we'll draw with, because we didn't take it into
+ // account when computing the matrices above.
+ if (!didSnap) {
+ transform = transform * currentMatrix;
+ }
+
+ ExtendMode extendMode = (aImageFlags & imgIContainer::FLAG_CLAMP)
+ ? ExtendMode::CLAMP
+ : aExtendMode;
+ // We were passed in the default extend mode but need to tile.
+ if (extendMode == ExtendMode::CLAMP && doTile) {
+ MOZ_ASSERT(!(aImageFlags & imgIContainer::FLAG_CLAMP));
+ extendMode = ExtendMode::REPEAT;
+ }
+
+ ImageRegion region = ImageRegion::CreateWithSamplingRestriction(
+ imageSpaceFill, subimage, extendMode);
+
+ return SnappedImageDrawingParameters(
+ transform, intImageSize, region,
+ CSSIntSize(svgViewportSize.width, svgViewportSize.height));
+}
+
+static ImgDrawResult DrawImageInternal(
+ gfxContext& aContext, nsPresContext* aPresContext, imgIContainer* aImage,
+ const SamplingFilter aSamplingFilter, const nsRect& aDest,
+ const nsRect& aFill, const nsPoint& aAnchor, const nsRect& aDirty,
+ const SVGImageContext& aSVGContext, uint32_t aImageFlags,
+ ExtendMode aExtendMode = ExtendMode::CLAMP, float aOpacity = 1.0) {
+ ImgDrawResult result = ImgDrawResult::SUCCESS;
+
+ aImageFlags |= imgIContainer::FLAG_ASYNC_NOTIFY;
+
+ if (aPresContext->Type() == nsPresContext::eContext_Print) {
+ // We want vector images to be passed on as vector commands, not a raster
+ // image.
+ aImageFlags |= imgIContainer::FLAG_BYPASS_SURFACE_CACHE;
+ }
+ if (aDest.Contains(aFill)) {
+ aImageFlags |= imgIContainer::FLAG_CLAMP;
+ }
+ int32_t appUnitsPerDevPixel = aPresContext->AppUnitsPerDevPixel();
+
+ SnappedImageDrawingParameters params = ComputeSnappedImageDrawingParameters(
+ &aContext, appUnitsPerDevPixel, aDest, aFill, aAnchor, aDirty, aImage,
+ aSamplingFilter, aImageFlags, aExtendMode);
+
+ if (!params.shouldDraw) {
+ return result;
+ }
+
+ {
+ gfxContextMatrixAutoSaveRestore contextMatrixRestorer(&aContext);
+
+ aContext.SetMatrixDouble(params.imageSpaceToDeviceSpace);
+
+ SVGImageContext newContext = aSVGContext;
+ if (!aSVGContext.GetViewportSize()) {
+ newContext.SetViewportSize(Some(params.svgViewportSize));
+ }
+
+ result = aImage->Draw(&aContext, params.size, params.region,
+ imgIContainer::FRAME_CURRENT, aSamplingFilter,
+ newContext, aImageFlags, aOpacity);
+ }
+
+ return result;
+}
+
+/* static */
+ImgDrawResult nsLayoutUtils::DrawSingleUnscaledImage(
+ gfxContext& aContext, nsPresContext* aPresContext, imgIContainer* aImage,
+ const SamplingFilter aSamplingFilter, const nsPoint& aDest,
+ const nsRect* aDirty, const SVGImageContext& aSVGContext,
+ uint32_t aImageFlags, const nsRect* aSourceArea) {
+ CSSIntSize imageSize;
+ aImage->GetWidth(&imageSize.width);
+ aImage->GetHeight(&imageSize.height);
+ aImage->GetResolution().ApplyTo(imageSize.width, imageSize.height);
+
+ if (imageSize.width < 1 || imageSize.height < 1) {
+ NS_WARNING("Image width or height is non-positive");
+ return ImgDrawResult::TEMPORARY_ERROR;
+ }
+
+ nsSize size(CSSPixel::ToAppUnits(imageSize));
+ nsRect source;
+ if (aSourceArea) {
+ source = *aSourceArea;
+ } else {
+ source.SizeTo(size);
+ }
+
+ nsRect dest(aDest - source.TopLeft(), size);
+ nsRect fill(aDest, source.Size());
+ // Ensure that only a single image tile is drawn. If aSourceArea extends
+ // outside the image bounds, we want to honor the aSourceArea-to-aDest
+ // translation but we don't want to actually tile the image.
+ fill.IntersectRect(fill, dest);
+ return DrawImageInternal(aContext, aPresContext, aImage, aSamplingFilter,
+ dest, fill, aDest, aDirty ? *aDirty : dest,
+ aSVGContext, aImageFlags);
+}
+
+/* static */
+ImgDrawResult nsLayoutUtils::DrawSingleImage(
+ gfxContext& aContext, nsPresContext* aPresContext, imgIContainer* aImage,
+ SamplingFilter aSamplingFilter, const nsRect& aDest, const nsRect& aDirty,
+ const SVGImageContext& aSVGContext, uint32_t aImageFlags,
+ const nsPoint* aAnchorPoint) {
+ // NOTE(emilio): We can hardcode resolution to 1 here, since we're interested
+ // in the actual image pixels, for snapping purposes, not on the adjusted
+ // size.
+ CSSIntSize pixelImageSize(ComputeSizeForDrawingWithFallback(
+ aImage, ImageResolution(), aDest.Size()));
+ if (pixelImageSize.width < 1 || pixelImageSize.height < 1) {
+ NS_ASSERTION(pixelImageSize.width >= 0 && pixelImageSize.height >= 0,
+ "Image width or height is negative");
+ return ImgDrawResult::SUCCESS; // no point in drawing a zero size image
+ }
+
+ const nsSize imageSize(CSSPixel::ToAppUnits(pixelImageSize));
+ const nsRect source(nsPoint(), imageSize);
+ const nsRect dest = GetWholeImageDestination(imageSize, source, aDest);
+
+ // Ensure that only a single image tile is drawn. If aSourceArea extends
+ // outside the image bounds, we want to honor the aSourceArea-to-aDest
+ // transform but we don't want to actually tile the image.
+ nsRect fill;
+ fill.IntersectRect(aDest, dest);
+ return DrawImageInternal(aContext, aPresContext, aImage, aSamplingFilter,
+ dest, fill,
+ aAnchorPoint ? *aAnchorPoint : fill.TopLeft(),
+ aDirty, aSVGContext, aImageFlags);
+}
+
+/* static */
+void nsLayoutUtils::ComputeSizeForDrawing(
+ imgIContainer* aImage, const ImageResolution& aResolution,
+ /* outparam */ CSSIntSize& aImageSize,
+ /* outparam */ AspectRatio& aIntrinsicRatio,
+ /* outparam */ bool& aGotWidth,
+ /* outparam */ bool& aGotHeight) {
+ aGotWidth = NS_SUCCEEDED(aImage->GetWidth(&aImageSize.width));
+ aGotHeight = NS_SUCCEEDED(aImage->GetHeight(&aImageSize.height));
+ Maybe<AspectRatio> intrinsicRatio = aImage->GetIntrinsicRatio();
+ aIntrinsicRatio = intrinsicRatio.valueOr(AspectRatio());
+
+ if (aGotWidth) {
+ aResolution.ApplyXTo(aImageSize.width);
+ }
+ if (aGotHeight) {
+ aResolution.ApplyYTo(aImageSize.height);
+ }
+
+ if (!(aGotWidth && aGotHeight) && intrinsicRatio.isNothing()) {
+ // We hit an error (say, because the image failed to load or couldn't be
+ // decoded) and should return zero size.
+ aGotWidth = aGotHeight = true;
+ aImageSize = CSSIntSize(0, 0);
+ }
+}
+
+/* static */
+CSSIntSize nsLayoutUtils::ComputeSizeForDrawingWithFallback(
+ imgIContainer* aImage, const ImageResolution& aResolution,
+ const nsSize& aFallbackSize) {
+ CSSIntSize imageSize;
+ AspectRatio imageRatio;
+ bool gotHeight, gotWidth;
+ ComputeSizeForDrawing(aImage, aResolution, imageSize, imageRatio, gotWidth,
+ gotHeight);
+
+ // If we didn't get both width and height, try to compute them using the
+ // intrinsic ratio of the image.
+ if (gotWidth != gotHeight) {
+ if (!gotWidth) {
+ if (imageRatio) {
+ imageSize.width = imageRatio.ApplyTo(imageSize.height);
+ gotWidth = true;
+ }
+ } else {
+ if (imageRatio) {
+ imageSize.height = imageRatio.Inverted().ApplyTo(imageSize.width);
+ gotHeight = true;
+ }
+ }
+ }
+
+ // If we still don't have a width or height, just use the fallback size the
+ // caller provided.
+ if (!gotWidth) {
+ imageSize.width =
+ nsPresContext::AppUnitsToIntCSSPixels(aFallbackSize.width);
+ }
+ if (!gotHeight) {
+ imageSize.height =
+ nsPresContext::AppUnitsToIntCSSPixels(aFallbackSize.height);
+ }
+
+ return imageSize;
+}
+
+/* static */ LayerIntRect SnapRectForImage(
+ const gfx::Matrix& aTransform, const gfx::MatrixScales& aScaleFactors,
+ const LayoutDeviceRect& aRect) {
+ // Attempt to snap pixels, the same as ComputeSnappedImageDrawingParameters.
+ // Any changes to the algorithm here will need to be reflected there.
+ bool snapped = false;
+ LayerIntRect snapRect;
+ if (!aTransform.HasNonAxisAlignedTransform() && aTransform._11 > 0.0 &&
+ aTransform._22 > 0.0) {
+ gfxRect rect(gfxPoint(aRect.X(), aRect.Y()),
+ gfxSize(aRect.Width(), aRect.Height()));
+
+ gfxPoint p1 =
+ ThebesPoint(aTransform.TransformPoint(ToPoint(rect.TopLeft())));
+ gfxPoint p2 =
+ ThebesPoint(aTransform.TransformPoint(ToPoint(rect.TopRight())));
+ gfxPoint p3 =
+ ThebesPoint(aTransform.TransformPoint(ToPoint(rect.BottomRight())));
+
+ if (p2 == gfxPoint(p1.x, p3.y) || p2 == gfxPoint(p3.x, p1.y)) {
+ p1.Round();
+ p3.Round();
+
+ IntPoint p1i(int32_t(p1.x), int32_t(p1.y));
+ IntPoint p3i(int32_t(p3.x), int32_t(p3.y));
+
+ snapRect.MoveTo(std::min(p1i.x, p3i.x), std::min(p1i.y, p3i.y));
+ snapRect.SizeTo(std::max(p1i.x, p3i.x) - snapRect.X(),
+ std::max(p1i.y, p3i.y) - snapRect.Y());
+ snapped = true;
+ }
+ }
+
+ if (!snapped) {
+ // If we couldn't snap directly with the transform, we need to go best
+ // effort in layer pixels.
+ snapRect = RoundedToInt(
+ aRect * LayoutDeviceToLayerScale2D::FromUnknownScale(aScaleFactors));
+ }
+
+ // An empty size is unacceptable so we ensure our suggested size is at least
+ // 1 pixel wide/tall.
+ if (snapRect.Width() < 1) {
+ snapRect.SetWidth(1);
+ }
+ if (snapRect.Height() < 1) {
+ snapRect.SetHeight(1);
+ }
+ return snapRect;
+}
+
+/* static */
+IntSize nsLayoutUtils::ComputeImageContainerDrawingParameters(
+ imgIContainer* aImage, nsIFrame* aForFrame,
+ const LayoutDeviceRect& aDestRect, const LayoutDeviceRect& aFillRect,
+ const StackingContextHelper& aSc, uint32_t aFlags,
+ SVGImageContext& aSVGContext, Maybe<ImageIntRegion>& aRegion) {
+ MOZ_ASSERT(aImage);
+ MOZ_ASSERT(aForFrame);
+
+ MatrixScales scaleFactors = aSc.GetInheritedScale();
+ SamplingFilter samplingFilter =
+ nsLayoutUtils::GetSamplingFilterForFrame(aForFrame);
+
+ // Compute our SVG context parameters, if any. Don't replace the viewport
+ // size if it was already set, prefer what the caller gave.
+ SVGImageContext::MaybeStoreContextPaint(aSVGContext, aForFrame, aImage);
+ if ((scaleFactors.xScale != 1.0 || scaleFactors.yScale != 1.0) &&
+ aImage->GetType() == imgIContainer::TYPE_VECTOR &&
+ (!aSVGContext.GetViewportSize())) {
+ gfxSize gfxDestSize(aDestRect.Width(), aDestRect.Height());
+ IntSize viewportSize = aImage->OptimalImageSizeForDest(
+ gfxDestSize, imgIContainer::FRAME_CURRENT, samplingFilter, aFlags);
+
+ CSSIntSize cssViewportSize(viewportSize.width, viewportSize.height);
+ aSVGContext.SetViewportSize(Some(cssViewportSize));
+ }
+
+ const gfx::Matrix& itm = aSc.GetInheritedTransform();
+ LayerIntRect destRect = SnapRectForImage(itm, scaleFactors, aDestRect);
+
+ // Since we always decode entire raster images, we only care about the
+ // ImageIntRegion for vector images when we are recording blobs, for which we
+ // may only draw part of in some cases.
+ if ((aImage->GetType() != imgIContainer::TYPE_VECTOR) ||
+ !(aFlags & imgIContainer::FLAG_RECORD_BLOB)) {
+ // If the transform scale of our stacking context helper is being animated
+ // on the compositor then the transform will have the current value of the
+ // scale, but the scale factors will have max value of the scale animation.
+ // So we want to ask for a decoded image that can fulfill that larger size.
+ int32_t scaleWidth = int32_t(ceil(aDestRect.Width() * scaleFactors.xScale));
+ if (scaleWidth > destRect.width + 2) {
+ destRect.width = scaleWidth;
+ }
+ int32_t scaleHeight =
+ int32_t(ceil(aDestRect.Height() * scaleFactors.yScale));
+ if (scaleHeight > destRect.height + 2) {
+ destRect.height = scaleHeight;
+ }
+
+ return aImage->OptimalImageSizeForDest(
+ gfxSize(destRect.Width(), destRect.Height()),
+ imgIContainer::FRAME_CURRENT, samplingFilter, aFlags);
+ }
+
+ // We only use the region rect with blob recordings. This is because when we
+ // rasterize an SVG image in process, we always create a complete
+ // rasterization of the whole image which can be given to any caller, while
+ // we support partial rasterization with the blob recordings.
+ if (aFlags & imgIContainer::FLAG_RECORD_BLOB) {
+ // If the dest rect contains the fill rect, then we are only displaying part
+ // of the vector image. We need to calculate the restriction region to avoid
+ // drawing more than we need, and sampling outside the desired bounds.
+ LayerIntRect clipRect = SnapRectForImage(itm, scaleFactors, aFillRect);
+ if (destRect.Contains(clipRect)) {
+ LayerIntRect restrictRect = destRect.Intersect(clipRect);
+ restrictRect.MoveBy(-destRect.TopLeft());
+
+ if (restrictRect.Width() < 1) {
+ restrictRect.SetWidth(1);
+ }
+ if (restrictRect.Height() < 1) {
+ restrictRect.SetHeight(1);
+ }
+
+ if (restrictRect.X() != 0 || restrictRect.Y() != 0 ||
+ restrictRect.Size() != destRect.Size()) {
+ IntRect sampleRect = restrictRect.ToUnknownRect();
+ aRegion = Some(ImageIntRegion::CreateWithSamplingRestriction(
+ sampleRect, sampleRect, ExtendMode::CLAMP));
+ }
+ }
+ }
+
+ // VectorImage::OptimalImageSizeForDest will just round up, but we already
+ // have an integer size.
+ return destRect.Size().ToUnknownSize();
+}
+
+/* static */
+nsPoint nsLayoutUtils::GetBackgroundFirstTilePos(const nsPoint& aDest,
+ const nsPoint& aFill,
+ const nsSize& aRepeatSize) {
+ return nsPoint(NSToIntFloor(float(aFill.x - aDest.x) / aRepeatSize.width) *
+ aRepeatSize.width,
+ NSToIntFloor(float(aFill.y - aDest.y) / aRepeatSize.height) *
+ aRepeatSize.height) +
+ aDest;
+}
+
+/* static */
+ImgDrawResult nsLayoutUtils::DrawBackgroundImage(
+ gfxContext& aContext, nsIFrame* aForFrame, nsPresContext* aPresContext,
+ imgIContainer* aImage, SamplingFilter aSamplingFilter, const nsRect& aDest,
+ const nsRect& aFill, const nsSize& aRepeatSize, const nsPoint& aAnchor,
+ const nsRect& aDirty, uint32_t aImageFlags, ExtendMode aExtendMode,
+ float aOpacity) {
+ AUTO_PROFILER_LABEL("nsLayoutUtils::DrawBackgroundImage",
+ GRAPHICS_Rasterization);
+
+ CSSIntSize destCSSSize{nsPresContext::AppUnitsToIntCSSPixels(aDest.width),
+ nsPresContext::AppUnitsToIntCSSPixels(aDest.height)};
+
+ SVGImageContext svgContext(Some(destCSSSize));
+ SVGImageContext::MaybeStoreContextPaint(svgContext, aForFrame, aImage);
+
+ /* Fast path when there is no need for image spacing */
+ if (aRepeatSize.width == aDest.width && aRepeatSize.height == aDest.height) {
+ return DrawImageInternal(aContext, aPresContext, aImage, aSamplingFilter,
+ aDest, aFill, aAnchor, aDirty, svgContext,
+ aImageFlags, aExtendMode, aOpacity);
+ }
+
+ const nsPoint firstTilePos =
+ GetBackgroundFirstTilePos(aDest.TopLeft(), aFill.TopLeft(), aRepeatSize);
+ const nscoord xMost = aFill.XMost();
+ const nscoord repeatWidth = aRepeatSize.width;
+ const nscoord yMost = aFill.YMost();
+ const nscoord repeatHeight = aRepeatSize.height;
+ nsRect dest(0, 0, aDest.width, aDest.height);
+ nsPoint anchor = aAnchor;
+ for (nscoord x = firstTilePos.x; x < xMost; x += repeatWidth) {
+ for (nscoord y = firstTilePos.y; y < yMost; y += repeatHeight) {
+ dest.x = x;
+ dest.y = y;
+ ImgDrawResult result = DrawImageInternal(
+ aContext, aPresContext, aImage, aSamplingFilter, dest, dest, anchor,
+ aDirty, svgContext, aImageFlags, ExtendMode::CLAMP, aOpacity);
+ anchor.y += repeatHeight;
+ if (result != ImgDrawResult::SUCCESS) {
+ return result;
+ }
+ }
+ anchor.x += repeatWidth;
+ anchor.y = aAnchor.y;
+ }
+
+ return ImgDrawResult::SUCCESS;
+}
+
+/* static */
+ImgDrawResult nsLayoutUtils::DrawImage(
+ gfxContext& aContext, ComputedStyle* aComputedStyle,
+ nsPresContext* aPresContext, imgIContainer* aImage,
+ const SamplingFilter aSamplingFilter, const nsRect& aDest,
+ const nsRect& aFill, const nsPoint& aAnchor, const nsRect& aDirty,
+ uint32_t aImageFlags, float aOpacity) {
+ SVGImageContext svgContext;
+ SVGImageContext::MaybeStoreContextPaint(svgContext, *aPresContext,
+ *aComputedStyle, aImage);
+
+ return DrawImageInternal(aContext, aPresContext, aImage, aSamplingFilter,
+ aDest, aFill, aAnchor, aDirty, svgContext,
+ aImageFlags, ExtendMode::CLAMP, aOpacity);
+}
+
+/* static */
+nsRect nsLayoutUtils::GetWholeImageDestination(const nsSize& aWholeImageSize,
+ const nsRect& aImageSourceArea,
+ const nsRect& aDestArea) {
+ double scaleX = double(aDestArea.width) / aImageSourceArea.width;
+ double scaleY = double(aDestArea.height) / aImageSourceArea.height;
+ nscoord destOffsetX = NSToCoordRound(aImageSourceArea.x * scaleX);
+ nscoord destOffsetY = NSToCoordRound(aImageSourceArea.y * scaleY);
+ nscoord wholeSizeX = NSToCoordRound(aWholeImageSize.width * scaleX);
+ nscoord wholeSizeY = NSToCoordRound(aWholeImageSize.height * scaleY);
+ return nsRect(aDestArea.TopLeft() - nsPoint(destOffsetX, destOffsetY),
+ nsSize(wholeSizeX, wholeSizeY));
+}
+
+/* static */
+already_AddRefed<imgIContainer> nsLayoutUtils::OrientImage(
+ imgIContainer* aContainer, const StyleImageOrientation& aOrientation) {
+ MOZ_ASSERT(aContainer, "Should have an image container");
+ nsCOMPtr<imgIContainer> img(aContainer);
+
+ switch (aOrientation) {
+ case StyleImageOrientation::FromImage:
+ break;
+ case StyleImageOrientation::None:
+ img = ImageOps::Unorient(img);
+ break;
+ }
+
+ return img.forget();
+}
+
+/* static */
+bool nsLayoutUtils::ImageRequestUsesCORS(imgIRequest* aRequest) {
+ int32_t corsMode = mozilla::CORS_NONE;
+ return NS_SUCCEEDED(aRequest->GetCORSMode(&corsMode)) &&
+ corsMode != mozilla::CORS_NONE;
+}
+
+static bool NonZeroCorner(const LengthPercentage& aLength) {
+ // Since negative results are clamped to 0, check > 0.
+ return aLength.Resolve(nscoord_MAX) > 0 || aLength.Resolve(0) > 0;
+}
+
+/* static */
+bool nsLayoutUtils::HasNonZeroCorner(const BorderRadius& aCorners) {
+ for (const auto corner : mozilla::AllPhysicalHalfCorners()) {
+ if (NonZeroCorner(aCorners.Get(corner))) return true;
+ }
+ return false;
+}
+
+// aCorner is a "full corner" value, i.e. eCornerTopLeft etc.
+static bool IsCornerAdjacentToSide(uint8_t aCorner, Side aSide) {
+ static_assert((int)eSideTop == eCornerTopLeft, "Check for Full Corner");
+ static_assert((int)eSideRight == eCornerTopRight, "Check for Full Corner");
+ static_assert((int)eSideBottom == eCornerBottomRight,
+ "Check for Full Corner");
+ static_assert((int)eSideLeft == eCornerBottomLeft, "Check for Full Corner");
+ static_assert((int)eSideTop == ((eCornerTopRight - 1) & 3),
+ "Check for Full Corner");
+ static_assert((int)eSideRight == ((eCornerBottomRight - 1) & 3),
+ "Check for Full Corner");
+ static_assert((int)eSideBottom == ((eCornerBottomLeft - 1) & 3),
+ "Check for Full Corner");
+ static_assert((int)eSideLeft == ((eCornerTopLeft - 1) & 3),
+ "Check for Full Corner");
+
+ return aSide == aCorner || aSide == ((aCorner - 1) & 3);
+}
+
+/* static */
+bool nsLayoutUtils::HasNonZeroCornerOnSide(const BorderRadius& aCorners,
+ Side aSide) {
+ static_assert(eCornerTopLeftX / 2 == eCornerTopLeft,
+ "Check for Non Zero on side");
+ static_assert(eCornerTopLeftY / 2 == eCornerTopLeft,
+ "Check for Non Zero on side");
+ static_assert(eCornerTopRightX / 2 == eCornerTopRight,
+ "Check for Non Zero on side");
+ static_assert(eCornerTopRightY / 2 == eCornerTopRight,
+ "Check for Non Zero on side");
+ static_assert(eCornerBottomRightX / 2 == eCornerBottomRight,
+ "Check for Non Zero on side");
+ static_assert(eCornerBottomRightY / 2 == eCornerBottomRight,
+ "Check for Non Zero on side");
+ static_assert(eCornerBottomLeftX / 2 == eCornerBottomLeft,
+ "Check for Non Zero on side");
+ static_assert(eCornerBottomLeftY / 2 == eCornerBottomLeft,
+ "Check for Non Zero on side");
+
+ for (const auto corner : mozilla::AllPhysicalHalfCorners()) {
+ // corner is a "half corner" value, so dividing by two gives us a
+ // "full corner" value.
+ if (NonZeroCorner(aCorners.Get(corner)) &&
+ IsCornerAdjacentToSide(corner / 2, aSide))
+ return true;
+ }
+ return false;
+}
+
+/* static */
+widget::TransparencyMode nsLayoutUtils::GetFrameTransparency(
+ nsIFrame* aBackgroundFrame, nsIFrame* aCSSRootFrame) {
+ if (!aCSSRootFrame->StyleEffects()->IsOpaque()) {
+ return TransparencyMode::Transparent;
+ }
+
+ if (HasNonZeroCorner(aCSSRootFrame->StyleBorder()->mBorderRadius)) {
+ return TransparencyMode::Transparent;
+ }
+
+ StyleAppearance appearance =
+ aCSSRootFrame->StyleDisplay()->EffectiveAppearance();
+
+ if (appearance == StyleAppearance::MozWinBorderlessGlass) {
+ return TransparencyMode::BorderlessGlass;
+ }
+
+ nsITheme::Transparency transparency;
+ if (aCSSRootFrame->IsThemed(&transparency)) {
+ return transparency == nsITheme::eTransparent
+ ? TransparencyMode::Transparent
+ : TransparencyMode::Opaque;
+ }
+
+ // We need an uninitialized window to be treated as opaque because doing
+ // otherwise breaks window display effects on some platforms, specifically
+ // Vista. (bug 450322)
+ if (aBackgroundFrame->IsViewportFrame() &&
+ !aBackgroundFrame->PrincipalChildList().FirstChild()) {
+ return TransparencyMode::Opaque;
+ }
+
+ const ComputedStyle* bgSC = nsCSSRendering::FindBackground(aBackgroundFrame);
+ if (!bgSC) {
+ return TransparencyMode::Transparent;
+ }
+ const nsStyleBackground* bg = bgSC->StyleBackground();
+ if (NS_GET_A(bg->BackgroundColor(bgSC)) < 255 ||
+ // bottom layer's clip is used for the color
+ bg->BottomLayer().mClip != StyleGeometryBox::BorderBox) {
+ return TransparencyMode::Transparent;
+ }
+ return TransparencyMode::Opaque;
+}
+
+/* static */
+bool nsLayoutUtils::IsPopup(const nsIFrame* aFrame) {
+ // Optimization: the frame can't possibly be a popup if it has no view.
+ if (!aFrame->HasView()) {
+ NS_ASSERTION(!aFrame->IsMenuPopupFrame(), "popup frame must have a view");
+ return false;
+ }
+ return aFrame->IsMenuPopupFrame();
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetDisplayRootFrame(nsIFrame* aFrame) {
+ return const_cast<nsIFrame*>(
+ nsLayoutUtils::GetDisplayRootFrame(const_cast<const nsIFrame*>(aFrame)));
+}
+
+/* static */
+const nsIFrame* nsLayoutUtils::GetDisplayRootFrame(const nsIFrame* aFrame) {
+ // We could use GetRootPresContext() here if the
+ // NS_FRAME_IN_POPUP frame bit is set.
+ const nsIFrame* f = aFrame;
+ for (;;) {
+ if (!f->HasAnyStateBits(NS_FRAME_IN_POPUP)) {
+ f = f->PresShell()->GetRootFrame();
+ if (!f) {
+ return aFrame;
+ }
+ } else if (IsPopup(f)) {
+ return f;
+ }
+ nsIFrame* parent = GetCrossDocParentFrameInProcess(f);
+ if (!parent) return f;
+ f = parent;
+ }
+}
+
+/* static */
+nsIFrame* nsLayoutUtils::GetReferenceFrame(nsIFrame* aFrame) {
+ nsIFrame* f = aFrame;
+ for (;;) {
+ if (f->IsTransformed() || f->IsPreserve3DLeaf() || IsPopup(f)) {
+ return f;
+ }
+ nsIFrame* parent = GetCrossDocParentFrameInProcess(f);
+ if (!parent) {
+ return f;
+ }
+ f = parent;
+ }
+}
+
+/* static */ gfx::ShapedTextFlags nsLayoutUtils::GetTextRunFlagsForStyle(
+ const ComputedStyle* aComputedStyle, nsPresContext* aPresContext,
+ const nsStyleFont* aStyleFont, const nsStyleText* aStyleText,
+ nscoord aLetterSpacing) {
+ gfx::ShapedTextFlags result = gfx::ShapedTextFlags();
+ if (aLetterSpacing != 0 ||
+ aStyleText->mTextJustify == StyleTextJustify::InterCharacter) {
+ result |= gfx::ShapedTextFlags::TEXT_DISABLE_OPTIONAL_LIGATURES;
+ }
+ if (aStyleText->mMozControlCharacterVisibility ==
+ StyleMozControlCharacterVisibility::Hidden) {
+ result |= gfx::ShapedTextFlags::TEXT_HIDE_CONTROL_CHARACTERS;
+ }
+ switch (aComputedStyle->StyleText()->mTextRendering) {
+ case StyleTextRendering::Optimizespeed:
+ result |= gfx::ShapedTextFlags::TEXT_OPTIMIZE_SPEED;
+ break;
+ case StyleTextRendering::Auto:
+ if (aPresContext && aStyleFont->mFont.size.ToCSSPixels() <
+ aPresContext->GetAutoQualityMinFontSize()) {
+ result |= gfx::ShapedTextFlags::TEXT_OPTIMIZE_SPEED;
+ }
+ break;
+ default:
+ break;
+ }
+ return result | GetTextRunOrientFlagsForStyle(aComputedStyle);
+}
+
+/* static */ gfx::ShapedTextFlags nsLayoutUtils::GetTextRunOrientFlagsForStyle(
+ const ComputedStyle* aComputedStyle) {
+ auto writingMode = aComputedStyle->StyleVisibility()->mWritingMode;
+ switch (writingMode) {
+ case StyleWritingModeProperty::HorizontalTb:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_HORIZONTAL;
+
+ case StyleWritingModeProperty::VerticalLr:
+ case StyleWritingModeProperty::VerticalRl:
+ switch (aComputedStyle->StyleVisibility()->mTextOrientation) {
+ case StyleTextOrientation::Mixed:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_VERTICAL_MIXED;
+ case StyleTextOrientation::Upright:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_VERTICAL_UPRIGHT;
+ case StyleTextOrientation::Sideways:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_VERTICAL_SIDEWAYS_RIGHT;
+ default:
+ MOZ_ASSERT_UNREACHABLE("unknown text-orientation");
+ return gfx::ShapedTextFlags();
+ }
+
+ case StyleWritingModeProperty::SidewaysLr:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_VERTICAL_SIDEWAYS_LEFT;
+
+ case StyleWritingModeProperty::SidewaysRl:
+ return gfx::ShapedTextFlags::TEXT_ORIENT_VERTICAL_SIDEWAYS_RIGHT;
+
+ default:
+ MOZ_ASSERT_UNREACHABLE("unknown writing-mode");
+ return gfx::ShapedTextFlags();
+ }
+}
+
+/* static */
+void nsLayoutUtils::GetRectDifferenceStrips(const nsRect& aR1,
+ const nsRect& aR2, nsRect* aHStrip,
+ nsRect* aVStrip) {
+ NS_ASSERTION(aR1.TopLeft() == aR2.TopLeft(),
+ "expected rects at the same position");
+ nsRect unionRect(aR1.x, aR1.y, std::max(aR1.width, aR2.width),
+ std::max(aR1.height, aR2.height));
+ nscoord VStripStart = std::min(aR1.width, aR2.width);
+ nscoord HStripStart = std::min(aR1.height, aR2.height);
+ *aVStrip = unionRect;
+ aVStrip->x += VStripStart;
+ aVStrip->width -= VStripStart;
+ *aHStrip = unionRect;
+ aHStrip->y += HStripStart;
+ aHStrip->height -= HStripStart;
+}
+
+nsDeviceContext* nsLayoutUtils::GetDeviceContextForScreenInfo(
+ nsPIDOMWindowOuter* aWindow) {
+ if (!aWindow) {
+ return nullptr;
+ }
+
+ nsCOMPtr<nsIDocShell> docShell = aWindow->GetDocShell();
+ while (docShell) {
+ // Now make sure our size is up to date. That will mean that the device
+ // context does the right thing on multi-monitor systems when we return it
+ // to the caller. It will also make sure that our prescontext has been
+ // created, if we're supposed to have one.
+ nsCOMPtr<nsPIDOMWindowOuter> win = docShell->GetWindow();
+ if (!win) {
+ // No reason to go on
+ return nullptr;
+ }
+
+ win->EnsureSizeAndPositionUpToDate();
+
+ RefPtr<nsPresContext> presContext = docShell->GetPresContext();
+ if (presContext) {
+ nsDeviceContext* context = presContext->DeviceContext();
+ if (context) {
+ return context;
+ }
+ }
+
+ nsCOMPtr<nsIDocShellTreeItem> parentItem;
+ docShell->GetInProcessParent(getter_AddRefs(parentItem));
+ docShell = do_QueryInterface(parentItem);
+ }
+
+ return nullptr;
+}
+
+/* static */
+bool nsLayoutUtils::IsReallyFixedPos(const nsIFrame* aFrame) {
+ MOZ_ASSERT(aFrame->StyleDisplay()->mPosition == StylePositionProperty::Fixed,
+ "IsReallyFixedPos called on non-'position:fixed' frame");
+ return MayBeReallyFixedPos(aFrame);
+}
+
+/* static */
+bool nsLayoutUtils::MayBeReallyFixedPos(const nsIFrame* aFrame) {
+ MOZ_ASSERT(aFrame->GetParent(),
+ "MayBeReallyFixedPos called on frame not in tree");
+ LayoutFrameType parentType = aFrame->GetParent()->Type();
+ return parentType == LayoutFrameType::Viewport ||
+ parentType == LayoutFrameType::PageContent;
+}
+
+/* static */
+bool nsLayoutUtils::IsInPositionFixedSubtree(const nsIFrame* aFrame) {
+ for (const nsIFrame* f = aFrame; f; f = f->GetParent()) {
+ if (f->StyleDisplay()->mPosition == StylePositionProperty::Fixed &&
+ nsLayoutUtils::IsReallyFixedPos(f)) {
+ return true;
+ }
+ }
+ return false;
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromOffscreenCanvas(
+ OffscreenCanvas* aOffscreenCanvas, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ SurfaceFromElementResult result;
+
+ IntSize size = aOffscreenCanvas->GetWidthHeight();
+
+ result.mSourceSurface =
+ aOffscreenCanvas->GetSurfaceSnapshot(&result.mAlphaType);
+ if (!result.mSourceSurface) {
+ // If the element doesn't have a context then we won't get a snapshot. The
+ // canvas spec wants us to not error and just draw nothing, so return an
+ // empty surface.
+ result.mAlphaType = gfxAlphaType::Opaque;
+ RefPtr<DrawTarget> ref =
+ aTarget ? aTarget : gfxPlatform::ThreadLocalScreenReferenceDrawTarget();
+ if (ref->CanCreateSimilarDrawTarget(size, SurfaceFormat::B8G8R8A8)) {
+ RefPtr<DrawTarget> dt =
+ ref->CreateSimilarDrawTarget(size, SurfaceFormat::B8G8R8A8);
+ if (dt) {
+ result.mSourceSurface = dt->Snapshot();
+ }
+ }
+ } else if (aTarget) {
+ RefPtr<SourceSurface> opt =
+ aTarget->OptimizeSourceSurface(result.mSourceSurface);
+ if (opt) {
+ result.mSourceSurface = opt;
+ }
+ }
+
+ result.mHasSize = true;
+ result.mSize = size;
+ result.mIntrinsicSize = size;
+ result.mIsWriteOnly = aOffscreenCanvas->IsWriteOnly();
+
+ nsIGlobalObject* global = aOffscreenCanvas->GetParentObject();
+ if (global) {
+ result.mPrincipal = global->PrincipalOrNull();
+ }
+
+ return result;
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromImageBitmap(
+ mozilla::dom::ImageBitmap* aImageBitmap, uint32_t aSurfaceFlags) {
+ SurfaceFromElementResult result;
+ RefPtr<DrawTarget> dt = Factory::CreateDrawTarget(
+ BackendType::SKIA, IntSize(1, 1), SurfaceFormat::B8G8R8A8);
+
+ // An ImageBitmap, not being a DOM element, only has `origin-clean`
+ // (via our `IsWriteOnly`), and does not participate in CORS.
+ // Right now we mark this by setting mCORSUsed to true.
+ result.mCORSUsed = true;
+ result.mIsWriteOnly = aImageBitmap->IsWriteOnly();
+ result.mSourceSurface = aImageBitmap->PrepareForDrawTarget(dt);
+
+ if (result.mSourceSurface) {
+ result.mSize = result.mIntrinsicSize = result.mSourceSurface->GetSize();
+ result.mHasSize = true;
+ result.mAlphaType = IsOpaque(result.mSourceSurface->GetFormat())
+ ? gfxAlphaType::Opaque
+ : gfxAlphaType::Premult;
+ }
+
+ nsCOMPtr<nsIGlobalObject> global = aImageBitmap->GetParentObject();
+ if (global) {
+ result.mPrincipal = global->PrincipalOrNull();
+ }
+
+ return result;
+}
+
+static RefPtr<SourceSurface> ScaleSourceSurface(SourceSurface& aSurface,
+ const IntSize& aTargetSize) {
+ const IntSize surfaceSize = aSurface.GetSize();
+
+ MOZ_ASSERT(surfaceSize != aTargetSize);
+ MOZ_ASSERT(!surfaceSize.IsEmpty());
+ MOZ_ASSERT(!aTargetSize.IsEmpty());
+
+ RefPtr<DrawTarget> dt = Factory::CreateDrawTarget(
+ gfxVars::ContentBackend(), aTargetSize, aSurface.GetFormat());
+
+ if (!dt || !dt->IsValid()) {
+ return nullptr;
+ }
+
+ dt->DrawSurface(&aSurface, Rect(Point(), Size(aTargetSize)),
+ Rect(Point(), Size(surfaceSize)));
+ return dt->GetBackingSurface();
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromElement(
+ nsIImageLoadingContent* aElement, const Maybe<int32_t>& aResizeWidth,
+ const Maybe<int32_t>& aResizeHeight, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ SurfaceFromElementResult result;
+ nsresult rv;
+
+ nsCOMPtr<imgIRequest> imgRequest;
+ rv = aElement->GetRequest(nsIImageLoadingContent::CURRENT_REQUEST,
+ getter_AddRefs(imgRequest));
+ if (NS_FAILED(rv)) {
+ return result;
+ }
+
+ if (!imgRequest) {
+ // There's no image request. This is either because a request for
+ // a non-empty URI failed, or the URI is the empty string.
+ nsCOMPtr<nsIURI> currentURI;
+ aElement->GetCurrentURI(getter_AddRefs(currentURI));
+ if (!currentURI) {
+ // Treat the empty URI as available instead of broken state.
+ result.mHasSize = true;
+ }
+ return result;
+ }
+
+ uint32_t status;
+ imgRequest->GetImageStatus(&status);
+ result.mHasSize = status & imgIRequest::STATUS_SIZE_AVAILABLE;
+ if ((status & imgIRequest::STATUS_LOAD_COMPLETE) == 0) {
+ // Spec says to use GetComplete, but that only works on
+ // HTMLImageElement, and we support all sorts of other stuff
+ // here. Do this for now pending spec clarification.
+ result.mIsStillLoading = (status & imgIRequest::STATUS_ERROR) == 0;
+ return result;
+ }
+
+ nsCOMPtr<nsIPrincipal> principal;
+ rv = imgRequest->GetImagePrincipal(getter_AddRefs(principal));
+ if (NS_FAILED(rv)) {
+ return result;
+ }
+
+ nsCOMPtr<imgIContainer> imgContainer;
+ rv = imgRequest->GetImage(getter_AddRefs(imgContainer));
+ if (NS_FAILED(rv)) {
+ return result;
+ }
+
+ nsCOMPtr<nsIContent> content = do_QueryInterface(aElement);
+
+ // Ensure that the image is oriented the same way as it's displayed
+ // if the image request is of the same origin.
+ auto orientation =
+ content->GetPrimaryFrame() &&
+ !(aSurfaceFlags & SFE_ORIENTATION_FROM_IMAGE)
+ ? content->GetPrimaryFrame()->StyleVisibility()->UsedImageOrientation(
+ imgRequest)
+ : nsStyleVisibility::UsedImageOrientation(
+ imgRequest, StyleImageOrientation::FromImage);
+ imgContainer = OrientImage(imgContainer, orientation);
+
+ const bool noRasterize = aSurfaceFlags & SFE_NO_RASTERIZING_VECTORS;
+
+ uint32_t whichFrame = aSurfaceFlags & SFE_WANT_FIRST_FRAME_IF_IMAGE
+ ? (uint32_t)imgIContainer::FRAME_FIRST
+ : (uint32_t)imgIContainer::FRAME_CURRENT;
+ const bool exactSize = aSurfaceFlags & SFE_EXACT_SIZE_SURFACE;
+
+ uint32_t frameFlags =
+ imgIContainer::FLAG_SYNC_DECODE | imgIContainer::FLAG_ASYNC_NOTIFY;
+ if (aSurfaceFlags & SFE_NO_COLORSPACE_CONVERSION)
+ frameFlags |= imgIContainer::FLAG_DECODE_NO_COLORSPACE_CONVERSION;
+ if (aSurfaceFlags & SFE_ALLOW_NON_PREMULT) {
+ frameFlags |= imgIContainer::FLAG_DECODE_NO_PREMULTIPLY_ALPHA;
+ }
+
+ int32_t imgWidth, imgHeight;
+ HTMLImageElement* element = HTMLImageElement::FromNodeOrNull(content);
+ if (aSurfaceFlags & SFE_USE_ELEMENT_SIZE_IF_VECTOR && element &&
+ imgContainer->GetType() == imgIContainer::TYPE_VECTOR) {
+ // We're holding a strong ref to "element" via "content".
+ imgWidth = MOZ_KnownLive(element)->Width();
+ imgHeight = MOZ_KnownLive(element)->Height();
+ } else {
+ auto res = imgContainer->GetResolution();
+ rv = imgContainer->GetWidth(&imgWidth);
+ if (NS_SUCCEEDED(rv)) {
+ res.ApplyXTo(imgWidth);
+ } else if (aResizeWidth.isSome()) {
+ imgWidth = *aResizeWidth;
+ rv = NS_OK;
+ }
+ nsresult rv2 = imgContainer->GetHeight(&imgHeight);
+ if (NS_SUCCEEDED(rv2)) {
+ res.ApplyYTo(imgHeight);
+ } else if (aResizeHeight.isSome()) {
+ imgHeight = *aResizeHeight;
+ rv2 = NS_OK;
+ }
+ if (NS_FAILED(rv) || NS_FAILED(rv2)) return result;
+ }
+ result.mSize = result.mIntrinsicSize = IntSize(imgWidth, imgHeight);
+
+ if (!noRasterize || imgContainer->GetType() == imgIContainer::TYPE_RASTER) {
+ result.mSourceSurface =
+ imgContainer->GetFrameAtSize(result.mSize, whichFrame, frameFlags);
+ if (!result.mSourceSurface) {
+ return result;
+ }
+ if (exactSize && result.mSourceSurface->GetSize() != result.mSize) {
+ result.mSourceSurface =
+ ScaleSourceSurface(*result.mSourceSurface, result.mSize);
+ if (!result.mSourceSurface) {
+ return result;
+ }
+ }
+ // The surface we return is likely to be cached. We don't want to have to
+ // convert to a surface that's compatible with aTarget each time it's used
+ // (that would result in terrible performance), so we convert once here
+ // upfront if aTarget is specified.
+ if (aTarget) {
+ RefPtr<SourceSurface> optSurface =
+ aTarget->OptimizeSourceSurface(result.mSourceSurface);
+ if (optSurface) {
+ result.mSourceSurface = optSurface;
+ }
+ }
+
+ const auto& format = result.mSourceSurface->GetFormat();
+ if (IsOpaque(format)) {
+ result.mAlphaType = gfxAlphaType::Opaque;
+ } else if (frameFlags & imgIContainer::FLAG_DECODE_NO_PREMULTIPLY_ALPHA) {
+ result.mAlphaType = gfxAlphaType::NonPremult;
+ } else {
+ result.mAlphaType = gfxAlphaType::Premult;
+ }
+ } else {
+ result.mDrawInfo.mImgContainer = imgContainer;
+ result.mDrawInfo.mWhichFrame = whichFrame;
+ result.mDrawInfo.mDrawingFlags = frameFlags;
+ }
+
+ result.mCORSUsed = nsLayoutUtils::ImageRequestUsesCORS(imgRequest);
+
+ bool hadCrossOriginRedirects = true;
+ imgRequest->GetHadCrossOriginRedirects(&hadCrossOriginRedirects);
+
+ result.mPrincipal = std::move(principal);
+ result.mHadCrossOriginRedirects = hadCrossOriginRedirects;
+ result.mImageRequest = std::move(imgRequest);
+ result.mIsWriteOnly = CanvasUtils::CheckWriteOnlySecurity(
+ result.mCORSUsed, result.mPrincipal, result.mHadCrossOriginRedirects);
+
+ return result;
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromElement(
+ HTMLImageElement* aElement, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ return SurfaceFromElement(static_cast<nsIImageLoadingContent*>(aElement),
+ Nothing(), Nothing(), aSurfaceFlags, aTarget);
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromElement(
+ HTMLCanvasElement* aElement, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ SurfaceFromElementResult result;
+
+ IntSize size = aElement->GetSize();
+
+ auto pAlphaType = &result.mAlphaType;
+ if (!(aSurfaceFlags & SFE_ALLOW_NON_PREMULT)) {
+ pAlphaType =
+ nullptr; // Coersce GetSurfaceSnapshot to give us Opaque/Premult only.
+ }
+ result.mSourceSurface = aElement->GetSurfaceSnapshot(pAlphaType, aTarget);
+ if (!result.mSourceSurface) {
+ // If the element doesn't have a context then we won't get a snapshot. The
+ // canvas spec wants us to not error and just draw nothing, so return an
+ // empty surface.
+ result.mAlphaType = gfxAlphaType::Opaque;
+ RefPtr<DrawTarget> ref =
+ aTarget ? aTarget
+ : gfxPlatform::GetPlatform()->ScreenReferenceDrawTarget();
+ if (ref->CanCreateSimilarDrawTarget(size, SurfaceFormat::B8G8R8A8)) {
+ RefPtr<DrawTarget> dt =
+ ref->CreateSimilarDrawTarget(size, SurfaceFormat::B8G8R8A8);
+ if (dt) {
+ result.mSourceSurface = dt->Snapshot();
+ }
+ }
+ } else if (aTarget) {
+ RefPtr<SourceSurface> opt =
+ aTarget->OptimizeSourceSurface(result.mSourceSurface);
+ if (opt) {
+ result.mSourceSurface = opt;
+ }
+ }
+
+ // Ensure that any future changes to the canvas trigger proper invalidation,
+ // in case this is being used by -moz-element()
+ aElement->MarkContextClean();
+
+ result.mHasSize = true;
+ result.mSize = size;
+ result.mIntrinsicSize = size;
+ result.mPrincipal = aElement->NodePrincipal();
+ result.mHadCrossOriginRedirects = false;
+ result.mIsWriteOnly = aElement->IsWriteOnly();
+
+ return result;
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromElement(
+ HTMLVideoElement* aElement, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ SurfaceFromElementResult result;
+ result.mAlphaType = gfxAlphaType::Opaque; // Assume opaque.
+
+ if (aElement->ContainsRestrictedContent()) {
+ return result;
+ }
+
+ uint16_t readyState = aElement->ReadyState();
+ if (readyState == HAVE_NOTHING || readyState == HAVE_METADATA) {
+ result.mIsStillLoading = true;
+ return result;
+ }
+
+ // If it doesn't have a principal, just bail
+ nsCOMPtr<nsIPrincipal> principal = aElement->GetCurrentVideoPrincipal();
+ if (!principal) {
+ return result;
+ }
+
+ result.mLayersImage = aElement->GetCurrentImage();
+ if (!result.mLayersImage) {
+ return result;
+ }
+
+ result.mCORSUsed = aElement->GetCORSMode() != CORS_NONE;
+ result.mHasSize = true;
+ result.mSize = result.mLayersImage->GetSize();
+ result.mIntrinsicSize =
+ gfx::IntSize(aElement->VideoWidth(), aElement->VideoHeight());
+ result.mPrincipal = std::move(principal);
+ result.mHadCrossOriginRedirects = aElement->HadCrossOriginRedirects();
+ result.mIsWriteOnly = CanvasUtils::CheckWriteOnlySecurity(
+ result.mCORSUsed, result.mPrincipal, result.mHadCrossOriginRedirects);
+
+ if (aTarget) {
+ // They gave us a DrawTarget to optimize for, so even though we have a
+ // layers::Image, we should unconditionally try to grab a SourceSurface and
+ // try to optimize it.
+ if ((result.mSourceSurface = result.mLayersImage->GetAsSourceSurface())) {
+ RefPtr<SourceSurface> opt =
+ aTarget->OptimizeSourceSurface(result.mSourceSurface);
+ if (opt) {
+ result.mSourceSurface = opt;
+ }
+ }
+ }
+
+ return result;
+}
+
+SurfaceFromElementResult nsLayoutUtils::SurfaceFromElement(
+ dom::Element* aElement, const Maybe<int32_t>& aResizeWidth,
+ const Maybe<int32_t>& aResizeHeight, uint32_t aSurfaceFlags,
+ RefPtr<DrawTarget>& aTarget) {
+ // If it's a <canvas>, we may be able to just grab its internal surface
+ if (HTMLCanvasElement* canvas = HTMLCanvasElement::FromNodeOrNull(aElement)) {
+ return SurfaceFromElement(canvas, aSurfaceFlags, aTarget);
+ }
+
+ // Maybe it's <video>?
+ if (HTMLVideoElement* video = HTMLVideoElement::FromNodeOrNull(aElement)) {
+ return SurfaceFromElement(video, aSurfaceFlags, aTarget);
+ }
+
+ // Finally, check if it's a normal image
+ nsCOMPtr<nsIImageLoadingContent> imageLoader = do_QueryInterface(aElement);
+
+ if (!imageLoader) {
+ return SurfaceFromElementResult();
+ }
+
+ return SurfaceFromElement(imageLoader, aResizeWidth, aResizeHeight,
+ aSurfaceFlags, aTarget);
+}
+
+/* static */
+Element* nsLayoutUtils::GetEditableRootContentByContentEditable(
+ Document* aDocument) {
+ // If the document is in designMode we should return nullptr.
+ if (!aDocument || aDocument->IsInDesignMode()) {
+ return nullptr;
+ }
+
+ // contenteditable only works with HTML document.
+ // XXXbz should this test IsHTMLOrXHTML(), or just IsHTML()?
+ if (!aDocument->IsHTMLOrXHTML()) {
+ return nullptr;
+ }
+
+ Element* rootElement = aDocument->GetRootElement();
+ if (rootElement && rootElement->IsEditable()) {
+ return rootElement;
+ }
+
+ // If there is no editable root element, check its <body> element.
+ // Note that the body element could be <frameset> element.
+ Element* bodyElement = aDocument->GetBody();
+ if (bodyElement && bodyElement->IsEditable()) {
+ return bodyElement;
+ }
+ return nullptr;
+}
+
+#ifdef DEBUG
+/* static */
+void nsLayoutUtils::AssertNoDuplicateContinuations(
+ nsIFrame* aContainer, const nsFrameList& aFrameList) {
+ for (nsIFrame* f : aFrameList) {
+ // Check only later continuations of f; we deal with checking the
+ // earlier continuations when we hit those earlier continuations in
+ // the frame list.
+ for (nsIFrame* c = f; (c = c->GetNextInFlow());) {
+ NS_ASSERTION(c->GetParent() != aContainer || !aFrameList.ContainsFrame(c),
+ "Two continuations of the same frame in the same "
+ "frame list");
+ }
+ }
+}
+
+// Is one of aFrame's ancestors a letter frame?
+static bool IsInLetterFrame(nsIFrame* aFrame) {
+ for (nsIFrame* f = aFrame->GetParent(); f; f = f->GetParent()) {
+ if (f->IsLetterFrame()) {
+ return true;
+ }
+ }
+ return false;
+}
+
+/* static */
+void nsLayoutUtils::AssertTreeOnlyEmptyNextInFlows(nsIFrame* aSubtreeRoot) {
+ NS_ASSERTION(aSubtreeRoot->GetPrevInFlow(),
+ "frame tree not empty, but caller reported complete status");
+
+ // Also assert that text frames map no text.
+ auto [start, end] = aSubtreeRoot->GetOffsets();
+ // In some cases involving :first-letter, we'll partially unlink a
+ // continuation in the middle of a continuation chain from its
+ // previous and next continuations before destroying it, presumably so
+ // that we don't also destroy the later continuations. Once we've
+ // done this, GetOffsets returns incorrect values.
+ // For examples, see list of tests in
+ // https://bugzilla.mozilla.org/show_bug.cgi?id=619021#c29
+ NS_ASSERTION(start == end || IsInLetterFrame(aSubtreeRoot),
+ "frame tree not empty, but caller reported complete status");
+
+ for (const auto& childList : aSubtreeRoot->ChildLists()) {
+ for (nsIFrame* child : childList.mList) {
+ nsLayoutUtils::AssertTreeOnlyEmptyNextInFlows(child);
+ }
+ }
+}
+#endif
+
+static void GetFontFacesForFramesInner(
+ nsIFrame* aFrame, nsLayoutUtils::UsedFontFaceList& aResult,
+ nsLayoutUtils::UsedFontFaceTable& aFontFaces, uint32_t aMaxRanges,
+ bool aSkipCollapsedWhitespace) {
+ MOZ_ASSERT(aFrame, "NULL frame pointer");
+
+ if (aFrame->IsTextFrame()) {
+ if (!aFrame->GetPrevContinuation()) {
+ nsLayoutUtils::GetFontFacesForText(aFrame, 0, INT32_MAX, true, aResult,
+ aFontFaces, aMaxRanges,
+ aSkipCollapsedWhitespace);
+ }
+ return;
+ }
+
+ FrameChildListID childLists[] = {FrameChildListID::Principal,
+ FrameChildListID::Popup};
+ for (size_t i = 0; i < ArrayLength(childLists); ++i) {
+ for (nsIFrame* child : aFrame->GetChildList(childLists[i])) {
+ child = nsPlaceholderFrame::GetRealFrameFor(child);
+ GetFontFacesForFramesInner(child, aResult, aFontFaces, aMaxRanges,
+ aSkipCollapsedWhitespace);
+ }
+ }
+}
+
+/* static */
+nsresult nsLayoutUtils::GetFontFacesForFrames(nsIFrame* aFrame,
+ UsedFontFaceList& aResult,
+ UsedFontFaceTable& aFontFaces,
+ uint32_t aMaxRanges,
+ bool aSkipCollapsedWhitespace) {
+ MOZ_ASSERT(aFrame, "NULL frame pointer");
+
+ while (aFrame) {
+ GetFontFacesForFramesInner(aFrame, aResult, aFontFaces, aMaxRanges,
+ aSkipCollapsedWhitespace);
+ aFrame = GetNextContinuationOrIBSplitSibling(aFrame);
+ }
+
+ return NS_OK;
+}
+
+static void AddFontsFromTextRun(gfxTextRun* aTextRun, nsTextFrame* aFrame,
+ gfxSkipCharsIterator& aSkipIter,
+ const gfxTextRun::Range& aRange,
+ nsLayoutUtils::UsedFontFaceList& aResult,
+ nsLayoutUtils::UsedFontFaceTable& aFontFaces,
+ uint32_t aMaxRanges) {
+ nsIContent* content = aFrame->GetContent();
+ int32_t contentLimit =
+ aFrame->GetContentOffset() + aFrame->GetInFlowContentLength();
+ for (gfxTextRun::GlyphRunIterator glyphRuns(aTextRun, aRange);
+ !glyphRuns.AtEnd(); glyphRuns.NextRun()) {
+ gfxFontEntry* fe = glyphRuns.GlyphRun()->mFont->GetFontEntry();
+ // if we have already listed this face, just make sure the match type is
+ // recorded
+ InspectorFontFace* fontFace = aFontFaces.Get(fe);
+ if (fontFace) {
+ fontFace->AddMatchType(glyphRuns.GlyphRun()->mMatchType);
+ } else {
+ // A new font entry we haven't seen before
+ fontFace = new InspectorFontFace(fe, aTextRun->GetFontGroup(),
+ glyphRuns.GlyphRun()->mMatchType);
+ aFontFaces.InsertOrUpdate(fe, fontFace);
+ aResult.AppendElement(fontFace);
+ }
+
+ // Add this glyph run to the fontFace's list of ranges, unless we have
+ // already collected as many as wanted.
+ if (fontFace->RangeCount() < aMaxRanges) {
+ int32_t start =
+ aSkipIter.ConvertSkippedToOriginal(glyphRuns.StringStart());
+ int32_t end = aSkipIter.ConvertSkippedToOriginal(glyphRuns.StringEnd());
+
+ // Mapping back from textrun offsets ("skipped" offsets that reflect the
+ // text after whitespace collapsing, etc) to DOM content offsets in the
+ // original text is ambiguous, because many original characters can
+ // map to a single skipped offset. aSkipIter.ConvertSkippedToOriginal()
+ // will return an "original" offset that corresponds to the *end* of
+ // a collapsed run of characters in this case; but that might extend
+ // beyond the current content node if the textrun mapped multiple nodes.
+ // So we clamp the end offset to keep it valid for the content node
+ // that corresponds to the current textframe.
+ end = std::min(end, contentLimit);
+
+ if (end > start) {
+ RefPtr<nsRange> range =
+ nsRange::Create(content, start, content, end, IgnoreErrors());
+ NS_WARNING_ASSERTION(range,
+ "nsRange::Create() failed to create valid range");
+ if (range) {
+ fontFace->AddRange(range);
+ }
+ }
+ }
+ }
+}
+
+/* static */
+void nsLayoutUtils::GetFontFacesForText(nsIFrame* aFrame, int32_t aStartOffset,
+ int32_t aEndOffset,
+ bool aFollowContinuations,
+ UsedFontFaceList& aResult,
+ UsedFontFaceTable& aFontFaces,
+ uint32_t aMaxRanges,
+ bool aSkipCollapsedWhitespace) {
+ MOZ_ASSERT(aFrame, "NULL frame pointer");
+
+ if (!aFrame->IsTextFrame()) {
+ return;
+ }
+
+ if (!aFrame->StyleVisibility()->IsVisible()) {
+ return;
+ }
+
+ nsTextFrame* curr = static_cast<nsTextFrame*>(aFrame);
+ do {
+ int32_t fstart = std::max(curr->GetContentOffset(), aStartOffset);
+ int32_t fend = std::min(curr->GetContentEnd(), aEndOffset);
+ if (fstart >= fend) {
+ curr = static_cast<nsTextFrame*>(curr->GetNextContinuation());
+ continue;
+ }
+
+ // curr is overlapping with the offset we want
+ gfxSkipCharsIterator iter = curr->EnsureTextRun(nsTextFrame::eInflated);
+ gfxTextRun* textRun = curr->GetTextRun(nsTextFrame::eInflated);
+ if (!textRun) {
+ NS_WARNING("failed to get textRun, low memory?");
+ return;
+ }
+
+ // include continuations in the range that share the same textrun
+ nsTextFrame* next = nullptr;
+ if (aFollowContinuations && fend < aEndOffset) {
+ next = static_cast<nsTextFrame*>(curr->GetNextContinuation());
+ while (next && next->GetTextRun(nsTextFrame::eInflated) == textRun) {
+ fend = std::min(next->GetContentEnd(), aEndOffset);
+ next = fend < aEndOffset
+ ? static_cast<nsTextFrame*>(next->GetNextContinuation())
+ : nullptr;
+ }
+ }
+
+ if (!aSkipCollapsedWhitespace || (curr->HasAnyNoncollapsedCharacters() &&
+ curr->HasNonSuppressedText())) {
+ gfxTextRun::Range range(iter.ConvertOriginalToSkipped(fstart),
+ iter.ConvertOriginalToSkipped(fend));
+ AddFontsFromTextRun(textRun, curr, iter, range, aResult, aFontFaces,
+ aMaxRanges);
+ }
+
+ curr = next;
+ } while (aFollowContinuations && curr);
+}
+
+/* static */
+size_t nsLayoutUtils::SizeOfTextRunsForFrames(nsIFrame* aFrame,
+ MallocSizeOf aMallocSizeOf,
+ bool clear) {
+ MOZ_ASSERT(aFrame, "NULL frame pointer");
+
+ size_t total = 0;
+
+ if (aFrame->IsTextFrame()) {
+ nsTextFrame* textFrame = static_cast<nsTextFrame*>(aFrame);
+ for (uint32_t i = 0; i < 2; ++i) {
+ gfxTextRun* run = textFrame->GetTextRun(
+ (i != 0) ? nsTextFrame::eInflated : nsTextFrame::eNotInflated);
+ if (run) {
+ if (clear) {
+ run->ResetSizeOfAccountingFlags();
+ } else {
+ total += run->MaybeSizeOfIncludingThis(aMallocSizeOf);
+ }
+ }
+ }
+ return total;
+ }
+
+ for (const auto& childList : aFrame->ChildLists()) {
+ for (nsIFrame* f : childList.mList) {
+ total += SizeOfTextRunsForFrames(f, aMallocSizeOf, clear);
+ }
+ }
+ return total;
+}
+
+/* static */
+void nsLayoutUtils::RecomputeSmoothScrollDefault() {
+ if (nsContentUtils::ShouldResistFingerprinting(
+ "We use the global RFP pref to maintain consistent scroll behavior "
+ "in the browser.",
+ RFPTarget::CSSPrefersReducedMotion)) {
+ // When resist fingerprinting is enabled, we should not default disable
+ // smooth scrolls when the user prefers-reduced-motion to avoid leaking
+ // the value of the OS pref to sites.
+ Preferences::SetBool(StaticPrefs::GetPrefName_general_smoothScroll(), true,
+ PrefValueKind::Default);
+ } else {
+ // We want prefers-reduced-motion to determine the default
+ // value of the general.smoothScroll pref. If the user
+ // changed the pref we want to respect the change.
+ Preferences::SetBool(
+ StaticPrefs::GetPrefName_general_smoothScroll(),
+ !LookAndFeel::GetInt(LookAndFeel::IntID::PrefersReducedMotion, 0),
+ PrefValueKind::Default);
+ }
+}
+
+/* static */
+bool nsLayoutUtils::IsSmoothScrollingEnabled() {
+ return StaticPrefs::general_smoothScroll();
+}
+
+/* static */
+void nsLayoutUtils::Initialize() {
+ nsComputedDOMStyle::RegisterPrefChangeCallbacks();
+}
+
+/* static */
+void nsLayoutUtils::Shutdown() {
+ if (sContentMap) {
+ sContentMap = nullptr;
+ }
+
+ nsComputedDOMStyle::UnregisterPrefChangeCallbacks();
+}
+
+/* static */
+void nsLayoutUtils::RegisterImageRequest(nsPresContext* aPresContext,
+ imgIRequest* aRequest,
+ bool* aRequestRegistered) {
+ if (!aPresContext) {
+ return;
+ }
+
+ if (aRequestRegistered && *aRequestRegistered) {
+ // Our request is already registered with the refresh driver, so
+ // no need to register it again.
+ return;
+ }
+
+ if (aRequest) {
+ aPresContext->RefreshDriver()->AddImageRequest(aRequest);
+ if (aRequestRegistered) {
+ *aRequestRegistered = true;
+ }
+ }
+}
+
+/* static */
+void nsLayoutUtils::RegisterImageRequestIfAnimated(nsPresContext* aPresContext,
+ imgIRequest* aRequest,
+ bool* aRequestRegistered) {
+ if (!aPresContext) {
+ return;
+ }
+
+ if (aRequestRegistered && *aRequestRegistered) {
+ // Our request is already registered with the refresh driver, so
+ // no need to register it again.
+ return;
+ }
+
+ if (aRequest) {
+ nsCOMPtr<imgIContainer> image;
+ if (NS_SUCCEEDED(aRequest->GetImage(getter_AddRefs(image)))) {
+ // Check to verify that the image is animated. If so, then add it to the
+ // list of images tracked by the refresh driver.
+ bool isAnimated = false;
+ nsresult rv = image->GetAnimated(&isAnimated);
+ if (NS_SUCCEEDED(rv) && isAnimated) {
+ aPresContext->RefreshDriver()->AddImageRequest(aRequest);
+ if (aRequestRegistered) {
+ *aRequestRegistered = true;
+ }
+ }
+ }
+ }
+}
+
+/* static */
+void nsLayoutUtils::DeregisterImageRequest(nsPresContext* aPresContext,
+ imgIRequest* aRequest,
+ bool* aRequestRegistered) {
+ if (!aPresContext) {
+ return;
+ }
+
+ // Deregister our imgIRequest with the refresh driver to
+ // complete tear-down, but only if it has been registered
+ if (aRequestRegistered && !*aRequestRegistered) {
+ return;
+ }
+
+ if (aRequest) {
+ nsCOMPtr<imgIContainer> image;
+ if (NS_SUCCEEDED(aRequest->GetImage(getter_AddRefs(image)))) {
+ aPresContext->RefreshDriver()->RemoveImageRequest(aRequest);
+
+ if (aRequestRegistered) {
+ *aRequestRegistered = false;
+ }
+ }
+ }
+}
+
+/* static */
+void nsLayoutUtils::PostRestyleEvent(Element* aElement,
+ RestyleHint aRestyleHint,
+ nsChangeHint aMinChangeHint) {
+ if (Document* doc = aElement->GetComposedDoc()) {
+ if (nsPresContext* presContext = doc->GetPresContext()) {
+ presContext->RestyleManager()->PostRestyleEvent(aElement, aRestyleHint,
+ aMinChangeHint);
+ }
+ }
+}
+
+nsSetAttrRunnable::nsSetAttrRunnable(Element* aElement, nsAtom* aAttrName,
+ const nsAString& aValue)
+ : mozilla::Runnable("nsSetAttrRunnable"),
+ mElement(aElement),
+ mAttrName(aAttrName),
+ mValue(aValue) {
+ NS_ASSERTION(aElement && aAttrName, "Missing stuff, prepare to crash");
+}
+
+nsSetAttrRunnable::nsSetAttrRunnable(Element* aElement, nsAtom* aAttrName,
+ int32_t aValue)
+ : mozilla::Runnable("nsSetAttrRunnable"),
+ mElement(aElement),
+ mAttrName(aAttrName) {
+ NS_ASSERTION(aElement && aAttrName, "Missing stuff, prepare to crash");
+ mValue.AppendInt(aValue);
+}
+
+NS_IMETHODIMP
+nsSetAttrRunnable::Run() {
+ return mElement->SetAttr(kNameSpaceID_None, mAttrName, mValue, true);
+}
+
+nsUnsetAttrRunnable::nsUnsetAttrRunnable(Element* aElement, nsAtom* aAttrName)
+ : mozilla::Runnable("nsUnsetAttrRunnable"),
+ mElement(aElement),
+ mAttrName(aAttrName) {
+ NS_ASSERTION(aElement && aAttrName, "Missing stuff, prepare to crash");
+}
+
+NS_IMETHODIMP
+nsUnsetAttrRunnable::Run() {
+ return mElement->UnsetAttr(kNameSpaceID_None, mAttrName, true);
+}
+
+/**
+ * Compute the minimum font size inside of a container with the given
+ * width, such that **when the user zooms the container to fill the full
+ * width of the device**, the fonts satisfy our minima.
+ */
+static nscoord MinimumFontSizeFor(nsPresContext* aPresContext,
+ WritingMode aWritingMode,
+ nscoord aContainerISize) {
+ PresShell* presShell = aPresContext->PresShell();
+
+ uint32_t emPerLine = presShell->FontSizeInflationEmPerLine();
+ uint32_t minTwips = presShell->FontSizeInflationMinTwips();
+ if (emPerLine == 0 && minTwips == 0) {
+ return 0;
+ }
+
+ nscoord byLine = 0, byInch = 0;
+ if (emPerLine != 0) {
+ byLine = aContainerISize / emPerLine;
+ }
+ if (minTwips != 0) {
+ // REVIEW: Is this giving us app units and sizes *not* counting
+ // viewport scaling?
+ gfxSize screenSize = aPresContext->ScreenSizeInchesForFontInflation();
+ float deviceISizeInches =
+ aWritingMode.IsVertical() ? screenSize.height : screenSize.width;
+ byInch =
+ NSToCoordRound(aContainerISize / (deviceISizeInches * 1440 / minTwips));
+ }
+ return std::max(byLine, byInch);
+}
+
+/* static */
+float nsLayoutUtils::FontSizeInflationInner(const nsIFrame* aFrame,
+ nscoord aMinFontSize) {
+ // Note that line heights should be inflated by the same ratio as the
+ // font size of the same text; thus we operate only on the font size
+ // even when we're scaling a line height.
+ nscoord styleFontSize = aFrame->StyleFont()->mFont.size.ToAppUnits();
+ if (styleFontSize <= 0) {
+ // Never scale zero font size.
+ return 1.0;
+ }
+
+ if (aMinFontSize <= 0) {
+ // No need to scale.
+ return 1.0;
+ }
+
+ // If between this current frame and its font inflation container there is a
+ // non-inline element with fixed width or height, then we should not inflate
+ // fonts for this frame.
+ for (const nsIFrame* f = aFrame; f && !f->IsContainerForFontSizeInflation();
+ f = f->GetParent()) {
+ nsIContent* content = f->GetContent();
+ LayoutFrameType fType = f->Type();
+ nsIFrame* parent = f->GetParent();
+ // Also, if there is more than one frame corresponding to a single
+ // content node, we want the outermost one.
+ if (!(parent && parent->GetContent() == content) &&
+ // ignore width/height on inlines since they don't apply
+ fType != LayoutFrameType::Inline &&
+ // ignore width on radios and checkboxes since we enlarge them and
+ // they have width/height in ua.css
+ fType != LayoutFrameType::CheckboxRadio) {
+ // ruby annotations should have the same inflation as its
+ // grandparent, which is the ruby frame contains the annotation.
+ if (fType == LayoutFrameType::RubyText) {
+ MOZ_ASSERT(parent && parent->IsRubyTextContainerFrame());
+ nsIFrame* grandparent = parent->GetParent();
+ MOZ_ASSERT(grandparent && grandparent->IsRubyFrame());
+ return FontSizeInflationFor(grandparent);
+ }
+ WritingMode wm = f->GetWritingMode();
+ const auto& stylePosISize = f->StylePosition()->ISize(wm);
+ const auto& stylePosBSize = f->StylePosition()->BSize(wm);
+ if (!stylePosISize.IsAuto() ||
+ !stylePosBSize.BehavesLikeInitialValueOnBlockAxis()) {
+ return 1.0;
+ }
+ }
+ }
+
+ int32_t interceptParam = StaticPrefs::font_size_inflation_mappingIntercept();
+ float maxRatio = (float)StaticPrefs::font_size_inflation_maxRatio() / 100.0f;
+
+ float ratio = float(styleFontSize) / float(aMinFontSize);
+ float inflationRatio;
+
+ // Given a minimum inflated font size m, a specified font size s, we want to
+ // find the inflated font size i and then return the ratio of i to s (i/s).
+ if (interceptParam >= 0) {
+ // Since the mapping intercept parameter P is greater than zero, we use it
+ // to determine the point where our mapping function intersects the i=s
+ // line. This means that we have an equation of the form:
+ //
+ // i = m + s*(P/2)/(1 + P/2), if s <= (1 + P/2)*m
+ // i = s, if s >= (1 + P/2)*m
+
+ float intercept = 1 + float(interceptParam) / 2.0f;
+ if (ratio >= intercept) {
+ // If we're already at 1+P/2 or more times the minimum, don't scale.
+ return 1.0;
+ }
+
+ // The point (intercept, intercept) is where the part of the i vs. s graph
+ // that's not slope 1 meets the i=s line. (This part of the
+ // graph is a line from (0, m), to that point). We calculate the
+ // intersection point to be ((1+P/2)m, (1+P/2)m), where P is the
+ // intercept parameter above. We then need to return i/s.
+ inflationRatio = (1.0f + (ratio * (intercept - 1) / intercept)) / ratio;
+ } else {
+ // This is the case where P is negative. We essentially want to implement
+ // the case for P=infinity here, so we make i = s + m, which means that
+ // i/s = s/s + m/s = 1 + 1/ratio
+ inflationRatio = 1 + 1.0f / ratio;
+ }
+
+ if (maxRatio > 1.0 && inflationRatio > maxRatio) {
+ return maxRatio;
+ } else {
+ return inflationRatio;
+ }
+}
+
+static bool ShouldInflateFontsForContainer(const nsIFrame* aFrame) {
+ // We only want to inflate fonts for text that is in a place
+ // with room to expand. The question is what the best heuristic for
+ // that is...
+ // For now, we're going to use NS_FRAME_IN_CONSTRAINED_BSIZE, which
+ // indicates whether the frame is inside something with a constrained
+ // block-size (propagating down the tree), but the propagation stops when
+ // we hit overflow-y [or -x, for vertical mode]: scroll or auto.
+ const nsStyleText* styleText = aFrame->StyleText();
+
+ return styleText->mTextSizeAdjust != StyleTextSizeAdjust::None &&
+ !aFrame->HasAnyStateBits(NS_FRAME_IN_CONSTRAINED_BSIZE) &&
+ // We also want to disable font inflation for containers that have
+ // preformatted text.
+ // MathML cells need special treatment. See bug 1002526 comment 56.
+ (styleText->WhiteSpaceCanWrap(aFrame) ||
+ aFrame->IsFrameOfType(nsIFrame::eMathML));
+}
+
+nscoord nsLayoutUtils::InflationMinFontSizeFor(const nsIFrame* aFrame) {
+ nsPresContext* presContext = aFrame->PresContext();
+ if (!FontSizeInflationEnabled(presContext) ||
+ presContext->mInflationDisabledForShrinkWrap) {
+ return 0;
+ }
+
+ for (const nsIFrame* f = aFrame; f; f = f->GetParent()) {
+ if (f->IsContainerForFontSizeInflation()) {
+ if (!ShouldInflateFontsForContainer(f)) {
+ return 0;
+ }
+
+ nsFontInflationData* data =
+ nsFontInflationData::FindFontInflationDataFor(aFrame);
+ // FIXME: The need to null-check here is sort of a bug, and might
+ // lead to incorrect results.
+ if (!data || !data->InflationEnabled()) {
+ return 0;
+ }
+
+ return MinimumFontSizeFor(aFrame->PresContext(), aFrame->GetWritingMode(),
+ data->UsableISize());
+ }
+ }
+
+ MOZ_ASSERT(false, "root should always be container");
+
+ return 0;
+}
+
+float nsLayoutUtils::FontSizeInflationFor(const nsIFrame* aFrame) {
+ if (aFrame->IsInSVGTextSubtree()) {
+ const nsIFrame* container = aFrame;
+ while (!container->IsSVGTextFrame()) {
+ container = container->GetParent();
+ }
+ NS_ASSERTION(container, "expected to find an ancestor SVGTextFrame");
+ return static_cast<const SVGTextFrame*>(container)
+ ->GetFontSizeScaleFactor();
+ }
+
+ if (!FontSizeInflationEnabled(aFrame->PresContext())) {
+ return 1.0f;
+ }
+
+ return FontSizeInflationInner(aFrame, InflationMinFontSizeFor(aFrame));
+}
+
+/* static */
+bool nsLayoutUtils::FontSizeInflationEnabled(nsPresContext* aPresContext) {
+ PresShell* presShell = aPresContext->GetPresShell();
+ if (!presShell) {
+ return false;
+ }
+ return presShell->FontSizeInflationEnabled();
+}
+
+/* static */
+nsRect nsLayoutUtils::GetBoxShadowRectForFrame(nsIFrame* aFrame,
+ const nsSize& aFrameSize) {
+ auto boxShadows = aFrame->StyleEffects()->mBoxShadow.AsSpan();
+ if (boxShadows.IsEmpty()) {
+ return nsRect();
+ }
+
+ nsRect inputRect(nsPoint(0, 0), aFrameSize);
+
+ // According to the CSS spec, box-shadow should be based on the border box.
+ // However, that looks broken when the background extends outside the border
+ // box, as can be the case with native theming. To fix that we expand the
+ // area that we shadow to include the bounds of any native theme drawing.
+ const nsStyleDisplay* styleDisplay = aFrame->StyleDisplay();
+ nsITheme::Transparency transparency;
+ if (aFrame->IsThemed(styleDisplay, &transparency)) {
+ // For opaque (rectangular) theme widgets we can take the generic
+ // border-box path with border-radius disabled.
+ if (transparency != nsITheme::eOpaque) {
+ nsPresContext* presContext = aFrame->PresContext();
+ presContext->Theme()->GetWidgetOverflow(
+ presContext->DeviceContext(), aFrame,
+ styleDisplay->EffectiveAppearance(), &inputRect);
+ }
+ }
+
+ nsRect shadows;
+ int32_t A2D = aFrame->PresContext()->AppUnitsPerDevPixel();
+ for (auto& shadow : boxShadows) {
+ nsRect tmpRect = inputRect;
+
+ // inset shadows are never painted outside the frame
+ if (shadow.inset) {
+ continue;
+ }
+
+ tmpRect.MoveBy(nsPoint(shadow.base.horizontal.ToAppUnits(),
+ shadow.base.vertical.ToAppUnits()));
+ tmpRect.Inflate(shadow.spread.ToAppUnits());
+ tmpRect.Inflate(nsContextBoxBlur::GetBlurRadiusMargin(
+ shadow.base.blur.ToAppUnits(), A2D));
+ shadows.UnionRect(shadows, tmpRect);
+ }
+ return shadows;
+}
+
+/* static */
+bool nsLayoutUtils::GetContentViewerSize(
+ const nsPresContext* aPresContext, LayoutDeviceIntSize& aOutSize,
+ SubtractDynamicToolbar aSubtractDynamicToolbar) {
+ nsCOMPtr<nsIDocShell> docShell = aPresContext->GetDocShell();
+ if (!docShell) {
+ return false;
+ }
+
+ nsCOMPtr<nsIContentViewer> cv;
+ docShell->GetContentViewer(getter_AddRefs(cv));
+ if (!cv) {
+ return false;
+ }
+
+ nsIntRect bounds;
+ cv->GetBounds(bounds);
+
+ if (aPresContext->IsRootContentDocumentCrossProcess() &&
+ aSubtractDynamicToolbar == SubtractDynamicToolbar::Yes &&
+ aPresContext->HasDynamicToolbar() && !bounds.IsEmpty()) {
+ MOZ_ASSERT(aPresContext->IsRootContentDocumentCrossProcess());
+ bounds.height -= aPresContext->GetDynamicToolbarMaxHeight();
+ // Collapse the size in the case the dynamic toolbar max height is greater
+ // than the content bound height so that hopefully embedders of GeckoView
+ // may notice they set wrong dynamic toolbar max height.
+ if (bounds.height < 0) {
+ bounds.height = 0;
+ }
+ }
+
+ aOutSize = LayoutDeviceIntRect::FromUnknownRect(bounds).Size();
+ return true;
+}
+
+bool nsLayoutUtils::UpdateCompositionBoundsForRCDRSF(
+ ParentLayerRect& aCompBounds, const nsPresContext* aPresContext,
+ IncludeDynamicToolbar aIncludeDynamicToolbar) {
+ SubtractDynamicToolbar shouldSubtractDynamicToolbar =
+ aIncludeDynamicToolbar == IncludeDynamicToolbar::Force
+ ? SubtractDynamicToolbar::No
+ : aPresContext->IsRootContentDocumentCrossProcess() &&
+ aPresContext->HasDynamicToolbar()
+ ? SubtractDynamicToolbar::Yes
+ : SubtractDynamicToolbar::No;
+
+ if (shouldSubtractDynamicToolbar == SubtractDynamicToolbar::Yes) {
+ if (RefPtr<MobileViewportManager> MVM =
+ aPresContext->PresShell()->GetMobileViewportManager()) {
+ // Convert the intrinsic composition size to app units here since
+ // the returned size of below CalculateScrollableRectForFrame call has
+ // been already converted/rounded to app units.
+ nsSize intrinsicCompositionSize =
+ CSSSize::ToAppUnits(MVM->GetIntrinsicCompositionSize());
+
+ if (nsIScrollableFrame* rootScrollableFrame =
+ aPresContext->PresShell()->GetRootScrollFrameAsScrollable()) {
+ // Expand the composition size to include the area initially covered by
+ // the dynamic toolbar only if the content is taller than the intrinsic
+ // composition size (i.e. the dynamic toolbar should be able to move
+ // only if the content is vertically scrollable).
+ if (intrinsicCompositionSize.height <
+ CalculateScrollableRectForFrame(rootScrollableFrame, nullptr)
+ .Height()) {
+ shouldSubtractDynamicToolbar = SubtractDynamicToolbar::No;
+ }
+ }
+ }
+ }
+
+ LayoutDeviceIntSize contentSize;
+ if (!GetContentViewerSize(aPresContext, contentSize,
+ shouldSubtractDynamicToolbar)) {
+ return false;
+ }
+ aCompBounds.SizeTo(ViewAs<ParentLayerPixel>(
+ LayoutDeviceSize(contentSize),
+ PixelCastJustification::LayoutDeviceIsParentLayerForRCDRSF));
+ return true;
+}
+
+/* static */
+nsMargin nsLayoutUtils::ScrollbarAreaToExcludeFromCompositionBoundsFor(
+ const nsIFrame* aScrollFrame) {
+ if (!aScrollFrame || !aScrollFrame->GetScrollTargetFrame()) {
+ return nsMargin();
+ }
+ nsPresContext* presContext = aScrollFrame->PresContext();
+ PresShell* presShell = presContext->GetPresShell();
+ if (!presShell) {
+ return nsMargin();
+ }
+ bool isRootScrollFrame = aScrollFrame == presShell->GetRootScrollFrame();
+ bool isRootContentDocRootScrollFrame =
+ isRootScrollFrame && presContext->IsRootContentDocumentCrossProcess();
+ if (!isRootContentDocRootScrollFrame) {
+ return nsMargin();
+ }
+ if (presContext->UseOverlayScrollbars()) {
+ return nsMargin();
+ }
+ nsIScrollableFrame* scrollableFrame = aScrollFrame->GetScrollTargetFrame();
+ if (!scrollableFrame) {
+ return nsMargin();
+ }
+ return scrollableFrame->GetActualScrollbarSizes(
+ nsIScrollableFrame::ScrollbarSizesOptions::
+ INCLUDE_VISUAL_VIEWPORT_SCROLLBARS);
+}
+
+/* static */
+nsSize nsLayoutUtils::CalculateCompositionSizeForFrame(
+ nsIFrame* aFrame, bool aSubtractScrollbars,
+ const nsSize* aOverrideScrollPortSize,
+ IncludeDynamicToolbar aIncludeDynamicToolbar) {
+ // If we have a scrollable frame, restrict the composition bounds to its
+ // scroll port. The scroll port excludes the frame borders and the scroll
+ // bars, which we don't want to be part of the composition bounds.
+ nsIScrollableFrame* scrollableFrame = aFrame->GetScrollTargetFrame();
+ nsRect rect = scrollableFrame ? scrollableFrame->GetScrollPortRect()
+ : aFrame->GetRect();
+ nsSize size =
+ aOverrideScrollPortSize ? *aOverrideScrollPortSize : rect.Size();
+
+ nsPresContext* presContext = aFrame->PresContext();
+ PresShell* presShell = presContext->PresShell();
+
+ bool isRootContentDocRootScrollFrame =
+ presContext->IsRootContentDocumentCrossProcess() &&
+ aFrame == presShell->GetRootScrollFrame();
+ if (isRootContentDocRootScrollFrame) {
+ ParentLayerRect compBounds;
+ if (UpdateCompositionBoundsForRCDRSF(compBounds, presContext,
+ aIncludeDynamicToolbar)) {
+ int32_t auPerDevPixel = presContext->AppUnitsPerDevPixel();
+ size = nsSize(compBounds.width * auPerDevPixel,
+ compBounds.height * auPerDevPixel);
+ }
+ }
+
+ if (aSubtractScrollbars) {
+ nsMargin margins = ScrollbarAreaToExcludeFromCompositionBoundsFor(aFrame);
+ size.width -= margins.LeftRight();
+ size.height -= margins.TopBottom();
+ }
+
+ return size;
+}
+
+/* static */
+CSSSize nsLayoutUtils::CalculateBoundingCompositionSize(
+ const nsIFrame* aFrame, bool aIsRootContentDocRootScrollFrame,
+ const FrameMetrics& aMetrics) {
+ if (aIsRootContentDocRootScrollFrame) {
+ return ViewAs<LayerPixel>(
+ aMetrics.GetCompositionBounds().Size(),
+ PixelCastJustification::ParentLayerToLayerForRootComposition) *
+ LayerToScreenScale(1.0f) / aMetrics.DisplayportPixelsPerCSSPixel();
+ }
+ nsPresContext* presContext = aFrame->PresContext();
+ ScreenSize rootCompositionSize;
+ nsPresContext* rootPresContext =
+ presContext->GetInProcessRootContentDocumentPresContext();
+ if (!rootPresContext) {
+ rootPresContext = presContext->GetRootPresContext();
+ }
+ PresShell* rootPresShell = nullptr;
+ if (rootPresContext) {
+ rootPresShell = rootPresContext->PresShell();
+ if (nsIFrame* rootFrame = rootPresShell->GetRootFrame()) {
+ ParentLayerRect compBounds;
+ if (UpdateCompositionBoundsForRCDRSF(compBounds, rootPresContext)) {
+ rootCompositionSize = ViewAs<ScreenPixel>(
+ compBounds.Size(),
+ PixelCastJustification::ScreenIsParentLayerForRoot);
+ } else {
+ // LayoutDeviceToScreenScale2D =
+ // LayoutDeviceToParentLayerScale *
+ // ParentLayerToScreenScale2D
+ LayoutDeviceToScreenScale2D cumulativeResolution =
+ LayoutDeviceToParentLayerScale(
+ rootPresShell->GetCumulativeResolution()) *
+ GetTransformToAncestorScaleCrossProcessForFrameMetrics(rootFrame);
+
+ int32_t rootAUPerDevPixel = rootPresContext->AppUnitsPerDevPixel();
+ rootCompositionSize = (LayoutDeviceRect::FromAppUnits(
+ rootFrame->GetRect(), rootAUPerDevPixel) *
+ cumulativeResolution)
+ .Size();
+ }
+ }
+ } else {
+ nsIWidget* widget = aFrame->GetNearestWidget();
+ LayoutDeviceIntRect widgetBounds = widget->GetBounds();
+ rootCompositionSize = ScreenSize(ViewAs<ScreenPixel>(
+ widgetBounds.Size(),
+ PixelCastJustification::LayoutDeviceIsScreenForBounds));
+ }
+
+ // Adjust composition size for the size of scroll bars.
+ nsIFrame* rootRootScrollFrame =
+ rootPresShell ? rootPresShell->GetRootScrollFrame() : nullptr;
+ nsMargin scrollbarMargins =
+ ScrollbarAreaToExcludeFromCompositionBoundsFor(rootRootScrollFrame);
+ LayoutDeviceMargin margins = LayoutDeviceMargin::FromAppUnits(
+ scrollbarMargins, rootPresContext->AppUnitsPerDevPixel());
+ // Scrollbars are not subject to resolution scaling, so LD pixels = layer
+ // pixels for them.
+ rootCompositionSize.width -= margins.LeftRight();
+ rootCompositionSize.height -= margins.TopBottom();
+
+ CSSSize result =
+ rootCompositionSize / aMetrics.DisplayportPixelsPerCSSPixel();
+
+ // If this is a nested content process, the in-process root content document's
+ // composition size may still be arbitrarily large, so bound it further by
+ // how much of the in-process RCD is visible in the top-level (cross-process
+ // RCD) viewport.
+ if (rootPresShell) {
+ if (BrowserChild* bc = BrowserChild::GetFrom(rootPresShell)) {
+ if (const auto& visibleRect =
+ bc->GetTopLevelViewportVisibleRectInSelfCoords()) {
+ CSSSize cssVisibleRect =
+ visibleRect->Size() / rootPresContext->CSSToDevPixelScale();
+ result = Min(result, cssVisibleRect);
+ }
+ }
+ }
+
+ return result;
+}
+
+/* static */
+nsRect nsLayoutUtils::CalculateScrollableRectForFrame(
+ const nsIScrollableFrame* aScrollableFrame, const nsIFrame* aRootFrame) {
+ nsRect contentBounds;
+ if (aScrollableFrame) {
+ contentBounds = aScrollableFrame->GetScrollRange();
+
+ nsPoint scrollPosition = aScrollableFrame->GetScrollPosition();
+ if (aScrollableFrame->GetScrollStyles().mVertical ==
+ StyleOverflow::Hidden) {
+ contentBounds.y = scrollPosition.y;
+ contentBounds.height = 0;
+ }
+ if (aScrollableFrame->GetScrollStyles().mHorizontal ==
+ StyleOverflow::Hidden) {
+ contentBounds.x = scrollPosition.x;
+ contentBounds.width = 0;
+ }
+
+ contentBounds.width += aScrollableFrame->GetScrollPortRect().width;
+ contentBounds.height += aScrollableFrame->GetScrollPortRect().height;
+ } else {
+ contentBounds = aRootFrame->GetRect();
+ // Clamp to (0, 0) if there is no corresponding scrollable frame for the
+ // given |aRootFrame|.
+ contentBounds.MoveTo(0, 0);
+ }
+ return contentBounds;
+}
+
+/* static */
+nsRect nsLayoutUtils::CalculateExpandedScrollableRect(nsIFrame* aFrame) {
+ nsRect scrollableRect = CalculateScrollableRectForFrame(
+ aFrame->GetScrollTargetFrame(), aFrame->PresShell()->GetRootFrame());
+ nsSize compSize = CalculateCompositionSizeForFrame(aFrame);
+
+ if (aFrame == aFrame->PresShell()->GetRootScrollFrame()) {
+ // the composition size for the root scroll frame does not include the
+ // local resolution, so we adjust.
+ float res = aFrame->PresShell()->GetResolution();
+ compSize.width = NSToCoordRound(compSize.width / res);
+ compSize.height = NSToCoordRound(compSize.height / res);
+ }
+
+ if (scrollableRect.width < compSize.width) {
+ scrollableRect.x =
+ std::max(0, scrollableRect.x - (compSize.width - scrollableRect.width));
+ scrollableRect.width = compSize.width;
+ }
+
+ if (scrollableRect.height < compSize.height) {
+ scrollableRect.y = std::max(
+ 0, scrollableRect.y - (compSize.height - scrollableRect.height));
+ scrollableRect.height = compSize.height;
+ }
+ return scrollableRect;
+}
+
+/* static */
+void nsLayoutUtils::DoLogTestDataForPaint(WebRenderLayerManager* aManager,
+ ViewID aScrollId,
+ const std::string& aKey,
+ const std::string& aValue) {
+ MOZ_ASSERT(nsLayoutUtils::IsAPZTestLoggingEnabled(), "don't call me");
+ aManager->LogTestDataForCurrentPaint(aScrollId, aKey, aValue);
+}
+
+void nsLayoutUtils::LogAdditionalTestData(nsDisplayListBuilder* aBuilder,
+ const std::string& aKey,
+ const std::string& aValue) {
+ WebRenderLayerManager* manager = aBuilder->GetWidgetLayerManager(nullptr);
+ if (!manager) {
+ return;
+ }
+ manager->LogAdditionalTestData(aKey, aValue);
+}
+
+/* static */
+bool nsLayoutUtils::IsAPZTestLoggingEnabled() {
+ return StaticPrefs::apz_test_logging_enabled();
+}
+
+////////////////////////////////////////
+// SurfaceFromElementResult
+
+SurfaceFromElementResult::SurfaceFromElementResult()
+ // Use safe default values here
+ : mHadCrossOriginRedirects(false),
+ mIsWriteOnly(true),
+ mIsStillLoading(false),
+ mHasSize(false),
+ mCORSUsed(false),
+ mAlphaType(gfxAlphaType::Opaque) {}
+
+const RefPtr<mozilla::gfx::SourceSurface>&
+SurfaceFromElementResult::GetSourceSurface() {
+ if (!mSourceSurface && mLayersImage) {
+ mSourceSurface = mLayersImage->GetAsSourceSurface();
+ }
+
+ return mSourceSurface;
+}
+
+////////////////////////////////////////
+
+bool nsLayoutUtils::IsNonWrapperBlock(nsIFrame* aFrame) {
+ MOZ_ASSERT(aFrame);
+ return aFrame->IsBlockFrameOrSubclass() && !aFrame->IsBlockWrapper();
+}
+
+AutoMaybeDisableFontInflation::AutoMaybeDisableFontInflation(nsIFrame* aFrame) {
+ // FIXME: Now that inflation calculations are based on the flow
+ // root's NCA's (nearest common ancestor of its inflatable
+ // descendants) width, we could probably disable inflation in
+ // fewer cases than we currently do.
+ // MathML cells need special treatment. See bug 1002526 comment 56.
+ if (aFrame->IsContainerForFontSizeInflation() &&
+ !aFrame->IsFrameOfType(nsIFrame::eMathML)) {
+ mPresContext = aFrame->PresContext();
+ mOldValue = mPresContext->mInflationDisabledForShrinkWrap;
+ mPresContext->mInflationDisabledForShrinkWrap = true;
+ } else {
+ // indicate we have nothing to restore
+ mPresContext = nullptr;
+ mOldValue = false;
+ }
+}
+
+AutoMaybeDisableFontInflation::~AutoMaybeDisableFontInflation() {
+ if (mPresContext) {
+ mPresContext->mInflationDisabledForShrinkWrap = mOldValue;
+ }
+}
+
+namespace mozilla {
+
+Rect NSRectToRect(const nsRect& aRect, double aAppUnitsPerPixel) {
+ // Note that by making aAppUnitsPerPixel a double we're doing floating-point
+ // division using a larger type and avoiding rounding error.
+ return Rect(Float(aRect.x / aAppUnitsPerPixel),
+ Float(aRect.y / aAppUnitsPerPixel),
+ Float(aRect.width / aAppUnitsPerPixel),
+ Float(aRect.height / aAppUnitsPerPixel));
+}
+
+Rect NSRectToSnappedRect(const nsRect& aRect, double aAppUnitsPerPixel,
+ const gfx::DrawTarget& aSnapDT) {
+ // Note that by making aAppUnitsPerPixel a double we're doing floating-point
+ // division using a larger type and avoiding rounding error.
+ Rect rect(Float(aRect.x / aAppUnitsPerPixel),
+ Float(aRect.y / aAppUnitsPerPixel),
+ Float(aRect.width / aAppUnitsPerPixel),
+ Float(aRect.height / aAppUnitsPerPixel));
+ MaybeSnapToDevicePixels(rect, aSnapDT, true);
+ return rect;
+}
+// Similar to a snapped rect, except an axis is left unsnapped if the snapping
+// process results in a length of 0.
+Rect NSRectToNonEmptySnappedRect(const nsRect& aRect, double aAppUnitsPerPixel,
+ const gfx::DrawTarget& aSnapDT) {
+ // Note that by making aAppUnitsPerPixel a double we're doing floating-point
+ // division using a larger type and avoiding rounding error.
+ Rect rect(Float(aRect.x / aAppUnitsPerPixel),
+ Float(aRect.y / aAppUnitsPerPixel),
+ Float(aRect.width / aAppUnitsPerPixel),
+ Float(aRect.height / aAppUnitsPerPixel));
+ MaybeSnapToDevicePixels(rect, aSnapDT, true, false);
+ return rect;
+}
+
+void StrokeLineWithSnapping(const nsPoint& aP1, const nsPoint& aP2,
+ int32_t aAppUnitsPerDevPixel,
+ DrawTarget& aDrawTarget, const Pattern& aPattern,
+ const StrokeOptions& aStrokeOptions,
+ const DrawOptions& aDrawOptions) {
+ Point p1 = NSPointToPoint(aP1, aAppUnitsPerDevPixel);
+ Point p2 = NSPointToPoint(aP2, aAppUnitsPerDevPixel);
+ SnapLineToDevicePixelsForStroking(p1, p2, aDrawTarget,
+ aStrokeOptions.mLineWidth);
+ aDrawTarget.StrokeLine(p1, p2, aPattern, aStrokeOptions, aDrawOptions);
+}
+
+} // namespace mozilla
+
+/* static */
+void nsLayoutUtils::SetBSizeFromFontMetrics(const nsIFrame* aFrame,
+ ReflowOutput& aMetrics,
+ const LogicalMargin& aFramePadding,
+ WritingMode aLineWM,
+ WritingMode aFrameWM) {
+ RefPtr<nsFontMetrics> fm =
+ nsLayoutUtils::GetInflatedFontMetricsForFrame(aFrame);
+
+ if (fm) {
+ // Compute final height of the frame.
+ //
+ // Do things the standard css2 way -- though it's hard to find it
+ // in the css2 spec! It's actually found in the css1 spec section
+ // 4.4 (you will have to read between the lines to really see
+ // it).
+ //
+ // The height of our box is the sum of our font size plus the top
+ // and bottom border and padding. The height of children do not
+ // affect our height.
+ aMetrics.SetBlockStartAscent(
+ aLineWM.IsAlphabeticalBaseline()
+ ? aLineWM.IsLineInverted() ? fm->MaxDescent() : fm->MaxAscent()
+ : fm->MaxHeight() / 2);
+ aMetrics.BSize(aLineWM) = fm->MaxHeight();
+ } else {
+ NS_WARNING("Cannot get font metrics - defaulting sizes to 0");
+ aMetrics.SetBlockStartAscent(aMetrics.BSize(aLineWM) = 0);
+ }
+ aMetrics.SetBlockStartAscent(aMetrics.BlockStartAscent() +
+ aFramePadding.BStart(aFrameWM));
+ aMetrics.BSize(aLineWM) += aFramePadding.BStartEnd(aFrameWM);
+}
+
+/* static */
+// _BOUNDARY because Dispatch() with `targets` must not handle the event.
+MOZ_CAN_RUN_SCRIPT_BOUNDARY bool
+nsLayoutUtils::HasDocumentLevelListenersForApzAwareEvents(
+ PresShell* aPresShell) {
+ if (Document* doc = aPresShell->GetDocument()) {
+ WidgetEvent event(true, eVoidEvent);
+ nsTArray<EventTarget*> targets;
+ // TODO: Bug 1506441
+ nsresult rv =
+ EventDispatcher::Dispatch(MOZ_KnownLive(ToSupports(doc)), nullptr,
+ &event, nullptr, nullptr, nullptr, &targets);
+ NS_ENSURE_SUCCESS(rv, false);
+ for (size_t i = 0; i < targets.Length(); i++) {
+ if (targets[i]->IsApzAware()) {
+ return true;
+ }
+ }
+ }
+ return false;
+}
+
+/* static */
+bool nsLayoutUtils::CanScrollOriginClobberApz(ScrollOrigin aScrollOrigin) {
+ switch (aScrollOrigin) {
+ case ScrollOrigin::None:
+ case ScrollOrigin::NotSpecified:
+ case ScrollOrigin::Apz:
+ case ScrollOrigin::Restore:
+ return false;
+ default:
+ return true;
+ }
+}
+
+/* static */
+ScrollMetadata nsLayoutUtils::ComputeScrollMetadata(
+ const nsIFrame* aForFrame, const nsIFrame* aScrollFrame,
+ nsIContent* aContent, const nsIFrame* aItemFrame,
+ const nsPoint& aOffsetToReferenceFrame,
+ WebRenderLayerManager* aLayerManager, ViewID aScrollParentId,
+ const nsSize& aScrollPortSize, bool aIsRootContent) {
+ const nsPresContext* presContext = aForFrame->PresContext();
+ int32_t auPerDevPixel = presContext->AppUnitsPerDevPixel();
+
+ PresShell* presShell = presContext->GetPresShell();
+ ScrollMetadata metadata;
+ FrameMetrics& metrics = metadata.GetMetrics();
+ metrics.SetLayoutViewport(
+ CSSRect(CSSPoint(), CSSSize::FromAppUnits(aScrollPortSize)));
+
+ nsIDocShell* docShell = presContext->GetDocShell();
+ const BrowsingContext* bc =
+ docShell ? docShell->GetBrowsingContext() : nullptr;
+ bool isTouchEventsEnabled =
+ bc &&
+ bc->TouchEventsOverride() == mozilla::dom::TouchEventsOverride::Enabled;
+
+ if (bc && bc->InRDMPane() && isTouchEventsEnabled) {
+ metadata.SetIsRDMTouchSimulationActive(true);
+ }
+
+ ViewID scrollId = ScrollableLayerGuid::NULL_SCROLL_ID;
+ if (aContent) {
+ if (void* paintRequestTime =
+ aContent->GetProperty(nsGkAtoms::paintRequestTime)) {
+ metrics.SetPaintRequestTime(*static_cast<TimeStamp*>(paintRequestTime));
+ aContent->RemoveProperty(nsGkAtoms::paintRequestTime);
+ }
+ scrollId = nsLayoutUtils::FindOrCreateIDFor(aContent);
+ nsRect dp;
+ if (DisplayPortUtils::GetDisplayPort(aContent, &dp)) {
+ metrics.SetDisplayPort(CSSRect::FromAppUnits(dp));
+ DisplayPortUtils::MarkDisplayPortAsPainted(aContent);
+ }
+
+ metrics.SetHasNonZeroDisplayPortMargins(false);
+ if (DisplayPortMarginsPropertyData* currentData =
+ static_cast<DisplayPortMarginsPropertyData*>(
+ aContent->GetProperty(nsGkAtoms::DisplayPortMargins))) {
+ if (currentData->mMargins.mMargins != ScreenMargin()) {
+ metrics.SetHasNonZeroDisplayPortMargins(true);
+ }
+ }
+
+ // Note: GetProperty() will return nullptr both in the case where
+ // the property hasn't been set, and in the case where the property
+ // has been set to false (in which case the property value is
+ // `reinterpret_cast<void*>(false)` which is nullptr.
+ if (aContent->GetProperty(nsGkAtoms::forceMousewheelAutodir)) {
+ metadata.SetForceMousewheelAutodir(true);
+ }
+
+ if (aContent->GetProperty(nsGkAtoms::forceMousewheelAutodirHonourRoot)) {
+ metadata.SetForceMousewheelAutodirHonourRoot(true);
+ }
+
+ if (IsAPZTestLoggingEnabled()) {
+ LogTestDataForPaint(aLayerManager, scrollId, "displayport",
+ metrics.GetDisplayPort());
+ }
+
+ metrics.SetMinimalDisplayPort(
+ aContent->GetProperty(nsGkAtoms::MinimalDisplayPort));
+ }
+
+ nsIScrollableFrame* scrollableFrame = nullptr;
+ if (aScrollFrame) scrollableFrame = aScrollFrame->GetScrollTargetFrame();
+
+ metrics.SetScrollableRect(
+ CSSRect::FromAppUnits(nsLayoutUtils::CalculateScrollableRectForFrame(
+ scrollableFrame, aForFrame)));
+
+ if (scrollableFrame) {
+ CSSPoint layoutScrollOffset =
+ CSSPoint::FromAppUnits(scrollableFrame->GetScrollPosition());
+ CSSPoint visualScrollOffset =
+ aIsRootContent
+ ? CSSPoint::FromAppUnits(presShell->GetVisualViewportOffset())
+ : layoutScrollOffset;
+ metrics.SetVisualScrollOffset(visualScrollOffset);
+ // APZ sometimes reads this even if we haven't set a visual scroll
+ // update type (specifically, in the isFirstPaint case), so always
+ // set it.
+ metrics.SetVisualDestination(visualScrollOffset);
+
+ if (aIsRootContent) {
+ if (aLayerManager->GetIsFirstPaint() &&
+ presShell->IsVisualViewportOffsetSet()) {
+ // Restore the visual viewport offset to the copy stored on the
+ // main thread.
+ presShell->ScrollToVisual(presShell->GetVisualViewportOffset(),
+ FrameMetrics::eRestore, ScrollMode::Instant);
+ }
+ }
+
+ if (scrollableFrame->IsRootScrollFrameOfDocument()) {
+ if (const Maybe<PresShell::VisualScrollUpdate>& visualUpdate =
+ presShell->GetPendingVisualScrollUpdate()) {
+ metrics.SetVisualDestination(
+ CSSPoint::FromAppUnits(visualUpdate->mVisualScrollOffset));
+ metrics.SetVisualScrollUpdateType(visualUpdate->mUpdateType);
+ presShell->AcknowledgePendingVisualScrollUpdate();
+ }
+ }
+
+ if (aIsRootContent) {
+ // Expand the layout viewport to the size including the area covered by
+ // the dynamic toolbar in the case where the dynamic toolbar is being
+ // used, otherwise when the dynamic toolbar transitions on the compositor,
+ // the layout viewport will be smaller than the visual viewport on the
+ // compositor, thus the layout viewport offset will be forced to be moved
+ // in FrameMetrics::KeepLayoutViewportEnclosingVisualViewport.
+ if (presContext->HasDynamicToolbar()) {
+ CSSRect viewport = metrics.GetLayoutViewport();
+ viewport.SizeTo(nsLayoutUtils::ExpandHeightForDynamicToolbar(
+ presContext, viewport.Size()));
+ metrics.SetLayoutViewport(viewport);
+
+ // We need to set 'fixed margins' to adjust 'fixed margins' value on the
+ // composiutor in the case where the dynamic toolbar is completely
+ // hidden because the margin value on the compositor is offset from the
+ // position where the dynamic toolbar is completely VISIBLE but now the
+ // toolbar is completely HIDDEN we need to adjust the difference on the
+ // compositor.
+ if (presContext->GetDynamicToolbarState() ==
+ DynamicToolbarState::Collapsed) {
+ metrics.SetFixedLayerMargins(
+ ScreenMargin(0, 0,
+ presContext->GetDynamicToolbarHeight() -
+ presContext->GetDynamicToolbarMaxHeight(),
+ 0));
+ }
+ }
+ }
+
+ metrics.SetScrollGeneration(scrollableFrame->CurrentScrollGeneration());
+
+ CSSRect viewport = metrics.GetLayoutViewport();
+ viewport.MoveTo(layoutScrollOffset);
+ metrics.SetLayoutViewport(viewport);
+
+ nsSize lineScrollAmount = scrollableFrame->GetLineScrollAmount();
+ LayoutDeviceIntSize lineScrollAmountInDevPixels =
+ LayoutDeviceIntSize::FromAppUnitsRounded(
+ lineScrollAmount, presContext->AppUnitsPerDevPixel());
+ metadata.SetLineScrollAmount(lineScrollAmountInDevPixels);
+
+ nsSize pageScrollAmount = scrollableFrame->GetPageScrollAmount();
+ LayoutDeviceIntSize pageScrollAmountInDevPixels =
+ LayoutDeviceIntSize::FromAppUnitsRounded(
+ pageScrollAmount, presContext->AppUnitsPerDevPixel());
+ metadata.SetPageScrollAmount(pageScrollAmountInDevPixels);
+
+ if (aScrollFrame->GetParent()) {
+ metadata.SetDisregardedDirection(
+ WheelHandlingUtils::GetDisregardedWheelScrollDirection(
+ aScrollFrame->GetParent()));
+ }
+
+ metadata.SetSnapInfo(scrollableFrame->GetScrollSnapInfo());
+ metadata.SetOverscrollBehavior(
+ scrollableFrame->GetOverscrollBehaviorInfo());
+ metadata.SetScrollUpdates(scrollableFrame->GetScrollUpdates());
+ }
+
+ // If we have the scrollparent being the same as the scroll id, the
+ // compositor-side code could get into an infinite loop while building the
+ // overscroll handoff chain.
+ MOZ_ASSERT(aScrollParentId == ScrollableLayerGuid::NULL_SCROLL_ID ||
+ scrollId != aScrollParentId);
+ metrics.SetScrollId(scrollId);
+ metrics.SetIsRootContent(aIsRootContent);
+ metadata.SetScrollParentId(aScrollParentId);
+
+ const nsIFrame* rootScrollFrame = presShell->GetRootScrollFrame();
+ bool isRootScrollFrame = aScrollFrame == rootScrollFrame;
+ Document* document = presShell->GetDocument();
+
+ if (scrollId != ScrollableLayerGuid::NULL_SCROLL_ID &&
+ !presContext->GetParentPresContext()) {
+ if ((aScrollFrame && isRootScrollFrame)) {
+ metadata.SetIsLayersIdRoot(true);
+ } else {
+ MOZ_ASSERT(document, "A non-root-scroll frame must be in a document");
+ if (aContent == document->GetDocumentElement()) {
+ metadata.SetIsLayersIdRoot(true);
+ }
+ }
+ }
+
+ // Get whether the root content is RTL(E.g. it's true either if
+ // "writing-mode: vertical-rl", or if
+ // "writing-mode: horizontal-tb; direction: rtl;" in CSS).
+ // For the concept of this and the reason why we need to get this kind of
+ // information, see the definition of |mIsAutoDirRootContentRTL| in struct
+ // |ScrollMetadata|.
+ const Element* bodyElement = document ? document->GetBodyElement() : nullptr;
+ const nsIFrame* primaryFrame =
+ bodyElement ? bodyElement->GetPrimaryFrame() : rootScrollFrame;
+ if (!primaryFrame) {
+ primaryFrame = rootScrollFrame;
+ }
+ if (primaryFrame) {
+ WritingMode writingModeOfRootScrollFrame = primaryFrame->GetWritingMode();
+ WritingMode::BlockDir blockDirOfRootScrollFrame =
+ writingModeOfRootScrollFrame.GetBlockDir();
+ WritingMode::InlineDir inlineDirOfRootScrollFrame =
+ writingModeOfRootScrollFrame.GetInlineDir();
+ if (blockDirOfRootScrollFrame == WritingMode::BlockDir::eBlockRL ||
+ (blockDirOfRootScrollFrame == WritingMode::BlockDir::eBlockTB &&
+ inlineDirOfRootScrollFrame == WritingMode::InlineDir::eInlineRTL)) {
+ metadata.SetIsAutoDirRootContentRTL(true);
+ }
+ }
+
+ // Only the root scrollable frame for a given presShell should pick up
+ // the presShell's resolution. All the other frames are 1.0.
+ if (isRootScrollFrame) {
+ metrics.SetPresShellResolution(presShell->GetResolution());
+ } else {
+ metrics.SetPresShellResolution(1.0f);
+ }
+
+ if (presShell->IsResolutionUpdated()) {
+ metadata.SetResolutionUpdated(true);
+ }
+
+ // The cumulative resolution is the resolution at which the scroll frame's
+ // content is actually rendered. It includes the pres shell resolutions of
+ // all the pres shells from here up to the root, as well as any css-driven
+ // resolution. We don't need to compute it as it's already stored in the
+ // container parameters... except if we're in WebRender in which case we
+ // don't have a aContainerParameters. In that case we're also not rasterizing
+ // in Gecko anyway, so the only resolution we care about here is the presShell
+ // resolution which we need to propagate to WebRender.
+ metrics.SetCumulativeResolution(
+ LayoutDeviceToLayerScale(presShell->GetCumulativeResolution()));
+
+ metrics.SetTransformToAncestorScale(
+ GetTransformToAncestorScaleCrossProcessForFrameMetrics(
+ aScrollFrame ? aScrollFrame : aForFrame));
+ metrics.SetDevPixelsPerCSSPixel(presContext->CSSToDevPixelScale());
+
+ // Initially, AsyncPanZoomController should render the content to the screen
+ // at the painted resolution.
+ const LayerToParentLayerScale layerToParentLayerScale(1.0f);
+ metrics.SetZoom(metrics.GetCumulativeResolution() *
+ metrics.GetDevPixelsPerCSSPixel() * layerToParentLayerScale);
+
+ // Calculate the composition bounds as the size of the scroll frame and
+ // its origin relative to the reference frame.
+ // If aScrollFrame is null, we are in a document without a root scroll frame,
+ // so it's a xul document. In this case, use the size of the viewport frame.
+ const nsIFrame* frameForCompositionBoundsCalculation =
+ aScrollFrame ? aScrollFrame : aForFrame;
+ nsRect compositionBounds(
+ frameForCompositionBoundsCalculation->GetOffsetToCrossDoc(aItemFrame) +
+ aOffsetToReferenceFrame,
+ frameForCompositionBoundsCalculation->GetSize());
+ if (scrollableFrame) {
+ // If we have a scrollable frame, restrict the composition bounds to its
+ // scroll port. The scroll port excludes the frame borders and the scroll
+ // bars, which we don't want to be part of the composition bounds.
+ nsRect scrollPort = scrollableFrame->GetScrollPortRect();
+ compositionBounds = nsRect(
+ compositionBounds.TopLeft() + scrollPort.TopLeft(), scrollPort.Size());
+ }
+ ParentLayerRect frameBounds =
+ LayoutDeviceRect::FromAppUnits(compositionBounds, auPerDevPixel) *
+ metrics.GetCumulativeResolution() * layerToParentLayerScale;
+
+ // For the root scroll frame of the root content document (RCD-RSF), the above
+ // calculation will yield the size of the viewport frame as the composition
+ // bounds, which doesn't actually correspond to what is visible when
+ // nsIDOMWindowUtils::setCSSViewport has been called to modify the visible
+ // area of the prescontext that the viewport frame is reflowed into. In that
+ // case if our document has a widget then the widget's bounds will correspond
+ // to what is visible. If we don't have a widget the root view's bounds
+ // correspond to what would be visible because they don't get modified by
+ // setCSSViewport.
+ bool isRootContentDocRootScrollFrame =
+ isRootScrollFrame && presContext->IsRootContentDocumentCrossProcess();
+ if (isRootContentDocRootScrollFrame) {
+ UpdateCompositionBoundsForRCDRSF(frameBounds, presContext);
+ if (RefPtr<MobileViewportManager> MVM =
+ presContext->PresShell()->GetMobileViewportManager()) {
+ metrics.SetCompositionSizeWithoutDynamicToolbar(
+ MVM->GetCompositionSizeWithoutDynamicToolbar());
+ }
+ }
+
+ metrics.SetCompositionBoundsWidthIgnoringScrollbars(frameBounds.width);
+
+ nsMargin sizes = ScrollbarAreaToExcludeFromCompositionBoundsFor(aScrollFrame);
+ // Scrollbars are not subject to resolution scaling, so LD pixels = layer
+ // pixels for them.
+ ParentLayerMargin boundMargins =
+ LayoutDeviceMargin::FromAppUnits(sizes, auPerDevPixel) *
+ LayoutDeviceToParentLayerScale(1.0f);
+ frameBounds.Deflate(boundMargins);
+
+ metrics.SetCompositionBounds(frameBounds);
+
+ metrics.SetBoundingCompositionSize(
+ nsLayoutUtils::CalculateBoundingCompositionSize(
+ aScrollFrame ? aScrollFrame : aForFrame,
+ isRootContentDocRootScrollFrame, metrics));
+
+ if (StaticPrefs::apz_printtree() || StaticPrefs::apz_test_logging_enabled()) {
+ if (const nsIContent* content =
+ frameForCompositionBoundsCalculation->GetContent()) {
+ nsAutoString contentDescription;
+ if (content->IsElement()) {
+ content->AsElement()->Describe(contentDescription);
+ } else {
+ contentDescription.AssignLiteral("(not an element)");
+ }
+ metadata.SetContentDescription(
+ NS_LossyConvertUTF16toASCII(contentDescription));
+ if (IsAPZTestLoggingEnabled()) {
+ LogTestDataForPaint(aLayerManager, scrollId, "contentDescription",
+ metadata.GetContentDescription().get());
+ }
+ }
+ }
+
+ metrics.SetPresShellId(presShell->GetPresShellId());
+
+ // If the scroll frame's content is marked 'scrollgrab', record this
+ // in the FrameMetrics so APZ knows to provide the scroll grabbing
+ // behaviour.
+ if (aScrollFrame &&
+ nsContentUtils::HasScrollgrab(aScrollFrame->GetContent())) {
+ metadata.SetHasScrollgrab(true);
+ }
+
+ if (ShouldDisableApzForElement(aContent)) {
+ metadata.SetForceDisableApz(true);
+ }
+
+ metadata.SetIsPaginatedPresentation(presContext->Type() !=
+ nsPresContext::eContext_Galley);
+
+ return metadata;
+}
+
+/*static*/
+Maybe<ScrollMetadata> nsLayoutUtils::GetRootMetadata(
+ nsDisplayListBuilder* aBuilder, WebRenderLayerManager* aLayerManager,
+ const std::function<bool(ViewID& aScrollId)>& aCallback) {
+ nsIFrame* frame = aBuilder->RootReferenceFrame();
+ nsPresContext* presContext = frame->PresContext();
+ PresShell* presShell = presContext->PresShell();
+ Document* document = presShell->GetDocument();
+
+ // There is one case where we want the root container layer to have metrics.
+ // If the parent process is using XUL windows, there is no root scrollframe,
+ // and without explicitly creating metrics there will be no guaranteed
+ // top-level APZC.
+ bool addMetrics = XRE_IsParentProcess() && !presShell->GetRootScrollFrame();
+
+ // Add metrics if there are none in the layer tree with the id (create an id
+ // if there isn't one already) of the root scroll frame/root content.
+ bool ensureMetricsForRootId = nsLayoutUtils::AsyncPanZoomEnabled(frame) &&
+ aBuilder->IsPaintingToWindow() &&
+ !presContext->GetParentPresContext();
+
+ nsIContent* content = nullptr;
+ nsIFrame* rootScrollFrame = presShell->GetRootScrollFrame();
+ if (rootScrollFrame) {
+ content = rootScrollFrame->GetContent();
+ } else {
+ // If there is no root scroll frame, pick the document element instead.
+ // The only case we don't want to do this is in non-APZ fennec, where
+ // we want the root xul document to get a null scroll id so that the root
+ // content document gets the first non-null scroll id.
+ content = document->GetDocumentElement();
+ }
+
+ if (ensureMetricsForRootId && content) {
+ ViewID scrollId = nsLayoutUtils::FindOrCreateIDFor(content);
+ if (aCallback(scrollId)) {
+ ensureMetricsForRootId = false;
+ }
+ }
+
+ if (addMetrics || ensureMetricsForRootId) {
+ bool isRootContent = presContext->IsRootContentDocumentCrossProcess();
+
+ nsSize scrollPortSize = frame->GetSize();
+ if (isRootContent && rootScrollFrame) {
+ nsIScrollableFrame* scrollableFrame =
+ rootScrollFrame->GetScrollTargetFrame();
+ scrollPortSize = scrollableFrame->GetScrollPortRect().Size();
+ }
+ return Some(nsLayoutUtils::ComputeScrollMetadata(
+ frame, rootScrollFrame, content, frame,
+ aBuilder->ToReferenceFrame(frame), aLayerManager,
+ ScrollableLayerGuid::NULL_SCROLL_ID, scrollPortSize, isRootContent));
+ }
+
+ return Nothing();
+}
+
+/* static */
+void nsLayoutUtils::TransformToAncestorAndCombineRegions(
+ const nsRegion& aRegion, nsIFrame* aFrame, const nsIFrame* aAncestorFrame,
+ nsRegion* aPreciseTargetDest, nsRegion* aImpreciseTargetDest,
+ Maybe<Matrix4x4Flagged>* aMatrixCache, const DisplayItemClip* aClip) {
+ if (aRegion.IsEmpty()) {
+ return;
+ }
+ bool isPrecise;
+ RegionBuilder<nsRegion> transformedRegion;
+ for (nsRegion::RectIterator it = aRegion.RectIter(); !it.Done(); it.Next()) {
+ nsRect transformed = TransformFrameRectToAncestor(
+ aFrame, it.Get(), aAncestorFrame, &isPrecise, aMatrixCache);
+ if (aClip) {
+ transformed = aClip->ApplyNonRoundedIntersection(transformed);
+ if (aClip->GetRoundedRectCount() > 0) {
+ isPrecise = false;
+ }
+ }
+ transformedRegion.OrWith(transformed);
+ }
+ nsRegion* dest = isPrecise ? aPreciseTargetDest : aImpreciseTargetDest;
+ dest->OrWith(transformedRegion.ToRegion());
+ // If the region becomes too complex this has a large performance impact.
+ // We limit its complexity here.
+ if (dest->GetNumRects() > 12) {
+ dest->SimplifyOutward(6);
+ if (isPrecise) {
+ aPreciseTargetDest->OrWith(*aImpreciseTargetDest);
+ *aImpreciseTargetDest = std::move(*aPreciseTargetDest);
+ aImpreciseTargetDest->SimplifyOutward(6);
+ *aPreciseTargetDest = nsRegion();
+ }
+ }
+}
+
+/* static */
+bool nsLayoutUtils::ShouldUseNoFramesSheet(Document* aDocument) {
+ bool allowSubframes = true;
+ nsIDocShell* docShell = aDocument->GetDocShell();
+ if (docShell) {
+ docShell->GetAllowSubframes(&allowSubframes);
+ }
+ return !allowSubframes;
+}
+
+/* static */
+void nsLayoutUtils::GetFrameTextContent(nsIFrame* aFrame, nsAString& aResult) {
+ aResult.Truncate();
+ AppendFrameTextContent(aFrame, aResult);
+}
+
+/* static */
+void nsLayoutUtils::AppendFrameTextContent(nsIFrame* aFrame,
+ nsAString& aResult) {
+ if (aFrame->IsTextFrame()) {
+ auto* const textFrame = static_cast<nsTextFrame*>(aFrame);
+ const auto offset = AssertedCast<uint32_t>(textFrame->GetContentOffset());
+ const auto length = AssertedCast<uint32_t>(textFrame->GetContentLength());
+ textFrame->TextFragment()->AppendTo(aResult, offset, length);
+ } else {
+ for (nsIFrame* child : aFrame->PrincipalChildList()) {
+ AppendFrameTextContent(child, aResult);
+ }
+ }
+}
+
+/* static */
+nsRect nsLayoutUtils::GetSelectionBoundingRect(const Selection* aSel) {
+ nsRect res;
+ // Bounding client rect may be empty after calling GetBoundingClientRect
+ // when range is collapsed. So we get caret's rect when range is
+ // collapsed.
+ if (aSel->IsCollapsed()) {
+ nsIFrame* frame = nsCaret::GetGeometry(aSel, &res);
+ if (frame) {
+ nsIFrame* relativeTo = GetContainingBlockForClientRect(frame);
+ res = TransformFrameRectToAncestor(frame, res, relativeTo);
+ }
+ } else {
+ RectAccumulator accumulator;
+ const uint32_t rangeCount = aSel->RangeCount();
+ for (const uint32_t idx : IntegerRange(rangeCount)) {
+ MOZ_ASSERT(aSel->RangeCount() == rangeCount);
+ nsRange* range = aSel->GetRangeAt(idx);
+ nsRange::CollectClientRectsAndText(
+ &accumulator, nullptr, range, range->GetStartContainer(),
+ range->StartOffset(), range->GetEndContainer(), range->EndOffset(),
+ true, false);
+ }
+ res = accumulator.mResultRect.IsEmpty() ? accumulator.mFirstRect
+ : accumulator.mResultRect;
+ }
+
+ return res;
+}
+
+/* static */
+nsBlockFrame* nsLayoutUtils::GetFloatContainingBlock(nsIFrame* aFrame) {
+ nsIFrame* ancestor = aFrame->GetParent();
+ while (ancestor && !ancestor->IsFloatContainingBlock()) {
+ ancestor = ancestor->GetParent();
+ }
+ MOZ_ASSERT(!ancestor || ancestor->IsBlockFrameOrSubclass(),
+ "Float containing block can only be block frame");
+ return static_cast<nsBlockFrame*>(ancestor);
+}
+
+// The implementations of this calculation are adapted from
+// Element::GetBoundingClientRect().
+/* static */
+CSSRect nsLayoutUtils::GetBoundingContentRect(
+ const nsIContent* aContent, const nsIScrollableFrame* aRootScrollFrame,
+ Maybe<CSSRect>* aOutNearestScrollClip) {
+ if (nsIFrame* frame = aContent->GetPrimaryFrame()) {
+ return GetBoundingFrameRect(frame, aRootScrollFrame, aOutNearestScrollClip);
+ }
+ return CSSRect();
+}
+
+/* static */
+CSSRect nsLayoutUtils::GetBoundingFrameRect(
+ nsIFrame* aFrame, const nsIScrollableFrame* aRootScrollFrame,
+ Maybe<CSSRect>* aOutNearestScrollClip) {
+ CSSRect result;
+ nsIFrame* relativeTo = aRootScrollFrame->GetScrolledFrame();
+ result = CSSRect::FromAppUnits(nsLayoutUtils::GetAllInFlowRectsUnion(
+ aFrame, relativeTo, nsLayoutUtils::RECTS_ACCOUNT_FOR_TRANSFORMS));
+
+ // If the element is contained in a scrollable frame that is not
+ // the root scroll frame, make sure to clip the result so that it is
+ // not larger than the containing scrollable frame's bounds.
+ nsIScrollableFrame* scrollFrame = nsLayoutUtils::GetNearestScrollableFrame(
+ aFrame, SCROLLABLE_INCLUDE_HIDDEN | SCROLLABLE_FIXEDPOS_FINDS_ROOT);
+ if (scrollFrame && scrollFrame != aRootScrollFrame) {
+ nsIFrame* subFrame = do_QueryFrame(scrollFrame);
+ MOZ_ASSERT(subFrame);
+ // Get the bounds of the scroll frame in the same coordinate space
+ // as |result|.
+ nsRect subFrameRect = subFrame->GetRectRelativeToSelf();
+ TransformResult res =
+ nsLayoutUtils::TransformRect(subFrame, relativeTo, subFrameRect);
+ MOZ_ASSERT(res == TRANSFORM_SUCCEEDED || res == NONINVERTIBLE_TRANSFORM);
+ if (res == TRANSFORM_SUCCEEDED) {
+ CSSRect subFrameRectCSS = CSSRect::FromAppUnits(subFrameRect);
+ if (aOutNearestScrollClip) {
+ *aOutNearestScrollClip = Some(subFrameRectCSS);
+ }
+
+ result = subFrameRectCSS.Intersect(result);
+ }
+ }
+ return result;
+}
+
+/* static */
+bool nsLayoutUtils::IsTransformed(nsIFrame* aForFrame, nsIFrame* aTopFrame) {
+ for (nsIFrame* f = aForFrame; f != aTopFrame; f = f->GetParent()) {
+ if (f->IsTransformed()) {
+ return true;
+ }
+ }
+ return false;
+}
+
+/*static*/
+CSSPoint nsLayoutUtils::GetCumulativeApzCallbackTransform(nsIFrame* aFrame) {
+ CSSPoint delta;
+ if (!aFrame) {
+ return delta;
+ }
+ nsIFrame* frame = aFrame;
+ nsCOMPtr<nsIContent> lastContent;
+ bool seenRcdRsf = false;
+
+ // Helper lambda to apply the callback transform for a single frame.
+ auto applyCallbackTransformForFrame = [&](nsIFrame* frame) {
+ if (frame) {
+ nsCOMPtr<nsIContent> content = frame->GetContent();
+ if (content && (content != lastContent)) {
+ void* property = content->GetProperty(nsGkAtoms::apzCallbackTransform);
+ if (property) {
+ delta += *static_cast<CSSPoint*>(property);
+ }
+ }
+ lastContent = content;
+ }
+ };
+
+ while (frame) {
+ // Apply the callback transform for the current frame.
+ applyCallbackTransformForFrame(frame);
+
+ // Keep track of whether we've encountered the RCD-RSF's content element.
+ nsPresContext* pc = frame->PresContext();
+ if (pc->IsRootContentDocumentCrossProcess()) {
+ if (PresShell* shell = pc->GetPresShell()) {
+ if (nsIFrame* rsf = shell->GetRootScrollFrame()) {
+ if (frame->GetContent() == rsf->GetContent()) {
+ seenRcdRsf = true;
+ }
+ }
+ }
+ }
+
+ // If we reach the RCD's viewport frame, but have not encountered
+ // the RCD-RSF, we were inside fixed content in the RCD.
+ // We still want to apply the RCD-RSF's callback transform because
+ // it contains the offset between the visual and layout viewports
+ // which applies to fixed content as well.
+ ViewportFrame* viewportFrame = do_QueryFrame(frame);
+ if (viewportFrame) {
+ if (pc->IsRootContentDocumentCrossProcess() && !seenRcdRsf) {
+ applyCallbackTransformForFrame(pc->PresShell()->GetRootScrollFrame());
+ }
+ }
+
+ // Proceed to the parent frame.
+ frame = GetCrossDocParentFrameInProcess(frame);
+ }
+ return delta;
+}
+
+static nsSize ComputeMaxSizeForPartialPrerender(nsIFrame* aFrame,
+ nsSize aMaxSize) {
+ Matrix4x4Flagged transform = nsLayoutUtils::GetTransformToAncestor(
+ RelativeTo{aFrame},
+ RelativeTo{nsLayoutUtils::GetDisplayRootFrame(aFrame)});
+
+ Matrix transform2D;
+ if (!transform.Is2D(&transform2D)) {
+ return aMaxSize;
+ }
+
+ gfx::Rect result(0, 0, aMaxSize.width, aMaxSize.height);
+ auto scale = transform2D.ScaleFactors();
+ if (scale.xScale != 0 && scale.yScale != 0) {
+ result.width /= scale.xScale;
+ result.height /= scale.yScale;
+ }
+
+ // Don't apply translate.
+ transform2D._31 = 0.0f;
+ transform2D._32 = 0.0f;
+
+ // Don't apply scale.
+ if (scale.xScale != 0 && scale.yScale != 0) {
+ transform2D._11 /= scale.xScale;
+ transform2D._12 /= scale.xScale;
+ transform2D._21 /= scale.yScale;
+ transform2D._22 /= scale.yScale;
+ }
+
+ // Theoretically we should use transform2D.Inverse() here but in this case
+ // |transform2D| is a pure rotation matrix, no scaling, no translate at all,
+ // so that the result bound's width and height would be pretty much same
+ // as the one rotated by the inverse matrix.
+ result = transform2D.TransformBounds(result);
+ return nsSize(
+ result.width < (float)nscoord_MAX ? result.width : nscoord_MAX,
+ result.height < (float)nscoord_MAX ? result.height : nscoord_MAX);
+}
+
+/* static */
+nsRect nsLayoutUtils::ComputePartialPrerenderArea(
+ nsIFrame* aFrame, const nsRect& aDirtyRect, const nsRect& aOverflow,
+ const nsSize& aPrerenderSize) {
+ nsSize maxSizeForPartialPrerender =
+ ComputeMaxSizeForPartialPrerender(aFrame, aPrerenderSize);
+ // Simple calculation for now: center the pre-render area on the dirty rect,
+ // and clamp to the overflow area. Later we can do more advanced things like
+ // redistributing from one axis to another, or from one side to another.
+ nscoord xExcess =
+ std::max(maxSizeForPartialPrerender.width - aDirtyRect.width, 0);
+ nscoord yExcess =
+ std::max(maxSizeForPartialPrerender.height - aDirtyRect.height, 0);
+ nsRect result = aDirtyRect;
+ result.Inflate(xExcess / 2, yExcess / 2);
+ return result.MoveInsideAndClamp(aOverflow);
+}
+
+static bool LineHasNonEmptyContentWorker(nsIFrame* aFrame) {
+ // Look for non-empty frames, but ignore inline and br frames.
+ // For inline frames, descend into the children, if any.
+ if (aFrame->IsInlineFrame()) {
+ for (nsIFrame* child : aFrame->PrincipalChildList()) {
+ if (LineHasNonEmptyContentWorker(child)) {
+ return true;
+ }
+ }
+ } else {
+ if (!aFrame->IsBrFrame() && !aFrame->IsEmpty()) {
+ return true;
+ }
+ }
+ return false;
+}
+
+static bool LineHasNonEmptyContent(nsLineBox* aLine) {
+ int32_t count = aLine->GetChildCount();
+ for (nsIFrame* frame = aLine->mFirstChild; count > 0;
+ --count, frame = frame->GetNextSibling()) {
+ if (LineHasNonEmptyContentWorker(frame)) {
+ return true;
+ }
+ }
+ return false;
+}
+
+/* static */
+bool nsLayoutUtils::IsInvisibleBreak(nsINode* aNode,
+ nsIFrame** aNextLineFrame) {
+ if (aNextLineFrame) {
+ *aNextLineFrame = nullptr;
+ }
+
+ if (!aNode->IsElement() || !aNode->IsEditable()) {
+ return false;
+ }
+ nsIFrame* frame = aNode->AsElement()->GetPrimaryFrame();
+ if (!frame || !frame->IsBrFrame()) {
+ return false;
+ }
+
+ nsContainerFrame* f = frame->GetParent();
+ while (f && f->IsFrameOfType(nsIFrame::eLineParticipant)) {
+ f = f->GetParent();
+ }
+ nsBlockFrame* blockAncestor = do_QueryFrame(f);
+ if (!blockAncestor) {
+ // The container frame doesn't support line breaking.
+ return false;
+ }
+
+ bool valid = false;
+ nsBlockInFlowLineIterator iter(blockAncestor, frame, &valid);
+ if (!valid) {
+ return false;
+ }
+
+ bool lineNonEmpty = LineHasNonEmptyContent(iter.GetLine());
+ if (!lineNonEmpty) {
+ return false;
+ }
+
+ while (iter.Next()) {
+ auto currentLine = iter.GetLine();
+ // Completely skip empty lines.
+ if (!currentLine->IsEmpty()) {
+ // If we come across an inline line, the BR has caused a visible line
+ // break.
+ if (currentLine->IsInline()) {
+ if (aNextLineFrame) {
+ *aNextLineFrame = currentLine->mFirstChild;
+ }
+ return false;
+ }
+ break;
+ }
+ }
+
+ return lineNonEmpty;
+}
+
+static nsRect ComputeSVGReferenceRect(nsIFrame* aFrame,
+ StyleGeometryBox aGeometryBox) {
+ MOZ_ASSERT(aFrame->GetContent()->IsSVGElement());
+ nsRect r;
+
+ // For SVG elements without associated CSS layout box, the used value for
+ // content-box, padding-box, border-box and margin-box is fill-box.
+ switch (aGeometryBox) {
+ case StyleGeometryBox::StrokeBox: {
+ // XXX Bug 1299876
+ // The size of stroke-box is not correct if this graphic element has
+ // specific stroke-linejoin or stroke-linecap.
+ gfxRect bbox =
+ SVGUtils::GetBBox(aFrame, SVGUtils::eBBoxIncludeFillGeometry |
+ SVGUtils::eBBoxIncludeStroke);
+ r = nsLayoutUtils::RoundGfxRectToAppRect(bbox, AppUnitsPerCSSPixel());
+ break;
+ }
+ case StyleGeometryBox::ViewBox: {
+ SVGViewportElement* viewportElement =
+ SVGElement::FromNode(aFrame->GetContent())->GetCtx();
+ if (!viewportElement) {
+ // We should not render without a viewport so return an empty rect.
+ break;
+ }
+
+ if (viewportElement->HasViewBox()) {
+ // If a `viewBox` attribute is specified for the SVG viewport creating
+ // element:
+ // 1. The reference box is positioned at the origin of the coordinate
+ // system established by the `viewBox` attribute.
+ // 2. The dimension of the reference box is set to the width and height
+ // values of the `viewBox` attribute.
+ const SVGViewBox& value =
+ viewportElement->GetAnimatedViewBox()->GetAnimValue();
+ r = nsRect(nsPresContext::CSSPixelsToAppUnits(value.x),
+ nsPresContext::CSSPixelsToAppUnits(value.y),
+ nsPresContext::CSSPixelsToAppUnits(value.width),
+ nsPresContext::CSSPixelsToAppUnits(value.height));
+ } else {
+ // No viewBox is specified, uses the nearest SVG viewport as reference
+ // box.
+ svgFloatSize viewportSize = viewportElement->GetViewportSize();
+ r = nsRect(0, 0, nsPresContext::CSSPixelsToAppUnits(viewportSize.width),
+ nsPresContext::CSSPixelsToAppUnits(viewportSize.height));
+ }
+
+ break;
+ }
+ case StyleGeometryBox::NoBox:
+ case StyleGeometryBox::BorderBox:
+ case StyleGeometryBox::ContentBox:
+ case StyleGeometryBox::PaddingBox:
+ case StyleGeometryBox::MarginBox:
+ case StyleGeometryBox::FillBox: {
+ gfxRect bbox =
+ SVGUtils::GetBBox(aFrame, SVGUtils::eBBoxIncludeFillGeometry);
+ r = nsLayoutUtils::RoundGfxRectToAppRect(bbox, AppUnitsPerCSSPixel());
+ break;
+ }
+ default: {
+ MOZ_ASSERT_UNREACHABLE("unknown StyleGeometryBox type");
+ gfxRect bbox =
+ SVGUtils::GetBBox(aFrame, SVGUtils::eBBoxIncludeFillGeometry);
+ r = nsLayoutUtils::RoundGfxRectToAppRect(bbox, AppUnitsPerCSSPixel());
+ break;
+ }
+ }
+
+ return r;
+}
+
+static nsRect ComputeHTMLReferenceRect(nsIFrame* aFrame,
+ StyleGeometryBox aGeometryBox) {
+ nsRect r;
+
+ // For elements with associated CSS layout box, the used value for fill-box,
+ // stroke-box and view-box is border-box.
+ switch (aGeometryBox) {
+ case StyleGeometryBox::ContentBox:
+ r = aFrame->GetContentRectRelativeToSelf();
+ break;
+ case StyleGeometryBox::PaddingBox:
+ r = aFrame->GetPaddingRectRelativeToSelf();
+ break;
+ case StyleGeometryBox::MarginBox:
+ r = aFrame->GetMarginRectRelativeToSelf();
+ break;
+ case StyleGeometryBox::NoBox:
+ case StyleGeometryBox::BorderBox:
+ case StyleGeometryBox::FillBox:
+ case StyleGeometryBox::StrokeBox:
+ case StyleGeometryBox::ViewBox:
+ r = aFrame->GetRectRelativeToSelf();
+ break;
+ default:
+ MOZ_ASSERT_UNREACHABLE("unknown StyleGeometryBox type");
+ r = aFrame->GetRectRelativeToSelf();
+ break;
+ }
+
+ return r;
+}
+
+static StyleGeometryBox ShapeBoxToGeometryBox(const StyleShapeBox& aBox) {
+ switch (aBox) {
+ case StyleShapeBox::BorderBox:
+ return StyleGeometryBox::BorderBox;
+ case StyleShapeBox::ContentBox:
+ return StyleGeometryBox::ContentBox;
+ case StyleShapeBox::MarginBox:
+ return StyleGeometryBox::MarginBox;
+ case StyleShapeBox::PaddingBox:
+ return StyleGeometryBox::PaddingBox;
+ }
+ MOZ_ASSERT_UNREACHABLE("Unknown shape box type");
+ return StyleGeometryBox::MarginBox;
+}
+
+static StyleGeometryBox ClipPathBoxToGeometryBox(
+ const StyleShapeGeometryBox& aBox) {
+ using Tag = StyleShapeGeometryBox::Tag;
+ switch (aBox.tag) {
+ case Tag::ShapeBox:
+ return ShapeBoxToGeometryBox(aBox.AsShapeBox());
+ case Tag::ElementDependent:
+ return StyleGeometryBox::NoBox;
+ case Tag::FillBox:
+ return StyleGeometryBox::FillBox;
+ case Tag::StrokeBox:
+ return StyleGeometryBox::StrokeBox;
+ case Tag::ViewBox:
+ return StyleGeometryBox::ViewBox;
+ }
+ MOZ_ASSERT_UNREACHABLE("Unknown shape box type");
+ return StyleGeometryBox::NoBox;
+}
+
+/* static */
+nsRect nsLayoutUtils::ComputeGeometryBox(nsIFrame* aFrame,
+ StyleGeometryBox aGeometryBox) {
+ // We use ComputeSVGReferenceRect for all SVG elements, except <svg>
+ // element, which does have an associated CSS layout box. In this case we
+ // should still use ComputeHTMLReferenceRect for region computing.
+ return aFrame->HasAnyStateBits(NS_FRAME_SVG_LAYOUT)
+ ? ComputeSVGReferenceRect(aFrame, aGeometryBox)
+ : ComputeHTMLReferenceRect(aFrame, aGeometryBox);
+}
+
+nsRect nsLayoutUtils::ComputeGeometryBox(nsIFrame* aFrame,
+ const StyleShapeBox& aBox) {
+ return ComputeGeometryBox(aFrame, ShapeBoxToGeometryBox(aBox));
+}
+
+nsRect nsLayoutUtils::ComputeGeometryBox(nsIFrame* aFrame,
+ const StyleShapeGeometryBox& aBox) {
+ return ComputeGeometryBox(aFrame, ClipPathBoxToGeometryBox(aBox));
+}
+
+/* static */
+nsPoint nsLayoutUtils::ComputeOffsetToUserSpace(nsDisplayListBuilder* aBuilder,
+ nsIFrame* aFrame) {
+ nsPoint offsetToBoundingBox =
+ aBuilder->ToReferenceFrame(aFrame) -
+ SVGIntegrationUtils::GetOffsetToBoundingBox(aFrame);
+ if (!aFrame->IsFrameOfType(nsIFrame::eSVG)) {
+ // Snap the offset if the reference frame is not a SVG frame, since other
+ // frames will be snapped to pixel when rendering.
+ offsetToBoundingBox =
+ nsPoint(aFrame->PresContext()->RoundAppUnitsToNearestDevPixels(
+ offsetToBoundingBox.x),
+ aFrame->PresContext()->RoundAppUnitsToNearestDevPixels(
+ offsetToBoundingBox.y));
+ }
+
+ // During SVG painting, the offset computed here is applied to the gfxContext
+ // "ctx" used to paint the mask. After applying only "offsetToBoundingBox",
+ // "ctx" would have its origin at the top left corner of frame's bounding box
+ // (over all continuations).
+ // However, SVG painting needs the origin to be located at the origin of the
+ // SVG frame's "user space", i.e. the space in which, for example, the
+ // frame's BBox lives.
+ // SVG geometry frames and foreignObject frames apply their own offsets, so
+ // their position is relative to their user space. So for these frame types,
+ // if we want "ctx" to be in user space, we first need to subtract the
+ // frame's position so that SVG painting can later add it again and the
+ // frame is painted in the right place.
+ gfxPoint toUserSpaceGfx =
+ SVGUtils::FrameSpaceInCSSPxToUserSpaceOffset(aFrame);
+ nsPoint toUserSpace =
+ nsPoint(nsPresContext::CSSPixelsToAppUnits(float(toUserSpaceGfx.x)),
+ nsPresContext::CSSPixelsToAppUnits(float(toUserSpaceGfx.y)));
+
+ return (offsetToBoundingBox - toUserSpace);
+}
+
+/* static */
+already_AddRefed<nsFontMetrics> nsLayoutUtils::GetMetricsFor(
+ nsPresContext* aPresContext, bool aIsVertical,
+ const nsStyleFont* aStyleFont, Length aFontSize, bool aUseUserFontSet) {
+ nsFont font = aStyleFont->mFont;
+ font.size = aFontSize;
+ gfxFont::Orientation orientation =
+ aIsVertical ? nsFontMetrics::eVertical : nsFontMetrics::eHorizontal;
+ nsFontMetrics::Params params;
+ params.language = aStyleFont->mLanguage;
+ params.explicitLanguage = aStyleFont->mExplicitLanguage;
+ params.orientation = orientation;
+ params.userFontSet =
+ aUseUserFontSet ? aPresContext->GetUserFontSet() : nullptr;
+ params.textPerf = aPresContext->GetTextPerfMetrics();
+ params.featureValueLookup = aPresContext->GetFontFeatureValuesLookup();
+ return aPresContext->GetMetricsFor(font, params);
+}
+
+/* static */
+void nsLayoutUtils::ComputeSystemFont(nsFont* aSystemFont,
+ LookAndFeel::FontID aFontID,
+ const nsFont& aDefaultVariableFont,
+ const Document* aDocument) {
+ gfxFontStyle fontStyle;
+ nsAutoString systemFontName;
+ if (!LookAndFeel::GetFont(aFontID, systemFontName, fontStyle)) {
+ return;
+ }
+ systemFontName.Trim("\"'");
+ NS_ConvertUTF16toUTF8 nameu8(systemFontName);
+ Servo_FontFamily_ForSystemFont(&nameu8, &aSystemFont->family);
+ aSystemFont->style = fontStyle.style;
+ aSystemFont->family.is_system_font = fontStyle.systemFont;
+ aSystemFont->weight = fontStyle.weight;
+ aSystemFont->stretch = fontStyle.stretch;
+ aSystemFont->size = Length::FromPixels(fontStyle.size);
+
+ // aSystemFont->langGroup = fontStyle.langGroup;
+ switch (StyleFontSizeAdjust::Tag(fontStyle.sizeAdjustBasis)) {
+ case StyleFontSizeAdjust::Tag::None:
+ aSystemFont->sizeAdjust = StyleFontSizeAdjust::None();
+ break;
+ case StyleFontSizeAdjust::Tag::ExHeight:
+ aSystemFont->sizeAdjust =
+ StyleFontSizeAdjust::ExHeight(fontStyle.sizeAdjust);
+ break;
+ case StyleFontSizeAdjust::Tag::CapHeight:
+ aSystemFont->sizeAdjust =
+ StyleFontSizeAdjust::CapHeight(fontStyle.sizeAdjust);
+ break;
+ case StyleFontSizeAdjust::Tag::ChWidth:
+ aSystemFont->sizeAdjust =
+ StyleFontSizeAdjust::ChWidth(fontStyle.sizeAdjust);
+ break;
+ case StyleFontSizeAdjust::Tag::IcWidth:
+ aSystemFont->sizeAdjust =
+ StyleFontSizeAdjust::IcWidth(fontStyle.sizeAdjust);
+ break;
+ case StyleFontSizeAdjust::Tag::IcHeight:
+ aSystemFont->sizeAdjust =
+ StyleFontSizeAdjust::IcHeight(fontStyle.sizeAdjust);
+ break;
+ }
+
+ if (aFontID == LookAndFeel::FontID::MozField ||
+ aFontID == LookAndFeel::FontID::MozButton ||
+ aFontID == LookAndFeel::FontID::MozList) {
+ // For textfields, buttons and selects, we use whatever font is defined by
+ // the system. Which it appears (and the assumption is) it is always a
+ // proportional font. Then we always use 2 points smaller than what the
+ // browser has defined as the default proportional font.
+ //
+ // This matches historical Windows behavior and other browsers.
+ auto newSize =
+ aDefaultVariableFont.size.ToCSSPixels() - CSSPixel::FromPoints(2.0f);
+ aSystemFont->size = Length::FromPixels(std::max(float(newSize), 0.0f));
+ }
+}
+
+/* static */
+bool nsLayoutUtils::ShouldHandleMetaViewport(const Document* aDocument) {
+ auto metaViewportOverride = nsIDocShell::META_VIEWPORT_OVERRIDE_NONE;
+ if (aDocument) {
+ if (nsIDocShell* docShell = aDocument->GetDocShell()) {
+ metaViewportOverride = docShell->GetMetaViewportOverride();
+ }
+ }
+ switch (metaViewportOverride) {
+ case nsIDocShell::META_VIEWPORT_OVERRIDE_ENABLED:
+ return true;
+ case nsIDocShell::META_VIEWPORT_OVERRIDE_DISABLED:
+ return false;
+ default:
+ MOZ_ASSERT(metaViewportOverride ==
+ nsIDocShell::META_VIEWPORT_OVERRIDE_NONE);
+ // The META_VIEWPORT_OVERRIDE_NONE case means that there is no override
+ // and we rely solely on the StaticPrefs.
+ return StaticPrefs::dom_meta_viewport_enabled();
+ }
+}
+
+/* static */
+ComputedStyle* nsLayoutUtils::StyleForScrollbar(
+ const nsIFrame* aScrollbarPart) {
+ // Get the closest content node which is not an anonymous scrollbar
+ // part. It should be the originating element of the scrollbar part.
+ nsIContent* content = aScrollbarPart->GetContent();
+ // Note that the content may be a normal element with scrollbar part
+ // value specified for its -moz-appearance, so don't rely on it being
+ // a native anonymous. Also note that we have to check the node name
+ // because anonymous element like generated content may originate a
+ // scrollbar.
+ MOZ_ASSERT(content, "No content for the scrollbar part?");
+ while (content && content->IsInNativeAnonymousSubtree() &&
+ content->IsAnyOfXULElements(
+ nsGkAtoms::scrollbar, nsGkAtoms::scrollbarbutton,
+ nsGkAtoms::scrollcorner, nsGkAtoms::slider, nsGkAtoms::thumb)) {
+ content = content->GetParent();
+ }
+ MOZ_ASSERT(content, "Native anonymous element with no originating node?");
+ // Use the style from the primary frame of the content.
+ // Note: it is important to use the primary frame rather than an
+ // ancestor frame of the scrollbar part for the correct handling of
+ // viewport scrollbar. The content of the scroll frame of the viewport
+ // is the root element, but its style inherits from the viewport.
+ // Since we need to use the style of root element for the viewport
+ // scrollbar, we have to get the style from the primary frame.
+ if (nsIFrame* primaryFrame = content->GetPrimaryFrame()) {
+ return primaryFrame->Style();
+ }
+ // If the element doesn't have primary frame, get the computed style
+ // from the element directly. This can happen on viewport, because
+ // the scrollbar of viewport may be shown when the root element has
+ // > display: none; overflow: scroll;
+ MOZ_ASSERT(
+ content == aScrollbarPart->PresContext()->Document()->GetRootElement(),
+ "Root element is the only case for this fallback "
+ "path to be triggered");
+ RefPtr<ComputedStyle> style =
+ ServoStyleSet::ResolveServoStyle(*content->AsElement());
+ // Dropping the strong reference is fine because the style should be
+ // held strongly by the element.
+ return style.get();
+}
+
+enum class FramePosition : uint8_t {
+ Unknown,
+ InView,
+ OutOfView,
+};
+
+// NOTE: Returns a pair of Nothing() and `FramePosition::Unknown` if |aFrame|
+// is not in out-of-process or if we haven't received enough information from
+// APZ.
+static std::pair<Maybe<ScreenRect>, FramePosition> GetFrameVisibleRectOnScreen(
+ const nsIFrame* aFrame) {
+ // We actually want the in-process top prescontext here.
+ nsPresContext* topContextInProcess =
+ aFrame->PresContext()->GetInProcessRootContentDocumentPresContext();
+ if (!topContextInProcess) {
+ // We are in chrome process.
+ return std::make_pair(Nothing(), FramePosition::Unknown);
+ }
+
+ if (topContextInProcess->Document()->IsTopLevelContentDocument()) {
+ // We are in the top of content document.
+ return std::make_pair(Nothing(), FramePosition::Unknown);
+ }
+
+ nsIDocShell* docShell = topContextInProcess->GetDocShell();
+ BrowserChild* browserChild = BrowserChild::GetFrom(docShell);
+ if (!browserChild) {
+ // We are not in out-of-process iframe.
+ return std::make_pair(Nothing(), FramePosition::Unknown);
+ }
+
+ if (!browserChild->GetEffectsInfo().IsVisible()) {
+ // There is no visible rect on this iframe at all.
+ return std::make_pair(Some(ScreenRect()), FramePosition::Unknown);
+ }
+
+ Maybe<ScreenRect> visibleRect =
+ browserChild->GetTopLevelViewportVisibleRectInBrowserCoords();
+ if (!visibleRect) {
+ // We are unsure if we haven't received the transformed rectangle of the
+ // iframe's visible area.
+ return std::make_pair(Nothing(), FramePosition::Unknown);
+ }
+
+ nsIFrame* rootFrame = topContextInProcess->PresShell()->GetRootFrame();
+ nsRect transformedToIFrame = nsLayoutUtils::TransformFrameRectToAncestor(
+ aFrame, aFrame->InkOverflowRectRelativeToSelf(), rootFrame);
+
+ LayoutDeviceRect rectInLayoutDevicePixel = LayoutDeviceRect::FromAppUnits(
+ transformedToIFrame, topContextInProcess->AppUnitsPerDevPixel());
+
+ ScreenRect transformedToRoot = ViewAs<ScreenPixel>(
+ browserChild->GetChildToParentConversionMatrix().TransformBounds(
+ rectInLayoutDevicePixel),
+ PixelCastJustification::ContentProcessIsLayerInUiProcess);
+
+ FramePosition position = FramePosition::Unknown;
+ // we need to check whether the transformed rect is outside the iframe
+ // visible rect or not because in some cases the rect size is (0x0), thus
+ // the intersection between the transformed rect and the iframe visible rect
+ // would also be (0x0), then we can't tell whether the given nsIFrame is
+ // inside the iframe visible rect or not by calling BaseRect::IsEmpty for the
+ // intersection.
+ if (transformedToRoot.x > visibleRect->XMost() ||
+ transformedToRoot.y > visibleRect->YMost() ||
+ visibleRect->x > transformedToRoot.XMost() ||
+ visibleRect->y > transformedToRoot.YMost()) {
+ position = FramePosition::OutOfView;
+ } else {
+ position = FramePosition::InView;
+ }
+
+ return std::make_pair(Some(visibleRect->Intersect(transformedToRoot)),
+ position);
+}
+
+// static
+bool nsLayoutUtils::FrameIsScrolledOutOfViewInCrossProcess(
+ const nsIFrame* aFrame) {
+ auto [visibleRect, framePosition] = GetFrameVisibleRectOnScreen(aFrame);
+ if (visibleRect.isNothing()) {
+ return false;
+ }
+
+ return visibleRect->IsEmpty() && framePosition != FramePosition::InView;
+}
+
+// static
+bool nsLayoutUtils::FrameIsMostlyScrolledOutOfViewInCrossProcess(
+ const nsIFrame* aFrame, nscoord aMargin) {
+ auto [visibleRect, framePosition] = GetFrameVisibleRectOnScreen(aFrame);
+ (void)framePosition;
+ if (visibleRect.isNothing()) {
+ return false;
+ }
+
+ nsPresContext* topContextInProcess =
+ aFrame->PresContext()->GetInProcessRootContentDocumentPresContext();
+ MOZ_ASSERT(topContextInProcess);
+
+ nsIDocShell* docShell = topContextInProcess->GetDocShell();
+ BrowserChild* browserChild = BrowserChild::GetFrom(docShell);
+ MOZ_ASSERT(browserChild);
+
+ auto scale =
+ browserChild->GetChildToParentConversionMatrix().As2D().ScaleFactors();
+ ScreenSize margin(scale.xScale * CSSPixel::FromAppUnits(aMargin),
+ scale.yScale * CSSPixel::FromAppUnits(aMargin));
+
+ return visibleRect->width < margin.width ||
+ visibleRect->height < margin.height;
+}
+
+// static
+nsSize nsLayoutUtils::ExpandHeightForViewportUnits(nsPresContext* aPresContext,
+ const nsSize& aSize) {
+ nsSize sizeForViewportUnits = aPresContext->GetSizeForViewportUnits();
+
+ // |aSize| might be the size expanded to the minimum-scale size whereas the
+ // size for viewport units is not scaled so that we need to expand the |aSize|
+ // height by multiplying by the ratio of the viewport units height to the
+ // visible area height.
+ float vhExpansionRatio = (float)sizeForViewportUnits.height /
+ aPresContext->GetVisibleArea().height;
+
+ MOZ_ASSERT(aSize.height <= NSCoordSaturatingNonnegativeMultiply(
+ aSize.height, vhExpansionRatio));
+ return nsSize(aSize.width, NSCoordSaturatingNonnegativeMultiply(
+ aSize.height, vhExpansionRatio));
+}
+
+template <typename SizeType>
+/* static */ SizeType ExpandHeightForDynamicToolbarImpl(
+ const nsPresContext* aPresContext, const SizeType& aSize) {
+ MOZ_ASSERT(aPresContext);
+
+ LayoutDeviceIntSize displaySize;
+ if (RefPtr<MobileViewportManager> MVM =
+ aPresContext->PresShell()->GetMobileViewportManager()) {
+ displaySize = MVM->DisplaySize();
+ } else if (!nsLayoutUtils::GetContentViewerSize(aPresContext, displaySize)) {
+ return aSize;
+ }
+
+ float toolbarHeightRatio =
+ mozilla::ScreenCoord(aPresContext->GetDynamicToolbarMaxHeight()) /
+ mozilla::ViewAs<mozilla::ScreenPixel>(
+ displaySize,
+ mozilla::PixelCastJustification::LayoutDeviceIsScreenForBounds)
+ .height;
+
+ SizeType expandedSize = aSize;
+ static_assert(std::is_same_v<nsSize, SizeType> ||
+ std::is_same_v<CSSSize, SizeType>);
+ if constexpr (std::is_same_v<nsSize, SizeType>) {
+ expandedSize.height =
+ NSCoordSaturatingAdd(aSize.height, aSize.height * toolbarHeightRatio);
+ } else if (std::is_same_v<CSSSize, SizeType>) {
+ expandedSize.height = aSize.height + aSize.height * toolbarHeightRatio;
+ }
+ return expandedSize;
+}
+
+CSSSize nsLayoutUtils::ExpandHeightForDynamicToolbar(
+ const nsPresContext* aPresContext, const CSSSize& aSize) {
+ return ExpandHeightForDynamicToolbarImpl(aPresContext, aSize);
+}
+nsSize nsLayoutUtils::ExpandHeightForDynamicToolbar(
+ const nsPresContext* aPresContext, const nsSize& aSize) {
+ return ExpandHeightForDynamicToolbarImpl(aPresContext, aSize);
+}