| /* |
| * Copyright (C) 1999 Lars Knoll (knoll@kde.org) |
| * (C) 1999 Antti Koivisto (koivisto@kde.org) |
| * (C) 2005 Allan Sandfeld Jensen (kde@carewolf.com) |
| * (C) 2005, 2006 Samuel Weinig (sam.weinig@gmail.com) |
| * Copyright (C) 2005, 2006, 2007, 2008, 2009, 2013 Apple Inc. All rights reserved. |
| * Copyright (C) 2010 Google Inc. All rights reserved. |
| * |
| * This library is free software; you can redistribute it and/or |
| * modify it under the terms of the GNU Library General Public |
| * License as published by the Free Software Foundation; either |
| * version 2 of the License, or (at your option) any later version. |
| * |
| * This library is distributed in the hope that it will be useful, |
| * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| * Library General Public License for more details. |
| * |
| * You should have received a copy of the GNU Library General Public License |
| * along with this library; see the file COPYING.LIB. If not, write to |
| * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
| * Boston, MA 02110-1301, USA. |
| * |
| */ |
| |
| #include "config.h" |
| #include "RenderBoxModelObject.h" |
| |
| #include "BitmapImage.h" |
| #include "BorderEdge.h" |
| #include "CachedImage.h" |
| #include "ColorBlending.h" |
| #include "Document.h" |
| #include "FloatRoundedRect.h" |
| #include "Frame.h" |
| #include "FrameView.h" |
| #include "GeometryUtilities.h" |
| #include "GraphicsContext.h" |
| #include "HTMLImageElement.h" |
| #include "HTMLNames.h" |
| #include "ImageBuffer.h" |
| #include "ImageQualityController.h" |
| #include "InlineIteratorInlineBox.h" |
| #include "Path.h" |
| #include "RenderBlock.h" |
| #include "RenderFlexibleBox.h" |
| #include "RenderFragmentContainer.h" |
| #include "RenderInline.h" |
| #include "RenderLayer.h" |
| #include "RenderLayerBacking.h" |
| #include "RenderLayerCompositor.h" |
| #include "RenderLayerScrollableArea.h" |
| #include "RenderMultiColumnFlow.h" |
| #include "RenderTable.h" |
| #include "RenderTableRow.h" |
| #include "RenderText.h" |
| #include "RenderTextFragment.h" |
| #include "RenderTreeBuilder.h" |
| #include "RenderView.h" |
| #include "ScrollingConstraints.h" |
| #include "Settings.h" |
| #include "Styleable.h" |
| #include "TextBoxPainter.h" |
| #include "TransformState.h" |
| #include <wtf/IsoMallocInlines.h> |
| #include <wtf/NeverDestroyed.h> |
| #if ASSERT_ENABLED |
| #include <wtf/SetForScope.h> |
| #endif |
| |
| #if PLATFORM(IOS_FAMILY) |
| #include "RuntimeApplicationChecks.h" |
| #endif |
| |
| namespace WebCore { |
| |
| using namespace HTMLNames; |
| |
| WTF_MAKE_ISO_ALLOCATED_IMPL(RenderBoxModelObject); |
| |
| // The HashMap for storing continuation pointers. |
| // An inline can be split with blocks occuring in between the inline content. |
| // When this occurs we need a pointer to the next object. We can basically be |
| // split into a sequence of inlines and blocks. The continuation will either be |
| // an anonymous block (that houses other blocks) or it will be an inline flow. |
| // <b><i><p>Hello</p></i></b>. In this example the <i> will have a block as |
| // its continuation but the <b> will just have an inline as its continuation. |
| RenderBoxModelObject::ContinuationChainNode::ContinuationChainNode(RenderBoxModelObject& renderer) |
| : renderer(renderer) |
| { |
| } |
| |
| RenderBoxModelObject::ContinuationChainNode::~ContinuationChainNode() |
| { |
| if (next) { |
| ASSERT(previous); |
| ASSERT(next->previous == this); |
| next->previous = previous; |
| } |
| if (previous) { |
| ASSERT(previous->next == this); |
| previous->next = next; |
| } |
| } |
| |
| void RenderBoxModelObject::ContinuationChainNode::insertAfter(ContinuationChainNode& after) |
| { |
| ASSERT(!previous); |
| ASSERT(!next); |
| if ((next = after.next)) { |
| ASSERT(next->previous == &after); |
| next->previous = this; |
| } |
| previous = &after; |
| after.next = this; |
| } |
| |
| using ContinuationChainNodeMap = HashMap<const RenderBoxModelObject*, std::unique_ptr<RenderBoxModelObject::ContinuationChainNode>>; |
| |
| static ContinuationChainNodeMap& continuationChainNodeMap() |
| { |
| static NeverDestroyed<ContinuationChainNodeMap> map; |
| return map; |
| } |
| |
| using FirstLetterRemainingTextMap = HashMap<const RenderBoxModelObject*, WeakPtr<RenderTextFragment>>; |
| |
| static FirstLetterRemainingTextMap& firstLetterRemainingTextMap() |
| { |
| static NeverDestroyed<FirstLetterRemainingTextMap> map; |
| return map; |
| } |
| |
| void RenderBoxModelObject::setSelectionState(HighlightState state) |
| { |
| if (state == HighlightState::Inside && selectionState() != HighlightState::None) |
| return; |
| |
| if ((state == HighlightState::Start && selectionState() == HighlightState::End) |
| || (state == HighlightState::End && selectionState() == HighlightState::Start)) |
| RenderLayerModelObject::setSelectionState(HighlightState::Both); |
| else |
| RenderLayerModelObject::setSelectionState(state); |
| |
| // FIXME: We should consider whether it is OK propagating to ancestor RenderInlines. |
| // This is a workaround for http://webkit.org/b/32123 |
| // The containing block can be null in case of an orphaned tree. |
| RenderBlock* containingBlock = this->containingBlock(); |
| if (containingBlock && !containingBlock->isRenderView()) |
| containingBlock->setSelectionState(state); |
| } |
| |
| void RenderBoxModelObject::contentChanged(ContentChangeType changeType) |
| { |
| if (!hasLayer()) |
| return; |
| |
| layer()->contentChanged(changeType); |
| } |
| |
| bool RenderBoxModelObject::hasAcceleratedCompositing() const |
| { |
| return view().compositor().hasAcceleratedCompositing(); |
| } |
| |
| RenderBoxModelObject::RenderBoxModelObject(Element& element, RenderStyle&& style, BaseTypeFlags baseTypeFlags) |
| : RenderLayerModelObject(element, WTFMove(style), baseTypeFlags | RenderBoxModelObjectFlag) |
| { |
| } |
| |
| RenderBoxModelObject::RenderBoxModelObject(Document& document, RenderStyle&& style, BaseTypeFlags baseTypeFlags) |
| : RenderLayerModelObject(document, WTFMove(style), baseTypeFlags | RenderBoxModelObjectFlag) |
| { |
| } |
| |
| RenderBoxModelObject::~RenderBoxModelObject() |
| { |
| // Do not add any code here. Add it to willBeDestroyed() instead. |
| ASSERT(!continuation()); |
| } |
| |
| void RenderBoxModelObject::willBeDestroyed() |
| { |
| if (hasContinuationChainNode()) |
| removeFromContinuationChain(); |
| |
| if (isFirstLetter()) |
| clearFirstLetterRemainingText(); |
| |
| if (!renderTreeBeingDestroyed()) |
| view().imageQualityController().rendererWillBeDestroyed(*this); |
| |
| RenderLayerModelObject::willBeDestroyed(); |
| } |
| |
| bool RenderBoxModelObject::hasVisibleBoxDecorationStyle() const |
| { |
| return hasBackground() || style().hasVisibleBorderDecoration() || style().hasEffectiveAppearance() || style().boxShadow(); |
| } |
| |
| void RenderBoxModelObject::updateFromStyle() |
| { |
| RenderLayerModelObject::updateFromStyle(); |
| |
| // Set the appropriate bits for a box model object. Since all bits are cleared in styleWillChange, |
| // we only check for bits that could possibly be set to true. |
| const RenderStyle& styleToUse = style(); |
| setHasVisibleBoxDecorations(hasVisibleBoxDecorationStyle()); |
| setInline(styleToUse.isDisplayInlineType()); |
| setPositionState(styleToUse.position()); |
| setHorizontalWritingMode(styleToUse.isHorizontalWritingMode()); |
| if (styleToUse.isFlippedBlocksWritingMode()) |
| view().frameView().setHasFlippedBlockRenderers(true); |
| setPaintContainmentApplies(shouldApplyPaintContainment()); |
| } |
| |
| static LayoutSize accumulateInFlowPositionOffsets(const RenderObject* child) |
| { |
| if (!child->isAnonymousBlock() || !child->isInFlowPositioned()) |
| return LayoutSize(); |
| LayoutSize offset; |
| for (RenderElement* parent = downcast<RenderBlock>(*child).inlineContinuation(); is<RenderInline>(parent); parent = parent->parent()) { |
| if (parent->isInFlowPositioned()) |
| offset += downcast<RenderInline>(*parent).offsetForInFlowPosition(); |
| } |
| return offset; |
| } |
| |
| static inline bool isOutOfFlowPositionedWithImplicitHeight(const RenderBoxModelObject& child) |
| { |
| return child.isOutOfFlowPositioned() && !child.style().logicalTop().isAuto() && !child.style().logicalBottom().isAuto(); |
| } |
| |
| RenderBlock* RenderBoxModelObject::containingBlockForAutoHeightDetection(Length logicalHeight) const |
| { |
| // For percentage heights: The percentage is calculated with respect to the |
| // height of the generated box's containing block. If the height of the |
| // containing block is not specified explicitly (i.e., it depends on content |
| // height), and this element is not absolutely positioned, the used height is |
| // calculated as if 'auto' was specified. |
| if (!logicalHeight.isPercentOrCalculated() || isOutOfFlowPositioned()) |
| return nullptr; |
| |
| // Anonymous block boxes are ignored when resolving percentage values that |
| // would refer to it: the closest non-anonymous ancestor box is used instead. |
| auto* cb = containingBlock(); |
| while (cb && cb->isAnonymous() && !is<RenderView>(cb)) |
| cb = cb->containingBlock(); |
| if (!cb) |
| return nullptr; |
| |
| // Matching RenderBox::percentageLogicalHeightIsResolvable() by |
| // ignoring table cell's attribute value, where it says that table cells |
| // violate what the CSS spec says to do with heights. Basically we don't care |
| // if the cell specified a height or not. |
| if (cb->isTableCell()) |
| return nullptr; |
| |
| // Match RenderBox::availableLogicalHeightUsing by special casing the layout |
| // view. The available height is taken from the frame. |
| if (cb->isRenderView()) |
| return nullptr; |
| |
| if (isOutOfFlowPositionedWithImplicitHeight(*cb)) |
| return nullptr; |
| |
| return cb; |
| } |
| |
| bool RenderBoxModelObject::hasAutoHeightOrContainingBlockWithAutoHeight() const |
| { |
| const auto* thisBox = isBox() ? downcast<RenderBox>(this) : nullptr; |
| Length logicalHeightLength = style().logicalHeight(); |
| auto* cb = containingBlockForAutoHeightDetection(logicalHeightLength); |
| |
| if (logicalHeightLength.isPercentOrCalculated() && cb && isBox()) |
| cb->addPercentHeightDescendant(*const_cast<RenderBox*>(downcast<RenderBox>(this))); |
| |
| if (thisBox && thisBox->isFlexItem() && downcast<RenderFlexibleBox>(*parent()).useChildOverridingLogicalHeightForPercentageResolution(*thisBox)) |
| return false; |
| |
| if (thisBox && thisBox->isGridItem() && thisBox->hasOverridingContainingBlockContentLogicalHeight()) |
| return thisBox->overridingContainingBlockContentLogicalHeight() == std::nullopt; |
| |
| if (logicalHeightLength.isAuto() && !isOutOfFlowPositionedWithImplicitHeight(*this)) |
| return true; |
| |
| // We need the containing block to have a definite block-size in order to resolve the block-size of the descendant, |
| // except when in quirks mode. Flexboxes follow strict behavior even in quirks mode, though. |
| if (!cb || (document().inQuirksMode() && !cb->isFlexibleBoxIncludingDeprecated())) |
| return false; |
| if (thisBox && thisBox->hasOverridingContainingBlockContentLogicalHeight()) |
| return thisBox->overridingContainingBlockContentLogicalHeight() == std::nullopt; |
| return !cb->hasDefiniteLogicalHeight(); |
| } |
| |
| DecodingMode RenderBoxModelObject::decodingModeForImageDraw(const Image& image, const PaintInfo& paintInfo) const |
| { |
| if (!is<BitmapImage>(image)) |
| return DecodingMode::Synchronous; |
| |
| const BitmapImage& bitmapImage = downcast<BitmapImage>(image); |
| if (bitmapImage.canAnimate()) { |
| // The DecodingMode for the current frame has to be Synchronous. The DecodingMode |
| // for the next frame will be calculated in BitmapImage::internalStartAnimation(). |
| return DecodingMode::Synchronous; |
| } |
| |
| // Large image case. |
| #if PLATFORM(IOS_FAMILY) |
| if (IOSApplication::isIBooksStorytime()) |
| return DecodingMode::Synchronous; |
| #endif |
| if (is<HTMLImageElement>(element())) { |
| auto decodingMode = downcast<HTMLImageElement>(*element()).decodingMode(); |
| if (decodingMode != DecodingMode::Auto) |
| return decodingMode; |
| } |
| if (bitmapImage.isLargeImageAsyncDecodingEnabledForTesting()) |
| return DecodingMode::Asynchronous; |
| if (document().isImageDocument()) |
| return DecodingMode::Synchronous; |
| if (paintInfo.paintBehavior.contains(PaintBehavior::Snapshotting)) |
| return DecodingMode::Synchronous; |
| if (!settings().largeImageAsyncDecodingEnabled()) |
| return DecodingMode::Synchronous; |
| if (!bitmapImage.canUseAsyncDecodingForLargeImages()) |
| return DecodingMode::Synchronous; |
| if (paintInfo.paintBehavior.contains(PaintBehavior::TileFirstPaint)) |
| return DecodingMode::Asynchronous; |
| // FIXME: isVisibleInViewport() is not cheap. Find a way to make this condition faster. |
| if (!isVisibleInViewport()) |
| return DecodingMode::Asynchronous; |
| return DecodingMode::Synchronous; |
| } |
| |
| LayoutSize RenderBoxModelObject::relativePositionOffset() const |
| { |
| // This function has been optimized to avoid calls to containingBlock() in the common case |
| // where all values are either auto or fixed. |
| |
| LayoutSize offset = accumulateInFlowPositionOffsets(this); |
| |
| // Objects that shrink to avoid floats normally use available line width when computing containing block width. However |
| // in the case of relative positioning using percentages, we can't do this. The offset should always be resolved using the |
| // available width of the containing block. Therefore we don't use containingBlockLogicalWidthForContent() here, but instead explicitly |
| // call availableWidth on our containing block. |
| // However for grid items the containing block is the grid area, so offsets should be resolved against that: |
| // https://drafts.csswg.org/css-grid/#grid-item-sizing |
| if (!style().left().isAuto() || !style().right().isAuto()) { |
| LayoutUnit availableWidth = hasOverridingContainingBlockContentWidth() |
| ? valueOrDefault(overridingContainingBlockContentWidth()) : containingBlock()->availableWidth(); |
| if (!style().left().isAuto()) { |
| if (!style().right().isAuto() && !containingBlock()->style().isLeftToRightDirection()) |
| offset.setWidth(-valueForLength(style().right(), !style().right().isFixed() ? availableWidth : 0_lu)); |
| else |
| offset.expand(valueForLength(style().left(), !style().left().isFixed() ? availableWidth : 0_lu), 0_lu); |
| } else if (!style().right().isAuto()) |
| offset.expand(-valueForLength(style().right(), !style().right().isFixed() ? availableWidth : 0_lu), 0_lu); |
| } |
| |
| // If the containing block of a relatively positioned element does not |
| // specify a height, a percentage top or bottom offset should be resolved as |
| // auto. An exception to this is if the containing block has the WinIE quirk |
| // where <html> and <body> assume the size of the viewport. In this case, |
| // calculate the percent offset based on this height. |
| // See <https://bugs.webkit.org/show_bug.cgi?id=26396>. |
| // Another exception is a grid item, as the containing block is the grid area: |
| // https://drafts.csswg.org/css-grid/#grid-item-sizing |
| if (!style().top().isAuto() |
| && (!style().top().isPercentOrCalculated() |
| || !containingBlock()->hasAutoHeightOrContainingBlockWithAutoHeight() |
| || containingBlock()->stretchesToViewport() |
| || hasOverridingContainingBlockContentHeight())) { |
| // FIXME: The computation of the available height is repeated later for "bottom". |
| // We could refactor this and move it to some common code for both ifs, however moving it outside of the ifs |
| // is not possible as it'd cause performance regressions. |
| offset.expand(0_lu, valueForLength(style().top(), !style().top().isFixed() |
| ? (hasOverridingContainingBlockContentHeight() ? overridingContainingBlockContentHeight().value_or(0_lu) : containingBlock()->availableHeight()) |
| : LayoutUnit())); |
| } else if (!style().bottom().isAuto() |
| && (!style().bottom().isPercentOrCalculated() |
| || !containingBlock()->hasAutoHeightOrContainingBlockWithAutoHeight() |
| || containingBlock()->stretchesToViewport() |
| || hasOverridingContainingBlockContentHeight())) { |
| // FIXME: Check comment above for "top", it applies here too. |
| offset.expand(0_lu, -valueForLength(style().bottom(), !style().bottom().isFixed() |
| ? (hasOverridingContainingBlockContentHeight() ? overridingContainingBlockContentHeight().value_or(0_lu) : containingBlock()->availableHeight()) |
| : LayoutUnit())); |
| } |
| |
| return offset; |
| } |
| |
| LayoutPoint RenderBoxModelObject::adjustedPositionRelativeToOffsetParent(const LayoutPoint& startPoint) const |
| { |
| // If the element is the HTML body element or doesn't have a parent |
| // return 0 and stop this algorithm. |
| if (isBody() || !parent()) |
| return LayoutPoint(); |
| |
| LayoutPoint referencePoint = startPoint; |
| |
| // If the offsetParent of the element is null, or is the HTML body element, |
| // return the distance between the canvas origin and the left border edge |
| // of the element and stop this algorithm. |
| if (const RenderBoxModelObject* offsetParent = this->offsetParent()) { |
| if (is<RenderBox>(*offsetParent) && !offsetParent->isBody() && !is<RenderTable>(*offsetParent)) |
| referencePoint.move(-downcast<RenderBox>(*offsetParent).borderLeft(), -downcast<RenderBox>(*offsetParent).borderTop()); |
| if (!isOutOfFlowPositioned() || enclosingFragmentedFlow()) { |
| if (isRelativelyPositioned()) |
| referencePoint.move(relativePositionOffset()); |
| else if (isStickilyPositioned()) |
| referencePoint.move(stickyPositionOffset()); |
| |
| // CSS regions specification says that region flows should return the body element as their offsetParent. |
| // Since we will bypass the body’s renderer anyway, just end the loop if we encounter a region flow (named flow thread). |
| // See http://dev.w3.org/csswg/css-regions/#cssomview-offset-attributes |
| auto* ancestor = parent(); |
| while (ancestor != offsetParent) { |
| // FIXME: What are we supposed to do inside SVG content? |
| |
| if (is<RenderMultiColumnFlow>(*ancestor)) { |
| // We need to apply a translation based off what region we are inside. |
| RenderFragmentContainer* fragment = downcast<RenderMultiColumnFlow>(*ancestor).physicalTranslationFromFlowToFragment(referencePoint); |
| if (fragment) |
| referencePoint.moveBy(fragment->topLeftLocation()); |
| } else if (!isOutOfFlowPositioned()) { |
| if (is<RenderBox>(*ancestor) && !is<RenderTableRow>(*ancestor)) |
| referencePoint.moveBy(downcast<RenderBox>(*ancestor).topLeftLocation()); |
| } |
| |
| ancestor = ancestor->parent(); |
| } |
| |
| if (is<RenderBox>(*offsetParent) && offsetParent->isBody() && !offsetParent->isPositioned()) |
| referencePoint.moveBy(downcast<RenderBox>(*offsetParent).topLeftLocation()); |
| } |
| } |
| |
| return referencePoint; |
| } |
| |
| std::pair<const RenderBox&, const RenderLayer*> RenderBoxModelObject::enclosingClippingBoxForStickyPosition() const |
| { |
| ASSERT(isStickilyPositioned()); |
| RenderLayer* clipLayer = hasLayer() ? layer()->enclosingOverflowClipLayer(ExcludeSelf) : nullptr; |
| const RenderBox& box = clipLayer ? downcast<RenderBox>(clipLayer->renderer()) : view(); |
| return { box, clipLayer }; |
| } |
| |
| void RenderBoxModelObject::computeStickyPositionConstraints(StickyPositionViewportConstraints& constraints, const FloatRect& constrainingRect) const |
| { |
| constraints.setConstrainingRectAtLastLayout(constrainingRect); |
| |
| // Do not use anonymous containing blocks to determine sticky constraints. We want the size |
| // of the first true containing block, because that is what imposes the limitation on the |
| // movement of stickily positioned items. |
| RenderBlock* containingBlock = this->containingBlock(); |
| while (containingBlock && (!is<RenderBlock>(*containingBlock) || containingBlock->isAnonymousBlock())) |
| containingBlock = containingBlock->containingBlock(); |
| ASSERT(containingBlock); |
| |
| auto [enclosingClippingBox, enclosingClippingLayer] = enclosingClippingBoxForStickyPosition(); |
| |
| LayoutRect containerContentRect; |
| if (!enclosingClippingLayer || (containingBlock != &enclosingClippingBox)) { |
| // In this case either the scrolling element is the view or there is another containing block in |
| // the hierarchy between this stickily positioned item and its scrolling ancestor. In both cases, |
| // we use the content box rectangle of the containing block, which is what should constrain the |
| // movement. |
| containerContentRect = containingBlock->computedCSSContentBoxRect(); |
| } else { |
| containerContentRect = containingBlock->layoutOverflowRect(); |
| containerContentRect.contract(LayoutBoxExtent { |
| containingBlock->computedCSSPaddingTop(), containingBlock->computedCSSPaddingRight(), |
| containingBlock->computedCSSPaddingBottom(), containingBlock->computedCSSPaddingLeft() }); |
| } |
| |
| LayoutUnit maxWidth = containingBlock->availableLogicalWidth(); |
| |
| // Sticky positioned element ignore any override logical width on the containing block (as they don't call |
| // containingBlockLogicalWidthForContent). It's unclear whether this is totally fine. |
| LayoutBoxExtent minMargin(minimumValueForLength(style().marginTop(), maxWidth), |
| minimumValueForLength(style().marginRight(), maxWidth), |
| minimumValueForLength(style().marginBottom(), maxWidth), |
| minimumValueForLength(style().marginLeft(), maxWidth)); |
| |
| // Compute the container-relative area within which the sticky element is allowed to move. |
| containerContentRect.contract(minMargin); |
| |
| // Finally compute container rect relative to the scrolling ancestor. We pass an empty |
| // mode here, because sticky positioning should ignore transforms. |
| FloatRect containerRectRelativeToScrollingAncestor = containingBlock->localToContainerQuad(FloatRect(containerContentRect), &enclosingClippingBox, { } /* ignore transforms */).boundingBox(); |
| if (enclosingClippingLayer) { |
| FloatPoint containerLocationRelativeToScrollingAncestor = containerRectRelativeToScrollingAncestor.location() - |
| FloatSize(enclosingClippingBox.borderLeft() + enclosingClippingBox.paddingLeft(), |
| enclosingClippingBox.borderTop() + enclosingClippingBox.paddingTop()); |
| if (&enclosingClippingBox != containingBlock) { |
| if (auto* scrollableArea = enclosingClippingLayer->scrollableArea()) |
| containerLocationRelativeToScrollingAncestor += scrollableArea->scrollOffset(); |
| } |
| containerRectRelativeToScrollingAncestor.setLocation(containerLocationRelativeToScrollingAncestor); |
| } |
| constraints.setContainingBlockRect(containerRectRelativeToScrollingAncestor); |
| |
| // Now compute the sticky box rect, also relative to the scrolling ancestor. |
| LayoutRect stickyBoxRect = frameRectForStickyPositioning(); |
| |
| // Ideally, it would be possible to call this->localToContainerQuad to determine the frame |
| // rectangle in the coordinate system of the scrolling ancestor, but localToContainerQuad |
| // itself depends on sticky positioning! Instead, start from the parent but first adjusting |
| // the rectangle for the writing mode of this stickily-positioned element. We also pass an |
| // empty mode here because sticky positioning should ignore transforms. |
| // |
| // FIXME: It would also be nice to not have to call localToContainerQuad again since we |
| // have already done a similar call to move from the containing block to the scrolling |
| // ancestor above, but localToContainerQuad takes care of a lot of complex situations |
| // involving inlines, tables, and transformations. |
| if (parent()->isBox()) |
| downcast<RenderBox>(parent())->flipForWritingMode(stickyBoxRect); |
| auto stickyBoxRelativeToScrollingAncestor = parent()->localToContainerQuad(FloatRect(stickyBoxRect), &enclosingClippingBox, { } /* ignore transforms */).boundingBox(); |
| |
| if (enclosingClippingLayer) { |
| stickyBoxRelativeToScrollingAncestor.move(-FloatSize(enclosingClippingBox.borderLeft() + enclosingClippingBox.paddingLeft(), |
| enclosingClippingBox.borderTop() + enclosingClippingBox.paddingTop())); |
| |
| if (&enclosingClippingBox != parent()) { |
| if (auto* scrollableArea = enclosingClippingLayer->scrollableArea()) |
| stickyBoxRelativeToScrollingAncestor.moveBy(scrollableArea->scrollOffset()); |
| } |
| } |
| constraints.setStickyBoxRect(stickyBoxRelativeToScrollingAncestor); |
| |
| if (!style().left().isAuto()) { |
| constraints.setLeftOffset(valueForLength(style().left(), constrainingRect.width())); |
| constraints.addAnchorEdge(ViewportConstraints::AnchorEdgeLeft); |
| } |
| |
| if (!style().right().isAuto()) { |
| constraints.setRightOffset(valueForLength(style().right(), constrainingRect.width())); |
| constraints.addAnchorEdge(ViewportConstraints::AnchorEdgeRight); |
| } |
| |
| if (!style().top().isAuto()) { |
| constraints.setTopOffset(valueForLength(style().top(), constrainingRect.height())); |
| constraints.addAnchorEdge(ViewportConstraints::AnchorEdgeTop); |
| } |
| |
| if (!style().bottom().isAuto()) { |
| constraints.setBottomOffset(valueForLength(style().bottom(), constrainingRect.height())); |
| constraints.addAnchorEdge(ViewportConstraints::AnchorEdgeBottom); |
| } |
| } |
| |
| FloatRect RenderBoxModelObject::constrainingRectForStickyPosition() const |
| { |
| RenderLayer* enclosingClippingLayer = hasLayer() ? layer()->enclosingOverflowClipLayer(ExcludeSelf) : nullptr; |
| |
| if (enclosingClippingLayer) { |
| RenderBox& enclosingClippingBox = downcast<RenderBox>(enclosingClippingLayer->renderer()); |
| LayoutRect clipRect = enclosingClippingBox.overflowClipRect(LayoutPoint(), nullptr); // FIXME: make this work in regions. |
| clipRect.contract(LayoutSize(enclosingClippingBox.paddingLeft() + enclosingClippingBox.paddingRight(), |
| enclosingClippingBox.paddingTop() + enclosingClippingBox.paddingBottom())); |
| |
| FloatRect constrainingRect = enclosingClippingBox.localToContainerQuad(FloatRect(clipRect), &view()).boundingBox(); |
| |
| auto* scrollableArea = enclosingClippingLayer->scrollableArea(); |
| FloatPoint scrollOffset; |
| if (scrollableArea) |
| scrollOffset = FloatPoint() + scrollableArea->scrollOffset(); |
| |
| float scrollbarOffset = 0; |
| if (enclosingClippingBox.hasLayer() && enclosingClippingBox.shouldPlaceVerticalScrollbarOnLeft() && scrollableArea) |
| scrollbarOffset = scrollableArea->verticalScrollbarWidth(IgnoreOverlayScrollbarSize); |
| |
| constrainingRect.setLocation(FloatPoint(scrollOffset.x() + scrollbarOffset, scrollOffset.y())); |
| return constrainingRect; |
| } |
| |
| return view().frameView().rectForFixedPositionLayout(); |
| } |
| |
| LayoutSize RenderBoxModelObject::stickyPositionOffset() const |
| { |
| FloatRect constrainingRect = constrainingRectForStickyPosition(); |
| StickyPositionViewportConstraints constraints; |
| computeStickyPositionConstraints(constraints, constrainingRect); |
| |
| // The sticky offset is physical, so we can just return the delta computed in absolute coords (though it may be wrong with transforms). |
| return LayoutSize(constraints.computeStickyOffset(constrainingRect)); |
| } |
| |
| LayoutSize RenderBoxModelObject::offsetForInFlowPosition() const |
| { |
| if (isRelativelyPositioned()) |
| return relativePositionOffset(); |
| |
| if (isStickilyPositioned()) |
| return stickyPositionOffset(); |
| |
| return LayoutSize(); |
| } |
| |
| LayoutUnit RenderBoxModelObject::offsetLeft() const |
| { |
| // Note that RenderInline and RenderBox override this to pass a different |
| // startPoint to adjustedPositionRelativeToOffsetParent. |
| return adjustedPositionRelativeToOffsetParent(LayoutPoint()).x(); |
| } |
| |
| LayoutUnit RenderBoxModelObject::offsetTop() const |
| { |
| // Note that RenderInline and RenderBox override this to pass a different |
| // startPoint to adjustedPositionRelativeToOffsetParent. |
| return adjustedPositionRelativeToOffsetParent(LayoutPoint()).y(); |
| } |
| |
| LayoutUnit RenderBoxModelObject::computedCSSPadding(const Length& padding) const |
| { |
| LayoutUnit w; |
| if (padding.isPercentOrCalculated()) |
| w = containingBlockLogicalWidthForContent(); |
| return minimumValueForLength(padding, w); |
| } |
| |
| RoundedRect RenderBoxModelObject::getBackgroundRoundedRect(const LayoutRect& borderRect, const InlineIterator::InlineBoxIterator& box, |
| bool includeLogicalLeftEdge, bool includeLogicalRightEdge) const |
| { |
| RoundedRect border = style().getRoundedBorderFor(borderRect, includeLogicalLeftEdge, includeLogicalRightEdge); |
| if (box && (box->nextInlineBox() || box->previousInlineBox())) { |
| RoundedRect segmentBorder = style().getRoundedBorderFor(LayoutRect(0_lu, 0_lu, borderRect.width(), borderRect.height()), includeLogicalLeftEdge, includeLogicalRightEdge); |
| border.setRadii(segmentBorder.radii()); |
| } |
| return border; |
| } |
| |
| void RenderBoxModelObject::clipRoundedInnerRect(GraphicsContext& context, const FloatRect& rect, const FloatRoundedRect& clipRect) |
| { |
| if (clipRect.isRenderable()) |
| context.clipRoundedRect(clipRect); |
| else { |
| // We create a rounded rect for each of the corners and clip it, while making sure we clip opposing corners together. |
| if (!clipRect.radii().topLeft().isEmpty() || !clipRect.radii().bottomRight().isEmpty()) { |
| FloatRect topCorner(clipRect.rect().x(), clipRect.rect().y(), rect.maxX() - clipRect.rect().x(), rect.maxY() - clipRect.rect().y()); |
| FloatRoundedRect::Radii topCornerRadii; |
| topCornerRadii.setTopLeft(clipRect.radii().topLeft()); |
| context.clipRoundedRect(FloatRoundedRect(topCorner, topCornerRadii)); |
| |
| FloatRect bottomCorner(rect.x(), rect.y(), clipRect.rect().maxX() - rect.x(), clipRect.rect().maxY() - rect.y()); |
| FloatRoundedRect::Radii bottomCornerRadii; |
| bottomCornerRadii.setBottomRight(clipRect.radii().bottomRight()); |
| context.clipRoundedRect(FloatRoundedRect(bottomCorner, bottomCornerRadii)); |
| } |
| |
| if (!clipRect.radii().topRight().isEmpty() || !clipRect.radii().bottomLeft().isEmpty()) { |
| FloatRect topCorner(rect.x(), clipRect.rect().y(), clipRect.rect().maxX() - rect.x(), rect.maxY() - clipRect.rect().y()); |
| FloatRoundedRect::Radii topCornerRadii; |
| topCornerRadii.setTopRight(clipRect.radii().topRight()); |
| context.clipRoundedRect(FloatRoundedRect(topCorner, topCornerRadii)); |
| |
| FloatRect bottomCorner(clipRect.rect().x(), rect.y(), rect.maxX() - clipRect.rect().x(), clipRect.rect().maxY() - rect.y()); |
| FloatRoundedRect::Radii bottomCornerRadii; |
| bottomCornerRadii.setBottomLeft(clipRect.radii().bottomLeft()); |
| context.clipRoundedRect(FloatRoundedRect(bottomCorner, bottomCornerRadii)); |
| } |
| } |
| } |
| |
| static LayoutRect shrinkRectByOneDevicePixel(const GraphicsContext& context, const LayoutRect& rect, float devicePixelRatio) |
| { |
| LayoutRect shrunkRect = rect; |
| AffineTransform transform = context.getCTM(); |
| shrunkRect.inflateX(-ceilToDevicePixel(1_lu / transform.xScale(), devicePixelRatio)); |
| shrunkRect.inflateY(-ceilToDevicePixel(1_lu / transform.yScale(), devicePixelRatio)); |
| return shrunkRect; |
| } |
| |
| LayoutRect RenderBoxModelObject::borderInnerRectAdjustedForBleedAvoidance(const GraphicsContext& context, const LayoutRect& rect, BackgroundBleedAvoidance bleedAvoidance) const |
| { |
| if (bleedAvoidance != BackgroundBleedBackgroundOverBorder) |
| return rect; |
| |
| // We shrink the rectangle by one device pixel on each side to make it fully overlap the anti-aliased background border |
| return shrinkRectByOneDevicePixel(context, rect, document().deviceScaleFactor()); |
| } |
| |
| RoundedRect RenderBoxModelObject::backgroundRoundedRectAdjustedForBleedAvoidance(const GraphicsContext& context, const LayoutRect& borderRect, BackgroundBleedAvoidance bleedAvoidance, const InlineIterator::InlineBoxIterator& box, bool includeLogicalLeftEdge, bool includeLogicalRightEdge) const |
| { |
| if (bleedAvoidance == BackgroundBleedShrinkBackground) { |
| // We shrink the rectangle by one device pixel on each side because the bleed is one pixel maximum. |
| return getBackgroundRoundedRect(shrinkRectByOneDevicePixel(context, borderRect, document().deviceScaleFactor()), box, |
| includeLogicalLeftEdge, includeLogicalRightEdge); |
| } |
| if (bleedAvoidance == BackgroundBleedBackgroundOverBorder) |
| return style().getRoundedInnerBorderFor(borderRect, includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| return getBackgroundRoundedRect(borderRect, box, includeLogicalLeftEdge, includeLogicalRightEdge); |
| } |
| |
| static void applyBoxShadowForBackground(GraphicsContext& context, const RenderStyle& style) |
| { |
| const ShadowData* boxShadow = style.boxShadow(); |
| while (boxShadow->style() != ShadowStyle::Normal) |
| boxShadow = boxShadow->next(); |
| |
| FloatSize shadowOffset(boxShadow->x().value(), boxShadow->y().value()); |
| context.setShadow(shadowOffset, boxShadow->radius().value(), style.colorByApplyingColorFilter(boxShadow->color()), boxShadow->isWebkitBoxShadow() ? ShadowRadiusMode::Legacy : ShadowRadiusMode::Default); |
| } |
| |
| InterpolationQuality RenderBoxModelObject::chooseInterpolationQuality(GraphicsContext& context, Image& image, const void* layer, const LayoutSize& size) |
| { |
| return view().imageQualityController().chooseInterpolationQuality(context, this, image, layer, size); |
| } |
| |
| void RenderBoxModelObject::paintMaskForTextFillBox(ImageBuffer* maskImage, const FloatRect& maskRect, const InlineIterator::InlineBoxIterator& inlineBox, const LayoutRect& scrolledPaintRect) |
| { |
| GraphicsContext& maskImageContext = maskImage->context(); |
| maskImageContext.translate(-maskRect.location()); |
| |
| // Now add the text to the clip. We do this by painting using a special paint phase that signals to |
| // the painter it should just modify the clip. |
| PaintInfo maskInfo(maskImageContext, LayoutRect { maskRect }, PaintPhase::TextClip, PaintBehavior::ForceBlackText); |
| if (inlineBox) { |
| auto paintOffset = scrolledPaintRect.location() - toLayoutSize(LayoutPoint(inlineBox->visualRectIgnoringBlockDirection().location())); |
| |
| for (auto box = inlineBox->firstLeafBox(), end = inlineBox->endLeafBox(); box != end; box.traverseNextOnLine()) { |
| if (!box->isText()) |
| continue; |
| if (auto* legacyTextBox = downcast<LegacyInlineTextBox>(box->legacyInlineBox())) { |
| LegacyTextBoxPainter textBoxPainter(*legacyTextBox, maskInfo, paintOffset); |
| textBoxPainter.paint(); |
| continue; |
| } |
| #if ENABLE(LAYOUT_FORMATTING_CONTEXT) |
| ModernTextBoxPainter textBoxPainter(box->modernPath().inlineContent(), box->modernPath().box(), maskInfo, paintOffset); |
| textBoxPainter.paint(); |
| #endif |
| } |
| return; |
| } |
| |
| LayoutSize localOffset = is<RenderBox>(*this) ? downcast<RenderBox>(*this).locationOffset() : LayoutSize(); |
| paint(maskInfo, scrolledPaintRect.location() - localOffset); |
| } |
| |
| void RenderBoxModelObject::paintFillLayerExtended(const PaintInfo& paintInfo, const Color& color, const FillLayer& bgLayer, const LayoutRect& rect, |
| BackgroundBleedAvoidance bleedAvoidance, const InlineIterator::InlineBoxIterator& box, const LayoutRect& backgroundImageStrip, CompositeOperator op, RenderElement* backgroundObject, BaseBackgroundColorUsage baseBgColorUsage) |
| { |
| GraphicsContext& context = paintInfo.context(); |
| |
| if ((context.paintingDisabled() && !context.detectingContentfulPaint()) || rect.isEmpty()) |
| return; |
| |
| auto [includeLeftEdge, includeRightEdge] = box ? box->hasClosedLeftAndRightEdge() : std::pair(true, true); |
| |
| bool hasRoundedBorder = style().hasBorderRadius() && (includeLeftEdge || includeRightEdge); |
| bool clippedWithLocalScrolling = hasNonVisibleOverflow() && bgLayer.attachment() == FillAttachment::LocalBackground; |
| bool isBorderFill = bgLayer.clip() == FillBox::Border; |
| bool isRoot = this->isDocumentElementRenderer(); |
| |
| Color bgColor = color; |
| StyleImage* bgImage = bgLayer.image(); |
| bool shouldPaintBackgroundImage = bgImage && bgImage->canRender(this, style().effectiveZoom()); |
| |
| if (context.detectingContentfulPaint()) { |
| if (!context.contenfulPaintDetected() && shouldPaintBackgroundImage && bgImage->cachedImage()) { |
| if (style().backgroundSizeType() != FillSizeType::Size || !style().backgroundSizeLength().isEmpty()) |
| context.setContentfulPaintDetected(); |
| return; |
| } |
| } |
| |
| if (context.invalidatingImagesWithAsyncDecodes()) { |
| if (shouldPaintBackgroundImage && bgImage->cachedImage()->isClientWaitingForAsyncDecoding(*this)) |
| bgImage->cachedImage()->removeAllClientsWaitingForAsyncDecoding(); |
| return; |
| } |
| |
| bool forceBackgroundToWhite = false; |
| if (document().printing()) { |
| if (style().printColorAdjust() == PrintColorAdjust::Economy) |
| forceBackgroundToWhite = true; |
| if (settings().shouldPrintBackgrounds()) |
| forceBackgroundToWhite = false; |
| } |
| |
| // When printing backgrounds is disabled or using economy mode, |
| // change existing background colors and images to a solid white background. |
| // If there's no bg color or image, leave it untouched to avoid affecting transparency. |
| // We don't try to avoid loading the background images, because this style flag is only set |
| // when printing, and at that point we've already loaded the background images anyway. (To avoid |
| // loading the background images we'd have to do this check when applying styles rather than |
| // while rendering.) |
| if (forceBackgroundToWhite) { |
| // Note that we can't reuse this variable below because the bgColor might be changed |
| bool shouldPaintBackgroundColor = !bgLayer.next() && bgColor.isVisible(); |
| if (shouldPaintBackgroundImage || shouldPaintBackgroundColor) { |
| bgColor = Color::white; |
| shouldPaintBackgroundImage = false; |
| } |
| } |
| |
| bool baseBgColorOnly = (baseBgColorUsage == BaseBackgroundColorOnly); |
| if (baseBgColorOnly && (!isRoot || bgLayer.next() || bgColor.isOpaque())) |
| return; |
| |
| bool colorVisible = bgColor.isVisible(); |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| FloatRect pixelSnappedRect = snapRectToDevicePixels(rect, deviceScaleFactor); |
| |
| // Fast path for drawing simple color backgrounds. |
| if (!isRoot && !clippedWithLocalScrolling && !shouldPaintBackgroundImage && isBorderFill && !bgLayer.next()) { |
| if (!colorVisible) |
| return; |
| |
| bool boxShadowShouldBeAppliedToBackground = this->boxShadowShouldBeAppliedToBackground(rect.location(), bleedAvoidance, box); |
| GraphicsContextStateSaver shadowStateSaver(context, boxShadowShouldBeAppliedToBackground); |
| if (boxShadowShouldBeAppliedToBackground) |
| applyBoxShadowForBackground(context, style()); |
| |
| if (hasRoundedBorder && bleedAvoidance != BackgroundBleedUseTransparencyLayer) { |
| FloatRoundedRect pixelSnappedBorder = backgroundRoundedRectAdjustedForBleedAvoidance(context, rect, bleedAvoidance, box, |
| includeLeftEdge, includeRightEdge).pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| if (pixelSnappedBorder.isRenderable()) { |
| CompositeOperator previousOperator = context.compositeOperation(); |
| bool saveRestoreCompositeOp = op != previousOperator; |
| if (saveRestoreCompositeOp) |
| context.setCompositeOperation(op); |
| |
| context.fillRoundedRect(pixelSnappedBorder, bgColor); |
| |
| if (saveRestoreCompositeOp) |
| context.setCompositeOperation(previousOperator); |
| } else { |
| context.save(); |
| clipRoundedInnerRect(context, pixelSnappedRect, pixelSnappedBorder); |
| context.fillRect(pixelSnappedBorder.rect(), bgColor, op); |
| context.restore(); |
| } |
| } else |
| context.fillRect(pixelSnappedRect, bgColor, op); |
| |
| return; |
| } |
| |
| // FillBox::Border radius clipping is taken care of by BackgroundBleedUseTransparencyLayer |
| bool clipToBorderRadius = hasRoundedBorder && !(isBorderFill && bleedAvoidance == BackgroundBleedUseTransparencyLayer); |
| GraphicsContextStateSaver clipToBorderStateSaver(context, clipToBorderRadius); |
| if (clipToBorderRadius) { |
| RoundedRect border = isBorderFill ? backgroundRoundedRectAdjustedForBleedAvoidance(context, rect, bleedAvoidance, box, includeLeftEdge, includeRightEdge) : getBackgroundRoundedRect(rect, box, includeLeftEdge, includeRightEdge); |
| |
| // Clip to the padding or content boxes as necessary. |
| if (bgLayer.clip() == FillBox::Content) { |
| border = style().getRoundedInnerBorderFor(border.rect(), |
| paddingTop() + borderTop(), paddingBottom() + borderBottom(), paddingLeft() + borderLeft(), paddingRight() + borderRight(), includeLeftEdge, includeRightEdge); |
| } else if (bgLayer.clip() == FillBox::Padding) |
| border = style().getRoundedInnerBorderFor(border.rect(), includeLeftEdge, includeRightEdge); |
| |
| clipRoundedInnerRect(context, pixelSnappedRect, border.pixelSnappedRoundedRectForPainting(deviceScaleFactor)); |
| } |
| |
| LayoutUnit bLeft = includeLeftEdge ? borderLeft() : 0_lu; |
| LayoutUnit bRight = includeRightEdge ? borderRight() : 0_lu; |
| LayoutUnit pLeft = includeLeftEdge ? paddingLeft() : 0_lu; |
| LayoutUnit pRight = includeRightEdge ? paddingRight() : 0_lu; |
| |
| GraphicsContextStateSaver clipWithScrollingStateSaver(context, clippedWithLocalScrolling); |
| LayoutRect scrolledPaintRect = rect; |
| if (clippedWithLocalScrolling) { |
| // Clip to the overflow area. |
| auto& thisBox = downcast<RenderBox>(*this); |
| context.clip(thisBox.overflowClipRect(rect.location())); |
| |
| // Adjust the paint rect to reflect a scrolled content box with borders at the ends. |
| scrolledPaintRect.moveBy(-thisBox.scrollPosition()); |
| scrolledPaintRect.setWidth(bLeft + layer()->scrollWidth() + bRight); |
| scrolledPaintRect.setHeight(borderTop() + layer()->scrollHeight() + borderBottom()); |
| } |
| |
| GraphicsContextStateSaver backgroundClipStateSaver(context, false); |
| RefPtr<ImageBuffer> maskImage; |
| FloatRect maskRect; |
| |
| if (bgLayer.clip() == FillBox::Padding || bgLayer.clip() == FillBox::Content) { |
| // Clip to the padding or content boxes as necessary. |
| if (!clipToBorderRadius) { |
| bool includePadding = bgLayer.clip() == FillBox::Content; |
| LayoutRect clipRect = LayoutRect(scrolledPaintRect.x() + bLeft + (includePadding ? pLeft : 0_lu), |
| scrolledPaintRect.y() + borderTop() + (includePadding ? paddingTop() : 0_lu), |
| scrolledPaintRect.width() - bLeft - bRight - (includePadding ? pLeft + pRight : 0_lu), |
| scrolledPaintRect.height() - borderTop() - borderBottom() - (includePadding ? paddingTop() + paddingBottom() : 0_lu)); |
| backgroundClipStateSaver.save(); |
| context.clip(clipRect); |
| } |
| } else if (bgLayer.clip() == FillBox::Text) { |
| // We have to draw our text into a mask that can then be used to clip background drawing. |
| // First figure out how big the mask has to be. It should be no bigger than what we need |
| // to actually render, so we should intersect the dirty rect with the border box of the background. |
| maskRect = snapRectToDevicePixels(rect, deviceScaleFactor); |
| maskRect.intersect(snapRectToDevicePixels(paintInfo.rect, deviceScaleFactor)); |
| |
| // Now create the mask. |
| maskImage = context.createAlignedImageBuffer(maskRect.size()); |
| if (!maskImage) |
| return; |
| paintMaskForTextFillBox(maskImage.get(), maskRect, box, scrolledPaintRect); |
| |
| // The mask has been created. Now we just need to clip to it. |
| backgroundClipStateSaver.save(); |
| context.clip(maskRect); |
| context.beginTransparencyLayer(1); |
| } |
| |
| auto isOpaqueRoot = false; |
| if (isRoot) { |
| isOpaqueRoot = bgLayer.next() || bgColor.isOpaque() || view().shouldPaintBaseBackground(); |
| view().frameView().setContentIsOpaque(isOpaqueRoot); |
| } |
| |
| // Paint the color first underneath all images, culled if background image occludes it. |
| // FIXME: In the bgLayer.hasFiniteBounds() case, we could improve the culling test |
| // by verifying whether the background image covers the entire layout rect. |
| if (!bgLayer.next()) { |
| LayoutRect backgroundRect(scrolledPaintRect); |
| bool boxShadowShouldBeAppliedToBackground = this->boxShadowShouldBeAppliedToBackground(rect.location(), bleedAvoidance, box); |
| if (boxShadowShouldBeAppliedToBackground || !shouldPaintBackgroundImage || !bgLayer.hasOpaqueImage(*this) || !bgLayer.hasRepeatXY() || bgLayer.isEmpty()) { |
| if (!boxShadowShouldBeAppliedToBackground) |
| backgroundRect.intersect(paintInfo.rect); |
| |
| // If we have an alpha and we are painting the root element, blend with the base background color. |
| Color baseColor; |
| bool shouldClearBackground = false; |
| if ((baseBgColorUsage != BaseBackgroundColorSkip) && isOpaqueRoot) { |
| baseColor = view().frameView().baseBackgroundColor(); |
| if (!baseColor.isVisible()) |
| shouldClearBackground = true; |
| } |
| |
| GraphicsContextStateSaver shadowStateSaver(context, boxShadowShouldBeAppliedToBackground); |
| if (boxShadowShouldBeAppliedToBackground) |
| applyBoxShadowForBackground(context, style()); |
| |
| FloatRect backgroundRectForPainting = snapRectToDevicePixels(backgroundRect, deviceScaleFactor); |
| if (baseColor.isVisible()) { |
| if (!baseBgColorOnly && bgColor.isVisible()) |
| baseColor = blendSourceOver(baseColor, bgColor); |
| context.fillRect(backgroundRectForPainting, baseColor, CompositeOperator::Copy); |
| } else if (!baseBgColorOnly && bgColor.isVisible()) { |
| auto operation = context.compositeOperation(); |
| if (shouldClearBackground) { |
| if (op == CompositeOperator::DestinationOut) // We're punching out the background. |
| operation = op; |
| else |
| operation = CompositeOperator::Copy; |
| } |
| context.fillRect(backgroundRectForPainting, bgColor, operation); |
| } else if (shouldClearBackground) |
| context.clearRect(backgroundRectForPainting); |
| } |
| } |
| |
| // no progressive loading of the background image |
| if (!baseBgColorOnly && shouldPaintBackgroundImage) { |
| // Multiline inline boxes paint like the image was one long strip spanning lines. The backgroundImageStrip is this fictional rectangle. |
| auto imageRect = backgroundImageStrip.isEmpty() ? scrolledPaintRect : backgroundImageStrip; |
| auto paintOffset = backgroundImageStrip.isEmpty() ? rect.location() : backgroundImageStrip.location(); |
| auto geometry = calculateBackgroundImageGeometry(paintInfo.paintContainer, bgLayer, paintOffset, imageRect, backgroundObject); |
| geometry.clip(LayoutRect(pixelSnappedRect)); |
| RefPtr<Image> image; |
| if (!geometry.destRect().isEmpty() && (image = bgImage->image(backgroundObject ? backgroundObject : this, geometry.tileSize()))) { |
| context.setDrawLuminanceMask(bgLayer.maskMode() == MaskMode::Luminance); |
| |
| if (is<BitmapImage>(image)) |
| downcast<BitmapImage>(*image).updateFromSettings(settings()); |
| |
| ImagePaintingOptions options = { |
| op == CompositeOperator::SourceOver ? bgLayer.compositeForPainting() : op, |
| bgLayer.blendMode(), |
| decodingModeForImageDraw(*image, paintInfo), |
| ImageOrientation::FromImage, |
| chooseInterpolationQuality(context, *image, &bgLayer, geometry.tileSize()) |
| }; |
| |
| auto drawResult = context.drawTiledImage(*image, geometry.destRect(), toLayoutPoint(geometry.relativePhase()), geometry.tileSize(), geometry.spaceSize(), options); |
| if (drawResult == ImageDrawResult::DidRequestDecoding) { |
| ASSERT(bgImage->hasCachedImage()); |
| bgImage->cachedImage()->addClientWaitingForAsyncDecoding(*this); |
| } |
| } |
| } |
| |
| if (maskImage && bgLayer.clip() == FillBox::Text) { |
| context.drawConsumingImageBuffer(WTFMove(maskImage), maskRect, CompositeOperator::DestinationIn); |
| context.endTransparencyLayer(); |
| } |
| } |
| |
| static inline LayoutUnit resolveWidthForRatio(LayoutUnit height, const LayoutSize& intrinsicRatio) |
| { |
| return height * intrinsicRatio.width() / intrinsicRatio.height(); |
| } |
| |
| static inline LayoutUnit resolveHeightForRatio(LayoutUnit width, const LayoutSize& intrinsicRatio) |
| { |
| return width * intrinsicRatio.height() / intrinsicRatio.width(); |
| } |
| |
| static inline LayoutSize resolveAgainstIntrinsicWidthOrHeightAndRatio(const LayoutSize& size, const LayoutSize& intrinsicRatio, LayoutUnit useWidth, LayoutUnit useHeight) |
| { |
| if (intrinsicRatio.isEmpty()) { |
| if (useWidth) |
| return LayoutSize(useWidth, size.height()); |
| return LayoutSize(size.width(), useHeight); |
| } |
| |
| if (useWidth) |
| return LayoutSize(useWidth, resolveHeightForRatio(useWidth, intrinsicRatio)); |
| return LayoutSize(resolveWidthForRatio(useHeight, intrinsicRatio), useHeight); |
| } |
| |
| static inline LayoutSize resolveAgainstIntrinsicRatio(const LayoutSize& size, const LayoutSize& intrinsicRatio) |
| { |
| // Two possible solutions: (size.width(), solutionHeight) or (solutionWidth, size.height()) |
| // "... must be assumed to be the largest dimensions..." = easiest answer: the rect with the largest surface area. |
| |
| LayoutUnit solutionWidth = resolveWidthForRatio(size.height(), intrinsicRatio); |
| LayoutUnit solutionHeight = resolveHeightForRatio(size.width(), intrinsicRatio); |
| if (solutionWidth <= size.width()) { |
| if (solutionHeight <= size.height()) { |
| // If both solutions fit, choose the one covering the larger area. |
| LayoutUnit areaOne = solutionWidth * size.height(); |
| LayoutUnit areaTwo = size.width() * solutionHeight; |
| if (areaOne < areaTwo) |
| return LayoutSize(size.width(), solutionHeight); |
| return LayoutSize(solutionWidth, size.height()); |
| } |
| |
| // Only the first solution fits. |
| return LayoutSize(solutionWidth, size.height()); |
| } |
| |
| // Only the second solution fits, assert that. |
| ASSERT(solutionHeight <= size.height()); |
| return LayoutSize(size.width(), solutionHeight); |
| } |
| |
| LayoutSize RenderBoxModelObject::calculateImageIntrinsicDimensions(StyleImage* image, const LayoutSize& positioningAreaSize, ScaleByEffectiveZoomOrNot shouldScaleOrNot) const |
| { |
| // A generated image without a fixed size, will always return the container size as intrinsic size. |
| if (image->isGeneratedImage() && image->usesImageContainerSize()) |
| return LayoutSize(positioningAreaSize.width(), positioningAreaSize.height()); |
| |
| Length intrinsicWidth; |
| Length intrinsicHeight; |
| FloatSize intrinsicRatio; |
| image->computeIntrinsicDimensions(this, intrinsicWidth, intrinsicHeight, intrinsicRatio); |
| |
| ASSERT(!intrinsicWidth.isPercentOrCalculated()); |
| ASSERT(!intrinsicHeight.isPercentOrCalculated()); |
| |
| LayoutSize resolvedSize(intrinsicWidth.value(), intrinsicHeight.value()); |
| LayoutSize minimumSize(resolvedSize.width() > 0 ? 1 : 0, resolvedSize.height() > 0 ? 1 : 0); |
| |
| if (shouldScaleOrNot == ScaleByEffectiveZoom) |
| resolvedSize.scale(style().effectiveZoom()); |
| resolvedSize.clampToMinimumSize(minimumSize); |
| |
| if (!resolvedSize.isEmpty()) |
| return resolvedSize; |
| |
| // If the image has one of either an intrinsic width or an intrinsic height: |
| // * and an intrinsic aspect ratio, then the missing dimension is calculated from the given dimension and the ratio. |
| // * and no intrinsic aspect ratio, then the missing dimension is assumed to be the size of the rectangle that |
| // establishes the coordinate system for the 'background-position' property. |
| if (resolvedSize.width() > 0 || resolvedSize.height() > 0) |
| return resolveAgainstIntrinsicWidthOrHeightAndRatio(positioningAreaSize, LayoutSize(intrinsicRatio), resolvedSize.width(), resolvedSize.height()); |
| |
| // If the image has no intrinsic dimensions and has an intrinsic ratio the dimensions must be assumed to be the |
| // largest dimensions at that ratio such that neither dimension exceeds the dimensions of the rectangle that |
| // establishes the coordinate system for the 'background-position' property. |
| if (!intrinsicRatio.isEmpty()) |
| return resolveAgainstIntrinsicRatio(positioningAreaSize, LayoutSize(intrinsicRatio)); |
| |
| // If the image has no intrinsic ratio either, then the dimensions must be assumed to be the rectangle that |
| // establishes the coordinate system for the 'background-position' property. |
| return positioningAreaSize; |
| } |
| |
| LayoutSize RenderBoxModelObject::calculateFillTileSize(const FillLayer& fillLayer, const LayoutSize& positioningAreaSize) const |
| { |
| StyleImage* image = fillLayer.image(); |
| FillSizeType type = fillLayer.size().type; |
| auto devicePixelSize = LayoutUnit { 1.0 / document().deviceScaleFactor() }; |
| |
| LayoutSize imageIntrinsicSize; |
| if (image) { |
| imageIntrinsicSize = calculateImageIntrinsicDimensions(image, positioningAreaSize, ScaleByEffectiveZoom); |
| imageIntrinsicSize.scale(1 / image->imageScaleFactor(), 1 / image->imageScaleFactor()); |
| } else |
| imageIntrinsicSize = positioningAreaSize; |
| |
| switch (type) { |
| case FillSizeType::Size: { |
| LayoutSize tileSize = positioningAreaSize; |
| |
| Length layerWidth = fillLayer.size().size.width; |
| Length layerHeight = fillLayer.size().size.height; |
| |
| if (layerWidth.isFixed()) |
| tileSize.setWidth(layerWidth.value()); |
| else if (layerWidth.isPercentOrCalculated()) { |
| auto resolvedWidth = valueForLength(layerWidth, positioningAreaSize.width()); |
| // Non-zero resolved value should always produce some content. |
| tileSize.setWidth(!resolvedWidth ? resolvedWidth : std::max(devicePixelSize, resolvedWidth)); |
| } |
| |
| if (layerHeight.isFixed()) |
| tileSize.setHeight(layerHeight.value()); |
| else if (layerHeight.isPercentOrCalculated()) { |
| auto resolvedHeight = valueForLength(layerHeight, positioningAreaSize.height()); |
| // Non-zero resolved value should always produce some content. |
| tileSize.setHeight(!resolvedHeight ? resolvedHeight : std::max(devicePixelSize, resolvedHeight)); |
| } |
| |
| // If one of the values is auto we have to use the appropriate |
| // scale to maintain our aspect ratio. |
| if (layerWidth.isAuto() && !layerHeight.isAuto()) { |
| if (imageIntrinsicSize.height()) |
| tileSize.setWidth(imageIntrinsicSize.width() * tileSize.height() / imageIntrinsicSize.height()); |
| } else if (!layerWidth.isAuto() && layerHeight.isAuto()) { |
| if (imageIntrinsicSize.width()) |
| tileSize.setHeight(imageIntrinsicSize.height() * tileSize.width() / imageIntrinsicSize.width()); |
| } else if (layerWidth.isAuto() && layerHeight.isAuto()) { |
| // If both width and height are auto, use the image's intrinsic size. |
| tileSize = imageIntrinsicSize; |
| } |
| |
| tileSize.clampNegativeToZero(); |
| return tileSize; |
| } |
| case FillSizeType::None: { |
| // If both values are ‘auto’ then the intrinsic width and/or height of the image should be used, if any. |
| if (!imageIntrinsicSize.isEmpty()) |
| return imageIntrinsicSize; |
| |
| // If the image has neither an intrinsic width nor an intrinsic height, its size is determined as for ‘contain’. |
| type = FillSizeType::Contain; |
| } |
| FALLTHROUGH; |
| case FillSizeType::Contain: |
| case FillSizeType::Cover: { |
| // Scale computation needs higher precision than what LayoutUnit can offer. |
| FloatSize localImageIntrinsicSize = imageIntrinsicSize; |
| FloatSize localPositioningAreaSize = positioningAreaSize; |
| |
| float horizontalScaleFactor = localImageIntrinsicSize.width() ? (localPositioningAreaSize.width() / localImageIntrinsicSize.width()) : 1; |
| float verticalScaleFactor = localImageIntrinsicSize.height() ? (localPositioningAreaSize.height() / localImageIntrinsicSize.height()) : 1; |
| float scaleFactor = type == FillSizeType::Contain ? std::min(horizontalScaleFactor, verticalScaleFactor) : std::max(horizontalScaleFactor, verticalScaleFactor); |
| |
| if (localImageIntrinsicSize.isEmpty()) |
| return { }; |
| |
| return LayoutSize(localImageIntrinsicSize.scaled(scaleFactor).expandedTo({ devicePixelSize, devicePixelSize })); |
| } |
| } |
| |
| ASSERT_NOT_REACHED(); |
| return { }; |
| } |
| |
| static void pixelSnapBackgroundImageGeometryForPainting(LayoutRect& destinationRect, LayoutSize& tileSize, LayoutSize& phase, LayoutSize& space, float scaleFactor) |
| { |
| tileSize = LayoutSize(snapRectToDevicePixels(LayoutRect(destinationRect.location(), tileSize), scaleFactor).size()); |
| phase = LayoutSize(snapRectToDevicePixels(LayoutRect(destinationRect.location(), phase), scaleFactor).size()); |
| space = LayoutSize(snapRectToDevicePixels(LayoutRect(LayoutPoint(), space), scaleFactor).size()); |
| destinationRect = LayoutRect(snapRectToDevicePixels(destinationRect, scaleFactor)); |
| } |
| |
| bool RenderBoxModelObject::fixedBackgroundPaintsInLocalCoordinates() const |
| { |
| if (!isDocumentElementRenderer()) |
| return false; |
| |
| if (view().frameView().paintBehavior().contains(PaintBehavior::FlattenCompositingLayers)) |
| return false; |
| |
| RenderLayer* rootLayer = view().layer(); |
| if (!rootLayer || !rootLayer->isComposited()) |
| return false; |
| |
| return rootLayer->backing()->backgroundLayerPaintsFixedRootBackground(); |
| } |
| |
| static inline LayoutUnit getSpace(LayoutUnit areaSize, LayoutUnit tileSize) |
| { |
| int numberOfTiles = areaSize / tileSize; |
| LayoutUnit space = -1; |
| |
| if (numberOfTiles > 1) |
| space = (areaSize - numberOfTiles * tileSize) / (numberOfTiles - 1); |
| |
| return space; |
| } |
| |
| static LayoutUnit resolveEdgeRelativeLength(const Length& length, Edge edge, LayoutUnit availableSpace, const LayoutSize& areaSize, const LayoutSize& tileSize) |
| { |
| LayoutUnit result = minimumValueForLength(length, availableSpace); |
| |
| if (edge == Edge::Right) |
| return areaSize.width() - tileSize.width() - result; |
| |
| if (edge == Edge::Bottom) |
| return areaSize.height() - tileSize.height() - result; |
| |
| return result; |
| } |
| |
| BackgroundImageGeometry RenderBoxModelObject::calculateBackgroundImageGeometry(const RenderLayerModelObject* paintContainer, const FillLayer& fillLayer, const LayoutPoint& paintOffset, |
| const LayoutRect& borderBoxRect, RenderElement* backgroundObject) const |
| { |
| LayoutUnit left; |
| LayoutUnit top; |
| LayoutSize positioningAreaSize; |
| // Determine the background positioning area and set destination rect to the background painting area. |
| // Destination rect will be adjusted later if the background is non-repeating. |
| // FIXME: transforms spec says that fixed backgrounds behave like scroll inside transforms. https://bugs.webkit.org/show_bug.cgi?id=15679 |
| LayoutRect destinationRect(borderBoxRect); |
| bool fixedAttachment = fillLayer.attachment() == FillAttachment::FixedBackground; |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| if (!fixedAttachment) { |
| LayoutUnit right; |
| LayoutUnit bottom; |
| // Scroll and Local. |
| if (fillLayer.origin() != FillBox::Border) { |
| left = borderLeft(); |
| right = borderRight(); |
| top = borderTop(); |
| bottom = borderBottom(); |
| if (fillLayer.origin() == FillBox::Content) { |
| left += paddingLeft(); |
| right += paddingRight(); |
| top += paddingTop(); |
| bottom += paddingBottom(); |
| } |
| } |
| |
| // The background of the box generated by the root element covers the entire canvas including |
| // its margins. Since those were added in already, we have to factor them out when computing |
| // the background positioning area. |
| if (isDocumentElementRenderer()) { |
| positioningAreaSize = downcast<RenderBox>(*this).size() - LayoutSize(left + right, top + bottom); |
| positioningAreaSize = LayoutSize(snapSizeToDevicePixel(positioningAreaSize, LayoutPoint(), deviceScaleFactor)); |
| if (view().frameView().hasExtendedBackgroundRectForPainting()) { |
| LayoutRect extendedBackgroundRect = view().frameView().extendedBackgroundRectForPainting(); |
| left += (marginLeft() - extendedBackgroundRect.x()); |
| top += (marginTop() - extendedBackgroundRect.y()); |
| } |
| } else { |
| positioningAreaSize = borderBoxRect.size() - LayoutSize(left + right, top + bottom); |
| positioningAreaSize = LayoutSize(snapRectToDevicePixels(LayoutRect(paintOffset, positioningAreaSize), deviceScaleFactor).size()); |
| } |
| } else { |
| LayoutRect viewportRect; |
| float topContentInset = 0; |
| if (settings().fixedBackgroundsPaintRelativeToDocument()) |
| viewportRect = view().unscaledDocumentRect(); |
| else { |
| FrameView& frameView = view().frameView(); |
| bool useFixedLayout = frameView.useFixedLayout() && !frameView.fixedLayoutSize().isEmpty(); |
| |
| if (useFixedLayout) { |
| // Use the fixedLayoutSize() when useFixedLayout() because the rendering will scale |
| // down the frameView to to fit in the current viewport. |
| viewportRect.setSize(frameView.fixedLayoutSize()); |
| } else |
| viewportRect.setSize(frameView.sizeForVisibleContent()); |
| |
| if (fixedBackgroundPaintsInLocalCoordinates()) { |
| if (!useFixedLayout) { |
| // Shifting location up by topContentInset is needed for layout tests which expect |
| // layout to be shifted down when calling window.internals.setTopContentInset(). |
| topContentInset = frameView.topContentInset(ScrollView::TopContentInsetType::WebCoreOrPlatformContentInset); |
| viewportRect.setLocation(LayoutPoint(0, -topContentInset)); |
| } |
| } else if (useFixedLayout || frameView.frameScaleFactor() != 1) { |
| // scrollPositionForFixedPosition() is adjusted for page scale and it does not include |
| // topContentInset so do not add it to the calculation below. |
| viewportRect.setLocation(frameView.scrollPositionForFixedPosition()); |
| } else { |
| // documentScrollPositionRelativeToViewOrigin() includes -topContentInset in its height |
| // so we need to account for that in calculating the phase size |
| topContentInset = frameView.topContentInset(ScrollView::TopContentInsetType::WebCoreOrPlatformContentInset); |
| viewportRect.setLocation(frameView.documentScrollPositionRelativeToViewOrigin()); |
| } |
| |
| top += topContentInset; |
| } |
| |
| if (paintContainer) |
| viewportRect.moveBy(LayoutPoint(-paintContainer->localToAbsolute(FloatPoint()))); |
| |
| destinationRect = viewportRect; |
| positioningAreaSize = destinationRect.size(); |
| positioningAreaSize.setHeight(positioningAreaSize.height() - topContentInset); |
| positioningAreaSize = LayoutSize(snapRectToDevicePixels(LayoutRect(destinationRect.location(), positioningAreaSize), deviceScaleFactor).size()); |
| } |
| |
| auto clientForBackgroundImage = backgroundObject ? backgroundObject : this; |
| LayoutSize tileSize = calculateFillTileSize(fillLayer, positioningAreaSize); |
| if (StyleImage* layerImage = fillLayer.image()) |
| layerImage->setContainerContextForRenderer(*clientForBackgroundImage, tileSize, style().effectiveZoom()); |
| |
| FillRepeat backgroundRepeatX = fillLayer.repeatX(); |
| FillRepeat backgroundRepeatY = fillLayer.repeatY(); |
| LayoutUnit availableWidth = positioningAreaSize.width() - tileSize.width(); |
| LayoutUnit availableHeight = positioningAreaSize.height() - tileSize.height(); |
| |
| LayoutSize spaceSize; |
| LayoutSize phase; |
| LayoutSize noRepeat; |
| LayoutUnit computedXPosition = resolveEdgeRelativeLength(fillLayer.xPosition(), fillLayer.backgroundXOrigin(), availableWidth, positioningAreaSize, tileSize); |
| if (backgroundRepeatX == FillRepeat::Round && positioningAreaSize.width() > 0 && tileSize.width() > 0) { |
| int numTiles = std::max(1, roundToInt(positioningAreaSize.width() / tileSize.width())); |
| if (fillLayer.size().size.height.isAuto() && backgroundRepeatY != FillRepeat::Round) |
| tileSize.setHeight(tileSize.height() * positioningAreaSize.width() / (numTiles * tileSize.width())); |
| |
| tileSize.setWidth(positioningAreaSize.width() / numTiles); |
| phase.setWidth(tileSize.width() ? tileSize.width() - fmodf((computedXPosition + left), tileSize.width()) : 0); |
| } |
| |
| LayoutUnit computedYPosition = resolveEdgeRelativeLength(fillLayer.yPosition(), fillLayer.backgroundYOrigin(), availableHeight, positioningAreaSize, tileSize); |
| if (backgroundRepeatY == FillRepeat::Round && positioningAreaSize.height() > 0 && tileSize.height() > 0) { |
| int numTiles = std::max(1, roundToInt(positioningAreaSize.height() / tileSize.height())); |
| if (fillLayer.size().size.width.isAuto() && backgroundRepeatX != FillRepeat::Round) |
| tileSize.setWidth(tileSize.width() * positioningAreaSize.height() / (numTiles * tileSize.height())); |
| |
| tileSize.setHeight(positioningAreaSize.height() / numTiles); |
| phase.setHeight(tileSize.height() ? tileSize.height() - fmodf((computedYPosition + top), tileSize.height()) : 0); |
| } |
| |
| if (backgroundRepeatX == FillRepeat::Repeat) { |
| phase.setWidth(tileSize.width() ? tileSize.width() - fmodf(computedXPosition + left, tileSize.width()) : 0); |
| spaceSize.setWidth(0); |
| } else if (backgroundRepeatX == FillRepeat::Space && tileSize.width() > 0) { |
| LayoutUnit space = getSpace(positioningAreaSize.width(), tileSize.width()); |
| if (space >= 0) { |
| LayoutUnit actualWidth = tileSize.width() + space; |
| computedXPosition = minimumValueForLength(Length(), availableWidth); |
| spaceSize.setWidth(space); |
| spaceSize.setHeight(0); |
| phase.setWidth(actualWidth ? actualWidth - fmodf((computedXPosition + left), actualWidth) : 0); |
| } else |
| backgroundRepeatX = FillRepeat::NoRepeat; |
| } |
| |
| if (backgroundRepeatX == FillRepeat::NoRepeat) { |
| LayoutUnit xOffset = left + computedXPosition; |
| if (xOffset > 0) |
| destinationRect.move(xOffset, 0_lu); |
| xOffset = std::min<LayoutUnit>(xOffset, 0); |
| phase.setWidth(-xOffset); |
| destinationRect.setWidth(tileSize.width() + xOffset); |
| spaceSize.setWidth(0); |
| } |
| |
| if (backgroundRepeatY == FillRepeat::Repeat) { |
| phase.setHeight(tileSize.height() ? tileSize.height() - fmodf(computedYPosition + top, tileSize.height()) : 0); |
| spaceSize.setHeight(0); |
| } else if (backgroundRepeatY == FillRepeat::Space && tileSize.height() > 0) { |
| LayoutUnit space = getSpace(positioningAreaSize.height(), tileSize.height()); |
| |
| if (space >= 0) { |
| LayoutUnit actualHeight = tileSize.height() + space; |
| computedYPosition = minimumValueForLength(Length(), availableHeight); |
| spaceSize.setHeight(space); |
| phase.setHeight(actualHeight ? actualHeight - fmodf((computedYPosition + top), actualHeight) : 0); |
| } else |
| backgroundRepeatY = FillRepeat::NoRepeat; |
| } |
| if (backgroundRepeatY == FillRepeat::NoRepeat) { |
| LayoutUnit yOffset = top + computedYPosition; |
| if (yOffset > 0) |
| destinationRect.move(0_lu, yOffset); |
| yOffset = std::min<LayoutUnit>(yOffset, 0); |
| phase.setHeight(-yOffset); |
| destinationRect.setHeight(tileSize.height() + yOffset); |
| spaceSize.setHeight(0); |
| } |
| |
| if (fixedAttachment) { |
| LayoutPoint attachmentPoint = borderBoxRect.location(); |
| phase.expand(std::max<LayoutUnit>(attachmentPoint.x() - destinationRect.x(), 0), std::max<LayoutUnit>(attachmentPoint.y() - destinationRect.y(), 0)); |
| } |
| |
| destinationRect.intersect(borderBoxRect); |
| pixelSnapBackgroundImageGeometryForPainting(destinationRect, tileSize, phase, spaceSize, deviceScaleFactor); |
| return BackgroundImageGeometry(destinationRect, tileSize, phase, spaceSize, fixedAttachment); |
| } |
| |
| void RenderBoxModelObject::getGeometryForBackgroundImage(const RenderLayerModelObject* paintContainer, const LayoutPoint& paintOffset, FloatRect& destRect, FloatSize& phase, FloatSize& tileSize) const |
| { |
| LayoutRect paintRect(destRect); |
| auto geometry = calculateBackgroundImageGeometry(paintContainer, style().backgroundLayers(), paintOffset, paintRect); |
| phase = geometry.phase(); |
| tileSize = geometry.tileSize(); |
| destRect = geometry.destRect(); |
| } |
| |
| bool RenderBoxModelObject::paintNinePieceImage(GraphicsContext& graphicsContext, const LayoutRect& rect, const RenderStyle& style, |
| const NinePieceImage& ninePieceImage, CompositeOperator op) |
| { |
| StyleImage* styleImage = ninePieceImage.image(); |
| if (!styleImage) |
| return false; |
| |
| if (!styleImage->isLoaded()) |
| return true; // Never paint a nine-piece image incrementally, but don't paint the fallback borders either. |
| |
| if (!styleImage->canRender(this, style.effectiveZoom())) |
| return false; |
| |
| // FIXME: border-image is broken with full page zooming when tiling has to happen, since the tiling function |
| // doesn't have any understanding of the zoom that is in effect on the tile. |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| |
| LayoutRect rectWithOutsets = rect; |
| rectWithOutsets.expand(style.imageOutsets(ninePieceImage)); |
| LayoutRect destination = LayoutRect(snapRectToDevicePixels(rectWithOutsets, deviceScaleFactor)); |
| |
| LayoutSize source = calculateImageIntrinsicDimensions(styleImage, destination.size(), DoNotScaleByEffectiveZoom); |
| |
| // If both values are ‘auto’ then the intrinsic width and/or height of the image should be used, if any. |
| styleImage->setContainerContextForRenderer(*this, source, style.effectiveZoom()); |
| |
| ninePieceImage.paint(graphicsContext, this, style, destination, source, deviceScaleFactor, op); |
| return true; |
| } |
| |
| static bool allCornersClippedOut(const RoundedRect& border, const LayoutRect& clipRect) |
| { |
| LayoutRect boundingRect = border.rect(); |
| if (clipRect.contains(boundingRect)) |
| return false; |
| |
| RoundedRect::Radii radii = border.radii(); |
| |
| LayoutRect topLeftRect(boundingRect.location(), radii.topLeft()); |
| if (clipRect.intersects(topLeftRect)) |
| return false; |
| |
| LayoutRect topRightRect(boundingRect.location(), radii.topRight()); |
| topRightRect.setX(boundingRect.maxX() - topRightRect.width()); |
| if (clipRect.intersects(topRightRect)) |
| return false; |
| |
| LayoutRect bottomLeftRect(boundingRect.location(), radii.bottomLeft()); |
| bottomLeftRect.setY(boundingRect.maxY() - bottomLeftRect.height()); |
| if (clipRect.intersects(bottomLeftRect)) |
| return false; |
| |
| LayoutRect bottomRightRect(boundingRect.location(), radii.bottomRight()); |
| bottomRightRect.setX(boundingRect.maxX() - bottomRightRect.width()); |
| bottomRightRect.setY(boundingRect.maxY() - bottomRightRect.height()); |
| if (clipRect.intersects(bottomRightRect)) |
| return false; |
| |
| return true; |
| } |
| |
| static bool borderWillArcInnerEdge(const LayoutSize& firstRadius, const LayoutSize& secondRadius) |
| { |
| return !firstRadius.isZero() || !secondRadius.isZero(); |
| } |
| |
| inline bool styleRequiresClipPolygon(BorderStyle style) |
| { |
| return style == BorderStyle::Dotted || style == BorderStyle::Dashed; // These are drawn with a stroke, so we have to clip to get corner miters. |
| } |
| |
| static bool borderStyleFillsBorderArea(BorderStyle style) |
| { |
| return !(style == BorderStyle::Dotted || style == BorderStyle::Dashed || style == BorderStyle::Double); |
| } |
| |
| static bool borderStyleHasInnerDetail(BorderStyle style) |
| { |
| return style == BorderStyle::Groove || style == BorderStyle::Ridge || style == BorderStyle::Double; |
| } |
| |
| static bool borderStyleIsDottedOrDashed(BorderStyle style) |
| { |
| return style == BorderStyle::Dotted || style == BorderStyle::Dashed; |
| } |
| |
| // BorderStyle::Outset darkens the bottom and right (and maybe lightens the top and left) |
| // BorderStyle::Inset darkens the top and left (and maybe lightens the bottom and right) |
| static inline bool borderStyleHasUnmatchedColorsAtCorner(BorderStyle style, BoxSide side, BoxSide adjacentSide) |
| { |
| // These styles match at the top/left and bottom/right. |
| if (style == BorderStyle::Inset || style == BorderStyle::Groove || style == BorderStyle::Ridge || style == BorderStyle::Outset) { |
| BoxSideSet topRightSides = { BoxSideFlag::Top, BoxSideFlag::Right }; |
| BoxSideSet bottomLeftSides = { BoxSideFlag::Bottom, BoxSideFlag::Left }; |
| |
| BoxSideSet usedSides { edgeFlagForSide(side), edgeFlagForSide(adjacentSide) }; |
| return usedSides == topRightSides || usedSides == bottomLeftSides; |
| } |
| return false; |
| } |
| |
| static inline bool colorsMatchAtCorner(BoxSide side, BoxSide adjacentSide, const BorderEdges& edges) |
| { |
| auto& edge = edges.at(side); |
| auto& adjacentEdge = edges.at(adjacentSide); |
| |
| if (edge.shouldRender() != adjacentEdge.shouldRender()) |
| return false; |
| |
| if (!edgesShareColor(edge, adjacentEdge)) |
| return false; |
| |
| return !borderStyleHasUnmatchedColorsAtCorner(edge.style(), side, adjacentSide); |
| } |
| |
| |
| static inline bool colorNeedsAntiAliasAtCorner(BoxSide side, BoxSide adjacentSide, const BorderEdges& edges) |
| { |
| auto& edge = edges.at(side); |
| auto& adjacentEdge = edges.at(adjacentSide); |
| |
| if (edge.color().isOpaque()) |
| return false; |
| |
| if (edge.shouldRender() != adjacentEdge.shouldRender()) |
| return false; |
| |
| if (!edgesShareColor(edge, adjacentEdge)) |
| return true; |
| |
| return borderStyleHasUnmatchedColorsAtCorner(edge.style(), side, adjacentSide); |
| } |
| |
| // This assumes that we draw in order: top, bottom, left, right. |
| static inline bool willBeOverdrawn(BoxSide side, BoxSide adjacentSide, const BorderEdges& edges) |
| { |
| switch (side) { |
| case BoxSide::Top: |
| case BoxSide::Bottom: { |
| auto& edge = edges.at(side); |
| auto& adjacentEdge = edges.at(adjacentSide); |
| |
| if (adjacentEdge.presentButInvisible()) |
| return false; |
| |
| if (!edgesShareColor(edge, adjacentEdge) && !adjacentEdge.color().isOpaque()) |
| return false; |
| |
| if (!borderStyleFillsBorderArea(adjacentEdge.style())) |
| return false; |
| |
| return true; |
| } |
| case BoxSide::Left: |
| case BoxSide::Right: |
| // These draw last, so are never overdrawn. |
| return false; |
| } |
| return false; |
| } |
| |
| static inline bool borderStylesRequireMitre(BoxSide side, BoxSide adjacentSide, BorderStyle style, BorderStyle adjacentStyle) |
| { |
| if (style == BorderStyle::Double || adjacentStyle == BorderStyle::Double || adjacentStyle == BorderStyle::Groove || adjacentStyle == BorderStyle::Ridge) |
| return true; |
| |
| if (borderStyleIsDottedOrDashed(style) != borderStyleIsDottedOrDashed(adjacentStyle)) |
| return true; |
| |
| if (style != adjacentStyle) |
| return true; |
| |
| return borderStyleHasUnmatchedColorsAtCorner(style, side, adjacentSide); |
| } |
| |
| static bool joinRequiresMitre(BoxSide side, BoxSide adjacentSide, const BorderEdges& edges, bool allowOverdraw) |
| { |
| auto& edge = edges.at(side); |
| auto& adjacentEdge = edges.at(adjacentSide); |
| |
| if ((edge.isTransparent() && adjacentEdge.isTransparent()) || !adjacentEdge.isPresent()) |
| return false; |
| |
| if (allowOverdraw && willBeOverdrawn(side, adjacentSide, edges)) |
| return false; |
| |
| if (!edgesShareColor(edge, adjacentEdge)) |
| return true; |
| |
| if (borderStylesRequireMitre(side, adjacentSide, edge.style(), adjacentEdge.style())) |
| return true; |
| |
| return false; |
| } |
| |
| static RoundedRect calculateAdjustedInnerBorder(const RoundedRect&innerBorder, BoxSide side) |
| { |
| // Expand the inner border as necessary to make it a rounded rect (i.e. radii contained within each edge). |
| // This function relies on the fact we only get radii not contained within each edge if one of the radii |
| // for an edge is zero, so we can shift the arc towards the zero radius corner. |
| RoundedRect::Radii newRadii = innerBorder.radii(); |
| LayoutRect newRect = innerBorder.rect(); |
| |
| float overshoot; |
| float maxRadii; |
| |
| switch (side) { |
| case BoxSide::Top: |
| overshoot = newRadii.topLeft().width() + newRadii.topRight().width() - newRect.width(); |
| if (overshoot > 0) { |
| ASSERT(!(newRadii.topLeft().width() && newRadii.topRight().width())); |
| newRect.setWidth(newRect.width() + overshoot); |
| if (!newRadii.topLeft().width()) |
| newRect.move(-overshoot, 0); |
| } |
| newRadii.setBottomLeft({ }); |
| newRadii.setBottomRight({ }); |
| maxRadii = std::max(newRadii.topLeft().height(), newRadii.topRight().height()); |
| if (maxRadii > newRect.height()) |
| newRect.setHeight(maxRadii); |
| break; |
| |
| case BoxSide::Bottom: |
| overshoot = newRadii.bottomLeft().width() + newRadii.bottomRight().width() - newRect.width(); |
| if (overshoot > 0) { |
| ASSERT(!(newRadii.bottomLeft().width() && newRadii.bottomRight().width())); |
| newRect.setWidth(newRect.width() + overshoot); |
| if (!newRadii.bottomLeft().width()) |
| newRect.move(-overshoot, 0); |
| } |
| newRadii.setTopLeft({ }); |
| newRadii.setTopRight({ }); |
| maxRadii = std::max(newRadii.bottomLeft().height(), newRadii.bottomRight().height()); |
| if (maxRadii > newRect.height()) { |
| newRect.move(0, newRect.height() - maxRadii); |
| newRect.setHeight(maxRadii); |
| } |
| break; |
| |
| case BoxSide::Left: |
| overshoot = newRadii.topLeft().height() + newRadii.bottomLeft().height() - newRect.height(); |
| if (overshoot > 0) { |
| ASSERT(!(newRadii.topLeft().height() && newRadii.bottomLeft().height())); |
| newRect.setHeight(newRect.height() + overshoot); |
| if (!newRadii.topLeft().height()) |
| newRect.move(0, -overshoot); |
| } |
| newRadii.setTopRight({ }); |
| newRadii.setBottomRight({ }); |
| maxRadii = std::max(newRadii.topLeft().width(), newRadii.bottomLeft().width()); |
| if (maxRadii > newRect.width()) |
| newRect.setWidth(maxRadii); |
| break; |
| |
| case BoxSide::Right: |
| overshoot = newRadii.topRight().height() + newRadii.bottomRight().height() - newRect.height(); |
| if (overshoot > 0) { |
| ASSERT(!(newRadii.topRight().height() && newRadii.bottomRight().height())); |
| newRect.setHeight(newRect.height() + overshoot); |
| if (!newRadii.topRight().height()) |
| newRect.move(0, -overshoot); |
| } |
| newRadii.setTopLeft({ }); |
| newRadii.setBottomLeft({ }); |
| maxRadii = std::max(newRadii.topRight().width(), newRadii.bottomRight().width()); |
| if (maxRadii > newRect.width()) { |
| newRect.move(newRect.width() - maxRadii, 0); |
| newRect.setWidth(maxRadii); |
| } |
| break; |
| } |
| |
| return RoundedRect(newRect, newRadii); |
| } |
| |
| void RenderBoxModelObject::paintOneBorderSide(GraphicsContext& graphicsContext, const RenderStyle& style, const RoundedRect& outerBorder, const RoundedRect& innerBorder, |
| const LayoutRect& sideRect, BoxSide side, BoxSide adjacentSide1, BoxSide adjacentSide2, const BorderEdges& edges, const Path* path, |
| BackgroundBleedAvoidance bleedAvoidance, bool includeLogicalLeftEdge, bool includeLogicalRightEdge, bool antialias, const Color* overrideColor) |
| { |
| auto& edgeToRender = edges.at(side); |
| ASSERT(edgeToRender.widthForPainting()); |
| auto& adjacentEdge1 = edges.at(adjacentSide1); |
| auto& adjacentEdge2 = edges.at(adjacentSide2); |
| |
| bool mitreAdjacentSide1 = joinRequiresMitre(side, adjacentSide1, edges, !antialias); |
| bool mitreAdjacentSide2 = joinRequiresMitre(side, adjacentSide2, edges, !antialias); |
| |
| bool adjacentSide1StylesMatch = colorsMatchAtCorner(side, adjacentSide1, edges); |
| bool adjacentSide2StylesMatch = colorsMatchAtCorner(side, adjacentSide2, edges); |
| |
| const Color& colorToPaint = overrideColor ? *overrideColor : edgeToRender.color(); |
| |
| if (path) { |
| GraphicsContextStateSaver stateSaver(graphicsContext); |
| |
| clipBorderSidePolygon(graphicsContext, outerBorder, innerBorder, side, adjacentSide1StylesMatch, adjacentSide2StylesMatch); |
| |
| if (!innerBorder.isRenderable()) |
| graphicsContext.clipOutRoundedRect(FloatRoundedRect(calculateAdjustedInnerBorder(innerBorder, side))); |
| |
| float thickness = std::max(std::max(edgeToRender.widthForPainting(), adjacentEdge1.widthForPainting()), adjacentEdge2.widthForPainting()); |
| drawBoxSideFromPath(graphicsContext, outerBorder.rect(), *path, edges, edgeToRender.widthForPainting(), thickness, side, style, |
| colorToPaint, edgeToRender.style(), bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge); |
| } else { |
| bool clipForStyle = styleRequiresClipPolygon(edgeToRender.style()) && (mitreAdjacentSide1 || mitreAdjacentSide2); |
| bool clipAdjacentSide1 = colorNeedsAntiAliasAtCorner(side, adjacentSide1, edges) && mitreAdjacentSide1; |
| bool clipAdjacentSide2 = colorNeedsAntiAliasAtCorner(side, adjacentSide2, edges) && mitreAdjacentSide2; |
| bool shouldClip = clipForStyle || clipAdjacentSide1 || clipAdjacentSide2; |
| |
| GraphicsContextStateSaver clipStateSaver(graphicsContext, shouldClip); |
| if (shouldClip) { |
| bool aliasAdjacentSide1 = clipAdjacentSide1 || (clipForStyle && mitreAdjacentSide1); |
| bool aliasAdjacentSide2 = clipAdjacentSide2 || (clipForStyle && mitreAdjacentSide2); |
| clipBorderSidePolygon(graphicsContext, outerBorder, innerBorder, side, !aliasAdjacentSide1, !aliasAdjacentSide2); |
| // Since we clipped, no need to draw with a mitre. |
| mitreAdjacentSide1 = false; |
| mitreAdjacentSide2 = false; |
| } |
| drawLineForBoxSide(graphicsContext, sideRect, side, colorToPaint, edgeToRender.style(), mitreAdjacentSide1 ? adjacentEdge1.widthForPainting() : 0, mitreAdjacentSide2 ? adjacentEdge2.widthForPainting() : 0, antialias); |
| } |
| } |
| |
| static LayoutRect calculateSideRect(const RoundedRect& outerBorder, const BorderEdges& edges, BoxSide side) |
| { |
| LayoutRect sideRect = outerBorder.rect(); |
| float width = edges.at(side).widthForPainting(); |
| |
| switch (side) { |
| case BoxSide::Top: |
| sideRect.setHeight(width); |
| break; |
| case BoxSide::Right: |
| sideRect.shiftXEdgeTo(sideRect.maxX() - width); |
| break; |
| case BoxSide::Bottom: |
| sideRect.shiftYEdgeTo(sideRect.maxY() - width); |
| break; |
| case BoxSide::Left: |
| sideRect.setWidth(width); |
| break; |
| } |
| |
| return sideRect; |
| } |
| |
| void RenderBoxModelObject::paintBorderSides(GraphicsContext& graphicsContext, const RenderStyle& style, const RoundedRect& outerBorder, const RoundedRect& innerBorder, |
| const IntPoint& innerBorderAdjustment, const BorderEdges& edges, BoxSideSet edgeSet, BackgroundBleedAvoidance bleedAvoidance, |
| bool includeLogicalLeftEdge, bool includeLogicalRightEdge, bool antialias, const Color* overrideColor) |
| { |
| bool renderRadii = outerBorder.isRounded(); |
| |
| Path roundedPath; |
| if (renderRadii) |
| roundedPath.addRoundedRect(outerBorder); |
| |
| // The inner border adjustment for bleed avoidance mode BackgroundBleedBackgroundOverBorder |
| // is only applied to sideRect, which is okay since BackgroundBleedBackgroundOverBorder |
| // is only to be used for solid borders and the shape of the border painted by drawBoxSideFromPath |
| // only depends on sideRect when painting solid borders. |
| |
| auto paintOneSide = [&](BoxSide side, BoxSide adjacentSide1, BoxSide adjacentSide2) { |
| auto& edge = edges.at(side); |
| if (!edge.shouldRender() || !edgeSet.contains(edgeFlagForSide(side))) |
| return; |
| |
| LayoutRect sideRect = outerBorder.rect(); |
| LayoutSize firstRadius; |
| LayoutSize secondRadius; |
| |
| switch (side) { |
| case BoxSide::Top: |
| sideRect.setHeight(edge.widthForPainting() + innerBorderAdjustment.y()); |
| firstRadius = innerBorder.radii().topLeft(); |
| secondRadius = innerBorder.radii().topRight(); |
| break; |
| case BoxSide::Right: |
| sideRect.shiftXEdgeTo(sideRect.maxX() - edge.widthForPainting() - innerBorderAdjustment.x()); |
| firstRadius = innerBorder.radii().bottomRight(); |
| secondRadius = innerBorder.radii().topRight(); |
| break; |
| case BoxSide::Bottom: |
| sideRect.shiftYEdgeTo(sideRect.maxY() - edge.widthForPainting() - innerBorderAdjustment.y()); |
| firstRadius = innerBorder.radii().bottomLeft(); |
| secondRadius = innerBorder.radii().bottomRight(); |
| break; |
| case BoxSide::Left: |
| sideRect.setWidth(edge.widthForPainting() + innerBorderAdjustment.x()); |
| firstRadius = innerBorder.radii().bottomLeft(); |
| secondRadius = innerBorder.radii().topLeft(); |
| break; |
| } |
| |
| bool usePath = renderRadii && (borderStyleHasInnerDetail(edge.style()) || borderWillArcInnerEdge(firstRadius, secondRadius)); |
| paintOneBorderSide(graphicsContext, style, outerBorder, innerBorder, sideRect, side, adjacentSide1, adjacentSide2, edges, usePath ? &roundedPath : nullptr, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge, antialias, overrideColor); |
| }; |
| |
| paintOneSide(BoxSide::Top, BoxSide::Left, BoxSide::Right); |
| paintOneSide(BoxSide::Bottom, BoxSide::Left, BoxSide::Right); |
| paintOneSide(BoxSide::Left, BoxSide::Top, BoxSide::Bottom); |
| paintOneSide(BoxSide::Right, BoxSide::Top, BoxSide::Bottom); |
| } |
| |
| void RenderBoxModelObject::paintTranslucentBorderSides(GraphicsContext& graphicsContext, const RenderStyle& style, const RoundedRect& outerBorder, const RoundedRect& innerBorder, const IntPoint& innerBorderAdjustment, |
| const BorderEdges& edges, BoxSideSet edgesToDraw, BackgroundBleedAvoidance bleedAvoidance, bool includeLogicalLeftEdge, bool includeLogicalRightEdge, bool antialias) |
| { |
| // willBeOverdrawn assumes that we draw in order: top, bottom, left, right. |
| // This is different from BoxSide enum order. |
| static constexpr std::array<BoxSide, 4> paintOrderSides = { BoxSide::Top, BoxSide::Bottom, BoxSide::Left, BoxSide::Right }; |
| |
| while (edgesToDraw) { |
| // Find undrawn edges sharing a color. |
| Color commonColor; |
| |
| BoxSideSet commonColorEdgeSet; |
| for (auto side : paintOrderSides) { |
| if (!edgesToDraw.contains(edgeFlagForSide(side))) |
| continue; |
| |
| auto& edge = edges.at(side); |
| bool includeEdge; |
| if (commonColorEdgeSet.isEmpty()) { |
| commonColor = edge.color(); |
| includeEdge = true; |
| } else |
| includeEdge = edge.color() == commonColor; |
| |
| if (includeEdge) |
| commonColorEdgeSet.add(edgeFlagForSide(side)); |
| } |
| |
| bool useTransparencyLayer = includesAdjacentEdges(commonColorEdgeSet) && !commonColor.isOpaque(); |
| if (useTransparencyLayer) { |
| graphicsContext.beginTransparencyLayer(commonColor.alphaAsFloat()); |
| commonColor = commonColor.opaqueColor(); |
| } |
| |
| paintBorderSides(graphicsContext, style, outerBorder, innerBorder, innerBorderAdjustment, edges, commonColorEdgeSet, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge, antialias, &commonColor); |
| |
| if (useTransparencyLayer) |
| graphicsContext.endTransparencyLayer(); |
| |
| edgesToDraw.remove(commonColorEdgeSet); |
| } |
| } |
| |
| void RenderBoxModelObject::paintBorder(const PaintInfo& info, const LayoutRect& rect, const RenderStyle& style, |
| BackgroundBleedAvoidance bleedAvoidance, bool includeLogicalLeftEdge, bool includeLogicalRightEdge) |
| { |
| GraphicsContext& graphicsContext = info.context(); |
| |
| if (graphicsContext.paintingDisabled()) |
| return; |
| |
| auto paintsBorderImage = [&](LayoutRect rect, const NinePieceImage& ninePieceImage) { |
| auto* styleImage = ninePieceImage.image(); |
| if (!styleImage) |
| return false; |
| |
| if (!styleImage->isLoaded()) |
| return false; |
| |
| if (!styleImage->canRender(this, style.effectiveZoom())) |
| return false; |
| |
| auto rectWithOutsets = rect; |
| rectWithOutsets.expand(style.imageOutsets(ninePieceImage)); |
| return !rectWithOutsets.isEmpty(); |
| }; |
| |
| if (rect.isEmpty() && !paintsBorderImage(rect, style.borderImage())) |
| return; |
| |
| auto rectToClipOut = paintRectToClipOutFromBorder(rect); |
| bool appliedClipAlready = !rectToClipOut.isEmpty(); |
| GraphicsContextStateSaver stateSave(graphicsContext, appliedClipAlready); |
| if (!rectToClipOut.isEmpty()) |
| graphicsContext.clipOut(snapRectToDevicePixels(rectToClipOut, document().deviceScaleFactor())); |
| |
| // border-image is not affected by border-radius. |
| if (paintNinePieceImage(graphicsContext, rect, style, style.borderImage())) |
| return; |
| |
| auto edges = borderEdges(style, document().deviceScaleFactor(), includeLogicalLeftEdge, includeLogicalRightEdge); |
| RoundedRect outerBorder = style.getRoundedBorderFor(rect, includeLogicalLeftEdge, includeLogicalRightEdge); |
| RoundedRect innerBorder = style.getRoundedInnerBorderFor(borderInnerRectAdjustedForBleedAvoidance(graphicsContext, rect, bleedAvoidance), includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| // If no borders intersects with the dirty area, we can skip the border painting. |
| if (innerBorder.contains(info.rect)) |
| return; |
| |
| bool haveAlphaColor = false; |
| bool haveAllSolidEdges = true; |
| bool haveAllDoubleEdges = true; |
| int numEdgesVisible = 4; |
| bool allEdgesShareColor = true; |
| std::optional<BoxSide> firstVisibleSide; |
| BoxSideSet edgesToDraw; |
| |
| for (auto side : allBoxSides) { |
| auto& currEdge = edges.at(side); |
| |
| if (currEdge.shouldRender()) |
| edgesToDraw.add(edgeFlagForSide(side)); |
| |
| if (currEdge.presentButInvisible()) { |
| --numEdgesVisible; |
| allEdgesShareColor = false; |
| continue; |
| } |
| |
| if (!currEdge.widthForPainting()) { |
| --numEdgesVisible; |
| continue; |
| } |
| |
| if (!firstVisibleSide) |
| firstVisibleSide = side; |
| else if (currEdge.color() != edges.at(*firstVisibleSide).color()) |
| allEdgesShareColor = false; |
| |
| if (!currEdge.color().isOpaque()) |
| haveAlphaColor = true; |
| |
| if (currEdge.style() != BorderStyle::Solid) |
| haveAllSolidEdges = false; |
| |
| if (currEdge.style() != BorderStyle::Double) |
| haveAllDoubleEdges = false; |
| } |
| |
| // If no corner intersects the clip region, we can pretend outerBorder is |
| // rectangular to improve performance. |
| if (haveAllSolidEdges && outerBorder.isRounded() && allCornersClippedOut(outerBorder, info.rect)) |
| outerBorder.setRadii(RoundedRect::Radii()); |
| |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| // isRenderable() check avoids issue described in https://bugs.webkit.org/show_bug.cgi?id=38787 |
| if ((haveAllSolidEdges || haveAllDoubleEdges) && allEdgesShareColor && innerBorder.isRenderable()) { |
| // Fast path for drawing all solid edges and all unrounded double edges |
| if (numEdgesVisible == 4 && (outerBorder.isRounded() || haveAlphaColor) |
| && (haveAllSolidEdges || (!outerBorder.isRounded() && !innerBorder.isRounded()))) { |
| Path path; |
| |
| FloatRoundedRect pixelSnappedOuterBorder = outerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| if (pixelSnappedOuterBorder.isRounded() && bleedAvoidance != BackgroundBleedUseTransparencyLayer) |
| path.addRoundedRect(pixelSnappedOuterBorder); |
| else |
| path.addRect(pixelSnappedOuterBorder.rect()); |
| |
| if (haveAllDoubleEdges) { |
| LayoutRect innerThirdRect = outerBorder.rect(); |
| LayoutRect outerThirdRect = outerBorder.rect(); |
| for (auto side : allBoxSides) { |
| LayoutUnit outerWidth; |
| LayoutUnit innerWidth; |
| edges.at(side).getDoubleBorderStripeWidths(outerWidth, innerWidth); |
| switch (side) { |
| case BoxSide::Top: |
| innerThirdRect.shiftYEdgeTo(innerThirdRect.y() + innerWidth); |
| outerThirdRect.shiftYEdgeTo(outerThirdRect.y() + outerWidth); |
| break; |
| case BoxSide::Right: |
| innerThirdRect.setWidth(innerThirdRect.width() - innerWidth); |
| outerThirdRect.setWidth(outerThirdRect.width() - outerWidth); |
| break; |
| case BoxSide::Bottom: |
| innerThirdRect.setHeight(innerThirdRect.height() - innerWidth); |
| outerThirdRect.setHeight(outerThirdRect.height() - outerWidth); |
| break; |
| case BoxSide::Left: |
| innerThirdRect.shiftXEdgeTo(innerThirdRect.x() + innerWidth); |
| outerThirdRect.shiftXEdgeTo(outerThirdRect.x() + outerWidth); |
| break; |
| } |
| } |
| |
| FloatRoundedRect pixelSnappedOuterThird = outerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| pixelSnappedOuterThird.setRect(snapRectToDevicePixels(outerThirdRect, deviceScaleFactor)); |
| |
| if (pixelSnappedOuterThird.isRounded() && bleedAvoidance != BackgroundBleedUseTransparencyLayer) |
| path.addRoundedRect(pixelSnappedOuterThird); |
| else |
| path.addRect(pixelSnappedOuterThird.rect()); |
| |
| FloatRoundedRect pixelSnappedInnerThird = innerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| pixelSnappedInnerThird.setRect(snapRectToDevicePixels(innerThirdRect, deviceScaleFactor)); |
| if (pixelSnappedInnerThird.isRounded() && bleedAvoidance != BackgroundBleedUseTransparencyLayer) |
| path.addRoundedRect(pixelSnappedInnerThird); |
| else |
| path.addRect(pixelSnappedInnerThird.rect()); |
| } |
| |
| FloatRoundedRect pixelSnappedInnerBorder = innerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| if (pixelSnappedInnerBorder.isRounded()) |
| path.addRoundedRect(pixelSnappedInnerBorder); |
| else |
| path.addRect(pixelSnappedInnerBorder.rect()); |
| |
| graphicsContext.setFillRule(WindRule::EvenOdd); |
| graphicsContext.setFillColor(edges.at(*firstVisibleSide).color()); |
| graphicsContext.fillPath(path); |
| return; |
| } |
| // Avoid creating transparent layers |
| if (haveAllSolidEdges && numEdgesVisible != 4 && !outerBorder.isRounded() && haveAlphaColor) { |
| Path path; |
| |
| for (auto side : allBoxSides) { |
| if (edges.at(side).shouldRender()) { |
| auto sideRect = calculateSideRect(outerBorder, edges, side); |
| path.addRect(sideRect); // FIXME: Need pixel snapping here. |
| } |
| } |
| |
| graphicsContext.setFillRule(WindRule::NonZero); |
| graphicsContext.setFillColor(edges.at(*firstVisibleSide).color()); |
| graphicsContext.fillPath(path); |
| return; |
| } |
| } |
| |
| bool clipToOuterBorder = outerBorder.isRounded(); |
| GraphicsContextStateSaver stateSaver(graphicsContext, clipToOuterBorder && !appliedClipAlready); |
| if (clipToOuterBorder) { |
| // Clip to the inner and outer radii rects. |
| if (bleedAvoidance != BackgroundBleedUseTransparencyLayer) |
| graphicsContext.clipRoundedRect(outerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor)); |
| // isRenderable() check avoids issue described in https://bugs.webkit.org/show_bug.cgi?id=38787 |
| // The inside will be clipped out later (in clipBorderSideForComplexInnerPath) |
| if (innerBorder.isRenderable()) |
| graphicsContext.clipOutRoundedRect(innerBorder.pixelSnappedRoundedRectForPainting(deviceScaleFactor)); |
| } |
| |
| // If only one edge visible antialiasing doesn't create seams |
| bool antialias = shouldAntialiasLines(graphicsContext) || numEdgesVisible == 1; |
| RoundedRect unadjustedInnerBorder = (bleedAvoidance == BackgroundBleedBackgroundOverBorder) ? style.getRoundedInnerBorderFor(rect, includeLogicalLeftEdge, includeLogicalRightEdge) : innerBorder; |
| IntPoint innerBorderAdjustment(innerBorder.rect().x() - unadjustedInnerBorder.rect().x(), innerBorder.rect().y() - unadjustedInnerBorder.rect().y()); |
| if (haveAlphaColor) |
| paintTranslucentBorderSides(graphicsContext, style, outerBorder, unadjustedInnerBorder, innerBorderAdjustment, edges, edgesToDraw, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge, antialias); |
| else |
| paintBorderSides(graphicsContext, style, outerBorder, unadjustedInnerBorder, innerBorderAdjustment, edges, edgesToDraw, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge, antialias); |
| } |
| |
| void RenderBoxModelObject::drawBoxSideFromPath(GraphicsContext& graphicsContext, const LayoutRect& borderRect, const Path& borderPath, const BorderEdges& edges, |
| float thickness, float drawThickness, BoxSide side, const RenderStyle& style, Color color, BorderStyle borderStyle, BackgroundBleedAvoidance bleedAvoidance, |
| bool includeLogicalLeftEdge, bool includeLogicalRightEdge) |
| { |
| if (thickness <= 0) |
| return; |
| |
| if (borderStyle == BorderStyle::Double && thickness < 3) |
| borderStyle = BorderStyle::Solid; |
| |
| switch (borderStyle) { |
| case BorderStyle::None: |
| case BorderStyle::Hidden: |
| return; |
| case BorderStyle::Dotted: |
| case BorderStyle::Dashed: { |
| graphicsContext.setStrokeColor(color); |
| |
| // The stroke is doubled here because the provided path is the |
| // outside edge of the border so half the stroke is clipped off. |
| // The extra multiplier is so that the clipping mask can antialias |
| // the edges to prevent jaggies. |
| graphicsContext.setStrokeThickness(drawThickness * 2 * 1.1f); |
| graphicsContext.setStrokeStyle(borderStyle == BorderStyle::Dashed ? DashedStroke : DottedStroke); |
| |
| // If the number of dashes that fit in the path is odd and non-integral then we |
| // will have an awkwardly-sized dash at the end of the path. To try to avoid that |
| // here, we simply make the whitespace dashes ever so slightly bigger. |
| // FIXME: This could be even better if we tried to manipulate the dash offset |
| // and possibly the gapLength to get the corners dash-symmetrical. |
| float dashLength = thickness * ((borderStyle == BorderStyle::Dashed) ? 3.0f : 1.0f); |
| float gapLength = dashLength; |
| float numberOfDashes = borderPath.length() / dashLength; |
| // Don't try to show dashes if we have less than 2 dashes + 2 gaps. |
| // FIXME: should do this test per side. |
| if (numberOfDashes >= 4) { |
| bool evenNumberOfFullDashes = !((int)numberOfDashes % 2); |
| bool integralNumberOfDashes = !(numberOfDashes - (int)numberOfDashes); |
| if (!evenNumberOfFullDashes && !integralNumberOfDashes) { |
| float numberOfGaps = numberOfDashes / 2; |
| gapLength += (dashLength / numberOfGaps); |
| } |
| |
| auto lineDash = DashArray::from(dashLength, gapLength); |
| graphicsContext.setLineDash(WTFMove(lineDash), dashLength); |
| } |
| |
| // FIXME: stroking the border path causes issues with tight corners: |
| // https://bugs.webkit.org/show_bug.cgi?id=58711 |
| // Also, to get the best appearance we should stroke a path between the two borders. |
| graphicsContext.strokePath(borderPath); |
| return; |
| } |
| case BorderStyle::Double: { |
| // Get the inner border rects for both the outer border line and the inner border line |
| LayoutUnit outerBorderTopWidth; |
| LayoutUnit innerBorderTopWidth; |
| edges.top().getDoubleBorderStripeWidths(outerBorderTopWidth, innerBorderTopWidth); |
| |
| LayoutUnit outerBorderRightWidth; |
| LayoutUnit innerBorderRightWidth; |
| edges.right().getDoubleBorderStripeWidths(outerBorderRightWidth, innerBorderRightWidth); |
| |
| LayoutUnit outerBorderBottomWidth; |
| LayoutUnit innerBorderBottomWidth; |
| edges.bottom().getDoubleBorderStripeWidths(outerBorderBottomWidth, innerBorderBottomWidth); |
| |
| LayoutUnit outerBorderLeftWidth; |
| LayoutUnit innerBorderLeftWidth; |
| edges.left().getDoubleBorderStripeWidths(outerBorderLeftWidth, innerBorderLeftWidth); |
| |
| // Draw inner border line |
| { |
| GraphicsContextStateSaver stateSaver(graphicsContext); |
| RoundedRect innerClip = style.getRoundedInnerBorderFor(borderRect, |
| innerBorderTopWidth, innerBorderBottomWidth, innerBorderLeftWidth, innerBorderRightWidth, |
| includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| graphicsContext.clipRoundedRect(FloatRoundedRect(innerClip)); |
| drawBoxSideFromPath(graphicsContext, borderRect, borderPath, edges, thickness, drawThickness, side, style, color, BorderStyle::Solid, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge); |
| } |
| |
| // Draw outer border line |
| { |
| GraphicsContextStateSaver stateSaver(graphicsContext); |
| LayoutRect outerRect = borderRect; |
| if (bleedAvoidance == BackgroundBleedUseTransparencyLayer) { |
| outerRect.inflate(1); |
| ++outerBorderTopWidth; |
| ++outerBorderBottomWidth; |
| ++outerBorderLeftWidth; |
| ++outerBorderRightWidth; |
| } |
| |
| RoundedRect outerClip = style.getRoundedInnerBorderFor(outerRect, |
| outerBorderTopWidth, outerBorderBottomWidth, outerBorderLeftWidth, outerBorderRightWidth, |
| includeLogicalLeftEdge, includeLogicalRightEdge); |
| graphicsContext.clipOutRoundedRect(FloatRoundedRect(outerClip)); |
| drawBoxSideFromPath(graphicsContext, borderRect, borderPath, edges, thickness, drawThickness, side, style, color, BorderStyle::Solid, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge); |
| } |
| return; |
| } |
| case BorderStyle::Ridge: |
| case BorderStyle::Groove: |
| { |
| BorderStyle s1; |
| BorderStyle s2; |
| if (borderStyle == BorderStyle::Groove) { |
| s1 = BorderStyle::Inset; |
| s2 = BorderStyle::Outset; |
| } else { |
| s1 = BorderStyle::Outset; |
| s2 = BorderStyle::Inset; |
| } |
| |
| // Paint full border |
| drawBoxSideFromPath(graphicsContext, borderRect, borderPath, edges, thickness, drawThickness, side, style, color, s1, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| // Paint inner only |
| GraphicsContextStateSaver stateSaver(graphicsContext); |
| LayoutUnit topWidth { edges.top().widthForPainting() / 2 }; |
| LayoutUnit bottomWidth { edges.bottom().widthForPainting() / 2 }; |
| LayoutUnit leftWidth { edges.left().widthForPainting() / 2 }; |
| LayoutUnit rightWidth { edges.right().widthForPainting() / 2 }; |
| |
| RoundedRect clipRect = style.getRoundedInnerBorderFor(borderRect, |
| topWidth, bottomWidth, leftWidth, rightWidth, |
| includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| graphicsContext.clipRoundedRect(FloatRoundedRect(clipRect)); |
| drawBoxSideFromPath(graphicsContext, borderRect, borderPath, edges, thickness, drawThickness, side, style, color, s2, bleedAvoidance, includeLogicalLeftEdge, includeLogicalRightEdge); |
| return; |
| } |
| case BorderStyle::Inset: |
| case BorderStyle::Outset: |
| calculateBorderStyleColor(borderStyle, side, color); |
| break; |
| default: |
| break; |
| } |
| |
| graphicsContext.setStrokeStyle(NoStroke); |
| graphicsContext.setFillColor(color); |
| graphicsContext.drawRect(snapRectToDevicePixels(borderRect, document().deviceScaleFactor())); |
| } |
| |
| void RenderBoxModelObject::clipBorderSidePolygon(GraphicsContext& graphicsContext, const RoundedRect& outerBorder, const RoundedRect& innerBorder, |
| BoxSide side, bool firstEdgeMatches, bool secondEdgeMatches) |
| { |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| const FloatRect& outerRect = snapRectToDevicePixels(outerBorder.rect(), deviceScaleFactor); |
| const FloatRect& innerRect = snapRectToDevicePixels(innerBorder.rect(), deviceScaleFactor); |
| |
| // For each side, create a quad that encompasses all parts of that side that may draw, |
| // including areas inside the innerBorder. |
| // |
| // 0----------------3 |
| // 0 \ / 0 |
| // |\ 1----------- 2 /| |
| // | 1 1 | |
| // | | | | |
| // | | | | |
| // | 2 2 | |
| // |/ 1------------2 \| |
| // 3 / \ 3 |
| // 0----------------3 |
| // |
| Vector<FloatPoint> quad; |
| switch (side) { |
| case BoxSide::Top: |
| quad = { outerRect.minXMinYCorner(), innerRect.minXMinYCorner(), innerRect.maxXMinYCorner(), outerRect.maxXMinYCorner() }; |
| |
| if (!innerBorder.radii().topLeft().isZero()) |
| findIntersection(outerRect.minXMinYCorner(), innerRect.minXMinYCorner(), innerRect.minXMaxYCorner(), innerRect.maxXMinYCorner(), quad[1]); |
| |
| if (!innerBorder.radii().topRight().isZero()) |
| findIntersection(outerRect.maxXMinYCorner(), innerRect.maxXMinYCorner(), innerRect.minXMinYCorner(), innerRect.maxXMaxYCorner(), quad[2]); |
| break; |
| |
| case BoxSide::Left: |
| quad = { outerRect.minXMinYCorner(), innerRect.minXMinYCorner(), innerRect.minXMaxYCorner(), outerRect.minXMaxYCorner() }; |
| |
| if (!innerBorder.radii().topLeft().isZero()) |
| findIntersection(outerRect.minXMinYCorner(), innerRect.minXMinYCorner(), innerRect.minXMaxYCorner(), innerRect.maxXMinYCorner(), quad[1]); |
| |
| if (!innerBorder.radii().bottomLeft().isZero()) |
| findIntersection(outerRect.minXMaxYCorner(), innerRect.minXMaxYCorner(), innerRect.minXMinYCorner(), innerRect.maxXMaxYCorner(), quad[2]); |
| break; |
| |
| case BoxSide::Bottom: |
| quad = { outerRect.minXMaxYCorner(), innerRect.minXMaxYCorner(), innerRect.maxXMaxYCorner(), outerRect.maxXMaxYCorner() }; |
| |
| if (!innerBorder.radii().bottomLeft().isZero()) |
| findIntersection(outerRect.minXMaxYCorner(), innerRect.minXMaxYCorner(), innerRect.minXMinYCorner(), innerRect.maxXMaxYCorner(), quad[1]); |
| |
| if (!innerBorder.radii().bottomRight().isZero()) |
| findIntersection(outerRect.maxXMaxYCorner(), innerRect.maxXMaxYCorner(), innerRect.maxXMinYCorner(), innerRect.minXMaxYCorner(), quad[2]); |
| break; |
| |
| case BoxSide::Right: |
| quad = { outerRect.maxXMinYCorner(), innerRect.maxXMinYCorner(), innerRect.maxXMaxYCorner(), outerRect.maxXMaxYCorner() }; |
| |
| if (!innerBorder.radii().topRight().isZero()) |
| findIntersection(outerRect.maxXMinYCorner(), innerRect.maxXMinYCorner(), innerRect.minXMinYCorner(), innerRect.maxXMaxYCorner(), quad[1]); |
| |
| if (!innerBorder.radii().bottomRight().isZero()) |
| findIntersection(outerRect.maxXMaxYCorner(), innerRect.maxXMaxYCorner(), innerRect.maxXMinYCorner(), innerRect.minXMaxYCorner(), quad[2]); |
| break; |
| } |
| |
| // If the border matches both of its adjacent sides, don't anti-alias the clip, and |
| // if neither side matches, anti-alias the clip. |
| if (firstEdgeMatches == secondEdgeMatches) { |
| bool wasAntialiased = graphicsContext.shouldAntialias(); |
| graphicsContext.setShouldAntialias(!firstEdgeMatches); |
| graphicsContext.clipPath(Path::polygonPathFromPoints(quad), WindRule::NonZero); |
| graphicsContext.setShouldAntialias(wasAntialiased); |
| return; |
| } |
| |
| // Square off the end which shouldn't be affected by antialiasing, and clip. |
| Vector<FloatPoint> firstQuad = { |
| quad[0], |
| quad[1], |
| quad[2], |
| side == BoxSide::Top || side == BoxSide::Bottom ? FloatPoint(quad[3].x(), quad[2].y()) : FloatPoint(quad[2].x(), quad[3].y()), |
| quad[3] |
| }; |
| bool wasAntialiased = graphicsContext.shouldAntialias(); |
| graphicsContext.setShouldAntialias(!firstEdgeMatches); |
| graphicsContext.clipPath(Path::polygonPathFromPoints(firstQuad), WindRule::NonZero); |
| |
| Vector<FloatPoint> secondQuad = { |
| quad[0], |
| side == BoxSide::Top || side == BoxSide::Bottom ? FloatPoint(quad[0].x(), quad[1].y()) : FloatPoint(quad[1].x(), quad[0].y()), |
| quad[1], |
| quad[2], |
| quad[3] |
| }; |
| // Antialiasing affects the second side. |
| graphicsContext.setShouldAntialias(!secondEdgeMatches); |
| graphicsContext.clipPath(Path::polygonPathFromPoints(secondQuad), WindRule::NonZero); |
| |
| graphicsContext.setShouldAntialias(wasAntialiased); |
| } |
| |
| bool RenderBoxModelObject::borderObscuresBackgroundEdge(const FloatSize& contextScale) const |
| { |
| auto edges = borderEdges(style(), document().deviceScaleFactor()); |
| |
| for (auto side : allBoxSides) { |
| auto& currEdge = edges.at(side); |
| // FIXME: for vertical text |
| float axisScale = (side == BoxSide::Top || side == BoxSide::Bottom) ? contextScale.height() : contextScale.width(); |
| if (!currEdge.obscuresBackgroundEdge(axisScale)) |
| return false; |
| } |
| |
| return true; |
| } |
| |
| bool RenderBoxModelObject::borderObscuresBackground() const |
| { |
| if (!style().hasBorder()) |
| return false; |
| |
| // Bail if we have any border-image for now. We could look at the image alpha to improve this. |
| if (style().borderImage().image()) |
| return false; |
| |
| auto edges = borderEdges(style(), document().deviceScaleFactor()); |
| |
| for (auto side : allBoxSides) { |
| if (!edges.at(side).obscuresBackground()) |
| return false; |
| } |
| |
| return true; |
| } |
| |
| bool RenderBoxModelObject::boxShadowShouldBeAppliedToBackground(const LayoutPoint&, BackgroundBleedAvoidance bleedAvoidance, const InlineIterator::InlineBoxIterator& inlineBox) const |
| { |
| if (bleedAvoidance != BackgroundBleedNone) |
| return false; |
| |
| if (style().hasEffectiveAppearance()) |
| return false; |
| |
| bool hasOneNormalBoxShadow = false; |
| for (const ShadowData* currentShadow = style().boxShadow(); currentShadow; currentShadow = currentShadow->next()) { |
| if (currentShadow->style() != ShadowStyle::Normal) |
| continue; |
| |
| if (hasOneNormalBoxShadow) |
| return false; |
| hasOneNormalBoxShadow = true; |
| |
| if (!currentShadow->spread().isZero()) |
| return false; |
| } |
| |
| if (!hasOneNormalBoxShadow) |
| return false; |
| |
| Color backgroundColor = style().visitedDependentColorWithColorFilter(CSSPropertyBackgroundColor); |
| if (!backgroundColor.isOpaque()) |
| return false; |
| |
| auto* lastBackgroundLayer = &style().backgroundLayers(); |
| while (auto* next = lastBackgroundLayer->next()) |
| lastBackgroundLayer = next; |
| |
| if (lastBackgroundLayer->clip() != FillBox::Border) |
| return false; |
| |
| if (lastBackgroundLayer->image() && style().hasBorderRadius()) |
| return false; |
| |
| auto applyToInlineBox = [&] { |
| // The checks here match how paintFillLayer() decides whether to clip (if it does, the shadow |
| // would be clipped out, so it has to be drawn separately). |
| if (inlineBox->isRootInlineBox()) |
| return true; |
| if (!inlineBox->previousInlineBox() && !inlineBox->nextInlineBox()) |
| return true; |
| auto* image = lastBackgroundLayer->image(); |
| auto& renderer = inlineBox->renderer(); |
| bool hasFillImage = image && image->canRender(&renderer, renderer.style().effectiveZoom()); |
| return !hasFillImage && !renderer.style().hasBorderRadius(); |
| }; |
| |
| if (inlineBox && !applyToInlineBox()) |
| return false; |
| |
| if (hasNonVisibleOverflow() && lastBackgroundLayer->attachment() == FillAttachment::LocalBackground) |
| return false; |
| |
| return true; |
| } |
| |
| static inline LayoutRect areaCastingShadowInHole(const LayoutRect& holeRect, LayoutUnit shadowExtent, LayoutUnit shadowSpread, const LayoutSize& shadowOffset) |
| { |
| LayoutRect bounds(holeRect); |
| |
| bounds.inflate(shadowExtent); |
| |
| if (shadowSpread < 0) |
| bounds.inflate(-shadowSpread); |
| |
| LayoutRect offsetBounds = bounds; |
| offsetBounds.move(-shadowOffset); |
| return unionRect(bounds, offsetBounds); |
| } |
| |
| void RenderBoxModelObject::paintBoxShadow(const PaintInfo& info, const LayoutRect& paintRect, const RenderStyle& style, ShadowStyle shadowStyle, bool includeLogicalLeftEdge, bool includeLogicalRightEdge) |
| { |
| // FIXME: Deal with border-image. Would be great to use border-image as a mask. |
| GraphicsContext& context = info.context(); |
| if (context.paintingDisabled() || !style.boxShadow()) |
| return; |
| |
| RoundedRect borderRect = (shadowStyle == ShadowStyle::Inset) ? style.getRoundedInnerBorderFor(paintRect, includeLogicalLeftEdge, includeLogicalRightEdge) |
| : style.getRoundedBorderFor(paintRect, includeLogicalLeftEdge, includeLogicalRightEdge); |
| |
| if (!borderRect.isRenderable()) |
| borderRect.adjustRadii(); |
| |
| bool hasBorderRadius = style.hasBorderRadius(); |
| float deviceScaleFactor = document().deviceScaleFactor(); |
| |
| bool hasOpaqueBackground = style.visitedDependentColorWithColorFilter(CSSPropertyBackgroundColor).isOpaque(); |
| for (const ShadowData* shadow = style.boxShadow(); shadow; shadow = shadow->next()) { |
| if (shadow->style() != shadowStyle) |
| continue; |
| |
| LayoutSize shadowOffset(shadow->x().value(), shadow->y().value()); |
| LayoutUnit shadowPaintingExtent = shadow->paintingExtent(); |
| LayoutUnit shadowSpread = LayoutUnit(shadow->spread().value()); |
| auto shadowRadius = shadow->radius().value(); |
| |
| if (shadowOffset.isZero() && !shadowRadius && !shadowSpread) |
| continue; |
| |
| Color shadowColor = style.colorByApplyingColorFilter(shadow->color()); |
| |
| if (shadow->style() == ShadowStyle::Normal) { |
| auto fillRect = borderRect; |
| fillRect.inflate(shadowSpread); |
| if (fillRect.isEmpty()) |
| continue; |
| |
| auto shadowRect = borderRect.rect(); |
| shadowRect.inflate(shadowPaintingExtent + shadowSpread); |
| shadowRect.move(shadowOffset); |
| auto pixelSnappedShadowRect = snapRectToDevicePixels(shadowRect, deviceScaleFactor); |
| |
| GraphicsContextStateSaver stateSaver(context); |
| context.clip(pixelSnappedShadowRect); |
| |
| // Move the fill just outside the clip, adding at least 1 pixel of separation so that the fill does not |
| // bleed in (due to antialiasing) if the context is transformed. |
| LayoutUnit xOffset = paintRect.width() + std::max<LayoutUnit>(0, shadowOffset.width()) + shadowPaintingExtent + 2 * shadowSpread + LayoutUnit(1); |
| LayoutSize extraOffset(xOffset.ceil(), 0); |
| shadowOffset -= extraOffset; |
| fillRect.move(extraOffset); |
| |
| auto pixelSnappedRectToClipOut = borderRect.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| auto pixelSnappedFillRect = fillRect.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| |
| LayoutPoint shadowRectOrigin = fillRect.rect().location() + shadowOffset; |
| FloatPoint snappedShadowOrigin = FloatPoint(roundToDevicePixel(shadowRectOrigin.x(), deviceScaleFactor), roundToDevicePixel(shadowRectOrigin.y(), deviceScaleFactor)); |
| FloatSize snappedShadowOffset = snappedShadowOrigin - pixelSnappedFillRect.rect().location(); |
| |
| context.setShadow(snappedShadowOffset, shadowRadius, shadowColor, shadow->isWebkitBoxShadow() ? ShadowRadiusMode::Legacy : ShadowRadiusMode::Default); |
| |
| if (hasBorderRadius) { |
| // If the box is opaque, it is unnecessary to clip it out. However, doing so saves time |
| // when painting the shadow. On the other hand, it introduces subpixel gaps along the |
| // corners. Those are avoided by insetting the clipping path by one pixel. |
| if (hasOpaqueBackground) |
| pixelSnappedRectToClipOut.inflateWithRadii(-1.0f); |
| |
| if (!pixelSnappedRectToClipOut.isEmpty()) |
| context.clipOutRoundedRect(pixelSnappedRectToClipOut); |
| |
| RoundedRect influenceRect(LayoutRect(pixelSnappedShadowRect), borderRect.radii()); |
| influenceRect.expandRadii(2 * shadowPaintingExtent + shadowSpread); |
| |
| if (allCornersClippedOut(influenceRect, info.rect)) |
| context.fillRect(pixelSnappedFillRect.rect(), Color::black); |
| else { |
| pixelSnappedFillRect.expandRadii(shadowSpread); |
| if (!pixelSnappedFillRect.isRenderable()) |
| pixelSnappedFillRect.adjustRadii(); |
| context.fillRoundedRect(pixelSnappedFillRect, Color::black); |
| } |
| } else { |
| // If the box is opaque, it is unnecessary to clip it out. However, doing so saves time |
| // when painting the shadow. On the other hand, it introduces subpixel gaps along the |
| // edges if they are not pixel-aligned. Those are avoided by insetting the clipping path |
| // by one pixel. |
| if (hasOpaqueBackground) { |
| // FIXME: The function to decide on the policy based on the transform should be a named function. |
| // FIXME: It's not clear if this check is right. What about integral scale factors? |
| AffineTransform transform = context.getCTM(); |
| if (transform.a() != 1 || (transform.d() != 1 && transform.d() != -1) || transform.b() || transform.c()) |
| pixelSnappedRectToClipOut.inflate(-1.0f); |
| } |
| |
| if (!pixelSnappedRectToClipOut.isEmpty()) |
| context.clipOut(pixelSnappedRectToClipOut.rect()); |
| |
| context.fillRect(pixelSnappedFillRect.rect(), Color::black); |
| } |
| } else { |
| // Inset shadow. |
| auto holeRect = borderRect.rect(); |
| holeRect.inflate(-shadowSpread); |
| |
| bool isHorizontal = style.isHorizontalWritingMode(); |
| if (!includeLogicalLeftEdge) { |
| if (isHorizontal) |
| holeRect.shiftXEdgeBy(-(std::max<LayoutUnit>(shadowOffset.width(), 0) + shadowPaintingExtent + shadowSpread)); |
| else |
| holeRect.shiftYEdgeBy(-(std::max<LayoutUnit>(shadowOffset.height(), 0) + shadowPaintingExtent + shadowSpread)); |
| } |
| |
| if (!includeLogicalRightEdge) { |
| if (isHorizontal) |
| holeRect.setWidth(holeRect.width() - std::min<LayoutUnit>(shadowOffset.width(), 0) + shadowPaintingExtent + shadowSpread); |
| else |
| holeRect.setHeight(holeRect.height() - std::min<LayoutUnit>(shadowOffset.height(), 0) + shadowPaintingExtent + shadowSpread); |
| } |
| |
| auto roundedHoleRect = RoundedRect { holeRect, borderRect.radii() }; |
| if (shadowSpread && roundedHoleRect.isRounded()) { |
| auto rounedRectCorrectingForSpread = [&]() { |
| bool horizontal = style.isHorizontalWritingMode(); |
| LayoutUnit leftWidth { (!horizontal || includeLogicalLeftEdge) ? style.borderLeftWidth() + shadowSpread : 0 }; |
| LayoutUnit rightWidth { (!horizontal || includeLogicalRightEdge) ? style.borderRightWidth() + shadowSpread : 0 }; |
| LayoutUnit topWidth { (horizontal || includeLogicalLeftEdge) ? style.borderTopWidth() + shadowSpread : 0 }; |
| LayoutUnit bottomWidth { (horizontal || includeLogicalRightEdge) ? style.borderBottomWidth() + shadowSpread : 0 }; |
| |
| return style.getRoundedInnerBorderFor(paintRect, topWidth, bottomWidth, leftWidth, rightWidth, includeLogicalLeftEdge, includeLogicalRightEdge); |
| }(); |
| roundedHoleRect.setRadii(rounedRectCorrectingForSpread.radii()); |
| } |
| |
| auto pixelSnappedHoleRect = roundedHoleRect.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| auto pixelSnappedBorderRect = borderRect.pixelSnappedRoundedRectForPainting(deviceScaleFactor); |
| if (pixelSnappedHoleRect.isEmpty()) { |
| if (hasBorderRadius) |
| context.fillRoundedRect(pixelSnappedBorderRect, shadowColor); |
| else |
| context.fillRect(pixelSnappedBorderRect.rect(), shadowColor); |
| continue; |
| } |
| |
| Color fillColor = shadowColor.opaqueColor(); |
| auto shadowCastingRect = areaCastingShadowInHole(borderRect.rect(), shadowPaintingExtent, shadowSpread, shadowOffset); |
| auto pixelSnappedOuterRect = snapRectToDevicePixels(shadowCastingRect, deviceScaleFactor); |
| |
| GraphicsContextStateSaver stateSaver(context); |
| if (hasBorderRadius) |
| context.clipRoundedRect(pixelSnappedBorderRect); |
| else |
| context.clip(pixelSnappedBorderRect.rect()); |
| |
| LayoutUnit xOffset = 2 * paintRect.width() + std::max<LayoutUnit>(0, shadowOffset.width()) + shadowPaintingExtent - 2 * shadowSpread + LayoutUnit(1); |
| LayoutSize extraOffset(xOffset.ceil(), 0); |
| |
| context.translate(extraOffset); |
| shadowOffset -= extraOffset; |
| |
| auto snappedShadowOffset = roundSizeToDevicePixels(shadowOffset, deviceScaleFactor); |
| context.setShadow(snappedShadowOffset, shadowRadius, shadowColor, shadow->isWebkitBoxShadow() ? ShadowRadiusMode::Legacy : ShadowRadiusMode::Default); |
| context.fillRectWithRoundedHole(pixelSnappedOuterRect, pixelSnappedHoleRect, fillColor); |
| } |
| } |
| } |
| |
| LayoutUnit RenderBoxModelObject::containingBlockLogicalWidthForContent() const |
| { |
| if (auto* containingBlock = this->containingBlock()) |
| return containingBlock->availableLogicalWidth(); |
| return { }; |
| } |
| |
| RenderBoxModelObject* RenderBoxModelObject::continuation() const |
| { |
| if (!hasContinuationChainNode()) |
| return nullptr; |
| |
| auto& continuationChainNode = *continuationChainNodeMap().get(this); |
| if (!continuationChainNode.next) |
| return nullptr; |
| return continuationChainNode.next->renderer.get(); |
| } |
| |
| RenderInline* RenderBoxModelObject::inlineContinuation() const |
| { |
| if (!hasContinuationChainNode()) |
| return nullptr; |
| |
| for (auto* next = continuationChainNodeMap().get(this)->next; next; next = next->next) { |
| if (is<RenderInline>(*next->renderer)) |
| return downcast<RenderInline>(next->renderer.get()); |
| } |
| return nullptr; |
| } |
| |
| void RenderBoxModelObject::forRendererAndContinuations(RenderBoxModelObject& renderer, const std::function<void(RenderBoxModelObject&)>& function) |
| { |
| function(renderer); |
| if (!renderer.hasContinuationChainNode()) |
| return; |
| |
| for (auto* next = continuationChainNodeMap().get(&renderer)->next; next; next = next->next) { |
| if (!next->renderer) |
| continue; |
| function(*next->renderer); |
| } |
| } |
| |
| RenderBoxModelObject::ContinuationChainNode* RenderBoxModelObject::continuationChainNode() const |
| { |
| return continuationChainNodeMap().get(this); |
| } |
| |
| void RenderBoxModelObject::insertIntoContinuationChainAfter(RenderBoxModelObject& afterRenderer) |
| { |
| ASSERT(isContinuation()); |
| ASSERT(!continuationChainNodeMap().contains(this)); |
| |
| auto& after = afterRenderer.ensureContinuationChainNode(); |
| ensureContinuationChainNode().insertAfter(after); |
| } |
| |
| void RenderBoxModelObject::removeFromContinuationChain() |
| { |
| ASSERT(hasContinuationChainNode()); |
| ASSERT(continuationChainNodeMap().contains(this)); |
| setHasContinuationChainNode(false); |
| continuationChainNodeMap().remove(this); |
| } |
| |
| auto RenderBoxModelObject::ensureContinuationChainNode() -> ContinuationChainNode& |
| { |
| setHasContinuationChainNode(true); |
| return *continuationChainNodeMap().ensure(this, [&] { |
| return makeUnique<ContinuationChainNode>(*this); |
| }).iterator->value; |
| } |
| |
| RenderTextFragment* RenderBoxModelObject::firstLetterRemainingText() const |
| { |
| if (!isFirstLetter()) |
| return nullptr; |
| return firstLetterRemainingTextMap().get(this).get(); |
| } |
| |
| void RenderBoxModelObject::setFirstLetterRemainingText(RenderTextFragment& remainingText) |
| { |
| ASSERT(isFirstLetter()); |
| firstLetterRemainingTextMap().set(this, remainingText); |
| } |
| |
| void RenderBoxModelObject::clearFirstLetterRemainingText() |
| { |
| ASSERT(isFirstLetter()); |
| firstLetterRemainingTextMap().remove(this); |
| } |
| |
| bool RenderBoxModelObject::shouldAntialiasLines(GraphicsContext& context) |
| { |
| // FIXME: We may want to not antialias when scaled by an integral value, |
| // and we may want to antialias when translated by a non-integral value. |
| return !context.getCTM().isIdentityOrTranslationOrFlipped(); |
| } |
| |
| void RenderBoxModelObject::mapAbsoluteToLocalPoint(OptionSet<MapCoordinatesMode> mode, TransformState& transformState) const |
| { |
| RenderElement* container = this->container(); |
| if (!container) |
| return; |
| |
| container->mapAbsoluteToLocalPoint(mode, transformState); |
| |
| LayoutSize containerOffset = offsetFromContainer(*container, LayoutPoint()); |
| |
| bool preserve3D = mode.contains(UseTransforms) && (container->style().preserves3D() || style().preserves3D()); |
| if (mode.contains(UseTransforms) && shouldUseTransformFromContainer(container)) { |
| TransformationMatrix t; |
| getTransformFromContainer(container, containerOffset, t); |
| transformState.applyTransform(t, preserve3D ? TransformState::AccumulateTransform : TransformState::FlattenTransform); |
| } else |
| transformState.move(containerOffset.width(), containerOffset.height(), preserve3D ? TransformState::AccumulateTransform : TransformState::FlattenTransform); |
| } |
| |
| bool RenderBoxModelObject::hasRunningAcceleratedAnimations() const |
| { |
| if (auto styleable = Styleable::fromRenderer(*this)) |
| return styleable->runningAnimationsAreAllAccelerated(); |
| return false; |
| } |
| |
| void RenderBoxModelObject::collectAbsoluteQuadsForContinuation(Vector<FloatQuad>& quads, bool* wasFixed) const |
| { |
| ASSERT(continuation()); |
| for (auto* nextInContinuation = this->continuation(); nextInContinuation; nextInContinuation = nextInContinuation->continuation()) { |
| if (is<RenderBlock>(*nextInContinuation)) { |
| auto& blockBox = downcast<RenderBlock>(*nextInContinuation); |
| // For blocks inside inlines, we include margins so that we run right up to the inline boxes |
| // above and below us (thus getting merged with them to form a single irregular shape). |
| auto logicalRect = FloatRect { 0, -blockBox.collapsedMarginBefore(), blockBox.width(), |
| blockBox.height() + blockBox.collapsedMarginBefore() + blockBox.collapsedMarginAfter() }; |
| nextInContinuation->absoluteQuadsIgnoringContinuation(logicalRect, quads, wasFixed); |
| continue; |
| } |
| nextInContinuation->absoluteQuadsIgnoringContinuation({ }, quads, wasFixed); |
| } |
| } |
| |
| void RenderBoxModelObject::applyTransform(TransformationMatrix&, const RenderStyle&, const FloatRect&, OptionSet<RenderStyle::TransformOperationOption>) const |
| { |
| // applyTransform() is only used through RenderLayer*, which only invokes this for RenderBox derived renderers, thus not for |
| // RenderInline/RenderLineBreak - the other two renderers that inherit from RenderBoxModelObject. |
| ASSERT_NOT_REACHED(); |
| } |
| |
| } // namespace WebCore |