зеркало из https://github.com/mozilla/gecko-dev.git
5806 строки
187 KiB
C++
5806 строки
187 KiB
C++
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
|
|
/* This Source Code Form is subject to the terms of the Mozilla Public
|
|
* License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
|
|
|
|
// Main header first:
|
|
#include "SVGTextFrame.h"
|
|
|
|
// Keep others in (case-insensitive) order:
|
|
#include "DOMSVGPoint.h"
|
|
#include "gfx2DGlue.h"
|
|
#include "gfxContext.h"
|
|
#include "gfxFont.h"
|
|
#include "gfxSkipChars.h"
|
|
#include "gfxTypes.h"
|
|
#include "gfxUtils.h"
|
|
#include "LookAndFeel.h"
|
|
#include "mozilla/gfx/2D.h"
|
|
#include "mozilla/gfx/PatternHelpers.h"
|
|
#include "mozilla/Likely.h"
|
|
#include "nsAlgorithm.h"
|
|
#include "nsBidiPresUtils.h"
|
|
#include "nsBlockFrame.h"
|
|
#include "nsCaret.h"
|
|
#include "nsContentUtils.h"
|
|
#include "nsGkAtoms.h"
|
|
#include "nsQuickSort.h"
|
|
#include "SVGObserverUtils.h"
|
|
#include "nsSVGOuterSVGFrame.h"
|
|
#include "nsSVGPaintServerFrame.h"
|
|
#include "mozilla/dom/Selection.h"
|
|
#include "mozilla/dom/SVGRect.h"
|
|
#include "mozilla/dom/SVGTextContentElementBinding.h"
|
|
#include "nsSVGIntegrationUtils.h"
|
|
#include "nsSVGUtils.h"
|
|
#include "nsTArray.h"
|
|
#include "nsTextFrame.h"
|
|
#include "nsTextNode.h"
|
|
#include "SVGAnimatedNumberList.h"
|
|
#include "SVGContentUtils.h"
|
|
#include "SVGContextPaint.h"
|
|
#include "SVGLengthList.h"
|
|
#include "SVGNumberList.h"
|
|
#include "SVGGeometryElement.h"
|
|
#include "SVGTextPathElement.h"
|
|
#include "nsLayoutUtils.h"
|
|
#include "nsFrameSelection.h"
|
|
#include "nsStyleStructInlines.h"
|
|
#include <algorithm>
|
|
#include <cmath>
|
|
#include <limits>
|
|
|
|
using namespace mozilla;
|
|
using namespace mozilla::dom;
|
|
using namespace mozilla::dom::SVGTextContentElementBinding;
|
|
using namespace mozilla::gfx;
|
|
using namespace mozilla::image;
|
|
|
|
// ============================================================================
|
|
// Utility functions
|
|
|
|
/**
|
|
* Using the specified gfxSkipCharsIterator, converts an offset and length
|
|
* in original char indexes to skipped char indexes.
|
|
*
|
|
* @param aIterator The gfxSkipCharsIterator to use for the conversion.
|
|
* @param aOriginalOffset The original offset.
|
|
* @param aOriginalLength The original length.
|
|
*/
|
|
static gfxTextRun::Range
|
|
ConvertOriginalToSkipped(gfxSkipCharsIterator& aIterator,
|
|
uint32_t aOriginalOffset, uint32_t aOriginalLength)
|
|
{
|
|
uint32_t start = aIterator.ConvertOriginalToSkipped(aOriginalOffset);
|
|
aIterator.AdvanceOriginal(aOriginalLength);
|
|
return gfxTextRun::Range(start, aIterator.GetSkippedOffset());
|
|
}
|
|
|
|
/**
|
|
* Converts an nsPoint from app units to user space units using the specified
|
|
* nsPresContext and returns it as a gfxPoint.
|
|
*/
|
|
static gfxPoint
|
|
AppUnitsToGfxUnits(const nsPoint& aPoint, const nsPresContext* aContext)
|
|
{
|
|
return gfxPoint(aContext->AppUnitsToGfxUnits(aPoint.x),
|
|
aContext->AppUnitsToGfxUnits(aPoint.y));
|
|
}
|
|
|
|
/**
|
|
* Converts a gfxRect that is in app units to CSS pixels using the specified
|
|
* nsPresContext and returns it as a gfxRect.
|
|
*/
|
|
static gfxRect
|
|
AppUnitsToFloatCSSPixels(const gfxRect& aRect, const nsPresContext* aContext)
|
|
{
|
|
return gfxRect(nsPresContext::AppUnitsToFloatCSSPixels(aRect.x),
|
|
nsPresContext::AppUnitsToFloatCSSPixels(aRect.y),
|
|
nsPresContext::AppUnitsToFloatCSSPixels(aRect.width),
|
|
nsPresContext::AppUnitsToFloatCSSPixels(aRect.height));
|
|
}
|
|
|
|
/**
|
|
* Scales a gfxRect around a given point.
|
|
*
|
|
* @param aRect The rectangle to scale.
|
|
* @param aPoint The point around which to scale.
|
|
* @param aScale The scale amount.
|
|
*/
|
|
static void
|
|
ScaleAround(gfxRect& aRect, const gfxPoint& aPoint, double aScale)
|
|
{
|
|
aRect.x = aPoint.x - aScale * (aPoint.x - aRect.x);
|
|
aRect.y = aPoint.y - aScale * (aPoint.y - aRect.y);
|
|
aRect.width *= aScale;
|
|
aRect.height *= aScale;
|
|
}
|
|
|
|
/**
|
|
* Returns whether a gfxPoint lies within a gfxRect.
|
|
*/
|
|
static bool
|
|
Inside(const gfxRect& aRect, const gfxPoint& aPoint)
|
|
{
|
|
return aPoint.x >= aRect.x &&
|
|
aPoint.x < aRect.XMost() &&
|
|
aPoint.y >= aRect.y &&
|
|
aPoint.y < aRect.YMost();
|
|
}
|
|
|
|
/**
|
|
* Gets the measured ascent and descent of the text in the given nsTextFrame
|
|
* in app units.
|
|
*
|
|
* @param aFrame The text frame.
|
|
* @param aAscent The ascent in app units (output).
|
|
* @param aDescent The descent in app units (output).
|
|
*/
|
|
static void
|
|
GetAscentAndDescentInAppUnits(nsTextFrame* aFrame,
|
|
gfxFloat& aAscent, gfxFloat& aDescent)
|
|
{
|
|
gfxSkipCharsIterator it = aFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = aFrame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
gfxTextRun::Range range = ConvertOriginalToSkipped(
|
|
it, aFrame->GetContentOffset(), aFrame->GetContentLength());
|
|
|
|
gfxTextRun::Metrics metrics =
|
|
textRun->MeasureText(range, gfxFont::LOOSE_INK_EXTENTS, nullptr, nullptr);
|
|
|
|
aAscent = metrics.mAscent;
|
|
aDescent = metrics.mDescent;
|
|
}
|
|
|
|
/**
|
|
* Updates an interval by intersecting it with another interval.
|
|
* The intervals are specified using a start index and a length.
|
|
*/
|
|
static void
|
|
IntersectInterval(uint32_t& aStart, uint32_t& aLength,
|
|
uint32_t aStartOther, uint32_t aLengthOther)
|
|
{
|
|
uint32_t aEnd = aStart + aLength;
|
|
uint32_t aEndOther = aStartOther + aLengthOther;
|
|
|
|
if (aStartOther >= aEnd || aStart >= aEndOther) {
|
|
aLength = 0;
|
|
} else {
|
|
if (aStartOther >= aStart)
|
|
aStart = aStartOther;
|
|
aLength = std::min(aEnd, aEndOther) - aStart;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Intersects an interval as IntersectInterval does but by taking
|
|
* the offset and length of the other interval from a
|
|
* nsTextFrame::TrimmedOffsets object.
|
|
*/
|
|
static void
|
|
TrimOffsets(uint32_t& aStart, uint32_t& aLength,
|
|
const nsTextFrame::TrimmedOffsets& aTrimmedOffsets)
|
|
{
|
|
IntersectInterval(aStart, aLength,
|
|
aTrimmedOffsets.mStart, aTrimmedOffsets.mLength);
|
|
}
|
|
|
|
/**
|
|
* Returns the closest ancestor-or-self node that is not an SVG <a>
|
|
* element.
|
|
*/
|
|
static nsIContent*
|
|
GetFirstNonAAncestor(nsIContent* aContent)
|
|
{
|
|
while (aContent && aContent->IsSVGElement(nsGkAtoms::a)) {
|
|
aContent = aContent->GetParent();
|
|
}
|
|
return aContent;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the given node is a text content element[1], taking into
|
|
* account whether it has a valid parent.
|
|
*
|
|
* For example, in:
|
|
*
|
|
* <svg xmlns="http://www.w3.org/2000/svg">
|
|
* <text><a/><text/></text>
|
|
* <tspan/>
|
|
* </svg>
|
|
*
|
|
* true would be returned for the outer <text> element and the <a> element,
|
|
* and false for the inner <text> element (since a <text> is not allowed
|
|
* to be a child of another <text>) and the <tspan> element (because it
|
|
* must be inside a <text> subtree).
|
|
*
|
|
* Note that we don't support the <tref> element yet and this function
|
|
* returns false for it.
|
|
*
|
|
* [1] https://svgwg.org/svg2-draft/intro.html#TermTextContentElement
|
|
*/
|
|
static bool
|
|
IsTextContentElement(nsIContent* aContent)
|
|
{
|
|
if (aContent->IsSVGElement(nsGkAtoms::text)) {
|
|
nsIContent* parent = GetFirstNonAAncestor(aContent->GetParent());
|
|
return !parent || !IsTextContentElement(parent);
|
|
}
|
|
|
|
if (aContent->IsSVGElement(nsGkAtoms::textPath)) {
|
|
nsIContent* parent = GetFirstNonAAncestor(aContent->GetParent());
|
|
return parent && parent->IsSVGElement(nsGkAtoms::text);
|
|
}
|
|
|
|
if (aContent->IsAnyOfSVGElements(nsGkAtoms::a,
|
|
nsGkAtoms::tspan)) {
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the specified frame is an nsTextFrame that has some text
|
|
* content.
|
|
*/
|
|
static bool
|
|
IsNonEmptyTextFrame(nsIFrame* aFrame)
|
|
{
|
|
nsTextFrame* textFrame = do_QueryFrame(aFrame);
|
|
if (!textFrame) {
|
|
return false;
|
|
}
|
|
|
|
return textFrame->GetContentLength() != 0;
|
|
}
|
|
|
|
/**
|
|
* Takes an nsIFrame and if it is a text frame that has some text content,
|
|
* returns it as an nsTextFrame and its corresponding nsTextNode.
|
|
*
|
|
* @param aFrame The frame to look at.
|
|
* @param aTextFrame aFrame as an nsTextFrame (output).
|
|
* @param aTextNode The nsTextNode content of aFrame (output).
|
|
* @return true if aFrame is a non-empty text frame, false otherwise.
|
|
*/
|
|
static bool
|
|
GetNonEmptyTextFrameAndNode(nsIFrame* aFrame,
|
|
nsTextFrame*& aTextFrame,
|
|
nsTextNode*& aTextNode)
|
|
{
|
|
nsTextFrame* text = do_QueryFrame(aFrame);
|
|
bool isNonEmptyTextFrame = text && text->GetContentLength() != 0;
|
|
|
|
if (isNonEmptyTextFrame) {
|
|
nsIContent* content = text->GetContent();
|
|
NS_ASSERTION(content && content->IsText(),
|
|
"unexpected content type for nsTextFrame");
|
|
|
|
nsTextNode* node = static_cast<nsTextNode*>(content);
|
|
MOZ_ASSERT(node->TextLength() != 0,
|
|
"frame's GetContentLength() should be 0 if the text node "
|
|
"has no content");
|
|
|
|
aTextFrame = text;
|
|
aTextNode = node;
|
|
}
|
|
|
|
MOZ_ASSERT(IsNonEmptyTextFrame(aFrame) == isNonEmptyTextFrame,
|
|
"our logic should agree with IsNonEmptyTextFrame");
|
|
return isNonEmptyTextFrame;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the specified atom is for one of the five
|
|
* glyph positioning attributes that can appear on SVG text
|
|
* elements -- x, y, dx, dy or rotate.
|
|
*/
|
|
static bool
|
|
IsGlyphPositioningAttribute(nsAtom* aAttribute)
|
|
{
|
|
return aAttribute == nsGkAtoms::x ||
|
|
aAttribute == nsGkAtoms::y ||
|
|
aAttribute == nsGkAtoms::dx ||
|
|
aAttribute == nsGkAtoms::dy ||
|
|
aAttribute == nsGkAtoms::rotate;
|
|
}
|
|
|
|
/**
|
|
* Returns the position in app units of a given baseline (using an
|
|
* SVG dominant-baseline property value) for a given nsTextFrame.
|
|
*
|
|
* @param aFrame The text frame to inspect.
|
|
* @param aTextRun The text run of aFrame.
|
|
* @param aDominantBaseline The dominant-baseline value to use.
|
|
*/
|
|
static nscoord
|
|
GetBaselinePosition(nsTextFrame* aFrame,
|
|
gfxTextRun* aTextRun,
|
|
uint8_t aDominantBaseline,
|
|
float aFontSizeScaleFactor)
|
|
{
|
|
WritingMode writingMode = aFrame->GetWritingMode();
|
|
gfxTextRun::Metrics metrics =
|
|
aTextRun->MeasureText(gfxFont::LOOSE_INK_EXTENTS, nullptr);
|
|
|
|
switch (aDominantBaseline) {
|
|
case NS_STYLE_DOMINANT_BASELINE_HANGING:
|
|
case NS_STYLE_DOMINANT_BASELINE_TEXT_BEFORE_EDGE:
|
|
return writingMode.IsVerticalRL()
|
|
? metrics.mAscent + metrics.mDescent : 0;
|
|
|
|
case NS_STYLE_DOMINANT_BASELINE_USE_SCRIPT:
|
|
case NS_STYLE_DOMINANT_BASELINE_NO_CHANGE:
|
|
case NS_STYLE_DOMINANT_BASELINE_RESET_SIZE:
|
|
// These three should not simply map to 'baseline', but we don't
|
|
// support the complex baseline model that SVG 1.1 has and which
|
|
// css3-linebox now defines.
|
|
// (fall through)
|
|
|
|
case NS_STYLE_DOMINANT_BASELINE_AUTO:
|
|
case NS_STYLE_DOMINANT_BASELINE_ALPHABETIC:
|
|
return writingMode.IsVerticalRL()
|
|
? metrics.mAscent + metrics.mDescent -
|
|
aFrame->GetLogicalBaseline(writingMode)
|
|
: aFrame->GetLogicalBaseline(writingMode);
|
|
|
|
case NS_STYLE_DOMINANT_BASELINE_MIDDLE:
|
|
return aFrame->GetLogicalBaseline(writingMode) -
|
|
SVGContentUtils::GetFontXHeight(aFrame) / 2.0 *
|
|
aFrame->PresContext()->AppUnitsPerCSSPixel() * aFontSizeScaleFactor;
|
|
|
|
case NS_STYLE_DOMINANT_BASELINE_TEXT_AFTER_EDGE:
|
|
case NS_STYLE_DOMINANT_BASELINE_IDEOGRAPHIC:
|
|
return writingMode.IsVerticalLR()
|
|
? 0 : metrics.mAscent + metrics.mDescent;
|
|
|
|
case NS_STYLE_DOMINANT_BASELINE_CENTRAL:
|
|
case NS_STYLE_DOMINANT_BASELINE_MATHEMATICAL:
|
|
return (metrics.mAscent + metrics.mDescent) / 2.0;
|
|
}
|
|
|
|
MOZ_ASSERT_UNREACHABLE("unexpected dominant-baseline value");
|
|
return aFrame->GetLogicalBaseline(writingMode);
|
|
}
|
|
|
|
/**
|
|
* For a given text run, returns the range of skipped characters that comprise
|
|
* the ligature group and/or cluster that includes the character represented
|
|
* by the specified gfxSkipCharsIterator.
|
|
*
|
|
* @param aTextRun The text run to use for determining whether a given character
|
|
* is part of a ligature or cluster.
|
|
* @param aIterator The gfxSkipCharsIterator to use for the current position
|
|
* in the text run.
|
|
*/
|
|
static gfxTextRun::Range
|
|
ClusterRange(gfxTextRun* aTextRun, const gfxSkipCharsIterator& aIterator)
|
|
{
|
|
uint32_t start = aIterator.GetSkippedOffset();
|
|
uint32_t end = start + 1;
|
|
while (end < aTextRun->GetLength() &&
|
|
(!aTextRun->IsLigatureGroupStart(end) ||
|
|
!aTextRun->IsClusterStart(end))) {
|
|
end++;
|
|
}
|
|
return gfxTextRun::Range(start, end);
|
|
}
|
|
|
|
/**
|
|
* Truncates an array to be at most the length of another array.
|
|
*
|
|
* @param aArrayToTruncate The array to truncate.
|
|
* @param aReferenceArray The array whose length will be used to truncate
|
|
* aArrayToTruncate to.
|
|
*/
|
|
template<typename T, typename U>
|
|
static void
|
|
TruncateTo(nsTArray<T>& aArrayToTruncate, const nsTArray<U>& aReferenceArray)
|
|
{
|
|
uint32_t length = aReferenceArray.Length();
|
|
if (aArrayToTruncate.Length() > length) {
|
|
aArrayToTruncate.TruncateLength(length);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Asserts that the anonymous block child of the SVGTextFrame has been
|
|
* reflowed (or does not exist). Returns null if the child has not been
|
|
* reflowed, and the frame otherwise.
|
|
*
|
|
* We check whether the kid has been reflowed and not the frame itself
|
|
* since we sometimes need to call this function during reflow, after the
|
|
* kid has been reflowed but before we have cleared the dirty bits on the
|
|
* frame itself.
|
|
*/
|
|
static SVGTextFrame*
|
|
FrameIfAnonymousChildReflowed(SVGTextFrame* aFrame)
|
|
{
|
|
MOZ_ASSERT(aFrame, "aFrame must not be null");
|
|
nsIFrame* kid = aFrame->PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
MOZ_ASSERT(false, "should have already reflowed the anonymous block child");
|
|
return nullptr;
|
|
}
|
|
return aFrame;
|
|
}
|
|
|
|
static double
|
|
GetContextScale(const gfxMatrix& aMatrix)
|
|
{
|
|
// The context scale is the ratio of the length of the transformed
|
|
// diagonal vector (1,1) to the length of the untransformed diagonal
|
|
// (which is sqrt(2)).
|
|
gfxPoint p = aMatrix.TransformPoint(gfxPoint(1, 1)) -
|
|
aMatrix.TransformPoint(gfxPoint(0, 0));
|
|
return SVGContentUtils::ComputeNormalizedHypotenuse(p.x, p.y);
|
|
}
|
|
|
|
// ============================================================================
|
|
// Utility classes
|
|
|
|
namespace mozilla {
|
|
|
|
// ----------------------------------------------------------------------------
|
|
// TextRenderedRun
|
|
|
|
/**
|
|
* A run of text within a single nsTextFrame whose glyphs can all be painted
|
|
* with a single call to nsTextFrame::PaintText. A text rendered run can
|
|
* be created for a sequence of two or more consecutive glyphs as long as:
|
|
*
|
|
* - Only the first glyph has (or none of the glyphs have) been positioned
|
|
* with SVG text positioning attributes
|
|
* - All of the glyphs have zero rotation
|
|
* - The glyphs are not on a text path
|
|
* - The glyphs correspond to content within the one nsTextFrame
|
|
*
|
|
* A TextRenderedRunIterator produces TextRenderedRuns required for painting a
|
|
* whole SVGTextFrame.
|
|
*/
|
|
struct TextRenderedRun
|
|
{
|
|
typedef gfxTextRun::Range Range;
|
|
|
|
/**
|
|
* Constructs a TextRenderedRun that is uninitialized except for mFrame
|
|
* being null.
|
|
*/
|
|
TextRenderedRun()
|
|
: mFrame(nullptr)
|
|
{
|
|
}
|
|
|
|
/**
|
|
* Constructs a TextRenderedRun with all of the information required to
|
|
* paint it. See the comments documenting the member variables below
|
|
* for descriptions of the arguments.
|
|
*/
|
|
TextRenderedRun(nsTextFrame* aFrame, const gfxPoint& aPosition,
|
|
float aLengthAdjustScaleFactor, double aRotate,
|
|
float aFontSizeScaleFactor, nscoord aBaseline,
|
|
uint32_t aTextFrameContentOffset,
|
|
uint32_t aTextFrameContentLength,
|
|
uint32_t aTextElementCharIndex)
|
|
: mFrame(aFrame),
|
|
mPosition(aPosition),
|
|
mLengthAdjustScaleFactor(aLengthAdjustScaleFactor),
|
|
mRotate(static_cast<float>(aRotate)),
|
|
mFontSizeScaleFactor(aFontSizeScaleFactor),
|
|
mBaseline(aBaseline),
|
|
mTextFrameContentOffset(aTextFrameContentOffset),
|
|
mTextFrameContentLength(aTextFrameContentLength),
|
|
mTextElementCharIndex(aTextElementCharIndex)
|
|
{
|
|
}
|
|
|
|
/**
|
|
* Returns the text run for the text frame that this rendered run is part of.
|
|
*/
|
|
gfxTextRun* GetTextRun() const
|
|
{
|
|
mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
return mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
}
|
|
|
|
/**
|
|
* Returns whether this rendered run is RTL.
|
|
*/
|
|
bool IsRightToLeft() const
|
|
{
|
|
return GetTextRun()->IsRightToLeft();
|
|
}
|
|
|
|
/**
|
|
* Returns whether this rendered run is vertical.
|
|
*/
|
|
bool IsVertical() const
|
|
{
|
|
return GetTextRun()->IsVertical();
|
|
}
|
|
|
|
/**
|
|
* Returns the transform that converts from a <text> element's user space into
|
|
* the coordinate space that rendered runs can be painted directly in.
|
|
*
|
|
* The difference between this method and GetTransformFromRunUserSpaceToUserSpace
|
|
* is that when calling in to nsTextFrame::PaintText, it will already take
|
|
* into account any left clip edge (that is, it doesn't just apply a visual
|
|
* clip to the rendered text, it shifts the glyphs over so that they are
|
|
* painted with their left edge at the x coordinate passed in to it).
|
|
* Thus we need to account for this in our transform.
|
|
*
|
|
*
|
|
* Assume that we have <text x="100" y="100" rotate="0 0 1 0 0 1">abcdef</text>.
|
|
* This would result in four text rendered runs:
|
|
*
|
|
* - one for "ab"
|
|
* - one for "c"
|
|
* - one for "de"
|
|
* - one for "f"
|
|
*
|
|
* Assume now that we are painting the third TextRenderedRun. It will have
|
|
* a left clip edge that is the sum of the advances of "abc", and it will
|
|
* have a right clip edge that is the advance of "f". In
|
|
* SVGTextFrame::PaintSVG(), we pass in nsPoint() (i.e., the origin)
|
|
* as the point at which to paint the text frame, and we pass in the
|
|
* clip edge values. The nsTextFrame will paint the substring of its
|
|
* text such that the top-left corner of the "d"'s glyph cell will be at
|
|
* (0, 0) in the current coordinate system.
|
|
*
|
|
* Thus, GetTransformFromUserSpaceForPainting must return a transform from
|
|
* whatever user space the <text> element is in to a coordinate space in
|
|
* device pixels (as that's what nsTextFrame works in) where the origin is at
|
|
* the same position as our user space mPositions[i].mPosition value for
|
|
* the "d" glyph, which will be (100 + userSpaceAdvance("abc"), 100).
|
|
* The translation required to do this (ignoring the scale to get from
|
|
* user space to device pixels, and ignoring the
|
|
* (100 + userSpaceAdvance("abc"), 100) translation) is:
|
|
*
|
|
* (-leftEdge, -baseline)
|
|
*
|
|
* where baseline is the distance between the baseline of the text and the top
|
|
* edge of the nsTextFrame. We translate by -leftEdge horizontally because
|
|
* the nsTextFrame will already shift the glyphs over by that amount and start
|
|
* painting glyphs at x = 0. We translate by -baseline vertically so that
|
|
* painting the top edges of the glyphs at y = 0 will result in their
|
|
* baselines being at our desired y position.
|
|
*
|
|
*
|
|
* Now for an example with RTL text. Assume our content is now
|
|
* <text x="100" y="100" rotate="0 0 1 0 0 1">WERBEH</text>. We'd have
|
|
* the following text rendered runs:
|
|
*
|
|
* - one for "EH"
|
|
* - one for "B"
|
|
* - one for "ER"
|
|
* - one for "W"
|
|
*
|
|
* Again, we are painting the third TextRenderedRun. The left clip edge
|
|
* is the advance of the "W" and the right clip edge is the sum of the
|
|
* advances of "BEH". Our translation to get the rendered "ER" glyphs
|
|
* in the right place this time is:
|
|
*
|
|
* (-frameWidth + rightEdge, -baseline)
|
|
*
|
|
* which is equivalent to:
|
|
*
|
|
* (-(leftEdge + advance("ER")), -baseline)
|
|
*
|
|
* The reason we have to shift left additionally by the width of the run
|
|
* of glyphs we are painting is that although the nsTextFrame is RTL,
|
|
* we still supply the top-left corner to paint the frame at when calling
|
|
* nsTextFrame::PaintText, even though our user space positions for each
|
|
* glyph in mPositions specifies the origin of each glyph, which for RTL
|
|
* glyphs is at the right edge of the glyph cell.
|
|
*
|
|
*
|
|
* For any other use of an nsTextFrame in the context of a particular run
|
|
* (such as hit testing, or getting its rectangle),
|
|
* GetTransformFromRunUserSpaceToUserSpace should be used.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
* @param aItem The nsCharClipDisplayItem that holds the amount of clipping
|
|
* from the left and right edges of the text frame for this rendered run.
|
|
* An appropriate nsCharClipDisplayItem can be obtained by constructing an
|
|
* SVGCharClipDisplayItem for the TextRenderedRun.
|
|
*/
|
|
gfxMatrix GetTransformFromUserSpaceForPainting(
|
|
nsPresContext* aContext,
|
|
const nsCharClipDisplayItem& aItem) const;
|
|
|
|
/**
|
|
* Returns the transform that converts from "run user space" to a <text>
|
|
* element's user space. Run user space is a coordinate system that has the
|
|
* same size as the <text>'s user space but rotated and translated such that
|
|
* (0,0) is the top-left of the rectangle that bounds the text.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
*/
|
|
gfxMatrix GetTransformFromRunUserSpaceToUserSpace(nsPresContext* aContext) const;
|
|
|
|
/**
|
|
* Returns the transform that converts from "run user space" to float pixels
|
|
* relative to the nsTextFrame that this rendered run is a part of.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
*/
|
|
gfxMatrix GetTransformFromRunUserSpaceToFrameUserSpace(nsPresContext* aContext) const;
|
|
|
|
/**
|
|
* Flag values used for the aFlags arguments of GetRunUserSpaceRect,
|
|
* GetFrameUserSpaceRect and GetUserSpaceRect.
|
|
*/
|
|
enum {
|
|
// Includes the fill geometry of the text in the returned rectangle.
|
|
eIncludeFill = 1,
|
|
// Includes the stroke geometry of the text in the returned rectangle.
|
|
eIncludeStroke = 2,
|
|
// Includes any text shadow in the returned rectangle.
|
|
eIncludeTextShadow = 4,
|
|
// Don't include any horizontal glyph overflow in the returned rectangle.
|
|
eNoHorizontalOverflow = 8
|
|
};
|
|
|
|
/**
|
|
* Returns a rectangle that bounds the fill and/or stroke of the rendered run
|
|
* in run user space.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
* @param aFlags A combination of the flags above (eIncludeFill and
|
|
* eIncludeStroke) indicating what parts of the text to include in
|
|
* the rectangle.
|
|
*/
|
|
SVGBBox GetRunUserSpaceRect(nsPresContext* aContext, uint32_t aFlags) const;
|
|
|
|
/**
|
|
* Returns a rectangle that covers the fill and/or stroke of the rendered run
|
|
* in "frame user space".
|
|
*
|
|
* Frame user space is a coordinate space of the same scale as the <text>
|
|
* element's user space, but with its rotation set to the rotation of
|
|
* the glyphs within this rendered run and its origin set to the position
|
|
* such that placing the nsTextFrame there would result in the glyphs in
|
|
* this rendered run being at their correct positions.
|
|
*
|
|
* For example, say we have <text x="100 150" y="100">ab</text>. Assume
|
|
* the advance of both the "a" and the "b" is 12 user units, and the
|
|
* ascent of the text is 8 user units and its descent is 6 user units,
|
|
* and that we are not measuing the stroke of the text, so that we stay
|
|
* entirely within the glyph cells.
|
|
*
|
|
* There will be two text rendered runs, one for "a" and one for "b".
|
|
*
|
|
* The frame user space for the "a" run will have its origin at
|
|
* (100, 100 - 8) in the <text> element's user space and will have its
|
|
* axes aligned with the user space (since there is no rotate="" or
|
|
* text path involve) and with its scale the same as the user space.
|
|
* The rect returned by this method will be (0, 0, 12, 14), since the "a"
|
|
* glyph is right at the left of the nsTextFrame.
|
|
*
|
|
* The frame user space for the "b" run will have its origin at
|
|
* (150 - 12, 100 - 8), and scale/rotation the same as above. The rect
|
|
* returned by this method will be (12, 0, 12, 14), since we are
|
|
* advance("a") horizontally in to the text frame.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
* @param aFlags A combination of the flags above (eIncludeFill and
|
|
* eIncludeStroke) indicating what parts of the text to include in
|
|
* the rectangle.
|
|
*/
|
|
SVGBBox GetFrameUserSpaceRect(nsPresContext* aContext, uint32_t aFlags) const;
|
|
|
|
/**
|
|
* Returns a rectangle that covers the fill and/or stroke of the rendered run
|
|
* in the <text> element's user space.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
* @param aFlags A combination of the flags above indicating what parts of the
|
|
* text to include in the rectangle.
|
|
* @param aAdditionalTransform An additional transform to apply to the
|
|
* frame user space rectangle before its bounds are transformed into
|
|
* user space.
|
|
*/
|
|
SVGBBox GetUserSpaceRect(nsPresContext* aContext, uint32_t aFlags,
|
|
const gfxMatrix* aAdditionalTransform = nullptr) const;
|
|
|
|
/**
|
|
* Gets the app unit amounts to clip from the left and right edges of
|
|
* the nsTextFrame in order to paint just this rendered run.
|
|
*
|
|
* Note that if clip edge amounts land in the middle of a glyph, the
|
|
* glyph won't be painted at all. The clip edges are thus more of
|
|
* a selection mechanism for which glyphs will be painted, rather
|
|
* than a geometric clip.
|
|
*/
|
|
void GetClipEdges(nscoord& aVisIStartEdge, nscoord& aVisIEndEdge) const;
|
|
|
|
/**
|
|
* Returns the advance width of the whole rendered run.
|
|
*/
|
|
nscoord GetAdvanceWidth() const;
|
|
|
|
/**
|
|
* Returns the index of the character into this rendered run whose
|
|
* glyph cell contains the given point, or -1 if there is no such
|
|
* character. This does not hit test against any overflow.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
* @param aPoint The point in the user space of the <text> element.
|
|
*/
|
|
int32_t GetCharNumAtPosition(nsPresContext* aContext,
|
|
const gfxPoint& aPoint) const;
|
|
|
|
/**
|
|
* The text frame that this rendered run lies within.
|
|
*/
|
|
nsTextFrame* mFrame;
|
|
|
|
/**
|
|
* The point in user space that the text is positioned at.
|
|
*
|
|
* For a horizontal run:
|
|
* The x coordinate is the left edge of a LTR run of text or the right edge of
|
|
* an RTL run. The y coordinate is the baseline of the text.
|
|
* For a vertical run:
|
|
* The x coordinate is the baseline of the text.
|
|
* The y coordinate is the top edge of a LTR run, or bottom of RTL.
|
|
*/
|
|
gfxPoint mPosition;
|
|
|
|
/**
|
|
* The horizontal scale factor to apply when painting glyphs to take
|
|
* into account textLength="".
|
|
*/
|
|
float mLengthAdjustScaleFactor;
|
|
|
|
/**
|
|
* The rotation in radians in the user coordinate system that the text has.
|
|
*/
|
|
float mRotate;
|
|
|
|
/**
|
|
* The scale factor that was used to transform the text run's original font
|
|
* size into a sane range for painting and measurement.
|
|
*/
|
|
double mFontSizeScaleFactor;
|
|
|
|
/**
|
|
* The baseline in app units of this text run. The measurement is from the
|
|
* top of the text frame. (From the left edge if vertical.)
|
|
*/
|
|
nscoord mBaseline;
|
|
|
|
/**
|
|
* The offset and length in mFrame's content nsTextNode that corresponds to
|
|
* this text rendered run. These are original char indexes.
|
|
*/
|
|
uint32_t mTextFrameContentOffset;
|
|
uint32_t mTextFrameContentLength;
|
|
|
|
/**
|
|
* The character index in the whole SVG <text> element that this text rendered
|
|
* run begins at.
|
|
*/
|
|
uint32_t mTextElementCharIndex;
|
|
};
|
|
|
|
gfxMatrix
|
|
TextRenderedRun::GetTransformFromUserSpaceForPainting(
|
|
nsPresContext* aContext,
|
|
const nsCharClipDisplayItem& aItem) const
|
|
{
|
|
// We transform to device pixels positioned such that painting the text frame
|
|
// at (0,0) with aItem will result in the text being in the right place.
|
|
|
|
gfxMatrix m;
|
|
if (!mFrame) {
|
|
return m;
|
|
}
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
// Glyph position in user space.
|
|
m.PreTranslate(mPosition / cssPxPerDevPx);
|
|
|
|
// Take into account any font size scaling and scaling due to textLength="".
|
|
m.PreScale(1.0 / mFontSizeScaleFactor, 1.0 / mFontSizeScaleFactor);
|
|
|
|
// Rotation due to rotate="" or a <textPath>.
|
|
m.PreRotate(mRotate);
|
|
|
|
m.PreScale(mLengthAdjustScaleFactor, 1.0);
|
|
|
|
// Translation to get the text frame in the right place.
|
|
nsPoint t;
|
|
if (IsVertical()) {
|
|
t = nsPoint(-mBaseline,
|
|
IsRightToLeft()
|
|
? -mFrame->GetRect().height + aItem.mVisIEndEdge
|
|
: -aItem.mVisIStartEdge);
|
|
} else {
|
|
t = nsPoint(IsRightToLeft()
|
|
? -mFrame->GetRect().width + aItem.mVisIEndEdge
|
|
: -aItem.mVisIStartEdge,
|
|
-mBaseline);
|
|
}
|
|
m.PreTranslate(AppUnitsToGfxUnits(t, aContext));
|
|
|
|
return m;
|
|
}
|
|
|
|
gfxMatrix
|
|
TextRenderedRun::GetTransformFromRunUserSpaceToUserSpace(
|
|
nsPresContext* aContext) const
|
|
{
|
|
gfxMatrix m;
|
|
if (!mFrame) {
|
|
return m;
|
|
}
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
nscoord start, end;
|
|
GetClipEdges(start, end);
|
|
|
|
// Glyph position in user space.
|
|
m.PreTranslate(mPosition);
|
|
|
|
// Rotation due to rotate="" or a <textPath>.
|
|
m.PreRotate(mRotate);
|
|
|
|
// Scale due to textLength="".
|
|
m.PreScale(mLengthAdjustScaleFactor, 1.0);
|
|
|
|
// Translation to get the text frame in the right place.
|
|
nsPoint t;
|
|
if (IsVertical()) {
|
|
t = nsPoint(-mBaseline,
|
|
IsRightToLeft()
|
|
? -mFrame->GetRect().height + start + end
|
|
: 0);
|
|
} else {
|
|
t = nsPoint(IsRightToLeft()
|
|
? -mFrame->GetRect().width + start + end
|
|
: 0,
|
|
-mBaseline);
|
|
}
|
|
m.PreTranslate(AppUnitsToGfxUnits(t, aContext) *
|
|
cssPxPerDevPx / mFontSizeScaleFactor);
|
|
|
|
return m;
|
|
}
|
|
|
|
gfxMatrix
|
|
TextRenderedRun::GetTransformFromRunUserSpaceToFrameUserSpace(
|
|
nsPresContext* aContext) const
|
|
{
|
|
gfxMatrix m;
|
|
if (!mFrame) {
|
|
return m;
|
|
}
|
|
|
|
nscoord start, end;
|
|
GetClipEdges(start, end);
|
|
|
|
// Translate by the horizontal distance into the text frame this
|
|
// rendered run is.
|
|
gfxFloat appPerCssPx = nsPresContext::AppUnitsPerCSSPixel();
|
|
gfxPoint t = IsVertical() ? gfxPoint(0, start / appPerCssPx)
|
|
: gfxPoint(start / appPerCssPx, 0);
|
|
return m.PreTranslate(t);
|
|
}
|
|
|
|
SVGBBox
|
|
TextRenderedRun::GetRunUserSpaceRect(nsPresContext* aContext,
|
|
uint32_t aFlags) const
|
|
{
|
|
SVGBBox r;
|
|
if (!mFrame) {
|
|
return r;
|
|
}
|
|
|
|
// Determine the amount of overflow above and below the frame's mRect.
|
|
//
|
|
// We need to call GetVisualOverflowRectRelativeToSelf because this includes
|
|
// overflowing decorations, which the MeasureText call below does not. We
|
|
// assume here the decorations only overflow above and below the frame, never
|
|
// horizontally.
|
|
nsRect self = mFrame->GetVisualOverflowRectRelativeToSelf();
|
|
nsRect rect = mFrame->GetRect();
|
|
bool vertical = IsVertical();
|
|
nscoord above = vertical ? -self.x : -self.y;
|
|
nscoord below = vertical ? self.XMost() - rect.width
|
|
: self.YMost() - rect.height;
|
|
|
|
gfxSkipCharsIterator it = mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
// Get the content range for this rendered run.
|
|
Range range = ConvertOriginalToSkipped(it, mTextFrameContentOffset,
|
|
mTextFrameContentLength);
|
|
if (range.Length() == 0) {
|
|
return r;
|
|
}
|
|
|
|
// Measure that range.
|
|
gfxTextRun::Metrics metrics =
|
|
textRun->MeasureText(range, gfxFont::LOOSE_INK_EXTENTS, nullptr, nullptr);
|
|
// Make sure it includes the font-box.
|
|
gfxRect fontBox(0, -metrics.mAscent,
|
|
metrics.mAdvanceWidth, metrics.mAscent + metrics.mDescent);
|
|
metrics.mBoundingBox.UnionRect(metrics.mBoundingBox, fontBox);
|
|
|
|
// Determine the rectangle that covers the rendered run's fill,
|
|
// taking into account the measured vertical overflow due to
|
|
// decorations.
|
|
nscoord baseline = metrics.mBoundingBox.y + metrics.mAscent;
|
|
gfxFloat x, width;
|
|
if (aFlags & eNoHorizontalOverflow) {
|
|
x = 0.0;
|
|
width = textRun->GetAdvanceWidth(range, nullptr);
|
|
} else {
|
|
x = metrics.mBoundingBox.x;
|
|
width = metrics.mBoundingBox.width;
|
|
}
|
|
nsRect fillInAppUnits(x, baseline - above,
|
|
width, metrics.mBoundingBox.height + above + below);
|
|
if (textRun->IsVertical()) {
|
|
// Swap line-relative textMetrics dimensions to physical coordinates.
|
|
Swap(fillInAppUnits.x, fillInAppUnits.y);
|
|
Swap(fillInAppUnits.width, fillInAppUnits.height);
|
|
}
|
|
|
|
// Account for text-shadow.
|
|
if (aFlags & eIncludeTextShadow) {
|
|
fillInAppUnits =
|
|
nsLayoutUtils::GetTextShadowRectsUnion(fillInAppUnits, mFrame);
|
|
}
|
|
|
|
// Convert the app units rectangle to user units.
|
|
gfxRect fill = AppUnitsToFloatCSSPixels(gfxRect(fillInAppUnits.x,
|
|
fillInAppUnits.y,
|
|
fillInAppUnits.width,
|
|
fillInAppUnits.height),
|
|
aContext);
|
|
|
|
// Scale the rectangle up due to any mFontSizeScaleFactor. We scale
|
|
// it around the text's origin.
|
|
ScaleAround(fill,
|
|
textRun->IsVertical()
|
|
? gfxPoint(nsPresContext::AppUnitsToFloatCSSPixels(baseline), 0.0)
|
|
: gfxPoint(0.0, nsPresContext::AppUnitsToFloatCSSPixels(baseline)),
|
|
1.0 / mFontSizeScaleFactor);
|
|
|
|
// Include the fill if requested.
|
|
if (aFlags & eIncludeFill) {
|
|
r = fill;
|
|
}
|
|
|
|
// Include the stroke if requested.
|
|
if ((aFlags & eIncludeStroke) &&
|
|
!fill.IsEmpty() &&
|
|
nsSVGUtils::GetStrokeWidth(mFrame) > 0) {
|
|
r.UnionEdges(nsSVGUtils::PathExtentsToMaxStrokeExtents(fill, mFrame,
|
|
gfxMatrix()));
|
|
}
|
|
|
|
return r;
|
|
}
|
|
|
|
SVGBBox
|
|
TextRenderedRun::GetFrameUserSpaceRect(nsPresContext* aContext,
|
|
uint32_t aFlags) const
|
|
{
|
|
SVGBBox r = GetRunUserSpaceRect(aContext, aFlags);
|
|
if (r.IsEmpty()) {
|
|
return r;
|
|
}
|
|
gfxMatrix m = GetTransformFromRunUserSpaceToFrameUserSpace(aContext);
|
|
return m.TransformBounds(r.ToThebesRect());
|
|
}
|
|
|
|
SVGBBox
|
|
TextRenderedRun::GetUserSpaceRect(nsPresContext* aContext,
|
|
uint32_t aFlags,
|
|
const gfxMatrix* aAdditionalTransform) const
|
|
{
|
|
SVGBBox r = GetRunUserSpaceRect(aContext, aFlags);
|
|
if (r.IsEmpty()) {
|
|
return r;
|
|
}
|
|
gfxMatrix m = GetTransformFromRunUserSpaceToUserSpace(aContext);
|
|
if (aAdditionalTransform) {
|
|
m *= *aAdditionalTransform;
|
|
}
|
|
return m.TransformBounds(r.ToThebesRect());
|
|
}
|
|
|
|
void
|
|
TextRenderedRun::GetClipEdges(nscoord& aVisIStartEdge,
|
|
nscoord& aVisIEndEdge) const
|
|
{
|
|
uint32_t contentLength = mFrame->GetContentLength();
|
|
if (mTextFrameContentOffset == 0 &&
|
|
mTextFrameContentLength == contentLength) {
|
|
// If the rendered run covers the entire content, we know we don't need
|
|
// to clip without having to measure anything.
|
|
aVisIStartEdge = 0;
|
|
aVisIEndEdge = 0;
|
|
return;
|
|
}
|
|
|
|
gfxSkipCharsIterator it = mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
// Get the covered content offset/length for this rendered run in skipped
|
|
// characters, since that is what GetAdvanceWidth expects.
|
|
Range runRange = ConvertOriginalToSkipped(it, mTextFrameContentOffset,
|
|
mTextFrameContentLength);
|
|
|
|
// Get the offset/length of the whole nsTextFrame.
|
|
uint32_t frameOffset = mFrame->GetContentOffset();
|
|
uint32_t frameLength = mFrame->GetContentLength();
|
|
|
|
// Trim the whole-nsTextFrame offset/length to remove any leading/trailing
|
|
// white space, as the nsTextFrame when painting does not include them when
|
|
// interpreting clip edges.
|
|
nsTextFrame::TrimmedOffsets trimmedOffsets =
|
|
mFrame->GetTrimmedOffsets(mFrame->GetContent()->GetText(), true);
|
|
TrimOffsets(frameOffset, frameLength, trimmedOffsets);
|
|
|
|
// Convert the trimmed whole-nsTextFrame offset/length into skipped
|
|
// characters.
|
|
Range frameRange = ConvertOriginalToSkipped(it, frameOffset, frameLength);
|
|
|
|
// Measure the advance width in the text run between the start of
|
|
// frame's content and the start of the rendered run's content,
|
|
nscoord startEdge = textRun->
|
|
GetAdvanceWidth(Range(frameRange.start, runRange.start), nullptr);
|
|
|
|
// and between the end of the rendered run's content and the end
|
|
// of the frame's content.
|
|
nscoord endEdge = textRun->
|
|
GetAdvanceWidth(Range(runRange.end, frameRange.end), nullptr);
|
|
|
|
if (textRun->IsRightToLeft()) {
|
|
aVisIStartEdge = endEdge;
|
|
aVisIEndEdge = startEdge;
|
|
} else {
|
|
aVisIStartEdge = startEdge;
|
|
aVisIEndEdge = endEdge;
|
|
}
|
|
}
|
|
|
|
nscoord
|
|
TextRenderedRun::GetAdvanceWidth() const
|
|
{
|
|
gfxSkipCharsIterator it = mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
Range range = ConvertOriginalToSkipped(it, mTextFrameContentOffset,
|
|
mTextFrameContentLength);
|
|
|
|
return textRun->GetAdvanceWidth(range, nullptr);
|
|
}
|
|
|
|
int32_t
|
|
TextRenderedRun::GetCharNumAtPosition(nsPresContext* aContext,
|
|
const gfxPoint& aPoint) const
|
|
{
|
|
if (mTextFrameContentLength == 0) {
|
|
return -1;
|
|
}
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
// Convert the point from user space into run user space, and take
|
|
// into account any mFontSizeScaleFactor.
|
|
gfxMatrix m = GetTransformFromRunUserSpaceToUserSpace(aContext);
|
|
if (!m.Invert()) {
|
|
return -1;
|
|
}
|
|
gfxPoint p = m.TransformPoint(aPoint) / cssPxPerDevPx * mFontSizeScaleFactor;
|
|
|
|
// First check that the point lies vertically between the top and bottom
|
|
// edges of the text.
|
|
gfxFloat ascent, descent;
|
|
GetAscentAndDescentInAppUnits(mFrame, ascent, descent);
|
|
|
|
WritingMode writingMode = mFrame->GetWritingMode();
|
|
if (writingMode.IsVertical()) {
|
|
gfxFloat leftEdge =
|
|
mFrame->GetLogicalBaseline(writingMode) -
|
|
(writingMode.IsVerticalRL() ? ascent : descent);
|
|
gfxFloat rightEdge = leftEdge + ascent + descent;
|
|
if (p.x < aContext->AppUnitsToGfxUnits(leftEdge) ||
|
|
p.x > aContext->AppUnitsToGfxUnits(rightEdge)) {
|
|
return -1;
|
|
}
|
|
} else {
|
|
gfxFloat topEdge = mFrame->GetLogicalBaseline(writingMode) - ascent;
|
|
gfxFloat bottomEdge = topEdge + ascent + descent;
|
|
if (p.y < aContext->AppUnitsToGfxUnits(topEdge) ||
|
|
p.y > aContext->AppUnitsToGfxUnits(bottomEdge)) {
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
gfxSkipCharsIterator it = mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
// Next check that the point lies horizontally within the left and right
|
|
// edges of the text.
|
|
Range range = ConvertOriginalToSkipped(it, mTextFrameContentOffset,
|
|
mTextFrameContentLength);
|
|
gfxFloat runAdvance =
|
|
aContext->AppUnitsToGfxUnits(textRun->GetAdvanceWidth(range, nullptr));
|
|
|
|
gfxFloat pos = writingMode.IsVertical() ? p.y : p.x;
|
|
if (pos < 0 || pos >= runAdvance) {
|
|
return -1;
|
|
}
|
|
|
|
// Finally, measure progressively smaller portions of the rendered run to
|
|
// find which glyph it lies within. This will need to change once we
|
|
// support letter-spacing and word-spacing.
|
|
bool rtl = textRun->IsRightToLeft();
|
|
for (int32_t i = mTextFrameContentLength - 1; i >= 0; i--) {
|
|
range = ConvertOriginalToSkipped(it, mTextFrameContentOffset, i);
|
|
gfxFloat advance =
|
|
aContext->AppUnitsToGfxUnits(textRun->GetAdvanceWidth(range, nullptr));
|
|
if ((rtl && pos < runAdvance - advance) ||
|
|
(!rtl && pos >= advance)) {
|
|
return i;
|
|
}
|
|
}
|
|
return -1;
|
|
}
|
|
|
|
// ----------------------------------------------------------------------------
|
|
// TextNodeIterator
|
|
|
|
enum SubtreePosition
|
|
{
|
|
eBeforeSubtree,
|
|
eWithinSubtree,
|
|
eAfterSubtree
|
|
};
|
|
|
|
/**
|
|
* An iterator class for nsTextNodes that are descendants of a given node, the
|
|
* root. Nodes are iterated in document order. An optional subtree can be
|
|
* specified, in which case the iterator will track whether the current state of
|
|
* the traversal over the tree is within that subtree or is past that subtree.
|
|
*/
|
|
class TextNodeIterator
|
|
{
|
|
public:
|
|
/**
|
|
* Constructs a TextNodeIterator with the specified root node and optional
|
|
* subtree.
|
|
*/
|
|
explicit TextNodeIterator(nsIContent* aRoot, nsIContent* aSubtree = nullptr)
|
|
: mRoot(aRoot),
|
|
mSubtree(aSubtree == aRoot ? nullptr : aSubtree),
|
|
mCurrent(aRoot),
|
|
mSubtreePosition(mSubtree ? eBeforeSubtree : eWithinSubtree)
|
|
{
|
|
NS_ASSERTION(aRoot, "expected non-null root");
|
|
if (!aRoot->IsText()) {
|
|
Next();
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Returns the current nsTextNode, or null if the iterator has finished.
|
|
*/
|
|
nsTextNode* Current() const
|
|
{
|
|
return static_cast<nsTextNode*>(mCurrent);
|
|
}
|
|
|
|
/**
|
|
* Advances to the next nsTextNode and returns it, or null if the end of
|
|
* iteration has been reached.
|
|
*/
|
|
nsTextNode* Next();
|
|
|
|
/**
|
|
* Returns whether the iterator is currently within the subtree rooted
|
|
* at mSubtree. Returns true if we are not tracking a subtree (we consider
|
|
* that we're always within the subtree).
|
|
*/
|
|
bool IsWithinSubtree() const
|
|
{
|
|
return mSubtreePosition == eWithinSubtree;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the iterator is past the subtree rooted at mSubtree.
|
|
* Returns false if we are not tracking a subtree.
|
|
*/
|
|
bool IsAfterSubtree() const
|
|
{
|
|
return mSubtreePosition == eAfterSubtree;
|
|
}
|
|
|
|
private:
|
|
/**
|
|
* The root under which all nsTextNodes will be iterated over.
|
|
*/
|
|
nsIContent* mRoot;
|
|
|
|
/**
|
|
* The node rooting the subtree to track.
|
|
*/
|
|
nsIContent* mSubtree;
|
|
|
|
/**
|
|
* The current node during iteration.
|
|
*/
|
|
nsIContent* mCurrent;
|
|
|
|
/**
|
|
* The current iterator position relative to mSubtree.
|
|
*/
|
|
SubtreePosition mSubtreePosition;
|
|
};
|
|
|
|
nsTextNode*
|
|
TextNodeIterator::Next()
|
|
{
|
|
// Starting from mCurrent, we do a non-recursive traversal to the next
|
|
// nsTextNode beneath mRoot, updating mSubtreePosition appropriately if we
|
|
// encounter mSubtree.
|
|
if (mCurrent) {
|
|
do {
|
|
nsIContent* next = IsTextContentElement(mCurrent) ?
|
|
mCurrent->GetFirstChild() :
|
|
nullptr;
|
|
if (next) {
|
|
mCurrent = next;
|
|
if (mCurrent == mSubtree) {
|
|
mSubtreePosition = eWithinSubtree;
|
|
}
|
|
} else {
|
|
for (;;) {
|
|
if (mCurrent == mRoot) {
|
|
mCurrent = nullptr;
|
|
break;
|
|
}
|
|
if (mCurrent == mSubtree) {
|
|
mSubtreePosition = eAfterSubtree;
|
|
}
|
|
next = mCurrent->GetNextSibling();
|
|
if (next) {
|
|
mCurrent = next;
|
|
if (mCurrent == mSubtree) {
|
|
mSubtreePosition = eWithinSubtree;
|
|
}
|
|
break;
|
|
}
|
|
if (mCurrent == mSubtree) {
|
|
mSubtreePosition = eAfterSubtree;
|
|
}
|
|
mCurrent = mCurrent->GetParent();
|
|
}
|
|
}
|
|
} while (mCurrent && !mCurrent->IsText());
|
|
}
|
|
|
|
return static_cast<nsTextNode*>(mCurrent);
|
|
}
|
|
|
|
// ----------------------------------------------------------------------------
|
|
// TextNodeCorrespondenceRecorder
|
|
|
|
/**
|
|
* TextNodeCorrespondence is used as the value of a frame property that
|
|
* is stored on all its descendant nsTextFrames. It stores the number of DOM
|
|
* characters between it and the previous nsTextFrame that did not have an
|
|
* nsTextFrame created for them, due to either not being in a correctly
|
|
* parented text content element, or because they were display:none.
|
|
* These are called "undisplayed characters".
|
|
*
|
|
* See also TextNodeCorrespondenceRecorder below, which is what sets the
|
|
* frame property.
|
|
*/
|
|
struct TextNodeCorrespondence
|
|
{
|
|
explicit TextNodeCorrespondence(uint32_t aUndisplayedCharacters)
|
|
: mUndisplayedCharacters(aUndisplayedCharacters)
|
|
{
|
|
}
|
|
|
|
uint32_t mUndisplayedCharacters;
|
|
};
|
|
|
|
NS_DECLARE_FRAME_PROPERTY_DELETABLE(TextNodeCorrespondenceProperty,
|
|
TextNodeCorrespondence)
|
|
|
|
/**
|
|
* Returns the number of undisplayed characters before the specified
|
|
* nsTextFrame.
|
|
*/
|
|
static uint32_t
|
|
GetUndisplayedCharactersBeforeFrame(nsTextFrame* aFrame)
|
|
{
|
|
void* value = aFrame->GetProperty(TextNodeCorrespondenceProperty());
|
|
TextNodeCorrespondence* correspondence =
|
|
static_cast<TextNodeCorrespondence*>(value);
|
|
if (!correspondence) {
|
|
// FIXME bug 903785
|
|
NS_ERROR("expected a TextNodeCorrespondenceProperty on nsTextFrame "
|
|
"used for SVG text");
|
|
return 0;
|
|
}
|
|
return correspondence->mUndisplayedCharacters;
|
|
}
|
|
|
|
/**
|
|
* Traverses the nsTextFrames for an SVGTextFrame and records a
|
|
* TextNodeCorrespondenceProperty on each for the number of undisplayed DOM
|
|
* characters between each frame. This is done by iterating simultaneously
|
|
* over the nsTextNodes and nsTextFrames and noting when nsTextNodes (or
|
|
* parts of them) are skipped when finding the next nsTextFrame.
|
|
*/
|
|
class TextNodeCorrespondenceRecorder
|
|
{
|
|
public:
|
|
/**
|
|
* Entry point for the TextNodeCorrespondenceProperty recording.
|
|
*/
|
|
static void RecordCorrespondence(SVGTextFrame* aRoot);
|
|
|
|
private:
|
|
explicit TextNodeCorrespondenceRecorder(SVGTextFrame* aRoot)
|
|
: mNodeIterator(aRoot->GetContent()),
|
|
mPreviousNode(nullptr),
|
|
mNodeCharIndex(0)
|
|
{
|
|
}
|
|
|
|
void Record(SVGTextFrame* aRoot);
|
|
void TraverseAndRecord(nsIFrame* aFrame);
|
|
|
|
/**
|
|
* Returns the next non-empty nsTextNode.
|
|
*/
|
|
nsTextNode* NextNode();
|
|
|
|
/**
|
|
* The iterator over the nsTextNodes that we use as we simultaneously
|
|
* iterate over the nsTextFrames.
|
|
*/
|
|
TextNodeIterator mNodeIterator;
|
|
|
|
/**
|
|
* The previous nsTextNode we iterated over.
|
|
*/
|
|
nsTextNode* mPreviousNode;
|
|
|
|
/**
|
|
* The index into the current nsTextNode's character content.
|
|
*/
|
|
uint32_t mNodeCharIndex;
|
|
};
|
|
|
|
/* static */ void
|
|
TextNodeCorrespondenceRecorder::RecordCorrespondence(SVGTextFrame* aRoot)
|
|
{
|
|
if (aRoot->GetStateBits() & NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY) {
|
|
// Resolve bidi so that continuation frames are created if necessary:
|
|
aRoot->MaybeResolveBidiForAnonymousBlockChild();
|
|
TextNodeCorrespondenceRecorder recorder(aRoot);
|
|
recorder.Record(aRoot);
|
|
aRoot->RemoveStateBits(NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY);
|
|
}
|
|
}
|
|
|
|
void
|
|
TextNodeCorrespondenceRecorder::Record(SVGTextFrame* aRoot)
|
|
{
|
|
if (!mNodeIterator.Current()) {
|
|
// If there are no nsTextNodes then there is nothing to do.
|
|
return;
|
|
}
|
|
|
|
// Traverse over all the nsTextFrames and record the number of undisplayed
|
|
// characters.
|
|
TraverseAndRecord(aRoot);
|
|
|
|
// Find how many undisplayed characters there are after the final nsTextFrame.
|
|
uint32_t undisplayed = 0;
|
|
if (mNodeIterator.Current()) {
|
|
if (mPreviousNode && mPreviousNode->TextLength() != mNodeCharIndex) {
|
|
// The last nsTextFrame ended part way through an nsTextNode. The
|
|
// remaining characters count as undisplayed.
|
|
NS_ASSERTION(mNodeCharIndex < mPreviousNode->TextLength(),
|
|
"incorrect tracking of undisplayed characters in "
|
|
"text nodes");
|
|
undisplayed += mPreviousNode->TextLength() - mNodeCharIndex;
|
|
}
|
|
// All the remaining nsTextNodes that we iterate must also be undisplayed.
|
|
for (nsTextNode* textNode = mNodeIterator.Current();
|
|
textNode;
|
|
textNode = NextNode()) {
|
|
undisplayed += textNode->TextLength();
|
|
}
|
|
}
|
|
|
|
// Record the trailing number of undisplayed characters on the
|
|
// SVGTextFrame.
|
|
aRoot->mTrailingUndisplayedCharacters = undisplayed;
|
|
}
|
|
|
|
nsTextNode*
|
|
TextNodeCorrespondenceRecorder::NextNode()
|
|
{
|
|
mPreviousNode = mNodeIterator.Current();
|
|
nsTextNode* next;
|
|
do {
|
|
next = mNodeIterator.Next();
|
|
} while (next && next->TextLength() == 0);
|
|
return next;
|
|
}
|
|
|
|
void
|
|
TextNodeCorrespondenceRecorder::TraverseAndRecord(nsIFrame* aFrame)
|
|
{
|
|
// Recursively iterate over the frame tree, for frames that correspond
|
|
// to text content elements.
|
|
if (IsTextContentElement(aFrame->GetContent())) {
|
|
for (nsIFrame* f : aFrame->PrincipalChildList()) {
|
|
TraverseAndRecord(f);
|
|
}
|
|
return;
|
|
}
|
|
|
|
nsTextFrame* frame; // The current text frame.
|
|
nsTextNode* node; // The text node for the current text frame.
|
|
if (!GetNonEmptyTextFrameAndNode(aFrame, frame, node)) {
|
|
// If this isn't an nsTextFrame, or is empty, nothing to do.
|
|
return;
|
|
}
|
|
|
|
NS_ASSERTION(frame->GetContentOffset() >= 0,
|
|
"don't know how to handle negative content indexes");
|
|
|
|
uint32_t undisplayed = 0;
|
|
if (!mPreviousNode) {
|
|
// Must be the very first text frame.
|
|
NS_ASSERTION(mNodeCharIndex == 0, "incorrect tracking of undisplayed "
|
|
"characters in text nodes");
|
|
if (!mNodeIterator.Current()) {
|
|
MOZ_ASSERT_UNREACHABLE("incorrect tracking of correspondence between "
|
|
"text frames and text nodes");
|
|
} else {
|
|
// Each whole nsTextNode we find before we get to the text node for the
|
|
// first text frame must be undisplayed.
|
|
while (mNodeIterator.Current() != node) {
|
|
undisplayed += mNodeIterator.Current()->TextLength();
|
|
NextNode();
|
|
}
|
|
// If the first text frame starts at a non-zero content offset, then those
|
|
// earlier characters are also undisplayed.
|
|
undisplayed += frame->GetContentOffset();
|
|
NextNode();
|
|
}
|
|
} else if (mPreviousNode == node) {
|
|
// Same text node as last time.
|
|
if (static_cast<uint32_t>(frame->GetContentOffset()) != mNodeCharIndex) {
|
|
// We have some characters in the middle of the text node
|
|
// that are undisplayed.
|
|
NS_ASSERTION(mNodeCharIndex <
|
|
static_cast<uint32_t>(frame->GetContentOffset()),
|
|
"incorrect tracking of undisplayed characters in "
|
|
"text nodes");
|
|
undisplayed = frame->GetContentOffset() - mNodeCharIndex;
|
|
}
|
|
} else {
|
|
// Different text node from last time.
|
|
if (mPreviousNode->TextLength() != mNodeCharIndex) {
|
|
NS_ASSERTION(mNodeCharIndex < mPreviousNode->TextLength(),
|
|
"incorrect tracking of undisplayed characters in "
|
|
"text nodes");
|
|
// Any trailing characters at the end of the previous nsTextNode are
|
|
// undisplayed.
|
|
undisplayed = mPreviousNode->TextLength() - mNodeCharIndex;
|
|
}
|
|
// Each whole nsTextNode we find before we get to the text node for
|
|
// the current text frame must be undisplayed.
|
|
while (mNodeIterator.Current() != node) {
|
|
undisplayed += mNodeIterator.Current()->TextLength();
|
|
NextNode();
|
|
}
|
|
// If the current text frame starts at a non-zero content offset, then those
|
|
// earlier characters are also undisplayed.
|
|
undisplayed += frame->GetContentOffset();
|
|
NextNode();
|
|
}
|
|
|
|
// Set the frame property.
|
|
frame->SetProperty(TextNodeCorrespondenceProperty(),
|
|
new TextNodeCorrespondence(undisplayed));
|
|
|
|
// Remember how far into the current nsTextNode we are.
|
|
mNodeCharIndex = frame->GetContentEnd();
|
|
}
|
|
|
|
// ----------------------------------------------------------------------------
|
|
// TextFrameIterator
|
|
|
|
/**
|
|
* An iterator class for nsTextFrames that are descendants of an
|
|
* SVGTextFrame. The iterator can optionally track whether the
|
|
* current nsTextFrame is for a descendant of, or past, a given subtree
|
|
* content node or frame. (This functionality is used for example by the SVG
|
|
* DOM text methods to get only the nsTextFrames for a particular <tspan>.)
|
|
*
|
|
* TextFrameIterator also tracks and exposes other information about the
|
|
* current nsTextFrame:
|
|
*
|
|
* * how many undisplayed characters came just before it
|
|
* * its position (in app units) relative to the SVGTextFrame's anonymous
|
|
* block frame
|
|
* * what nsInlineFrame corresponding to a <textPath> element it is a
|
|
* descendant of
|
|
* * what computed dominant-baseline value applies to it
|
|
*
|
|
* Note that any text frames that are empty -- whose ContentLength() is 0 --
|
|
* will be skipped over.
|
|
*/
|
|
class TextFrameIterator
|
|
{
|
|
public:
|
|
/**
|
|
* Constructs a TextFrameIterator for the specified SVGTextFrame
|
|
* with an optional frame subtree to restrict iterated text frames to.
|
|
*/
|
|
explicit TextFrameIterator(SVGTextFrame* aRoot, const nsIFrame* aSubtree = nullptr)
|
|
: mRootFrame(aRoot),
|
|
mSubtree(aSubtree),
|
|
mCurrentFrame(aRoot),
|
|
mCurrentPosition(),
|
|
mSubtreePosition(mSubtree ? eBeforeSubtree : eWithinSubtree)
|
|
{
|
|
Init();
|
|
}
|
|
|
|
/**
|
|
* Constructs a TextFrameIterator for the specified SVGTextFrame
|
|
* with an optional frame content subtree to restrict iterated text frames to.
|
|
*/
|
|
TextFrameIterator(SVGTextFrame* aRoot, nsIContent* aSubtree)
|
|
: mRootFrame(aRoot),
|
|
mSubtree(aRoot && aSubtree && aSubtree != aRoot->GetContent() ?
|
|
aSubtree->GetPrimaryFrame() :
|
|
nullptr),
|
|
mCurrentFrame(aRoot),
|
|
mCurrentPosition(),
|
|
mSubtreePosition(mSubtree ? eBeforeSubtree : eWithinSubtree)
|
|
{
|
|
Init();
|
|
}
|
|
|
|
/**
|
|
* Returns the root SVGTextFrame this TextFrameIterator is iterating over.
|
|
*/
|
|
SVGTextFrame* Root() const
|
|
{
|
|
return mRootFrame;
|
|
}
|
|
|
|
/**
|
|
* Returns the current nsTextFrame.
|
|
*/
|
|
nsTextFrame* Current() const
|
|
{
|
|
return do_QueryFrame(mCurrentFrame);
|
|
}
|
|
|
|
/**
|
|
* Returns the number of undisplayed characters in the DOM just before the
|
|
* current frame.
|
|
*/
|
|
uint32_t UndisplayedCharacters() const;
|
|
|
|
/**
|
|
* Returns the current frame's position, in app units, relative to the
|
|
* root SVGTextFrame's anonymous block frame.
|
|
*/
|
|
nsPoint Position() const
|
|
{
|
|
return mCurrentPosition;
|
|
}
|
|
|
|
/**
|
|
* Advances to the next nsTextFrame and returns it.
|
|
*/
|
|
nsTextFrame* Next();
|
|
|
|
/**
|
|
* Returns whether the iterator is within the subtree.
|
|
*/
|
|
bool IsWithinSubtree() const
|
|
{
|
|
return mSubtreePosition == eWithinSubtree;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the iterator is past the subtree.
|
|
*/
|
|
bool IsAfterSubtree() const
|
|
{
|
|
return mSubtreePosition == eAfterSubtree;
|
|
}
|
|
|
|
/**
|
|
* Returns the frame corresponding to the <textPath> element, if we
|
|
* are inside one.
|
|
*/
|
|
nsIFrame* TextPathFrame() const
|
|
{
|
|
return mTextPathFrames.IsEmpty() ?
|
|
nullptr :
|
|
mTextPathFrames.ElementAt(mTextPathFrames.Length() - 1);
|
|
}
|
|
|
|
/**
|
|
* Returns the current frame's computed dominant-baseline value.
|
|
*/
|
|
uint8_t DominantBaseline() const
|
|
{
|
|
return mBaselines.ElementAt(mBaselines.Length() - 1);
|
|
}
|
|
|
|
/**
|
|
* Finishes the iterator.
|
|
*/
|
|
void Close()
|
|
{
|
|
mCurrentFrame = nullptr;
|
|
}
|
|
|
|
private:
|
|
/**
|
|
* Initializes the iterator and advances to the first item.
|
|
*/
|
|
void Init()
|
|
{
|
|
if (!mRootFrame) {
|
|
return;
|
|
}
|
|
|
|
mBaselines.AppendElement(mRootFrame->StyleSVGReset()->mDominantBaseline);
|
|
Next();
|
|
}
|
|
|
|
/**
|
|
* Pushes the specified frame's computed dominant-baseline value.
|
|
* If the value of the property is "auto", then the parent frame's
|
|
* computed value is used.
|
|
*/
|
|
void PushBaseline(nsIFrame* aNextFrame);
|
|
|
|
/**
|
|
* Pops the current dominant-baseline off the stack.
|
|
*/
|
|
void PopBaseline();
|
|
|
|
/**
|
|
* The root frame we are iterating through.
|
|
*/
|
|
SVGTextFrame* mRootFrame;
|
|
|
|
/**
|
|
* The frame for the subtree we are also interested in tracking.
|
|
*/
|
|
const nsIFrame* mSubtree;
|
|
|
|
/**
|
|
* The current value of the iterator.
|
|
*/
|
|
nsIFrame* mCurrentFrame;
|
|
|
|
/**
|
|
* The position, in app units, of the current frame relative to mRootFrame.
|
|
*/
|
|
nsPoint mCurrentPosition;
|
|
|
|
/**
|
|
* Stack of frames corresponding to <textPath> elements that are in scope
|
|
* for the current frame.
|
|
*/
|
|
AutoTArray<nsIFrame*, 1> mTextPathFrames;
|
|
|
|
/**
|
|
* Stack of dominant-baseline values to record as we traverse through the
|
|
* frame tree.
|
|
*/
|
|
AutoTArray<uint8_t, 8> mBaselines;
|
|
|
|
/**
|
|
* The iterator's current position relative to mSubtree.
|
|
*/
|
|
SubtreePosition mSubtreePosition;
|
|
};
|
|
|
|
uint32_t
|
|
TextFrameIterator::UndisplayedCharacters() const
|
|
{
|
|
MOZ_ASSERT(!(mRootFrame->GetStateBits() &
|
|
NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY),
|
|
"Text correspondence must be up to date");
|
|
|
|
if (!mCurrentFrame) {
|
|
return mRootFrame->mTrailingUndisplayedCharacters;
|
|
}
|
|
|
|
nsTextFrame* frame = do_QueryFrame(mCurrentFrame);
|
|
return GetUndisplayedCharactersBeforeFrame(frame);
|
|
}
|
|
|
|
nsTextFrame*
|
|
TextFrameIterator::Next()
|
|
{
|
|
// Starting from mCurrentFrame, we do a non-recursive traversal to the next
|
|
// nsTextFrame beneath mRoot, updating mSubtreePosition appropriately if we
|
|
// encounter mSubtree.
|
|
if (mCurrentFrame) {
|
|
do {
|
|
nsIFrame* next = IsTextContentElement(mCurrentFrame->GetContent()) ?
|
|
mCurrentFrame->PrincipalChildList().FirstChild() :
|
|
nullptr;
|
|
if (next) {
|
|
// Descend into this frame, and accumulate its position.
|
|
mCurrentPosition += next->GetPosition();
|
|
if (next->GetContent()->IsSVGElement(nsGkAtoms::textPath)) {
|
|
// Record this <textPath> frame.
|
|
mTextPathFrames.AppendElement(next);
|
|
}
|
|
// Record the frame's baseline.
|
|
PushBaseline(next);
|
|
mCurrentFrame = next;
|
|
if (mCurrentFrame == mSubtree) {
|
|
// If the current frame is mSubtree, we have now moved into it.
|
|
mSubtreePosition = eWithinSubtree;
|
|
}
|
|
} else {
|
|
for (;;) {
|
|
// We want to move past the current frame.
|
|
if (mCurrentFrame == mRootFrame) {
|
|
// If we've reached the root frame, we're finished.
|
|
mCurrentFrame = nullptr;
|
|
break;
|
|
}
|
|
// Remove the current frame's position.
|
|
mCurrentPosition -= mCurrentFrame->GetPosition();
|
|
if (mCurrentFrame->GetContent()->IsSVGElement(nsGkAtoms::textPath)) {
|
|
// Pop off the <textPath> frame if this is a <textPath>.
|
|
mTextPathFrames.TruncateLength(mTextPathFrames.Length() - 1);
|
|
}
|
|
// Pop off the current baseline.
|
|
PopBaseline();
|
|
if (mCurrentFrame == mSubtree) {
|
|
// If this was mSubtree, we have now moved past it.
|
|
mSubtreePosition = eAfterSubtree;
|
|
}
|
|
next = mCurrentFrame->GetNextSibling();
|
|
if (next) {
|
|
// Moving to the next sibling.
|
|
mCurrentPosition += next->GetPosition();
|
|
if (next->GetContent()->IsSVGElement(nsGkAtoms::textPath)) {
|
|
// Record this <textPath> frame.
|
|
mTextPathFrames.AppendElement(next);
|
|
}
|
|
// Record the frame's baseline.
|
|
PushBaseline(next);
|
|
mCurrentFrame = next;
|
|
if (mCurrentFrame == mSubtree) {
|
|
// If the current frame is mSubtree, we have now moved into it.
|
|
mSubtreePosition = eWithinSubtree;
|
|
}
|
|
break;
|
|
}
|
|
if (mCurrentFrame == mSubtree) {
|
|
// If there is no next sibling frame, and the current frame is
|
|
// mSubtree, we have now moved past it.
|
|
mSubtreePosition = eAfterSubtree;
|
|
}
|
|
// Ascend out of this frame.
|
|
mCurrentFrame = mCurrentFrame->GetParent();
|
|
}
|
|
}
|
|
} while (mCurrentFrame &&
|
|
!IsNonEmptyTextFrame(mCurrentFrame));
|
|
}
|
|
|
|
return Current();
|
|
}
|
|
|
|
void
|
|
TextFrameIterator::PushBaseline(nsIFrame* aNextFrame)
|
|
{
|
|
uint8_t baseline = aNextFrame->StyleSVGReset()->mDominantBaseline;
|
|
if (baseline == NS_STYLE_DOMINANT_BASELINE_AUTO) {
|
|
baseline = mBaselines.LastElement();
|
|
}
|
|
mBaselines.AppendElement(baseline);
|
|
}
|
|
|
|
void
|
|
TextFrameIterator::PopBaseline()
|
|
{
|
|
NS_ASSERTION(!mBaselines.IsEmpty(), "popped too many baselines");
|
|
mBaselines.TruncateLength(mBaselines.Length() - 1);
|
|
}
|
|
|
|
// -----------------------------------------------------------------------------
|
|
// TextRenderedRunIterator
|
|
|
|
/**
|
|
* Iterator for TextRenderedRun objects for the SVGTextFrame.
|
|
*/
|
|
class TextRenderedRunIterator
|
|
{
|
|
public:
|
|
/**
|
|
* Values for the aFilter argument of the constructor, to indicate which frames
|
|
* we should be limited to iterating TextRenderedRun objects for.
|
|
*/
|
|
enum RenderedRunFilter {
|
|
// Iterate TextRenderedRuns for all nsTextFrames.
|
|
eAllFrames,
|
|
// Iterate only TextRenderedRuns for nsTextFrames that are
|
|
// visibility:visible.
|
|
eVisibleFrames
|
|
};
|
|
|
|
/**
|
|
* Constructs a TextRenderedRunIterator with an optional frame subtree to
|
|
* restrict iterated rendered runs to.
|
|
*
|
|
* @param aSVGTextFrame The SVGTextFrame whose rendered runs to iterate
|
|
* through.
|
|
* @param aFilter Indicates whether to iterate rendered runs for non-visible
|
|
* nsTextFrames.
|
|
* @param aSubtree An optional frame subtree to restrict iterated rendered
|
|
* runs to.
|
|
*/
|
|
explicit TextRenderedRunIterator(SVGTextFrame* aSVGTextFrame,
|
|
RenderedRunFilter aFilter = eAllFrames,
|
|
const nsIFrame* aSubtree = nullptr)
|
|
: mFrameIterator(FrameIfAnonymousChildReflowed(aSVGTextFrame), aSubtree),
|
|
mFilter(aFilter),
|
|
mTextElementCharIndex(0),
|
|
mFrameStartTextElementCharIndex(0),
|
|
mFontSizeScaleFactor(aSVGTextFrame->mFontSizeScaleFactor),
|
|
mCurrent(First())
|
|
{
|
|
}
|
|
|
|
/**
|
|
* Constructs a TextRenderedRunIterator with a content subtree to restrict
|
|
* iterated rendered runs to.
|
|
*
|
|
* @param aSVGTextFrame The SVGTextFrame whose rendered runs to iterate
|
|
* through.
|
|
* @param aFilter Indicates whether to iterate rendered runs for non-visible
|
|
* nsTextFrames.
|
|
* @param aSubtree A content subtree to restrict iterated rendered runs to.
|
|
*/
|
|
TextRenderedRunIterator(SVGTextFrame* aSVGTextFrame,
|
|
RenderedRunFilter aFilter,
|
|
nsIContent* aSubtree)
|
|
: mFrameIterator(FrameIfAnonymousChildReflowed(aSVGTextFrame), aSubtree),
|
|
mFilter(aFilter),
|
|
mTextElementCharIndex(0),
|
|
mFrameStartTextElementCharIndex(0),
|
|
mFontSizeScaleFactor(aSVGTextFrame->mFontSizeScaleFactor),
|
|
mCurrent(First())
|
|
{
|
|
}
|
|
|
|
/**
|
|
* Returns the current TextRenderedRun.
|
|
*/
|
|
TextRenderedRun Current() const
|
|
{
|
|
return mCurrent;
|
|
}
|
|
|
|
/**
|
|
* Advances to the next TextRenderedRun and returns it.
|
|
*/
|
|
TextRenderedRun Next();
|
|
|
|
private:
|
|
/**
|
|
* Returns the root SVGTextFrame this iterator is for.
|
|
*/
|
|
SVGTextFrame* Root() const
|
|
{
|
|
return mFrameIterator.Root();
|
|
}
|
|
|
|
/**
|
|
* Advances to the first TextRenderedRun and returns it.
|
|
*/
|
|
TextRenderedRun First();
|
|
|
|
/**
|
|
* The frame iterator to use.
|
|
*/
|
|
TextFrameIterator mFrameIterator;
|
|
|
|
/**
|
|
* The filter indicating which TextRenderedRuns to return.
|
|
*/
|
|
RenderedRunFilter mFilter;
|
|
|
|
/**
|
|
* The character index across the entire <text> element we are currently
|
|
* up to.
|
|
*/
|
|
uint32_t mTextElementCharIndex;
|
|
|
|
/**
|
|
* The character index across the entire <text> for the start of the current
|
|
* frame.
|
|
*/
|
|
uint32_t mFrameStartTextElementCharIndex;
|
|
|
|
/**
|
|
* The font-size scale factor we used when constructing the nsTextFrames.
|
|
*/
|
|
double mFontSizeScaleFactor;
|
|
|
|
/**
|
|
* The current TextRenderedRun.
|
|
*/
|
|
TextRenderedRun mCurrent;
|
|
};
|
|
|
|
TextRenderedRun
|
|
TextRenderedRunIterator::Next()
|
|
{
|
|
if (!mFrameIterator.Current()) {
|
|
// If there are no more frames, then there are no more rendered runs to
|
|
// return.
|
|
mCurrent = TextRenderedRun();
|
|
return mCurrent;
|
|
}
|
|
|
|
// The values we will use to initialize the TextRenderedRun with.
|
|
nsTextFrame* frame;
|
|
gfxPoint pt;
|
|
double rotate;
|
|
nscoord baseline;
|
|
uint32_t offset, length;
|
|
uint32_t charIndex;
|
|
|
|
// We loop, because we want to skip over rendered runs that either aren't
|
|
// within our subtree of interest, because they don't match the filter,
|
|
// or because they are hidden due to having fallen off the end of a
|
|
// <textPath>.
|
|
for (;;) {
|
|
if (mFrameIterator.IsAfterSubtree()) {
|
|
mCurrent = TextRenderedRun();
|
|
return mCurrent;
|
|
}
|
|
|
|
frame = mFrameIterator.Current();
|
|
|
|
charIndex = mTextElementCharIndex;
|
|
|
|
// Find the end of the rendered run, by looking through the
|
|
// SVGTextFrame's positions array until we find one that is recorded
|
|
// as a run boundary.
|
|
uint32_t runStart, runEnd; // XXX Replace runStart with mTextElementCharIndex.
|
|
runStart = mTextElementCharIndex;
|
|
runEnd = runStart + 1;
|
|
while (runEnd < Root()->mPositions.Length() &&
|
|
!Root()->mPositions[runEnd].mRunBoundary) {
|
|
runEnd++;
|
|
}
|
|
|
|
// Convert the global run start/end indexes into an offset/length into the
|
|
// current frame's nsTextNode.
|
|
offset = frame->GetContentOffset() + runStart -
|
|
mFrameStartTextElementCharIndex;
|
|
length = runEnd - runStart;
|
|
|
|
// If the end of the frame's content comes before the run boundary we found
|
|
// in SVGTextFrame's position array, we need to shorten the rendered run.
|
|
uint32_t contentEnd = frame->GetContentEnd();
|
|
if (offset + length > contentEnd) {
|
|
length = contentEnd - offset;
|
|
}
|
|
|
|
NS_ASSERTION(offset >= uint32_t(frame->GetContentOffset()), "invalid offset");
|
|
NS_ASSERTION(offset + length <= contentEnd, "invalid offset or length");
|
|
|
|
// Get the frame's baseline position.
|
|
frame->EnsureTextRun(nsTextFrame::eInflated);
|
|
baseline = GetBaselinePosition(frame,
|
|
frame->GetTextRun(nsTextFrame::eInflated),
|
|
mFrameIterator.DominantBaseline(),
|
|
mFontSizeScaleFactor);
|
|
|
|
// Trim the offset/length to remove any leading/trailing white space.
|
|
uint32_t untrimmedOffset = offset;
|
|
uint32_t untrimmedLength = length;
|
|
nsTextFrame::TrimmedOffsets trimmedOffsets =
|
|
frame->GetTrimmedOffsets(frame->GetContent()->GetText(), true);
|
|
TrimOffsets(offset, length, trimmedOffsets);
|
|
charIndex += offset - untrimmedOffset;
|
|
|
|
// Get the position and rotation of the character that begins this
|
|
// rendered run.
|
|
pt = Root()->mPositions[charIndex].mPosition;
|
|
rotate = Root()->mPositions[charIndex].mAngle;
|
|
|
|
// Determine if we should skip this rendered run.
|
|
bool skip = !mFrameIterator.IsWithinSubtree() ||
|
|
Root()->mPositions[mTextElementCharIndex].mHidden;
|
|
if (mFilter == eVisibleFrames) {
|
|
skip = skip || !frame->StyleVisibility()->IsVisible();
|
|
}
|
|
|
|
// Update our global character index to move past the characters
|
|
// corresponding to this rendered run.
|
|
mTextElementCharIndex += untrimmedLength;
|
|
|
|
// If we have moved past the end of the current frame's content, we need to
|
|
// advance to the next frame.
|
|
if (offset + untrimmedLength >= contentEnd) {
|
|
mFrameIterator.Next();
|
|
mTextElementCharIndex += mFrameIterator.UndisplayedCharacters();
|
|
mFrameStartTextElementCharIndex = mTextElementCharIndex;
|
|
}
|
|
|
|
if (!mFrameIterator.Current()) {
|
|
if (skip) {
|
|
// That was the last frame, and we skipped this rendered run. So we
|
|
// have no rendered run to return.
|
|
mCurrent = TextRenderedRun();
|
|
return mCurrent;
|
|
}
|
|
break;
|
|
}
|
|
|
|
if (length && !skip) {
|
|
// Only return a rendered run if it didn't get collapsed away entirely
|
|
// (due to it being all white space) and if we don't want to skip it.
|
|
break;
|
|
}
|
|
}
|
|
|
|
mCurrent = TextRenderedRun(frame, pt, Root()->mLengthAdjustScaleFactor,
|
|
rotate, mFontSizeScaleFactor, baseline,
|
|
offset, length, charIndex);
|
|
return mCurrent;
|
|
}
|
|
|
|
TextRenderedRun
|
|
TextRenderedRunIterator::First()
|
|
{
|
|
if (!mFrameIterator.Current()) {
|
|
return TextRenderedRun();
|
|
}
|
|
|
|
if (Root()->mPositions.IsEmpty()) {
|
|
mFrameIterator.Close();
|
|
return TextRenderedRun();
|
|
}
|
|
|
|
// Get the character index for the start of this rendered run, by skipping
|
|
// any undisplayed characters.
|
|
mTextElementCharIndex = mFrameIterator.UndisplayedCharacters();
|
|
mFrameStartTextElementCharIndex = mTextElementCharIndex;
|
|
|
|
return Next();
|
|
}
|
|
|
|
// -----------------------------------------------------------------------------
|
|
// CharIterator
|
|
|
|
/**
|
|
* Iterator for characters within an SVGTextFrame.
|
|
*/
|
|
class CharIterator
|
|
{
|
|
typedef gfxTextRun::Range Range;
|
|
|
|
public:
|
|
/**
|
|
* Values for the aFilter argument of the constructor, to indicate which
|
|
* characters we should be iterating over.
|
|
*/
|
|
enum CharacterFilter {
|
|
// Iterate over all original characters from the DOM that are within valid
|
|
// text content elements.
|
|
eOriginal,
|
|
// Iterate only over characters that are addressable by the positioning
|
|
// attributes x="", y="", etc. This includes all characters after
|
|
// collapsing white space as required by the value of 'white-space'.
|
|
eAddressable,
|
|
// Iterate only over characters that are the first of clusters or ligature
|
|
// groups.
|
|
eClusterAndLigatureGroupStart,
|
|
// Iterate only over characters that are part of a cluster or ligature
|
|
// group but not the first character.
|
|
eClusterOrLigatureGroupMiddle
|
|
};
|
|
|
|
/**
|
|
* Constructs a CharIterator.
|
|
*
|
|
* @param aSVGTextFrame The SVGTextFrame whose characters to iterate
|
|
* through.
|
|
* @param aFilter Indicates which characters to iterate over.
|
|
* @param aSubtree A content subtree to track whether the current character
|
|
* is within.
|
|
*/
|
|
CharIterator(SVGTextFrame* aSVGTextFrame,
|
|
CharacterFilter aFilter,
|
|
nsIContent* aSubtree,
|
|
bool aPostReflow = true);
|
|
|
|
/**
|
|
* Returns whether the iterator is finished.
|
|
*/
|
|
bool AtEnd() const
|
|
{
|
|
return !mFrameIterator.Current();
|
|
}
|
|
|
|
/**
|
|
* Advances to the next matching character. Returns true if there was a
|
|
* character to advance to, and false otherwise.
|
|
*/
|
|
bool Next();
|
|
|
|
/**
|
|
* Advances ahead aCount matching characters. Returns true if there were
|
|
* enough characters to advance past, and false otherwise.
|
|
*/
|
|
bool Next(uint32_t aCount);
|
|
|
|
/**
|
|
* Advances ahead up to aCount matching characters.
|
|
*/
|
|
void NextWithinSubtree(uint32_t aCount);
|
|
|
|
/**
|
|
* Advances to the character with the specified index. The index is in the
|
|
* space of original characters (i.e., all DOM characters under the <text>
|
|
* that are within valid text content elements).
|
|
*/
|
|
bool AdvanceToCharacter(uint32_t aTextElementCharIndex);
|
|
|
|
/**
|
|
* Advances to the first matching character after the current nsTextFrame.
|
|
*/
|
|
bool AdvancePastCurrentFrame();
|
|
|
|
/**
|
|
* Advances to the first matching character after the frames within
|
|
* the current <textPath>.
|
|
*/
|
|
bool AdvancePastCurrentTextPathFrame();
|
|
|
|
/**
|
|
* Advances to the first matching character of the subtree. Returns true
|
|
* if we successfully advance to the subtree, or if we are already within
|
|
* the subtree. Returns false if we are past the subtree.
|
|
*/
|
|
bool AdvanceToSubtree();
|
|
|
|
/**
|
|
* Returns the nsTextFrame for the current character.
|
|
*/
|
|
nsTextFrame* TextFrame() const
|
|
{
|
|
return mFrameIterator.Current();
|
|
}
|
|
|
|
/**
|
|
* Returns whether the iterator is within the subtree.
|
|
*/
|
|
bool IsWithinSubtree() const
|
|
{
|
|
return mFrameIterator.IsWithinSubtree();
|
|
}
|
|
|
|
/**
|
|
* Returns whether the iterator is past the subtree.
|
|
*/
|
|
bool IsAfterSubtree() const
|
|
{
|
|
return mFrameIterator.IsAfterSubtree();
|
|
}
|
|
|
|
/**
|
|
* Returns whether the current character is a skipped character.
|
|
*/
|
|
bool IsOriginalCharSkipped() const
|
|
{
|
|
return mSkipCharsIterator.IsOriginalCharSkipped();
|
|
}
|
|
|
|
/**
|
|
* Returns whether the current character is the start of a cluster and
|
|
* ligature group.
|
|
*/
|
|
bool IsClusterAndLigatureGroupStart() const;
|
|
|
|
/**
|
|
* Returns whether the current character is trimmed away when painting,
|
|
* due to it being leading/trailing white space.
|
|
*/
|
|
bool IsOriginalCharTrimmed() const;
|
|
|
|
/**
|
|
* Returns whether the current character is unaddressable from the SVG glyph
|
|
* positioning attributes.
|
|
*/
|
|
bool IsOriginalCharUnaddressable() const
|
|
{
|
|
return IsOriginalCharSkipped() || IsOriginalCharTrimmed();
|
|
}
|
|
|
|
/**
|
|
* Returns the text run for the current character.
|
|
*/
|
|
gfxTextRun* TextRun() const
|
|
{
|
|
return mTextRun;
|
|
}
|
|
|
|
/**
|
|
* Returns the current character index.
|
|
*/
|
|
uint32_t TextElementCharIndex() const
|
|
{
|
|
return mTextElementCharIndex;
|
|
}
|
|
|
|
/**
|
|
* Returns the character index for the start of the cluster/ligature group it
|
|
* is part of.
|
|
*/
|
|
uint32_t GlyphStartTextElementCharIndex() const
|
|
{
|
|
return mGlyphStartTextElementCharIndex;
|
|
}
|
|
|
|
/**
|
|
* Returns the number of undisplayed characters between the beginning of
|
|
* the glyph and the current character.
|
|
*/
|
|
uint32_t GlyphUndisplayedCharacters() const
|
|
{
|
|
return mGlyphUndisplayedCharacters;
|
|
}
|
|
|
|
/**
|
|
* Gets the original character offsets within the nsTextNode for the
|
|
* cluster/ligature group the current character is a part of.
|
|
*
|
|
* @param aOriginalOffset The offset of the start of the cluster/ligature
|
|
* group (output).
|
|
* @param aOriginalLength The length of cluster/ligature group (output).
|
|
*/
|
|
void GetOriginalGlyphOffsets(uint32_t& aOriginalOffset,
|
|
uint32_t& aOriginalLength) const;
|
|
|
|
/**
|
|
* Gets the advance, in user units, of the glyph the current character is
|
|
* part of.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
*/
|
|
gfxFloat GetGlyphAdvance(nsPresContext* aContext) const;
|
|
|
|
/**
|
|
* Gets the advance, in user units, of the current character. If the
|
|
* character is a part of ligature, then the advance returned will be
|
|
* a fraction of the ligature glyph's advance.
|
|
*
|
|
* @param aContext The context to use for unit conversions.
|
|
*/
|
|
gfxFloat GetAdvance(nsPresContext* aContext) const;
|
|
|
|
/**
|
|
* Gets the specified partial advance of the glyph the current character is
|
|
* part of. The partial advance is measured from the first character
|
|
* corresponding to the glyph until the specified part length.
|
|
*
|
|
* The part length value does not include any undisplayed characters in the
|
|
* middle of the cluster/ligature group. For example, if you have:
|
|
*
|
|
* <text>f<tspan display="none">x</tspan>i</text>
|
|
*
|
|
* and the "f" and "i" are ligaturized, then calling GetGlyphPartialAdvance
|
|
* with aPartLength values will have the following results:
|
|
*
|
|
* 0 => 0
|
|
* 1 => adv("fi") / 2
|
|
* 2 => adv("fi")
|
|
*
|
|
* @param aPartLength The number of characters in the cluster/ligature group
|
|
* to measure.
|
|
* @param aContext The context to use for unit conversions.
|
|
*/
|
|
gfxFloat GetGlyphPartialAdvance(uint32_t aPartLength,
|
|
nsPresContext* aContext) const;
|
|
|
|
/**
|
|
* Returns the frame corresponding to the <textPath> that the current
|
|
* character is within.
|
|
*/
|
|
nsIFrame* TextPathFrame() const
|
|
{
|
|
return mFrameIterator.TextPathFrame();
|
|
}
|
|
|
|
private:
|
|
/**
|
|
* Advances to the next character without checking it against the filter.
|
|
* Returns true if there was a next character to advance to, or false
|
|
* otherwise.
|
|
*/
|
|
bool NextCharacter();
|
|
|
|
/**
|
|
* Returns whether the current character matches the filter.
|
|
*/
|
|
bool MatchesFilter() const;
|
|
|
|
/**
|
|
* If this is the start of a glyph, record it.
|
|
*/
|
|
void UpdateGlyphStartTextElementCharIndex() {
|
|
if (!IsOriginalCharSkipped() && IsClusterAndLigatureGroupStart()) {
|
|
mGlyphStartTextElementCharIndex = mTextElementCharIndex;
|
|
mGlyphUndisplayedCharacters = 0;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* The filter to use.
|
|
*/
|
|
CharacterFilter mFilter;
|
|
|
|
/**
|
|
* The iterator for text frames.
|
|
*/
|
|
TextFrameIterator mFrameIterator;
|
|
|
|
/**
|
|
* A gfxSkipCharsIterator for the text frame the current character is
|
|
* a part of.
|
|
*/
|
|
gfxSkipCharsIterator mSkipCharsIterator;
|
|
|
|
// Cache for information computed by IsOriginalCharTrimmed.
|
|
mutable nsTextFrame* mFrameForTrimCheck;
|
|
mutable uint32_t mTrimmedOffset;
|
|
mutable uint32_t mTrimmedLength;
|
|
|
|
/**
|
|
* The text run the current character is a part of.
|
|
*/
|
|
gfxTextRun* mTextRun;
|
|
|
|
/**
|
|
* The current character's index.
|
|
*/
|
|
uint32_t mTextElementCharIndex;
|
|
|
|
/**
|
|
* The index of the character that starts the cluster/ligature group the
|
|
* current character is a part of.
|
|
*/
|
|
uint32_t mGlyphStartTextElementCharIndex;
|
|
|
|
/**
|
|
* If we are iterating in mode eClusterOrLigatureGroupMiddle, then
|
|
* this tracks how many undisplayed characters were encountered
|
|
* between the start of this glyph (at mGlyphStartTextElementCharIndex)
|
|
* and the current character (at mTextElementCharIndex).
|
|
*/
|
|
uint32_t mGlyphUndisplayedCharacters;
|
|
|
|
/**
|
|
* The scale factor to apply to glyph advances returned by
|
|
* GetGlyphAdvance etc. to take into account textLength="".
|
|
*/
|
|
float mLengthAdjustScaleFactor;
|
|
|
|
/**
|
|
* Whether the instance of this class is being used after reflow has occurred
|
|
* or not.
|
|
*/
|
|
bool mPostReflow;
|
|
};
|
|
|
|
CharIterator::CharIterator(SVGTextFrame* aSVGTextFrame,
|
|
CharIterator::CharacterFilter aFilter,
|
|
nsIContent* aSubtree,
|
|
bool aPostReflow)
|
|
: mFilter(aFilter),
|
|
mFrameIterator(aSVGTextFrame, aSubtree),
|
|
mFrameForTrimCheck(nullptr),
|
|
mTrimmedOffset(0),
|
|
mTrimmedLength(0),
|
|
mTextRun(nullptr),
|
|
mTextElementCharIndex(0),
|
|
mGlyphStartTextElementCharIndex(0),
|
|
mGlyphUndisplayedCharacters(0),
|
|
mLengthAdjustScaleFactor(aSVGTextFrame->mLengthAdjustScaleFactor),
|
|
mPostReflow(aPostReflow)
|
|
{
|
|
if (!AtEnd()) {
|
|
mSkipCharsIterator = TextFrame()->EnsureTextRun(nsTextFrame::eInflated);
|
|
mTextRun = TextFrame()->GetTextRun(nsTextFrame::eInflated);
|
|
mTextElementCharIndex = mFrameIterator.UndisplayedCharacters();
|
|
UpdateGlyphStartTextElementCharIndex();
|
|
if (!MatchesFilter()) {
|
|
Next();
|
|
}
|
|
}
|
|
}
|
|
|
|
bool
|
|
CharIterator::Next()
|
|
{
|
|
while (NextCharacter()) {
|
|
if (MatchesFilter()) {
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
bool
|
|
CharIterator::Next(uint32_t aCount)
|
|
{
|
|
if (aCount == 0 && AtEnd()) {
|
|
return false;
|
|
}
|
|
while (aCount) {
|
|
if (!Next()) {
|
|
return false;
|
|
}
|
|
aCount--;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void
|
|
CharIterator::NextWithinSubtree(uint32_t aCount)
|
|
{
|
|
while (IsWithinSubtree() && aCount) {
|
|
--aCount;
|
|
if (!Next()) {
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
bool
|
|
CharIterator::AdvanceToCharacter(uint32_t aTextElementCharIndex)
|
|
{
|
|
while (mTextElementCharIndex < aTextElementCharIndex) {
|
|
if (!Next()) {
|
|
return false;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
CharIterator::AdvancePastCurrentFrame()
|
|
{
|
|
// XXX Can do this better than one character at a time if it matters.
|
|
nsTextFrame* currentFrame = TextFrame();
|
|
do {
|
|
if (!Next()) {
|
|
return false;
|
|
}
|
|
} while (TextFrame() == currentFrame);
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
CharIterator::AdvancePastCurrentTextPathFrame()
|
|
{
|
|
nsIFrame* currentTextPathFrame = TextPathFrame();
|
|
NS_ASSERTION(currentTextPathFrame,
|
|
"expected AdvancePastCurrentTextPathFrame to be called only "
|
|
"within a text path frame");
|
|
do {
|
|
if (!AdvancePastCurrentFrame()) {
|
|
return false;
|
|
}
|
|
} while (TextPathFrame() == currentTextPathFrame);
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
CharIterator::AdvanceToSubtree()
|
|
{
|
|
while (!IsWithinSubtree()) {
|
|
if (IsAfterSubtree()) {
|
|
return false;
|
|
}
|
|
if (!AdvancePastCurrentFrame()) {
|
|
return false;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
CharIterator::IsClusterAndLigatureGroupStart() const
|
|
{
|
|
return mTextRun->IsLigatureGroupStart(mSkipCharsIterator.GetSkippedOffset()) &&
|
|
mTextRun->IsClusterStart(mSkipCharsIterator.GetSkippedOffset());
|
|
}
|
|
|
|
bool
|
|
CharIterator::IsOriginalCharTrimmed() const
|
|
{
|
|
if (mFrameForTrimCheck != TextFrame()) {
|
|
// Since we do a lot of trim checking, we cache the trimmed offsets and
|
|
// lengths while we are in the same frame.
|
|
mFrameForTrimCheck = TextFrame();
|
|
uint32_t offset = mFrameForTrimCheck->GetContentOffset();
|
|
uint32_t length = mFrameForTrimCheck->GetContentLength();
|
|
nsIContent* content = mFrameForTrimCheck->GetContent();
|
|
nsTextFrame::TrimmedOffsets trim =
|
|
mFrameForTrimCheck->GetTrimmedOffsets(content->GetText(),
|
|
/* aTrimAfter */ true,
|
|
mPostReflow);
|
|
TrimOffsets(offset, length, trim);
|
|
mTrimmedOffset = offset;
|
|
mTrimmedLength = length;
|
|
}
|
|
|
|
// A character is trimmed if it is outside the mTrimmedOffset/mTrimmedLength
|
|
// range and it is not a significant newline character.
|
|
uint32_t index = mSkipCharsIterator.GetOriginalOffset();
|
|
return !((index >= mTrimmedOffset &&
|
|
index < mTrimmedOffset + mTrimmedLength) ||
|
|
(index >= mTrimmedOffset + mTrimmedLength &&
|
|
mFrameForTrimCheck->StyleText()->
|
|
NewlineIsSignificant(mFrameForTrimCheck) &&
|
|
mFrameForTrimCheck->GetContent()->GetText()->CharAt(index) == '\n'));
|
|
}
|
|
|
|
void
|
|
CharIterator::GetOriginalGlyphOffsets(uint32_t& aOriginalOffset,
|
|
uint32_t& aOriginalLength) const
|
|
{
|
|
gfxSkipCharsIterator it = TextFrame()->EnsureTextRun(nsTextFrame::eInflated);
|
|
it.SetOriginalOffset(mSkipCharsIterator.GetOriginalOffset() -
|
|
(mTextElementCharIndex -
|
|
mGlyphStartTextElementCharIndex -
|
|
mGlyphUndisplayedCharacters));
|
|
|
|
while (it.GetSkippedOffset() > 0 &&
|
|
(!mTextRun->IsClusterStart(it.GetSkippedOffset()) ||
|
|
!mTextRun->IsLigatureGroupStart(it.GetSkippedOffset()))) {
|
|
it.AdvanceSkipped(-1);
|
|
}
|
|
|
|
aOriginalOffset = it.GetOriginalOffset();
|
|
|
|
// Find the end of the cluster/ligature group.
|
|
it.SetOriginalOffset(mSkipCharsIterator.GetOriginalOffset());
|
|
do {
|
|
it.AdvanceSkipped(1);
|
|
} while (it.GetSkippedOffset() < mTextRun->GetLength() &&
|
|
(!mTextRun->IsClusterStart(it.GetSkippedOffset()) ||
|
|
!mTextRun->IsLigatureGroupStart(it.GetSkippedOffset())));
|
|
|
|
aOriginalLength = it.GetOriginalOffset() - aOriginalOffset;
|
|
}
|
|
|
|
gfxFloat
|
|
CharIterator::GetGlyphAdvance(nsPresContext* aContext) const
|
|
{
|
|
uint32_t offset, length;
|
|
GetOriginalGlyphOffsets(offset, length);
|
|
|
|
gfxSkipCharsIterator it = TextFrame()->EnsureTextRun(nsTextFrame::eInflated);
|
|
Range range = ConvertOriginalToSkipped(it, offset, length);
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
gfxFloat advance = mTextRun->GetAdvanceWidth(range, nullptr);
|
|
return aContext->AppUnitsToGfxUnits(advance) *
|
|
mLengthAdjustScaleFactor * cssPxPerDevPx;
|
|
}
|
|
|
|
gfxFloat
|
|
CharIterator::GetAdvance(nsPresContext* aContext) const
|
|
{
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
uint32_t offset = mSkipCharsIterator.GetSkippedOffset();
|
|
gfxFloat advance = mTextRun->
|
|
GetAdvanceWidth(Range(offset, offset + 1), nullptr);
|
|
return aContext->AppUnitsToGfxUnits(advance) *
|
|
mLengthAdjustScaleFactor * cssPxPerDevPx;
|
|
}
|
|
|
|
gfxFloat
|
|
CharIterator::GetGlyphPartialAdvance(uint32_t aPartLength,
|
|
nsPresContext* aContext) const
|
|
{
|
|
uint32_t offset, length;
|
|
GetOriginalGlyphOffsets(offset, length);
|
|
|
|
NS_ASSERTION(aPartLength <= length, "invalid aPartLength value");
|
|
length = aPartLength;
|
|
|
|
gfxSkipCharsIterator it = TextFrame()->EnsureTextRun(nsTextFrame::eInflated);
|
|
Range range = ConvertOriginalToSkipped(it, offset, length);
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(aContext->AppUnitsPerDevPixel());
|
|
|
|
gfxFloat advance = mTextRun->GetAdvanceWidth(range, nullptr);
|
|
return aContext->AppUnitsToGfxUnits(advance) *
|
|
mLengthAdjustScaleFactor * cssPxPerDevPx;
|
|
}
|
|
|
|
bool
|
|
CharIterator::NextCharacter()
|
|
{
|
|
if (AtEnd()) {
|
|
return false;
|
|
}
|
|
|
|
mTextElementCharIndex++;
|
|
|
|
// Advance within the current text run.
|
|
mSkipCharsIterator.AdvanceOriginal(1);
|
|
if (mSkipCharsIterator.GetOriginalOffset() < TextFrame()->GetContentEnd()) {
|
|
// We're still within the part of the text run for the current text frame.
|
|
UpdateGlyphStartTextElementCharIndex();
|
|
return true;
|
|
}
|
|
|
|
// Advance to the next frame.
|
|
mFrameIterator.Next();
|
|
|
|
// Skip any undisplayed characters.
|
|
uint32_t undisplayed = mFrameIterator.UndisplayedCharacters();
|
|
mGlyphUndisplayedCharacters += undisplayed;
|
|
mTextElementCharIndex += undisplayed;
|
|
if (!TextFrame()) {
|
|
// We're at the end.
|
|
mSkipCharsIterator = gfxSkipCharsIterator();
|
|
return false;
|
|
}
|
|
|
|
mSkipCharsIterator = TextFrame()->EnsureTextRun(nsTextFrame::eInflated);
|
|
mTextRun = TextFrame()->GetTextRun(nsTextFrame::eInflated);
|
|
UpdateGlyphStartTextElementCharIndex();
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
CharIterator::MatchesFilter() const
|
|
{
|
|
if (mFilter == eOriginal) {
|
|
return true;
|
|
}
|
|
|
|
if (IsOriginalCharSkipped()) {
|
|
return false;
|
|
}
|
|
|
|
if (mFilter == eAddressable) {
|
|
return !IsOriginalCharUnaddressable();
|
|
}
|
|
|
|
return (mFilter == eClusterAndLigatureGroupStart) ==
|
|
IsClusterAndLigatureGroupStart();
|
|
}
|
|
|
|
// -----------------------------------------------------------------------------
|
|
// nsCharClipDisplayItem
|
|
|
|
/**
|
|
* An nsCharClipDisplayItem that obtains its left and right clip edges from a
|
|
* TextRenderedRun object.
|
|
*/
|
|
class SVGCharClipDisplayItem : public nsCharClipDisplayItem {
|
|
public:
|
|
explicit SVGCharClipDisplayItem(const TextRenderedRun& aRun)
|
|
: nsCharClipDisplayItem(aRun.mFrame)
|
|
{
|
|
aRun.GetClipEdges(mVisIStartEdge, mVisIEndEdge);
|
|
}
|
|
|
|
NS_DISPLAY_DECL_NAME("SVGCharClip", TYPE_SVG_CHAR_CLIP)
|
|
};
|
|
|
|
// -----------------------------------------------------------------------------
|
|
// SVGTextDrawPathCallbacks
|
|
|
|
/**
|
|
* Text frame draw callback class that paints the text and text decoration parts
|
|
* of an nsTextFrame using SVG painting properties, and selection backgrounds
|
|
* and decorations as they would normally.
|
|
*
|
|
* An instance of this class is passed to nsTextFrame::PaintText if painting
|
|
* cannot be done directly (e.g. if we are using an SVG pattern fill, stroking
|
|
* the text, etc.).
|
|
*/
|
|
class SVGTextDrawPathCallbacks : public nsTextFrame::DrawPathCallbacks
|
|
{
|
|
typedef mozilla::image::imgDrawingParams imgDrawingParams;
|
|
|
|
public:
|
|
/**
|
|
* Constructs an SVGTextDrawPathCallbacks.
|
|
*
|
|
* @param aSVGTextFrame The ancestor text frame.
|
|
* @param aContext The context to use for painting.
|
|
* @param aFrame The nsTextFrame to paint.
|
|
* @param aCanvasTM The transformation matrix to set when painting; this
|
|
* should be the FOR_OUTERSVG_TM canvas TM of the text, so that
|
|
* paint servers are painted correctly.
|
|
* @param aImgParams Whether we need to synchronously decode images.
|
|
* @param aShouldPaintSVGGlyphs Whether SVG glyphs should be painted.
|
|
*/
|
|
SVGTextDrawPathCallbacks(SVGTextFrame* aSVGTextFrame,
|
|
gfxContext& aContext,
|
|
nsTextFrame* aFrame,
|
|
const gfxMatrix& aCanvasTM,
|
|
imgDrawingParams& aImgParams,
|
|
bool aShouldPaintSVGGlyphs)
|
|
: DrawPathCallbacks(aShouldPaintSVGGlyphs),
|
|
mSVGTextFrame(aSVGTextFrame),
|
|
mContext(aContext),
|
|
mFrame(aFrame),
|
|
mCanvasTM(aCanvasTM),
|
|
mImgParams(aImgParams),
|
|
mColor(0)
|
|
{
|
|
}
|
|
|
|
void NotifySelectionBackgroundNeedsFill(const Rect& aBackgroundRect,
|
|
nscolor aColor,
|
|
DrawTarget& aDrawTarget) override;
|
|
void PaintDecorationLine(Rect aPath, nscolor aColor) override;
|
|
void PaintSelectionDecorationLine(Rect aPath, nscolor aColor) override;
|
|
void NotifyBeforeText(nscolor aColor) override;
|
|
void NotifyGlyphPathEmitted() override;
|
|
void NotifyAfterText() override;
|
|
|
|
private:
|
|
void SetupContext();
|
|
|
|
bool IsClipPathChild() const {
|
|
return mSVGTextFrame->HasAnyStateBits(NS_STATE_SVG_CLIPPATH_CHILD);
|
|
}
|
|
|
|
/**
|
|
* Paints a piece of text geometry. This is called when glyphs
|
|
* or text decorations have been emitted to the gfxContext.
|
|
*/
|
|
void HandleTextGeometry();
|
|
|
|
/**
|
|
* Sets the gfxContext paint to the appropriate color or pattern
|
|
* for filling text geometry.
|
|
*/
|
|
void MakeFillPattern(GeneralPattern* aOutPattern);
|
|
|
|
/**
|
|
* Fills and strokes a piece of text geometry, using group opacity
|
|
* if the selection style requires it.
|
|
*/
|
|
void FillAndStrokeGeometry();
|
|
|
|
/**
|
|
* Fills a piece of text geometry.
|
|
*/
|
|
void FillGeometry();
|
|
|
|
/**
|
|
* Strokes a piece of text geometry.
|
|
*/
|
|
void StrokeGeometry();
|
|
|
|
SVGTextFrame* mSVGTextFrame;
|
|
gfxContext& mContext;
|
|
nsTextFrame* mFrame;
|
|
const gfxMatrix& mCanvasTM;
|
|
imgDrawingParams& mImgParams;
|
|
|
|
/**
|
|
* The color that we were last told from one of the path callback functions.
|
|
* This color can be the special NS_SAME_AS_FOREGROUND_COLOR,
|
|
* NS_40PERCENT_FOREGROUND_COLOR and NS_TRANSPARENT colors when we are
|
|
* painting selections or IME decorations.
|
|
*/
|
|
nscolor mColor;
|
|
};
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::NotifySelectionBackgroundNeedsFill(
|
|
const Rect& aBackgroundRect,
|
|
nscolor aColor,
|
|
DrawTarget& aDrawTarget)
|
|
{
|
|
if (IsClipPathChild()) {
|
|
// Don't paint selection backgrounds when in a clip path.
|
|
return;
|
|
}
|
|
|
|
mColor = aColor; // currently needed by MakeFillPattern
|
|
|
|
GeneralPattern fillPattern;
|
|
MakeFillPattern(&fillPattern);
|
|
if (fillPattern.GetPattern()) {
|
|
DrawOptions drawOptions(aColor == NS_40PERCENT_FOREGROUND_COLOR ? 0.4 : 1.0);
|
|
aDrawTarget.FillRect(aBackgroundRect, fillPattern, drawOptions);
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::NotifyBeforeText(nscolor aColor)
|
|
{
|
|
mColor = aColor;
|
|
SetupContext();
|
|
mContext.NewPath();
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::NotifyGlyphPathEmitted()
|
|
{
|
|
HandleTextGeometry();
|
|
mContext.NewPath();
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::NotifyAfterText()
|
|
{
|
|
mContext.Restore();
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::PaintDecorationLine(Rect aPath, nscolor aColor)
|
|
{
|
|
mColor = aColor;
|
|
AntialiasMode aaMode =
|
|
nsSVGUtils::ToAntialiasMode(mFrame->StyleText()->mTextRendering);
|
|
|
|
mContext.Save();
|
|
mContext.NewPath();
|
|
mContext.SetAntialiasMode(aaMode);
|
|
mContext.Rectangle(ThebesRect(aPath));
|
|
HandleTextGeometry();
|
|
mContext.NewPath();
|
|
mContext.Restore();
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::PaintSelectionDecorationLine(Rect aPath,
|
|
nscolor aColor)
|
|
{
|
|
if (IsClipPathChild()) {
|
|
// Don't paint selection decorations when in a clip path.
|
|
return;
|
|
}
|
|
|
|
mColor = aColor;
|
|
|
|
mContext.Save();
|
|
mContext.NewPath();
|
|
mContext.Rectangle(ThebesRect(aPath));
|
|
FillAndStrokeGeometry();
|
|
mContext.Restore();
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::SetupContext()
|
|
{
|
|
mContext.Save();
|
|
|
|
// XXX This is copied from nsSVGGlyphFrame::Render, but cairo doesn't actually
|
|
// seem to do anything with the antialias mode. So we can perhaps remove it,
|
|
// or make SetAntialiasMode set cairo text antialiasing too.
|
|
switch (mFrame->StyleText()->mTextRendering) {
|
|
case NS_STYLE_TEXT_RENDERING_OPTIMIZESPEED:
|
|
mContext.SetAntialiasMode(AntialiasMode::NONE);
|
|
break;
|
|
default:
|
|
mContext.SetAntialiasMode(AntialiasMode::SUBPIXEL);
|
|
break;
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::HandleTextGeometry()
|
|
{
|
|
if (IsClipPathChild()) {
|
|
RefPtr<Path> path = mContext.GetPath();
|
|
ColorPattern white(Color(1.f, 1.f, 1.f, 1.f)); // for masking, so no ToDeviceColor
|
|
mContext.GetDrawTarget()->Fill(path, white);
|
|
} else {
|
|
// Normal painting.
|
|
gfxContextMatrixAutoSaveRestore saveMatrix(&mContext);
|
|
mContext.SetMatrixDouble(mCanvasTM);
|
|
|
|
FillAndStrokeGeometry();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::MakeFillPattern(GeneralPattern* aOutPattern)
|
|
{
|
|
if (mColor == NS_SAME_AS_FOREGROUND_COLOR ||
|
|
mColor == NS_40PERCENT_FOREGROUND_COLOR) {
|
|
nsSVGUtils::MakeFillPatternFor(mFrame, &mContext, aOutPattern, mImgParams);
|
|
return;
|
|
}
|
|
|
|
if (mColor == NS_TRANSPARENT) {
|
|
return;
|
|
}
|
|
|
|
aOutPattern->InitColorPattern(ToDeviceColor(mColor));
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::FillAndStrokeGeometry()
|
|
{
|
|
bool pushedGroup = false;
|
|
if (mColor == NS_40PERCENT_FOREGROUND_COLOR) {
|
|
pushedGroup = true;
|
|
mContext.PushGroupForBlendBack(gfxContentType::COLOR_ALPHA, 0.4f);
|
|
}
|
|
|
|
uint32_t paintOrder = mFrame->StyleSVG()->mPaintOrder;
|
|
if (paintOrder == NS_STYLE_PAINT_ORDER_NORMAL) {
|
|
FillGeometry();
|
|
StrokeGeometry();
|
|
} else {
|
|
while (paintOrder) {
|
|
uint32_t component =
|
|
paintOrder & ((1 << NS_STYLE_PAINT_ORDER_BITWIDTH) - 1);
|
|
switch (component) {
|
|
case NS_STYLE_PAINT_ORDER_FILL:
|
|
FillGeometry();
|
|
break;
|
|
case NS_STYLE_PAINT_ORDER_STROKE:
|
|
StrokeGeometry();
|
|
break;
|
|
}
|
|
paintOrder >>= NS_STYLE_PAINT_ORDER_BITWIDTH;
|
|
}
|
|
}
|
|
|
|
if (pushedGroup) {
|
|
mContext.PopGroupAndBlend();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::FillGeometry()
|
|
{
|
|
GeneralPattern fillPattern;
|
|
MakeFillPattern(&fillPattern);
|
|
if (fillPattern.GetPattern()) {
|
|
RefPtr<Path> path = mContext.GetPath();
|
|
FillRule fillRule = nsSVGUtils::ToFillRule(IsClipPathChild() ?
|
|
mFrame->StyleSVG()->mClipRule :
|
|
mFrame->StyleSVG()->mFillRule);
|
|
if (fillRule != path->GetFillRule()) {
|
|
RefPtr<PathBuilder> builder = path->CopyToBuilder(fillRule);
|
|
path = builder->Finish();
|
|
}
|
|
mContext.GetDrawTarget()->Fill(path, fillPattern);
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextDrawPathCallbacks::StrokeGeometry()
|
|
{
|
|
// We don't paint the stroke when we are filling with a selection color.
|
|
if (mColor == NS_SAME_AS_FOREGROUND_COLOR ||
|
|
mColor == NS_40PERCENT_FOREGROUND_COLOR) {
|
|
if (nsSVGUtils::HasStroke(mFrame, /*aContextPaint*/ nullptr)) {
|
|
GeneralPattern strokePattern;
|
|
nsSVGUtils::MakeStrokePatternFor(mFrame, &mContext, &strokePattern,
|
|
mImgParams, /*aContextPaint*/ nullptr);
|
|
if (strokePattern.GetPattern()) {
|
|
if (!mFrame->GetParent()->GetContent()->IsSVGElement()) {
|
|
// The cast that follows would be unsafe
|
|
MOZ_ASSERT(false, "Our nsTextFrame's parent's content should be SVG");
|
|
return;
|
|
}
|
|
nsSVGElement* svgOwner =
|
|
static_cast<nsSVGElement*>(mFrame->GetParent()->GetContent());
|
|
|
|
// Apply any stroke-specific transform
|
|
gfxMatrix outerSVGToUser;
|
|
if (nsSVGUtils::GetNonScalingStrokeTransform(mFrame, &outerSVGToUser) &&
|
|
outerSVGToUser.Invert()) {
|
|
mContext.Multiply(outerSVGToUser);
|
|
}
|
|
|
|
RefPtr<Path> path = mContext.GetPath();
|
|
SVGContentUtils::AutoStrokeOptions strokeOptions;
|
|
SVGContentUtils::GetStrokeOptions(&strokeOptions, svgOwner,
|
|
mFrame->Style(),
|
|
/*aContextPaint*/ nullptr);
|
|
DrawOptions drawOptions;
|
|
drawOptions.mAntialiasMode =
|
|
nsSVGUtils::ToAntialiasMode(mFrame->StyleText()->mTextRendering);
|
|
mContext.GetDrawTarget()->Stroke(path, strokePattern, strokeOptions);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
} // namespace mozilla
|
|
|
|
|
|
// ============================================================================
|
|
// SVGTextFrame
|
|
|
|
// ----------------------------------------------------------------------------
|
|
// Display list item
|
|
|
|
class nsDisplaySVGText : public nsDisplayItem {
|
|
public:
|
|
nsDisplaySVGText(nsDisplayListBuilder* aBuilder, SVGTextFrame* aFrame)
|
|
: nsDisplayItem(aBuilder, aFrame)
|
|
{
|
|
MOZ_COUNT_CTOR(nsDisplaySVGText);
|
|
MOZ_ASSERT(aFrame, "Must have a frame!");
|
|
}
|
|
#ifdef NS_BUILD_REFCNT_LOGGING
|
|
virtual ~nsDisplaySVGText() {
|
|
MOZ_COUNT_DTOR(nsDisplaySVGText);
|
|
}
|
|
#endif
|
|
|
|
NS_DISPLAY_DECL_NAME("nsDisplaySVGText", TYPE_SVG_TEXT)
|
|
|
|
virtual void HitTest(nsDisplayListBuilder* aBuilder, const nsRect& aRect,
|
|
HitTestState* aState,
|
|
nsTArray<nsIFrame*> *aOutFrames) override;
|
|
virtual void Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) override;
|
|
nsDisplayItemGeometry* AllocateGeometry(nsDisplayListBuilder* aBuilder) override
|
|
{
|
|
return new nsDisplayItemGenericImageGeometry(this, aBuilder);
|
|
}
|
|
|
|
virtual nsRect GetComponentAlphaBounds(nsDisplayListBuilder* aBuilder) const override
|
|
{
|
|
bool snap;
|
|
return GetBounds(aBuilder, &snap);
|
|
}
|
|
};
|
|
|
|
void
|
|
nsDisplaySVGText::HitTest(nsDisplayListBuilder* aBuilder, const nsRect& aRect,
|
|
HitTestState* aState, nsTArray<nsIFrame*> *aOutFrames)
|
|
{
|
|
SVGTextFrame *frame = static_cast<SVGTextFrame*>(mFrame);
|
|
nsPoint pointRelativeToReferenceFrame = aRect.Center();
|
|
// ToReferenceFrame() includes frame->GetPosition(), our user space position.
|
|
nsPoint userSpacePtInAppUnits = pointRelativeToReferenceFrame -
|
|
(ToReferenceFrame() - frame->GetPosition());
|
|
|
|
gfxPoint userSpacePt =
|
|
gfxPoint(userSpacePtInAppUnits.x, userSpacePtInAppUnits.y) /
|
|
frame->PresContext()->AppUnitsPerCSSPixel();
|
|
|
|
nsIFrame* target = frame->GetFrameForPoint(userSpacePt);
|
|
if (target) {
|
|
aOutFrames->AppendElement(target);
|
|
}
|
|
}
|
|
|
|
void
|
|
nsDisplaySVGText::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx)
|
|
{
|
|
DrawTargetAutoDisableSubpixelAntialiasing
|
|
disable(aCtx->GetDrawTarget(), mDisableSubpixelAA);
|
|
|
|
uint32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
// ToReferenceFrame includes our mRect offset, but painting takes
|
|
// account of that too. To avoid double counting, we subtract that
|
|
// here.
|
|
nsPoint offset = ToReferenceFrame() - mFrame->GetPosition();
|
|
|
|
gfxPoint devPixelOffset =
|
|
nsLayoutUtils::PointToGfxPoint(offset, appUnitsPerDevPixel);
|
|
|
|
gfxMatrix tm = nsSVGUtils::GetCSSPxToDevPxMatrix(mFrame) *
|
|
gfxMatrix::Translation(devPixelOffset);
|
|
|
|
gfxContext* ctx = aCtx;
|
|
imgDrawingParams imgParams(aBuilder->ShouldSyncDecodeImages()
|
|
? imgIContainer::FLAG_SYNC_DECODE
|
|
: imgIContainer::FLAG_SYNC_DECODE_IF_FAST);
|
|
static_cast<SVGTextFrame*>(mFrame)->PaintSVG(*ctx, tm, imgParams);
|
|
nsDisplayItemGenericImageGeometry::UpdateDrawResult(this, imgParams.result);
|
|
}
|
|
|
|
// ---------------------------------------------------------------------
|
|
// nsQueryFrame methods
|
|
|
|
NS_QUERYFRAME_HEAD(SVGTextFrame)
|
|
NS_QUERYFRAME_ENTRY(SVGTextFrame)
|
|
NS_QUERYFRAME_TAIL_INHERITING(nsSVGDisplayContainerFrame)
|
|
|
|
// ---------------------------------------------------------------------
|
|
// Implementation
|
|
|
|
nsIFrame*
|
|
NS_NewSVGTextFrame(nsIPresShell* aPresShell, ComputedStyle* aStyle)
|
|
{
|
|
return new (aPresShell) SVGTextFrame(aStyle);
|
|
}
|
|
|
|
NS_IMPL_FRAMEARENA_HELPERS(SVGTextFrame)
|
|
|
|
// ---------------------------------------------------------------------
|
|
// nsIFrame methods
|
|
|
|
void
|
|
SVGTextFrame::Init(nsIContent* aContent,
|
|
nsContainerFrame* aParent,
|
|
nsIFrame* aPrevInFlow)
|
|
{
|
|
NS_ASSERTION(aContent->IsSVGElement(nsGkAtoms::text), "Content is not an SVG text");
|
|
|
|
nsSVGDisplayContainerFrame::Init(aContent, aParent, aPrevInFlow);
|
|
AddStateBits((aParent->GetStateBits() & NS_STATE_SVG_CLIPPATH_CHILD) |
|
|
NS_FRAME_SVG_LAYOUT | NS_FRAME_IS_SVG_TEXT);
|
|
|
|
mMutationObserver = new MutationObserver(this);
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::BuildDisplayList(nsDisplayListBuilder* aBuilder,
|
|
const nsDisplayListSet& aLists)
|
|
{
|
|
if (NS_SUBTREE_DIRTY(this)) {
|
|
// We can sometimes be asked to paint before reflow happens and we
|
|
// have updated mPositions, etc. In this case, we just avoid
|
|
// painting.
|
|
return;
|
|
}
|
|
if (!IsVisibleForPainting(aBuilder) &&
|
|
aBuilder->IsForPainting()) {
|
|
return;
|
|
}
|
|
DisplayOutline(aBuilder, aLists);
|
|
aLists.Content()->AppendToTop(
|
|
MakeDisplayItem<nsDisplaySVGText>(aBuilder, this));
|
|
}
|
|
|
|
nsresult
|
|
SVGTextFrame::AttributeChanged(int32_t aNameSpaceID,
|
|
nsAtom* aAttribute,
|
|
int32_t aModType)
|
|
{
|
|
if (aNameSpaceID != kNameSpaceID_None)
|
|
return NS_OK;
|
|
|
|
if (aAttribute == nsGkAtoms::transform) {
|
|
// We don't invalidate for transform changes (the layers code does that).
|
|
// Also note that SVGTransformableElement::GetAttributeChangeHint will
|
|
// return nsChangeHint_UpdateOverflow for "transform" attribute changes
|
|
// and cause DoApplyRenderingChangeToTree to make the SchedulePaint call.
|
|
|
|
if (!(mState & NS_FRAME_FIRST_REFLOW) &&
|
|
mCanvasTM && mCanvasTM->IsSingular()) {
|
|
// We won't have calculated the glyph positions correctly.
|
|
NotifyGlyphMetricsChange();
|
|
}
|
|
mCanvasTM = nullptr;
|
|
} else if (IsGlyphPositioningAttribute(aAttribute) ||
|
|
aAttribute == nsGkAtoms::textLength ||
|
|
aAttribute == nsGkAtoms::lengthAdjust) {
|
|
NotifyGlyphMetricsChange();
|
|
}
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DidSetComputedStyle(ComputedStyle* aOldComputedStyle)
|
|
{
|
|
if (mState & NS_FRAME_IS_NONDISPLAY) {
|
|
// We need this DidSetComputedStyle override to handle cases like this:
|
|
//
|
|
// <defs>
|
|
// <g>
|
|
// <mask>
|
|
// <text>...</text>
|
|
// </mask>
|
|
// </g>
|
|
// </defs>
|
|
//
|
|
// where the <text> is non-display, and a style change occurs on the <defs>,
|
|
// the <g>, the <mask>, or the <text> itself. If the style change happened
|
|
// on the parent of the <defs>, then in
|
|
// nsSVGDisplayContainerFrame::ReflowSVG, we would find the non-display
|
|
// <defs> container and then call ReflowSVGNonDisplayText on it. If we do
|
|
// not actually reflow the parent of the <defs>, then without this
|
|
// DidSetComputedStyle we would (a) not cause the <text>'s anonymous block
|
|
// child to be reflowed when it is next painted, and (b) not cause the
|
|
// <text> to be repainted anyway since the user of the <mask> would not
|
|
// know it needs to be repainted.
|
|
ScheduleReflowSVGNonDisplayText(nsIPresShell::eStyleChange);
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::ReflowSVGNonDisplayText()
|
|
{
|
|
MOZ_ASSERT(nsSVGUtils::AnyOuterSVGIsCallingReflowSVG(this),
|
|
"only call ReflowSVGNonDisplayText when an outer SVG frame is "
|
|
"under ReflowSVG");
|
|
MOZ_ASSERT(mState & NS_FRAME_IS_NONDISPLAY,
|
|
"only call ReflowSVGNonDisplayText if the frame is "
|
|
"NS_FRAME_IS_NONDISPLAY");
|
|
|
|
// We had a style change, so we mark this frame as dirty so that the next
|
|
// time it is painted, we reflow the anonymous block frame.
|
|
AddStateBits(NS_FRAME_IS_DIRTY);
|
|
|
|
// We also need to call InvalidateRenderingObservers, so that if the <text>
|
|
// element is within a <mask>, say, the element referencing the <mask> will
|
|
// be updated, which will then cause this SVGTextFrame to be painted and
|
|
// in doing so cause the anonymous block frame to be reflowed.
|
|
nsLayoutUtils::PostRestyleEvent(
|
|
mContent->AsElement(), nsRestyleHint(0),
|
|
nsChangeHint_InvalidateRenderingObservers);
|
|
|
|
// Finally, we need to actually reflow the anonymous block frame and update
|
|
// mPositions, in case we are being reflowed immediately after a DOM
|
|
// mutation that needs frame reconstruction.
|
|
MaybeReflowAnonymousBlockChild();
|
|
UpdateGlyphPositioning();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::ScheduleReflowSVGNonDisplayText(nsIPresShell::IntrinsicDirty aReason)
|
|
{
|
|
MOZ_ASSERT(!nsSVGUtils::OuterSVGIsCallingReflowSVG(this),
|
|
"do not call ScheduleReflowSVGNonDisplayText when the outer SVG "
|
|
"frame is under ReflowSVG");
|
|
MOZ_ASSERT(!(mState & NS_STATE_SVG_TEXT_IN_REFLOW),
|
|
"do not call ScheduleReflowSVGNonDisplayText while reflowing the "
|
|
"anonymous block child");
|
|
|
|
// We need to find an ancestor frame that we can call FrameNeedsReflow
|
|
// on that will cause the document to be marked as needing relayout,
|
|
// and for that ancestor (or some further ancestor) to be marked as
|
|
// a root to reflow. We choose the closest ancestor frame that is not
|
|
// NS_FRAME_IS_NONDISPLAY and which is either an outer SVG frame or a
|
|
// non-SVG frame. (We don't consider displayed SVG frame ancestors toerh
|
|
// than nsSVGOuterSVGFrame, since calling FrameNeedsReflow on those other
|
|
// SVG frames would do a bunch of unnecessary work on the SVG frames up to
|
|
// the nsSVGOuterSVGFrame.)
|
|
|
|
nsIFrame* f = this;
|
|
while (f) {
|
|
if (!(f->GetStateBits() & NS_FRAME_IS_NONDISPLAY)) {
|
|
if (NS_SUBTREE_DIRTY(f)) {
|
|
// This is a displayed frame, so if it is already dirty, we will be reflowed
|
|
// soon anyway. No need to call FrameNeedsReflow again, then.
|
|
return;
|
|
}
|
|
if (!f->IsFrameOfType(eSVG) ||
|
|
(f->GetStateBits() & NS_STATE_IS_OUTER_SVG)) {
|
|
break;
|
|
}
|
|
f->AddStateBits(NS_FRAME_HAS_DIRTY_CHILDREN);
|
|
}
|
|
f = f->GetParent();
|
|
}
|
|
|
|
MOZ_ASSERT(f, "should have found an ancestor frame to reflow");
|
|
|
|
PresShell()->FrameNeedsReflow(f, aReason, NS_FRAME_IS_DIRTY);
|
|
}
|
|
|
|
NS_IMPL_ISUPPORTS(SVGTextFrame::MutationObserver, nsIMutationObserver)
|
|
|
|
void
|
|
SVGTextFrame::MutationObserver::ContentAppended(nsIContent* aFirstNewContent)
|
|
{
|
|
mFrame->NotifyGlyphMetricsChange();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MutationObserver::ContentInserted(nsIContent* aChild)
|
|
{
|
|
mFrame->NotifyGlyphMetricsChange();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MutationObserver::ContentRemoved(nsIContent* aChild,
|
|
nsIContent* aPreviousSibling)
|
|
{
|
|
mFrame->NotifyGlyphMetricsChange();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MutationObserver::CharacterDataChanged(nsIContent* aContent,
|
|
const CharacterDataChangeInfo&)
|
|
{
|
|
mFrame->NotifyGlyphMetricsChange();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MutationObserver::AttributeChanged(
|
|
Element* aElement,
|
|
int32_t aNameSpaceID,
|
|
nsAtom* aAttribute,
|
|
int32_t aModType,
|
|
const nsAttrValue* aOldValue)
|
|
{
|
|
if (!aElement->IsSVGElement()) {
|
|
return;
|
|
}
|
|
|
|
// Attribute changes on this element will be handled by
|
|
// SVGTextFrame::AttributeChanged.
|
|
if (aElement == mFrame->GetContent()) {
|
|
return;
|
|
}
|
|
|
|
mFrame->HandleAttributeChangeInDescendant(aElement, aNameSpaceID, aAttribute);
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::HandleAttributeChangeInDescendant(Element* aElement,
|
|
int32_t aNameSpaceID,
|
|
nsAtom* aAttribute)
|
|
{
|
|
if (aElement->IsSVGElement(nsGkAtoms::textPath)) {
|
|
if (aNameSpaceID == kNameSpaceID_None &&
|
|
(aAttribute == nsGkAtoms::startOffset ||
|
|
aAttribute == nsGkAtoms::path ||
|
|
aAttribute == nsGkAtoms::side_)) {
|
|
NotifyGlyphMetricsChange();
|
|
} else if ((aNameSpaceID == kNameSpaceID_XLink ||
|
|
aNameSpaceID == kNameSpaceID_None) &&
|
|
aAttribute == nsGkAtoms::href) {
|
|
// Blow away our reference, if any
|
|
nsIFrame* childElementFrame = aElement->GetPrimaryFrame();
|
|
if (childElementFrame) {
|
|
childElementFrame->DeleteProperty(
|
|
SVGObserverUtils::HrefAsTextPathProperty());
|
|
NotifyGlyphMetricsChange();
|
|
}
|
|
}
|
|
} else {
|
|
if (aNameSpaceID == kNameSpaceID_None &&
|
|
IsGlyphPositioningAttribute(aAttribute)) {
|
|
NotifyGlyphMetricsChange();
|
|
}
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::FindCloserFrameForSelection(const nsPoint& aPoint,
|
|
FrameWithDistance* aCurrentBestFrame)
|
|
{
|
|
if (GetStateBits() & NS_FRAME_IS_NONDISPLAY) {
|
|
return;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
// Find the frame that has the closest rendered run rect to aPoint.
|
|
TextRenderedRunIterator it(this);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
uint32_t flags = TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeStroke |
|
|
TextRenderedRun::eNoHorizontalOverflow;
|
|
SVGBBox userRect = run.GetUserSpaceRect(presContext, flags);
|
|
float devPxPerCSSPx = presContext->CSSPixelsToDevPixels(1.f);
|
|
userRect.Scale(devPxPerCSSPx);
|
|
|
|
if (!userRect.IsEmpty()) {
|
|
gfxMatrix m;
|
|
if (!NS_SVGDisplayListHitTestingEnabled()) {
|
|
m = GetCanvasTM();
|
|
}
|
|
nsRect rect = nsSVGUtils::ToCanvasBounds(userRect.ToThebesRect(), m,
|
|
presContext);
|
|
|
|
if (nsLayoutUtils::PointIsCloserToRect(aPoint, rect,
|
|
aCurrentBestFrame->mXDistance,
|
|
aCurrentBestFrame->mYDistance)) {
|
|
aCurrentBestFrame->mFrame = run.mFrame;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
//----------------------------------------------------------------------
|
|
// nsSVGDisplayableFrame methods
|
|
|
|
void
|
|
SVGTextFrame::NotifySVGChanged(uint32_t aFlags)
|
|
{
|
|
MOZ_ASSERT(aFlags & (TRANSFORM_CHANGED | COORD_CONTEXT_CHANGED),
|
|
"Invalidation logic may need adjusting");
|
|
|
|
bool needNewBounds = false;
|
|
bool needGlyphMetricsUpdate = false;
|
|
bool needNewCanvasTM = false;
|
|
|
|
if ((aFlags & COORD_CONTEXT_CHANGED) &&
|
|
(mState & NS_STATE_SVG_POSITIONING_MAY_USE_PERCENTAGES)) {
|
|
needGlyphMetricsUpdate = true;
|
|
}
|
|
|
|
if (aFlags & TRANSFORM_CHANGED) {
|
|
needNewCanvasTM = true;
|
|
if (mCanvasTM && mCanvasTM->IsSingular()) {
|
|
// We won't have calculated the glyph positions correctly.
|
|
needNewBounds = true;
|
|
needGlyphMetricsUpdate = true;
|
|
}
|
|
if (StyleSVGReset()->HasNonScalingStroke()) {
|
|
// Stroke currently contributes to our mRect, and our stroke depends on
|
|
// the transform to our outer-<svg> if |vector-effect:non-scaling-stroke|.
|
|
needNewBounds = true;
|
|
}
|
|
}
|
|
|
|
// If the scale at which we computed our mFontSizeScaleFactor has changed by
|
|
// at least a factor of two, reflow the text. This avoids reflowing text
|
|
// at every tick of a transform animation, but ensures our glyph metrics
|
|
// do not get too far out of sync with the final font size on the screen.
|
|
if (needNewCanvasTM && mLastContextScale != 0.0f) {
|
|
mCanvasTM = nullptr;
|
|
// If we are a non-display frame, then we don't want to call
|
|
// GetCanvasTM(), since the context scale does not use it.
|
|
gfxMatrix newTM =
|
|
(mState & NS_FRAME_IS_NONDISPLAY) ? gfxMatrix() :
|
|
GetCanvasTM();
|
|
// Compare the old and new context scales.
|
|
float scale = GetContextScale(newTM);
|
|
float change = scale / mLastContextScale;
|
|
if (change >= 2.0f || change <= 0.5f) {
|
|
needNewBounds = true;
|
|
needGlyphMetricsUpdate = true;
|
|
}
|
|
}
|
|
|
|
if (needNewBounds) {
|
|
// Ancestor changes can't affect how we render from the perspective of
|
|
// any rendering observers that we may have, so we don't need to
|
|
// invalidate them. We also don't need to invalidate ourself, since our
|
|
// changed ancestor will have invalidated its entire area, which includes
|
|
// our area.
|
|
ScheduleReflowSVG();
|
|
}
|
|
|
|
if (needGlyphMetricsUpdate) {
|
|
// If we are positioned using percentage values we need to update our
|
|
// position whenever our viewport's dimensions change. But only do this if
|
|
// we have been reflowed once, otherwise the glyph positioning will be
|
|
// wrong. (We need to wait until bidi reordering has been done.)
|
|
if (!(mState & NS_FRAME_FIRST_REFLOW)) {
|
|
NotifyGlyphMetricsChange();
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Gets the offset into a DOM node that the specified caret is positioned at.
|
|
*/
|
|
static int32_t
|
|
GetCaretOffset(nsCaret* aCaret)
|
|
{
|
|
RefPtr<Selection> selection = aCaret->GetSelection();
|
|
if (!selection) {
|
|
return -1;
|
|
}
|
|
|
|
return selection->AnchorOffset();
|
|
}
|
|
|
|
/**
|
|
* Returns whether the caret should be painted for a given TextRenderedRun
|
|
* by checking whether the caret is in the range covered by the rendered run.
|
|
*
|
|
* @param aThisRun The TextRenderedRun to be painted.
|
|
* @param aCaret The caret.
|
|
*/
|
|
static bool
|
|
ShouldPaintCaret(const TextRenderedRun& aThisRun, nsCaret* aCaret)
|
|
{
|
|
int32_t caretOffset = GetCaretOffset(aCaret);
|
|
|
|
if (caretOffset < 0) {
|
|
return false;
|
|
}
|
|
|
|
if (uint32_t(caretOffset) >= aThisRun.mTextFrameContentOffset &&
|
|
uint32_t(caretOffset) < aThisRun.mTextFrameContentOffset +
|
|
aThisRun.mTextFrameContentLength) {
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::PaintSVG(gfxContext& aContext,
|
|
const gfxMatrix& aTransform,
|
|
imgDrawingParams& aImgParams,
|
|
const nsIntRect *aDirtyRect)
|
|
{
|
|
DrawTarget& aDrawTarget = *aContext.GetDrawTarget();
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (!kid) {
|
|
return;
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
gfxMatrix initialMatrix = aContext.CurrentMatrixDouble();
|
|
|
|
if (mState & NS_FRAME_IS_NONDISPLAY) {
|
|
// If we are in a canvas DrawWindow call that used the
|
|
// DRAWWINDOW_DO_NOT_FLUSH flag, then we may still have out
|
|
// of date frames. Just don't paint anything if they are
|
|
// dirty.
|
|
if (presContext->PresShell()->InDrawWindowNotFlushing() &&
|
|
NS_SUBTREE_DIRTY(this)) {
|
|
return;
|
|
}
|
|
// Text frames inside <clipPath>, <mask>, etc. will never have had
|
|
// ReflowSVG called on them, so call UpdateGlyphPositioning to do this now.
|
|
UpdateGlyphPositioning();
|
|
} else if (NS_SUBTREE_DIRTY(this)) {
|
|
// If we are asked to paint before reflow has recomputed mPositions etc.
|
|
// directly via PaintSVG, rather than via a display list, then we need
|
|
// to bail out here too.
|
|
return;
|
|
}
|
|
|
|
if (aTransform.IsSingular()) {
|
|
NS_WARNING("Can't render text element!");
|
|
return;
|
|
}
|
|
|
|
gfxMatrix matrixForPaintServers = aTransform * initialMatrix;
|
|
|
|
// Check if we need to draw anything.
|
|
if (aDirtyRect) {
|
|
NS_ASSERTION(!NS_SVGDisplayListPaintingEnabled() ||
|
|
(mState & NS_FRAME_IS_NONDISPLAY),
|
|
"Display lists handle dirty rect intersection test");
|
|
nsRect dirtyRect(aDirtyRect->x, aDirtyRect->y,
|
|
aDirtyRect->width, aDirtyRect->height);
|
|
|
|
gfxFloat appUnitsPerDevPixel = presContext->AppUnitsPerDevPixel();
|
|
gfxRect frameRect(mRect.x / appUnitsPerDevPixel,
|
|
mRect.y / appUnitsPerDevPixel,
|
|
mRect.width / appUnitsPerDevPixel,
|
|
mRect.height / appUnitsPerDevPixel);
|
|
|
|
nsRect canvasRect = nsLayoutUtils::RoundGfxRectToAppRect(
|
|
GetCanvasTM().TransformBounds(frameRect), 1);
|
|
if (!canvasRect.Intersects(dirtyRect)) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
// SVG frames' PaintSVG methods paint in CSS px, but normally frames paint in
|
|
// dev pixels. Here we multiply a CSS-px-to-dev-pixel factor onto aTransform
|
|
// so our non-SVG nsTextFrame children paint correctly.
|
|
auto auPerDevPx = presContext->AppUnitsPerDevPixel();
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(auPerDevPx);
|
|
gfxMatrix canvasTMForChildren = aTransform;
|
|
canvasTMForChildren.PreScale(cssPxPerDevPx, cssPxPerDevPx);
|
|
initialMatrix.PreScale(1 / cssPxPerDevPx, 1 / cssPxPerDevPx);
|
|
|
|
gfxContextMatrixAutoSaveRestore matSR(&aContext);
|
|
aContext.NewPath();
|
|
aContext.Multiply(canvasTMForChildren);
|
|
gfxMatrix currentMatrix = aContext.CurrentMatrixDouble();
|
|
|
|
RefPtr<nsCaret> caret = presContext->PresShell()->GetCaret();
|
|
nsRect caretRect;
|
|
nsIFrame* caretFrame = caret->GetPaintGeometry(&caretRect);
|
|
|
|
gfxContextAutoSaveRestore ctxSR;
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eVisibleFrames);
|
|
TextRenderedRun run = it.Current();
|
|
|
|
SVGContextPaint* outerContextPaint =
|
|
SVGContextPaint::GetContextPaint(GetContent());
|
|
|
|
while (run.mFrame) {
|
|
nsTextFrame* frame = run.mFrame;
|
|
|
|
// Determine how much of the left and right edges of the text frame we
|
|
// need to ignore.
|
|
SVGCharClipDisplayItem item(run);
|
|
|
|
RefPtr<SVGContextPaintImpl> contextPaint = new SVGContextPaintImpl();
|
|
DrawMode drawMode = contextPaint->Init(&aDrawTarget,
|
|
initialMatrix,
|
|
frame, outerContextPaint,
|
|
aImgParams);
|
|
if (drawMode & DrawMode::GLYPH_STROKE) {
|
|
ctxSR.EnsureSaved(&aContext);
|
|
// This may change the gfxContext's transform (for non-scaling stroke),
|
|
// in which case this needs to happen before we call SetMatrix() below.
|
|
nsSVGUtils::SetupStrokeGeometry(frame, &aContext, outerContextPaint);
|
|
}
|
|
|
|
// Set up the transform for painting the text frame for the substring
|
|
// indicated by the run.
|
|
gfxMatrix runTransform =
|
|
run.GetTransformFromUserSpaceForPainting(presContext, item) *
|
|
currentMatrix;
|
|
aContext.SetMatrixDouble(runTransform);
|
|
|
|
if (drawMode != DrawMode(0)) {
|
|
bool paintSVGGlyphs;
|
|
nsTextFrame::PaintTextParams params(&aContext);
|
|
params.framePt = Point();
|
|
params.dirtyRect = LayoutDevicePixel::
|
|
FromAppUnits(frame->GetVisualOverflowRect(), auPerDevPx);
|
|
params.contextPaint = contextPaint;
|
|
if (ShouldRenderAsPath(frame, paintSVGGlyphs)) {
|
|
SVGTextDrawPathCallbacks callbacks(this, aContext, frame,
|
|
matrixForPaintServers,
|
|
aImgParams,
|
|
paintSVGGlyphs);
|
|
params.callbacks = &callbacks;
|
|
frame->PaintText(params, item);
|
|
} else {
|
|
frame->PaintText(params, item);
|
|
}
|
|
}
|
|
|
|
if (frame == caretFrame && ShouldPaintCaret(run, caret)) {
|
|
// XXX Should we be looking at the fill/stroke colours to paint the
|
|
// caret with, rather than using the color property?
|
|
caret->PaintCaret(aDrawTarget, frame, nsPoint());
|
|
aContext.NewPath();
|
|
}
|
|
|
|
run = it.Next();
|
|
}
|
|
}
|
|
|
|
nsIFrame*
|
|
SVGTextFrame::GetFrameForPoint(const gfxPoint& aPoint)
|
|
{
|
|
NS_ASSERTION(PrincipalChildList().FirstChild(), "must have a child frame");
|
|
|
|
if (mState & NS_FRAME_IS_NONDISPLAY) {
|
|
// Text frames inside <clipPath> will never have had ReflowSVG called on
|
|
// them, so call UpdateGlyphPositioning to do this now. (Text frames
|
|
// inside <mask> and other non-display containers will never need to
|
|
// be hit tested.)
|
|
UpdateGlyphPositioning();
|
|
} else {
|
|
NS_ASSERTION(!NS_SUBTREE_DIRTY(this), "reflow should have happened");
|
|
}
|
|
|
|
// Hit-testing any clip-path will typically be a lot quicker than the
|
|
// hit-testing of our text frames in the loop below, so we do the former up
|
|
// front to avoid unnecessarily wasting cycles on the latter.
|
|
if (!nsSVGUtils::HitTestClip(this, aPoint)) {
|
|
return nullptr;
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
// Ideally we'd iterate backwards so that we can just return the first frame
|
|
// that is under aPoint. In practice this will rarely matter though since it
|
|
// is rare for text in/under an SVG <text> element to overlap (i.e. the first
|
|
// text frame that is hit will likely be the only text frame that is hit).
|
|
|
|
TextRenderedRunIterator it(this);
|
|
nsIFrame* hit = nullptr;
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
uint16_t hitTestFlags = nsSVGUtils::GetGeometryHitTestFlags(run.mFrame);
|
|
if (!(hitTestFlags & (SVG_HIT_TEST_FILL | SVG_HIT_TEST_STROKE))) {
|
|
continue;
|
|
}
|
|
|
|
gfxMatrix m = run.GetTransformFromRunUserSpaceToUserSpace(presContext);
|
|
if (!m.Invert()) {
|
|
return nullptr;
|
|
}
|
|
|
|
gfxPoint pointInRunUserSpace = m.TransformPoint(aPoint);
|
|
gfxRect frameRect =
|
|
run.GetRunUserSpaceRect(presContext, TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeStroke).ToThebesRect();
|
|
|
|
if (Inside(frameRect, pointInRunUserSpace)) {
|
|
hit = run.mFrame;
|
|
}
|
|
}
|
|
return hit;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::ReflowSVG()
|
|
{
|
|
NS_ASSERTION(nsSVGUtils::OuterSVGIsCallingReflowSVG(this),
|
|
"This call is probaby a wasteful mistake");
|
|
|
|
MOZ_ASSERT(!(GetStateBits() & NS_FRAME_IS_NONDISPLAY),
|
|
"ReflowSVG mechanism not designed for this");
|
|
|
|
if (!nsSVGUtils::NeedsReflowSVG(this)) {
|
|
MOZ_ASSERT(!HasAnyStateBits(NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY |
|
|
NS_STATE_SVG_POSITIONING_DIRTY),
|
|
"How did this happen?");
|
|
return;
|
|
}
|
|
|
|
MaybeReflowAnonymousBlockChild();
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
SVGBBox r;
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eAllFrames);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
uint32_t runFlags = 0;
|
|
if (run.mFrame->StyleSVG()->mFill.Type() != eStyleSVGPaintType_None) {
|
|
runFlags |= TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeTextShadow;
|
|
}
|
|
if (nsSVGUtils::HasStroke(run.mFrame)) {
|
|
runFlags |= TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeTextShadow;
|
|
}
|
|
// Our "visual" overflow rect needs to be valid for building display lists
|
|
// for hit testing, which means that for certain values of 'pointer-events'
|
|
// it needs to include the geometry of the fill or stroke even when the fill/
|
|
// stroke don't actually render (e.g. when stroke="none" or
|
|
// stroke-opacity="0"). GetGeometryHitTestFlags accounts for 'pointer-events'.
|
|
// The text-shadow is not part of the hit-test area.
|
|
uint16_t hitTestFlags = nsSVGUtils::GetGeometryHitTestFlags(run.mFrame);
|
|
if (hitTestFlags & SVG_HIT_TEST_FILL) {
|
|
runFlags |= TextRenderedRun::eIncludeFill;
|
|
}
|
|
if (hitTestFlags & SVG_HIT_TEST_STROKE) {
|
|
runFlags |= TextRenderedRun::eIncludeStroke;
|
|
}
|
|
|
|
if (runFlags) {
|
|
r.UnionEdges(run.GetUserSpaceRect(presContext, runFlags));
|
|
}
|
|
}
|
|
|
|
if (r.IsEmpty()) {
|
|
mRect.SetEmpty();
|
|
} else {
|
|
mRect =
|
|
nsLayoutUtils::RoundGfxRectToAppRect(r.ToThebesRect(), nsPresContext::AppUnitsPerCSSPixel());
|
|
|
|
// Due to rounding issues when we have a transform applied, we sometimes
|
|
// don't include an additional row of pixels. For now, just inflate our
|
|
// covered region.
|
|
mRect.Inflate(presContext->AppUnitsPerDevPixel());
|
|
}
|
|
|
|
if (mState & NS_FRAME_FIRST_REFLOW) {
|
|
// Make sure we have our filter property (if any) before calling
|
|
// FinishAndStoreOverflow (subsequent filter changes are handled off
|
|
// nsChangeHint_UpdateEffects):
|
|
SVGObserverUtils::UpdateEffects(this);
|
|
}
|
|
|
|
// Now unset the various reflow bits. Do this before calling
|
|
// FinishAndStoreOverflow since FinishAndStoreOverflow can require glyph
|
|
// positions (to resolve transform-origin).
|
|
RemoveStateBits(NS_FRAME_FIRST_REFLOW | NS_FRAME_IS_DIRTY |
|
|
NS_FRAME_HAS_DIRTY_CHILDREN);
|
|
|
|
nsRect overflow = nsRect(nsPoint(0,0), mRect.Size());
|
|
nsOverflowAreas overflowAreas(overflow, overflow);
|
|
FinishAndStoreOverflow(overflowAreas, mRect.Size());
|
|
|
|
// XXX nsSVGContainerFrame::ReflowSVG only looks at its nsSVGDisplayableFrame
|
|
// children, and calls ConsiderChildOverflow on them. Does it matter
|
|
// that ConsiderChildOverflow won't be called on our children?
|
|
nsSVGDisplayContainerFrame::ReflowSVG();
|
|
}
|
|
|
|
/**
|
|
* Converts nsSVGUtils::eBBox* flags into TextRenderedRun flags appropriate
|
|
* for the specified rendered run.
|
|
*/
|
|
static uint32_t
|
|
TextRenderedRunFlagsForBBoxContribution(const TextRenderedRun& aRun,
|
|
uint32_t aBBoxFlags)
|
|
{
|
|
uint32_t flags = 0;
|
|
if ((aBBoxFlags & nsSVGUtils::eBBoxIncludeFillGeometry) ||
|
|
((aBBoxFlags & nsSVGUtils::eBBoxIncludeFill) &&
|
|
aRun.mFrame->StyleSVG()->mFill.Type() != eStyleSVGPaintType_None)) {
|
|
flags |= TextRenderedRun::eIncludeFill;
|
|
}
|
|
if ((aBBoxFlags & nsSVGUtils::eBBoxIncludeStrokeGeometry) ||
|
|
((aBBoxFlags & nsSVGUtils::eBBoxIncludeStroke) &&
|
|
nsSVGUtils::HasStroke(aRun.mFrame))) {
|
|
flags |= TextRenderedRun::eIncludeStroke;
|
|
}
|
|
return flags;
|
|
}
|
|
|
|
SVGBBox
|
|
SVGTextFrame::GetBBoxContribution(const Matrix &aToBBoxUserspace,
|
|
uint32_t aFlags)
|
|
{
|
|
NS_ASSERTION(PrincipalChildList().FirstChild(), "must have a child frame");
|
|
SVGBBox bbox;
|
|
|
|
if (aFlags & nsSVGUtils::eForGetClientRects) {
|
|
Rect rect = NSRectToRect(mRect, PresContext()->AppUnitsPerCSSPixel());
|
|
if (!rect.IsEmpty()) {
|
|
bbox = aToBBoxUserspace.TransformBounds(rect);
|
|
}
|
|
return bbox;
|
|
}
|
|
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (kid && NS_SUBTREE_DIRTY(kid)) {
|
|
// Return an empty bbox if our kid's subtree is dirty. This may be called
|
|
// in that situation, e.g. when we're building a display list after an
|
|
// interrupted reflow. This can also be called during reflow before we've
|
|
// been reflowed, e.g. if an earlier sibling is calling FinishAndStoreOverflow and
|
|
// needs our parent's perspective matrix, which depends on the SVG bbox
|
|
// contribution of this frame. In the latter situation, when all siblings have
|
|
// been reflowed, the parent will compute its perspective and rerun
|
|
// FinishAndStoreOverflow for all its children.
|
|
return bbox;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
TextRenderedRunIterator it(this);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
uint32_t flags = TextRenderedRunFlagsForBBoxContribution(run, aFlags);
|
|
gfxMatrix m = ThebesMatrix(aToBBoxUserspace);
|
|
SVGBBox bboxForRun =
|
|
run.GetUserSpaceRect(presContext, flags, &m);
|
|
bbox.UnionEdges(bboxForRun);
|
|
}
|
|
|
|
return bbox;
|
|
}
|
|
|
|
//----------------------------------------------------------------------
|
|
// SVGTextFrame SVG DOM methods
|
|
|
|
/**
|
|
* Returns whether the specified node has any non-empty nsTextNodes
|
|
* beneath it.
|
|
*/
|
|
static bool
|
|
HasTextContent(nsIContent* aContent)
|
|
{
|
|
NS_ASSERTION(aContent, "expected non-null aContent");
|
|
|
|
TextNodeIterator it(aContent);
|
|
for (nsTextNode* text = it.Current(); text; text = it.Next()) {
|
|
if (text->TextLength() != 0) {
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Returns the number of DOM characters beneath the specified node.
|
|
*/
|
|
static uint32_t
|
|
GetTextContentLength(nsIContent* aContent)
|
|
{
|
|
NS_ASSERTION(aContent, "expected non-null aContent");
|
|
|
|
uint32_t length = 0;
|
|
TextNodeIterator it(aContent);
|
|
for (nsTextNode* text = it.Current(); text; text = it.Next()) {
|
|
length += text->TextLength();
|
|
}
|
|
return length;
|
|
}
|
|
|
|
int32_t
|
|
SVGTextFrame::ConvertTextElementCharIndexToAddressableIndex(
|
|
int32_t aIndex,
|
|
nsIContent* aContent)
|
|
{
|
|
CharIterator it(this, CharIterator::eOriginal, aContent);
|
|
if (!it.AdvanceToSubtree()) {
|
|
return -1;
|
|
}
|
|
int32_t result = 0;
|
|
int32_t textElementCharIndex;
|
|
while (!it.AtEnd() &&
|
|
it.IsWithinSubtree()) {
|
|
bool addressable = !it.IsOriginalCharUnaddressable();
|
|
textElementCharIndex = it.TextElementCharIndex();
|
|
it.Next();
|
|
uint32_t delta = it.TextElementCharIndex() - textElementCharIndex;
|
|
aIndex -= delta;
|
|
if (addressable) {
|
|
if (aIndex < 0) {
|
|
return result;
|
|
}
|
|
result += delta;
|
|
}
|
|
}
|
|
return -1;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetNumberOfChars method for the specified
|
|
* text content element.
|
|
*/
|
|
uint32_t
|
|
SVGTextFrame::GetNumberOfChars(nsIContent* aContent)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return 0;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
uint32_t n = 0;
|
|
CharIterator it(this, CharIterator::eAddressable, aContent);
|
|
if (it.AdvanceToSubtree()) {
|
|
while (!it.AtEnd() && it.IsWithinSubtree()) {
|
|
n++;
|
|
it.Next();
|
|
}
|
|
}
|
|
return n;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetComputedTextLength method for the specified
|
|
* text child element.
|
|
*/
|
|
float
|
|
SVGTextFrame::GetComputedTextLength(nsIContent* aContent)
|
|
{
|
|
UpdateGlyphPositioning();
|
|
|
|
float cssPxPerDevPx = PresContext()->
|
|
AppUnitsToFloatCSSPixels(PresContext()->AppUnitsPerDevPixel());
|
|
|
|
nscoord length = 0;
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eAllFrames,
|
|
aContent);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
length += run.GetAdvanceWidth();
|
|
}
|
|
|
|
return PresContext()->AppUnitsToGfxUnits(length) *
|
|
cssPxPerDevPx * mLengthAdjustScaleFactor / mFontSizeScaleFactor;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM SelectSubString method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::SelectSubString(nsIContent* aContent,
|
|
uint32_t charnum, uint32_t nchars)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
// Convert charnum/nchars from addressable characters relative to
|
|
// aContent to global character indices.
|
|
CharIterator chit(this, CharIterator::eAddressable, aContent);
|
|
if (!chit.AdvanceToSubtree() ||
|
|
!chit.Next(charnum) ||
|
|
chit.IsAfterSubtree()) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
charnum = chit.TextElementCharIndex();
|
|
nsIContent* content = chit.TextFrame()->GetContent();
|
|
chit.NextWithinSubtree(nchars);
|
|
nchars = chit.TextElementCharIndex() - charnum;
|
|
|
|
RefPtr<nsFrameSelection> frameSelection = GetFrameSelection();
|
|
|
|
frameSelection->HandleClick(content, charnum, charnum + nchars,
|
|
false, false, CARET_ASSOCIATE_BEFORE);
|
|
return NS_OK;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetSubStringLength method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::GetSubStringLength(nsIContent* aContent,
|
|
uint32_t charnum, uint32_t nchars,
|
|
float* aResult)
|
|
{
|
|
// For some content we cannot (or currently cannot) compute the length
|
|
// without reflowing. In those cases we need to fall back to using
|
|
// GetSubStringLengthSlowFallback.
|
|
//
|
|
// We fall back for textPath since we need glyph positioning in order to
|
|
// tell if any characters should be ignored due to having fallen off the
|
|
// end of the textPath.
|
|
//
|
|
// We fall back for bidi because GetTrimmedOffsets does not produce the
|
|
// correct results for bidi continuations when passed aPostReflow = false.
|
|
// XXX It may be possible to determine which continuations to trim from (and
|
|
// which sides), but currently we don't do that. It would require us to
|
|
// identify the visual (rather than logical) start and end of the line, to
|
|
// avoid trimming at line-internal frame boundaries. Maybe nsBidiPresUtils
|
|
// methods like GetFrameToRightOf and GetFrameToLeftOf would help?
|
|
//
|
|
TextFrameIterator frameIter(this);
|
|
for (nsTextFrame* frame = frameIter.Current(); frame; frame = frameIter.Next()) {
|
|
if (frameIter.TextPathFrame() ||
|
|
frame->GetNextContinuation()) {
|
|
return GetSubStringLengthSlowFallback(aContent, charnum, nchars, aResult);
|
|
}
|
|
}
|
|
|
|
// We only need our text correspondence to be up to date (no need to call
|
|
// UpdateGlyphPositioning).
|
|
TextNodeCorrespondenceRecorder::RecordCorrespondence(this);
|
|
|
|
// Convert charnum/nchars from addressable characters relative to
|
|
// aContent to global character indices.
|
|
CharIterator chit(this, CharIterator::eAddressable, aContent,
|
|
/* aPostReflow */ false);
|
|
if (!chit.AdvanceToSubtree() ||
|
|
!chit.Next(charnum) ||
|
|
chit.IsAfterSubtree()) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
// We do this after the NS_ERROR_DOM_INDEX_SIZE_ERR return so JS calls
|
|
// correctly throw when necessary.
|
|
if (nchars == 0) {
|
|
*aResult = 0.0f;
|
|
return NS_OK;
|
|
}
|
|
|
|
charnum = chit.TextElementCharIndex();
|
|
chit.NextWithinSubtree(nchars);
|
|
nchars = chit.TextElementCharIndex() - charnum;
|
|
|
|
// Sum of the substring advances.
|
|
nscoord textLength = 0;
|
|
|
|
TextFrameIterator frit(this); // aSubtree = nullptr
|
|
|
|
// Index of the first non-skipped char in the frame, and of a subsequent char
|
|
// that we're interested in. Both are relative to the index of the first
|
|
// non-skipped char in the ancestor <text> element.
|
|
uint32_t frameStartTextElementCharIndex = 0;
|
|
uint32_t textElementCharIndex;
|
|
|
|
for (nsTextFrame* frame = frit.Current(); frame; frame = frit.Next()) {
|
|
frameStartTextElementCharIndex += frit.UndisplayedCharacters();
|
|
textElementCharIndex = frameStartTextElementCharIndex;
|
|
|
|
// Offset into frame's nsTextNode:
|
|
const uint32_t untrimmedOffset = frame->GetContentOffset();
|
|
const uint32_t untrimmedLength = frame->GetContentEnd() - untrimmedOffset;
|
|
|
|
// Trim the offset/length to remove any leading/trailing white space.
|
|
uint32_t trimmedOffset = untrimmedOffset;
|
|
uint32_t trimmedLength = untrimmedLength;
|
|
nsTextFrame::TrimmedOffsets trimmedOffsets =
|
|
frame->GetTrimmedOffsets(frame->GetContent()->GetText(),
|
|
/* aTrimAfter */ true,
|
|
/* aPostReflow */ false);
|
|
TrimOffsets(trimmedOffset, trimmedLength, trimmedOffsets);
|
|
|
|
textElementCharIndex += trimmedOffset - untrimmedOffset;
|
|
|
|
if (textElementCharIndex >= charnum + nchars) {
|
|
break; // we're past the end of the substring
|
|
}
|
|
|
|
uint32_t offset = textElementCharIndex;
|
|
|
|
// Intersect the substring we are interested in with the range covered by
|
|
// the nsTextFrame.
|
|
IntersectInterval(offset, trimmedLength, charnum, nchars);
|
|
|
|
if (trimmedLength != 0) {
|
|
// Convert offset into an index into the frame.
|
|
offset += trimmedOffset - textElementCharIndex;
|
|
|
|
gfxSkipCharsIterator skipCharsIter =
|
|
frame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = frame->GetTextRun(nsTextFrame::eInflated);
|
|
Range range =
|
|
ConvertOriginalToSkipped(skipCharsIter, offset, trimmedLength);
|
|
|
|
// Accumulate the advance.
|
|
textLength += textRun->GetAdvanceWidth(range, nullptr);
|
|
}
|
|
|
|
// Advance, ready for next call:
|
|
frameStartTextElementCharIndex += untrimmedLength;
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
|
|
*aResult = presContext->AppUnitsToGfxUnits(textLength) *
|
|
cssPxPerDevPx / mFontSizeScaleFactor;
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult
|
|
SVGTextFrame::GetSubStringLengthSlowFallback(nsIContent* aContent,
|
|
uint32_t charnum, uint32_t nchars,
|
|
float* aResult)
|
|
{
|
|
// We need to make sure that we've been reflowed before updating the glyph
|
|
// positioning.
|
|
// XXX perf: It may be possible to limit reflow to just calling ReflowSVG,
|
|
// but we would still need to resort to full reflow for percentage
|
|
// positioning attributes. For now we just do a full reflow regardless since
|
|
// the cases that would cause us to be called are relatively uncommon.
|
|
PresShell()->FlushPendingNotifications(FlushType::Layout);
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
// Convert charnum/nchars from addressable characters relative to
|
|
// aContent to global character indices.
|
|
CharIterator chit(this, CharIterator::eAddressable, aContent);
|
|
if (!chit.AdvanceToSubtree() ||
|
|
!chit.Next(charnum) ||
|
|
chit.IsAfterSubtree()) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
if (nchars == 0) {
|
|
*aResult = 0.0f;
|
|
return NS_OK;
|
|
}
|
|
|
|
charnum = chit.TextElementCharIndex();
|
|
chit.NextWithinSubtree(nchars);
|
|
nchars = chit.TextElementCharIndex() - charnum;
|
|
|
|
// Find each rendered run that intersects with the range defined
|
|
// by charnum/nchars.
|
|
nscoord textLength = 0;
|
|
TextRenderedRunIterator runIter(this, TextRenderedRunIterator::eAllFrames);
|
|
TextRenderedRun run = runIter.Current();
|
|
while (run.mFrame) {
|
|
// If this rendered run is past the substring we are interested in, we
|
|
// are done.
|
|
uint32_t offset = run.mTextElementCharIndex;
|
|
if (offset >= charnum + nchars) {
|
|
break;
|
|
}
|
|
|
|
// Intersect the substring we are interested in with the range covered by
|
|
// the rendered run.
|
|
uint32_t length = run.mTextFrameContentLength;
|
|
IntersectInterval(offset, length, charnum, nchars);
|
|
|
|
if (length != 0) {
|
|
// Convert offset into an index into the frame.
|
|
offset += run.mTextFrameContentOffset - run.mTextElementCharIndex;
|
|
|
|
gfxSkipCharsIterator skipCharsIter =
|
|
run.mFrame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = run.mFrame->GetTextRun(nsTextFrame::eInflated);
|
|
Range range = ConvertOriginalToSkipped(skipCharsIter, offset, length);
|
|
|
|
// Accumulate the advance.
|
|
textLength += textRun->GetAdvanceWidth(range, nullptr);
|
|
}
|
|
|
|
run = runIter.Next();
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
|
|
*aResult = presContext->AppUnitsToGfxUnits(textLength) *
|
|
cssPxPerDevPx / mFontSizeScaleFactor;
|
|
return NS_OK;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetCharNumAtPosition method for the specified
|
|
* text content element.
|
|
*/
|
|
int32_t
|
|
SVGTextFrame::GetCharNumAtPosition(nsIContent* aContent,
|
|
nsISVGPoint* aPoint)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return -1;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* context = PresContext();
|
|
|
|
gfxPoint p(aPoint->X(), aPoint->Y());
|
|
|
|
int32_t result = -1;
|
|
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eAllFrames, aContent);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
// Hit test this rendered run. Later runs will override earlier ones.
|
|
int32_t index = run.GetCharNumAtPosition(context, p);
|
|
if (index != -1) {
|
|
result = index + run.mTextElementCharIndex;
|
|
}
|
|
}
|
|
|
|
if (result == -1) {
|
|
return result;
|
|
}
|
|
|
|
return ConvertTextElementCharIndexToAddressableIndex(result, aContent);
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetStartPositionOfChar method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::GetStartPositionOfChar(nsIContent* aContent,
|
|
uint32_t aCharNum,
|
|
nsISVGPoint** aResult)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
CharIterator it(this, CharIterator::eAddressable, aContent);
|
|
if (!it.AdvanceToSubtree() ||
|
|
!it.Next(aCharNum)) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
// We need to return the start position of the whole glyph.
|
|
uint32_t startIndex = it.GlyphStartTextElementCharIndex();
|
|
|
|
RefPtr<DOMSVGPoint> point =
|
|
new DOMSVGPoint(ToPoint(mPositions[startIndex].mPosition));
|
|
point.forget(aResult);
|
|
return NS_OK;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetEndPositionOfChar method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::GetEndPositionOfChar(nsIContent* aContent,
|
|
uint32_t aCharNum,
|
|
nsISVGPoint** aResult)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
CharIterator it(this, CharIterator::eAddressable, aContent);
|
|
if (!it.AdvanceToSubtree() ||
|
|
!it.Next(aCharNum)) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
// We need to return the end position of the whole glyph.
|
|
uint32_t startIndex = it.GlyphStartTextElementCharIndex();
|
|
|
|
// Get the advance of the glyph.
|
|
gfxFloat advance = it.GetGlyphAdvance(PresContext());
|
|
if (it.TextRun()->IsRightToLeft()) {
|
|
advance = -advance;
|
|
}
|
|
|
|
// The end position is the start position plus the advance in the direction
|
|
// of the glyph's rotation.
|
|
Matrix m =
|
|
Matrix::Rotation(mPositions[startIndex].mAngle) *
|
|
Matrix::Translation(ToPoint(mPositions[startIndex].mPosition));
|
|
Point p = m.TransformPoint(Point(advance / mFontSizeScaleFactor, 0));
|
|
|
|
RefPtr<DOMSVGPoint> point = new DOMSVGPoint(p);
|
|
point.forget(aResult);
|
|
return NS_OK;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetExtentOfChar method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::GetExtentOfChar(nsIContent* aContent,
|
|
uint32_t aCharNum,
|
|
SVGIRect** aResult)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
CharIterator it(this, CharIterator::eAddressable, aContent);
|
|
if (!it.AdvanceToSubtree() ||
|
|
!it.Next(aCharNum)) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
|
|
// We need to return the extent of the whole glyph.
|
|
uint32_t startIndex = it.GlyphStartTextElementCharIndex();
|
|
|
|
// The ascent and descent gives the height of the glyph.
|
|
gfxFloat ascent, descent;
|
|
GetAscentAndDescentInAppUnits(it.TextFrame(), ascent, descent);
|
|
|
|
// Get the advance of the glyph.
|
|
gfxFloat advance = it.GetGlyphAdvance(presContext);
|
|
gfxFloat x = it.TextRun()->IsRightToLeft() ? -advance : 0.0;
|
|
|
|
// The horizontal extent is the origin of the glyph plus the advance
|
|
// in the direction of the glyph's rotation.
|
|
gfxMatrix m;
|
|
m.PreTranslate(mPositions[startIndex].mPosition);
|
|
m.PreRotate(mPositions[startIndex].mAngle);
|
|
m.PreScale(1 / mFontSizeScaleFactor, 1 / mFontSizeScaleFactor);
|
|
|
|
gfxRect glyphRect;
|
|
if (it.TextRun()->IsVertical()) {
|
|
glyphRect =
|
|
gfxRect(-presContext->AppUnitsToGfxUnits(descent) * cssPxPerDevPx, x,
|
|
presContext->AppUnitsToGfxUnits(ascent + descent) * cssPxPerDevPx,
|
|
advance);
|
|
} else {
|
|
glyphRect =
|
|
gfxRect(x, -presContext->AppUnitsToGfxUnits(ascent) * cssPxPerDevPx,
|
|
advance,
|
|
presContext->AppUnitsToGfxUnits(ascent + descent) * cssPxPerDevPx);
|
|
}
|
|
|
|
// Transform the glyph's rect into user space.
|
|
gfxRect r = m.TransformBounds(glyphRect);
|
|
|
|
RefPtr<SVGRect> rect = new SVGRect(aContent, r.x, r.y, r.width, r.height);
|
|
rect.forget(aResult);
|
|
return NS_OK;
|
|
}
|
|
|
|
/**
|
|
* Implements the SVG DOM GetRotationOfChar method for the specified
|
|
* text content element.
|
|
*/
|
|
nsresult
|
|
SVGTextFrame::GetRotationOfChar(nsIContent* aContent,
|
|
uint32_t aCharNum,
|
|
float* aResult)
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (NS_SUBTREE_DIRTY(kid)) {
|
|
// We're never reflowed if we're under a non-SVG element that is
|
|
// never reflowed (such as the HTML 'caption' element).
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
CharIterator it(this, CharIterator::eAddressable, aContent);
|
|
if (!it.AdvanceToSubtree() ||
|
|
!it.Next(aCharNum)) {
|
|
return NS_ERROR_DOM_INDEX_SIZE_ERR;
|
|
}
|
|
|
|
*aResult = mPositions[it.TextElementCharIndex()].mAngle * 180.0 / M_PI;
|
|
return NS_OK;
|
|
}
|
|
|
|
//----------------------------------------------------------------------
|
|
// SVGTextFrame text layout methods
|
|
|
|
/**
|
|
* Given the character position array before values have been filled in
|
|
* to any unspecified positions, and an array of dx/dy values, returns whether
|
|
* a character at a given index should start a new rendered run.
|
|
*
|
|
* @param aPositions The array of character positions before unspecified
|
|
* positions have been filled in and dx/dy values have been added to them.
|
|
* @param aDeltas The array of dx/dy values.
|
|
* @param aIndex The character index in question.
|
|
*/
|
|
static bool
|
|
ShouldStartRunAtIndex(const nsTArray<CharPosition>& aPositions,
|
|
const nsTArray<gfxPoint>& aDeltas,
|
|
uint32_t aIndex)
|
|
{
|
|
if (aIndex == 0) {
|
|
return true;
|
|
}
|
|
|
|
if (aIndex < aPositions.Length()) {
|
|
// If an explicit x or y value was given, start a new run.
|
|
if (aPositions[aIndex].IsXSpecified() ||
|
|
aPositions[aIndex].IsYSpecified()) {
|
|
return true;
|
|
}
|
|
|
|
// If a non-zero rotation was given, or the previous character had a non-
|
|
// zero rotation, start a new run.
|
|
if ((aPositions[aIndex].IsAngleSpecified() &&
|
|
aPositions[aIndex].mAngle != 0.0f) ||
|
|
(aPositions[aIndex - 1].IsAngleSpecified() &&
|
|
(aPositions[aIndex - 1].mAngle != 0.0f))) {
|
|
return true;
|
|
}
|
|
}
|
|
|
|
if (aIndex < aDeltas.Length()) {
|
|
// If a non-zero dx or dy value was given, start a new run.
|
|
if (aDeltas[aIndex].x != 0.0 ||
|
|
aDeltas[aIndex].y != 0.0) {
|
|
return true;
|
|
}
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
bool
|
|
SVGTextFrame::ResolvePositionsForNode(nsIContent* aContent,
|
|
uint32_t& aIndex,
|
|
bool aInTextPath,
|
|
bool& aForceStartOfChunk,
|
|
nsTArray<gfxPoint>& aDeltas)
|
|
{
|
|
if (aContent->IsText()) {
|
|
// We found a text node.
|
|
uint32_t length = static_cast<nsTextNode*>(aContent)->TextLength();
|
|
if (length) {
|
|
uint32_t end = aIndex + length;
|
|
if (MOZ_UNLIKELY(end > mPositions.Length())) {
|
|
MOZ_ASSERT_UNREACHABLE("length of mPositions does not match characters "
|
|
"found by iterating content");
|
|
return false;
|
|
}
|
|
if (aForceStartOfChunk) {
|
|
// Note this character as starting a new anchored chunk.
|
|
mPositions[aIndex].mStartOfChunk = true;
|
|
aForceStartOfChunk = false;
|
|
}
|
|
while (aIndex < end) {
|
|
// Record whether each of these characters should start a new rendered
|
|
// run. That is always the case for characters on a text path.
|
|
//
|
|
// Run boundaries due to rotate="" values are handled in
|
|
// DoGlyphPositioning.
|
|
if (aInTextPath || ShouldStartRunAtIndex(mPositions, aDeltas, aIndex)) {
|
|
mPositions[aIndex].mRunBoundary = true;
|
|
}
|
|
aIndex++;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
// Skip past elements that aren't text content elements.
|
|
if (!IsTextContentElement(aContent)) {
|
|
return true;
|
|
}
|
|
|
|
if (aContent->IsSVGElement(nsGkAtoms::textPath)) {
|
|
// <textPath> elements behave as if they have x="0" y="0" on them, but only
|
|
// if there is not a value for the coordinates that got inherited from a
|
|
// parent. We skip this if there is no text content, so that empty
|
|
// <textPath>s don't interrupt the layout of text in the parent element.
|
|
if (HasTextContent(aContent)) {
|
|
if (MOZ_UNLIKELY(aIndex >= mPositions.Length())) {
|
|
MOZ_ASSERT_UNREACHABLE("length of mPositions does not match characters "
|
|
"found by iterating content");
|
|
return false;
|
|
}
|
|
if (!mPositions[aIndex].IsXSpecified()) {
|
|
mPositions[aIndex].mPosition.x = 0.0;
|
|
}
|
|
if (!mPositions[aIndex].IsYSpecified()) {
|
|
mPositions[aIndex].mPosition.y = 0.0;
|
|
}
|
|
mPositions[aIndex].mStartOfChunk = true;
|
|
}
|
|
} else if (!aContent->IsSVGElement(nsGkAtoms::a)) {
|
|
MOZ_ASSERT(aContent->IsSVGElement());
|
|
|
|
// We have a text content element that can have x/y/dx/dy/rotate attributes.
|
|
nsSVGElement* element = static_cast<nsSVGElement*>(aContent);
|
|
|
|
// Get x, y, dx, dy.
|
|
SVGUserUnitList x, y, dx, dy;
|
|
element->GetAnimatedLengthListValues(&x, &y, &dx, &dy, nullptr);
|
|
|
|
// Get rotate.
|
|
const SVGNumberList* rotate = nullptr;
|
|
SVGAnimatedNumberList* animatedRotate =
|
|
element->GetAnimatedNumberList(nsGkAtoms::rotate);
|
|
if (animatedRotate) {
|
|
rotate = &animatedRotate->GetAnimValue();
|
|
}
|
|
|
|
bool percentages = false;
|
|
uint32_t count = GetTextContentLength(aContent);
|
|
|
|
if (MOZ_UNLIKELY(aIndex + count > mPositions.Length())) {
|
|
MOZ_ASSERT_UNREACHABLE("length of mPositions does not match characters "
|
|
"found by iterating content");
|
|
return false;
|
|
}
|
|
|
|
// New text anchoring chunks start at each character assigned a position
|
|
// with x="" or y="", or if we forced one with aForceStartOfChunk due to
|
|
// being just after a <textPath>.
|
|
uint32_t newChunkCount = std::max(x.Length(), y.Length());
|
|
if (!newChunkCount && aForceStartOfChunk) {
|
|
newChunkCount = 1;
|
|
}
|
|
for (uint32_t i = 0, j = 0; i < newChunkCount && j < count; j++) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
mPositions[aIndex + j].mStartOfChunk = true;
|
|
i++;
|
|
}
|
|
}
|
|
|
|
// Copy dx="" and dy="" values into aDeltas.
|
|
if (!dx.IsEmpty() || !dy.IsEmpty()) {
|
|
// Any unspecified deltas when we grow the array just get left as 0s.
|
|
aDeltas.EnsureLengthAtLeast(aIndex + count);
|
|
for (uint32_t i = 0, j = 0; i < dx.Length() && j < count; j++) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
aDeltas[aIndex + j].x = dx[i];
|
|
percentages = percentages || dx.HasPercentageValueAt(i);
|
|
i++;
|
|
}
|
|
}
|
|
for (uint32_t i = 0, j = 0; i < dy.Length() && j < count; j++) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
aDeltas[aIndex + j].y = dy[i];
|
|
percentages = percentages || dy.HasPercentageValueAt(i);
|
|
i++;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Copy x="" and y="" values.
|
|
for (uint32_t i = 0, j = 0; i < x.Length() && j < count; j++) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
mPositions[aIndex + j].mPosition.x = x[i];
|
|
percentages = percentages || x.HasPercentageValueAt(i);
|
|
i++;
|
|
}
|
|
}
|
|
for (uint32_t i = 0, j = 0; i < y.Length() && j < count; j++) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
mPositions[aIndex + j].mPosition.y = y[i];
|
|
percentages = percentages || y.HasPercentageValueAt(i);
|
|
i++;
|
|
}
|
|
}
|
|
|
|
// Copy rotate="" values.
|
|
if (rotate && !rotate->IsEmpty()) {
|
|
uint32_t i = 0, j = 0;
|
|
while (i < rotate->Length() && j < count) {
|
|
if (!mPositions[aIndex + j].mUnaddressable) {
|
|
mPositions[aIndex + j].mAngle = M_PI * (*rotate)[i] / 180.0;
|
|
i++;
|
|
}
|
|
j++;
|
|
}
|
|
// Propagate final rotate="" value to the end of this element.
|
|
while (j < count) {
|
|
mPositions[aIndex + j].mAngle = mPositions[aIndex + j - 1].mAngle;
|
|
j++;
|
|
}
|
|
}
|
|
|
|
if (percentages) {
|
|
AddStateBits(NS_STATE_SVG_POSITIONING_MAY_USE_PERCENTAGES);
|
|
}
|
|
}
|
|
|
|
// Recurse to children.
|
|
bool inTextPath = aInTextPath || aContent->IsSVGElement(nsGkAtoms::textPath);
|
|
for (nsIContent* child = aContent->GetFirstChild();
|
|
child;
|
|
child = child->GetNextSibling()) {
|
|
bool ok = ResolvePositionsForNode(child, aIndex, inTextPath,
|
|
aForceStartOfChunk, aDeltas);
|
|
if (!ok) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
if (aContent->IsSVGElement(nsGkAtoms::textPath)) {
|
|
// Force a new anchored chunk just after a <textPath>.
|
|
aForceStartOfChunk = true;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool
|
|
SVGTextFrame::ResolvePositions(nsTArray<gfxPoint>& aDeltas,
|
|
bool aRunPerGlyph)
|
|
{
|
|
NS_ASSERTION(mPositions.IsEmpty(), "expected mPositions to be empty");
|
|
RemoveStateBits(NS_STATE_SVG_POSITIONING_MAY_USE_PERCENTAGES);
|
|
|
|
CharIterator it(this, CharIterator::eOriginal, /* aSubtree */ nullptr);
|
|
if (it.AtEnd()) {
|
|
return false;
|
|
}
|
|
|
|
// We assume the first character position is (0,0) unless we later see
|
|
// otherwise, and note it as unaddressable if it is.
|
|
bool firstCharUnaddressable = it.IsOriginalCharUnaddressable();
|
|
mPositions.AppendElement(CharPosition::Unspecified(firstCharUnaddressable));
|
|
|
|
// Fill in unspecified positions for all remaining characters, noting
|
|
// them as unaddressable if they are.
|
|
uint32_t index = 0;
|
|
while (it.Next()) {
|
|
while (++index < it.TextElementCharIndex()) {
|
|
mPositions.AppendElement(CharPosition::Unspecified(false));
|
|
}
|
|
mPositions.AppendElement(CharPosition::Unspecified(
|
|
it.IsOriginalCharUnaddressable()));
|
|
}
|
|
while (++index < it.TextElementCharIndex()) {
|
|
mPositions.AppendElement(CharPosition::Unspecified(false));
|
|
}
|
|
|
|
// Recurse over the content and fill in character positions as we go.
|
|
bool forceStartOfChunk = false;
|
|
index = 0;
|
|
bool ok = ResolvePositionsForNode(mContent, index, aRunPerGlyph,
|
|
forceStartOfChunk, aDeltas);
|
|
return ok && index > 0;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DetermineCharPositions(nsTArray<nsPoint>& aPositions)
|
|
{
|
|
NS_ASSERTION(aPositions.IsEmpty(), "expected aPositions to be empty");
|
|
|
|
nsPoint position, lastPosition;
|
|
|
|
TextFrameIterator frit(this);
|
|
for (nsTextFrame* frame = frit.Current(); frame; frame = frit.Next()) {
|
|
gfxSkipCharsIterator it = frame->EnsureTextRun(nsTextFrame::eInflated);
|
|
gfxTextRun* textRun = frame->GetTextRun(nsTextFrame::eInflated);
|
|
|
|
// Reset the position to the new frame's position.
|
|
position = frit.Position();
|
|
if (textRun->IsVertical()) {
|
|
if (textRun->IsRightToLeft()) {
|
|
position.y += frame->GetRect().height;
|
|
}
|
|
position.x += GetBaselinePosition(frame, textRun,
|
|
frit.DominantBaseline(),
|
|
mFontSizeScaleFactor);
|
|
} else {
|
|
if (textRun->IsRightToLeft()) {
|
|
position.x += frame->GetRect().width;
|
|
}
|
|
position.y += GetBaselinePosition(frame, textRun,
|
|
frit.DominantBaseline(),
|
|
mFontSizeScaleFactor);
|
|
}
|
|
|
|
// Any characters not in a frame, e.g. when display:none.
|
|
for (uint32_t i = 0; i < frit.UndisplayedCharacters(); i++) {
|
|
aPositions.AppendElement(position);
|
|
}
|
|
|
|
// Any white space characters trimmed at the start of the line of text.
|
|
nsTextFrame::TrimmedOffsets trimmedOffsets =
|
|
frame->GetTrimmedOffsets(frame->GetContent()->GetText(), true);
|
|
while (it.GetOriginalOffset() < trimmedOffsets.mStart) {
|
|
aPositions.AppendElement(position);
|
|
it.AdvanceOriginal(1);
|
|
}
|
|
|
|
// If a ligature was started in the previous frame, we should record
|
|
// the ligature's start position, not any partial position.
|
|
while (it.GetOriginalOffset() < frame->GetContentEnd() &&
|
|
!it.IsOriginalCharSkipped() &&
|
|
(!textRun->IsLigatureGroupStart(it.GetSkippedOffset()) ||
|
|
!textRun->IsClusterStart(it.GetSkippedOffset()))) {
|
|
uint32_t offset = it.GetSkippedOffset();
|
|
nscoord advance = textRun->
|
|
GetAdvanceWidth(Range(offset, offset + 1), nullptr);
|
|
(textRun->IsVertical() ? position.y : position.x) +=
|
|
textRun->IsRightToLeft() ? -advance : advance;
|
|
aPositions.AppendElement(lastPosition);
|
|
it.AdvanceOriginal(1);
|
|
}
|
|
|
|
// The meat of the text frame.
|
|
while (it.GetOriginalOffset() < frame->GetContentEnd()) {
|
|
aPositions.AppendElement(position);
|
|
if (!it.IsOriginalCharSkipped() &&
|
|
textRun->IsLigatureGroupStart(it.GetSkippedOffset()) &&
|
|
textRun->IsClusterStart(it.GetSkippedOffset())) {
|
|
// A real visible character.
|
|
nscoord advance = textRun->
|
|
GetAdvanceWidth(ClusterRange(textRun, it), nullptr);
|
|
(textRun->IsVertical() ? position.y : position.x) +=
|
|
textRun->IsRightToLeft() ? -advance : advance;
|
|
lastPosition = position;
|
|
}
|
|
it.AdvanceOriginal(1);
|
|
}
|
|
}
|
|
|
|
// Finally any characters at the end that are not in a frame.
|
|
for (uint32_t i = 0; i < frit.UndisplayedCharacters(); i++) {
|
|
aPositions.AppendElement(position);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Physical text-anchor values.
|
|
*/
|
|
enum TextAnchorSide {
|
|
eAnchorLeft,
|
|
eAnchorMiddle,
|
|
eAnchorRight
|
|
};
|
|
|
|
/**
|
|
* Converts a logical text-anchor value to its physical value, based on whether
|
|
* it is for an RTL frame.
|
|
*/
|
|
static TextAnchorSide
|
|
ConvertLogicalTextAnchorToPhysical(uint8_t aTextAnchor, bool aIsRightToLeft)
|
|
{
|
|
NS_ASSERTION(aTextAnchor <= 3, "unexpected value for aTextAnchor");
|
|
if (!aIsRightToLeft)
|
|
return TextAnchorSide(aTextAnchor);
|
|
return TextAnchorSide(2 - aTextAnchor);
|
|
}
|
|
|
|
/**
|
|
* Shifts the recorded character positions for an anchored chunk.
|
|
*
|
|
* @param aCharPositions The recorded character positions.
|
|
* @param aChunkStart The character index the starts the anchored chunk. This
|
|
* character's initial position is the anchor point.
|
|
* @param aChunkEnd The character index just after the end of the anchored
|
|
* chunk.
|
|
* @param aVisIStartEdge The left/top-most edge of any of the glyphs within the
|
|
* anchored chunk.
|
|
* @param aVisIEndEdge The right/bottom-most edge of any of the glyphs within
|
|
* the anchored chunk.
|
|
* @param aAnchorSide The direction to anchor.
|
|
*/
|
|
static void
|
|
ShiftAnchoredChunk(nsTArray<CharPosition>& aCharPositions,
|
|
uint32_t aChunkStart,
|
|
uint32_t aChunkEnd,
|
|
gfxFloat aVisIStartEdge,
|
|
gfxFloat aVisIEndEdge,
|
|
TextAnchorSide aAnchorSide,
|
|
bool aVertical)
|
|
{
|
|
NS_ASSERTION(aVisIStartEdge <= aVisIEndEdge,
|
|
"unexpected anchored chunk edges");
|
|
NS_ASSERTION(aChunkStart < aChunkEnd,
|
|
"unexpected values for aChunkStart and aChunkEnd");
|
|
|
|
gfxFloat shift = aVertical ? aCharPositions[aChunkStart].mPosition.y
|
|
: aCharPositions[aChunkStart].mPosition.x;
|
|
switch (aAnchorSide) {
|
|
case eAnchorLeft:
|
|
shift -= aVisIStartEdge;
|
|
break;
|
|
case eAnchorMiddle:
|
|
shift -= (aVisIStartEdge + aVisIEndEdge) / 2;
|
|
break;
|
|
case eAnchorRight:
|
|
shift -= aVisIEndEdge;
|
|
break;
|
|
default:
|
|
MOZ_ASSERT_UNREACHABLE("unexpected value for aAnchorSide");
|
|
}
|
|
|
|
if (shift != 0.0) {
|
|
if (aVertical) {
|
|
for (uint32_t i = aChunkStart; i < aChunkEnd; i++) {
|
|
aCharPositions[i].mPosition.y += shift;
|
|
}
|
|
} else {
|
|
for (uint32_t i = aChunkStart; i < aChunkEnd; i++) {
|
|
aCharPositions[i].mPosition.x += shift;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::AdjustChunksForLineBreaks()
|
|
{
|
|
nsBlockFrame* block = nsLayoutUtils::GetAsBlock(PrincipalChildList().FirstChild());
|
|
NS_ASSERTION(block, "expected block frame");
|
|
|
|
nsBlockFrame::LineIterator line = block->LinesBegin();
|
|
|
|
CharIterator it(this, CharIterator::eOriginal, /* aSubtree */ nullptr);
|
|
while (!it.AtEnd() && line != block->LinesEnd()) {
|
|
if (it.TextFrame() == line->mFirstChild) {
|
|
mPositions[it.TextElementCharIndex()].mStartOfChunk = true;
|
|
line++;
|
|
}
|
|
it.AdvancePastCurrentFrame();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::AdjustPositionsForClusters()
|
|
{
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
CharIterator it(this, CharIterator::eClusterOrLigatureGroupMiddle,
|
|
/* aSubtree */ nullptr);
|
|
while (!it.AtEnd()) {
|
|
// Find the start of the cluster/ligature group.
|
|
uint32_t charIndex = it.TextElementCharIndex();
|
|
uint32_t startIndex = it.GlyphStartTextElementCharIndex();
|
|
|
|
mPositions[charIndex].mClusterOrLigatureGroupMiddle = true;
|
|
|
|
// Don't allow different rotations on ligature parts.
|
|
bool rotationAdjusted = false;
|
|
double angle = mPositions[startIndex].mAngle;
|
|
if (mPositions[charIndex].mAngle != angle) {
|
|
mPositions[charIndex].mAngle = angle;
|
|
rotationAdjusted = true;
|
|
}
|
|
|
|
// Find out the partial glyph advance for this character and update
|
|
// the character position.
|
|
uint32_t partLength =
|
|
charIndex - startIndex - it.GlyphUndisplayedCharacters();
|
|
gfxFloat advance =
|
|
it.GetGlyphPartialAdvance(partLength, presContext) / mFontSizeScaleFactor;
|
|
gfxPoint direction = gfxPoint(cos(angle), sin(angle)) *
|
|
(it.TextRun()->IsRightToLeft() ? -1.0 : 1.0);
|
|
if (it.TextRun()->IsVertical()) {
|
|
Swap(direction.x, direction.y);
|
|
}
|
|
mPositions[charIndex].mPosition = mPositions[startIndex].mPosition +
|
|
direction * advance;
|
|
|
|
// Ensure any runs that would end in the middle of a ligature now end just
|
|
// after the ligature.
|
|
if (mPositions[charIndex].mRunBoundary) {
|
|
mPositions[charIndex].mRunBoundary = false;
|
|
if (charIndex + 1 < mPositions.Length()) {
|
|
mPositions[charIndex + 1].mRunBoundary = true;
|
|
}
|
|
} else if (rotationAdjusted) {
|
|
if (charIndex + 1 < mPositions.Length()) {
|
|
mPositions[charIndex + 1].mRunBoundary = true;
|
|
}
|
|
}
|
|
|
|
// Ensure any anchored chunks that would begin in the middle of a ligature
|
|
// now begin just after the ligature.
|
|
if (mPositions[charIndex].mStartOfChunk) {
|
|
mPositions[charIndex].mStartOfChunk = false;
|
|
if (charIndex + 1 < mPositions.Length()) {
|
|
mPositions[charIndex + 1].mStartOfChunk = true;
|
|
}
|
|
}
|
|
|
|
it.Next();
|
|
}
|
|
}
|
|
|
|
SVGGeometryElement*
|
|
SVGTextFrame::GetTextPathGeometryElement(nsIFrame* aTextPathFrame)
|
|
{
|
|
nsSVGTextPathProperty *property =
|
|
aTextPathFrame->GetProperty(SVGObserverUtils::HrefAsTextPathProperty());
|
|
|
|
if (!property) {
|
|
nsIContent* content = aTextPathFrame->GetContent();
|
|
SVGTextPathElement* tp = static_cast<SVGTextPathElement*>(content);
|
|
nsAutoString href;
|
|
if (tp->mStringAttributes[SVGTextPathElement::HREF].IsExplicitlySet()) {
|
|
tp->mStringAttributes[SVGTextPathElement::HREF]
|
|
.GetAnimValue(href, tp);
|
|
} else {
|
|
tp->mStringAttributes[SVGTextPathElement::XLINK_HREF]
|
|
.GetAnimValue(href, tp);
|
|
}
|
|
|
|
if (href.IsEmpty()) {
|
|
return nullptr; // no URL
|
|
}
|
|
|
|
nsCOMPtr<nsIURI> targetURI;
|
|
nsCOMPtr<nsIURI> base = content->GetBaseURI();
|
|
nsContentUtils::NewURIWithDocumentCharset(getter_AddRefs(targetURI), href,
|
|
content->GetUncomposedDoc(), base);
|
|
|
|
property = SVGObserverUtils::GetTextPathProperty(
|
|
targetURI,
|
|
aTextPathFrame,
|
|
SVGObserverUtils::HrefAsTextPathProperty());
|
|
if (!property)
|
|
return nullptr;
|
|
}
|
|
|
|
Element* element = property->GetReferencedElement();
|
|
return (element && element->IsNodeOfType(nsINode::eSHAPE)) ?
|
|
static_cast<SVGGeometryElement*>(element) : nullptr;
|
|
}
|
|
|
|
already_AddRefed<Path>
|
|
SVGTextFrame::GetTextPath(nsIFrame* aTextPathFrame)
|
|
{
|
|
nsIContent* content = aTextPathFrame->GetContent();
|
|
SVGTextPathElement* tp = static_cast<SVGTextPathElement*>(content);
|
|
if (tp->mPath.IsRendered()) {
|
|
// This is just an attribute so there's no transform that can apply
|
|
// so we can just return the path directly.
|
|
return tp->mPath.GetAnimValue().BuildPathForMeasuring();
|
|
}
|
|
|
|
SVGGeometryElement* geomElement = GetTextPathGeometryElement(aTextPathFrame);
|
|
if (!geomElement) {
|
|
return nullptr;
|
|
}
|
|
|
|
RefPtr<Path> path = geomElement->GetOrBuildPathForMeasuring();
|
|
if (!path) {
|
|
return nullptr;
|
|
}
|
|
|
|
gfxMatrix matrix = geomElement->PrependLocalTransformsTo(gfxMatrix());
|
|
if (!matrix.IsIdentity()) {
|
|
// Apply the geometry element's transform
|
|
RefPtr<PathBuilder> builder =
|
|
path->TransformedCopyToBuilder(ToMatrix(matrix));
|
|
path = builder->Finish();
|
|
}
|
|
|
|
return path.forget();
|
|
}
|
|
|
|
gfxFloat
|
|
SVGTextFrame::GetOffsetScale(nsIFrame* aTextPathFrame)
|
|
{
|
|
nsIContent* content = aTextPathFrame->GetContent();
|
|
SVGTextPathElement* tp = static_cast<SVGTextPathElement*>(content);
|
|
if (tp->mPath.IsRendered()) {
|
|
// A path attribute has no pathLength or transform
|
|
// so we return a unit scale.
|
|
return 1.0;
|
|
}
|
|
|
|
SVGGeometryElement* geomElement = GetTextPathGeometryElement(aTextPathFrame);
|
|
if (!geomElement)
|
|
return 1.0;
|
|
|
|
return geomElement->GetPathLengthScale(SVGGeometryElement::eForTextPath);
|
|
}
|
|
|
|
gfxFloat
|
|
SVGTextFrame::GetStartOffset(nsIFrame* aTextPathFrame)
|
|
{
|
|
SVGTextPathElement *tp =
|
|
static_cast<SVGTextPathElement*>(aTextPathFrame->GetContent());
|
|
nsSVGLength2 *length =
|
|
&tp->mLengthAttributes[SVGTextPathElement::STARTOFFSET];
|
|
|
|
if (length->IsPercentage()) {
|
|
RefPtr<Path> data = GetTextPath(aTextPathFrame);
|
|
return data ?
|
|
length->GetAnimValInSpecifiedUnits() * data->ComputeLength() / 100.0 :
|
|
0.0;
|
|
}
|
|
return length->GetAnimValue(tp) * GetOffsetScale(aTextPathFrame);
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DoTextPathLayout()
|
|
{
|
|
nsPresContext* context = PresContext();
|
|
|
|
CharIterator it(this, CharIterator::eClusterAndLigatureGroupStart,
|
|
/* aSubtree */ nullptr);
|
|
while (!it.AtEnd()) {
|
|
nsIFrame* textPathFrame = it.TextPathFrame();
|
|
if (!textPathFrame) {
|
|
// Skip past this frame if we're not in a text path.
|
|
it.AdvancePastCurrentFrame();
|
|
continue;
|
|
}
|
|
|
|
// Get the path itself.
|
|
RefPtr<Path> path = GetTextPath(textPathFrame);
|
|
if (!path) {
|
|
uint32_t start = it.TextElementCharIndex();
|
|
it.AdvancePastCurrentTextPathFrame();
|
|
uint32_t end = it.TextElementCharIndex();
|
|
for (uint32_t i = start; i < end; i++) {
|
|
mPositions[i].mHidden = true;
|
|
}
|
|
continue;
|
|
}
|
|
|
|
SVGTextPathElement* textPath =
|
|
static_cast<SVGTextPathElement*>(textPathFrame->GetContent());
|
|
uint16_t side =
|
|
textPath->EnumAttributes()[SVGTextPathElement::SIDE].GetAnimValue();
|
|
|
|
gfxFloat offset = GetStartOffset(textPathFrame);
|
|
Float pathLength = path->ComputeLength();
|
|
|
|
// Loop for each text frame in the text path.
|
|
do {
|
|
uint32_t i = it.TextElementCharIndex();
|
|
gfxFloat halfAdvance =
|
|
it.GetGlyphAdvance(context) / mFontSizeScaleFactor / 2.0;
|
|
gfxFloat sign = it.TextRun()->IsRightToLeft() ? -1.0 : 1.0;
|
|
bool vertical = it.TextRun()->IsVertical();
|
|
gfxFloat midx = (vertical ? mPositions[i].mPosition.y
|
|
: mPositions[i].mPosition.x) +
|
|
sign * halfAdvance + offset;
|
|
|
|
// Hide the character if it falls off the end of the path.
|
|
mPositions[i].mHidden = midx < 0 || midx > pathLength;
|
|
|
|
// Position the character on the path at the right angle.
|
|
Point tangent; // Unit vector tangent to the point we find.
|
|
Point pt;
|
|
if (side == TEXTPATH_SIDETYPE_RIGHT) {
|
|
pt = path->ComputePointAtLength(Float(pathLength - midx), &tangent);
|
|
tangent = -tangent;
|
|
} else {
|
|
pt = path->ComputePointAtLength(Float(midx), &tangent);
|
|
}
|
|
Float rotation = vertical ? atan2f(-tangent.x, tangent.y)
|
|
: atan2f(tangent.y, tangent.x);
|
|
Point normal(-tangent.y, tangent.x); // Unit vector normal to the point.
|
|
Point offsetFromPath = normal * (vertical ? -mPositions[i].mPosition.x
|
|
: mPositions[i].mPosition.y);
|
|
pt += offsetFromPath;
|
|
Point direction = tangent * sign;
|
|
mPositions[i].mPosition = ThebesPoint(pt) - ThebesPoint(direction) * halfAdvance;
|
|
mPositions[i].mAngle += rotation;
|
|
|
|
// Position any characters for a partial ligature.
|
|
for (uint32_t j = i + 1;
|
|
j < mPositions.Length() && mPositions[j].mClusterOrLigatureGroupMiddle;
|
|
j++) {
|
|
gfxPoint partialAdvance =
|
|
ThebesPoint(direction) * it.GetGlyphPartialAdvance(j - i, context) /
|
|
mFontSizeScaleFactor;
|
|
mPositions[j].mPosition = mPositions[i].mPosition + partialAdvance;
|
|
mPositions[j].mAngle = mPositions[i].mAngle;
|
|
mPositions[j].mHidden = mPositions[i].mHidden;
|
|
}
|
|
it.Next();
|
|
} while (it.TextPathFrame() &&
|
|
it.TextPathFrame()->GetContent() == textPath);
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DoAnchoring()
|
|
{
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
CharIterator it(this, CharIterator::eOriginal, /* aSubtree */ nullptr);
|
|
|
|
// Don't need to worry about skipped or trimmed characters.
|
|
while (!it.AtEnd() &&
|
|
(it.IsOriginalCharSkipped() || it.IsOriginalCharTrimmed())) {
|
|
it.Next();
|
|
}
|
|
|
|
bool vertical = GetWritingMode().IsVertical();
|
|
uint32_t start = it.TextElementCharIndex();
|
|
while (start < mPositions.Length()) {
|
|
it.AdvanceToCharacter(start);
|
|
nsTextFrame* chunkFrame = it.TextFrame();
|
|
|
|
// Measure characters in this chunk to find the left-most and right-most
|
|
// edges of all glyphs within the chunk.
|
|
uint32_t index = it.TextElementCharIndex();
|
|
uint32_t end = start;
|
|
gfxFloat left = std::numeric_limits<gfxFloat>::infinity();
|
|
gfxFloat right = -std::numeric_limits<gfxFloat>::infinity();
|
|
do {
|
|
if (!it.IsOriginalCharSkipped() && !it.IsOriginalCharTrimmed()) {
|
|
gfxFloat advance = it.GetAdvance(presContext) / mFontSizeScaleFactor;
|
|
gfxFloat pos =
|
|
it.TextRun()->IsVertical() ? mPositions[index].mPosition.y
|
|
: mPositions[index].mPosition.x;
|
|
if (it.TextRun()->IsRightToLeft()) {
|
|
left = std::min(left, pos - advance);
|
|
right = std::max(right, pos);
|
|
} else {
|
|
left = std::min(left, pos);
|
|
right = std::max(right, pos + advance);
|
|
}
|
|
}
|
|
it.Next();
|
|
index = end = it.TextElementCharIndex();
|
|
} while (!it.AtEnd() && !mPositions[end].mStartOfChunk);
|
|
|
|
if (left != std::numeric_limits<gfxFloat>::infinity()) {
|
|
bool isRTL =
|
|
chunkFrame->StyleVisibility()->mDirection == NS_STYLE_DIRECTION_RTL;
|
|
TextAnchorSide anchor =
|
|
ConvertLogicalTextAnchorToPhysical(chunkFrame->StyleSVG()->mTextAnchor,
|
|
isRTL);
|
|
|
|
ShiftAnchoredChunk(mPositions, start, end, left, right, anchor,
|
|
vertical);
|
|
}
|
|
|
|
start = it.TextElementCharIndex();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DoGlyphPositioning()
|
|
{
|
|
mPositions.Clear();
|
|
RemoveStateBits(NS_STATE_SVG_POSITIONING_DIRTY);
|
|
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (kid && NS_SUBTREE_DIRTY(kid)) {
|
|
MOZ_ASSERT(false, "should have already reflowed the kid");
|
|
return;
|
|
}
|
|
|
|
// Since we can be called directly via GetBBoxContribution, our correspondence
|
|
// may not be up to date.
|
|
TextNodeCorrespondenceRecorder::RecordCorrespondence(this);
|
|
|
|
// Determine the positions of each character in app units.
|
|
nsTArray<nsPoint> charPositions;
|
|
DetermineCharPositions(charPositions);
|
|
|
|
if (charPositions.IsEmpty()) {
|
|
// No characters, so nothing to do.
|
|
return;
|
|
}
|
|
|
|
// If the textLength="" attribute was specified, then we need ResolvePositions
|
|
// to record that a new run starts with each glyph.
|
|
SVGTextContentElement* element = static_cast<SVGTextContentElement*>(GetContent());
|
|
nsSVGLength2* textLengthAttr =
|
|
element->GetAnimatedLength(nsGkAtoms::textLength);
|
|
bool adjustingTextLength = textLengthAttr->IsExplicitlySet();
|
|
float expectedTextLength = textLengthAttr->GetAnimValue(element);
|
|
|
|
if (adjustingTextLength && expectedTextLength < 0.0f) {
|
|
// If textLength="" is less than zero, ignore it.
|
|
adjustingTextLength = false;
|
|
}
|
|
|
|
// Get the x, y, dx, dy, rotate values for the subtree.
|
|
nsTArray<gfxPoint> deltas;
|
|
if (!ResolvePositions(deltas, adjustingTextLength)) {
|
|
// If ResolvePositions returned false, it means either there were some
|
|
// characters in the DOM but none of them are displayed, or there was
|
|
// an error in processing mPositions. Clear out mPositions so that we don't
|
|
// attempt to do any painting later.
|
|
mPositions.Clear();
|
|
return;
|
|
}
|
|
|
|
// XXX We might be able to do less work when there is at most a single
|
|
// x/y/dx/dy position.
|
|
|
|
// Truncate the positioning arrays to the actual number of characters present.
|
|
TruncateTo(deltas, charPositions);
|
|
TruncateTo(mPositions, charPositions);
|
|
|
|
// Fill in an unspecified character position at index 0.
|
|
if (!mPositions[0].IsXSpecified()) {
|
|
mPositions[0].mPosition.x = 0.0;
|
|
}
|
|
if (!mPositions[0].IsYSpecified()) {
|
|
mPositions[0].mPosition.y = 0.0;
|
|
}
|
|
if (!mPositions[0].IsAngleSpecified()) {
|
|
mPositions[0].mAngle = 0.0;
|
|
}
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
bool vertical = GetWritingMode().IsVertical();
|
|
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
double factor = cssPxPerDevPx / mFontSizeScaleFactor;
|
|
|
|
// Determine how much to compress or expand glyph positions due to
|
|
// textLength="" and lengthAdjust="".
|
|
double adjustment = 0.0;
|
|
mLengthAdjustScaleFactor = 1.0f;
|
|
if (adjustingTextLength) {
|
|
nscoord frameLength = vertical ? PrincipalChildList().FirstChild()->GetRect().height
|
|
: PrincipalChildList().FirstChild()->GetRect().width;
|
|
float actualTextLength =
|
|
static_cast<float>(presContext->AppUnitsToGfxUnits(frameLength) * factor);
|
|
|
|
uint16_t lengthAdjust =
|
|
element->EnumAttributes()[SVGTextContentElement::LENGTHADJUST].GetAnimValue();
|
|
switch (lengthAdjust) {
|
|
case LENGTHADJUST_SPACINGANDGLYPHS:
|
|
// Scale the glyphs and their positions.
|
|
if (actualTextLength > 0) {
|
|
mLengthAdjustScaleFactor = expectedTextLength / actualTextLength;
|
|
}
|
|
break;
|
|
|
|
default:
|
|
MOZ_ASSERT(lengthAdjust == LENGTHADJUST_SPACING);
|
|
// Just add space between each glyph.
|
|
int32_t adjustableSpaces = 0;
|
|
for (uint32_t i = 1; i < mPositions.Length(); i++) {
|
|
if (!mPositions[i].mUnaddressable) {
|
|
adjustableSpaces++;
|
|
}
|
|
}
|
|
if (adjustableSpaces) {
|
|
adjustment = (expectedTextLength - actualTextLength) / adjustableSpaces;
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
|
|
// Fill in any unspecified character positions based on the positions recorded
|
|
// in charPositions, and also add in the dx/dy values.
|
|
if (!deltas.IsEmpty()) {
|
|
mPositions[0].mPosition += deltas[0];
|
|
}
|
|
|
|
gfxFloat xLengthAdjustFactor = vertical ? 1.0 : mLengthAdjustScaleFactor;
|
|
gfxFloat yLengthAdjustFactor = vertical ? mLengthAdjustScaleFactor : 1.0;
|
|
for (uint32_t i = 1; i < mPositions.Length(); i++) {
|
|
// Fill in unspecified x position.
|
|
if (!mPositions[i].IsXSpecified()) {
|
|
nscoord d = charPositions[i].x - charPositions[i - 1].x;
|
|
mPositions[i].mPosition.x =
|
|
mPositions[i - 1].mPosition.x +
|
|
presContext->AppUnitsToGfxUnits(d) * factor * xLengthAdjustFactor;
|
|
if (!vertical && !mPositions[i].mUnaddressable) {
|
|
mPositions[i].mPosition.x += adjustment;
|
|
}
|
|
}
|
|
// Fill in unspecified y position.
|
|
if (!mPositions[i].IsYSpecified()) {
|
|
nscoord d = charPositions[i].y - charPositions[i - 1].y;
|
|
mPositions[i].mPosition.y =
|
|
mPositions[i - 1].mPosition.y +
|
|
presContext->AppUnitsToGfxUnits(d) * factor * yLengthAdjustFactor;
|
|
if (vertical && !mPositions[i].mUnaddressable) {
|
|
mPositions[i].mPosition.y += adjustment;
|
|
}
|
|
}
|
|
// Add in dx/dy.
|
|
if (i < deltas.Length()) {
|
|
mPositions[i].mPosition += deltas[i];
|
|
}
|
|
// Fill in unspecified rotation values.
|
|
if (!mPositions[i].IsAngleSpecified()) {
|
|
mPositions[i].mAngle = 0.0f;
|
|
}
|
|
}
|
|
|
|
MOZ_ASSERT(mPositions.Length() == charPositions.Length());
|
|
|
|
AdjustChunksForLineBreaks();
|
|
AdjustPositionsForClusters();
|
|
DoAnchoring();
|
|
DoTextPathLayout();
|
|
}
|
|
|
|
bool
|
|
SVGTextFrame::ShouldRenderAsPath(nsTextFrame* aFrame,
|
|
bool& aShouldPaintSVGGlyphs)
|
|
{
|
|
// Rendering to a clip path.
|
|
if (HasAnyStateBits(NS_STATE_SVG_CLIPPATH_CHILD)) {
|
|
aShouldPaintSVGGlyphs = false;
|
|
return true;
|
|
}
|
|
|
|
aShouldPaintSVGGlyphs = true;
|
|
|
|
const nsStyleSVG* style = aFrame->StyleSVG();
|
|
|
|
// Fill is a non-solid paint, has a non-default fill-rule or has
|
|
// non-1 opacity.
|
|
if (!(style->mFill.Type() == eStyleSVGPaintType_None ||
|
|
(style->mFill.Type() == eStyleSVGPaintType_Color &&
|
|
style->mFillOpacity == 1))) {
|
|
return true;
|
|
}
|
|
|
|
// Text has a stroke.
|
|
if (style->HasStroke() &&
|
|
SVGContentUtils::CoordToFloat(static_cast<nsSVGElement*>(GetContent()),
|
|
style->mStrokeWidth) > 0) {
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::ScheduleReflowSVG()
|
|
{
|
|
if (mState & NS_FRAME_IS_NONDISPLAY) {
|
|
ScheduleReflowSVGNonDisplayText(nsIPresShell::eStyleChange);
|
|
} else {
|
|
nsSVGUtils::ScheduleReflowSVG(this);
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::NotifyGlyphMetricsChange()
|
|
{
|
|
// TODO: perf - adding NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY is overly
|
|
// aggressive here. Ideally we would only set that bit when our descendant
|
|
// frame tree changes (i.e. after frame construction).
|
|
AddStateBits(NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY |
|
|
NS_STATE_SVG_POSITIONING_DIRTY);
|
|
nsLayoutUtils::PostRestyleEvent(
|
|
mContent->AsElement(), nsRestyleHint(0),
|
|
nsChangeHint_InvalidateRenderingObservers);
|
|
ScheduleReflowSVG();
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::UpdateGlyphPositioning()
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (!kid) {
|
|
return;
|
|
}
|
|
|
|
if (mState & NS_STATE_SVG_POSITIONING_DIRTY) {
|
|
DoGlyphPositioning();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MaybeResolveBidiForAnonymousBlockChild()
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
|
|
if (kid &&
|
|
kid->GetStateBits() & NS_BLOCK_NEEDS_BIDI_RESOLUTION &&
|
|
PresContext()->BidiEnabled()) {
|
|
MOZ_ASSERT(static_cast<nsBlockFrame*>(do_QueryFrame(kid)),
|
|
"Expect anonymous child to be an nsBlockFrame");
|
|
nsBidiPresUtils::Resolve(static_cast<nsBlockFrame*>(kid));
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::MaybeReflowAnonymousBlockChild()
|
|
{
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (!kid)
|
|
return;
|
|
|
|
NS_ASSERTION(!(kid->GetStateBits() & NS_FRAME_IN_REFLOW),
|
|
"should not be in reflow when about to reflow again");
|
|
|
|
if (NS_SUBTREE_DIRTY(this)) {
|
|
if (mState & NS_FRAME_IS_DIRTY) {
|
|
// If we require a full reflow, ensure our kid is marked fully dirty.
|
|
// (Note that our anonymous nsBlockFrame is not an nsSVGDisplayableFrame, so
|
|
// even when we are called via our ReflowSVG this will not be done for us
|
|
// by nsSVGDisplayContainerFrame::ReflowSVG.)
|
|
kid->AddStateBits(NS_FRAME_IS_DIRTY);
|
|
}
|
|
|
|
TextNodeCorrespondenceRecorder::RecordCorrespondence(this);
|
|
|
|
MOZ_ASSERT(nsSVGUtils::AnyOuterSVGIsCallingReflowSVG(this),
|
|
"should be under ReflowSVG");
|
|
nsPresContext::InterruptPreventer noInterrupts(PresContext());
|
|
DoReflow();
|
|
}
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::DoReflow()
|
|
{
|
|
// Since we are going to reflow the anonymous block frame, we will
|
|
// need to update mPositions.
|
|
// We also mark our text correspondence as dirty since we can end up needing
|
|
// reflow in ways that do not set NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY.
|
|
// (We'd then fail the "expected a TextNodeCorrespondenceProperty" assertion
|
|
// when UpdateGlyphPositioning() is called after we return.)
|
|
AddStateBits(NS_STATE_SVG_TEXT_CORRESPONDENCE_DIRTY |
|
|
NS_STATE_SVG_POSITIONING_DIRTY);
|
|
|
|
if (mState & NS_FRAME_IS_NONDISPLAY) {
|
|
// Normally, these dirty flags would be cleared in ReflowSVG(), but that
|
|
// doesn't get called for non-display frames. We don't want to reflow our
|
|
// descendants every time SVGTextFrame::PaintSVG makes sure that we have
|
|
// valid positions by calling UpdateGlyphPositioning(), so we need to clear
|
|
// these dirty bits. Note that this also breaks an invalidation loop where
|
|
// our descendants invalidate as they reflow, which invalidates rendering
|
|
// observers, which reschedules the frame that is currently painting by
|
|
// referencing us to paint again. See bug 839958 comment 7. Hopefully we
|
|
// will break that loop more convincingly at some point.
|
|
RemoveStateBits(NS_FRAME_IS_DIRTY | NS_FRAME_HAS_DIRTY_CHILDREN);
|
|
}
|
|
|
|
nsPresContext *presContext = PresContext();
|
|
nsIFrame* kid = PrincipalChildList().FirstChild();
|
|
if (!kid)
|
|
return;
|
|
|
|
RefPtr<gfxContext> renderingContext =
|
|
presContext->PresShell()->CreateReferenceRenderingContext();
|
|
|
|
if (UpdateFontSizeScaleFactor()) {
|
|
// If the font size scale factor changed, we need the block to report
|
|
// an updated preferred width.
|
|
kid->MarkIntrinsicISizesDirty();
|
|
}
|
|
|
|
AddStateBits(NS_STATE_SVG_TEXT_IN_REFLOW);
|
|
|
|
nscoord inlineSize = kid->GetPrefISize(renderingContext);
|
|
WritingMode wm = kid->GetWritingMode();
|
|
ReflowInput reflowInput(presContext, kid,
|
|
renderingContext,
|
|
LogicalSize(wm, inlineSize,
|
|
NS_UNCONSTRAINEDSIZE));
|
|
ReflowOutput desiredSize(reflowInput);
|
|
nsReflowStatus status;
|
|
|
|
NS_ASSERTION(reflowInput.ComputedPhysicalBorderPadding() == nsMargin(0, 0, 0, 0) &&
|
|
reflowInput.ComputedPhysicalMargin() == nsMargin(0, 0, 0, 0),
|
|
"style system should ensure that :-moz-svg-text "
|
|
"does not get styled");
|
|
|
|
kid->Reflow(presContext, desiredSize, reflowInput, status);
|
|
kid->DidReflow(presContext, &reflowInput);
|
|
kid->SetSize(wm, desiredSize.Size(wm));
|
|
|
|
RemoveStateBits(NS_STATE_SVG_TEXT_IN_REFLOW);
|
|
}
|
|
|
|
// Usable font size range in devpixels / user-units
|
|
#define CLAMP_MIN_SIZE 8.0
|
|
#define CLAMP_MAX_SIZE 200.0
|
|
#define PRECISE_SIZE 200.0
|
|
|
|
bool
|
|
SVGTextFrame::UpdateFontSizeScaleFactor()
|
|
{
|
|
double oldFontSizeScaleFactor = mFontSizeScaleFactor;
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
bool geometricPrecision = false;
|
|
nscoord min = nscoord_MAX,
|
|
max = nscoord_MIN;
|
|
|
|
// Find the minimum and maximum font sizes used over all the
|
|
// nsTextFrames.
|
|
TextFrameIterator it(this);
|
|
nsTextFrame* f = it.Current();
|
|
while (f) {
|
|
if (!geometricPrecision) {
|
|
// Unfortunately we can't treat text-rendering:geometricPrecision
|
|
// separately for each text frame.
|
|
geometricPrecision = f->StyleText()->mTextRendering ==
|
|
NS_STYLE_TEXT_RENDERING_GEOMETRICPRECISION;
|
|
}
|
|
nscoord size = f->StyleFont()->mFont.size;
|
|
if (size) {
|
|
min = std::min(min, size);
|
|
max = std::max(max, size);
|
|
}
|
|
f = it.Next();
|
|
}
|
|
|
|
if (min == nscoord_MAX) {
|
|
// No text, so no need for scaling.
|
|
mFontSizeScaleFactor = 1.0;
|
|
return mFontSizeScaleFactor != oldFontSizeScaleFactor;
|
|
}
|
|
|
|
double minSize = nsPresContext::AppUnitsToFloatCSSPixels(min);
|
|
|
|
if (geometricPrecision) {
|
|
// We want to ensure minSize is scaled to PRECISE_SIZE.
|
|
mFontSizeScaleFactor = PRECISE_SIZE / minSize;
|
|
return mFontSizeScaleFactor != oldFontSizeScaleFactor;
|
|
}
|
|
|
|
// When we are non-display, we could be painted in different coordinate
|
|
// spaces, and we don't want to have to reflow for each of these. We
|
|
// just assume that the context scale is 1.0 for them all, so we don't
|
|
// get stuck with a font size scale factor based on whichever referencing
|
|
// frame happens to reflow first.
|
|
double contextScale = 1.0;
|
|
if (!(mState & NS_FRAME_IS_NONDISPLAY)) {
|
|
gfxMatrix m(GetCanvasTM());
|
|
if (!m.IsSingular()) {
|
|
contextScale = GetContextScale(m);
|
|
}
|
|
}
|
|
mLastContextScale = contextScale;
|
|
|
|
double maxSize = nsPresContext::AppUnitsToFloatCSSPixels(max);
|
|
|
|
// But we want to ignore any scaling required due to HiDPI displays, since
|
|
// regular CSS text frames will still create text runs using the font size
|
|
// in CSS pixels, and we want SVG text to have the same rendering as HTML
|
|
// text for regular font sizes.
|
|
float cssPxPerDevPx =
|
|
nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
contextScale *= cssPxPerDevPx;
|
|
|
|
double minTextRunSize = minSize * contextScale;
|
|
double maxTextRunSize = maxSize * contextScale;
|
|
|
|
if (minTextRunSize >= CLAMP_MIN_SIZE &&
|
|
maxTextRunSize <= CLAMP_MAX_SIZE) {
|
|
// We are already in the ideal font size range for all text frames,
|
|
// so we only have to take into account the contextScale.
|
|
mFontSizeScaleFactor = contextScale;
|
|
} else if (maxSize / minSize > CLAMP_MAX_SIZE / CLAMP_MIN_SIZE) {
|
|
// We can't scale the font sizes so that all of the text frames lie
|
|
// within our ideal font size range, so we treat the minimum as more
|
|
// important and just scale so that minSize = CLAMP_MIN_SIZE.
|
|
mFontSizeScaleFactor = CLAMP_MIN_SIZE / minTextRunSize;
|
|
} else if (minTextRunSize < CLAMP_MIN_SIZE) {
|
|
mFontSizeScaleFactor = CLAMP_MIN_SIZE / minTextRunSize;
|
|
} else {
|
|
mFontSizeScaleFactor = CLAMP_MAX_SIZE / maxTextRunSize;
|
|
}
|
|
|
|
return mFontSizeScaleFactor != oldFontSizeScaleFactor;
|
|
}
|
|
|
|
double
|
|
SVGTextFrame::GetFontSizeScaleFactor() const
|
|
{
|
|
return mFontSizeScaleFactor;
|
|
}
|
|
|
|
/**
|
|
* Take aPoint, which is in the <text> element's user space, and convert
|
|
* it to the appropriate frame user space of aChildFrame according to
|
|
* which rendered run the point hits.
|
|
*/
|
|
Point
|
|
SVGTextFrame::TransformFramePointToTextChild(const Point& aPoint,
|
|
nsIFrame* aChildFrame)
|
|
{
|
|
NS_ASSERTION(aChildFrame &&
|
|
nsLayoutUtils::GetClosestFrameOfType
|
|
(aChildFrame->GetParent(), LayoutFrameType::SVGText) == this,
|
|
"aChildFrame must be a descendant of this frame");
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
// Add in the mRect offset to aPoint, as that will have been taken into
|
|
// account when transforming the point from the ancestor frame down
|
|
// to this one.
|
|
float cssPxPerDevPx = nsPresContext::AppUnitsToFloatCSSPixels(presContext->AppUnitsPerDevPixel());
|
|
float factor = nsPresContext::AppUnitsPerCSSPixel();
|
|
Point framePosition(NSAppUnitsToFloatPixels(mRect.x, factor),
|
|
NSAppUnitsToFloatPixels(mRect.y, factor));
|
|
Point pointInUserSpace = aPoint * cssPxPerDevPx + framePosition;
|
|
|
|
// Find the closest rendered run for the text frames beneath aChildFrame.
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eAllFrames,
|
|
aChildFrame);
|
|
TextRenderedRun hit;
|
|
gfxPoint pointInRun;
|
|
nscoord dx = nscoord_MAX;
|
|
nscoord dy = nscoord_MAX;
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
uint32_t flags = TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeStroke |
|
|
TextRenderedRun::eNoHorizontalOverflow;
|
|
gfxRect runRect = run.GetRunUserSpaceRect(presContext, flags).ToThebesRect();
|
|
|
|
gfxMatrix m = run.GetTransformFromRunUserSpaceToUserSpace(presContext);
|
|
if (!m.Invert()) {
|
|
return aPoint;
|
|
}
|
|
gfxPoint pointInRunUserSpace = m.TransformPoint(ThebesPoint(pointInUserSpace));
|
|
|
|
if (Inside(runRect, pointInRunUserSpace)) {
|
|
// The point was inside the rendered run's rect, so we choose it.
|
|
dx = 0;
|
|
dy = 0;
|
|
pointInRun = pointInRunUserSpace;
|
|
hit = run;
|
|
} else if (nsLayoutUtils::PointIsCloserToRect(pointInRunUserSpace,
|
|
runRect, dx, dy)) {
|
|
// The point was closer to this rendered run's rect than any others
|
|
// we've seen so far.
|
|
pointInRun.x = clamped(pointInRunUserSpace.x,
|
|
runRect.X(), runRect.XMost());
|
|
pointInRun.y = clamped(pointInRunUserSpace.y,
|
|
runRect.Y(), runRect.YMost());
|
|
hit = run;
|
|
}
|
|
}
|
|
|
|
if (!hit.mFrame) {
|
|
// We didn't find any rendered runs for the frame.
|
|
return aPoint;
|
|
}
|
|
|
|
// Return the point in user units relative to the nsTextFrame,
|
|
// but taking into account mFontSizeScaleFactor.
|
|
gfxMatrix m = hit.GetTransformFromRunUserSpaceToFrameUserSpace(presContext);
|
|
m.PreScale(mFontSizeScaleFactor, mFontSizeScaleFactor);
|
|
return ToPoint(m.TransformPoint(pointInRun) / cssPxPerDevPx);
|
|
}
|
|
|
|
/**
|
|
* For each rendered run beneath aChildFrame, translate aRect from
|
|
* aChildFrame to the run's text frame, transform it then into
|
|
* the run's frame user space, intersect it with the run's
|
|
* frame user space rect, then transform it up to user space.
|
|
* The result is the union of all of these.
|
|
*/
|
|
gfxRect
|
|
SVGTextFrame::TransformFrameRectFromTextChild(const nsRect& aRect,
|
|
const nsIFrame* aChildFrame)
|
|
{
|
|
NS_ASSERTION(aChildFrame &&
|
|
nsLayoutUtils::GetClosestFrameOfType
|
|
(aChildFrame->GetParent(), LayoutFrameType::SVGText) == this,
|
|
"aChildFrame must be a descendant of this frame");
|
|
|
|
UpdateGlyphPositioning();
|
|
|
|
nsPresContext* presContext = PresContext();
|
|
|
|
gfxRect result;
|
|
TextRenderedRunIterator it(this, TextRenderedRunIterator::eAllFrames,
|
|
aChildFrame);
|
|
for (TextRenderedRun run = it.Current(); run.mFrame; run = it.Next()) {
|
|
// First, translate aRect from aChildFrame to this run's frame.
|
|
nsRect rectInTextFrame = aRect + aChildFrame->GetOffsetTo(run.mFrame);
|
|
|
|
// Scale it into frame user space.
|
|
gfxRect rectInFrameUserSpace =
|
|
AppUnitsToFloatCSSPixels(gfxRect(rectInTextFrame.x,
|
|
rectInTextFrame.y,
|
|
rectInTextFrame.width,
|
|
rectInTextFrame.height), presContext);
|
|
|
|
// Intersect it with the run.
|
|
uint32_t flags = TextRenderedRun::eIncludeFill |
|
|
TextRenderedRun::eIncludeStroke;
|
|
|
|
if (rectInFrameUserSpace.IntersectRect(rectInFrameUserSpace,
|
|
run.GetFrameUserSpaceRect(presContext, flags).ToThebesRect())) {
|
|
// Transform it up to user space of the <text>, also taking into
|
|
// account the font size scale.
|
|
gfxMatrix m = run.GetTransformFromRunUserSpaceToUserSpace(presContext);
|
|
m.PreScale(mFontSizeScaleFactor, mFontSizeScaleFactor);
|
|
gfxRect rectInUserSpace = m.TransformRect(rectInFrameUserSpace);
|
|
|
|
// Union it into the result.
|
|
result.UnionRect(result, rectInUserSpace);
|
|
}
|
|
}
|
|
|
|
// Subtract the mRect offset from the result, as our user space for
|
|
// this frame is relative to the top-left of mRect.
|
|
float factor = nsPresContext::AppUnitsPerCSSPixel();
|
|
gfxPoint framePosition(NSAppUnitsToFloatPixels(mRect.x, factor),
|
|
NSAppUnitsToFloatPixels(mRect.y, factor));
|
|
|
|
return result - framePosition;
|
|
}
|
|
|
|
void
|
|
SVGTextFrame::AppendDirectlyOwnedAnonBoxes(nsTArray<OwnedAnonBox>& aResult)
|
|
{
|
|
MOZ_ASSERT(PrincipalChildList().FirstChild(), "Must have our anon box");
|
|
aResult.AppendElement(OwnedAnonBox(PrincipalChildList().FirstChild()));
|
|
}
|